* doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
[official-gcc.git] / gcc / ChangeLog
blobc5b794e6ad91555ab7350d69e4b6d219b37c2192
1 2013-10-21  Ian Lance Taylor  <iant@google.com>
3         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
4         don't imply that attributes can solve all problems.
5         (Directory Options): Fix typo.
7 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
10         extend and extend_add.
12 2013-10-21  Richard Biener  <rguenther@suse.de>
14         PR tree-optimization/58794
15         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
16         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
18 2013-10-21  Richard Biener  <rguenther@suse.de>
20         PR middle-end/58742
21         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
22         to (T) X for sign-changing conversions (or no conversion).
24 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
26         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
28 2013-10-20  Jan Hubicka  <jh@suse.cz>
30         * config/i386/i386-tune.def: Add comment; organize into categories
32 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
34         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
35         argument.  Update function comment.
36         (expand_set_or_movmem_via_rep): New function combining
37         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
38         (expand_movmem_via_rep_mov): Remove.
39         expand_setmem_via_rep_stos): Remove.
40         (expand_movmem_epilogue): Update calls correspondingly.
41         (expand_setmem_epilogue_via_loop): Likewise.
42         (emit_memset): New.
43         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
44         (expand_set_or_movmem_prologue): New function combining
45         expand_movmem_prologue and expand_setmem_prologue.
46         (expand_movmem_prologue): Remove.
47         (expand_setmem_prologue): Remove.
48         (expand_set_or_movmem_constant_prologue): New function combining
49         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
50         (expand_constant_movmem_prologue): Remove.
51         (expand_constant_setmem_prologue): Remove.
52         (promote_duplicated_reg): Allow vector-const0 value.
53         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
54         and ix86_expand_setmem.
55         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
56         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
58 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
60         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
61         little endian.
62         (vec_unpacku_hi_v8hi): Likewise.
63         (vec_unpacku_lo_v16qi): Likewise.
64         (vec_unpacku_lo_v8hi): Likewise.
66 2013-10-20  Jan Hubicka  <jh@suse.cz>
68         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
69         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
70         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
72 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
74         * config/mips/mips.h (ISA_HAS_WSBH): Define.
75         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
76         constants.
77         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
79 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
81         PR target/58603
82         * system.h: Undef m_slot.
84 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
86         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
87         all elements for both endian flavors.
89 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
91         PR target/58792
92         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
93         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
94         and SI_REG for 64bit SYSV ABI targets.
96 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
98         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
99         to multi_reg_return.  Clarify that we are skipping USEs of multiple
100         return registers.  Use bool type where appropriate.
102 2013-10-18  Jan Hubicka  <jh@suse.cz>
104         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
105         for cold functions.
106         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
107         (X86_TUNE_PUSH_MEMORY): Likewise.
108         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
109         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
110         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
111         New.
112         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
113         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
114         Remove.
115         (ix86_option_override_internal): Update to use tune features instead
116         of variables.
118 2013-10-18  Cong Hou  <congh@google.com>
120         PR tree-optimization/58508
121         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
122         statement that contains data refs with zero-step.
124 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
126         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
127         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
128         * gimple-low.c (gimple_check_call_arg,
129         gimple_check_call_matching_types): Move to cgraph.c.
130         * gimple-low.h: Remove prototype.
131         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
132         Relocate from gimple-low.c.
133         * cgraph.h: Add prototype.  Don't include basic-block.h.
134         * gimplify.c: Add gimple-low to include list.
135         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
136         * tree-eh.c: Add gimple-low to include list.
137         * tree-nested.c: Likewise.
138         * cfgexpand.c: Add tree-ssa-address.h to include list.
139         * expr.c: Likewise.
140         * gimple-fold.c: Likewise.
141         * gimple-ssa-strength-reduction.c: Likewise.
142         * trans-mem.c: Likewise.
143         * tree-mudflap.c: Likewise.
144         * tree-ssa-loop-ivopts.c: Likewise.
145         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
146         (degenerate_phi_result): Move to tree-phinodes.c.
147         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
148         * tree-ssa-threadedge.c: Likewise.
149         * tree-vrp.c: Likewise.
150         * tree-phinodes.c (degenerate_phi_result): Relocate here.
151         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
152         * tree-phinodes.h (degenerate_phi_result): Add prototype.
153         * tree-ssa-copy.c: Include tree-ssa-dom.h.
154         * tree-ssa-forwprop.c: Likewise.
155         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
156         pass_data_cleanup_cfg_post_optimizing,
157         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
158         * tree-optimize.c: Delete File.
159         * graphite.c: Include tree-cfgcleanup.h.
160         * passes.c: Likewise.
161         * tree-cfg.c: Likewise.
162         * tree-profile.c: Likewise.
163         * tree-ssa-dse.c: Likewise.
164         * tree-ssa-loop-ivcanon.c: Likewise.
165         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
166         * tree-outof-ssa.c: Include sbitmap.h.
167         * tree-ssa-live.c: Likewise.
168         * tree-ssa-propagate.c: Likewise.
169         * tree-ssa-structalias.c: Likewise.
170         * tree-stdarg.c: Likewise.
171         * Makefile.in (OBJS): Delete tree-optimize.o.
172         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
173         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
174         * varasm.c: Include basic-block.h.
175         * cfgloop.h: Include function.h instead of basic-block.h
176         (bb_loop_depth): Move to cfgloop.c.
177         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
179 2013-10-18  Teresa Johnson  <tejohnson@google.com>
181         * predict.c (probably_never_executed): Compare frequency-based
182         count to number of training runs.
183         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
185 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
187         * config/arm/arm.c (cortexa9_extra_costs): New table.
188         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
190 2013-10-18  Jeff Law  <law@redhat.com>
192         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
194         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
195         (register_jump_thread): Add "registered_jump_thread" debug
196         counter support.
197         * dbgcnt.def (registered_jump_thread): New debug counter.
199 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
201         * config/rs6000/rs6000.c: Include cgraph.h.
203 2013-10-18  Teresa Johnson  <tejohnson@google.com>
205         * tree-ssa-tail-merge.c (replace_block_by): Update edge
206         weights during merging.
208 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
210         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
211         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
212         * tree-ssa-operands.h: Remove prototype.
213         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
214         * gimple.c (virtual_operand_p): Relocate from gimple.c.
215         * gimple.h: Add prototype.
216         * gimple-ssa.h: Include tree-ssa-operands.h.
217         * tree-dump.c: Add tree-cfg.h to include list.
218         * tree-ssa-alias.c: Add ipa-reference.h to include list.
219         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
220         * config/i386/i386.c: Don't include tree-flow.h.
221         * config/rs6000/rs6000.c: Likewise.
223 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
225         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
227 2013-10-18  Richard Biener  <rguenther@suse.de>
229         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
230         or TYPE_UNSIGNED of integral types.
231         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
232         NULL_TREE for zero-precision integral types.
234 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
236         * config/aarch64/arm_neon.h
237         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
239 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
241         * opth-gen.awk: Define target_flags_explicit.
243 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
245         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
246         fields to the reg_addr array that describes the valid addressing
247         mode for any register, general purpose registers, floating point
248         registers, and Altivec registers.
249         (FIRST_RELOAD_REG_CLASS): Likewise.
250         (LAST_RELOAD_REG_CLASS): Likewise.
251         (struct reload_reg_map_type): Likewise.
252         (reload_reg_map_type): Likewise.
253         (RELOAD_REG_VALID): Likewise.
254         (RELOAD_REG_MULTIPLE): Likewise.
255         (RELOAD_REG_INDEXED): Likewise.
256         (RELOAD_REG_OFFSET): Likewise.
257         (RELOAD_REG_PRE_INCDEC): Likewise.
258         (RELOAD_REG_PRE_MODIFY): Likewise.
259         (reg_addr): Likewise.
260         (mode_supports_pre_incdec_p): New helper functions to say whether
261         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
262         (mode_supports_pre_modify_p): Likewise.
263         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
264         print the valid address mode bits for each mode.
265         (rs6000_debug_print_mode): Likewise.
266         (rs6000_debug_reg_global): Likewise.
267         (rs6000_setup_reg_addr_masks): New function to set up the address
268         mask bits for each type.
269         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
270         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
271         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
272         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
273         PRE_MODIFY are supported.
274         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
275         registers, instead of {src,dest}_av_p.
276         (rs6000_print_options_internal): Tweak the debug output slightly.
278 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
280         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
281         isa attribute.
283 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
285         * tree-flow.h (struct omp_region): Move to omp-low.c.
286         Remove omp_ prototypes and variables.
287         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
288         (copy_var_decl): Relocate prototype from tree-flow.h.
289         * gimple.c (copy_var_decl): Relocate from omp-low.c.
290         * tree.h: Move prototype to omp-low.h.
291         * omp-low.h: New File.  Relocate prototypes here.
292         * omp-low.c (struct omp_region): Make local here.
293         (root_omp_region): Make static.
294         (copy_var_decl) Move to gimple.c.
295         (new_omp_region): Make static.
296         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
297         * tree-cfg.c: Include omp-low.h.
298         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
299         * gimplify.c: Include omp-low.h.
300         * tree-parloops.c: Likewise.
302 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
304         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
305         operands are in memory, prefer to force non-matched operand 1 to
306         the register.
308 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
310         PR target/58673
311         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
312         restrict TImode addresses to single indirect registers if both
313         -mquad-memory and -mvsx-timode are used.
314         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
315         we should emit load/store quad.  Remove using %y for quad memory
316         addresses.
318         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
319         constraints to allow load/store quad on machines where TImode is
320         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
321         for TImode to load integer constants.
323 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
325         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
327 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
329         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
330         handling of STACK_REG.
332 2013-10-17  Richard Biener  <rguenther@suse.de>
334         PR tree-optimization/58143
335         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
336         New function.
337         (rewrite_to_defined_overflow): Likewise.
338         (move_computations_dom_walker::before_dom): Rewrite stmts
339         with undefined signed overflow that are not always executed
340         into unsigned arithmetic.
342 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
344         PR target/57756
345         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
346         explicit isa flag to be an options variable, instead of using
347         global_options_set.  Remove define from rs6000.h.
348         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
350         * config/rs6000/rs6000.c (rs6000_option_override_internal):
351         Initialize rs6000_isa_flags_explicit.
352         (rs6000_function_specific_save): Add gcc_options* parameter, so
353         that the powerpc builds after the 2013-10-15 changes.
354         (rs6000_function_specific_restore): Likewise.
356 2013-10-16  DJ Delorie  <dj@redhat.com>
358         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
359         op is a REG before checking REGNO.
360         (rl78_alloc_physical_registers): Verify pattern is a SET before
361         checking SET_SRC.
363 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
365         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
366         endianness.
367         (vec_unpacks_lo_v4sf): Likewise.
368         (vec_unpacks_float_hi_v4si): Likewise.
369         (vec_unpacks_float_lo_v4si): Likewise.
370         (vec_unpacku_float_hi_v4si): Likewise.
371         (vec_unpacku_float_lo_v4si): Likewise.
373 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
375         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
376         (vsx_concat_v2sf): Likewise.
378 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
380         * config/aarch64/aarch64.md
381         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
383 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
385         PR tree-optimization/58697
386         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
387         estimated_loop_iterations_int.
388         (max_stmt_executions_int): Call get_max_loop_iterations_int.
389         (get_max_loop_iterations_int): New.  HWINT version of
390         get_max_loop_iterations.
391         * cfgloop.h: Add prototypes.
392         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
393         * loop-unroll.c (decide_peel_once_rolling): Call
394         get_estimated_loop_iterations_int.
395         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
396         * tree-ssa-loop-niter.h: Tweak prototypes.
398 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
400         * gengtype-parse.c (struct_field_seq): Ignore access-control
401         keywords ("public:" etc).
403 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
405         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
406         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
408 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
410         * config/arm/arm.opt (mlra): New option.
411         * config/arm/arm.c (arm_lra_p): New function.
412         (TARGET_LRA_P): Define.
414 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
416         * tree-core.h (tree_code_name): Remove.
417         * tree.h (get_tree_code_name): New prototype.
418         * tree.c (tree_code_name): Make static.
419         (get_tree_code_name): New function.
420         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
421         tree_class_check_failed, tree_range_check_failed,
422         tree_not_class_check_failed, omp_clause_check_failed,
423         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
424         wrapper get_tree_code_name instead of calling tree_code_name directly.
425         * tree-vrp.c (dump_asserts_for): Likewise.
426         * tree-dump.c (dequeue_and_dump): Likewise.
427         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
428         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
429         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
430         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
431         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
432         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
433         dump_gimple_omp_for): Likewise.
434         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
435         * tree-ssa-pre.c (print_pre_expr): Likewise.
436         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
437         * print-tree.c (print_node_brief, print_node): Likewise.
438         * gimple.c (gimple_check_failed): Likewise.
439         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
440         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
441         * config/mep/mep.c (mep_validate_vliw): Likewise.
442         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
443         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
445 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
447         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
448         for AMD bdver3.
450 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
452         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
453         (MULTILIB_MATCHES): Add multilib for -march=v8.
455 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
457         PR target/57756
458         * optc-save-gen.awk: Add extra parameter to the save and restore
459         target calls.
460         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
461         * tree.c (build_optimization_node): New parameter.  Add extra parameter
462         to call to cl_optimization_save.
463         (build_target_option_node): New parameter. Add extra parameter
464         to call to cl_target_option_save.
465         * tree.h (build_optimization_node): New parameter.
466         (build_target_option_node): New parameter.
467         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
468         build_optimization_node and build_target_option_node.
469         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
470         (handle_pragma_push_options): Ditto.
471         * toplev.c (process_options): Ditto.
472         * opts.c (init_options_struct): Check for opts_set non-null.
473         * target.def (target_option.save): New parameter.
474         (target_option.restore): New parameter.
475         * tm.texi: Generate.
476         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
477         (ix86_pragma_target_parse): Ditto.
478         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
479         parameters.
480         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
481         to  build_optimization_node and build_target_option_node.
482         (rs6000_valid_attribute_p): Ditto.
483         (rs6000_pragma_target_parse): Ditto.
484         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
485         data.
486         * config/i386/i386.h:
487         TARGET_64BIT_P: New Macro
488         TARGET_MMX_P: New Macro.
489         TARGET_3DNOW_P: New Macro.
490         TARGET_3DNOW_A_P: New Macro.
491         TARGET_SSE_P: New Macro.
492         TARGET_SSE2_P: New Macro.
493         TARGET_SSE3_P: New Macro.
494         TARGET_SSSE3_P: New Macro.
495         TARGET_SSE4_1_P: New Macro.
496         TARGET_SSE4_2_P: New Macro.
497         TARGET_AVX_P: New Macro.
498         TARGET_AVX2_P: New Macro.
499         TARGET_AVX512F_P: New Macro.
500         TARGET_AVX512PF_P: New Macro.
501         TARGET_AVX512ER_P: New Macro.
502         TARGET_AVX512CD_P: New Macro.
503         TARGET_FMA_P: New Macro.
504         TARGET_SSE4A_P: New Macro.
505         TARGET_FMA4_P: New Macro.
506         TARGET_XOP_P: New Macro.
507         TARGET_LWP_P: New Macro.
508         TARGET_ABM_P: New Macro.
509         TARGET_BMI_P: New Macro.
510         TARGET_BMI2_P: New Macro.
511         TARGET_LZCNT_P: New Macro.
512         TARGET_TBM_P: New Macro.
513         TARGET_POPCNT_P: New Macro.
514         TARGET_SAHF_P: New Macro.
515         TARGET_MOVBE_P: New Macro.
516         TARGET_CRC32_P: New Macro.
517         TARGET_AES_P: New Macro.
518         TARGET_PCLMUL_P: New Macro.
519         TARGET_CMPXCHG16B_P: New Macro.
520         TARGET_FSGSBASE_P: New Macro.
521         TARGET_RDRND_P: New Macro.
522         TARGET_F16C_P: New Macro.
523         TARGET_RTM_P: New Macro.
524         TARGET_HLE_P: New Macro.
525         TARGET_RDSEED_P: New Macro.
526         TARGET_PRFCHW_P: New Macro.
527         TARGET_ADX_P: New Macro.
528         TARGET_FXSR_P: New Macro.
529         TARGET_XSAVE_P: New Macro.
530         TARGET_XSAVEOPT_P: New Macro.
531         TARGET_LP64_P: New Macro.
532         TARGET_X32_P: New Macro.
533         TARGET_FPMATH_DEFAULT_P: New Macro.
534         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
535         * config/i386/i386.c (ix86_option_override_internal): New parameters.
536         opts and opts_set.
537         Change ix86_tune_string to access opts->x_ix86_tune_string.
538         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
539         Change ix86_arch_string to access opts->x_ix86_arch_string.
540         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
541         Change ix86_pmode to access opts->x_ix86_pmode.
542         Change ix86_abi to access opts->x_ix86_abi.
543         Change ix86_cmodel to access opts->x_ix86_cmodel.
544         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
545         Change ix86_isa_flags_explicit to access
546         opts->x_ix86_isa_flags_explicit.
547         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
548         Change ix86_regparm to access opts->x_ix86_regparm.
549         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
550         Change ix86_preferred_stack_boundary_arg to access
551         opts->x_ix86_preferred_stack_boundary_arg.
552         Change ix86_force_align_arg_pointer to access
553         opts->x_ix86_force_align_arg_pointer.
554         Change ix86_incoming_stack_boundar_arg to access
555         opts->x_ix86_incoming_stack_boundar_arg.
556         Change ix86_fpmath to access opts->x_ix86_fpmath.
557         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
558         Change ix86_recip_name to access opts->x_ix86_recip_name.
559         Change ix86_stack_protector_guard to access
560         opts->x_ix86_stack_protector_guard.
561         Change ix86_tune_memcpy_strategy to access
562         opts->x_ix86_tune_memcpy_strategy.
563         Change ix86_tune_memset_strategy to access
564         opts->x_ix86_tune_memset_strategy.
565         Change global_options to access opts.
566         Change global_options_set to access opts_set.
567         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
568         Change TARGET_MMX to TARGET_MMX_P (opts->...).
569         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
570         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
571         Change TARGET_SSE to TARGET_SSE_P (opts->...).
572         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
573         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
574         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
575         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
576         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
577         Change TARGET_AVX to TARGET_AVX_P (opts->...).
578         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
579         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
580         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
581         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
582         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
583         Change TARGET_FMA to TARGET_FMA_P (opts->...).
584         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
585         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
586         Change TARGET_XOP to TARGET_XOP_P (opts->...).
587         Change TARGET_LWP to TARGET_LWP_P (opts->...).
588         Change TARGET_ABM to TARGET_ABM_P (opts->...).
589         Change TARGET_BMI to TARGET_BMI_P (opts->...).
590         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
591         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
592         Change TARGET_TBM to TARGET_TBM_P (opts->...).
593         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
594         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
595         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
596         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
597         Change TARGET_AES to TARGET_AES_P (opts->...).
598         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
599         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
600         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
601         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
602         Change TARGET_F16C to TARGET_F16C_P (opts->...).
603         Change TARGET_RTM to TARGET_RTM_P (opts->...).
604         Change TARGET_HLE to TARGET_HLE_P (opts->...).
605         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
606         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
607         Change TARGET_ADX to TARGET_ADX_P (opts->...).
608         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
609         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
610         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
611         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
612         Change TARGET_X32 to TARGET_X32_P (opts->...).
613         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
614         Change TARGET_FLOAT_RETURNS_IN_80387 to
615         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
616         (ix86_function_specific_save): New parameter. Use opts-> fields
617         to replace global fields.
618         (ix86_function_specific_restore): Ditto.
619         (ix86_valid_target_attribute_inner_p): New parameters.
620         Fix recursive call.
621         Fix call to ix86_handle_option and set_option.
622         (ix86_valid_target_attribute_tree): New parameters.
623         Change global_options to access opts.
624         Change global_options_set to access opts_set.
625         Fix call to ix86_valid_target_attribute_inner_p.
626         Change ix86_tune_string to access opts->x_ix86_tune_string.
627         Change ix86_arch_string to access opts->x_ix86_arch_string.
628         Change ix86_fpmath to access opts->x_ix86_fpmath
629         Fix call to ix86_option_override_internal.
630         Fix call to ix86_add_new_builtins.
631         Fix calls to build_optimization_node and build_target_option_node.
632         (ix86_valid_target_attribute_p): Remove access to global_options.
633         Use new gcc_options structure func_options.
634         Fix call to ix86_valid_target_attribute_tree.
635         Fix call to  build_optimization_node.
636         (get_builtin_code_for_version): Fix call to
637         ix86_valid_target_attribute_tree.
639 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
641         * Makefile.in (PICFLAG): New.
642         (enable_host_shared): New.
643         (INTERNAL_CFLAGS): Use PICFLAG.
644         (LIBIBERTY): Use pic build of libiberty.a if configured with
645         --enable-host-shared.
646         * configure.ac: Add --enable-host-shared, setting up new
647         PICFLAG variable.
648         * configure: Regenerate.
649         * doc/install.texi (--enable-shared): Add note contrasting it with ...
650         (--enable-host-shared): New option.
652 2013-10-15  Richard Biener  <rguenther@suse.de>
654         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
655         for built-in functions.
657 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
659         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
660         (optimize_range_tests_1): New function,
661         extracted from optimize_range_tests.
662         (optimize_range_tests_xor): Similarly.
663         (optimize_range_tests_diff): New function.
664         (optimize_range_tests): Use optimize_range_tests_1.
666 2013-10-15  Cong Hou  <congh@google.com>
668         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
669         requirement of the reduction pattern so that one operand of the
670         reduction operation can come from outside of the loop.
672 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
674         * config/arm/neon-schedgen.ml: Remove.
675         * config/arm/cortex-a9-neon.md: Remove comment regarding
676         neon-schedgen.ml.
678 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
680         * config/arm/types: Remove old neon types.
682 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
684         * config/arm/cortex-a7.md
685         (cortex_a7_neon_type): New.
686         (cortex_a7_neon_mul): Update for new types.
687         (cortex_a7_neon_mla): Likewise.
688         (cortex_a7_neon): Likewise.
690 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
692         * config/arm/cortex-a15-neon.md
693         (cortex_a15_neon_type): New,
695         (cortex_a15_neon_int_1): Remove.
696         (cortex_a15_neon_int_2): Likewise.
697         (cortex_a15_neon_int_3): Likewise.
698         (cortex_a15_neon_int_4): Likewise.
699         (cortex_a15_neon_int_5): Likewise.
700         (cortex_a15_neon_vqneg_vqabs): Likewise.
701         (cortex_a15_neon_vmov): Likewise.
702         (cortex_a15_neon_vaba): Likewise.
703         (cortex_a15_neon_vaba_qqq): Likewise.
704         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
705         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
706         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
707         Likewise.
708         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
709         (cortex_a15_neon_mla_qqq_8_16): Likewise.
710         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
711         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
712         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
713         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
714         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
715         (cortex_a15_neon_shift_1): Likewise.
716         (cortex_a15_neon_shift_2): Likewise.
717         (cortex_a15_neon_shift_3): Likewise.
718         (cortex_a15_neon_vshl_ddd): Likewise.
719         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
720         (cortex_a15_neon_vsra_vrsra): Likewise.
721         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
722         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
723         (cortex_a15_neon_bp_3cycle): Likewise.
724         (cortex_a15_neon_ldm_2): Likewise.
725         (cortex_a15_neon_stm_2): Likewise.
726         (cortex_a15_neon_mcr): Likewise.
727         (cortex_a15_neon_mrc): Likewise.
728         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
729         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
730         (cortex_a15_neon_fp_vmul_ddd): Likewise.
731         (cortex_a15_neon_fp_vmul_qqd): Likewise.
732         (cortex_a15_neon_fp_vmla_ddd): Likewise.
733         (cortex_a15_neon_fp_vmla_qqq): Likewise.
734         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
735         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
736         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
737         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
738         (cortex_a15_neon_bp_simple): Likewise.
739         (cortex_a15_neon_bp_2cycle): Likewise.
740         (cortex_a15_neon_bp_3cycle): Likewise.
741         (cortex_a15_neon_vld1_1_2_regs): Likewise.
742         (cortex_a15_neon_vld1_3_4_regs): Likewise.
743         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
744         (cortex_a15_neon_vld2_4_regs): Likewise.
745         (cortex_a15_neon_vld3_vld4): Likewise.
746         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
747         (cortex_a15_neon_vst1_3_4_regs): Likewise.
748         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
749         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
750         (cortex_a15_neon_vst3_vst4): Rename to...
751         (cortex_a15_neon_vst4): This, update for new attributes.
752         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
753         (cortex_a15_neon_vld3_vld4_lane): Likewise.
754         (cortex_a15_neon_vst1_vst2_lane): Likewise.
755         (cortex_a15_neon_vst3_vst4_lane): Likewise.
756         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
757         (cortex_a15_neon_ldm_2): Likewise.
758         (cortex_a15_neon_stm_2): Likewise.
759         (cortex_a15_neon_mcr): Likewise.
760         (cortex_a15_neon_mcr_2_mcrr): Likewise.
761         (cortex_a15_neon_mrc): Likewise.
762         (cortex_a15_neon_mrrc): Likewise.
764         (cortex_a15_neon_abd): New.
765         (cortex_a15_neon_abd_q): Likewise.
766         (cortex_a15_neon_aba): Likewise.
767         (cortex_a15_neon_aba_q): Likewise.
768         (cortex_a15_neon_acc): Likewise.
769         (cortex_a15_neon_acc_q): Likewise.
770         (cortex_a15_neon_arith_basic): Likewise.
771         (cortex_a15_neon_arith_complex): Likewise.
772         (cortex_a15_neon_multiply): Likewise.
773         (cortex_a15_neon_multiply_q): Likewise.
774         (cortex_a15_neon_mla): Likewise.
775         (cortex_a15_neon_mla_q): Likewise.
776         (cortex_a15_neon_sat_mla_long): Likewise.
777         (cortex_a15_neon_shift_acc): Likewise.
778         (cortex_a15_neon_shift_imm_basic): Likewise.
779         (cortex_a15_neon_shift_imm_complex): Likewise.
780         (cortex_a15_neon_shift_reg_basic): Likewise.
781         (cortex_a15_neon_shift_reg_basic_q): Likewise.
782         (cortex_a15_neon_shift_reg_complex): Likewise.
783         (cortex_a15_neon_shift_reg_complex_q): Likewise.
784         (cortex_a15_neon_fp_negabs): Likewise
785         (cortex_a15_neon_fp_arith): Likewise
786         (cortex_a15_neon_fp_arith_q): Likewise
787         (cortex_a15_neon_fp_cvt_int): Likewise
788         (cortex_a15_neon_fp_cvt_int_q): Likewise
789         (cortex_a15_neon_fp_cvt_16): Likewise
790         (cortex_a15_neon_fp_mul): Likewise
791         (cortex_a15_neon_fp_mul_q): Likewise
792         (cortex_a15_neon_fp_mla): Likewise
793         (cortex_a15_neon_fp_mla_q): Likewise
794         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
795         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
796         (cortex_a15_neon_bitops): Likewise.
797         (cortex_a15_neon_bitops_q): Likewise.
798         (cortex_a15_neon_from_gp): Likewise.
799         (cortex_a15_neon_from_gp_q): Likewise.
800         (cortex_a15_neon_tbl3_tbl4): Likewise.
801         (cortex_a15_neon_zip_q): Likewise.
802         (cortex_a15_neon_to_gp): Likewise.
803         (cortex_a15_neon_load_a): Likewise.
804         (cortex_a15_neon_load_b): Likewise.
805         (cortex_a15_neon_load_c): Likewise.
806         (cortex_a15_neon_load_d): Likewise.
807         (cortex_a15_neon_load_e): Likewise.
808         (cortex_a15_neon_load_f): Likewise.
809         (cortex_a15_neon_store_a): Likewise.
810         (cortex_a15_neon_store_b): Likewise.
811         (cortex_a15_neon_store_c): Likewise.
812         (cortex_a15_neon_store_d): Likewise.
813         (cortex_a15_neon_store_e): Likewise.
814         (cortex_a15_neon_store_f): Likewise.
815         (cortex_a15_neon_store_g): Likewise.
816         (cortex_a15_neon_store_h): Likewise.
817         (cortex_a15_vfp_to_from_gp): Likewise.
819 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
821         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
823         (cortex_a9_neon_vshl_ddd): Remove.
824         (cortex_a9_neon_vst3_vst4): Likewise.
825         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
827         (cortex_a9_neon_bit_ops_q): New.
829         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
830         (cortex_a9_neon_int_2): Likewise.
831         (cortex_a9_neon_int_3): Likewise.
832         (cortex_a9_neon_int_4): Likewise.
833         (cortex_a9_neon_int_5): Likewise.
834         (cortex_a9_neon_vqneg_vqabs): Likewise.
835         (cortex_a9_neon_vmov): Likewise.
836         (cortex_a9_neon_vaba): Likewise.
837         (cortex_a9_neon_vaba_qqq): Likewise.
838         (cortex_a9_neon_shift_1): Likewise.
839         (cortex_a9_neon_shift_2): Likewise.
840         (cortex_a9_neon_shift_3): Likewise.
841         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
842         (cortex_a9_neon_vsra_vrsra): Likewise.
843         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
844         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
845         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
846         Likewise.
847         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
848         (cortex_a9_neon_mla_qqq_8_16): Likewise.
849         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
850         Likewise.
851         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
852         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
853         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
854         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
855         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
856         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
857         (cortex_a9_neon_fp_vsum): Likewise.
858         (cortex_a9_neon_fp_vmul_ddd): Likewise.
859         (cortex_a9_neon_fp_vmul_qqd): Likewise.
860         (cortex_a9_neon_fp_vmla_ddd): Likewise.
861         (cortex_a9_neon_fp_vmla_qqq): Likewise.
862         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
863         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
864         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
865         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
866         (cortex_a9_neon_bp_simple): Likewise.
867         (cortex_a9_neon_bp_2cycle): Likewise.
868         (cortex_a9_neon_bp_3cycle): Likewise.
869         (cortex_a9_neon_ldr): Likewise.
870         (cortex_a9_neon_str): Likewise.
871         (cortex_a9_neon_vld1_1_2_regs): Likewise.
872         (cortex_a9_neon_vld1_3_4_regs): Likewise.
873         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
874         (cortex_a9_neon_vld2_4_regs): Likewise.
875         (cortex_a9_neon_vld3_vld4): Likewise.
876         (cortex_a9_neon_vld1_vld2_lane): Likewise.
877         (cortex_a9_neon_vld3_vld4_lane): Likewise.
878         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
879         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
880         (cortex_a9_neon_vst1_3_4_regs): Likewise.
881         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
882         (cortex_a9_neon_vst1_vst2_lane): Likewise.
883         (cortex_a9_neon_vst3_vst4_lane): Likewise.
884         (cortex_a9_neon_mcr): Likewise.
885         (cortex_a9_neon_mcr_2_mcrr): Likewise.
886         (cortex_a9_neon_mrc): Likewise.
887         (cortex_a9_neon_mrrc): Likewise.
889 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
891         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
893         (cortex_a8_neon_vshl_ddd): Remove.
894         (cortex_a8_neon_vst3_vst4): Likewise.
895         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
897         (cortex_a8_neon_bit_ops_q): New.
899         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
900         (cortex_a8_neon_int_2): Likewise..
901         (cortex_a8_neon_int_3): Likewise.
902         (cortex_a8_neon_int_5): Likewise.
903         (cortex_a8_neon_vqneg_vqabs): Likewise.
904         (cortex_a8_neon_int_4): Likewise.
905         (cortex_a8_neon_vaba): Likewise.
906         (cortex_a8_neon_vaba_qqq): Likewise.
907         (cortex_a8_neon_shift_1): Likewise.
908         (cortex_a8_neon_shift_2): Likewise.
909         (cortex_a8_neon_shift_3): Likewise.
910         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
911         (cortex_a8_neon_vsra_vrsra): Likewise.
912         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
913         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
914         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
915         Likewise.
916         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
917         (cortex_a8_neon_mla_qqq_8_16): Likewise.
918         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
919         Likewise.
920         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
921         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
922         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
923         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
924         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
925         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
926         (cortex_a8_neon_fp_vsum): Likewise.
927         (cortex_a8_neon_fp_vmul_ddd): Likewise.
928         (cortex_a8_neon_fp_vmul_qqd): Likewise.
929         (cortex_a8_neon_fp_vmla_ddd): Likewise.
930         (cortex_a8_neon_fp_vmla_qqq): Likewise.
931         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
932         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
933         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
934         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
935         (cortex_a8_neon_bp_simple): Likewise.
936         (cortex_a8_neon_bp_2cycle): Likewise.
937         (cortex_a8_neon_bp_3cycle): Likewise.
938         (cortex_a8_neon_ldr): Likewise.
939         (cortex_a8_neon_str): Likewise.
940         (cortex_a8_neon_vld1_1_2_regs): Likewise.
941         (cortex_a8_neon_vld1_3_4_regs): Likewise.
942         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
943         (cortex_a8_neon_vld2_4_regs): Likewise.
944         (cortex_a8_neon_vld3_vld4): Likewise.
945         (cortex_a8_neon_vld1_vld2_lane): Likewise.
946         (cortex_a8_neon_vld3_vld4_lane): Likewise.
947         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
948         (cortex_a8_neon_vst1_3_4_regs): Likewise.
949         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
950         (cortex_a8_neon_vst1_vst2_lane): Likewise.
951         (cortex_a8_neon_vst3_vst4_lane): Likewise.
952         (cortex_a8_neon_mcr): Likewise.
953         (cortex_a8_neon_mcr_2_mcrr): Likewise.
954         (cortex_a8_neon_mrc): Likewise.
955         (cortex_a8_neon_mrrc): Likewise.
957 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
959         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
960         (fp): New.
961         * config/aarch64/aarch64-simd.md (neon_type): Remove.
962         (aarch64_simd_dup<mode>): Add "type" attribute.
963         (aarch64_dup_lane<mode>): Likewise.
964         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
965         (*aarch64_simd_mov<mode>): Likewise.
966         (aarch64_simd_mov_from_<mode>low): Likewise.
967         (aarch64_simd_mov_from_<mode>high): Likewise.
968         (orn<mode>3): Likewise.
969         (bic<mode>3): Likewise.
970         (add<mode>3): Likewise.
971         (sub<mode>3): Likewise.
972         (mul<mode>3): Likewise.
973         (*aarch64_mul3_elt<mode>): Likewise.
974         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
975         (*aarch64_mul3_elt_to_128df): Likewise.
976         (*aarch64_mul3_elt_to_64v2df): Likewise.
977         (neg<mode>2): Likewise.
978         (abs<mode>2): Likewise.
979         (abd<mode>_3): Likewise.
980         (aba<mode>_3): Likewise.
981         (fabd<mode>_3): Likewise.
982         (*fabd_scalar<mode>3): Likewise.
983         (and<mode>3): Likewise.
984         (ior<mode>3): Likewise.
985         (xor<mode>3): Likewise.
986         (one_cmpl<mode>2): Likewise.
987         (aarch64_simd_vec_set<mode>): Likewise.
988         (aarch64_simd_lshr<mode>): Likewise.
989         (aarch64_simd_ashr<mode>): Likewise.
990         (aarch64_simd_imm_shl<mode>): Likewise.
991         (aarch64_simd_reg_sshl<mode): Likewise.
992         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
993         (aarch64_simd_reg_shl<mode>_signed): Likewise.
994         (aarch64_simd_vec_setv2di): Likewise.
995         (aarch64_simd_vec_set<mode>): Likewise.
996         (aarch64_mla<mode>): Likewise.
997         (*aarch64_mla_elt<mode>): Likewise.
998         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
999         (aarch64_mls<mode>): Likewise.
1000         (*aarch64_mls_elt<mode>): Likewise.
1001         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1002         (<su><maxmin><mode>3): Likewise.
1003         (move_lo_quad_<mode>): Likewise.
1004         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1005         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1006         (vec_pack_trunc_<mode>): Likewise.
1007         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1008         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1009         (*aarch64_<su>mlal_lo<mode>): Likewise.
1010         (*aarch64_<su>mlal_hi<mode>): Likewise.
1011         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1012         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1013         (*aarch64_<su>mlal<mode>): Likewise.
1014         (*aarch64_<su>mlsl<mode>): Likewise.
1015         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1016         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1017         (add<mode>3): Likewise.
1018         (sub<mode>3): Likewise.
1019         (mul<mode>3): Likewise.
1020         (div<mode>3): Likewise.
1021         (neg<mode>2): Likewise.
1022         (abs<mode>2): Likewise.
1023         (fma<mode>4): Likewise.
1024         (*aarch64_fma4_elt<mode>): Likewise.
1025         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1026         (*aarch64_fma4_elt_to_128df): Likewise.
1027         (*aarch64_fma4_elt_to_64v2df): Likewise.
1028         (fnma<mode>4): Likewise.
1029         (*aarch64_fnma4_elt<mode>): Likewise.
1030         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1031         (*aarch64_fnma4_elt_to_128df): Likewise.
1032         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1033         (<frint_pattern><mode>2): Likewise.
1034         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1035         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1036         (vec_unpacks_lo_v4sf): Likewise.
1037         (aarch64_float_extend_lo_v2df): Likewise.
1038         (vec_unpacks_hi_v4sf): Likewise.
1039         (aarch64_float_truncate_lo_v2sf): Likewise.
1040         (aarch64_float_truncate_hi_v4sf): Likewise.
1041         (aarch64_vmls<mode>): Likewise.
1042         (<su><maxmin><mode>3): Likewise.
1043         (<maxmin_uns><mode>3): Likewise.
1044         (reduc_<sur>plus_<mode>): Likewise.
1045         (reduc_<sur>plus_v2di): Likewise.
1046         (reduc_<sur>plus_v2si): Likewise.
1047         (reduc_<sur>plus_<mode>): Likewise.
1048         (aarch64_addpv4sf): Likewise.
1049         (clz<mode>2): Likewise.
1050         (reduc_<maxmin_uns>_<mode>): Likewise.
1051         (reduc_<maxmin_uns>_v2di): Likewise.
1052         (reduc_<maxmin_uns>_v2si): Likewise.
1053         (reduc_<maxmin_uns>_<mode>): Likewise.
1054         (reduc_<maxmin_uns>_v4sf): Likewise.
1055         (aarch64_simd_bsl<mode>_internal): Likewise.
1056         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1057         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1058         (aarch64_get_lane<mode>): Likewise.
1059         (*aarch64_combinez<mode>): Likewise.
1060         (aarch64_combine<mode>): Likewise.
1061         (aarch64_simd_combine<mode>): Likewise.
1062         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1063         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1064         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1065         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1066         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1067         (aarch64_<sur>h<addsub><mode>): Likewise.
1068         (aarch64_<sur><addsub>hn<mode>): Likewise.
1069         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1070         (aarch64_pmul<mode>): Likewise.
1071         (aarch64_<su_optab><optab><mode>): Likewise.
1072         (aarch64_<sur>qadd<mode>): Likewise.
1073         (aarch64_sqmovun<mode>): Likewise.
1074         (aarch64_<sur>qmovn<mode>): Likewise.
1075         (aarch64_s<optab><mode>): Likewise.
1076         (aarch64_sq<r>dmulh<mode>): Likewise.
1077         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1078         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1079         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1080         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1081         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1082         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1083         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1084         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1085         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1086         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1087         (aarch64_sqdmull<mode>): Likewise.
1088         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1089         (aarch64_sqdmull_n<mode>): Likewise.
1090         (aarch64_sqdmull2<mode>_internal): Likewise.
1091         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1092         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1093         (aarch64_<sur>shl<mode>): Likewise.
1094         (aarch64_<sur>q<r>shl<mode>
1095         (aarch64_<sur>shll_n<mode>): Likewise.
1096         (aarch64_<sur>shll2_n<mode>): Likewise.
1097         (aarch64_<sur>shr_n<mode>): Likewise.
1098         (aarch64_<sur>sra_n<mode>): Likewise.
1099         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1100         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1101         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1102         (aarch64_cm<optab><mode>): Likewise.
1103         (aarch64_cm<optab>di): Likewise.
1104         (aarch64_cm<optab><mode>): Likewise.
1105         (aarch64_cm<optab>di): Likewise.
1106         (aarch64_cmtst<mode>): Likewise.
1107         (aarch64_cmtstdi): Likewise.
1108         (aarch64_cm<optab><mode>): Likewise.
1109         (*aarch64_fac<optab><mode>): Likewise.
1110         (aarch64_addp<mode>): Likewise.
1111         (aarch64_addpdi): Likewise.
1112         (sqrt<mode>2): Likewise.
1113         (vec_load_lanesoi<mode>): Likewise.
1114         (vec_store_lanesoi<mode>): Likewise.
1115         (vec_load_lanesci<mode>): Likewise.
1116         (vec_store_lanesci<mode>): Likewise.
1117         (vec_load_lanesxi<mode>): Likewise.
1118         (vec_store_lanesxi<mode>): Likewise.
1119         (*aarch64_mov<mode>): Likewise.
1120         (aarch64_ld2<mode>_dreg): Likewise.
1121         (aarch64_ld2<mode>_dreg): Likewise.
1122         (aarch64_ld3<mode>_dreg): Likewise.
1123         (aarch64_ld3<mode>_dreg): Likewise.
1124         (aarch64_ld4<mode>_dreg): Likewise.
1125         (aarch64_ld4<mode>_dreg): Likewise.
1126         (aarch64_tbl1<mode>): Likewise.
1127         (aarch64_tbl2v16qi): Likewise.
1128         (aarch64_combinev16qi): Likewise.
1129         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1130         (aarch64_st2<mode>_dreg): Likewise.
1131         (aarch64_st2<mode>_dreg): Likewise.
1132         (aarch64_st3<mode>_dreg): Likewise.
1133         (aarch64_st3<mode>_dreg): Likewise.
1134         (aarch64_st4<mode>_dreg): Likewise.
1135         (aarch64_st4<mode>_dreg): Likewise.
1136         (*aarch64_simd_ld1r<mode>): Likewise.
1137         (aarch64_frecpe<mode>): Likewise.
1138         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1139         (aarch64_frecps<mode>): Likewise.
1141 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1143         * config/arm/iterators.md (V_elem_ch): New.
1144         (q): Likewise.
1145         (VQH_type): Likewise.
1146         * config/arm/arm.md (is_neon_type): New.
1147         (conds): Use is_neon_type.
1148         (anddi3_insn): Update type attribute.
1149         (xordi3_insn): Likewise.
1150         (one_cmpldi2): Likewise.
1151         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1152         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1153         (*movmisalign<mode>_neon_store): Likewise.
1154         (*movmisalign<mode>_neon_load): Likewise.
1155         (vec_set<mode>_internal): Likewise.
1156         (vec_set<mode>_internal): Likewise.
1157         (vec_setv2di_internal): Likewise.
1158         (vec_extract<mode>): Likewise.
1159         (vec_extract<mode>): Likewise.
1160         (vec_extractv2di): Likewise.
1161         (*add<mode>3_neon): Likewise.
1162         (adddi3_neon): Likewise.
1163         (*sub<mode>3_neon): Likewise.
1164         (subdi3_neon): Likewise.
1165         (fma<VCVTF:mode>4): Likewise.
1166         (fma<VCVTF:mode>4_intrinsic): Likewise.
1167         (*fmsub<VCVTF:mode>4): Likewise.
1168         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1169         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1170         (ior<mode>3): Likewise.
1171         (and<mode>3): Likewise.
1172         (orn<mode>3_neon): Likewise.
1173         (orndi3_neon): Likewise.
1174         (bic<mode>3_neon): Likewise.
1175         (bicdi3_neon): Likewise.
1176         (xor<mode>3): Likewise.
1177         (one_cmpl<mode>2): Likewise.
1178         (abs<mode>2): Likewise.
1179         (neg<mode>2): Likewise.
1180         (negdi2_neon): Likewise.
1181         (*umin<mode>3_neon): Likewise.
1182         (*umax<mode>3_neon): Likewise.
1183         (*smin<mode>3_neon): Likewise.
1184         (*smax<mode>3_neon): Likewise.
1185         (vashl<mode>3): Likewise.
1186         (vashr<mode>3_imm): Likewise.
1187         (vlshr<mode>3_imm): Likewise.
1188         (ashl<mode>3_signed): Likewise.
1189         (ashl<mode>3_unsigned): Likewise.
1190         (neon_load_count): Likewise.
1191         (ashldi3_neon_noclobber): Likewise.
1192         (ashldi3_neon): Likewise.
1193         (signed_shift_di3_neon): Likewise.
1194         (unsigned_shift_di3_neon): Likewise.
1195         (ashrdi3_neon_imm_noclobber): Likewise.
1196         (lshrdi3_neon_imm_noclobber): Likewise.
1197         (<shift>di3_neon): Likewise.
1198         (widen_ssum<mode>3): Likewise.
1199         (widen_usum<mode>3): Likewise.
1200         (quad_halves_<code>v4si): Likewise.
1201         (quad_halves_<code>v4sf): Likewise.
1202         (quad_halves_<code>v8hi): Likewise.
1203         (quad_halves_<code>v16qi): Likewise.
1204         (reduc_splus_v2di): Likewise.
1205         (neon_vpadd_internal<mode>): Likewise.
1206         (neon_vpsmin<mode>): Likewise.
1207         (neon_vpsmax<mode>): Likewise.
1208         (neon_vpumin<mode>): Likewise.
1209         (neon_vpumax<mode>): Likewise.
1210         (*ss_add<mode>_neon): Likewise.
1211         (*us_add<mode>_neon): Likewise.
1212         (*ss_sub<mode>_neon): Likewise.
1213         (*us_sub<mode>_neon): Likewise.
1214         (neon_vadd<mode>_unspec): Likewise.
1215         (neon_vaddl<mode>): Likewise.
1216         (neon_vaddw<mode>): Likewise.
1217         (neon_vhadd<mode>): Likewise.
1218         (neon_vqadd<mode>): Likewise.
1219         (neon_vaddhn<mode>): Likewise.
1220         (neon_vmul<mode>): Likewise.
1221         (neon_vfms<VCVTF:mode>): Likewise.
1222         (neon_vmlal<mode>): Likewise.
1223         (neon_vmls<mode>): Likewise.
1224         (neon_vmlsl<mode>): Likewise.
1225         (neon_vqdmulh<mode>): Likewise.
1226         (neon_vqdmlal<mode>): Likewise.
1227         (neon_vqdmlsl<mode>): Likewise.
1228         (neon_vmull<mode>): Likewise.
1229         (neon_vqdmull<mode>): Likewise.
1230         (neon_vsub<mode>_unspec): Likewise.
1231         (neon_vsubl<mode>): Likewise.
1232         (neon_vsubw<mode>): Likewise.
1233         (neon_vqsub<mode>): Likewise.
1234         (neon_vhsub<mode>): Likewise.
1235         (neon_vsubhn<mode>): Likewise.
1236         (neon_vceq<mode>): Likewise.
1237         (neon_vcge<mode>): Likewise.
1238         (neon_vcgeu<mode>): Likewise.
1239         (neon_vcgt<mode>): Likewise.
1240         (neon_vcgtu<mode>): Likewise.
1241         (neon_vcle<mode>): Likewise.
1242         (neon_vclt<mode>): Likewise.
1243         (neon_vcage<mode>): Likewise.
1244         (neon_vcagt<mode>): Likewise.
1245         (neon_vtst<mode>): Likewise.
1246         (neon_vabd<mode>): Likewise.
1247         (neon_vabdl<mode>): Likewise.
1248         (neon_vaba<mode>): Likewise.
1249         (neon_vabal<mode>): Likewise.
1250         (neon_vmax<mode>): Likewise.
1251         (neon_vmin<mode>): Likewise.
1252         (neon_vpaddl<mode>): Likewise.
1253         (neon_vpadal<mode>): Likewise.
1254         (neon_vpmax<mode>): Likewise.
1255         (neon_vpmin<mode>): Likewise.
1256         (neon_vrecps<mode>): Likewise.
1257         (neon_vrsqrts<mode>): Likewise.
1258         (neon_vqabs<mode>): Likewise.
1259         (neon_vqneg<mode>): Likewise.
1260         (neon_vcls<mode>): Likewise.
1261         (clz<mode>2): Likewise.
1262         (popcount<mode>2): Likewise.
1263         (neon_vrecpe<mode>): Likewise.
1264         (neon_vrsqrte<mode>): Likewise.
1265         (neon_vget_lane<mode>_sext_internal): Likewise.
1266         (neon_vget_lane<mode>_zext_internal): Likewise.
1267         (neon_vdup_n<mode>): Likewise.
1268         (neon_vdup_n<mode>): Likewise.
1269         (neon_vdup_nv2di): Likewise.
1270         (neon_vdup_lane<mode>_interal): Likewise.
1271         (*neon_vswp<mode>): Likewise.
1272         (neon_vcombine<mode>): Likewise.
1273         (float<mode><V_cvtto>2): Likewise.
1274         (floatuns<mode><V_cvtto>2): Likewise.
1275         (fix_trunc<mode><V_cvtto>2): Likewise.
1276         (fixuns_trunc<mode><V_cvtto>2
1277         (neon_vcvt<mode>): Likewise.
1278         (neon_vcvt<mode>): Likewise.
1279         (neon_vcvtv4sfv4hf): Likewise.
1280         (neon_vcvtv4hfv4sf): Likewise.
1281         (neon_vcvt_n<mode>): Likewise.
1282         (neon_vcvt_n<mode>): Likewise.
1283         (neon_vmovn<mode>): Likewise.
1284         (neon_vqmovn<mode>): Likewise.
1285         (neon_vqmovun<mode>): Likewise.
1286         (neon_vmovl<mode>): Likewise.
1287         (neon_vmul_lane<mode>): Likewise.
1288         (neon_vmul_lane<mode>): Likewise.
1289         (neon_vmull_lane<mode>): Likewise.
1290         (neon_vqdmull_lane<mode>): Likewise.
1291         (neon_vqdmulh_lane<mode>): Likewise.
1292         (neon_vqdmulh_lane<mode>): Likewise.
1293         (neon_vmla_lane<mode>): Likewise.
1294         (neon_vmla_lane<mode>): Likewise.
1295         (neon_vmlal_lane<mode>): Likewise.
1296         (neon_vqdmlal_lane<mode>): Likewise.
1297         (neon_vmls_lane<mode>): Likewise.
1298         (neon_vmls_lane<mode>): Likewise.
1299         (neon_vmlsl_lane<mode>): Likewise.
1300         (neon_vqdmlsl_lane<mode>): Likewise.
1301         (neon_vext<mode>): Likewise.
1302         (neon_vrev64<mode>): Likewise.
1303         (neon_vrev32<mode>): Likewise.
1304         (neon_vrev16<mode>): Likewise.
1305         (neon_vbsl<mode>_internal): Likewise.
1306         (neon_vshl<mode>): Likewise.
1307         (neon_vqshl<mode>): Likewise.
1308         (neon_vshr_n<mode>): Likewise.
1309         (neon_vshrn_n<mode>): Likewise.
1310         (neon_vqshrn_n<mode>): Likewise.
1311         (neon_vqshrun_n<mode>): Likewise.
1312         (neon_vshl_n<mode>): Likewise.
1313         (neon_vqshl_n<mode>): Likewise.
1314         (neon_vqshlu_n<mode>): Likewise.
1315         (neon_vshll_n<mode>): Likewise.
1316         (neon_vsra_n<mode>): Likewise.
1317         (neon_vsri_n<mode>): Likewise.
1318         (neon_vsli_n<mode>): Likewise.
1319         (neon_vtbl1v8qi): Likewise.
1320         (neon_vtbl2v8qi): Likewise.
1321         (neon_vtbl3v8qi): Likewise.
1322         (neon_vtbl4v8qi): Likewise.
1323         (neon_vtbl1v16qi): Likewise.
1324         (neon_vtbl2v16qi): Likewise.
1325         (neon_vcombinev16qi): Likewise.
1326         (neon_vtbx1v8qi): Likewise.
1327         (neon_vtbx2v8qi): Likewise.
1328         (neon_vtbx3v8qi): Likewise.
1329         (neon_vtbx4v8qi): Likewise.
1330         (*neon_vtrn<mode>_insn): Likewise.
1331         (*neon_vzip<mode>_insn): Likewise.
1332         (*neon_vuzp<mode>_insn): Likewise.
1333         (neon_vld1<mode>): Likewise.
1334         (neon_vld1_lane<mode>): Likewise.
1335         (neon_vld1_lane<mode>): Likewise.
1336         (neon_vld1_dup<mode>): Likewise.
1337         (neon_vld1_dup<mode>): Likewise.
1338         (neon_vld1_dupv2di): Likewise.
1339         (neon_vst1<mode>): Likewise.
1340         (neon_vst1_lane<mode>): Likewise.
1341         (neon_vst1_lane<mode>): Likewise.
1342         (neon_vld2<mode>): Likewise.
1343         (neon_vld2<mode>): Likewise.
1344         (neon_vld2_lane<mode>): Likewise.
1345         (neon_vld2_lane<mode>): Likewise.
1346         (neon_vld2_dup<mode>): Likewise.
1347         (neon_vst2<mode>): Likewise.
1348         (neon_vst2<mode>): Likewise.
1349         (neon_vst2_lane<mode>): Likewise.
1350         (neon_vst2_lane<mode>): Likewise.
1351         (neon_vld3<mode>): Likewise.
1352         (neon_vld3qa<mode>): Likewise.
1353         (neon_vld3qb<mode>): Likewise.
1354         (neon_vld3_lane<mode>): Likewise.
1355         (neon_vld3_lane<mode>): Likewise.
1356         (neon_vld3_dup<mode>): Likewise.
1357         (neon_vst3<mode>): Likewise.
1358         (neon_vst3qa<mode>): Likewise.
1359         (neon_vst3qb<mode>): Likewise.
1360         (neon_vst3_lane<mode>): Likewise.
1361         (neon_vst3_lane<mode>): Likewise.
1362         (neon_vld4<mode>): Likewise.
1363         (neon_vld4qa<mode>): Likewise.
1364         (neon_vld4qb<mode>): Likewise.
1365         (neon_vld4_lane<mode>): Likewise.
1366         (neon_vld4_lane<mode>): Likewise.
1367         (neon_vld4_dup<mode>): Likewise.
1368         (neon_vst4<mode>): Likewise.
1369         (neon_vst4qa<mode>): Likewise.
1370         (neon_vst4qb<mode>): Likewise.
1371         (neon_vst4_lane<mode>): Likewise.
1372         (neon_vst4_lane<mode>): Likewise.
1373         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1374         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1375         (neon_vec_<US>mult_lo_<mode>): Likewise.
1376         (neon_vec_<US>mult_hi_<mode>): Likewise.
1377         (neon_vec_<US>shiftl_<mode>): Likewise.
1378         (neon_unpack<US>_<mode>): Likewise.
1379         (neon_vec_<US>mult_<mode>): Likewise.
1380         (vec_pack_trunc_<mode>): Likewise.
1381         (neon_vec_pack_trunc_<mode>): Likewise.
1382         (neon_vabd<mode>_2): Likewise.
1383         (neon_vabd<mode>_3): Likewise.
1385 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1387         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1388         (load_pair): Update type attribute.
1389         (store_pair): Update type attribute.
1390         * config/aarch64/iterators.md (q): New.
1392 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1394         * config/arm/types.md: Add new types for Neon insns.
1396 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1397             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1398             Sergey Lega  <sergey.s.lega@intel.com>
1399             Anna Tikhonova  <anna.tikhonova@intel.com>
1400             Ilya Tocar  <ilya.tocar@intel.com>
1401             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1402             Ilya Verbin  <ilya.verbin@intel.com>
1403             Kirill Yukhin  <kirill.yukhin@intel.com>
1404             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1406         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1407         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1408         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1409         (rcp14<mode>): New.
1410         (srcp14<mode>): Ditto.
1411         (rsqrt14<mode>): Ditto.
1412         (rsqrt14<mode>): Ditto.
1413         (avx512f_vmscalef<mode>): Ditto.
1414         (avx512f_scalef<mode>): Ditto.
1415         (avx512f_getexp<mode>): Ditto.
1416         (avx512f_sgetexp<mode>): Ditto.
1417         (avx512f_fixupimm<mode>): Ditto.
1418         (avx512f_sfixupimm<mode>): Ditto.
1419         (avx512f_rndscale<mode>): Ditto.
1420         (*avx512er_exp2<mode>): Ditto.
1421         (*avx512er_rcp28<mode>): Ditto.
1422         (avx512er_rsqrt28<mode>): Ditto.
1423         (avx512f_getmant<mode>): Ditto.
1424         (avx512f_getmant<mode>): Ditto.
1425         (avx512f_rndscale<mode>): Fix formatting.
1427 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1429         * ipa-utils.h (ipa_edge_within_scc): Declare.
1430         * ipa-cp.c (edge_within_scc): Moved...
1431         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1433 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1434             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1435             Sergey Lega  <sergey.s.lega@intel.com>
1436             Anna Tikhonova  <anna.tikhonova@intel.com>
1437             Ilya Tocar  <ilya.tocar@intel.com>
1438             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1439             Ilya Verbin  <ilya.verbin@intel.com>
1440             Kirill Yukhin  <kirill.yukhin@intel.com>
1441             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1443         * config/i386/predicates.md (const_8_to_15_operand): New.
1444         (const_16_to_31_operand): Ditto.
1445         * config/i386/sse.md (V8FI): New.
1446         (V16FI): Ditto.
1447         (reduc_splus_v8df): Ditto.
1448         (reduc_splus_v16sf): Ditto.
1449         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1450         (vec_extract_hi_<mode>): Ditto.
1451         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1452         (vec_set_lo_<mode>): Ditto.
1453         (vec_set_hi_<mode>): Ditto.
1454         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1455         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1456         (avx512f_pshufd_1): Ditto.
1457         (avx512f_broadcast<mode>): Ditto.
1458         (avx512f_broadcast<mode>): Ditto.
1459         (define_split): Split vec_extract_lo into move.
1460         (ssequartermode): Ditto.
1461         (ssedoublemode): Extened with wider modes.
1462         (vec_extract_lo_<mode>): Ditto.
1464 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1465             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1466             Sergey Lega  <sergey.s.lega@intel.com>
1467             Anna Tikhonova  <anna.tikhonova@intel.com>
1468             Ilya Tocar  <ilya.tocar@intel.com>
1469             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1470             Ilya Verbin  <ilya.verbin@intel.com>
1471             Kirill Yukhin  <kirill.yukhin@intel.com>
1472             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1474         * config/i386/predicates.md (register_or_constm1_operand): New.
1475         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1476         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1477         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1478         UNSPEC_SCATTER_PREFETCH
1479         (VI48_512): New.
1480         (avx512f_ucmp<mode>3): Ditto.
1481         (avx512f_vternlog<mode>): Ditto.
1482         (avx512f_align<mode>): Ditto.
1483         (<shift_insn><mode>3): Ditto.
1484         (avx512f_<rotate>v<mode>): Ditto.
1485         (avx512f_<rotate><mode>): Ditto.
1486         (avx512f_eq<mode>3): Ditto.
1487         (avx512f_eq<mode>3_1): Ditto.
1488         (avx512f_gt<mode>3): Ditto.
1489         (avx512f_testm<mode>3): Ditto.
1490         (avx512f_testnm<mode>3): Ditto.
1491         (avx512pf_gatherpf<mode>): Ditto.
1492         (*avx512pf_gatherpf<mode>_mask): Ditto.
1493         (*avx512pf_gatherpf<mode>): Ditto.
1494         (avx512pf_scatterpf<mode>): Ditto.
1495         (*avx512pf_scatterpf<mode>_mask): Ditto.
1496         (*avx512pf_scatterpf<mode>): Ditto.
1497         (avx512f_vec_dup_gpr<mode>): Ditto.
1498         (clz<mode>2): Ditto.
1499         (conflict<mode>): Ditto.
1500         (REDUC_SMINMAX_MODE): Extened with wider modes.
1501         (reduc_<code>_<mode>): Ditto.
1502         (vlshr<mode>3): Ditto.
1503         (vashl<mode>3): Ditto.
1505 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1506             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1507             Sergey Lega  <sergey.s.lega@intel.com>
1508             Anna Tikhonova  <anna.tikhonova@intel.com>
1509             Ilya Tocar  <ilya.tocar@intel.com>
1510             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1511             Ilya Verbin  <ilya.verbin@intel.com>
1512             Kirill Yukhin  <kirill.yukhin@intel.com>
1513             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1515         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1516         UNSPEC_SCATTER.
1517         (VI48F_512): New.
1518         (avx512fmaskmode): Ditto.
1519         (bcstscalarsuff): Ditto.
1520         (avx512f_blendm<mode>): Ditto.
1521         (cmp_imm_predicate): Ditto.
1522         (avx512f_cmp<mode>3): Ditto.
1523         (avx512f_vec_dup<mode>): Ditto.
1524         (avx512f_vec_dup_mem<mode>): Ditto.
1525         (avx512f_vpermi2var<mode>3): Ditto.
1526         (avx512f_vpermt2var<mode>3): Ditto.
1527         (vec_init<mode>): Ditto.
1528         (avx512f_gathersi<mode>): Ditto.
1529         (*avx512f_gathersi<mode>): Ditto.
1530         (*avx512f_gathersi<mode>_2): Ditto.
1531         (avx512f_gatherdi<mode>): Ditto.
1532         (*avx512f_gatherdi<mode>): Ditto.
1533         (*avx512f_gatherdi<mode>_2): Ditto.
1534         (avx512f_scattersi<mode>): Ditto.
1535         (*avx512f_scattersi<mode>): Ditto.
1536         (avx512f_scatterdi<mode>): Ditto.
1537         (*avx512f_scatterdi<mode>): Ditto.
1538         (sseintprefix): Extened with wider modes.
1539         (VEC_GATHER_IDXSI): Ditto.
1540         (VEC_GATHER_IDXDI): Ditto.
1541         (VEC_GATHER_SRCDI): Ditto.
1543 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1544             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1546         * config/arm/t-aprofile: New file.
1547         * config.gcc: Handle --with-multilib-list option.
1549 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1551         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1552         the input for r2.
1554 2013-10-15  Richard Biener  <rguenther@suse.de>
1556         * tree-loop-distribution.c (build_empty_rdg): Inline into
1557         single user.
1558         (rdg_flag_vertex): Inline into single user.
1559         (rdg_flag_vertex_and_dependent): Likewise.
1560         (build_rdg_partition_for_vertex): Remove processed bitmap.
1561         (rdg_build_partitions): Simplify.
1563 2013-10-15  Richard Biener  <rguenther@suse.de>
1565         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1566         Restructure forwarding through conversions and copies to
1567         avoid performing copy-propagation the wrong way.  Adjust
1568         recursion invocations.
1569         (forward_propagate_addr_expr): Add argument stating if we
1570         are recursing from a single-use.
1571         (ssa_forward_propagate_and_combine): Adjust.
1573 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1575         * dumpfile.h (gcc::dump_manager): New class, to hold state
1576         relating to dumpfile management.
1577         (get_dump_file_name): Remove in favor of method of dump_manager.
1578         (dump_initialized_p): Likewise.
1579         (dump_start): Likewise.
1580         (dump_finish): Likewise.
1581         (dump_switch_p): Likewise.
1582         (dump_register): Likewise.
1583         (get_dump_file_info): Likewise.
1584         * context.c (gcc::context::context): Construct the dump_manager
1585         instance.
1586         * context.h (gcc::context::get_dumps): New.
1587         (gcc::context::m_dumps): New.
1588         * coverage.c (coverage_init): Port to dump_manager API.
1589         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1590         (extra_dump_files_in_use): Likewise.
1591         (extra_dump_files_alloced): Likewise.
1592         (gcc::dump_manager::dump_manager): New.
1593         (dump_register): Convert to...
1594         (gcc::dump_manager::dump_register): ...method, replacing
1595         function-static next_dump with m_next_dump field.
1596         (get_dump_file_info): Convert to...
1597         (gcc::dump_manager::get_dump_file_info): ...method.
1598         (get_dump_file_name): Convert to...
1599         (gcc::dump_manager::get_dump_file_name): ...method.
1600         (dump_start): Convert to...
1601         (gcc::dump_manager::dump_start): ...method.
1602         (dump_finish): Convert to...
1603         (gcc::dump_manager::dump_finish): ...method.
1604         (dump_begin): Replace body with...
1605         (gcc::dump_manager::dump_begin): ...new method.
1606         (dump_phase_enabled_p): Convert to...
1607         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1608         (dump_phase_enabled_p): Convert to...
1609         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1610         (dump_initialized_p):  Convert to...
1611         (gcc::dump_manager::dump_initialized_p): ...method.
1612         (dump_flag_name): Replace body with...
1613         (gcc::dump_manager::dump_flag_name): ...new method.
1614         (dump_enable_all): Convert to...
1615         (gcc::dump_manager::dump_enable_all): ...new method.
1616         (opt_info_enable_passes): Convert to...
1617         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1618         (dump_switch_p_1): Convert to...
1619         (gcc::dump_manager::dump_switch_p_1): ...new method.
1620         (dump_switch_p):  Convert to...
1621         (gcc::dump_manager::dump_switch_p): ...new method.
1622         (opt_info_switch_p): Port to dump_manager API.
1623         (enable_rtl_dump_file): Likewise.
1624         * opts-global.c (handle_common_deferred_options): Port to new
1625         dump_manager API.
1626         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1627         (pass_manager::register_one_dump_file): Likewise.
1628         (pass_manager::register_pass): Likewise.
1629         (pass_init_dump_file): Likewise.
1630         (pass_fini_dump_file): Likewise.
1631         * statistics.c (statistics_early_init): Likewise.
1633 2013-10-14  Richard Biener  <rguenther@suse.de>
1635         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1636         iterative_hash_canonical_type, gimple_canonical_type_hash,
1637         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1638         gimple_register_canonical_type, print_gimple_types_stats,
1639         free_gimple_type_tables): Move to lto/lto.c
1640         (gt-gimple.h): Do not include.
1641         * gimple.h (gimple_register_canonical_type,
1642         print_gimple_types_stats, free_gimple_type_tables): Remove.
1643         * Makefile.in (GTFILES): Remove gimple.c.
1645 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1647         PR target/58716
1648         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1649         scanning for msp430x targets.
1651 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1653         PR bootstrap/58509
1654         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1655         (registers_ok_for_ldd_peep): Move around.
1656         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1657         * config/sparc/sparc.md (widening peepholes): Use it.
1659 2013-10-14  Richard Biener  <rguenther@suse.de>
1661         PR middle-end/58712
1662         PR middle-end/55358
1663         * gimple.c (iterative_hash_canonical_type): Make sure to
1664         record the hash into the correct hashtable slot.
1666 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1668         PR rtl-optimization/58662
1669         * combine.c (try_combine): Take into account death nodes on I2 when
1670         splitting a PARALLEL of two independent SETs.  Fix dump message.
1672 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1674         PR target/51244
1675         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1676         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1677         * config/sh/t-sh (sh_treg_combine.o): New entry.
1678         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1679         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1680         (register_sh_passes): New function.  Register sh_treg_combine pass.
1681         (sh_option_override): Invoke it.
1682         (sh_canonicalize_comparison): Handle op0_preserve_value.
1683         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1684         opportunities.  Canonicalize branch condition.
1685         (nott): Allow only if pseudos can be created for non-SH2A.
1687 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1689         PR target/58690
1690         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1691         (ix86_expand_movmem): Replace copy_addr_to_reg with
1692         ix86_copy_addr_to_reg.
1693         (ix86_expand_setmem): Likewise.
1695 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1697         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1698         provided by ix86_fp_compare_mode instead of CCFPUmode.
1700 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1702         * config/aarch64/arm_neon.h
1703         (vtbx<1,3>_<psu>8): Fix register constriants.
1705 2013-10-11  Jeff Law  <law@redhat.com>
1707         PR tree-optimization/58640
1708         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1709         threading paths that cross over two loop entry points.
1711 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1713         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1714         handle vector float as well.
1715         (*vsx_le_perm_load_v4si): Likewise.
1716         (*vsx_le_perm_store_v2di): Likewise.
1717         (*vsx_le_perm_store_v4si): Likewise.
1719 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1721         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1722         directly to circumvent subtract from splat{31} workaround.
1723         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1724         prototype.
1725         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1726         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1727         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1728         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1729         separate big and little endian logic.
1730         (*altivec_vperm_<mode>_internal): New define_insn.
1731         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1732         separate big and little endian logic.
1733         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1734         (vec_permv16qi): Add little endian logic.
1736 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1738         * doc/extend.texi (returns_nonnull): Remove arguments.
1740 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1741             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1742             Sergey Lega  <sergey.s.lega@intel.com>
1743             Anna Tikhonova  <anna.tikhonova@intel.com>
1744             Ilya Tocar  <ilya.tocar@intel.com>
1745             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1746             Ilya Verbin  <ilya.verbin@intel.com>
1747             Kirill Yukhin  <kirill.yukhin@intel.com>
1748             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1750         * config/i386/sse.md (VI48F_256_512): New.
1751         (avx2_permvar<mode>): Change to ...
1752         (<avx2_avx512f>_permvar<mode>): This.
1754 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1755             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1756             Sergey Lega  <sergey.s.lega@intel.com>
1757             Anna Tikhonova  <anna.tikhonova@intel.com>
1758             Ilya Tocar  <ilya.tocar@intel.com>
1759             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1760             Ilya Verbin  <ilya.verbin@intel.com>
1761             Kirill Yukhin  <kirill.yukhin@intel.com>
1762             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1764         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1765         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1766         * config/i386/sse.md (VI4_AVX): New.
1767         (sf2simodelower): Ditto.
1768         (sse2_cvtps2dq): Change to ...
1769         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1771 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1772             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1773             Sergey Lega  <sergey.s.lega@intel.com>
1774             Anna Tikhonova  <anna.tikhonova@intel.com>
1775             Ilya Tocar  <ilya.tocar@intel.com>
1776             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1777             Ilya Verbin  <ilya.verbin@intel.com>
1778             Kirill Yukhin  <kirill.yukhin@intel.com>
1779             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1781         * config/i386/sse.md (V_512): New.
1782         (VI_512): Ditto.
1783         (vcond<V_512:mode><VF_512:mode>): Ditto.
1784         (vcond<V_512:mode><VI_512:mode>): Ditto.
1785         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1787 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1788             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1789             Sergey Lega  <sergey.s.lega@intel.com>
1790             Anna Tikhonova  <anna.tikhonova@intel.com>
1791             Ilya Tocar  <ilya.tocar@intel.com>
1792             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1793             Ilya Verbin  <ilya.verbin@intel.com>
1794             Kirill Yukhin  <kirill.yukhin@intel.com>
1795             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1797         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1798         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1799         (FMAMODE): Ditto.
1800         (fma<mode>4): Removed condition.
1801         (fms<mode>4): Ditto.
1802         (fnma<mode>4): Ditto.
1803         (fnms<mode>4): Ditto.
1804         (fma4i_fmadd_<mode>): Ditto.
1805         (*fma_fmadd_<mode>): Ditto.
1806         (*fma_fmsub_<mode>): Ditto.
1807         (*fma_fnmadd_<mode>): Ditto.
1808         (*fma_fnmsub_<mode>): Ditto.
1809         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1810         (*fma_fmaddsub_<mode>): Ditto.
1811         (*fma_fmsubadd_<mode>): Ditto.
1812         (*fmai_fmadd_<mode>): Ditto.
1813         (*fmai_fmsub_<mode>): Ditto.
1814         (*fmai_fnmadd_<mode>): Ditto.
1815         (*fmai_fnmsub_<mode>): Ditto.
1817 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1818             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1819             Sergey Lega  <sergey.s.lega@intel.com>
1820             Anna Tikhonova  <anna.tikhonova@intel.com>
1821             Ilya Tocar  <ilya.tocar@intel.com>
1822             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1823             Ilya Verbin  <ilya.verbin@intel.com>
1824             Kirill Yukhin  <kirill.yukhin@intel.com>
1825             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1827         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1828         (VI124_256): Changed to ...
1829         (VI124_256_48_512): This.
1830         (ssepackmode): Extended with wider modes.
1831         (<code><mode>3): Changed iterator.
1832         (*avx2_<code><mode>3): Ditto.
1833         (vec_pack_trunc_<mode>): Ditto.
1835 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1836             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1837             Sergey Lega  <sergey.s.lega@intel.com>
1838             Anna Tikhonova  <anna.tikhonova@intel.com>
1839             Ilya Tocar  <ilya.tocar@intel.com>
1840             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1841             Ilya Verbin  <ilya.verbin@intel.com>
1842             Kirill Yukhin  <kirill.yukhin@intel.com>
1843             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1845         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1846         (VI8F_256_512): Ditto.
1847         (abs<mode>2): Changed iterator.
1848         (avx2_perm<mode>): Changed to ...
1849         (<avx2_avx512f>_perm<mode>): This.
1850         (avx2_perm<mode>_1): Changed to ...
1851         (<avx2_avx512f>_perm<mode>_1): This.
1853 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1854             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1855             Sergey Lega  <sergey.s.lega@intel.com>
1856             Anna Tikhonova  <anna.tikhonova@intel.com>
1857             Ilya Tocar  <ilya.tocar@intel.com>
1858             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1859             Ilya Verbin  <ilya.verbin@intel.com>
1860             Kirill Yukhin  <kirill.yukhin@intel.com>
1861             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1863         * config/i386/sse.md (VI48_AVX512F): New.
1864         (VI48_AVX2): Changed to ...
1865         (VI48_AVX2_48_AVX512F): This.
1866         (avx2_ashrv<mode>): Changed to ...
1867         (<avx2_avx512f>_ashrv<mode>): This.
1868         (avx2_<shift_insn>v<mode>): Changed to ...
1869         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1871 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1872             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1873             Sergey Lega  <sergey.s.lega@intel.com>
1874             Anna Tikhonova  <anna.tikhonova@intel.com>
1875             Ilya Tocar  <ilya.tocar@intel.com>
1876             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1877             Ilya Verbin  <ilya.verbin@intel.com>
1878             Kirill Yukhin  <kirill.yukhin@intel.com>
1879             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1881         * config/i386/sse.md (VI4_AVX512F): New.
1882         (VI8_AVX2_AVX512F): Ditto.
1883         (mul<mode>3): Extended with wider modes.
1884         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1885         (mul<mode>3): Ditto.
1886         (vec_widen_<s>mult_odd_<mode>): Ditto.
1888 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1889             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1890             Sergey Lega  <sergey.s.lega@intel.com>
1891             Anna Tikhonova  <anna.tikhonova@intel.com>
1892             Ilya Tocar  <ilya.tocar@intel.com>
1893             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1894             Ilya Verbin  <ilya.verbin@intel.com>
1895             Kirill Yukhin  <kirill.yukhin@intel.com>
1896             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1898         * config/i386/sse.md (VI2_AVX512F): New.
1899         (VI124_AVX512F): Ditto.
1900         (sseunpackmode): Extended with wider modes.
1901         (sseunpackfltmode): Ditto.
1902         (vec_unpacks_float_hi_<mode>): Ditto.
1903         (vec_unpacks_float_lo_<mode>): Ditto.
1904         (vec_unpacku_float_hi_<mode>): Ditto.
1905         (vec_unpacku_float_lo_<mode>): Ditto.
1906         (vec_unpacks_lo_<mode>): Ditto.
1907         (vec_unpacks_hi_<mode>): Ditto.
1908         (vec_unpacku_lo_<mode>): Ditto.
1909         (vec_unpacku_hi_<mode>): Ditto.
1911 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1912             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1913             Sergey Lega  <sergey.s.lega@intel.com>
1914             Anna Tikhonova  <anna.tikhonova@intel.com>
1915             Ilya Tocar  <ilya.tocar@intel.com>
1916             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1917             Ilya Verbin  <ilya.verbin@intel.com>
1918             Kirill Yukhin  <kirill.yukhin@intel.com>
1919             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1921         * config/i386/i386.md (multdiv): New.
1922         (multdiv_mnemonic): Ditto.
1923         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1924         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1925         (<sse>_vmdiv<mode>3): Removed.
1927 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1928             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1929             Sergey Lega  <sergey.s.lega@intel.com>
1930             Anna Tikhonova  <anna.tikhonova@intel.com>
1931             Ilya Tocar  <ilya.tocar@intel.com>
1932             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1933             Ilya Verbin  <ilya.verbin@intel.com>
1934             Kirill Yukhin  <kirill.yukhin@intel.com>
1935             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1937         * config/i386/sse.md (V): Extended with wider modes.
1938         (VF2): Ditto.
1939         (ssehalfvecmode): Ditto.
1940         (i128): Ditto.
1941         (ssepackfltmode): Ditto.
1942         (avx_vec_concat<mode>): Ditto.
1943         (V_256_512): New iterator.
1944         (VF2_512_256): Ditto.
1945         (si2dfmode): New attribute.
1946         (si2dfmodelower): Ditto.
1947         (sf2dfmode): Ditto.
1948         (concat_tg_mode): Ditto.
1949         (floatv4siv4df2): Changed to ...
1950         (float<si2dfmodelower><mode>2): This.
1951         (avx_cvtps2pd256): Changed to ...
1952         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1953         (vec_pack_trunc_v4df): Changed to ...
1954         (vec_pack_trunc_<mode>): This.
1955         (avx_vpermil<mode>): Changed to ...
1956         (<sse2_avx_avx512f>_vpermil<mode>): This.
1957         (<fixsuffix>fix_truncv8dfv8si2): New.
1958         (vec_pack_sfix_trunc_v8df): Ditto.
1959         (avx512f_rndscale<mode>): Ditto.
1960         (avx512f_roundpd512): Ditto.
1961         (vec_pack_ufix_trunc_<mode>): Updated iterator.
1963 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1964             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1965             Sergey Lega  <sergey.s.lega@intel.com>
1966             Anna Tikhonova  <anna.tikhonova@intel.com>
1967             Ilya Tocar  <ilya.tocar@intel.com>
1968             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1969             Ilya Verbin  <ilya.verbin@intel.com>
1970             Kirill Yukhin  <kirill.yukhin@intel.com>
1971             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1973         * config/i386/i386.md (any_fix): New iterator.
1974         (fixsuffix): New attribute.
1975         * config/i386/sse.md (VF1): Extened with wider modes.
1976         (VI): Ditto.
1977         (VI_AVX2): Ditto.
1978         (VI8): Ditto.
1979         (sseintvecmodelower): Ditto.
1980         (ssescalarmode): Ditto.
1981         (ssescalarnum): Ditto.
1982         (VF1_128_256): New.
1983         (ssexmmmode): Ditto.
1984         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
1985         (<sse>_rcp<mode>2): Change iterator.
1986         (rsqrt<mode>2): Ditto.
1987         (<sse>_rsqrt<mode>2): Ditto.
1988         (avx2_vec_dup<mode>): Ditto.
1989         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1990         (round<mode>2_sfix): Ditto.
1991         (avx2_pbroadcast<mode>): Ditto.
1992         (*andnot<mode>3): Handle XI mode.
1993         (*<code><mode>3): Ditto.
1994         (AVXTOSSEMODE): Removed.
1995         (avx_vpermil<mode>): Changed to ...
1996         (<sse2_avx_avx512f>_vpermil<mode>): This.
1998 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1999             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2000             Sergey Lega  <sergey.s.lega@intel.com>
2001             Anna Tikhonova  <anna.tikhonova@intel.com>
2002             Ilya Tocar  <ilya.tocar@intel.com>
2003             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2004             Ilya Verbin  <ilya.verbin@intel.com>
2005             Kirill Yukhin  <kirill.yukhin@intel.com>
2006             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2008         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2009         (<sse>_comi): Ditto.
2010         (<sse>_ucomi): Ditto.
2011         (sse_cvtss2siq_2): Ditto.
2012         (sse2_cvtsd2si): Ditto.
2013         (sse2_cvtsd2siq): Ditto.
2014         (sse2_cvttsd2si): Ditto.
2015         (sse2_cvttsd2siq): Ditto.
2016         (<shift_insn><mode>3): Ditto.
2017         (sse2_cvtsi2sdq): Update constraint and prefix.
2018         (sse_cvtsi2ss): Update prefix.
2019         (sse_cvtsi2ssq): Ditto.
2021 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2023         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2024         ignore internal calls.
2026 2013-10-11  Richard Biener  <rguenther@suse.de>
2028         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2029         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2030         TREE_OVERFLOW is set.
2032 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2034         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2036         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2037         * gimple.h: Likewise.
2039         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2041         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2042         macros.
2044         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2046         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2047         * configure: Regenerate.
2049 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2051         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2052         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2053         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2054         similarly pp_right_paren.
2055         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2056         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2057         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2058         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2059         clauses.
2060         (omp_declare_simd_clauses_equal,
2061         omp_remove_redundant_declare_simd_attrs): New functions.
2062         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2063         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2064         * tree.h (OMP_LOOP_CHECK): Define.
2065         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2066         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2067         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2068         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2069         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2070         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2071         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2072         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2073         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2074         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2075         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2076         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2077         (omp_remove_redundant_declare_simd_attrs): New prototype.
2078         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2079         GIMPLE_OMP_TEAMS): New codes.
2080         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2081         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2082         fields.
2083         (target_nesting_level): New variable.
2084         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2085         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2086         for collapse > 1 static schedule unless ordered.
2087         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2088         (determine_parallel_type): Adjust get_ws_args_for caller.
2089         (install_var_field): Handle mask & 4 for double indirection.
2090         (scan_sharing_clauses): Ignore shared clause on teams construct.
2091         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2092         (create_omp_child_function): If inside target or declare target
2093         constructs, set "omp declare target" attribute on the child function.
2094         (find_combined_for): New function.
2095         (scan_omp_parallel): Handle combined loops.
2096         (scan_omp_target, scan_omp_teams): New functions.
2097         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2098         restrictions and set ctx->cancellable for cancellable constructs.
2099         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2100         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2101         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2102         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2103         (omp_clause_aligned_alignment): New function.
2104         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2105         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2106         on teams constructs.  Handle user defined reductions and new
2107         OpenMP 4.0 clauses.
2108         (lower_reduction_clauses): Don't set placeholder to address of ref
2109         if it has already the right type.
2110         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2111         (expand_parallel_call): Use the new non-_start suffixed builtins,
2112         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2113         and GOMP_parallel_end after the call.
2114         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2115         (expand_omp_for_init_counts): Handle combined loops.
2116         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2117         loops.
2118         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2119         end of cancellable loops.
2120         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2121         Likewise.  Handle collapse > 1 loops.
2122         (expand_omp_simd): Handle combined loops.
2123         (expand_omp_for): Add inner_stmt argument, adjust callers of
2124         expand_omp_for* functions, use expand_omp_for_static*chunk even
2125         for collapse > 1 unless ordered.
2126         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2127         of cancellable sections.
2128         (expand_omp_single): Remove need_barrier variable, just rely on
2129         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2130         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2131         (expand_omp_atomic_load, expand_omp_atomic_store,
2132         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2133         (expand_omp_target): New function.
2134         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2135         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2136         (build_omp_regions_1): Immediately close region for
2137         GF_OMP_TARGET_KIND_UPDATE.
2138         (maybe_add_implicit_barrier_cancel): New function.
2139         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2140         cancellation.
2141         (lower_omp_single): Likewise.  Add clobber after the barrier.
2142         (lower_omp_taskgroup): New function.
2143         (lower_omp_for): Handle combined loops.  Adjust
2144         lower_rec_input_clauses caller.  Handle cancellation.
2145         (lower_depend_clauses): New function.
2146         (lower_omp_taskreg): Lower depend clauses.  Adjust
2147         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2148         cancellation.
2149         (lower_omp_target, lower_omp_teams): New functions.
2150         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2151         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2152         and GOMP_cancellation_point calls.
2153         (lower_omp): Fold stmts inside of target region.
2154         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2155         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2156         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2157         (BT_FN_VOID_OMPFN_PTR_UINT,
2158         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2159         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2160         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2161         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2162         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2163         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2164         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2165         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2166         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2167         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2168         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2169         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2170         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2171         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2172         BUILT_IN_GOMP_PARALLEL_END.
2173         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2174         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2175         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2176         GF_OMP_FOR_KIND_DISTRIBUTE.
2177         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2178         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2179         (dump_gimple_omp_return): Print lhs if it has any.
2180         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2181         gimple_omp_atomic_seq_cst_p.
2182         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2183         and GIMPLE_OMP_TEAMS.
2184         * langhooks.c (lhd_omp_mappable_type): New function.
2185         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2186         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2187         hook.
2188         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2189         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2190         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2191         (struct gimplify_omp_ctx): Add combined_loop field.
2192         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2193         on stmts inside of target region.
2194         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2195         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2196         ORT_TARGET and ORT_TARGET_DATA.
2197         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2198         Handle GOVD_MAP.
2199         (omp_notice_threadprivate_variable): Complain about threadprivate
2200         variables in target region.
2201         (omp_notice_variable): Complain about vars with non-mappable type
2202         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2203         (omp_check_private): Ignore ORT_TARGET* regions.
2204         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2205         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2206         (find_combined_omp_for): New function.
2207         (gimplify_omp_for): Handle gimplification of combined loops.
2208         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2209         OMP_TEAMS.
2210         (gimplify_omp_target_update): New function.
2211         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2212         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2213         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2214         (gimplify_body): If fndecl has "omp declare target" attribute, add
2215         implicit ORT_TARGET context around it.
2216         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2217         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2218         * tree-nested.c (convert_nonlocal_reference_stmt,
2219         convert_local_reference_stmt, convert_gimple_call): Handle
2220         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2221         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2222         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2223         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2224         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2225         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2226         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2227         BUILT_IN_GOMP_LOOP_END_CANCEL,
2228         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2229         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2230         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2231         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2232         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2233         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2234         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2235         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2236         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2237         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2238         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2239         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2240         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2241         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2242         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2243         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2244         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2245         gimple_build_omp_teams): New functions.
2246         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2247         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2248         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2249         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2250         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2251         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2252         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2253         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2254         GF_OMP_ATOMIC_SEQ_CST): New.
2255         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2256         gimple_build_omp_teams): New prototypes.
2257         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2258         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2259         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2260         GIMPLE_OMP_SINGLE as end of range.
2261         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2262         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2263         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2264         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2265         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2266         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2267         gimple_omp_target_kind, gimple_omp_target_set_kind,
2268         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2269         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2270         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2271         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2272         gimple_omp_teams_set_clauses): New inlines.
2273         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2274         and GIMPLE_OMP_TASKGROUP.
2275         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2276         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2277         enum omp_clause_proc_bind_kind): New.
2278         (union omp_clause_subcode): Add depend_kind, map_kind and
2279         proc_bind_kind fields.
2280         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2281         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2282         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2283         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2284         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2286 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2288         * predict.c (tree_estimate_probability): Add new parameter
2289         for estimate_bb_frequencies.
2290         (estimate_bb_frequencies): Add new parameter to force estimation.
2291         (rebuild_frequencies): When max frequency in function is small,
2292         recompute counts from frequencies.
2293         * predict.h (estimate_bb_frequencies): New parameter.
2295 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2297         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2298         optimizations are disabled.
2300 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2302         * coverage.c (coverage_finish): Fix leak of da_file_name.
2304 2013-10-10  Jan Hubicka  <jh@suse.cz>
2306         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2307         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2308         Bobcat and generic.
2310 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2312         PR middle-end/58670
2313         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2314         if any labels are in FALLTHRU_BB, use a special label emitted
2315         immediately after the asm goto insn rather than label_rtx
2316         of the LABEL_DECL.
2317         (expand_asm_stmt): Adjust caller.
2318         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2319         edge if the last insn in predecessor is a jump with single successor,
2320         but it isn't simplejump_p.
2322 2013-10-10  Richard Biener  <rguenther@suse.de>
2324         PR tree-optimization/58656
2325         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2327 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2329         * gimplify.c: Include expr.h and tm_p.h for targets with special
2330         va-arg padding requirements.
2332 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2334         * tree-flow.h: Move some prototypes to gimple.h.
2335         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2336         * gimple.h: Relocate some prototypes from tree-flow.h
2337         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2338         Move to gimplify.c.
2339         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2340         (build_va_arg_indirect_ref): Relocate and make static.
2341         (std_gimplify_va_arg_expr): Relocate here.
2342         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2343         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2345 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2347         * doc/md.texi: Document the mnemonic attribute.
2349 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2351         PR target/57377
2352         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2353         (set_attr_alternative x ...) when searching for user defined
2354         mnemonic attribute.
2356 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2358         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2360 2013-10-09  Easwaran Raman <eraman@google.com>
2362         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2363         * cfgexpand.c (defer_stack_allocation): ...use here
2364         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2366 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2368         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2369         (operand_equal_for_value_replacement): New function, extracted from
2370         value_replacement and enhanced to catch more cases.
2371         (value_replacement): Use operand_equal_for_value_replacement.
2373 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2375         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2376         get_max_loop_iterations.
2378 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2380         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2381         Place comment above function.
2383 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2385         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2386         are not related to tree-cfg.c.
2387         * tree-ssa-address.h: New file.  Relocate prototypes.
2388         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2389         (addr_for_mem_ref): New.  Combine call to get_address_description and
2390         return addr_for_mem_ref.
2391         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2392         * tree-ssa-live.h: Adjust prototypes.
2393         * passes.c: Include tree-ssa-live.h.
2394         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2395         * graphite.c (graphite_transform_loops): Make static.
2396         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2397         make_pass_graphite, pass_data_graphite_transforms,
2398         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2399         * ipa-pure-const.c (warn_function_noreturn): Make static.
2400         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2401         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2402         Relocate from tree-cfg.c
2403         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2404         static.
2405         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2406         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2407         Move to ipa-pure-const.c.
2408         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2409         Relocate from tree-optimize.c.
2410         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2411         make_pass_fixup_cfg): Move to tree-cfg.c.
2412         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2413         Relocate some prototypes.
2414         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2415         * tree-dump.c (dump_function_to_file): Remove prototype.
2416         Add tree-flow.h to the include file.
2417         * tree-dump.h: Remove prototype.
2418         * tree-parloops.h: New File.  Add prototypes.
2419         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2420         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2421         from tree-ssa-loop.c.
2422         * tree-predcom.c (run_tree_predictive_commoning,
2423         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2424         Relocate here from tree-ssa-loop.c.
2425         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2426         ssa_name_values.release ().
2427         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2428         (ssa_name_values): Relocate from tree-flow.h.
2429         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2430         * tree-ssa-loop.c (run_tree_predictive_commoning,
2431         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2432         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2433         make_pass_graphite, pass_data_graphite_transforms,
2434         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2435         tree_parallelize_loops, pass_data_parallelize_loops,
2436         make_pass_parallelize_loops): Move to other files.
2437         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2438         moved here.
2439         * tree.h: Remove prototypes from tree-address.c.
2441 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2443         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2444         (struct int_tree_map): Move to tree-hasher.h
2445         (SCALE, LABEL, PERCENT): Move to gimple.h
2446         * tree-flow-inline.h: Delete.  Move functions to other files.
2447         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2448         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2449         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2450         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2451         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2452         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2453         * tree-hasher.h: Don't include tree-flow.h.
2454         (struct int_tree_map): Relocate from tree-flow.h.
2455         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2456         tree-flow-inline.h and make static.
2457         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2458         tree-flow-inline.h.
2459         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2460         tree-flow-inline.h and make static.
2461         * tree.h (is_global_var, may_be_aliased): Relocate from
2462         tree-flow-inline.h.
2463         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2464         * value-prof.c: No longer include tree-flow-inline.h.
2465         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2467 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2469         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2470         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2471         (enum move_pos): Move to tree-ssa-loop-im.h
2472         * cfgloop.h: Move some prototypes.
2473         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2474         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2475         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2476         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2477         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2478         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2479         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2480         (enum move_pos): Relocate here.
2481         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2482         tree-ssa-loop.c.
2483         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2484         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2485         make_pass_lim): Relocate here from tree-ssa-loop.c.
2486         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2487         tree-ssa-loop.c.
2488         (loop_edge_to_cancel, unloop_loops): Make static.
2489         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2490         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2491         (tree_complete_unroll, gate_tree_complete_unroll,
2492         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2493         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2494         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2495         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2496         (stmt_invariant_in_loop_p): Remove unused function.
2497         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2498         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2499         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2500         (gcov_type_to_double_int): Move to cfgloop.h.
2501         (double_int_cmp, bound_index,
2502         estimate_numbers_of_iterations_loop): Make static.
2503         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2504         (max_loop_iterations): Factor out get_max_loop_iterations.
2505         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2506         cfgloop.c.
2507         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2508         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2509         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2510         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2511         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2512         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2513         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2514         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2515         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2516         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2517         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2518         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2519         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2520         pass_data_complete_unroll, make_pass_complete_unroll,
2521         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2522         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2523         tree-ssa-loop-ivcanon.c.
2524         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2525         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2526         tree-ssa-loop-prefetch.c.
2527         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2528         tree-ssa-loop-im.c.
2529         (get_lsm_tmp_name): Relocate and add suffix parameter.
2530         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2531         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2532         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2533         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2534         (get_estimated_loop_iterations): Factor out accessor from
2535         estimated_loop_iterations in tree-ssa-loop-niter.c.
2536         (get_max_loop_iterations): Factor out accessor from
2537         _max_loop_iterations in tree-ssa-niter.c.
2538         * loop-unroll.c (decide_unroll_constant_iterations,
2539         decide_unroll_runtime_iterations, decide_peel_simple,
2540         decide_unroll_stupid): Use new get_* accessors.
2542 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2544         PR tree-optimization/20318
2545         * doc/extend.texi (returns_nonnull): New function attribute.
2546         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2547         attribute.
2548         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2549         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2551 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2553         PR middle-end/58570
2554         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2555         false if both components are bitfields.
2557 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2559         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2560         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2561         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2562         * config/aarch64/aarch64.h
2563         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2564         * config/aarch64/aarch64-simd-builtins.def
2565         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2567 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2569         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2571 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2573         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2575 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2577         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2578         (vneg_f64): New intrinsic.
2579         (vneg_s8): Asm replaced with C.
2580         (vneg_s16): Likewise.
2581         (vneg_s32): Likewise.
2582         (vneg_s64): New intrinsic.
2583         (vnegq_f32): Asm replaced with C.
2584         (vnegq_f64): Likewise.
2585         (vnegq_s8): Likewise.
2586         (vnegq_s16): Likewise.
2587         (vnegq_s32): Likewise.
2588         (vnegq_s64): Likewise.
2590 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2592         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2594 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2596         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2597         packed stack special handling.
2598         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2599         back to fixed stack slots for FPRs saved due to stdarg.
2601 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2603         * config/s390/s390.c (s390_frame_info): Restructure function.
2605 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2607         * config/s390/s390.c (struct s390_frame_layout): New field
2608         gpr_save_slots.
2609         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2610         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2611         regs_ever_clobbered to char*.
2612         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2613         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2614         (s390_register_info): Enable FPR save slots.  Move/Copy some
2615         functionality into ...
2616         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2617         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2618         function.
2619         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2620         not imply a stack frame.
2621         (s390_init_frame_layout): Remove variable clobbered_regs.
2622         (s390_update_register_info): Remove function.
2623         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2624         cannot be used for register renaming.
2625         (s390_hard_regno_scratch_ok): New function.
2626         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2627         (s390_initial_elimination_offset): Change offset calculation of
2628         the return address pointer.
2629         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2630         (restore_gprs): Set frame related flag.
2631         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2632         (s390_emit_prologue): Call s390_register_info instead of
2633         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2634         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2635         (s390_optimize_prologue): Call s390_optimize_register_info.
2636         Try to remove also FPR slot save/restore INSNs.  Remove frame
2637         related flags from restore INSNs.
2639 2013-10-08  DJ Delorie  <dj@redhat.com>
2641         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2642         (movhi): Likewise.
2644         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2645         avoid conflict with the MI use of %c.
2646         * config/rl78/rl78-real.md: change %c to %C throughout.
2647         * config/rl78/rl78-virt.md: Likewise.
2649 2013-10-08  Jan Hubicka  <jh@suse.cz>
2651         * config/i386/i386.c (ix86_option_override_internal): Switch
2652         to SSE math for -ffast-math when target ISA supports SSE2.
2654 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2656         * tree-flow.h: Remove some prototypes.
2657         * tree.h: Remove some protypes, add a couple.
2658         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2659         using_eh_for_cleanups_p): Add interface routines for front ends.
2660         * tree-eh.h: New file.  Add protoptyes.
2661         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2662         (add_stmt_to_eh_lp_fn): Make static.
2663         (lower_try_finally): Use new using_eh_for_cleanups_p.
2664         * emit-rtl.c: Include tree-eh.h.
2665         * gimple.h: Include tree-eh.h.
2667 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2669         PR tree-optimization/58480
2670         * tree-vrp.c (infer_nonnull_range): New function.
2671         (infer_value_range): Call infer_nonnull_range.
2673 2013-10-08  Dehao Chen  <dehao@google.com>
2675         PR tree-optimization/58619
2676         * tree-inline.c (copy_phis_for_bb): Combine location data
2677         only if non-null.
2679 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2681         PR target/58423
2682         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2683         RTX_FRAME_RELATED_P on INSN.
2685 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2687         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2688         (altivec_expand_vec_perm_const): Call it.
2690 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2692         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2693         sequences for LE VSX loads and stores at expand time.
2694         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2695         prototype.
2696         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2697         (rs6000_gen_le_vsx_permute): New.
2698         (rs6000_gen_le_vsx_load): New.
2699         (rs6000_gen_le_vsx_store): New.
2700         (rs6000_gen_le_vsx_move): New.
2701         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2702         (*vsx_le_perm_load_v4si): New.
2703         (*vsx_le_perm_load_v8hi): New.
2704         (*vsx_le_perm_load_v16qi): New.
2705         (*vsx_le_perm_store_v2di): New.
2706         (*vsx_le_perm_store_v4si): New.
2707         (*vsx_le_perm_store_v8hi): New.
2708         (*vsx_le_perm_store_v16qi): New.
2709         (*vsx_xxpermdi2_le_<mode>): New.
2710         (*vsx_xxpermdi4_le_<mode>): New.
2711         (*vsx_xxpermdi8_le_V8HI): New.
2712         (*vsx_xxpermdi16_le_V16QI): New.
2713         (*vsx_lxvd2x2_le_<mode>): New.
2714         (*vsx_lxvd2x4_le_<mode>): New.
2715         (*vsx_lxvd2x8_le_V8HI): New.
2716         (*vsx_lxvd2x16_le_V16QI): New.
2717         (*vsx_stxvd2x2_le_<mode>): New.
2718         (*vsx_stxvd2x4_le_<mode>): New.
2719         (*vsx_stxvd2x8_le_V8HI): New.
2720         (*vsx_stxvd2x16_le_V16QI): New.
2722 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2724         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2726 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2728         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2729         loop to work also for 31bit ABI.
2730         Save the stack pointer for frame_size > 0.
2732 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2734         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2735         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2736         constraint letters from expanders.
2737         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2738         retry count to general_operand.
2739         ("tabort"): Give operand 0 a mode.
2740         ("tabort_1"): Add mode and constraint letter for operand 0.
2741         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2743 2013-10-04  Jeff Law  <law@redhat.com>
2745         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2747         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2748         out of ...
2749         (thread_across_edge): Here.  Call it.
2751 2013-10-04  Cary Coutant  <ccoutant@google.com>
2753         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2754         discarding a location list expression (or a piece of one).
2756 2013-10-03  Jan Hubicka  <jh@suse.cz>
2758         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2759         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2760         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2762 2013-10-03  Jan Hubicka  <jh@suse.cz>
2764         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2765         accumulate-outgoing-args when producing unwind info.
2767 2013-10-03  Wei Mi  <wmi@google.com>
2769         * lra-constraints.c (insert_move_for_subreg): New function
2770         extracted from simplify_operand_subreg.
2771         (simplify_operand_subreg): Add reload for paradoxical subreg.
2773 2013-10-03  Rong Xu  <xur@google.com>
2775         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2776         the candidate of builtin_expect such that we should fix the
2777         size/time estimation.
2778         (estimate_function_body_sizes): Do the acutally size/time fix-up
2779         for builtin_expect.
2781 2013-10-03  Rong Xu  <xur@google.com>
2783         * predict.c (tree_predict_by_opcode): Get the probability
2784         for builtin_expect from param builtin_expect_probability.
2785         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2786         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2787         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2789 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2791         PR c++/19476
2792         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2793         * calls.c (alloca_call_p): Use get_callee_fndecl.
2794         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2795         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2796         Likewise.
2797         (vrp_visit_stmt): Remove duplicated code.
2799 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2801         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2802         ceildf2, btruncdf2, instead of vsx_* name.
2804         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2805         iterators to only do V2DF and V4SF here.  Move the DF code to
2806         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2807         just 'v' since only vector operations are handled with these insns
2808         after moving the DF support to rs6000.md.
2809         (vsx_sub<mode>3): Likewise.
2810         (vsx_mul<mode>3): Likewise.
2811         (vsx_div<mode>3): Likewise.
2812         (vsx_fre<mode>2): Likewise.
2813         (vsx_neg<mode>2): Likewise.
2814         (vsx_abs<mode>2): Likewise.
2815         (vsx_nabs<mode>2): Likewise.
2816         (vsx_smax<mode>3): Likewise.
2817         (vsx_smin<mode>3): Likewise.
2818         (vsx_sqrt<mode>2): Likewise.
2819         (vsx_rsqrte<mode>2): Likewise.
2820         (vsx_fms<mode>4): Likewise.
2821         (vsx_nfma<mode>4): Likewise.
2822         (vsx_copysign<mode>3): Likewise.
2823         (vsx_btrunc<mode>2): Likewise.
2824         (vsx_floor<mode>2): Likewise.
2825         (vsx_ceil<mode>2): Likewise.
2826         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2827         (vsx_sminsf3): Likewise.
2828         (vsx_fmadf4): Likewise.
2829         (vsx_fmsdf4): Likewise.
2830         (vsx_nfmadf4): Likewise.
2831         (vsx_nfmsdf4): Likewise.
2832         (vsx_cmpdf_internal1): Likewise.
2834         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2835         simpler to select whether a target has SPE or traditional floating
2836         point support in iterators.
2837         (TARGET_DF_SPE): Likewise.
2838         (TARGET_SF_FPR): Likewise.
2839         (TARGET_DF_FPR): Likewise.
2840         (TARGET_SF_INSN): Macros to say whether floating point support
2841         exists for a given operation for expanders.
2842         (TARGET_DF_INSN): Likewise.
2844         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2845         combining of SF/DF mode operations, using both traditional and VSX
2846         registers.
2847         (Fvsx): Likewise.
2848         (Ff): Likewise.
2849         (Fv): Likewise.
2850         (Fs): Likewise.
2851         (Ffre): Likewise.
2852         (FFRE): Likewise.
2853         (abs<mode>2): Combine SF/DF modes using traditional floating point
2854         instructions.  Add support for using the upper DF registers with
2855         VSX support, and SF registers with power8-vector support.  Update
2856         expanders for operations supported by both the SPE and traditional
2857         floating point units.
2858         (abs<mode>2_fpr): Likewise.
2859         (nabs<mode>2): Likewise.
2860         (nabs<mode>2_fpr): Likewise.
2861         (neg<mode>2): Likewise.
2862         (neg<mode>2_fpr): Likewise.
2863         (add<mode>3): Likewise.
2864         (add<mode>3_fpr): Likewise.
2865         (sub<mode>3): Likewise.
2866         (sub<mode>3_fpr): Likewise.
2867         (mul<mode>3): Likewise.
2868         (mul<mode>3_fpr): Likewise.
2869         (div<mode>3): Likewise.
2870         (div<mode>3_fpr): Likewise.
2871         (sqrt<mode>3): Likewise.
2872         (sqrt<mode>3_fpr): Likewise.
2873         (fre<Fs>): Likewise.
2874         (rsqrt<mode>2): Likewise.
2875         (cmp<mode>_fpr): Likewise.
2876         (smax<mode>3): Likewise.
2877         (smin<mode>3): Likewise.
2878         (smax<mode>3_vsx): Likewise.
2879         (smin<mode>3_vsx): Likewise.
2880         (negsf2): Delete SF operations that are merged with DF.
2881         (abssf2): Likewise.
2882         (addsf3): Likewise.
2883         (subsf3): Likewise.
2884         (mulsf3): Likewise.
2885         (divsf3): Likewise.
2886         (fres): Likewise.
2887         (fmasf4_fpr): Likewise.
2888         (fmssf4_fpr): Likewise.
2889         (nfmasf4_fpr): Likewise.
2890         (nfmssf4_fpr): Likewise.
2891         (sqrtsf2): Likewise.
2892         (rsqrtsf_internal1): Likewise.
2893         (smaxsf3): Likewise.
2894         (sminsf3): Likewise.
2895         (cmpsf_internal1): Likewise.
2896         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2897         (negdf2): Delete DF operations that are merged with SF.
2898         (absdf2): Likewise.
2899         (nabsdf2): Likewise.
2900         (adddf3): Likewise.
2901         (subdf3): Likewise.
2902         (muldf3): Likewise.
2903         (divdf3): Likewise.
2904         (fred): Likewise.
2905         (rsqrtdf_internal1): Likewise.
2906         (fmadf4_fpr): Likewise.
2907         (fmsdf4_fpr): Likewise.
2908         (nfmadf4_fpr): Likewise.
2909         (nfmsdf4_fpr): Likewise.
2910         (sqrtdf2): Likewise.
2911         (smaxdf3): Likewise.
2912         (smindf3): Likewise.
2913         (cmpdf_internal1): Likewise.
2914         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2915         (btrunc<mode>2): Delete separate expander, and combine with the
2916         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2917         (btrunc<mode>2_fpr): Likewise.
2918         (ceil<mode>2): Likewise.
2919         (ceil<mode>2_fpr): Likewise.
2920         (floor<mode>2): Likewise.
2921         (floor<mode>2_fpr): Likewise.
2922         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2923         Add support for using the upper registers with VSX and
2924         power8-vector.  Move insns to be closer to the define_expands. On
2925         VSX systems, prefer the traditional form of FMA over the VSX
2926         version, since the traditional form allows the target not to
2927         overlap with the inputs.
2928         (fms<mode>4_fpr): Likewise.
2929         (nfma<mode>4_fpr): Likewise.
2930         (nfms<mode>4_fpr): Likewise.
2932 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2933             Richard Earnshaw  <richard.earnshaw@arm.com>
2935         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2936         (struct mult_cost_table): Likewise.
2937         (struct mem_cost_table): Likewise.
2938         (struct fp_cost_table): Likewise.
2939         (struct vector_cost_table): Likewise.
2940         (cpu_cost_table): Likewise.
2941         * config/arm/arm.opt (mold-rts-costs): New option.
2942         (mnew-generic-costs): Likewise.
2943         * config/arm/arm.c (generic_extra_costs): New table.
2944         (cortexa15_extra_costs): Likewise.
2945         (arm_slowmul_tune): Use NULL as new costs.
2946         (arm_fastmul_tune): Likewise.
2947         (arm_strongarm_tune): Likewise.
2948         (arm_xscale_tune): Likewise.
2949         (arm_9e_tune): Likewise.
2950         (arm_v6t2_tune): Likewise.
2951         (arm_cortex_a5_tune): Likewise.
2952         (arm_cortex_a9_tune): Likewise.
2953         (arm_v6m_tune): Likewise.
2954         (arm_fa726te_tune): Likewise.
2955         (arm_cortex_a15_tune): Use cortex15_extra_costs.
2956         (arm_cortex_tune): Use generict_extra_costs.
2957         (shifter_op_p): New function.
2958         (arm_unspec_cost): Likewise.
2959         (LIBCALL_COST): Define.
2960         (arm_new_rtx_costs): New function.
2961         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
2962         table is available. Use old costs otherwise unless mnew-generic-costs
2963         is specified.
2964         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
2965         (cpu_cost_table): Declare.
2967 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2969         PR target/58460
2970         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
2971         (*subs_mul_imm_<mode>)
2972         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
2973         (*sub_<shift>_<mode>)
2974         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
2975         Remove k constraint.
2977 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
2979         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
2980         code.
2981         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
2983 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2985         * predict.c (probably_never_executed): New function.
2986         (probably_never_executed_bb_p): Invoke probably_never_executed.
2987         (probably_never_executed_edge_p): Ditto.
2988         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
2989         Treat profile insanities conservatively.
2991 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
2993         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
2995 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
2997         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
2998         use it.  Use smaller increase for scratch.  Don't increase reject
2999         for early clobber scratch.
3000         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3001         setting eliminated regs except setting fp from hfp.
3002         (lra_eliminate): Check lra_insn_recog_data on NULL.
3004 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3006         PR target/58587
3007         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3008         setting -mvsx-timode by default until the underlying problem is fixed.
3009         (RS6000_CPU, power7 defaults): Likewise.
3011 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3013         * config/x-linux (host-linux.o): Remove header dependencies.
3014         Use $(COMPILE) and $(POSTCOMPILE).
3015         * config/t-linux-android (linux-android.o): Ditto.
3017 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3019         * Makefile.in (expmed.o-warn): Remove.
3021 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3023         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3024         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3026 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3028         * dojump.c (do_jump_1): Divide probability between
3029         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3031 2013-10-02  Tom Tromey  <tromey@redhat.com>
3033         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3035 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3037         * tree-flow.h: Remove some prototypes.
3038         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3039         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3040         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3041         mark_virtual_phi_result_for_renaming): Relocate here.
3042         * tree-into-ssa.h: Add prototypes.
3043         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3044         single_pred_before_succ_order.
3045         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3046         (nonfreeing_call_p) Move to gimple.c.
3047         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3048         tree-ssa-phiopt.c.
3049         * basic-block.h (single_pred_before_succ_order): Add prototype.
3050         * gimple.c (nonfreeing_call_p): Relocate here.
3051         * gimple.h: Add prototype.
3052         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3053         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3054         * tree-ssa.h: Include tree-ssa-dom.h.
3056 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3058         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3059         Use $(COMPILE) and $(POSTCOMPILE).
3061         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3063 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3065         * tree-flow.h: Remove some prototypes.
3066         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3067         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3068         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3069         propagate_tree_value*): Move from here to...
3070         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3071         propagate_tree_value*): Relocate here.
3072         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3073         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3074         * gimple-fold.c: Remove gimple-fold.h from include list.
3075         * tree-vrp.c: Remove gimple-fold.h from include list.
3076         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3077         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3078         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3079         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3080         * sese.c: Add tree-ssa-propagate.h to include list.
3082 2013-10-02  Richard Biener  <rguenther@suse.de>
3084         * tree-loop-distribution.c: Include tree-vectorizer.h for
3085         find_loop_location.
3086         (enum partition_kind): Remove PKIND_REDUCTION.
3087         (struct partition_s): Remove has_writes member, add reduction_p member.
3088         (partition_alloc): Adjust.
3089         (partition_builtin_p): Likewise.
3090         (partition_has_writes): Remove.
3091         (partition_reduction_p): New function.
3092         (partition_merge_into): Likewise.
3093         (generate_code_for_partition): Commonize builtin partition
3094         handling tail.
3095         (rdg_cannot_recompute_vertex_p): Remove.
3096         (already_processed_vertex_p): Likewise.
3097         (rdg_flag_vertex): Do not set has_writes.
3098         (classify_partition): Adjust.
3099         (rdg_build_partitions): Do not set has_writes, treat all
3100         partitions as useful.
3101         (distribute_loop): Record number of library calls generated.  Adjust.
3102         (tree_loop_distribution): Report number of loops and library
3103         calls generated as opt-info.
3105 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3107         * tree-flow.h: Include new .h files.  Move prototypes.
3108         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3109         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3110         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3111         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3112         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3113         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3114         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3115         * tree.h (get_ref_base_and_extent): Move prototype out.
3116         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3117         tree-dfa.h.
3118         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3119         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3120         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3121         * tree-scalar-evolution.c: Include tree.h.
3122         * sese.c: Include tree.h.
3123         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3124         * dwarf2out.c: Include tree-dfa.h.
3125         * tree-chrec.c: Include tree.h.
3126         * tree-data-ref.c: Include tree.h.
3128 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3130         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3131         Fix whitespace.
3133 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3135         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3136         Use $(COMPILE) and $(POSTCOMPILE).
3137         (sol2-cxx.o): Likewise.
3138         (sol2-stubs.o): Likewise.
3139         (sol2.o): Likewise.
3140         * config/x-solaris (host-solaris.o): Likewise.
3142         * config/sparc/t-sparc (sparc.o): Remove.
3143         (sparc-c.o): Remove header dependencies.
3144         Use $(COMPILE) and $(POSTCOMPILE).
3145         * config/sparc/x-sparc: Likewise.
3147 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3149         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3150         * config/arc/arc700.md: Likewise.
3151         * config/arc/arc-modes.def: Likewise.
3152         * config/arc/arc-simd.h: Likewise.
3153         * config/arc/t-arc-uClibc: Likewise.
3154         * config/arc/t-arc-newlib: Likewise.
3156 2013-10-02  Renlin Li  <renlin.li@arm.com>
3158         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3159         plus_constant.
3160         (aarch64_expand_epilogue): Likewise.
3162 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3163             Yufeng Zhang  <yufeng.zhang@arm.com>
3165         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3166         declaration.
3167         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3168         'base_in' represent a conversion and legal_cast_p_1 holds; set
3169         'base_in' with the returned value from get_unwidened.
3171 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3173         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3174         plus_constant is not used.
3176 2013-10-01  Wei Mi  <wmi@google.com>
3178         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3179         * config/i386/i386.md: Add define_peephole2 to
3180         break partial reg stall for cvtss2sd/cvtsd2ss.
3182 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3184         * config/arc/arc.c (pass_arc_ifcvt::clone):
3185         Update for ctxt_ -> m_ctxt change.
3187 2013-10-01  Jeff Law  <law@redhat.com>
3189         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3190         outgoing_edge and intermediate_edge fields.  Instead store the path.
3191         (redirection_data::hash): Hash on the last edge's destination index.
3192         (redirection_data::equal): Check the entire thread path.
3193         (lookup_redirectio_data): Corresponding changes.
3194         (create_edge_and_update_destination_phis): Likewise.
3195         (thread_single_edge): Likewise.
3197 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3198             Diego Novillo <dnovillo@google.com>
3200         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3201         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3202         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3203         versions of these patterns.
3205         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3206         (__builtin_arc_aligned): Likewise.
3208         * config/arc/arc.md: Expand adc_0 comment stating the intended
3209         purpose and why it isn't ready.
3210         Replace commented out call_value_via_label_mixed with a
3211         plain comment about bl_s.
3213         * config/arc/arc.c (stdio.h): Don't include directly.
3214         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3215         Remove if (1) condition.
3216         (arc_encode_section_info): Fix comment.
3218 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3220         * config/arc/arc.c (arc_conditional_register_usage):
3221         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3222         Also set reg_alloc_order for DMA config regs.
3224 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3225             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3227         * doc/install.texi (--with-cpu): Mention ARC.
3228         (arc-*-elf32): New paragraph.
3229         (arc-linux-uclibc): Likewise.
3230         * doc/md.texi (Machine Constraints): Add ARC part.
3231         * doc/invoke.texi: (menu): Add ARC Options.
3232         (Machine Dependent Options) <ARC Options>: Add synopsis.
3233         (node ARC Options): Add.
3234         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3235         (ARC Built-in Functions): New section defining
3236         generic ARC built-in functions.
3237         (ARC SIMD Built-in Functions): New section defining SIMD specific
3238         built-in functions.
3239         (Declaring Attributes of Functions): Extended
3240         description of short_call and long_call attributes for ARC and
3241         added index entries.
3243 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3244             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3245             Joern Rennecke  <joern.rennecke@embecosm.com>
3246             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3247             Brendan Kehoe  <brendan@zen.org>
3248             Michael Eager  <eager@eagercon.com>
3249             Simon Cook  <simon.cook@embecosm.com>
3250             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3252         * config/arc, common/config/arc: New directories.
3254 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3255             Brendan Kehoe  <brendan@zen.org>
3256             Simon Cook  <simon.cook@embecosm.com>
3258         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3260 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3262         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3263         * tree-ssa-coalesce.h: New. Move prototype to here.
3264         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3265         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3266         (gimple_can_coalesce_p): Move to...
3267         * gimple.c (gimple_can_coalesce_p): Here.
3269 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3271         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3272         (dump_decl_set): Move to gimple.c.
3273         * gimple.h: Don't include tree-ssa-operands.h.
3274         (dump_decl_set): Add prototype.
3275         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3276         Move to gimple-ssa.h.
3277         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3278         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3279         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3280         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3281         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3282         than PHI_ARG_DEF.
3283         (dump_decl_set): Relocate here.
3284         * gimple-ssa.h: New file.
3285         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3286         Relocate from gimple.h.
3287         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3288         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3289         swap_tree_operands and remove non-ssa path.
3290         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3291         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3292         swap_ssa_operands.
3293         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3294         vect_is_simple_reduction_1): Use swap_ssa_operands.
3295         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3296         (enum need_phi_state): Move to tree-into-ssa.c.
3297         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3298         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3299         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3300         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3301         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3302         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3303         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3304         num_imm_uses): Move to ssa-iterators.h.
3305         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3306         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3307         tree-phinodes.h.
3308         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3309         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3310         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3311         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3312         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3313         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3314         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3315         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3316         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3317         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3318         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3319         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3320         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3321         (set_phi_nodes): Move to tree-phinodes.h.
3322         * tree-ssa-operands.h (enum ssa_op_iter_type,
3323         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3324         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3325         (dump_decl_set): Remove prototype.
3326         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3327         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3328         (set_phi_nodes): Relocate from tree-flow-inline.h.
3329         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3330         tree-flow-inline.h
3331         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3332         include list.  Temporarily add gimple.h to include list.
3333         * ssa-iterators.h: New file.
3334         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3335         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3336         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3337         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3338         Relocate from tree-ssa-operands.h.
3339         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3340         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3341         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3342         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3343         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3344         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3345         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3346         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3347         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3348         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3349         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3350         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3351         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3352         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3353         Relocate from tree-flow-inline.h.
3354         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3356 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3358         * aarch64-simd.md
3359         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3360         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3361         Insert '\t' to output template.
3362         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3363         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3364         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3365         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3367 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3369         * doc/install.texi (Host/target specific installation notes for GCC):
3370         Put @anchor before @heading.
3371         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3372         Use @email for email addresses.
3374 2013-10-01  Jeff Law  <law@redhat.com>
3376         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3377         a vec.  Only delete the path if we create one without successfully
3378         registering a jump thread.
3379         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3380         as a pointer.
3381         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3382         (paths): New vector of jump threading paths.
3383         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3384         (THREAD_PATH): New accessor macro for the entire thread path.
3385         (lookup_redirection_data): Get intermediate and final outgoing edge
3386         from the thread path.
3387         (create_edge_and_update_destination_phis): Copy the threading path.
3388         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3389         jump threading path.
3390         (ssa_redirect_edges): Get edges and block types from the jump threading
3391         path.  Free the path vector.
3392         (thread_block): Get edges from the jump threading path.  Look at the
3393         entire path to see if we thread to a loop exit.  If we cancel a jump
3394         thread request, then free the path vector.
3395         (thread_single_edge): Get edges and block types from the jump threading
3396         path.  Free the path vector.
3397         (thread_through_loop_header): Get edges and block types from the jump
3398         threading path.  Free the path vector.
3399         (mark_threaded_blocks): Iterate over the vector of paths and store
3400         the path on the appropriate edge.  Get edges and block types from the
3401         jump threading path.
3402         (mark_threaded_blocks): Get edges and block types from the jump
3403         threading path.  Free the path vector.
3404         (thread_through_all_blocks): Use the vector of paths rather than
3405         a vector of 3-edge sets.
3406         (register_jump_thread): Accept pointer to a path vector rather
3407         than the path vector itself.  Store the path vector for later use.
3408         Simplify.
3410 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3411             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3413         PR target/58574
3414         * config/s390/s390.c (s390_split_branches): Modify check for table
3415         jump insns.
3416         (s390_chunkify_start): Rearrange table jump insn check in order to
3417         deal with compare and branch insns correctly.
3419 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3421         PR target/58578
3422         Revert
3423         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3424         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3425         define_insn_and_split.
3426         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3427         (shiftsi3_compare): New pattern.
3428         (rrx): New pattern.
3429         * config/arm/unspecs.md (UNSPEC_RRX): New.
3431 2013-10-01  Alan Modra  <amodra@gmail.com>
3433         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3434         casing inout operands.
3436 2013-10-01  Richard Biener  <rguenther@suse.de>
3438         PR tree-optimization/58553
3439         * tree-loop-distribution.c (struct partition_s): Add niter member.
3440         (classify_partition): Populate niter member for the partition
3441         and properly identify whether the relevant store happens before
3442         or after the loop exit.
3443         (generate_memset_builtin): Use niter member from the partition.
3444         (generate_memcpy_builtin): Likewise.
3446 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3448         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3449         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3451 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3453         * basic-block.h (edge_list): Prefix member names with "m_".
3454         * context.h (context): Likewise.
3455         * domwalk.h (dom_walker): Likewise.
3456         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3457         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3458         * hash-table.h (hash_table): Likewise.
3459         * machmode.h (bit_field_mode_iterator): Likewise.
3460         * pass_manager.h (pass_list): Likewise.
3461         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3462         * tree-pass.h (pass_data): Likewise.
3463         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3464         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3465         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3466         * asan.c (pass_data_asan): Update accordingly.
3467         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3468         (control_dependences::control_dependences): Likewise.
3469         (control_dependences::~control_dependences): Likewise.
3470         (control_dependences::~control_dependences): Likewise.
3471         (control_dependences::get_edges_dependent_on): Likewise.
3472         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3473         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3474         * context.c (gcc::context::context): Likewise.
3475         * cprop.c (pass_rtl_cprop::clone): Likewise.
3476         * domwalk.c (dom_walker::walk): Likewise.
3477         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3478         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3479         * mode-switching.c (pass_mode_switching::clone): Likewise.
3480         * passes.c (opt_pass::opt_pass): Likewise.
3481         (pass_manager::pass_manager): Likewise.
3482         * predict.c (pass_strip_predict_hints::clone): Likewise.
3483         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3484         (pass_split_all_insns::clone): Likewise.
3485         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3486         Likewise.
3487         (bit_field_mode_iterator::next_mode): Likewise.
3488         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3489         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3490         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3491         * tree-complex.c (pass_lower_complex::clone): Likewise.
3492         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3493         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3494         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3495         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3496         (pass_fold_builtins::clone): Likewise.
3497         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3498         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3499         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3500         (pass_cd_dce::clone): Likewise.
3501         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3502         (pass_phi_only_cprop::clone): Likewise.
3503         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3504         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3505         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3506         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3507         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3508         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3509         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3510         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3511         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3512         * tree-vrp.c (pass_vrp::clone): Likewise.
3513         * tsan.c (pass_tsan::clone): Likewise.
3515 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3517         PR middle-end/58564
3518         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3519         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3521         PR middle-end/58564
3522         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3523         optimization, punt if sign_bit_p looked through any zero extension.
3525 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3527         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3528         Update redirected out edge count in joiner case.
3529         (ssa_redirect_edges): Common the joiner and non-joiner cases
3530         so that joiner case gets profile updates.
3532 2013-09-30  Richard Biener  <rguenther@suse.de>
3534         PR tree-optimization/58554
3535         * tree-loop-distribution.c (classify_partition): Require
3536         unconditionally executed stores for memcpy and memset recognition.
3537         (tree_loop_distribution): Calculate dominance info.
3539 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3541         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3542         (NO_PROFILE_COUNTERS): Likewise.
3543         (PROFILE_HOOK): Likewise.
3544         (FUNCTION_PROFILER): Likewise.
3545         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3547 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3549         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3550         calls and defines in TARGET_MACHO conditional.
3551         (load_macho_picbase_di): Likewise.
3552         (reload_macho_picbase): Likewise.
3553         (reload_macho_picbase_si): Likewise.
3554         (reload_macho_picbase_di): Likewise.
3555         (nonlocal_goto_receiver): Likewise.
3557 2013-09-30  Nick Clifton  <nickc@redhat.com>
3559         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3560         that use the MSP430X ISA.
3561         (msp430_option_override): Scan -mmcu command line option for any
3562         MCU name that supports the MSP430X ISA.
3563         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3564         -mmcu options which enable the MSP430X ISA.
3566 2013-09-30  Richard Biener  <rguenther@suse.de>
3568         PR middle-end/58532
3569         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3570         before looking for setjmp-like calls.
3572 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3574         PR target/10901
3575         * config/darwin-protos.h (machopic_get_function_picbase): New.
3576         * config/darwin.c (machopic_get_function_picbase): New.
3577         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3578         label for a new func.  (load_macho_picbase_di): Likewise.
3579         (reload_macho_picbase): New expand.
3580         (reload_macho_picbase_si): New insn.
3581         (reload_macho_picbase_di): New insn.
3582         (nonlocal_goto_receiver): New define and split.
3583         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3584         (unspecv enum): Add UNSPECV_NLGR.
3586 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3588         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3589         that altivec registers are correctly sized on Darwin.
3591 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3593         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3594         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3595         * config/x-darwin (host-darwin.o): Likewise.
3596         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3597         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3598         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3600 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3602         * doc/invoke.texi: Fix usage of @tie{} command.
3604 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3606         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3608 2013-09-28  Jan Hubicka  <jh@suse.cz>
3610         * config/i386/x86-tune.def: Add documentation for each of the options;
3611         add whitespace.
3613 2013-09-28  Jan Hubicka  <jh@suse.cz>
3615         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3616         generic.
3617         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3618         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3619         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3620         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3621         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3623 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3625         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3626         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3627         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3628         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3629         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3630         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3631         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3632         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3633         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3634         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3635         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3636         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3637         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3638         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3639         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3640         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3641         gimple.h, godump.c, graphite-clast-to-gimple.c,
3642         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3643         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3644         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3645         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3646         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3647         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3648         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3649         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3650         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3651         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3652         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3653         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3654         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3655         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3656         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3657         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3658         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3659         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3660         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3661         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3662         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3663         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3664         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3665         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3666         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3667         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3668         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3669         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3670         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3671         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3672         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3673         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3674         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3675         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3676         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3677         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3678         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3679         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3680         whitespace before "(".
3682 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3684         * expr.h (extract_bit_field): Remove packedp parameter.
3685         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3686         from forward declaration.
3687         (store_split_bit_field): Remove packedp arg from calls to
3688         extract_fixed_bit_field.
3689         (extract_bit_field_1): Remove packedp parameter and packedp
3690         argument from recursive calls and calls to extract_fixed_bit_field.
3691         (extract_bit_field): Remove packedp parameter and corresponding
3692         arg to extract_bit_field_1.
3693         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3694         to issue warnings.
3695         (extract_split_bit_field): Remove packedp arg from call to
3696         extract_fixed_bit_field.
3697         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3698         (copy_blkmode_from_reg): Likewise.
3699         (copy_blkmode_to_reg): Likewise.
3700         (read_complex_part): Likewise.
3701         (store_field): Likewise.
3702         (expand_expr_real_1): Likewise.
3703         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3704         to extract_bit_field.
3705         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3706         call to extract_bit_field.
3707         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3708         call to extract_bit_field.
3709         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3710         and special packedp behavior from -fstrict-volatile-bitfields
3711         documentation.
3713 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3715         * lra-eliminations.c (init_elim_table): Guard value_p.
3717 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3719         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3720         DFmode, DImode, and SFmode in the upper VSX registers based on the
3721         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3722         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3723         of the VSX register handling.
3725         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3726         (f32_sv): Likewise.
3727         (zero_extendsidi2_lfiwzx): Add support for loading into the
3728         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3729         only do VSX memory options on Altivec registers.
3730         (extendsidi2_lfiwax): Likewise.
3731         (extendsfdf2_fpr): Likewise.
3732         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3733         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3734         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3735         (movdi_internal64): Likewise.
3737 2013-09-27  Xinliang David Li  <davidxl@google.com>
3739         * opts.c (finish_options): Adjust parameters
3740         according to vect cost model.
3741         (common_handle_option): Set dynamic vect cost
3742         model for FDO.
3743         targhooks.c (default_add_stmt_cost): Compute stmt cost
3744         unconditionally.
3745         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3746         Use helper function.
3747         * tree-vectorizer.h (unlimited_cost_model): New function.
3748         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3749         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3750         function.
3751         (vect_enhance_data_refs_alignment): Ditto.
3752         * flag-types.h: New enum.
3753         * common/config/i386/i386-common.c (ix86_option_init_struct):
3754         No need to initialize vect_cost_model flag.
3755         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3756         unconditionally.
3758 2013-09-27  Diego Novillo  <dnovillo@google.com>
3760         * gimple.h (enum ssa_mode): Remove.
3762 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3764         * cfgloop.h (number_of_loops): Fix typo in check for null.
3766 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3768         PR middle-end/58551
3769         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3770         are children of outermost saved_cfun's loop, and set it up to
3771         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3772         if loop != loop0 and SESE region contains bbs that belong to loop0.
3774 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3776         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3777         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3778         (set_address_segment, set_address_base, set_address_index)
3779         (set_address_disp): Accept the argument unconditionally.
3780         (baseness): Remove must_be_base_p and must_be_index_p checks.
3781         (decompose_normal_address): Classify as much as possible in the
3782         main loop.
3784 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3786         * cse.c (count_reg_usage): Handle INT_LIST.
3787         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3788         * reginfo.c (reg_scan_mark_refs): Likewise.
3789         * reload1.c (eliminate_regs_1): Likewise.
3791 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3793         PR middle-end/58547
3794         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3795         signed.
3797 2013-09-27  Richard Biener  <rguenther@suse.de>
3799         PR tree-optimization/58459
3800         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3801         restriction not propagating into loops.
3803 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3805         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3806         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3807         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3809 2013-09-26  Richard Biener  <rguenther@suse.de>
3811         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3813 2013-09-26  Richard Biener  <rguenther@suse.de>
3815         * alias.h (component_uses_parent_alias_set): Rename to ...
3816         (component_uses_parent_alias_set_from): ... this.
3817         * alias.c (component_uses_parent_alias_set): Rename to ...
3818         (component_uses_parent_alias_set_from): ... this and return
3819         the desired parent.
3820         (reference_alias_ptr_type_1): Use the result from
3821         component_uses_parent_alias_set_from instead of stripping
3822         components one at a time.
3823         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3825 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3827         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3828         Move prototypes to...
3829         * tree-ssa-ter.h: New File.  Move prototypes here.
3830         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3831         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3832         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3833         from is_replaceable_p.
3834         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3835         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3836         (process_replaceable): Use ter_is_replaceable_p.
3837         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3838         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3839         newly refactored ssa_is_replaceable_p.
3840         * cfgexpand.c: Include tree-outof-ssa.h.
3841         * ssaexpand.h: Delete.
3843 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3845         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3846         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3847         tree-ssa.c
3848         (create_gimple_tmp): Delete.
3849         (get_expr_type, build_assign, build_type_cast): Move to...
3850         * gimple-builder.c: New File.
3851         (get_expr_type): Relocate from gimple.c.
3852         (build_assign, build_type_cast): Change to only create ssanames.
3853         * gimple.h: Move prototypes to...
3854         * gimple-builder.h: New File. Here.
3855         * tree-ssa.h: And here.
3856         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3857         count_uses_and_derefs): Relocate from gimple.c.
3858         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3859         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3860         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3861         gimple_replace_ssa_lhs.
3862         * asan.c: Include gimple-builder.h.
3863         * Makefile.in: Add gimple-builder.o.
3865 2013-09-26  Richard Biener  <rguenther@suse.de>
3867         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3868         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3869         (loe_visit_block): Use gcc_checking_assert.
3870         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3871         gimple_assign_ssa_name_copy_p.
3872         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3873         change.
3875 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3877         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3878         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3880 2013-09-26  Richard Biener  <rguenther@suse.de>
3882         PR tree-optimization/58539
3883         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3884         the fact that debug statements are not taking part in loop-closed
3885         SSA construction.
3887 2013-09-26  Nick Clifton  <nickc@redhat.com>
3889         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3890         time warning message.
3891         (msp430_print_operand_raw): Delete unused letter parameter.
3892         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3893         (msp430_print_operand_address): New function.
3894         (msp430_print_operand): Move address printing code from here to
3895         new function.
3896         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3897         assembler.
3898         (zero_extendpsisi2): Likewise.
3899         (extendpsisi2): New pattern.
3900         (andneghi3): New pattern.
3902 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3904         * config/aarch64/aarch64.opt (mlra): New option.
3905         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3906         (TARGET_LRA_P): Define.
3908 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3910         * expr.c (expand_assignment): Remove obsolete comment.
3912 2013-09-25  Jeff Law  <law@redhat.com>
3914         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3915         tree-ssa-threadupdate.h.
3916         (register_jump_thread): Similarly.
3917         * tree-ssa-threadupdate.h: New header file.
3918         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3919         * tree-vrp.c: Likewise.
3920         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3921         (thread_around_empty_blocks): Change type of path vector argument to
3922         an edge,type pair from just an edge.  Initialize both elements when
3923         appending to a jump threading path.  Tweak references to elements
3924         appropriately.
3925         (thread_across_edge): Similarly.  Release memory for the elements
3926         as needed.
3927         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3928         (dump_jump_thread_path): New function broken out from
3929         register_jump_thread.
3930         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3931         path vector entries.  Search the path for NULL edges and dump
3932         the path if one is found.  Tweak the conversion of path to 3-edge
3933         form to use the block copy type information embedded in the path.
3935 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3937         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3939 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3940             Vladimir Makarov  <vmakarov@redhat.com>
3942         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3943         from the least significant bit.
3944         (strip_address_mutations): Add bitfield operations handling.
3945         (must_be_index_p): Add shifting and rotate operations handling.
3946         (set_address_base): Use must_be_base_p predicate.
3947         (set_address_index): Use must_be_index_p predicate.
3949 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3950             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3951             Sergey Lega  <sergey.s.lega@intel.com>
3952             Anna Tikhonova  <anna.tikhonova@intel.com>
3953             Ilya Tocar  <ilya.tocar@intel.com>
3954             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3955             Ilya Verbin  <ilya.verbin@intel.com>
3956             Kirill Yukhin  <kirill.yukhin@intel.com>
3957             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3959         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
3960         Use new names.
3961         (ix86_expand_vector_move_misalign): Support new unaligned load and
3962         stores and use new names.
3963         (CODE_FOR_sse2_storedqu): Rename to ...
3964         (CODE_FOR_sse2_storedquv16qi): ... this.
3965         (CODE_FOR_sse2_loaddqu): Rename to ...
3966         (CODE_FOR_sse2_loaddquv16qi): ... this.
3967         (CODE_FOR_avx_loaddqu256): Rename to ...
3968         (CODE_FOR_avx_loaddquv32qi): ... this.
3969         (CODE_FOR_avx_storedqu256): Rename to ...
3970         (CODE_FOR_avx_storedquv32qi): ... this.
3971         * config/i386/i386.md (fpint_logic): New.
3972         * config/i386/sse.md (VMOVE): Extend for AVX512.
3973         (VF): Ditto.
3974         (VF_128_256): New.
3975         (VF_512): Ditto.
3976         (VI_UNALIGNED_LOADSTORE): Ditto.
3977         (sse2_avx_avx512f): Ditto.
3978         (sse2_avx2): Extend for AVX512.
3979         (sse4_1_avx2): Ditto.
3980         (avx2_avx512f): New.
3981         (sse): Extend for AVX512.
3982         (sse2): Ditto.
3983         (sse4_1): Ditto.
3984         (avxsizesuffix): Ditto.
3985         (sseintvecmode): Ditto.
3986         (ssePSmode): Ditto.
3987         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
3988         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
3989         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3990         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
3991         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3992         (<sse2_avx_avx512f>_storedqu<mode): ... this.
3993         (<sse>_movnt<mode>): Replace constraint "x" with "v".
3994         (STORENT_MODE): Extend for AVX512.
3995         (*absneg<mode>2): Replace constraint "x" with "v".
3996         (*mul<mode>3): Ditto.
3997         (*ieee_smin<mode>3): Ditto.
3998         (*ieee_smax<mode>3): Ditto.
3999         (avx_cmp<mode>3): Replace VF with VF_128_256.
4000         (*<sse>_maskcmp<mode>3_comm): Ditto.
4001         (<sse>_maskcmp<mode>3): Ditto.
4002         (<sse>_andnot<mode>3): Extend for AVX512.
4003         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4004         (<code><mode>3, fpint_logic): New.
4005         (*<code><mode>3): Extend for AVX512.
4006         (avx512flogicsuff): New.
4007         (avx512f_<logic><mode>): Ditto.
4008         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4009         VF_128_256.
4010         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4011         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4012         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4013         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4014         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4015         (xop_vpermil2<mode>3): Ditto.
4016         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4017         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4018         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4019         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4021 2013-09-25  Tom Tromey  <tromey@redhat.com>
4023         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4024         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4025         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4026         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4027         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4028         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4029         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4030         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4031         (GRAPHITE_HTAB_H): Remove.
4033 2013-09-25  Tom Tromey  <tromey@redhat.com>
4035         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4037 2013-09-25  Tom Tromey  <tromey@redhat.com>
4039         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4041 2013-09-25  Tom Tromey  <tromey@redhat.com>
4043         * config/i386/t-i386 (i386.o): Remove.
4044         (i386-c.o): Use COMPILE and POSTCOMPILE.
4046 2013-09-25  Tom Tromey  <tromey@redhat.com>
4048         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4050 2013-09-25  Tom Tromey  <tromey@redhat.com>
4052         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4053         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4054         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4055         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4056         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4057         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4058         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4059         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4060         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4061         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4062         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4063         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4064         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4065         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4066         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4067         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4068         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4069         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4070         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4071         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4072         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4073         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4074         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4075         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4076         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4077         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4078         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4079         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4080         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4081         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4082         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4083         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4084         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4085         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4086         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4087         (graphite-clast-to-gimple.o, graphite-dependences.o)
4088         (graphite-interchange.o, graphite-poly.o)
4089         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4090         (graphite-optimize-isl.o, tree-vect-loop.o)
4091         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4092         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4093         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4094         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4095         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4096         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4097         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4098         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4099         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4100         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4101         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4102         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4103         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4104         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4105         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4106         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4107         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4108         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4109         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4110         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4111         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4112         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4113         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4114         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4115         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4116         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4117         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4118         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4119         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4120         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4121         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4122         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4123         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4124         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4125         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4126         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4127         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4128         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4129         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4130         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4131         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4132         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4133         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4134         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4135         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4136         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4137         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4138         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4139         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4140         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4141         (gcov-dump.o): Remove.
4142         (default-c.o): Use COMPILE and POSTCOMPILE.
4143         (CFLAGS-gcc.o): New variable.
4144         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4146 2013-09-25  Tom Tromey  <tromey@redhat.com>
4148         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4149         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4150         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4151         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4152         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4153         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4154         (c-family/array-notation-common.o, c-family/stub-objc.o)
4155         (c-family/c-ubsan.o): Remove.
4157 2013-09-25  Tom Tromey  <tromey@redhat.com>
4159         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4161 2013-09-25  Tom Tromey  <tromey@redhat.com>
4163         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4164         to add -DENABLE_SHARED_LIBGCC.
4165         (gcc.o): Don't use subshell.
4167 2013-09-25  Tom Tromey  <tromey@redhat.com>
4169         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4170         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4171         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4172         * configure, config.in: Rebuild.
4174 2013-09-25  Tom Tromey  <tromey@redhat.com>
4176         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4177         (COMPILE, POSTCOMPILE): New variables.
4178         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4179         (DEPFILES): New variable.
4180         Include ".Po" files.
4181         * configure.ac: Add checks for dependency checking.
4182         * configure, aclocal.m4: Regenerate.
4184 2013-09-25  Tom Tromey  <tromey@redhat.com>
4186         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4187         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4189 2013-09-25  Tom Tromey  <tromey@redhat.com>
4191         * Makefile.in (generated_files): Add options.h,
4192         target-hooks-def.h, insn-opinit.h,
4193         common/common-target-hooks-def.h, pass-instances.def,
4194         c-family/c-target-hooks-def.h.
4196 2013-09-25  Jeff Law  <law@redhat.com>
4198         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4199         than foo[foo.length () - 1] to access last member in a vec.
4200         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4202 2013-09-25  Richard Biener  <rguenther@suse.de>
4204         PR middle-end/58521
4205         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4207 2013-09-25  Jan Hubicka  <jh@suse.cz>
4209         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4210         test.
4212 2013-09-25  Marek Polacek  <polacek@redhat.com>
4214         PR sanitizer/58420
4215         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4216         when determining the type name.
4218 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4220         * config/sh/sh.md: Fix formatting.
4222 2013-09-24  Xinliang David Li  <davidxl@google.com>
4224         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4225         max peel iterations parameter.
4226         * param.def: New parameter.
4227         * doc/invoke.texi: Document New parameter.
4229 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4231         * gimple-pretty-print.c: Various whitespace tweaks.
4232         * tree-core.h: Likewise.
4233         * tree-pretty-print.c: Likewise.
4234         * tree-ssa-alias.c: Likewise.
4235         * tree-ssa-copy.c: Likewise.
4236         * tree-ssanames.c: Likewise.
4237         * tree-ssanames.h: Likewise.
4238         * tree-vrp.c: Likewise.
4240 2013-09-24  Alan Modra  <amodra@gmail.com>
4242         PR middle-end/57134
4243         PR middle-end/57586
4244         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4245         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4246         inout operands.
4248 2013-09-24  Richard Biener  <rguenther@suse.de>
4250         PR middle-end/58513
4251         * tree.c (reference_alias_ptr_type): Move ...
4252         * alias.c (reference_alias_ptr_type): ... here and implement
4253         in terms of the new reference_alias_ptr_type_1.
4254         (ref_all_alias_ptr_type_p): New helper.
4255         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4256         use ref_all_alias_ptr_type_p.
4257         (get_deref_alias_set): Add flag_strict_aliasing check here.
4258         (reference_alias_ptr_type_1): New function, split out from ...
4259         (get_alias_set): ... here.
4260         (alias_ptr_types_compatible_p): New function.
4261         * alias.h (reference_alias_ptr_type): Declare.
4262         (alias_ptr_types_compatible_p): Likewise.
4263         * tree.h (reference_alias_ptr_type): Remove.
4264         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4265         to compare MEM_REF alias types.
4267 2013-09-24  Richard Biener  <rguenther@suse.de>
4269         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4271 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4273         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4274         reload helper function arrays into a single array reg_addr.
4275         (reload_fpr_gpr): Likewise.
4276         (reload_gpr_vsx): Likewise.
4277         (reload_vsx_gpr): Likewise.
4278         (struct rs6000_reg_addr): Likewise.
4279         (reg_addr): Likewise.
4280         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4281         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4282         (rs6000_init_hard_regno_mode_ok): Likewise.
4283         (rs6000_secondary_reload_direct_move): Likewise.
4284         (rs6000_secondary_reload): Likewise.
4286         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4287         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4288         power8 changes.  Put wg constraint in alphabetical order.
4290         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4291         for future work to add ISA 2.07 VSX single precision support.
4292         (-mvsx-scalar-double): Change default from -1 to 1, update
4293         documentation comment.
4294         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4295         (-mupper-regs-df): New debug switch to control whether DF values
4296         can go in the traditional Altivec registers.
4297         (-mupper-regs-sf): New debug switch to control whether SF values
4298         can go in the traditional Altivec registers.
4300         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4301         and wy constraints.
4302         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4303         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4304         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4305         (rs6000_debug_legitimate_address_p): Print if we are running
4306         before, during, or after reload.
4307         (rs6000_secondary_reload): Add a comment.
4308         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4310         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4311         constraints.  Update documentation string.
4312         (wd constraint): Likewise.
4313         (wf constraint): Likewise.
4314         (wg constraint): Likewise.
4315         (wn constraint): Likewise.
4316         (ws constraint): Likewise.
4317         (wt constraint): Likewise.
4318         (wx constraint): Likewise.
4319         (wz constraint): Likewise.
4320         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4321         instructions.
4322         (ww constraint): Likewise.
4323         (wy constraint): Likewise.
4324         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4325         in the previous submissions.
4326         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4328 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4330         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4331         (REG_BR_PROB): Say that the probability is stored in an int_list.
4332         * reg-notes.def: Update commentary to mention INT_LIST.
4333         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4334         (INT_LIST): New rtx.
4335         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4336         * rtlanal.c (int_reg_note_p): New function.
4337         (alloc_reg_note): Assert that the note does not have an int argument.
4338         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4339         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4340         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4341         rather than an INSN_LIST.  Handle INT_LIST.
4342         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4343         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4344         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4345         Manipulate them as ints rather than rtxes.
4346         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4347         * regmove.c (copy_src_to_dest): Likewise.
4348         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4350         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4351         into the cases that need it.
4352         * config/arm/arm.c (arm_unwind_emit): Likewise.
4354         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4355         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4356         * loop-doloop.c (add_test, doloop_modify): Likewise.
4357         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4358         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4359         * predict.c (combine_predictions_for_insn): Likewise.
4360         * print-rtl.c (print_rtx): Handle INT_LIST.
4361         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4362         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4363         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4364         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4365         (ix86_split_fp_branch, predict_jump): Likewise.
4366         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4367         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4368         * config/spu/spu.c (ea_load_store_inline): Likewise.
4370         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4371         value of a REG_BR_PROB note.
4372         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4373         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4374         * emit-rtl.c (try_split): Likewise.
4375         * predict.c (br_prob_note_reliable_p): Likewise.
4376         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4377         * reorg.c (mostly_true_jump): Likewise.
4378         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4379         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4380         * config/i386/i386.c (ix86_print_operand): Likewise.
4381         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4382         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4383         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4384         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4385         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4386         * config/sparc/sparc.c (output_cbranch): Likewise.
4387         * config/spu/spu.c (get_branch_target): Likewise.
4388         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4389         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4391 2013-09-23  Jan Hubicka  <jh@suse.cz>
4393         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4394         for ipa-devirt.
4395         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4396         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4397         of external calls
4398         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4399         (gimple_fold_call): Dump inconsistent devirtualizations; add
4400         sanity check for type based devirtualizations.
4401         * ipa-prop.c: Include ipa-utils.h
4402         (ipa_intraprocedural_devirtualization): Add sanity check.
4403         (try_make_edge_direct_virtual_call): Likewise.
4405 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4407         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4408         assignment statements.
4410 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4412         * gimple-pretty-print.c (dump_ssaname_info): New function.
4413         (dump_gimple_phi): Call it.
4414         (pp_gimple_stmt_1): Likewise.
4415         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4416         (range_info_def): Declare.
4417         * tree-pretty-print.c (pp_double_int): New function.
4418         (dump_generic_node): Call it.
4419         * tree-pretty-print.h (pp_double_int): Declare.
4420         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4421         * tree-ssanames.h (range_info_def): New structure.
4422         (value_range_type): Move definition here.
4423         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4424         Declare.
4425         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4426         initialization.
4427         (set_range_info): New function.
4428         (get_range_info): Likewise.
4429         (duplicate_ssa_name_range_info): Likewise.
4430         (duplicate_ssa_name_fn): Check pointer type and call
4431         duplicate_ssa_name_range_info.
4432         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4433         * tree-vrp.c (value_range_type): Remove definition, now in
4434         tree-ssanames.h.
4435         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4436         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4437         (SSA_NAME_RANGE_INFO): New macro.
4439 2013-09-23  Richard Biener  <rguenther@suse.de>
4441         PR tree-optimization/58464
4442         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4443         (phi_trans_add): Change to add conditionally on being not present.
4444         (phi_translate_1): Remove recursion detection here.
4445         (phi_translate): Pre-seed the cache with NULL to catch
4446         recursion here in a more generic way.
4447         (bitmap_find_leader): Adjust comment.
4448         (get_representative_for): Dump value-numbers.
4449         (create_expression_by_pieces): Likewise.
4450         (insert_into_preds_of_block): Likewise.
4452 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4454         PR target/58475
4455         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4456         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4458 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4460         Revert r202780:
4461         2013-09-20  Renlin Li  <renlin.li@arm.com>
4463         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4464         plus_constant.
4465         (aarch64_expand_epilogue): Likewise.
4466         (aarch64_legitimize_reload_address): Likewise.
4468 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4470         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4471         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4473 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4475         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4476         comparison operators when -fno-trapping-math is in effect.
4477         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4478         comparison operators in TFmode and assert that unsupported operators
4479         cannot reach here.
4480         (ia64_print_operand): Likewise.
4482 2013-09-21  Jan Hubicka  <jh@suse.cz>
4484         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4485         (sse_typeless_stores): Enable for core
4486         (sse_load0_by_pxor): Likewise.
4487         (four_jump_limit): Disable for core.
4488         (pad_returns): Likewise.
4489         (avoid_vector_decode): Likewise.
4490         (fuse_cmp_and_branch): Enable for cores.
4491         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4493 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4495         PR middle-end/56791
4496         * config/pa/pa.c (pa_option_override): Disable auto increment and
4497         decrement instructions until reload is completed.
4499         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4500         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4501         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4503 2013-09-20  DJ Delorie  <dj@redhat.com>
4504             Nick Clifton  <nickc@redhat.com>
4506         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4507         (need_to_save): Save bank 0 on interrupts.
4508         (characterize_address): Strip far address wrappers.
4509         (rl78_as_legitimate_address): Likewise.
4510         (transcode_memory_rtx): Likewise.
4511         (rl78_peep_movhi_p): Disable this peephole after devirt.
4512         (rl78_propogate_register_origins): Forget all origins when a
4513         CLOBBER is seen.
4514         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4515         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4516         peephole2's.
4517         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4518         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4519         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4520         relocating.
4521         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4523 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4525         * config/pa/pa.md: In "scc" insn patterns, change output template to
4526         handle const0_rtx in reg_or_0_operand operands.
4528 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4530         PR target/56875
4531         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4532         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4534 2013-09-20  Richard Biener  <rguenther@suse.de>
4536         PR middle-end/58484
4537         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4538         remembering SSA name version and block index.
4539         (new_scev_info_str): Adjust.
4540         (hash_scev_info): Likewise.  Also hash the block index.
4541         (eq_scev_info): Adjust.
4542         (find_var_scev_info): Likewise.
4543         (struct instantiate_cache_entry): Remove.
4544         (struct instantiate_cache_type): Use a htab to map name, block
4545         to chrec.
4546         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4547         (get_instantiated_value_entry): Likewise.
4548         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4549         (instantiate_scev_name): Adjust.
4551 2013-09-20  Jeff Law  <law@redhat.com>
4553         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4555 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4557         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4558         Call aarch64_simd_expand_args to update op[argc].
4560 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4562         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4563         plugin argument.
4565 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4567         * gengtype.c (file_rules): Added rule for *.cc files.
4568         (get_output_file_with_visibility): Give fatal message when no
4569         rules found.
4571 2013-09-20  Renlin Li  <renlin.li@arm.com>
4573         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4574         (aarch64_expand_epilogue): Likewise.
4575         (aarch64_legitimize_reload_address): Likewise.
4577 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4579         PR middle-end/57748
4580         * expr.c (expand_assignment): Remove misalignp code path.
4582 2013-09-20  Marek Polacek  <polacek@redhat.com>
4584         PR sanitizer/58413
4585         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4586         TYPE_PRECISION.  Add asserts.
4588 2013-09-20  Richard Biener  <rguenther@suse.de>
4590         PR tree-optimization/58453
4591         * tree-loop-distribution.c (distribute_loop): Apply the cost
4592         model for -ftree-loop-distribute-patterns, too.
4594 2013-09-20  Richard Biener  <rguenther@suse.de>
4596         PR middle-end/58473
4597         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4598         make type comparison less strict.
4600 2013-09-20  Alan Modra  <amodra@gmail.com>
4602         * configure: Regenerate.
4603         * aclocal.m4: Regenerate.
4605 2013-09-20  Marek Polacek  <polacek@redhat.com>
4607         PR other/58467
4608         * doc/extend.texi: Document that attribute used is meant to be used
4609         on variables with static storage duration.
4611 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4613         PR tree-optimization/58472
4614         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4615         simd_lane_access set inv_p = false.
4616         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4617         the simduid magic VAR_DECL.
4619 2013-09-19  Jan Hubicka  <jh@suse.cz>
4621         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4623 2013-09-17  Jeff Law  <law@redhat.com>
4625         * tree-ssa-dom.c (record_temporary_equivalences): New function
4626         split out of dom_opt_dom_walker::after_dom_children.
4627         (dom_opt_dom_walker::thread_across_edge): Move common code
4628         in here from dom_opt_dom_walker::after_dom_children.
4629         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4631 2013-09-19  Jan Hubicka  <jh@suse.cz>
4633         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4634         (TARGET_GENERIC): Use PROCESOR_GENERIC
4635         (enum processor_type): Unify generic32 and 64.
4636         * i386.md (cpu): Likewise.
4637         * x86-tune.def (use_leave): Enable for generic32.
4638         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4639         * athlon.md: Change generic64 to generic in all occurences.
4640         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4641         (ix86_target_macros_internal): Likewise.
4642         * driver-i386.c (host_detect_local_cpu): Likewise.
4643         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4644         to ..
4645         (generic_memcpy, generic_memset, generic_cost): This one.
4646         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4647         (m_GENERIC32, m_GENERIC64): Remove.
4648         (m_GENERIC): Turn into one flag.
4649         (processor_target): Unify generic tunnings.
4650         (ix86_option_override_internal): Replace generic32/64 by generic.
4651         (ix86_issue_rate): Likewise.
4652         (ix86_adjust_cost): Likewise.
4654 2013-09-19  Jan Hubicka  <jh@suse.cz>
4656         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4657         of speculative flag.
4659 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4661         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4662         GOMP_sections_start, even if !exit_reachable.
4664 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4666         * lra-constraints.c (need_for_all_save_p): Use macro
4667         HARD_REGNO_CALL_PART_CLOBBERED.
4668         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4669         set up pseudo conflict hard regs.
4671 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4673         PR target/58452
4674         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4675         operands.
4677 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4679         PR rtl-optimization/58438
4680         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4681         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4682         from previous subpasses.
4684 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4686         * arm.c (arm_get_frame_offsets): Validate architecture supports
4687         LDRD/STRD before accepting the tuning preference.
4688         (arm_expand_prologue): Likewise.
4689         (arm_expand_epilogue): Likewise.
4691 2013-09-18  Richard Biener  <rguenther@suse.de>
4693         PR tree-optimization/58417
4694         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4695         have chrecs with symbols defined in the loop as operands.
4696         (chrec_fold_multiply): Likewise.
4697         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4698         parameters before folding binary operations.
4699         (struct instantiate_cache_entry_hasher): Remove.
4700         (struct instantiate_cache_type): Use a pointer-map.
4701         (instantiate_cache_type::instantiate_cache_type): New function.
4702         (instantiate_cache_type::get): Likewise.
4703         (instantiate_cache_type::set): Likewise.
4704         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4705         (get_instantiated_value_entry): Likewise.
4706         (global_cache): New global.
4707         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4708         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4709         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4710         (instantiate_scev_name): Adjust.
4711         (instantiate_scev): Construct global instead of local cache.
4712         (resolve_mixers): Likewise.
4714 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4715             Paolo Carlini  <paolo.carlini@oracle.com>
4717         PR c++/58458
4718         * doc/implement-cxx.texi: Fix references to the C++ standards.
4720 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4722         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4723         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4724         simd_lane_access drs, update also DR_ALIGNED_TO.
4726 2013-09-18  Marek Polacek  <polacek@redhat.com>
4728         PR sanitizer/58411
4729         * doc/extend.texi: Document no_sanitize_undefined attribute.
4730         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4731         no_sanitize_undefined attribute.
4733 2013-09-18  Nick Clifton  <nickc@redhat.com>
4735         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4736         (ASM_DECLARE_FUNCTION_NAME): Define.
4738 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4740         * compare-elim.c (find_comparison_dom_walker): New class
4741         (find_comparisons_in_bb): Rename to
4742         find_comparison_dom_walker::before_dom_children
4743         (find_comparisons): Adjust
4744         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4745         adjust.
4746         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4747         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4748         (init_walk_dominator_tree): Remove declaration.
4749         (fini_walk_dominator_tree): Remove declaration.
4750         * fwprop.c (single_def_use_dom_walker): New class
4751         (single_def_use_enter_block): Convert to
4752         single_def_use_dom_walker::before_dom_children.
4753         (single_def_use_leave_block): Convert to
4754         single_def_use_dom_walker::after_dom_children.
4755         (build_single_def_use_links): Adjust.
4756         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4757         class.
4758         (find_candidates_in_block): Convert to
4759         find_candidates_dom_walker::before_dom_children.
4760         (execute_strength_reduction): Adjust.
4761         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4762         (sese_dom_walker): New class.
4763         (sese_dom_walker::sese_dom_walker): New constructor.
4764         (sese_dom_walker::~sese_dom_walker): New destructor.
4765         (build_sese_conditions_before): Convert to
4766         sese_dom_walker::before_dom_children.
4767         (build_sese_conditions_after): Convert to
4768         sese_dom_walker::after_dom_children.
4769         (build_poly_scop): Adjust
4770         * tree-into-ssa.c (rewrite_dom_walker): New class
4771         (rewrite_enter_block): Convert to
4772         rewrite_dom_walker::before_dom_children.
4773         (rewrite_leave_block): Convert to
4774         rewrite_dom_walker::after_dom_children.
4775         (rewrite_update_dom_walker): New class.
4776         (rewrite_update_enter_block): Convert to
4777         rewrite_update_dom_walker::before_dom_children.
4778         (rewrite_update_leave_block): Convert to
4779         rewrite_update_dom_walker::after_dom_children.
4780         (rewrite_blocks, rewrite_into_ssa): Adjust.
4781         (mark_def_dom_walker): New class.
4782         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4783         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4784         (mark_def_sites_blocks): Convert to
4785         mark_def_dom_walker::before_dom_children.
4786         (mark_def_site_blocks): Remove.
4787         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4788         (tree_ssa_dominator_optimize): Adjust.
4789         (dom_thread_across_edge): Convert to method
4790         dom_opt_dom_walker::thread_across_edge.
4791         (dom_opt_enter_block): Convert to member function
4792         dom_opt_dom_walker::before_dom_children.
4793         (dom_opt_leave_block): Convert to member function
4794         dom_opt_dom_walker::after_dom_children.
4795         * tree-ssa-dse.c (dse_dom_walker): New class.
4796         (dse_enter_block): Convert to member function
4797         dse_dom_walker::before_dom_children.
4798         (tree_ssa_dse): Adjust.
4799         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4800         (determine_invariantness_stmt): Convert to method
4801         invariantness_dom_walker::before_dom_children.
4802         (determine_invariantness): Remove
4803         (move_computations_dom_walker): New class.
4804         (move_computations_stmt): Convert to method
4805         move_computations_dom_walker::before_dom_children.
4806         (move_computations, tree_ssa_lim): Adjust.
4807         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4808         (nt_init_block): Make method
4809         notrappping_dom_walker::before_dom_children.
4810         (nt_fini_block): Make
4811         method nontrapping_dom_walker::after_dom_children.
4812         (get_non_trapping): Adjust.
4813         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4814         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4815         (eliminate_leave_block): Make method.
4816         eliminate_dom_walker::after_dom_children.
4817         (eliminate): Adjust
4818         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4819         (strlen_enter_block): Make method
4820         strlen_dom_walker::before_dom_children.
4821         (strlen_leave_block): Make
4822         method strlen_dom_walker::after_dom_children.
4823         (tree_ssa_strlen): Adjust.
4824         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4825         (tree_ssa_uncprop): Adjust.
4826         (uncprop_leave_block): Make method
4827         uncprop_dom_walker::after_dom_children.
4828         (uncprop_leave_block): Make method
4829         uncprop_dom_walker::before_dom_children.
4831 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4833         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4834         branch in same basic block.  Check both src and dest of the move insn.
4836 2013-09-17  Nick Clifton  <nickc@redhat.com>
4838         * config/rl78/rl78-real.md (bf): New pattern.
4839         (bt): New pattern.
4840         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4841         (rl78_print_operand): Do not put a # before a %B.
4842         * config/rl78/rl78.opt: Tweak doc strings.
4844 2013-09-17  DJ Delorie  <dj@redhat.com>
4846         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4847         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4848         (rl78_option_override): Likewise, if -mallregs.
4849         (is_virtual_register): Likewise.
4850         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4851         (REGNO_OK_FOR_BASE_P): Likewise.
4852         * config/rl78/rl78.opt (-mallregs): New.
4854 2013-09-17  Nick Clifton  <nickc@redhat.com>
4856         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4857         For interrupt functions: save all call clobbered registers if the
4858         interrupt handler is not a leaf function.
4859         (rl78_expand_prologue): Always recompute the frame information.
4860         For interrupt functions: only select bank 0 if one of the bank 0
4861         registers is going to be psuhed.
4863 2013-09-17  DJ Delorie  <dj@redhat.com>
4865         * config/rl78/constraints.md: For each W* constraint, rename to C*
4866         and create a W* constraint that checks for an optional ES: prefix
4867         pattern also.
4868         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4869         (es_addr): New.  Used to wrap far addresses.
4870         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4871         (rl78_es_base): New.
4872         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4873         wrapped far addresses.
4874         (rl78_print_operand_1): Unwrap far addresses before processing.
4875         (rl78_lo16): Wrap far addresses in unspecs.
4876         (rl78_es_addr): New.
4877         (rl78_es_base): New.
4878         (insn_ok_now): Check for not-yet-wrapped far addresses.
4879         (transcode_memory_rtx): Properly re-wrap far addresses.
4881 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4883         * config/sparc/t-rtems: Add leon3 multilibs.
4885 2013-09-17  Cong Hou  <congh@google.com>
4887         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4888         when checking the dot production pattern. The type of rhs operand
4889         of multiply is now checked correctly.
4891 2013-09-17  Jeff Law  <law@redhat.com>
4893         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4894         edge implied equivalences into successor phis.
4895         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4896         here from tree-ssa-threadedge.c.
4897         (mark_threaded_blocks): When threading through a joiner, if both
4898         successors of the joiner's clone reach the same block, verify the
4899         PHI arguments are equal.  If not, cancel the jump threading request.
4900         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4901         tree-ssa-threadupdate.c
4902         (thread_across_edge): Don't check PHI argument equality when
4903         threading through joiner block here.
4905 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4907         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4908         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4909         (warn_uninit, warn_uninitialized_vars,
4910         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4911         Move to tree-ssa-uninit.c.
4912         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4913         (has_undefined_value_p): New.  Pass dependant parts of
4914         ssa_undefined_value_p.
4915         (uninit_undefined_value_p): Use has_undefined_value_p.
4916         (warn_uninit, warn_uninitialized_vars,
4917         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4918         Move from tree-ssa.c.
4919         * tree-ssa.h: Adjust prototypes.
4921 2013-09-17  Jan Hubicka  <jh@suse.cz>
4923         PR middle-end/58332
4924         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4925         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4926         FUNCTION_NOT_OPTIMIZED.
4927         * ipa-inline-analysis.c (compute_inline_parameters): Function
4928         not optimized is not inlinable unless it is alwaysinline.
4929         (inline_analyze_function): Force calls in not optimized
4930         function not inlinable.
4932 2013-09-17  Jan Hubicka  <jh@suse.cz>
4934         PR middle-end/58329
4935         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4936         to return NULL.
4937         * ipa.c (function_and_variable_visibility): Likewise.
4938         * ipa-profile.c (ipa_profile): Likewise.
4940 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4942         PR ipa/58398
4943         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4944         attribute, and don't inline the resolver in this case.
4946 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4948         * coverage.c (get_coverage_counts): Add missing newline.
4950 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4952         PR tree-optimization/58088
4953         * fold-const.c (mask_with_trailing_zeros): New function.
4954         (fold_binary_loc): Make sure we don't recurse infinitely
4955         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
4956         Use mask_with_trailing_zeros where appropriate.
4958 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
4960         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
4961         of 'prev' var to get better distance estimation.
4963 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
4965         * tree-inline.h (struct copy_body_data): Add transform_parameter.
4966         * tree-inline.c (is_parameter_of): New predicate.
4967         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
4968         a parameter has been remapped.
4969         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
4970         (optimize_inline_calls): Initialize transform_parameter.
4971         (copy_gimple_seq_and_replace_locals): Likewise.
4972         (tree_function_versioning): Likewise.
4973         (maybe_inline_call_in_expr): Likewise.
4975 2013-09-17  Nick Clifton  <nickc@redhat.com>
4977         * config/msp430/msp430-protos.h: Add prototypes for new functions.
4978         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
4979         interrupt handlers.
4980         (is_attr_func): New function.
4981         (msp430_is_interrupt_func): New function.
4982         (is_naked_func): New function.
4983         (is_reentrant_func): New function.
4984         (is_critical_func): New function.
4985         (msp430_start_function): Add annotations for function attributes.
4986         (msp430_attr): New function.
4987         (msp430_attribute_table): New.
4988         (msp430_function_section): New function.
4989         (TARGET_ASM_FUNCTION_SECTION): Define.
4990         (msp430_builtin): New enum.
4991         (msp430_init_builtins): New function.
4992         (msp430_builtin_devl): New function.
4993         (msp430_expand_builtin): New function.
4994         (TARGET_INIT_BUILTINS): Define.
4995         (TARGET_EXPAND_BUILTINS): Define.
4996         (TARGET_BUILTIN_DECL): Define.
4997         (msp430_expand_prologue): Add support for naked, interrupt,
4998         critical and reentrant functions.
4999         (msp430_expand_epilogue): Likewise.
5000         (msp430_print_operand): Handle 'O' character.
5001         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5002         NO_TRAMPOLINES.
5003         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5004         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5005         (pushm): Use a 'n' rather than an 'i' constraint.
5006         (msp_return): Add generation of the interrupt return instruction.
5007         (disable_interrupts): New pattern.
5008         (enable_interrupts): New pattern.
5009         (push_intr_state): New pattern.
5010         (pop_intr_state): New pattern.
5011         (bic_SR): New pattern.
5012         (bis_SR): New pattern.
5013         * doc/extend.texi: Document MSP430 function attributes and builtin
5014         functions.
5016 2013-09-17  Richard Biener  <rguenther@suse.de>
5018         PR tree-optimization/58432
5019         * tree-loop-distribution.c (tree_loop_distribution): Also
5020         scan PHIs for outside loop uses and seed a partition from them.
5022 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5024         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5025         (restructure_reference): Call backtrace_base_for_ref.
5027 2013-09-17  Alan Modra  <amodra@gmail.com>
5029         PR target/57589
5030         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5031         patch.
5033 2013-09-16  DJ Delorie  <dj@redhat.com>
5035         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5036         vregs location for RL78/G10.
5037         (rl78_expand_prologue): Avoid SEL on G10.
5038         (rl78_expand_epilogue): Likewise.
5039         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5040         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5041         __RL78_G10__ when appropriate.
5042         (ASM_SPEC): Pass -mg10 along to the assembler.
5043         * config/rl78/rl78.md (sel_rb): Disable for G10.
5044         * config/rl78/rl78.opt: Add -mg10 option.
5045         * config/rl78/t-rl78: Add -mg10 multilib.
5047 2013-09-16  Xinliang David Li  <davidxl@google.com>
5049         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5050         * omp-low.c (omp_max_vf): Ditto.
5051         (expand_omp_simd): Ditto.
5052         * tree-vectorizer.c (vectorize_loops): Ditto.
5053         (gate_vect_slp): Ditto.
5054         (gate_increase_alignment): Ditto.
5055         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5056         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5057         (gate_tree_vectorize): Name change.
5058         (tree_vectorize): Ditto.
5059         (pass_vectorize::gate): Call new function.
5060         (pass_vectorize::execute): Ditto.
5061         * opts.c: O3 default setting change.
5062         (finish_options): Check new flag.
5063         * doc/invoke.texi: Document new flags.
5064         * common.opt: New flags.
5066 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5068         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5069         (Condition Code): Don't reference it.
5070         * doc/tm.texi: Regenerate.
5072 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5074         PR middle-end/58418
5075         * lra-constraints.c (undo_optional_reloads): Consider all optional
5076         reload even if it did not get a hard reg.
5078 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5080         * dumpfile.c (dump_loc): Remove newline emission.
5081         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5082         emission to dump_printf_loc calls where missing.
5083         (vect_mark_for_runtime_alias_test): Ditto.
5084         (vect_analyze_data_ref_dependence): Ditto.
5085         (vect_analyze_data_ref_dependences): Ditto.
5086         (vect_slp_analyze_data_ref_dependence): Ditto.
5087         (vect_slp_analyze_data_ref_dependences): Ditto.
5088         (vect_compute_data_ref_alignment): Ditto.
5089         (vect_update_misalignment_for_peel): Ditto.
5090         (vect_verify_datarefs_alignment): Ditto.
5091         (vector_alignment_reachable_p): Ditto.
5092         (vect_get_data_access_cost): Ditto.
5093         (vect_enhance_data_refs_alignment): Ditto.
5094         (vect_find_same_alignment_drs): Ditto.
5095         (vect_analyze_data_refs_alignment): Ditto.
5096         (vect_analyze_group_access): Ditto.
5097         (vect_analyze_data_ref_access): Ditto.
5098         (vect_analyze_data_ref_accesses): Ditto.
5099         (vect_prune_runtime_alias_test_list): Ditto.
5100         (vect_analyze_data_refs): Ditto.
5101         (vect_create_addr_base_for_vector_ref): Ditto.
5102         (vect_create_data_ref_ptr): Ditto.
5103         (vect_grouped_store_supported): Ditto.
5104         (vect_grouped_load_supported): Ditto.
5105         * value-prof.c (check_counter): Ditto.
5106         (check_ic_target): Ditto.
5107         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5108         (vect_recog_widen_mult_pattern): Ditto.
5109         (vect_recog_widen_sum_pattern): Ditto.
5110         (vect_recog_over_widening_pattern): Ditto.
5111         (vect_recog_widen_shift_pattern): Ditto.
5112         (vect_recog_rotate_pattern): Ditto.
5113         (vect_recog_vector_vector_shift_pattern): Ditto.
5114         (vect_recog_divmod_pattern): Ditto.
5115         (vect_recog_mixed_size_cond_pattern): Ditto.
5116         (vect_recog_bool_pattern): Ditto.
5117         (vect_pattern_recog_1): Ditto.
5118         (vect_pattern_recog): Ditto.
5119         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5120         (vect_is_simple_iv_evolution): Ditto.
5121         (vect_analyze_scalar_cycles_1): Ditto.
5122         (vect_get_loop_niters): Ditto.
5123         (vect_analyze_loop_1): Ditto.
5124         (vect_analyze_loop_form): Ditto.
5125         (vect_analyze_loop_operations): Ditto.
5126         (vect_analyze_loop_2): Ditto.
5127         (vect_analyze_loop): Ditto.
5128         (report_vect_op): Ditto.
5129         (vect_is_slp_reduction): Ditto.
5130         (vect_is_simple_reduction_1): Ditto.
5131         (vect_get_known_peeling_cost): Ditto.
5132         (vect_estimate_min_profitable_iters): Ditto.
5133         (vect_model_reduction_cost): Ditto.
5134         (vect_model_induction_cost): Ditto.
5135         (get_initial_def_for_induction): Ditto.
5136         (vect_create_epilog_for_reduction): Ditto.
5137         (vectorizable_reduction): Ditto.
5138         (vectorizable_induction): Ditto.
5139         (vectorizable_live_operation): Ditto.
5140         (vect_loop_kill_debug_uses): Ditto.
5141         (vect_transform_loop): Ditto.
5142         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5143         (vect_stmt_relevant_p): Ditto.
5144         (process_use): Ditto.
5145         (vect_mark_stmts_to_be_vectorized): Ditto.
5146         (vect_model_simple_cost): Ditto.
5147         (vect_model_promotion_demotion_cost): Ditto.
5148         (vect_model_store_cost): Ditto.
5149         (vect_get_store_cost): Ditto.
5150         (vect_model_load_cost): Ditto.
5151         (vect_get_load_cost): Ditto.
5152         (vect_init_vector_1): Ditto.
5153         (vect_get_vec_def_for_operand): Ditto.
5154         (vect_finish_stmt_generation): Ditto.
5155         (vectorizable_call): Ditto.
5156         (vectorizable_conversion): Ditto.
5157         (vectorizable_assignment): Ditto.
5158         (vectorizable_shift): Ditto.
5159         (vectorizable_operation): Ditto.
5160         (vectorizable_store): Ditto.
5161         (vectorizable_load): Ditto.
5162         (vectorizable_condition): Ditto.
5163         (vect_analyze_stmt): Ditto.
5164         (vect_transform_stmt): Ditto.
5165         (vect_is_simple_use): Ditto.
5166         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5167         (vect_can_advance_ivs_p): Ditto.
5168         (vect_update_ivs_after_vectorizer): Ditto.
5169         (vect_do_peeling_for_loop_bound): Ditto.
5170         (vect_gen_niters_for_prolog_loop): Ditto.
5171         (vect_update_inits_of_drs): Ditto.
5172         (vect_create_cond_for_alias_checks): Ditto.
5173         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5174         (vect_build_slp_tree_1): Ditto.
5175         (vect_supported_load_permutation_p): Ditto.
5176         (vect_analyze_slp_instance): Ditto.
5177         (vect_analyze_slp): Ditto.
5178         (vect_make_slp_decision): Ditto.
5179         (vect_detect_hybrid_slp): Ditto.
5180         (vect_bb_vectorization_profitable_p): Ditto.
5181         (vect_slp_analyze_bb_1): Ditto.
5182         (vect_update_slp_costs_according_to_vf): Ditto.
5183         (vect_get_mask_element): Ditto.
5184         (vect_transform_slp_perm_load): Ditto.
5185         (vect_schedule_slp_instance): Ditto.
5186         (vect_schedule_slp): Ditto.
5187         (vect_slp_transform_bb): Ditto.
5188         * profile.c (read_profile_edge_counts): Ditto.
5189         (compute_branch_probabilities): Ditto.
5190         * coverage.c (get_coverage_counts): Ditto.
5192 2013-09-16  Diego Novillo  <dnovillo@google.com>
5194         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5196 2013-09-16  Jan Hubicka  <jh@suse.cz>
5198         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5199         abstract functions; for static functions check the presence of body.
5201 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5203         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5204         * config/aarch64/aarch64-simd.md
5205         (aarch64_mla_elt<mode>): New.
5206         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5207         (aarch64_mls_elt<mode>): Likewise.
5208         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5209         (aarch64_fma4_elt<mode>): Likewise.
5210         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5211         (aarch64_fma4_elt_to_128v2df): Likewise.
5212         (aarch64_fma4_elt_to_64df): Likewise.
5213         (fnma<mode>4): Likewise.
5214         (aarch64_fnma4_elt<mode>): Likewise.
5215         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5216         (aarch64_fnma4_elt_to_128v2df): Likewise.
5217         (aarch64_fnma4_elt_to_64df): Likewise.
5218         * config/aarch64/iterators.md (VDQSF): New.
5219         * config/aarch64/arm_neon.h
5220         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5221         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5223 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5225         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5226         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5227         (aarch64_mul3_elt_to_128df): Likewise.
5228         (aarch64_mul3_elt_to_64v2df): Likewise.
5229         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5230         (VMUL): New.
5231         (VMUL_CHANGE_NLANES) Likewise.
5232         (h_con): Likewise.
5233         (f): Likewise.
5234         * config/aarch64/arm_neon.h
5235         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5237 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5239         * config/aarch64/arm_neon.h
5240         (vcvtx_high_f32_f64): Fix parameters.
5242 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5243             Uros Bizjak  <ubizjak@gmail.com>
5245         * config/alpha.c: Include tree-ssanames.h.
5247 2013-09-16  Richard Biener  <rguenther@suse.de>
5249         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5250         (dot_rdg_1): Handle control_dd.
5251         (create_edge_for_control_dependence): New function.
5252         (create_rdg_edges): Add control dependences if asked for.
5253         (build_rdg): Likewise.
5254         (generate_loops_for_partition): If there are not necessary
5255         control stmts remove all their dependencies.
5256         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5257         (distribute_loop): Pass on control dependences.
5258         (tree_loop_distribution): Compute control dependences and remove
5259         restriction on number of loop nodes.
5261 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5263         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5264         for internal calls.
5266 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5268         * cse.c (try_const_anchors): Punt on CC modes.
5270 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5272         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5274 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5276         PR target/58382
5277         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5278         calls to word_mode.
5280 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5282         PR target/48094
5283         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5284         seen.
5285         (darwin_objc1_section): Likewise.
5286         (darwin_file_end): Emit Image Info section when required.
5288 2013-09-14  Jan Hubicka  <jh@suse.cz>
5290         * tree-into-ssa.c (gate_into_ssa): New.
5291         (pass_data_build_ssa): Use it.
5292         * cgraph.h (expand_thunk): Update prototype.
5293         * cgraphunit.c (analyze_function): Expand thunks early.
5294         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5295         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5296         set lowered flag; do not add new function.
5297         (assemble_thunks_and_aliases): Update.
5298         * tree-ssa.c (gate_init_datastructures): New gate.
5299         (pass_data_init_datastructures): Use it.
5301 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5303         PR target/58269
5304         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5305         xmm register set described in the psABI.
5307 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5309         * dwarf2out.c (should_emit_struct_debug): Add check
5310         for type_decl variable is not NULL.
5312 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5314         * config.gcc: Use new winnt-c.c target hooks
5315         * config/t-winnt: New file
5316         * config/winnt-c.c: New file
5317         * doc/tm.texi.in: Document new hook
5318         * doc/tm.texi: Regenerated
5320 2013-09-13  Jan Hubicka  <jh@suse.cz>
5322         PR middle-end/58094
5323         * ipa-inline.c (check_callers): New function.
5324         (check_caller_edge): Remove.
5325         (want_inline_function_to_all_callers_p): Also permit alises that are
5326         called dirrectly.
5327         (inline_to_all_callers): Terminate the walk when devirtualization
5328         introduce new calls.
5330 2013-09-13  Jan Hubicka  <jh@suse.cz>
5332         * ipa-inline-analysis.c (struct growth_data): Add node.
5333         (do_estimate_growth_1): Fix detection of recursion.
5335 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5337         PR tree-optimization/58392
5338         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5339         to avoid shadowing of outer loop variable.  If
5340         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5341         replace_by_duplicate_decl simduid of loops that have it set and
5342         set dest_cfun->has_simduid_loops and/or
5343         dest_cfun->has_force_vect_loops.
5344         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5345         instead of maybe_lookup_decl.
5346         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5347         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5348         call.  Copy over force_vect and copy and remap simduid.  Set
5349         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5350         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5351         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5352         caller.  Don't set cfun->has_simduid_loops and/or
5353         cfun->has_force_vect_loops here.
5354         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5355         caller.
5356         (expand_call_inline, tree_function_versioning): Adjust copy_body
5357         callers.
5359 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5361         PR bootstrap/58388
5362         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5363         the assert if the edge was a speculative one.
5365 2013-09-13  Richard Biener  <rguenther@suse.de>
5367         * tree-data-ref.h (known_dependences_p): Move here ...
5368         * tree-loop-distribution.c (known_dependences_p): ... from here.
5369         (dump_rdg_component, debug_rdg_component): Remove.
5370         (dump_rdg): Adjust.
5371         (generate_loops_for_partition): Use gimple_uid instead of
5372         relying on matching stmt visit order.
5373         (rdg_build_partitions): Take starting stmt vector.
5374         (ldist_gen): Merge into ...
5375         (distribute_loop): ... this function.  Do not compute starting
5376         vertices vector.
5377         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5379 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5381         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5382         Set type attribute correctly. Set predicable_short_it attribute.
5383         (cmpsi_shiftsi): Remove %? from output template.
5385 2013-09-13  Richard Biener  <rguenther@suse.de>
5387         * tree-loop-distribution.c (struct rdg_component,
5388         rdg_defs_used_in_other_loops_p, free_rdg_components,
5389         rdg_build_components): Remove.
5390         (stmts_from_loop): Do not record virtual PHIs.
5391         (generate_loops_for_partition): Skip virtual PHIs.
5392         (build_rdg_partition_for_component): Rename to ...
5393         (build_rdg_partition_for_vertex): ... this and adjust.
5394         (rdg_build_partitions): Take a vector of starting vertices
5395         instead of components.  Remove unnecessary leftover handling.
5396         (ldist_gen): Do not build components or record other stores.
5397         (distribute_loop): Do not distribute loops containing stmts
5398         with side-effects.
5400 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5402         PR target/58314
5403         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5405 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5407         * config.gcc: Separate cases for mingw and cygwin targets,
5408         and add 64-bit cygwin target case.
5410         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5411         dll-export inline-functions.
5412         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5415 2013-09-13  Jeff Law  <law@redhat.com>
5417         PR middle-end/58387
5418         Revert:
5419         2013-09-06  Jeff Law  <law@redhat.com>
5421         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5422         edge implied equivalences into successor phis.
5424 2013-09-12  DJ Delorie  <dj@redhat.com>
5426         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5427         separators.
5429 2013-09-12  Brooks Moses  <bmoses@google.com>
5431         PR driver/42955
5432         * Makefile.in: Do not install driver binaries in $(target)/bin.
5434 2013-09-12  DJ Delorie  <dj@redhat.com>
5436         * config/rl78/rl78.opt (mrelax): New.
5437         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5438         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5440         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5441         between SP and FP.
5442         (rl78_expand_epilogue): Likewise.
5444 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5446         PR middle-end/58335
5447         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5448         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5449         pointer elimination with using remove_reg_equal_offset_note.
5451 2013-09-12  DJ Delorie  <dj@redhat.com>
5453         * config/msp430/: New port.
5454         * config.gcc (msp430): Added.
5455         * doc/invoke.texi: Document MSP430 options.
5456         * doc/install.texi: Document msp430-elf
5457         * doc/md.texi: Document msp430-elf
5458         * doc/contrib.texi: Document msp430-elf
5460         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5461         PSImode.
5463 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5465         PR ipa/58389
5466         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5467         reference descriptor is NULL.
5468         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5469         edge to NULL.
5471 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5473         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5474         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5475         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5476         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5477         make_temp_ssa_name): move to tree-ssanames.h
5478         * tree-ssa-alias.h: Move prototype.
5479         * tree-ssa.h: Include tree-ssanames.h.
5480         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5481         * tree-ssanames.h: New.  Move items from tree-flow*.h
5482         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5484 2013-09-12  Richard Biener  <rguenther@suse.de>
5486         PR tree-optimization/58404
5487         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5488         propagate non-invariant addresses into dereferences wrapped
5489         in component references.
5491 2013-09-12  Richard Biener  <rguenther@suse.de>
5493         PR tree-optimization/58402
5494         * passes.def: Move pass_late_warn_uninitialized later.
5496 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5498         * tree-ssa.h: New.  Move content from tree-flow.h and
5499         tree-flow-inline.h.
5500         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5501         Move prototypes belonging to tree-ssa.c.
5502         * tree-flow-inline.h (redirect_edge_var_map_def,
5503         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5504         tree-ssa.h.
5505         * gimple.h: Adjust prototypes.
5506         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5507         to...
5508         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5509         * tree.h: Move prototype to tree-ssa.h.
5510         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5511         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5512         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5513         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5514         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5515         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5516         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5517         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5518         graphite-blocking.c, graphite-clast-to-gimple.c,
5519         graphite-dependences.c, graphite-interchange.c,
5520         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5521         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5522         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5523         ipa-reference.c, ipa-split.c, ipa-utils.c,
5524         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5525         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5526         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5527         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5528         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5529         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5530         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5531         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5532         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5533         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5534         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5535         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5536         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5537         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5538         value-prof.c, var-tracking.c,
5539         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5541 2013-09-12  Richard Biener  <rguenther@suse.de>
5543         PR tree-optimization/58396
5544         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5545         (build_rdg): Take a loop-nest parameter, fix memleaks.
5546         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5548 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5550         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5551         for SLM.
5553 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5555         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5556         The FMA instruction names should have a 'v' prefix.
5558 2013-09-12  Richard Biener  <rguenther@suse.de>
5560         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5561         (dot_rdg): Use popen instead of system in optional code.
5562         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5563         (already_processed_vertex_p): Adjust.
5564         (has_anti_or_output_dependence, predecessor_has_mem_write,
5565         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5566         rdg_flag_uses): Remove.
5567         (rdg_flag_vertex): Simplify.
5568         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5569         remove recursion.
5570         (build_rdg_partition_for_component): Process the first vertex
5571         of a component only.
5572         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5574 2013-09-12  Alan Modra  <amodra@gmail.com>
5576         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5578 2013-09-11  DJ Delorie  <dj@redhat.com>
5579             Nick Clifton  <nickc@redhat.com>
5581         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5582         (rl78_stack_based_mem): New.
5583         * config/rl78/constraints.md (Iv08): New.
5584         (Iv16): New.
5585         (Iv24): New.
5586         (Is09): New.
5587         (Is17): New.
5588         (Is25): New.
5589         (ISsi): New.
5590         (IShi): New.
5591         (ISqi): New.
5592         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5593         (movhi): Likewise.
5594         (movsi): Change from expand to insn-and-split.
5595         (ashrsi3): Clobber AX.
5596         (lshrsi3): New.
5597         (ashlsi3): New.
5598         (cbranchsi4): New.
5599         * config/rl78/rl78.md (CC_REG): Fix.
5600         (addsi3): Allow memory and immediate operands.
5601         (addsi3_internal): Split into...
5602         (addsi3_internal_virt): ...new, and ...
5603         (addsi3_internal_real): ...new.
5604         (subsi): New.
5605         (subsi3_internal_virt): New.
5606         (subsi3_internal_real): New.
5607         (mulsi3): Add memory operand.
5608         (mulsi3_rl78): Likewise.
5609         (mulsi3_g13): Likewise.
5610         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5611         (cbranchqi4_real): Add more constraint options.
5612         (cbranchhi4_real): Expand pattern.
5613         (cbranchhi4_real_signed): New.
5614         (cbranchhi4_real_inverted): New.
5615         (cbranchsi4_real_lt): New.
5616         (cbranchsi4_real_ge): New.
5617         (cbranchsi4_real_signed): New.
5618         (cbranchsi4_real): New.
5619         (peephole2): New.
5620         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5621         constant shifts.
5622         (lshrsi3_virt): Likewise.
5623         (ashlsi3_virt): Likewise.
5624         (cbranchqi4_virt_signed): New.
5625         (cbranchhi4_virt_signed): New.
5626         (cbranchsi4_virt): New.
5627         * config/rl78/rl78.c: Whitespace fixes throughout.
5628         (move_elim_pass): New.
5629         (pass_data_rl78_move_elim): New.
5630         (pass_rl78_move_elim): New.
5631         (make_pass_rl78_move_elim): New.
5632         (rl78_devirt_info): Run devirt earlier.
5633         (rl78_move_elim_info): New.
5634         (rl78_asm_file_start): Register it.
5635         (rl78_split_movsi): New.
5636         (rl78_as_legitimate_address): Allow virtual base registers when
5637         appropriate.
5638         (rl78_addr_space_convert): Remove spurious debug stuff.
5639         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5640         (rl78_print_operand): More cases for not printing '#'.
5641         (rl78_expand_compare): Remove most of the logic.
5642         (content_memory): New.
5643         (clear_content_memory): New.
5644         (get_content_index): New.
5645         (get_content_name): New.
5646         (display_content_memory): New.
5647         (update_content): New.
5648         (record_content): New.
5649         (already_contains): New.
5650         (insn_ok_now): Re-recog insns with virtual registers.
5651         (add_postponed_content_update): New.
5652         (process_postponed_content_update): New.
5653         (gen_and_emit_move): New.
5654         (transcode_memory_rtx): Record new location content.
5655         Use gen_and_emit_move.
5656         (force_into_acc): New.
5657         (move_to_acc): Use gen_and_emit_move.
5658         (move_from_acc): Likewise.
5659         (move_acc_to_reg): Likewise.
5660         (move_to_x): Likewise.
5661         (move_to_hl): Likewise.
5662         (move_to_de): Likewise.
5663         (rl78_alloc_physical_registers_op1): Record location content.
5664         (has_constraint): New.
5665         (rl78_alloc_physical_registers_op2): Record location content.
5666         Optimize use of HL.
5667         (rl78_alloc_physical_registers_ro1): Likewise.
5668         (rl78_alloc_physical_registers_cmp): Likewise.
5669         (rl78_alloc_physical_registers_umul): Likewise.
5670         (rl78_alloc_address_registers_macax): New.
5671         (rl78_alloc_physical_registers): Initialize and set location
5672         content memory as needed.
5673         (rl78_reorg): Make sure split2 is called.
5674         (rl78_rtx_costs): New.
5676 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5678         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5679         for (not (neg ...)) and (neg (not ...)) cases.
5681 2013-09-11  Richard Biener  <rguenther@suse.de>
5683         PR middle-end/58377
5684         * passes.def: Split critical edges before late uninit warning passes.
5685         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5687 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5689         PR tree-optimization/58385
5690         * fold-const.c (build_range_check): If both low and high are NULL,
5691         use omit_one_operand_loc to preserve exp side-effects.
5693 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5695         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5697 2013-09-11  Richard Biener  <rguenther@suse.de>
5699         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5700         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5701         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5702         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5703         stmts_from_loop, known_dependences_p, build_empty_rdg,
5704         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5705         * tree-loop-distribution.c: ... here.
5706         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5707         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5708         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5709         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5710         * tree-loop-distribution.c: ... here.
5711         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5712         (struct partition_s): Add loops member.
5713         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5714         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5715         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5717 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5718             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5719             Sergey Lega  <sergey.s.lega@intel.com>
5720             Anna Tikhonova  <anna.tikhonova@intel.com>
5721             Ilya Tocar  <ilya.tocar@intel.com>
5722             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5723             Ilya Verbin  <ilya.verbin@intel.com>
5724             Kirill Yukhin  <kirill.yukhin@intel.com>
5725             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5727         * config/i386/constraints.md (k): New.
5728         (Yk): Ditto.
5729         * config/i386/i386.c (const regclass_map): Add new mask registers.
5730         (dbx_register_map): Ditto.
5731         (dbx64_register_map): Ditto.
5732         (svr4_dbx_register_map): Ditto.
5733         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5734         disabled.
5735         (ix86_preferred_reload_class): Disable constants for mask registers.
5736         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5737         (ix86_hard_regno_mode_ok): Support new mask registers.
5738         (x86_order_regs_for_local_alloc): Ditto.
5739         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5740         (FIXED_REGISTERS): Add new mask registers.
5741         (CALL_USED_REGISTERS): Ditto.
5742         (REG_ALLOC_ORDER): Ditto.
5743         (VALID_MASK_REG_MODE): New.
5744         (FIRST_MASK_REG): Ditto.
5745         (LAST_MASK_REG): Ditto.
5746         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5747         (MAYBE_MASK_CLASS_P): New.
5748         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5749         (REG_CLASS_CONTENTS): Ditto.
5750         (MASK_REGNO_P): New.
5751         (ANY_MASK_REG_P): Ditto.
5752         (HI_REGISTER_NAMES): Add new mask registers.
5753         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5754         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5755         mask registers.
5756         (attribute "type"): Add mskmov, msklog.
5757         (attribute "length_immediate"): Support them.
5758         (attribute "memory"): Ditto.
5759         (attribute "prefix_0f"): Ditto.
5760         (*movhi_internal): Support new mask registers.
5761         (*movqi_internal): Ditto.
5762         (define_split): Split out clobber pattern is a logic
5763         insn on mask registers.
5764         (*k<logic><mode>): New.
5765         (*andhi_1): Extend to support mask regs.
5766         (*andqi_1): Extend to support mask regs.
5767         (kandn<mode>): New.
5768         (define_split): Split and-not to and and not if operands
5769         are not mask regs.
5770         (*<code><mode>_1): Separate HI mode to new pattern...
5771         (*<code>hi_1): This.
5772         (*<code>qi_1): Extend to support mask regs.
5773         (kxnor<mode>): New.
5774         (kortestzhi): Ditto.
5775         (kortestchi): Ditto.
5776         (kunpckhi): Ditto.
5777         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5778         (*one_cmplhi2_1): ...Here, now with mask registers support.
5779         (*one_cmplqi2_1): Support new mask registers.
5780         (HI/QImode arithmetics splitter): Don't split if mask registers
5781         are used.
5782         (HI/QImode not splitter): Ditto.
5783         * config/i386/predicated.md (mask_reg_operand): New.
5784         (general_reg_operand): Ditto.
5786 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5788         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5789         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5791 2013-09-10  Jeff Law  <law@redhat.com>
5793         PR tree-optimization/58380
5794         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5795         of threading through a buried loop header.
5797         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5798         return value for single successor case.
5800 2013-09-10  Jan Hubicka  <jh@suse.cz>
5802         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5804 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5806         PR target/58361
5807         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5808         support conditional execution.
5809         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5811 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5813         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5814         constraint pass.
5815         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5816         Check destination too to check move insn.
5817         (undo_optional_reloads): Add check that the original peudo did not
5818         changed its allocation and the optional reload was inherited on last
5819         inheritance pass.  Break loop after deciding to keep optional reload.
5820         (lra_undo_inheritance): Add check that inherited pseudo still in
5821         memory.
5823 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5825         * config/aarch64/aarch64.md (generic_sched): New.
5826         * config/aarch64/aarch64-generic.md (load): Make conditional
5827         on generic_sched attribute.
5828         (nonload): Likewise.
5830 2013-09-10  Jan Hubicka  <jh@suse.cz>
5832         * lto-cgraph.c: Include ipa-utils.h.
5833         (compute_ltrans_boundary): Also add possible targets into the boundary.
5835 2013-09-10  Jan Hubicka  <jh@suse.cz>
5837         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5838         VAR_DECL of vtable rather than full expression.
5840 2013-09-10  Jan Hubicka  <jh@suse.cz>
5841             Paolo Carlini  <paolo.carlini@oracle.com>
5843         * cgraphunit.c (analyze_functions): Save input_location, set it
5844         to UNKNOWN_LOCATION and restore it at the end.
5846 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5848         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5849         represented by a thunk.
5851 2013-09-10  Jeff Law  <law@redhat.com>
5853         PR tree-optimization/58343
5854         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5855         jump threading requests through loop headers buried in the middle
5856         of a jump threading path.
5858         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5859         in return value/type.
5861 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5863         PR rtl-optimization/58365
5864         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5865         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5866         it differs.
5868 2013-09-10  Richard Biener  <rguenther@suse.de>
5870         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5871         * tree-data-ref.c (create_rdg_vertices): Collect all data
5872         references, signal failure to the caller, use data-ref API.
5873         (build_rdg): Compute data references only once.  Maintain lifetime
5874         of data references and data dependences from within RDG.
5875         (free_rdg): Free dependence relations.
5876         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5877         inventing extra dependences.
5878         (distribute_loop): Update for RDG API changes.
5880 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5882         * doc/invoke.texi (fms-extensions): Document changed
5883         behavior for ms-abi targets.
5884         * config/i386/i386.c (ix86_option_override_internal):
5885         Set default value of option -fms-extension for ms-abi targets.
5887 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5889         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5891 2013-09-10  Alan Modra  <amodra@gmail.com>
5893         PR target/58330
5894         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5896 2013-09-10  Alan Modra  <amodra@gmail.com>
5898         * config/rs6000/predicates.md (add_cint_operand): New.
5899         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5900         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5901         using add_cint_operand.
5902         (largetoc_high_plus_aix): Likewise.
5904 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5906         PR tree-optimization/58364
5907         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5908         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5909         the current range can't be an unconditional true or false.
5911 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5913         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5915 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5917         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5919 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5921         PR c++/43452
5922         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5924 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5926         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5927         NO_REGS for immediate that can't be moved directly into FP_REGS.
5929 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5931         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5932         comparison with negated operand.
5933         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5934         RTL form.
5936 2013-09-09  Richard Biener  <rguenther@suse.de>
5938         PR middle-end/58326
5939         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5940         of a subloop record all its block as affecting loop-closed SSA form.
5942 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5944         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5945         of an rtx/bitpos pair.
5946         (store_fixed_bit_field): Update accordingly.
5948 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5950         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5951         GEN_INT.
5952         * builtins.c (expand_errno_check): Likewise.
5953         * dwarf2cfi.c (init_return_column_size): Likewise.
5954         * except.c (sjlj_mark_call_sites): Likewise.
5955         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
5956         * lra-constraints.c (emit_inc): Likewise.
5957         * ree.c (combine_set_extension): Likewise.
5958         * regmove.c (fixup_match_2): Likewise.
5959         * reload1.c (inc_for_reload): Likewise.
5961 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5963         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
5964         (force_to_mode, simplify_shift_const_1, simplify_comparison):
5965         Use gen_int_mode with the mode of the associated simplify_* call.
5966         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
5967         * expmed.c (expand_shift_1): Likewise.
5968         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5969         * loop-iv.c (iv_number_of_iterations): Likewise.
5970         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5971         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
5973 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5975         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
5976         of the associated expand_* call.
5977         (asan_emit_stack_protection): Likewise.
5978         * builtins.c (round_trampoline_addr): Likewise.
5979         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
5980         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
5981         (emit_store_flag): Likewise.
5982         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
5983         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
5984         Likewise.
5985         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5986         * ifcvt.c (noce_try_store_flag_constants): Likewise.
5987         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5988         * modulo-sched.c (generate_prolog_epilog): Likewise.
5989         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
5990         (expand_ctz, expand_ffs, expand_unop): Likewise.
5992 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5994         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
5995         of the associated gen_rtx_* call.
5996         * caller-save.c (init_caller_save): Likewise.
5997         * combine.c (find_split_point, make_extraction): Likewise.
5998         (make_compound_operation): Likewise.
5999         * dwarf2out.c (mem_loc_descriptor): Likewise.
6000         * explow.c (plus_constant, probe_stack_range): Likewise.
6001         * expmed.c (expand_mult_const): Likewise.
6002         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6003         * reload1.c (init_reload): Likewise.
6004         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6005         * var-tracking.c (adjust_mems): Likewise.
6006         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6007         rather than gen_rtx_fmt_ee.
6009 2013-09-09  Jan Hubicka  <jh@suse.cz>
6011         PR middle-end/58294
6012         * value-prof.c (gimple_ic): Copy also abnormal edges.
6014 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6016         * asan.c (asan_shadow_cst): Use gen_int_mode.
6018 2013-09-08  Jan Hubicka  <jh@suse.cz>
6020         * ipa-profile.c: Add toplevel comment.
6021         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6022         (contains_hot_call_p): New function.
6023         (ipa_propagate_frequency): Set frequencies based on counts when
6024         profile is read.
6025         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6026         profile; do not tamper with profile after inlining if it is read.
6028 2013-09-08  Jan Hubicka  <jh@suse.cz>
6030         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6031         speculative edges.
6033 2013-09-08  Jan Hubicka  <jh@suse.cz>
6035         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6036         summary generation.
6038 2013-09-08  Jeff Law  <law@redhat.com>
6040         PR bootstrap/58340
6041         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6042         of 'found'.
6044 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6046         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6048 2013-09-08  Jan Hubicka  <jh@suse.cz>
6050         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6052 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6054         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6055         for non-debug insns.
6056         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6057         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6058         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6059         accordingly.
6061 2013-09-08  Jan Hubicka  <jh@suse.cz>
6063         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6064         targets and devirtualize to BUILT_IN_UNREACHABLE.
6065         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6066         * ipa.c (walk_polymorphic_call_targets): New function.
6067         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6068         functions; use the new timevar.
6069         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6070         was removed from the program.
6071         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6072         not consider it in the walk when its vtable is dead.
6073         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6074         record_binfo.
6075         (devirt_variable_node_removal_hook): New function.
6076         (possible_polymorphic_call_targets): Also register
6077         devirt_variable_node_removal_hook.
6078         (ipa_devirt): Do not do non-speculative devirtualization.
6079         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6081 2013-09-08  Jan Hubicka  <jh@suse.cz>
6083         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6084         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6085         varpool_remove_variable_insertion_hook): Declare.
6086         * varpool.c (varpool_node_hook_list): New structure.
6087         (first_varpool_node_removal_hook,
6088         first_varpool_variable_insertion_hook): New variables.
6089         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6090         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6091         varpool_remove_variable_insertion_hook,
6092         varpool_call_variable_insertion_hooks): New functions.
6093         (varpool_remove_node): Use it.
6095 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6097         PR c++/54941
6098         * diagnostic.c (diagnostic_build_prefix): When s.file is
6099         "<built-in>" don't output line and column numbers.
6101 2013-09-06  Jan Hubicka  <jh@suse.cz>
6103         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6104         * lto-streamer-out.c: Stream thunks, too.
6105         * lto-streamer-in.c (input_function): Pop cfun here
6106         (lto_read_body): Instead of here.
6108 2013-09-06  Caroline Tice  <cmtice@google.com>
6110         * doc/install.texi: Add documentation for the --enable-vtable-verify
6111         and the --disable-libvtv configure options.
6113 2013-09-06  Jeff Law  <law@redhat.com>
6115         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6116         edge implied equivalences into successor phis.
6118 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6120         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6122 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6124         * resource.c (mark_target_live_regs): Compute resources taking
6125         into account if a call is predicated or not.
6127 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6129         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6130         the compiler in the function names.
6132 2013-09-06  Jan Hubicka  <jh@suse.cz>
6134         PR middle-end/58094
6135         * ipa-inline.c (has_caller_p): New function.
6136         (want_inline_function_to_all_callers_p): Use it.
6137         (sum_callers, inline_to_all_callers): Break out from ...
6138         (ipa_inline): ... here.
6140 2013-09-06  Jan Hubicka  <jh@suse.cz>
6142         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6143         only when AVX is enabled.
6145 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6147         * config/aarch64/aarch64.md
6148         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6149         is fpsimd_<load/store>2.
6150         (load_pair<mode>): Likewise.
6151         (store_pair<mode>): Likewise.
6153 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6155         * config/arm/types.md (type): Add "mrs" type.
6156         * config/aarch64/aarch64.md
6157         (aarch64_load_tp_hard): Make type "mrs".
6158         * config/arm/arm.md
6159         (load_tp_hard): Make type "mrs".
6160         * config/arm/cortex-a15.md: Update with new attributes.
6161         * config/arm/cortex-a5.md: Update with new attributes.
6162         * config/arm/cortex-a53.md: Update with new attributes.
6163         * config/arm/cortex-a7.md: Update with new attributes.
6164         * config/arm/cortex-a8.md: Update with new attributes.
6165         * config/arm/cortex-a9.md: Update with new attributes.
6166         * config/arm/cortex-m4.md: Update with new attributes.
6167         * config/arm/cortex-r4.md: Update with new attributes.
6168         * config/arm/fa526.md: Update with new attributes.
6169         * config/arm/fa606te.md: Update with new attributes.
6170         * config/arm/fa626te.md: Update with new attributes.
6171         * config/arm/fa726te.md: Update with new attributes.
6173 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6175         * config/aarch64/aarch64.md
6176         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6177         (*movtf_aarch64): Likewise.
6178         * config/arm/arm.md
6179         (thumb1_movdi_insn): Use "multiple" for type where more than one
6180         instruction is used for a move.
6181         (*arm32_movhf): Likewise.
6182         (*thumb_movdf_insn): Likewise.
6184 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6186         * config/arm/types.md (type): Rename fcpys to fmov.
6187         * config/arm/vfp.md
6188         (*arm_movsi_vfp): Rename type fcpys as fmov.
6189         (*thumb2_movsi_vfp): Likewise
6190         (*movhf_vfp_neon): Likewise
6191         (*movhf_vfp): Likewise
6192         (*movsf_vfp): Likewise
6193         (*thumb2_movsf_vfp): Likewise
6194         (*movsfcc_vfp): Likewise
6195         (*thumb2_movsfcc_vfp): Likewise
6196         * config/aarch64/aarch64-simd.md
6197         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6198         * config/aarch64/aarch64.md
6199         (*movsi_aarch64): Replace type mov_reg with fmovs.
6200         (*movdi_aarch64): Likewise
6201         (*movsf_aarch64): Likewise
6202         (*movdf_aarch64): Likewise
6203         * config/arm/arm.c
6204         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6205         * config/arm/iwmmxt.md
6206         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6207         * config/arm/arm1020e.md: Update with new attributes.
6208         * config/arm/cortex-a15-neon.md: Update with new attributes.
6209         * config/arm/cortex-a5.md: Update with new attributes.
6210         * config/arm/cortex-a53.md: Update with new attributes.
6211         * config/arm/cortex-a7.md: Update with new attributes.
6212         * config/arm/cortex-a8-neon.md: Update with new attributes.
6213         * config/arm/cortex-a9.md: Update with new attributes.
6214         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6215         * config/arm/cortex-r4f.md: Update with new attributes.
6216         * config/arm/marvell-pj4.md: Update with new attributes.
6217         * config/arm/vfp11.md: Update with new attributes.
6219 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6221         * config/aarch64/aarch64.md
6222         (*madd<mode>): Fix type attribute.
6223         (*maddsi_uxtw): Likewise.
6224         (*msub<mode>): Likewise.
6225         (*msubsi_uxtw): Likewise.
6226         (<su_optab>maddsidi4): Likewise.
6227         (<su_optab>msubsidi4): Likewise.
6229 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6231         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6232         * config/arm/arm.md (core_cycles): Remove fdiv.
6233         * config/arm/vfp.md:
6234         (*sqrtsf2_vfp): Update for attribute changes.
6235         (*sqrtdf2_vfp): Likewise.
6236         * config/aarch64/aarch64.md:
6237         (sqrt<mode>2): Update for attribute changes.
6238         * config/arm/arm1020e.md: Update with new attributes.
6239         * config/arm/cortex-a15-neon.md: Update with new attributes.
6240         * config/arm/cortex-a5.md: Update with new attributes.
6241         * config/arm/cortex-a53.md: Update with new attributes.
6242         * config/arm/cortex-a7.md: Update with new attributes.
6243         * config/arm/cortex-a8-neon.md: Update with new attributes.
6244         * config/arm/cortex-a9.md: Update with new attributes.
6245         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6246         * config/arm/cortex-r4f.md: Update with new attributes.
6247         * config/arm/marvell-pj4.md: Update with new attributes.
6248         * config/arm/vfp11.md: Update with new attributes.
6250 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6252         * config/arm/types.md
6253         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6254         * config/aarch64/aarch64.md
6255         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6256         new attributes.
6257         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6258         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6259         (float<GPI:mode><GPF:mode>2): Likewise.
6260         * config/arm/vfp.md
6261         (*truncsisf2_vfp): Update with new attributes.
6262         (*truncsidf2_vfp): Likewise.
6263         (fixuns_truncsfsi2): Likewise.
6264         (fixuns_truncdfsi2): Likewise.
6265         (*floatsisf2_vfp): Likewise.
6266         (*floatsidf2_vfp): Likewise.
6267         (floatunssisf2): Likewise.
6268         (floatunssidf2): Likewise.
6269         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6270         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6271         * config/arm/arm1020e.md: Update with new attributes.
6272         * config/arm/cortex-a15-neon.md: Update with new attributes.
6273         * config/arm/cortex-a5.md: Update with new attributes.
6274         * config/arm/cortex-a53.md: Update with new attributes.
6275         * config/arm/cortex-a7.md: Update with new attributes.
6276         * config/arm/cortex-a8-neon.md: Update with new attributes.
6277         * config/arm/cortex-a9.md: Update with new attributes.
6278         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6279         * config/arm/cortex-r4f.md: Update with new attributes.
6280         * config/arm/marvell-pj4.md: Update with new attributes.
6281         * config/arm/vfp11.md: Update with new attributes.
6283 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6285         * config/aarch64/arm_neon.h
6286         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6287         (vqtbx<1,2,3,4><q>_s8): Likewise.
6289 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6291         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6292         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6293         (add<mode>3): Add type attribute.
6294         (add<mode>3): Likewise.
6295         (usadd<mode>3): Likewise.
6296         (ssadd<mode>3): Likewise.
6297         (sub<mode>3): Likewise.
6298         (sub<mode>3): Likewise.
6299         (ussub<mode>3): Likewise.
6300         (sssub<mode>3): Likewise.
6301         (ssmulsa3): Likewise.
6302         (usmulusa3): Likewise.
6303         (arm_usatsihi): Likewise.
6304         * config/arm/vfp.md
6305         (*movdi_vfp): Add types for all instructions.
6306         (*movdi_vfp_cortexa8): Likewise.
6307         (*movhf_vfp_neon): Likewise.
6308         (*movhf_vfp): Likewise.
6309         (*movdf_vfp): Likewise.
6310         (*thumb2_movdf_vfp): Likewise.
6311         (*thumb2_movdfcc_vfp): Likewise.
6312         * config/arm/arm.md: Add type attribute to all insn patterns.
6313         (*thumb1_adddi3): Add type attribute.
6314         (*arm_adddi3): Likewise.
6315         (*adddi_sesidi_di): Likewise.
6316         (*adddi_zesidi_di): Likewise.
6317         (*thumb1_addsi3): Likewise.
6318         (addsi3_compare0): Likewise.
6319         (*addsi3_compare0_scratch): Likewise.
6320         (*compare_negsi_si): Likewise.
6321         (cmpsi2_addneg): Likewise.
6322         (*addsi3_carryin_<optab>): Likewise.
6323         (*addsi3_carryin_alt2_<optab>): Likewise.
6324         (*addsi3_carryin_clobercc_<optab>): Likewise.
6325         (*subsi3_carryin): Likewise.
6326         (*subsi3_carryin_const): Likewise.
6327         (*subsi3_carryin_compare): Likewise.
6328         (*subsi3_carryin_compare_const): Likewise.
6329         (*arm_subdi3): Likewise.
6330         (*thumb_subdi3): Likewise.
6331         (*subdi_di_zesidi): Likewise.
6332         (*subdi_di_sesidi): Likewise.
6333         (*subdi_zesidi_di): Likewise.
6334         (*subdi_sesidi_di): Likewise.
6335         (*subdi_zesidi_ze): Likewise.
6336         (thumb1_subsi3_insn): Likewise.
6337         (*arm_subsi3_insn): Likewise.
6338         (*anddi3_insn): Likewise.
6339         (*anddi_zesidi_di): Likewise.
6340         (*anddi_sesdi_di): Likewise.
6341         (*ne_zeroextracts): Likewise.
6342         (*ne_zeroextracts): Likewise.
6343         (*ite_ne_zeroextr): Likewise.
6344         (*ite_ne_zeroextr): Likewise.
6345         (*anddi_notdi_di): Likewise.
6346         (*anddi_notzesidi): Likewise.
6347         (*anddi_notsesidi): Likewise.
6348         (andsi_notsi_si): Likewise.
6349         (thumb1_bicsi3): Likewise.
6350         (*iordi3_insn): Likewise.
6351         (*iordi_zesidi_di): Likewise.
6352         (*iordi_sesidi_di): Likewise.
6353         (*thumb1_iorsi3_insn): Likewise.
6354         (*xordi3_insn): Likewise.
6355         (*xordi_zesidi_di): Likewise.
6356         (*xordi_sesidi_di): Likewise.
6357         (*arm_xorsi3): Likewise.
6358         (*andsi_iorsi3_no): Likewise.
6359         (*smax_0): Likewise.
6360         (*smax_m1): Likewise.
6361         (*arm_smax_insn): Likewise.
6362         (*smin_0): Likewise.
6363         (*arm_smin_insn): Likewise.
6364         (*arm_umaxsi3): Likewise.
6365         (*arm_uminsi3): Likewise.
6366         (*minmax_arithsi): Likewise.
6367         (*minmax_arithsi_): Likewise.
6368         (*satsi_<SAT:code>): Likewise.
6369         (arm_ashldi3_1bit): Likewise.
6370         (arm_ashrdi3_1bit): Likewise.
6371         (arm_lshrdi3_1bit): Likewise.
6372         (*arm_negdi2): Likewise.
6373         (*thumb1_negdi2): Likewise.
6374         (*arm_negsi2): Likewise.
6375         (*thumb1_negsi2): Likewise.
6376         (*negdi_extendsid): Likewise.
6377         (*negdi_zero_extend): Likewise.
6378         (*arm_abssi2): Likewise.
6379         (*thumb1_abssi2): Likewise.
6380         (*arm_neg_abssi2): Likewise.
6381         (*thumb1_neg_abss): Likewise.
6382         (one_cmpldi2): Likewise.
6383         (extend<mode>di2): Likewise.
6384         (*compareqi_eq0): Likewise.
6385         (*arm_extendhisi2addsi): Likewise.
6386         (*arm_movdi): Likewise.
6387         (*thumb1_movdi_insn): Likewise.
6388         (*arm_movt): Likewise.
6389         (*thumb1_movsi_insn): Likewise.
6390         (pic_add_dot_plus_four): Likewise.
6391         (pic_add_dot_plus_eight): Likewise.
6392         (tls_load_dot_plus_eight): Likewise.
6393         (*thumb1_movhi_insn): Likewise.
6394         (*thumb1_movsf_insn): Likewise.
6395         (*movdf_soft_insn): Likewise.
6396         (*thumb_movdf_insn): Likewise.
6397         (cbranchsi4_insn): Likewise.
6398         (cbranchsi4_scratch): Likewise.
6399         (*negated_cbranchsi4): Likewise.
6400         (*tbit_cbranch): Likewise.
6401         (*tlobits_cbranch): Likewise.
6402         (*tstsi3_cbranch): Likewise.
6403         (*cbranchne_decr1): Likewise.
6404         (*addsi3_cbranch): Likewise.
6405         (*addsi3_cbranch_scratch): Likewise.
6406         (*arm_cmpdi_insn): Likewise.
6407         (*arm_cmpdi_unsig): Likewise.
6408         (*arm_cmpdi_zero): Likewise.
6409         (*thumb_cmpdi_zero): Likewise.
6410         (*deleted_compare): Likewise.
6411         (*mov_scc): Likewise.
6412         (*mov_negscc): Likewise.
6413         (*mov_notscc): Likewise.
6414         (*cstoresi_eq0_thumb1_insn): Likewise.
6415         (cstoresi_nltu_thumb1): Likewise.
6416         (cstoresi_ltu_thu): Likewise.
6417         (thumb1_addsi3_addgeu): Likewise.
6418         (*arm_jump): Likewise.
6419         (*thumb_jump): Likewise.
6420         (*check_arch2): Likewise.
6421         (arm_casesi_internal): Likewise.
6422         (thumb1_casesi_dispatch): Likewise.
6423         (*arm_indirect_jump): Likewise.
6424         (*thumb1_indirect_jump): Likewise.
6425         (nop): Likewise.
6426         (*and_scc): Likewise.
6427         (*ior_scc): Likewise.
6428         (*compare_scc): Likewise.
6429         (*cond_move): Likewise.
6430         (*cond_arith): Likewise.
6431         (*cond_sub): Likewise.
6432         (*cmp_ite0): Likewise.
6433         (*cmp_ite1): Likewise.
6434         (*cmp_and): Likewise.
6435         (*cmp_ior): Likewise.
6436         (*ior_scc_scc): Likewise.
6437         (*ior_scc_scc_cmp): Likewise.
6438         (*and_scc_scc): Likewise.
6439         (*and_scc_scc_cmp): Likewise.
6440         (*and_scc_scc_nod): Likewise.
6441         (*negscc): Likewise.
6442         (movcond_addsi): Likewise.
6443         (movcond): Likewise.
6444         (*ifcompare_plus_move): Likewise.
6445         (*if_plus_move): Likewise.
6446         (*ifcompare_move_plus): Likewise.
6447         (*if_move_plus): Likewise.
6448         (*ifcompare_arith_arith): Likewise.
6449         (*if_arith_arith): Likewise.
6450         (*ifcompare_arith_move): Likewise.
6451         (*if_arith_move): Likewise.
6452         (*ifcompare_move_arith): Likewise.
6453         (*if_move_arith): Likewise.
6454         (*ifcompare_move_not): Likewise.
6455         (*if_move_not): Likewise.
6456         (*ifcompare_not_move): Likewise.
6457         (*if_not_move): Likewise.
6458         (*ifcompare_shift_move): Likewise.
6459         (*if_shift_move): Likewise.
6460         (*ifcompare_move_shift): Likewise.
6461         (*if_move_shift): Likewise.
6462         (*ifcompare_shift_shift): Likewise.
6463         (*ifcompare_not_arith): Likewise.
6464         (*ifcompare_arith_not): Likewise.
6465         (*if_arith_not): Likewise.
6466         (*ifcompare_neg_move): Likewise.
6467         (*if_neg_move): Likewise.
6468         (*ifcompare_move_neg): Likewise.
6469         (*if_move_neg): Likewise.
6470         (prologue_thumb1_interwork): Likewise.
6471         (*cond_move_not): Likewise.
6472         (*sign_extract_onebit): Likewise.
6473         (*not_signextract_onebit): Likewise.
6474         (stack_tie): Likewise.
6475         (align_4): Likewise.
6476         (align_8): Likewise.
6477         (consttable_end): Likewise.
6478         (consttable_1): Likewise.
6479         (consttable_2): Likewise.
6480         (consttable_4): Likewise.
6481         (consttable_8): Likewise.
6482         (consttable_16): Likewise.
6483         (*thumb1_tablejump): Likewise.
6484         (prefetch): Likewise.
6485         (force_register_use): Likewise.
6486         (thumb_eh_return): Likewise.
6487         (load_tp_hard): Likewise.
6488         (load_tp_soft): Likewise.
6489         (tlscall): Likewise.
6490         (*arm_movtas_ze): Likewise.
6491         (*arm_rev): Likewise.
6492         (*arm_revsh): Likewise.
6493         (*arm_rev16): Likewise.
6494         * config/arm/thumb2.md
6495         (*thumb2_smaxsi3): Likewise.
6496         (*thumb2_sminsi3): Likewise.
6497         (*thumb32_umaxsi3): Likewise.
6498         (*thumb2_uminsi3): Likewise.
6499         (*thumb2_negdi2): Likewise.
6500         (*thumb2_abssi2): Likewise.
6501         (*thumb2_neg_abss): Likewise.
6502         (*thumb2_movsi_insn): Likewise.
6503         (tls_load_dot_plus_four): Likewise.
6504         (*thumb2_movhi_insn): Likewise.
6505         (*thumb2_mov_scc): Likewise.
6506         (*thumb2_mov_negs): Likewise.
6507         (*thumb2_mov_negs): Likewise.
6508         (*thumb2_mov_nots): Likewise.
6509         (*thumb2_mov_nots): Likewise.
6510         (*thumb2_movsicc_): Likewise.
6511         (*thumb2_movsfcc_soft_insn): Likewise.
6512         (*thumb2_indirect_jump): Likewise.
6513         (*thumb2_and_scc): Likewise.
6514         (*thumb2_ior_scc): Likewise.
6515         (*thumb2_ior_scc_strict_it): Likewise.
6516         (*thumb2_cond_move): Likewise.
6517         (*thumb2_cond_arith): Likewise.
6518         (*thumb2_cond_ari): Likewise.
6519         (*thumb2_cond_sub): Likewise.
6520         (*thumb2_negscc): Likewise.
6521         (*thumb2_movcond): Likewise.
6522         (thumb2_casesi_internal): Likewise.
6523         (thumb2_casesi_internal_pic): Likewise.
6524         (*thumb2_alusi3_short): Likewise.
6525         (*thumb2_mov<mode>_shortim): Likewise.
6526         (*thumb2_addsi_short): Likewise.
6527         (*thumb2_subsi_short): Likewise.
6528         (thumb2_addsi3_compare0): Likewise.
6529         (*thumb2_cbz): Likewise.
6530         (*thumb2_cbnz): Likewise.
6531         (*thumb2_one_cmplsi2_short): Likewise.
6532         (*thumb2_negsi2_short): Likewise.
6533         (*orsi_notsi_si): Likewise.
6534         * config/arm/arm1020e.md: Update with new attributes.
6535         * config/arm/arm1026ejs.md: Update with new attributes.
6536         * config/arm/arm1136jfs.md: Update with new attributes.
6537         * config/arm/arm926ejs.md: Update with new attributes.
6538         * config/arm/cortex-a15.md: Update with new attributes.
6539         * config/arm/cortex-a5.md: Update with new attributes.
6540         * config/arm/cortex-a53.md: Update with new attributes.
6541         * config/arm/cortex-a7.md: Update with new attributes.
6542         * config/arm/cortex-a8.md: Update with new attributes.
6543         * config/arm/cortex-a9.md: Update with new attributes.
6544         * config/arm/cortex-m4.md: Update with new attributes.
6545         * config/arm/cortex-r4.md: Update with new attributes.
6546         * config/arm/fa526.md: Update with new attributes.
6547         * config/arm/fa606te.md: Update with new attributes.
6548         * config/arm/fa626te.md: Update with new attributes.
6549         * config/arm/fa726te.md: Update with new attributes.
6551 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6553         * config/aarch64/aarch64-simd.md
6554         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6555         <vwx> iterator to ensure correct register choice.
6556         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6557         (aarch64_sqdmull_n<mode>): Likewise.
6558         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6559         * config/aarch64/arm_neon.h
6560         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6561         (vml<as><q>_n_<su>16): Likewise.
6562         (vml<as>l_high_lane<q>_<su>16): Likewise.
6563         (vml<as>l_high_n_<su>16): Likewise.
6564         (vml<as>l_lane<q>_<su>16): Likewise.
6565         (vml<as>l_n_<su>16): Likewise.
6566         (vmul<q>_lane<q>_<su>16): Likewise.
6567         (vmul<q>_n_<su>16): Likewise.
6568         (vmull_lane<q>_<su>16): Likewise.
6569         (vmull_n_<su>16): Likewise.
6570         (vmull_high_lane<q>_<su>16): Likewise.
6571         (vmull_high_n_<su>16): Likewise.
6572         (vqrdmulh<q>_n_s16): Likewise.
6574 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6576         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6577         have the correct lane parameter.
6579 2013-09-06  Richard Biener <rguenther@suse.de>
6581         * cfganal.c (control_dependences::~control_dependences):
6582         Properly free all of the vector.
6584 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6586         PR target/58269
6587         * config/i386/i386.c (ix86_conditional_register_usage):
6588         Proper initialize extended SSE registers.
6590 2013-09-06  Jan Hubicka  <jh@suse.cz>
6592         PR tree-optimization/58311
6593         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6595 2013-09-06  Jan Hubicka  <jh@suse.cz>
6597         * Makefile.in (tree-sra.o): Update dependencies.
6598         * tree-sra.c: Include ipa-utils.h
6599         (scan_function): Use recursive_call_p.
6600         (has_caller_p): New function.
6601         (cgraph_for_node_and_aliases): Count also callers of aliases.
6603 2013-09-06  Jan Hubicka  <jh@suse.cz>
6605         PR middle-end/58094
6606         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6607         * tree-tailcall.c: Include ipa-utils.h.
6608         (find_tail_calls): Use it.
6609         * ipa-pure-const.c (check_call): Likewise.
6610         * ipa-utils.c (recursive_call_p): New function.
6611         * ipa-utils.h (recursive_call_p): Dclare.
6612         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6613         (symtab_semantically_equivalent_p): New function.
6614         * Makefile.in (tree-tailcall.o): Update dependencies.
6616 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6618         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6619         non-inlinable part.
6621 2013-09-06  Richard Biener  <rguenther@suse.de>
6623         * lto-streamer.h (lto_global_var_decls): Remove.
6624         * Makefile.in (OBJS): Remove lto-symtab.o.
6625         (lto-symtab.o): Remove.
6626         (GTFILES): Remove lto-symtab.c
6627         * lto-symtab.c: Move to lto/
6629 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6631         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6632         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6633         (UNSPEC_FPINT_RINT): New constant definitions.
6634         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6635         definition with 2 attributes.
6636         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6637         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6638         definitions.
6640 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6642         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6643         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6644         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6645         attribute to "z196_alone".
6646         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6647         "zEC12_simple".
6649 2013-09-06  Richard Biener  <rguenther@suse.de>
6651         * basic-block.h (class control_dependences): New.
6652         * tree-ssa-dce.c (control_dependence_map): Remove.
6653         (cd): New global.
6654         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6655         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6656         find_pdom, find_control_dependence, find_all_control_dependences):
6657         Move to cfganal.c.
6658         (mark_control_dependent_edges_necessary,
6659         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6660         tree_dce_done, perform_tree_ssa_dce): Adjust.
6661         * cfganal.c (set_control_dependence_map_bit,
6662         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6663         find_all_control_dependences): Move from tree-ssa-dce.c and
6664         implement as methods of control_dependences class.
6665         (control_dependences::control_dependences): New.
6666         (control_dependences::~control_dependences): Likewise.
6667         (control_dependences::get_edges_dependent_on): Likewise.
6668         (control_dependences::get_edge): Likewise.
6670 2013-09-04  Jan Hubicka  <jh@suse.cz>
6672         * tree.c (types_same_for_odr): Drop overactive check.
6673         * ipa-devirt.c (hash_type_name): Likewise.
6675 2013-09-04  Jan Hubicka  <jh@suse.cz>
6677         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6678         (analyze_functions): ... here.
6680 2013-09-04  Jan Hubicka  <jh@suse.cz>
6682         PR middle-end/58201
6683         * cgraphunit.c (analyze_functions): Clear AUX fields
6684         after processing; initialize assembler name has.
6686 2013-09-05  Jeff Law  <law@redhat.com>
6688         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6689         from thread_around_empty_block.  Record threading path into PATH.
6690         Recurse if threading through the initial block is successful.
6691         (thread_across_edge): Corresponding changes to slightly simplify.
6693 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6695         * config/aarch64/aarch64.md
6696         (type): Remove frecpe, frecps, frecpx.
6697         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6698         fix to be a TARGET_SIMD instruction.
6699         (aarch64_frecps): Remove.
6700         * config/aarch64/aarch64-simd.md
6701         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6702         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6704 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6705             Sofiane Naci  <sofiane.naci@arm.com>
6707         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6708         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6709         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6710         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6711         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6712         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6713         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6714         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6715         into "clz, "rbit".  Rename "shift" to "shift_imm".
6716         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6717         changes.  Update for attribute changes all occurrences of arlo_* and
6718         shift* types.
6719         * config/arm/arm-fixed.md: Update for attribute changes
6720         all occurrences of arlo_* types.
6721         * config/arm/thumb2.md: Update for attribute changes all occurrences
6722         of arlo_* types.
6723         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6724         (cortexa7_older_only): Likewise.
6725         (cortexa7_younger):  Likewise.
6726         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6727         (1020alu_shift_op): Likewise.
6728         (1020alu_shift_reg_op): Likewise.
6729         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6730         (alu_shift_op): Likewise.
6731         (alu_shift_reg_op): Likewise.
6732         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6733         (11_alu_shift_op): Likewise.
6734         (11_alu_shift_reg_op): Likewise.
6735         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6736         (9_alu_shift_reg_op): Likewise.
6737         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6738         attribute changes.
6739         (cortex_a15_alu_shift): Likewise.
6740         (cortex_a15_alu_shift_reg): Likewise.
6741         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6742         attribute changes.
6743         (cortex_a5_alu_shift): Likewise.
6744         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6745         attribute changes.
6746         (cortex_a53_alu_shift): Likewise.
6747         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6748         attribute changes.
6749         (cortex_a7_alu_reg): Likewise.
6750         (cortex_a7_alu_shift): Likewise.
6751         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6752         attribute changes.
6753         (cortex_a8_alu_shift): Likewise.
6754         (cortex_a8_alu_shift_reg): Likewise.
6755         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6756         (cortex_a9_dp_shift): Likewise.
6757         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6758         attribute changes.
6759         * config/arm/cortex-r4.md
6760         (cortex_r4_alu): Update for attribute changes.
6761         (cortex_r4_mov): Likewise.
6762         (cortex_r4_alu_shift_reg): Likewise.
6763         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6764         (526_alu_shift_op): Likewise.
6765         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6766         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6767         (626te_alu_shift_op): Likewise.
6768         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6769         (726te_alu_shift_op): Likewise.
6770         (726te_alu_shift_reg_op): Likewise.
6771         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6772         (mp626_alu_shift_op): Likewise.
6773         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6774         (pj4_alu_conds): Likewise.
6775         (pj4_shift): Likewise.
6776         (pj4_shift_conds): Likewise.
6777         (pj4_alu_shift): Likewise.
6778         (pj4_alu_shift_conds): Likewise.
6779         * config/aarch64/aarch64.md: Update for attribute change
6780         all occurrences of arlo_* and shift* types.
6782 2013-09-05  Mike Stump  <mikestump@comcast.net>
6784         * tree.h: Move documentation for tree_function_decl to tree-core.h
6785         with the declaration.
6787 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6789         PR target/58139
6790         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6791         looking for widest mode.
6793 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6795         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6797 2013-09-05  Richard Biener  <rguenther@suse.de>
6799         PR tree-optimization/58137
6800         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6801         Do not create vectors of pointers.
6802         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6803         types for the components of the vector initializer.
6804         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6805         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6807 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6809         * ipa-prop.c (remove_described_reference): Accept missing references,
6810         return false if that hppens, otherwise return true.
6811         (cgraph_node_for_jfunc): New function.
6812         (try_decrement_rdesc_refcount): Likewise.
6813         (try_make_edge_direct_simple_call): Use them.
6814         (ipa_edge_removal_hook): Remove references from rdescs.
6815         (ipa_edge_duplication_hook): Clone rdescs and their references
6816         when the new edge has the same caller as the old one.
6817         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6818         reference before removing any edges.
6820 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6822         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6823         initial store.
6824         * thumb2.md (thumb2_storewb_parisi): New pattern.
6826 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6828         * config/aarch64/aarch64-option-extensions.def: Add
6829         AARCH64_OPT_EXTENSION of 'crc'.
6830         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6831         (AARCH64_ISA_CRC): Ditto.
6832         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6833         description of the CRC extension.
6835 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6837         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6838         * config/rs6000/linux.h: Ditto.
6839         * alpha/linux.h: Ditto.
6840         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6841         no_c99_libc_has_function.
6842         * config/c6x/uclinux-elf.h: Ditto.
6843         * config/lm32/uclinux-elf.h: Ditto.
6844         * config/m68k/uclinux.h: Ditto.
6845         * config/moxie/uclinux.h: Ditto.
6846         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6847         (crisv32-*-linux*, cris-*-linux*): Ditto.
6848         * config/bfin/bfin.c: Include "tm_p.h".
6850 2013-09-05  Richard Biener  <rguenther@suse.de>
6852         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6853         check for a definition without a basic-block.
6855 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6856             Sofiane Naci <sofiane.naci@arm.com>
6858         * config/aarch64/aarch64.md
6859         (*movti_aarch64): Rename r_2_f and f_2_r.
6860         (*movsf_aarch64): Likewise.
6861         (*movdf_aarch64): Likewise.
6862         (*movtf_aarch64): Likewise.
6863         (aarch64_movdi_<mode>low): Likewise.
6864         (aarch64_movdi_<mode>high): Likewise.
6865         (aarch64_mov<mode>high_di): Likewise.
6866         (aarch64_mov<mode>low_di): Likewise.
6867         (aarch64_movtilow_tilow): Likewise.
6868         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6869         values to config/arm/types.md
6870         (attribute "conds"): Update for attribute change.
6871         (anddi3_insn): Likewise.
6872         (iordi3_insn): Likewise.
6873         (xordi3_insn): Likewise.
6874         (one_cmpldi2): Likewise.
6875         * config/arm/types.md (type): Add Neon types.
6876         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6877         use "type" attribute.
6878         (movmisalign<mode>_neon_store): Likewise.
6879         (movmisalign<mode>_neon_load): Likewise.
6880         (vec_set<mode>_internal): Likewise.
6881         (vec_setv2di_internal): Likewise.
6882         (vec_extract<mode>): Likewise.
6883         (vec_extractv2di): Likewise.
6884         (add<mode>3_neon): Likewise.
6885         (adddi3_neon): Likewise.
6886         (sub<mode>3_neon): Likewise.
6887         (subdi3_neon): Likewise.
6888         (mul<mode>3_neon): Likewise.
6889         (mul<mode>3add<mode>_neon): Likewise.
6890         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6891         (fma<VCVTF:mode>4)): Likewise.
6892         (fma<VCVTF:mode>4_intrinsic): Likewise.
6893         (fmsub<VCVTF:mode>4)): Likewise.
6894         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6895         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6896         (ior<mode>3): Likewise.
6897         (and<mode>3): Likewise.
6898         (anddi3_neon): Likewise.
6899         (orn<mode>3_neon): Likewise.
6900         (orndi3_neon): Likewise.
6901         (bic<mode>3_neon): Likewise.
6902         (bicdi3_neon): Likewise.
6903         (xor<mode>3): Likewise.
6904         (one_cmpl<mode>2): Likewise.
6905         (abs<mode>2): Likewise.
6906         (neg<mode>2): Likewise.
6907         (umin<mode>3_neon): Likewise.
6908         (umax<mode>3_neon): Likewise.
6909         (smin<mode>3_neon): Likewise.
6910         (smax<mode>3_neon): Likewise.
6911         (vashl<mode>3): Likewise.
6912         (vashr<mode>3_imm): Likewise.
6913         (vlshr<mode>3_imm): Likewise.
6914         (ashl<mode>3_signed): Likewise.
6915         (ashl<mode>3_unsigned): Likewise.
6916         (neon_load_count): Likewise.
6917         (ashldi3_neon_noclobber): Likewise.
6918         (signed_shift_di3_neon): Likewise.
6919         (unsigned_shift_di3_neon): Likewise.
6920         (ashrdi3_neon_imm_noclobber): Likewise.
6921         (lshrdi3_neon_imm_noclobber): Likewise.
6922         (widen_ssum<mode>3): Likewise.
6923         (widen_usum<mode>3): Likewise.
6924         (quad_halves_<code>v4si): Likewise.
6925         (quad_halves_<code>v4sf): Likewise.
6926         (quad_halves_<code>v8hi): Likewise.
6927         (quad_halves_<code>v16qi): Likewise.
6928         (reduc_splus_v2di): Likewise.
6929         (neon_vpadd_internal<mode>): Likewise.
6930         (neon_vpsmin<mode>): Likewise.
6931         (neon_vpsmax<mode>): Likewise.
6932         (neon_vpumin<mode>): Likewise.
6933         (neon_vpumax<mode>): Likewise.
6934         (ss_add<mode>_neon): Likewise.
6935         (us_add<mode>_neon): Likewise.
6936         (ss_sub<mode>_neon): Likewise.
6937         (us_sub<mode>_neon): Likewise.
6938         (neon_vadd<mode>_unspec): Likewise.
6939         (neon_vaddl<mode>): Likewise.
6940         (neon_vaddw<mode>): Likewise.
6941         (neon_vhadd<mode>): Likewise.
6942         (neon_vqadd<mode>): Likewise.
6943         (neon_vaddhn<mode>): Likewise.
6944         (neon_vmul<mode>): Likewise.
6945         (neon_vmla<mode>): Likewise.
6946         (neon_vmlal<mode>): Likewise.
6947         (neon_vmls<mode>): Likewise.
6948         (neon_vmlsl<mode>): Likewise.
6949         (neon_vqdmulh<mode>): Likewise.
6950         (neon_vqdmlal<mode>): Likewise.
6951         (neon_vqdmlsl<mode>): Likewise.
6952         (neon_vmull<mode>): Likewise.
6953         (neon_vqdmull<mode>): Likewise.
6954         (neon_vsub<mode>_unspec): Likewise.
6955         (neon_vsubl<mode>): Likewise.
6956         (neon_vsubw<mode>): Likewise.
6957         (neon_vqsub<mode>): Likewise.
6958         (neon_vhsub<mode>): Likewise.
6959         (neon_vsubhn<mode>): Likewise.
6960         (neon_vceq<mode>): Likewise.
6961         (neon_vcge<mode>): Likewise.
6962         (neon_vcgeu<mode>): Likewise.
6963         (neon_vcgt<mode>): Likewise.
6964         (neon_vcgtu<mode>): Likewise.
6965         (neon_vcle<mode>): Likewise.
6966         (neon_vclt<mode>): Likewise.
6967         (neon_vcage<mode>): Likewise.
6968         (neon_vcagt<mode>): Likewise.
6969         (neon_vtst<mode>): Likewise.
6970         (neon_vabd<mode>): Likewise.
6971         (neon_vabdl<mode>): Likewise.
6972         (neon_vaba<mode>): Likewise.
6973         (neon_vabal<mode>): Likewise.
6974         (neon_vmax<mode>): Likewise.
6975         (neon_vmin<mode>): Likewise.
6976         (neon_vpaddl<mode>): Likewise.
6977         (neon_vpadal<mode>): Likewise.
6978         (neon_vpmax<mode>): Likewise.
6979         (neon_vpmin<mode>): Likewise.
6980         (neon_vrecps<mode>): Likewise.
6981         (neon_vrsqrts<mode>): Likewise.
6982         (neon_vqabs<mode>): Likewise.
6983         (neon_vqneg<mode>): Likewise.
6984         (neon_vcls<mode>): Likewise.
6985         (clz<mode>2): Likewise.
6986         (popcount<mode>2): Likewise.
6987         (neon_vrecpe): Likewise.
6988         (neon_vrsqrte): Likewise.
6989         (neon_vget_lane<mode>_sext_internal): Likewise.
6990         (neon_vget_lane<mode>_zext_internal): Likewise.
6991         (neon_vdup_n<mode>): Likewise.
6992         (neon_vdup_nv2di): Likewise.
6993         (neon_vdpu_lane<mode>_internal): Likewise.
6994         (neon_vswp<mode>): Likewise.
6995         (float<mode><V_cvtto>2): Likewise.
6996         (floatuns<mode><V_cvtto>2): Likewise.
6997         (fix_trunc<mode><V_cvtto>)2): Likewise
6998         (fixuns_trunc<mode><V_cvtto)2): Likewise.
6999         (neon_vcvt<mode>): Likewise.
7000         (neon_vcvtv4sfv4hf): Likewise.
7001         (neon_vcvtv4hfv4sf): Likewise.
7002         (neon_vcvt_n<mode>): Likewise.
7003         (neon_vmovn<mode>): Likewise.
7004         (neon_vqmovn<mode>): Likewise.
7005         (neon_vqmovun<mode>): Likewise.
7006         (neon_vmovl<mode>): Likewise.
7007         (neon_vmul_lane<mode>): Likewise.
7008         (neon_vmull_lane<mode>): Likewise.
7009         (neon_vqdmull_lane<mode>): Likewise.
7010         (neon_vqdmulh_lane<mode>): Likewise.
7011         (neon_vmla_lane<mode>): Likewise.
7012         (neon_vmlal_lane<mode>): Likewise.
7013         (neon_vqdmlal_lane<mode>): Likewise.
7014         (neon_vmls_lane<mode>): Likewise.
7015         (neon_vmlsl_lane<mode>): Likewise.
7016         (neon_vqdmlsl_lane<mode>): Likewise.
7017         (neon_vext<mode>): Likewise.
7018         (neon_vrev64<mode>): Likewise.
7019         (neon_vrev32<mode>): Likewise.
7020         (neon_vrev16<mode>): Likewise.
7021         (neon_vbsl<mode>_internal): Likewise.
7022         (neon_vshl<mode>): Likewise.
7023         (neon_vqshl<mode>): Likewise.
7024         (neon_vshr_n<mode>): Likewise.
7025         (neon_vshrn_n<mode>): Likewise.
7026         (neon_vqshrn_n<mode>): Likewise.
7027         (neon_vqshrun_n<mode>): Likewise.
7028         (neon_vshl_n<mode>): Likewise.
7029         (neon_vqshl_n<mode>): Likewise.
7030         (neon_vqshlu_n<mode>): Likewise.
7031         (neon_vshll_n<mode>): Likewise.
7032         (neon_vsra_n<mode>): Likewise.
7033         (neon_vsri_n<mode>): Likewise.
7034         (neon_vsli_n<mode>): Likewise.
7035         (neon_vtbl1v8qi): Likewise.
7036         (neon_vtbl2v8qi): Likewise.
7037         (neon_vtbl3v8qi): Likewise.
7038         (neon_vtbl4v8qi): Likewise.
7039         (neon_vtbx1v8qi): Likewise.
7040         (neon_vtbx2v8qi): Likewise.
7041         (neon_vtbx3v8qi): Likewise.
7042         (neon_vtbx4v8qi): Likewise.
7043         (neon_vtrn<mode>_internal): Likewise.
7044         (neon_vzip<mode>_internal): Likewise.
7045         (neon_vuzp<mode>_internal): Likewise.
7046         (neon_vld1<mode>): Likewise.
7047         (neon_vld1_lane<mode>): Likewise.
7048         (neon_vld1_dup<mode>): Likewise.
7049         (neon_vld1_dupv2di): Likewise.
7050         (neon_vst1<mode>): Likewise.
7051         (neon_vst1_lane<mode>): Likewise.
7052         (neon_vld2<mode>): Likewise.
7053         (neon_vld2_lane<mode>): Likewise.
7054         (neon_vld2_dup<mode>): Likewise.
7055         (neon_vst2<mode>): Likewise.
7056         (neon_vst2_lane<mode>): Likewise.
7057         (neon_vld3<mode>): Likewise.
7058         (neon_vld3qa<mode>): Likewise.
7059         (neon_vld3qb<mode>): Likewise.
7060         (neon_vld3_lane<mode>): Likewise.
7061         (neon_vld3_dup<mode>): Likewise.
7062         (neon_vst3<mode>): Likewise.
7063         (neon_vst3qa<mode>): Likewise.
7064         (neon_vst3qb<mode>): Likewise.
7065         (neon_vst3_lane<mode>): Likewise.
7066         (neon_vld4<mode>): Likewise.
7067         (neon_vld4qa<mode>): Likewise.
7068         (neon_vld4qb<mode>): Likewise.
7069         (neon_vld4_lane<mode>): Likewise.
7070         (neon_vld4_dup<mode>): Likewise.
7071         (neon_vst4<mode>): Likewise.
7072         (neon_vst4qa<mode>): Likewise.
7073         (neon_vst4qb<mode>): Likewise.
7074         (neon_vst4_lane<mode>): Likewise.
7075         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7076         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7077         (neon_vec_<US>mult_lo_<mode>): Likewise.
7078         (neon_vec_<US>mult_hi_<mode>): Likewise.
7079         (neon_vec_<US>shiftl_<mode>): Likewise.
7080         (neon_unpack<US>_<mode>): Likewise.
7081         (neon_vec_<US>mult_<mode>): Likewise.
7082         (vec_pack_trunc_<mode>): Likewise.
7083         (neon_vec_pack_trunk_<mode>): Likewise.
7084         (neon_vabd<mode>_2): Likewise.
7085         (neon_vabd<mode>_3): Likewise.
7086         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7087         (thumb2_movsi_vfp): Likewise.
7088         (movdi_vfp): Likewise.
7089         (movdi_vfp_cortexa8): Likewise.
7090         (movhf_vfp_neon): Likewise.
7091         (movhf_vfp): Likewiwse.
7092         (movsf_vfp): Likewiwse.
7093         (thumb2_movsf_vfp): Likewiwse.
7094         (movdf_vfp): Likewise.
7095         (thumb2_movdf_vfp): Likewise.
7096         (movsfcc_vfp): Likewise.
7097         (thumb2_movsfcc_vfp): Likewise.
7098         (movdfcc_vfp): Likewise.
7099         (thumb2_movdfcc_vfp): Likewise.
7100         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7101         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7102         (v10_v2c): Likewise.
7103         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7104         attribute change.
7105         (cortex_a15_neon_int_2): Likewise.
7106         (cortex_a15_neon_int_3): Likewise.
7107         (cortex_a15_neon_int_4): Likewise.
7108         (cortex_a15_neon_int_5): Likewise.
7109         (cortex_a15_neon_vqneg_vqabs): Likewise.
7110         (cortex_a15_neon_vmov): Likewise.
7111         (cortex_a15_neon_vaba): Likewise.
7112         (cortex_a15_neon_vaba_qqq): Likewise.
7113         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7114         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7115         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7116         scalar_64_32_long_scalar): Likewise.
7117         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7118         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7119         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7120         lotype_qdd_64_32_long): Likewise.
7121         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7122         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7123         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7124         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7125         (cortex_a15_neon_shift_1): Likewise.
7126         (cortex_a15_neon_shift_2): Likewise.
7127         (cortex_a15_neon_shift_3): Likewise.
7128         (cortex_a15_neon_vshl_ddd): Likewise.
7129         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7130         (cortex_a15_neon_vsra_vrsra): Likewise.
7131         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7132         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7133         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7134         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7135         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7136         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7137         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7138         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7139         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7140         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7141         (cortex_a15_neon_bp_simple): Likewise.
7142         (cortex_a15_neon_bp_2cycle): Likewise.
7143         (cortex_a15_neon_bp_3cycle): Likewise.
7144         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7145         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7146         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7147         (cortex_a15_neon_vld2_4_regs): Likewise.
7148         (cortex_a15_neon_vld3_vld4): Likewise.
7149         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7150         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7151         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7152         (cortex_a15_neon_vst3_vst4): Likewise.
7153         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7154         (cortex_a15_neon_vld3_vld4_lane" 10
7155         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7156         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7157         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7158         (cortex_a15_neon_ldm_2): Likewise.0
7159         (cortex_a15_neon_stm_2): Likewise.
7160         (cortex_a15_neon_mcr): Likewise.
7161         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7162         (cortex_a15_neon_mrc): Likewise.
7163         (cortex_a15_neon_mrrc): Likewise.
7164         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7165         change.
7166         (cortex_a15_alu_shift): Likewise.
7167         (cortex_a15_alu_shift_reg): Likewise.
7168         (cortex_a15_mult32): Likewise.
7169         (cortex_a15_mult64): Likewise.
7170         (cortex_a15_block): Likewise.
7171         (cortex_a15_branch): Likewise.
7172         (cortex_a15_load1): Likewise.
7173         (cortex_a15_load3): Likewise.
7174         (cortex_a15_store1): Likewise.
7175         (cortex_a15_store3): Likewise.
7176         (cortex_a15_call): Likewise.
7177         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7178         (cortex_a5_f2r): Likewise.
7179         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7180         change.
7181         (cortex_a53_f2r): Likewise.
7182         * config/arm/cortex-a7.md
7183         (cortex_a7_branch): Update for attribute change.
7184         (cortex_a7_call): Likewise.
7185         (cortex_a7_alu_imm): Likewise.
7186         (cortex_a7_alu_reg): Likewise.
7187         (cortex_a7_alu_shift): Likewise.
7188         (cortex_a7_mul): Likewise.
7189         (cortex_a7_load1): Likewise.
7190         (cortex_a7_store1): Likewise.
7191         (cortex_a7_load2): Likewise.
7192         (cortex_a7_store2): Likewise.
7193         (cortex_a7_load3): Likewise.
7194         (cortex_a7_store3): Likewise.
7195         (cortex_a7_load4): Likewise.
7196         (cortex_a7_store4): Likewise.
7197         (cortex_a7_fpalu): Likewise.
7198         (cortex_a7_fconst): Likewise.
7199         (cortex_a7_fpmuls): Likewise.
7200         (cortex_a7_neon_mul): Likewise.
7201         (cortex_a7_fpmacs): Likewise.
7202         (cortex_a7_neon_mla: Likewise.
7203         (cortex_a7_fpmuld: Likewise.
7204         (cortex_a7_fpmacd: Likewise.
7205         (cortex_a7_fpfmad: Likewise.
7206         (cortex_a7_fdivs: Likewise.
7207         (cortex_a7_fdivd: Likewise.
7208         (cortex_a7_r2f: Likewise.
7209         (cortex_a7_f2r: Likewise.
7210         (cortex_a7_f_flags: Likewise.
7211         (cortex_a7_f_loads: Likewise.
7212         (cortex_a7_f_loadd: Likewise.
7213         (cortex_a7_f_stores: Likewise.
7214         (cortex_a7_f_stored: Likewise.
7215         (cortex_a7_neon): Likewise.
7216         * config/arm/cortex-a8-neon.md
7217         (cortex_a8_neon_mrc): Update for attribute change.
7218         (cortex_a8_neon_mrrc): Likewise.
7219         (cortex_a8_neon_int_1): Likewise.
7220         (cortex_a8_neon_int_2): Likewise.
7221         (cortex_a8_neon_int_3): Likewise.
7222         (cortex_a8_neon_int_4): Likewise.
7223         (cortex_a8_neon_int_5): Likewise.
7224         (cortex_a8_neon_vqneg_vqabs): Likewise.
7225         (cortex_a8_neon_vmov): Likewise.
7226         (cortex_a8_neon_vaba): Likewise.
7227         (cortex_a8_neon_vaba_qqq): Likewise.
7228         (cortex_a8_neon_vsma): Likewise.
7229         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7230         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7231         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7232         long_scalar): Likewise.
7233         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7234         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7235         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7236         long_scalar_qdd_64_32_long): Likewise.
7237         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7238         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7239         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7240         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7241         (cortex_a8_neon_shift_1): Likewise.
7242         (cortex_a8_neon_shift_2): Likewise.
7243         (cortex_a8_neon_shift_3): Likewise.
7244         (cortex_a8_neon_vshl_ddd): Likewise.
7245         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7246         (cortex_a8_neon_vsra_vrsra): Likewise.
7247         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7248         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7249         (cortex_a8_neon_fp_vsum): Likewise.
7250         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7251         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7252         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7253         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7254         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7255         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7256         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7257         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7258         (cortex_a8_neon_bp_simple): Likewise.
7259         (cortex_a8_neon_bp_2cycle): Likewise.
7260         (cortex_a8_neon_bp_3cycle): Likewise.
7261         (cortex_a8_neon_ldr): Likewise.
7262         (cortex_a8_neon_str): Likewise.
7263         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7264         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7265         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7266         (cortex_a8_neon_vld2_4_regs): Likewise.
7267         (cortex_a8_neon_vld3_vld4): Likewise.
7268         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7269         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7270         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7271         (cortex_a8_neon_vst3_vst4): Likewise.
7272         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7273         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7274         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7275         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7276         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7277         (cortex_a8_neon_mcr): Likewise.
7278         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7279         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7280         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7281         change.
7282         (ca9_neon_mrrc): Likewise.
7283         (cortex_a9_neon_int_1): Likewise.
7284         (cortex_a9_neon_int_2): Likewise.
7285         (cortex_a9_neon_int_3): Likewise.
7286         (cortex_a9_neon_int_4): Likewise.
7287         (cortex_a9_neon_int_5): Likewise.
7288         (cortex_a9_neon_vqneg_vqabs): Likewise.
7289         (cortex_a9_neon_vmov): Likewise.
7290         (cortex_a9_neon_vaba): Likewise.
7291         (cortex_a9_neon_vaba_qqq): Likewise.
7292         (cortex_a9_neon_vsma): Likewise.
7293         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7294         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7295         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7296         long_scalar): Likewise.
7297         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7298         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7299         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7300         long_scalar_qdd_64_32_long): Likewise.
7301         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7302         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7303         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7304         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7305         (cortex_a9_neon_shift_1): Likewise.
7306         (cortex_a9_neon_shift_2): Likewise.
7307         (cortex_a9_neon_shift_3): Likewise.
7308         (cortex_a9_neon_vshl_ddd): Likewise.
7309         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7310         (cortex_a9_neon_vsra_vrsra): Likewise.
7311         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7312         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7313         (cortex_a9_neon_fp_vsum): Likewise.
7314         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7315         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7316         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7317         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7318         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7319         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7320         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7321         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7322         (cortex_a9_neon_bp_simple): Likewise.
7323         (cortex_a9_neon_bp_2cycle): Likewise.
7324         (cortex_a9_neon_bp_3cycle): Likewise.
7325         (cortex_a9_neon_ldr): Likewise.
7326         (cortex_a9_neon_str): Likewise.
7327         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7328         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7329         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7330         (cortex_a9_neon_vld2_4_regs): Likewise.
7331         (cortex_a9_neon_vld3_vld4): Likewise.
7332         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7333         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7334         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7335         (cortex_a9_neon_vst3_vst4): Likewise.
7336         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7337         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7338         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7339         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7340         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7341         (cortex_a9_neon_mcr): Likewise.
7342         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7343         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7344         (cortex_a9_fps): Likewise.
7345         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7346         change.
7347         (cortex_m4_fmuls): Likewise.
7348         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7349         change.
7350         (cortex_r4_mrc): Likewise.
7351         * config/arm/iterators.md: Update comment referring to neon_type.
7352         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7353         (iwmmxt_movsi_insn): Likewise.
7354         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7355         attribute change.
7356         (pj4_core_to_vfp): Likewise.
7357         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7358         attribute change.
7359         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7360         (vfp_fstore): Likewise.
7361         * doc/md.texi: Change references to neon_type to refer to type.
7363 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7365         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7367 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7369         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7370         lglobl if not weak.
7372 2013-09-04  Easwaran Raman  <eraman@google.com>
7374         PR middle-end/57370
7375         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7376         (build_and_add_sum): Use it.
7377         (appears_later_in_bb): Simplify code.
7379 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7381         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7383 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7385         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7387 2013-09-04  Jan Hubicka  <jh@suse.cz>
7389         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7390         * ipa-devirt.c: Include diganostic.h
7391         (odr_type_d): Add types and types_set.
7392         (hash_type_name): Work for types with vtables during LTO.
7393         (odr_hasher::remove): Fix comment; destroy types_set.
7394         (add_type_duplicate): New function,
7395         (get_odr_type): Use it.
7396         (dump_type_inheritance_graph): Dump type duplicates.
7397         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7398         graph.
7399         * tree.c (types_same_for_odr): Give exact answers on types with
7400         virtual tables.
7402 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7404         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7405         explaining their differences.
7407 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7409         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7411 2013-09-03  Jeff Law  <law@redhat.com>
7413         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7414         when not threading through a joiner block.  Pass joiner/no joiner
7415         state to register_jump_thread.
7416         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7417         state from argument rather than implying on path length.
7418         Dump the entire jump thread path into debugging dump.
7419         * tree-flow.h (register_jump_thread): Update prototype.
7421 2013-08-29  Xinliang David Li  <davidxl@google.com>
7423         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7424         Remove a trivial gcc_assert.
7426 2013-08-29  Xinliang David Li  <davidxl@google.com>
7428         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7429         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7430         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7431         Delay base decl alignment adjustment.
7432         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7433         * tree-vectorizer.h: New data structure.
7434         (set_dr_misalignment): New function.
7435         (dr_misalignment): Ditto.
7436         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7437         (vectorizable_load): Ditto.
7438         (ensure_base_align): New function.
7439         (vectorize_loops): Add dbg_cnt support.
7440         (execute_vect_slp): Ditto.
7441         * dbgcnt.def: New debug counter.
7442         * Makefile: New dependency.
7444 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7446         Revert:
7448         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7450         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7452 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7454         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7455         function descriptor.
7457 2013-09-03  Richard Biener  <rguenther@suse.de>
7459         * tree-affine.c (add_elt_to_tree): Fix association issue,
7460         avoid useless converts and make sure to always return a
7461         properly typed result.
7463 2013-09-03  Richard Biener  <rguenther@suse.de>
7465         PR middle-end/57656
7466         * fold-const.c (negate_expr_p): Fix division case.
7467         (negate_expr): Likewise.
7469 2013-09-03  Richard Biener  <rguenther@suse.de>
7471         PR lto/58285
7472         * tree-streamer-out.c: Include tm.h.
7473         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7475 2013-09-03  Jan Hubicka  <jh@suse.cz>
7477         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7479 2013-09-03  Alan Modra  <amodra@gmail.com>
7481         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7482         multilibs to big-endian target and vice versa.
7483         * config/rs6000/t-linux64: Use := assignment on all vars.
7484         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7485         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7486         * config/rs6000/t-linux64le: New file.
7487         * config/rs6000/t-linux64bele: New file.
7488         * config/rs6000/t-linux64lebe: New file.
7490 2013-09-02  Jan Hubicka  <jh@suse.cz>
7492         * ipa-inline-transform.c (inline_transform): Do not
7493         optimize_inline_calls when not optimizing.
7495 2013-09-02  Jan Hubicka  <jh@suse.cz>
7497         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7498         duplicated nodes for assembler names.
7499         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7500         hard registers from assembler name hash.
7502 2013-09-02  Jan Hubicka  <jh@suse.cz>
7504         * ipa-split.c (execute_split_functions): Split externally visible
7505         functions called once.
7507 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7509         PR ipa/58106
7510         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7511         linked list.  When finding the correct duplicate, also consider also
7512         the caller in additon to its inlined_to node.
7514 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7516         * config/aarch64/aarch64-simd-builtins.def
7517         (dup_lane_scalar): Remove.
7518         * config/aarch64/aarch64-simd.md
7519         (aarch64_simd_dup): Add 'w->w' alternative.
7520         (aarch64_dup_lane<mode>): Allow for VALL.
7521         (aarch64_dup_lane_scalar<mode>): Remove.
7522         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7523         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7524         (aarch64_get_lane_unsigned<mode>): Likewise.
7525         (aarch64_get_lane<mode>): Likewise.
7526         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7527         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7528         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7529         (VCON): Change container of V2SF.
7530         (vswap_width_name): Likewise.
7531         * config/aarch64/arm_neon.h
7532         (__aarch64_vdup_lane_any): New.
7533         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7534         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7535         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7537 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7539         PR middle-end/56382
7540         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7541         the source or the destination is a single hard register.
7543 2013-09-02  Richard Biener  <rguenther@suse.de>
7545         PR middle-end/57511
7546         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7547         non-linear SCEVs.
7549 2013-09-02  Richard Biener  <rguenther@suse.de>
7551         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7552         arithmetic to sizetype.
7554 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7556         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7557         Find auto-increment use both before and after candidate.
7559 2013-09-02  Marek Polacek  <polacek@redhat.com>
7561         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7563 2013-09-01  Jan Hubicka  <jh@suse.cz>
7565         * Makefile.in: Add ipa-profile.o
7566         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7567         * cgraph.c (struct cgraph_propagate_frequency_data,
7568         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7569         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7570         * cgraph.h (cgraph_propagate_frequency): Remove.
7571         * ipa-inline-analysis.c: Include ipa-utils.h;
7572         drop duplicated cfgloop.h.
7573         (inline_update_callee_summaries): Update.
7574         * ipa-profile.c: New file.
7575         * ipa-utils.h (ipa_propagate_frequency): Declare.
7576         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7577         data-streamer.h, value-prof.h.
7578         (symtab_remove_unreachable_nodes): Update profile.
7579         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7580         account_time_size, cmp_counts, dump_histogram,
7581         ipa_profile_generate_summary, ipa_profile_write_summary,
7582         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7583         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7584         Move to ipa-profile.c.
7586 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7588         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7590 2013-09-01  Jan Hubicka  <jh@suse.cz>
7592         * common.opt (fdevirtualize-speculatively): New function.
7593         * invoke.texi (fdevirtualize-speculatively): Document.
7594         * ipa-devirt.c: Include ipa-inline.h
7595         (likely_target_p): New function.
7596         (ipa_devirt): New function.
7597         (gate_ipa_devirt): New function.
7598         (pass_data_ipa_devirt): New static var.
7599         (pass_ipa_devirt): Likewise.
7600         (make_pass_ipa_devirt): New function.
7601         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7602         (common_handle_option): Disable devirtualization when
7603         value range profiling is available.
7604         * passes.def (pass_ipa_devirt): Add.
7605         * timever.def (TV_IPA_DEVIRT): New timevar.
7606         * tree-pass.h (make_pass_ipa_devirt):
7608 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7610         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7611         include sanitize(undefined).
7613 2013-08-31  Diego Novillo  <dnovillo@google.com>
7615         * Makefile.in (TREE_CORE_H): Define.
7616         (TREE_H): Use.
7617         (GTFILES): Add tree-core.h.
7618         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7619         size the array.
7620         * tree-core.h: New file.
7621         Move all data structures, enum, typedefs, global
7622         declarations and constants from ...
7623         * tree.h: ... here.
7625 2013-08-31  Jan Hubicka  <jh@suse.cz>
7627         * bulitins.c (expand_builtin): Do not early exit for gcov
7628         instrumented functions.
7630 2013-08-31  Marek Polacek  <polacek@redhat.com>
7632         * ubsan.c: Include tm_p.h.
7634 2013-08-31  Jan Hubicka  <jh@suse.cz>
7636         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7637         warning.
7639         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7640         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7641         * tree-cfg.c (verify_gimple_label): ... here.
7642         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7643         (ipa_merge_profiles): New function.
7644         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7645         (lto_input_function_body): Likewise.
7646         * ipa-utils.h (ipa_merge_profiles): Declare.
7647         * lto-streamer.h (lto_input_function_body): Update prototype.
7648         (emit_label_in_global_context_p): Remove.
7649         * lto-symtab.c: Include ipa-utils.h
7650         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7652 2013-08-31  Jan Hubicka  <jh@suse.cz>
7654         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7656 2013-08-31  Jan Hubicka  <jh@suse.cz>
7658         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7660 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7662         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7663         "cmp" RTX before signed_comparison_operator check to account
7664         for "code" changes.
7666 2013-08-30  Jan Hubicka  <jh@suse.cz>
7668         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7669         (detect_type_change_1): Rename to ...
7670         (detect_type_change): ... this one; early return on non-polymorphic
7671         types.
7672         (detect_type_change_ssa): Add comp_type parameter; update
7673         use of detect_type_change.
7674         (compute_complex_assign_jump_func): Add param_type parameter;
7675         update use of detect_type_change_ssa.
7676         (compute_complex_ancestor_jump_func): Likewise.
7677         (ipa_get_callee_param_type): New function.
7678         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7679         update calls to the jump function computation functions.
7681 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7682             Steven Bosscher  <steven@gcc.gnu.org>
7684         * cfgrtl.c (fixup_new_cold_bb): New routine.
7685         (commit_edge_insertions): Invoke fixup_partitions.
7686         (find_partition_fixes): New routine.
7687         (fixup_partitions): Ditto.
7688         (verify_hot_cold_block_grouping): Update comments.
7689         (rtl_verify_edges): Invoke find_partition_fixes.
7690         (rtl_verify_bb_pointers): Update comments.
7691         (rtl_verify_bb_layout): Ditto.
7692         * basic-block.h (probably_never_executed_edge_p): Declare.
7693         (fixup_partitions): Ditto.
7694         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7695         * bb-reorder.c (sanitize_hot_paths): New function.
7696         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7697         sanitize_hot_paths.
7698         * predict.c (probably_never_executed_edge_p): New routine.
7699         * cfg.c (check_bb_profile): Add partition insanity warnings.
7701 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7703         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7705 2013-08-30  Marek Polacek  <polacek@redhat.com>
7707         * Makefile.in (ubsan.o): Add.
7708         (c-family/c-ubsan.o): Add.
7709         (builtins.o): Add ubsan.h dependency.
7710         * ubsan.h: New file.
7711         * ubsan.c: New file.
7712         * common.opt: Add -fsanitize=undefined option.
7713         (flag_sanitize): Add variable.
7714         (fsanitize=): Add option.  Add Driver.
7715         (fsanitize=thread): Remove option.
7716         (fsanitize=address): Likewise.
7717         (static-libubsan): New option.
7718         * doc/invoke.texi: Document the new flag and -static-libubsan.
7719         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7720         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7721         * builtin-attrs.def (ATTR_COLD): Define.
7722         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7723         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7724         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7725         * flag-types.h (sanitize_code): New enum.
7726         * opts.c (common_handle_option): Parse command line arguments
7727         of -fsanitize=.  Add -fsanitize=unreachable option.
7728         * varasm.c (get_variable_section): Adjust.
7729         (assemble_noswitch_variable): Likewise.
7730         (assemble_variable): Likewise.
7731         (output_constant_def_contents): Likewise.
7732         (categorize_decl_for_section): Likewise.
7733         (place_block_symbol): Likewise.
7734         (output_object_block): Likewise.
7735         * builtins.def: Likewise.
7736         * toplev.c (compile_file): Likewise.
7737         (process_options): Likewise.
7738         * cppbuiltin.c: Likewise.
7739         * tsan.c (tsan_pass): Likewise.
7740         (tsan_gate): Likewise.
7741         (tsan_gate_O0): Likewise.
7742         * cfgexpand.c (partition_stack_vars): Likewise.
7743         (expand_stack_vars): Likewise.
7744         (defer_stack_allocation): Likewise.
7745         (expand_used_vars): Likewise.
7746         * cfgcleanup.c (old_insns_match_p): Likewise.
7747         * asan.c (asan_finish_file): Likewise.
7748         (asan_instrument): Likewise.
7749         (gate_asan): Likewise.
7750         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7751         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7752         (asan_global_struct): Use pointer_sized_int_node instead
7753         calling build_nonstandard_integer_type.
7754         (initialize_sanitizer_builtins): Likewise.
7755         (asan_finish_file): Likewise.
7756         * gcc.c: Document %{%:function(args):X}.
7757         (static_spec_functions): Add sanitize.
7758         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7759         store funcval != NULL there.
7760         (do_spec_1): Adjust handle_spec_function caller.
7761         (handle_braces): Allow %:function(args) as condition.
7762         (sanitize_spec_function): New function.
7763         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7764         (LIBUBSAN_SPEC): Likewise.
7765         (LIBUBSAN_EARLY_SPEC): Likewise.
7766         (SANITIZER_SPEC): Handle libubsan.
7767         (SANITIZER_EARLY_SPEC): Likewise.
7768         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7769         instead of fsanitize=address.
7770         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7771         instead of fsanitize=address*.
7772         * builtins.c: Include ubsan.h.
7773         (fold_builtin_0): Instrument __builtin_unreachable.
7774         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7775         instead of flag_asan.
7776         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7777         (pointer_sized_int_node): Define.
7778         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7780 2013-08-30  Mike Stump  <mikestump@comcast.net>
7782         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7783         with RE patterns.
7785 2013-08-29  Jan Hubicka  <jh@suse.cz>
7787         * cgraph.c (cgraph_function_body_availability): Handle weakref
7788         correctly.
7789         * passes.def: Remove pass_fixup_cfg.
7790         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7791         track when we need to remove functions.
7792         (gate_ipa_inline): Execute inlining always; add comment why.
7793         (pass_data_ipa_inline): Remove TODO_remove_functions.
7794         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7795         do not produce summaries.
7796         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7797         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7798         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7799         constant pool and vtable.
7801 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7803         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7804         New arm_neon.h's internal macros to specify 64-bit constants.
7805         Avoid using stdint.h's macros.
7807 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7809         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7810         before applying REG_P.
7812 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7814         PR tree-optimization/58277
7815         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7816         after seeing too many stmts with vdef in between dombb and current
7817         bb, invalidate everything.
7819 2013-08-30  Richard Biener  <rguenther@suse.de>
7821         * fold-const.c (fold_single_bit_test): Fix overflow test.
7823 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7825         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7826         and which can live in a register, always retrieve the value on entry.
7827         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7828         passed by invisible reference specially.
7829         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7830         (vt_add_function_parameter): Correctly deal with a parameter passed by
7831         invisible reference.
7833 2013-08-30  Jan Hubicka  <jh@suse.cz>
7835         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7837 2013-08-30  Richard Biener  <rguenther@suse.de>
7839         PR tree-optimization/58228
7840         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7841         allow invariant loads in nested loop vectorization.
7843 2013-08-30  Richard Biener  <rguenther@suse.de>
7845         PR tree-optimization/58223
7846         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7847         (has_anti_or_output_dependence): ... this and adjust to also
7848         look for output dependences.
7849         (mark_nodes_having_upstream_mem_writes): Adjust.
7850         (rdg_flag_uses): Likewise.
7852 2013-08-30  Richard Biener  <rguenther@suse.de>
7854         PR tree-optimization/58010
7855         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7856         assert that we have a loop-closed PHI.
7858 2013-08-29  Jan Hubicka  <jh@suse.cz>
7860         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7861         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7862         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7863         (lto_free_function_in_decl_state_for_node): New function.
7865 2013-08-29  Xinliang David Li  <davidxl@google.com>
7867         * loop-unroll.c (report_unroll_peel): Minor message change.
7868         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7869         Emit alignment peeling message with default -fopt-info.
7870         (vect_loop_versioning): Emit loop version info message.
7871         * tree-vectorizer.c (vectorize_loops): Minor message change.
7872         (execute_vect_slp): Ditto.
7874 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7876         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7877         of the clone from the DECL_NAME of the original function.
7879 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7881         * passes.c (register_pass): Add overload.
7882         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7884 2013-08-29  Jan Hubicka  <jh@suse.cz>
7886         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7887         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7888         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7889         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7890         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7891         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7892         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7893         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7895 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7897         * dumpfile.c (dump_loc): Output column number.
7898         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7899         * doc/invoke.texi: Document optall -fopt-info flag.
7900         * profile.c (read_profile_edge_counts): Use new dump framework.
7901         (compute_branch_probabilities): Ditto.
7902         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7903         when pass not in any opt group.
7904         * pass_manager.h (pass_manager::get_pass_profile): New method.
7905         * value-prof.c (check_counter): Use new dump framework.
7906         (check_ic_target): Ditto.
7907         * coverage.c (get_coverage_counts): Ditto.
7908         (coverage_init): Setup new dump framework.
7910 2013-08-29  Richard Biener  <rguenther@suse.de>
7912         PR tree-optimization/58246
7913         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7914         handle the dominance check inside a basic-block.
7916 2013-08-29  Richard Biener  <rguenther@suse.de>
7918         PR middle-end/57287
7919         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7920         of default defs that appear in abnormal PHI nodes.
7922 2013-08-29  Richard Biener  <rguenther@suse.de>
7924         PR tree-optimization/57685
7925         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7926         single-use operands to avoid exponential complexity.
7928 2013-08-28  Dehao Chen  <dehao@google.com>
7930         * ipa-inline.c (edge_badness): Fix integer underflow.
7932 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7934         * gtm-builtins.def (_ITM_free): Declare leaf.
7936 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7938         PR target/58067
7939         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7940         (*tls_local_dynamic_base_64_largepic): Likewise.
7941         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7942         Remove predicate from call operand.
7943         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7944         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7946 2013-08-28  Jeff Law  <law@redhat.com>
7948         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7949         checks for the number of predecessors and successors allowed.
7950         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7951         which require copying a joiner block if there is a request which
7952         is a subpath that requires no joiner block copying.
7954 2013-08-28  Jan Hubicka  <jh@suse.cz>
7956         * lto-streamer-out.c (DFS_write_tree_body): Drop
7957         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
7958         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
7959         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
7960         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
7961         Do not read DECL_ERROR_ISSUED.
7962         (unpack_ts_decl_with_vis_value_fields): Do not read
7963         DECL_DEFER_OUTPUT.
7964         (lto_input_ts_binfo_tree_pointers): Do not read
7965         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
7966         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
7967         write DECL_ERROR_ISSUED..
7968         (pack_ts_decl_with_vis_value_fields): Do not write
7969         DECL_DEFER_OUTPUT.
7970         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
7971         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
7972         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
7973         * tree.h (tree_decl_common): Update comment.
7974         (DECL_ERROR_ISSUED): Remove.
7976 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7978         PR middle-end/58257
7979         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
7981 2013-08-28  Jan Hubicka  <jh@suse.cz>
7983         * builtins.def (free): Declare leaf.
7985 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
7987         * gdbhooks.py: New.
7988         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
7989         * configure: Regenerate.
7991 2013-08-27  Martin Jambor  <mjambor@suse.cz>
7993         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
7994         (ipa_ancestor_jf_data): Likewise.
7995         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
7996         (ipa_get_jf_pass_through_type_preserved): New function.
7997         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
7998         (ipa_get_jf_ancestor_type_preserved): New function.
7999         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8000         (ipa_get_jf_ancestor_result): Likewise.
8001         (propagate_vals_accross_pass_through): Use
8002         ipa_get_jf_pass_through_result to do all the value mappings.
8003         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8004         type_preserved flag.
8005         (ipa_set_jf_cst_copy): New function.
8006         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8007         (ipa_set_jf_arith_pass_through): Likewise.
8008         (ipa_set_ancestor_jf): Likewise.
8009         (compute_complex_assign_jump_func): Set type_preserved instead of
8010         punting.
8011         (ipa_compute_jump_functions_for_edge): Likewise.
8012         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8013         (update_jump_functions_after_inlining): Update type_preserved.
8014         Explicitely create jump functions when combining one with pass_through.
8015         (ipa_write_jump_function): Stream the type_preserved flags.
8016         (ipa_read_jump_function): Likewise.
8018 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8019             Aldy Hernandez  <aldyh@redhat.com>
8021         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8022         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8023         * function.h (struct function): Add has_force_vect_loops and
8024         has_simduid_loops.
8025         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8026         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8027         handle it.
8028         * gimple.def: Update CLAUSES comments.
8029         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8030         (gimple_build_omp_for): Add argument to prototype.
8031         (gimple_omp_for_kind): New.
8032         (gimple_omp_for_set_kind): New.
8033         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8034         GOVD_DATA_SHARE_CLASS.
8035         (enum omp_region_type): Add ORT_SIMD.
8036         (gimple_add_tmp_var): Handle ORT_SIMD.
8037         (gimplify_var_or_parm_decl): Same.
8038         (is_gimple_stmt): Same.
8039         (omp_firstprivatize_variable): Same.
8040         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8041         (omp_notice_variable): Handle ORT_SIMD.
8042         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8043         (omp_check_private): Handle ORT_SIMD.
8044         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8045         OMP_CLAUSE_SAFELEN.
8046         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8047         Handle OMP_CLAUSE_LASTPRIVATE.
8048         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8049         OMP_CLAUSE_SAFELEN.
8050         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8051         (gimplify_expr): Handle OMP_SIMD.
8052         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8053         (expand_GOMP_SIMD_VF): New.
8054         (expand_GOMP_SIMD_LAST_LANE): New.
8055         * internal-fn.def (GOMP_SIMD_LANE): New.
8056         (GOMP_SIMD_VF): New.
8057         (GOMP_SIMD_LAST_LANE): New.
8058         * omp-low.c: Include target.h.
8059         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8060         OMP_CLAUSE_SAFELEN.
8061         (check_omp_nesting_restrictions): Same.
8062         (omp_max_vf): New.
8063         (lower_rec_simd_input_clauses): New.
8064         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8065         OMP_CLAUSE_LINEAR.
8066         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8067         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8068         (expand_omp_build_assign): New.
8069         (expand_omp_for_init_counts): New.
8070         (expand_omp_for_init_vars): New.
8071         (extract_omp_for_update_vars): New.
8072         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8073         and rewrite accordingly.
8074         (expand_omp_simd): New.
8075         (expand_omp_for): Use expand_omp_simd.
8076         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8077         (lower_omp_for): Do not lower the body.
8078         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8079         in their own loops.
8080         * tree-flow.h (find_omp_clause): Remove prototype.
8081         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8082         forcing vectorization of the loop, or if flag_tree_vectorize.
8083         (gate_tree_if_conversion): Similarly.
8084         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8085         gimple_build_omp_for.
8086         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8087         * tree-parloops (create_parallel_loop): Pass kind argument to
8088         gimple_build_omp_for.
8089         * tree-pretty-print.c (dump_omp_clause): Add cases for
8090         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8091         OMP_CLAUSE__SIMDUID_.
8092         (dump_generic_node): Handle OMP_SIMD.
8093         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8094         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8095         unroll OMP_SIMD loops here.
8096         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8097         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8098         loop->safelen.
8099         (vect_analyze_data_refs): Handle simd loops.
8100         * tree-vect-loop.c (vectorizable_live_operation): Handle
8101         IFN_GOMP_SIMD*.
8102         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8103         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8104         (vectorizable_load): Same.
8105         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8106         (struct simduid_to_vf): New.
8107         (simduid_to_vf::hash): New.
8108         (simduid_to-vf::equal): New.
8109         (struct simd_array_to_simduid): New.
8110         (simd_array_to_simduid::hash): New.
8111         (simd_array_to_simduid::equal): New.
8112         (adjust_simduid_builtins): New.
8113         (struct note_simd_array_uses_struct): New.
8114         (note_simd_array_uses_cb): New.
8115         (note_simd_array_uses): New.
8116         (vectorize_loops): Handle simd hints and adjust simd builtins
8117         accordingly.
8118         * tree-vectorizer.h (struct _stmt_vec_info): Add
8119         simd_lane_access_p field.
8120         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8121         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8122         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8123         (omp_clause_code_name): Same.
8124         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8125         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8126         * tree.def (OMP_SIMD): New entry.
8127         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8128         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8129         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8130         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8131         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8132         (OMP_CLAUSE_LINEAR_STEP): New.
8133         (OMP_CLAUSE_SAFELEN_EXPR): New.
8134         (OMP_CLAUSE__SIMDUID__DECL): New.
8135         (find_omp_clause): New prototype.
8137 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8139         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8140         Haswell processor detection.
8142 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8144         PR target/57927
8145         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8146         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8147         AVX2 capable processors.
8149 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8151         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8152         of vget_low_* with implementations in terms of other intrinsics.
8154 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8156         PR middle-end/57219
8157         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8158         values to -1, 0 and 1.
8160 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8162         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8163         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8164         (<optab><mode>3_insn): Remove.
8165         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8166         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8167         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8168         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8169         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8170         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8171         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8172         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8173         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8174         (ror<mode>3_insn): Likewise.
8175         * config/aarch64/predicates.md (aarch64_simd_register): New.
8177 2013-08-27  Richard Biener  <rguenther@suse.de>
8179         PR tree-optimization/57521
8180         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8181         one edge is non-critical.
8182         (find_phi_replacement_condition): Make sure to use a non-critical
8183         edge.  Cleanup and remove old bug workarounds.
8184         (bb_postdominates_preds): Remove.
8185         (if_convertible_loop_p_1): Do not compute post-dominators.
8186         (combine_blocks): Do not free post-dominators.
8187         (main_tree_if_conversion): Likewise.
8188         (pass_data_if_conversion): Add TODO_verify_ssa.
8190 2013-08-27  DJ Delorie  <dj@redhat.com>
8192         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8194 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8196         * function.c (assign_parm_find_data_types): Set passed_mode and
8197         nominal_mode to the TYPE_MODE of nominal_type for the built
8198         pointer type in case of the struct-pass-by-reference.
8200 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8202         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8203         (UINT16_TYPE): Change default to "unsigned int".
8205         * config/avr/avr.opt (mfract-convert-truncate): New option.
8206         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8207         is set, round negative fractional integers according to n1169
8208         when converting to integer types.
8210 2013-08-26  Jan Hubicka  <jh@suse.cz>
8212         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8213         methods can not be called indirectly when their address is not taken.
8215 2013-08-26  Jan Hubicka  <jh@suse.cz>
8217         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8218         ctor_for_folding.
8220 2013-08-26  Jan Hubicka  <jh@suse.cz>
8222         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8223         can be unshared.
8225 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8227         * reload.c (find_valid_class): Allow classes that do not include
8228         FIRST_PSEUDO_REGISTER - 1.
8230 2013-08-26  Jan Hubicka  <jh@suse.cz>
8232         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8233         fix edge count/frequency when speculation failed; fix type check
8234         for the direct call.
8236 2013-08-26  Jan Hubicka  <jh@suse.cz>
8238         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8240 2013-08-26  Jan Hubicka  <jh@suse.cz>
8242         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8243         to be changed by edge redirection.
8245 2013-08-26  Jan Hubicka  <jh@suse.cz>
8247         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8248         formating; add sanity check.
8249         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8250         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8251         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8252         (resolve_noninline_speculation): Update callee keys, too.
8254 2013-08-26  Jan Hubicka  <jh@suse.cz>
8256         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8257         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8259 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8261         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8262         into proper place.
8264 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8266         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8267         (x86_64_elf_select_section): Ditto.
8268         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8269         arguments.
8270         (ix86_pass_by_reference): Ditto.
8271         (output_set_got): Ditto.
8272         (ix86_unary_operator_ok): Ditto.
8273         (ix86_expand_builtin): Ditto.
8275 2013-08-23  Jan Hubicka  <jh@suse.cz>
8277         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8279 2013-08-23  Jan Hubicka  <jh@suse.cz>
8281         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8282         (tree_decl_with_vis): Add FINAL field.
8284 2013-08-23  Jeff Law  <law@redhat.com>
8286         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8287         the debugging dump when the expression is fully redundant.
8289 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8291         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8292         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8293         * pretty-print.c (pp_formatted_text_data): Likewise.
8294         (pp_write_text_to_stream): Likewise.
8295         (pp_write_text_as_dot_label_to_stream): Likewise.
8296         (pp_append_r): Likewise.
8297         (pp_format): Likewise.
8298         (pp_flush): Likewise.
8299         (pp_clear_output_area): Likewise.
8300         (pp_append_text): Likewise.
8301         (pp_formatted_text): Likewise.
8302         (pp_remaining_character_count_for_line): Likewise.
8303         (pp_newline): Likewise.
8304         (pp_character): Likewise.
8305         (output_buffer::~output_buffer): Define.
8306         (pretty_printer::~pretty_printer): Destruct output buffer.
8307         * pretty-print.h (output_buffer::~output_buffer): Declare.
8308         (pretty_printer::~pretty_printer): Declare virtual.
8310 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8312         PR other/57324
8313         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8314         HOST_WIDE_INT_M1U): New macros.
8315         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8316         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8317         unsigned -1 for lshift.
8318         * cse.c (cse_insn): Likewise.
8319         * double-int.c (rshift_double, lshift_double): Likewise.
8320         * builtins.c (fold_builtin_bitop): Likewise.
8321         * combine.c (force_to_mode): Likewise.
8322         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8323         * simplify-rtx.c (simplify_const_unary_operation,
8324         simplify_const_binary_operation): Likewise.
8325         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8326         check_va_list_escapes): Likewise.
8327         * rtlanal.c (nonzero_bits1): Likewise.
8328         * expmed.c (expand_smod_pow2): Likewise.
8329         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8331 2013-08-23  Jan Hubicka  <jh@suse.cz>
8333         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8334         as having address taken.
8336 2013-08-23  Jan Hubicka  <jh@suse.cz>
8338         * ipa-utils.h (method_class_type): Declare.
8339         * ipa-devirt.c (method_class_type): Export.
8341         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8342         do not walk base classes of anonymous types.
8344 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8346         PR rtl-optimization/58220
8347         PR regression/58221
8348         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8349         handle SEQUENCE insns properly.
8351 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8353         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8354         definition.
8355         (pp_newline_and_indent): Likewise.
8356         (pp_separate_with): Likewise.
8357         * pretty-print.c (pp_newline_and_flush): Define.
8358         (pp_newline_and_indent): Likewise.
8359         (pp_separate_with): Likewise.
8361 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8363         PR target/58218
8364         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8365         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8367 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8369         * config/i386/predicates.md (ext_sse_reg_operand): New.
8370         * config/i386/i386.md (*movti_internal): Use
8371         predicate to determine if EVEX is needed.
8372         (*movsi_internal): Ditto.
8373         (*movdf_internal): Ditto.
8374         (*movsf_internal): Ditto.
8375         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8377 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8379         PR tree-optimization/58209
8380         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8381         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8382         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8383         emit POINTER_PLUS_EXPR.
8384         (create_tailcall_accumulator): For pointer result type accumulate in
8385         sizetype type.
8387 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8389         * configure.ac: Add backslashes missing from the last change.
8390         * configure: Regenerate.
8392 2013-08-22  Jan Hubicka  <jh@susue.cz>
8394         * ipa.c (function_and_variable_visibility): First remember function
8395         was global and then make it local.
8397 2013-08-22  Julian Brown  <julian@codesourcery.com>
8399         * configure.ac: Add aarch64 to list of arches which use "nop" in
8400         debug_line test.
8401         * configure: Regenerate.
8403 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8405         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8406         gnu_libc_has_function.
8407         * config/s390/tpf.h: Likewise.
8409 2013-08-22  Jan Hubicka  <jh@susue.cz>
8411         * timevar.c (validate_phases): Add cast.
8413 2013-08-22  Jan Hubicka  <jh@susue.cz>
8415         * timevar.c (validate_phases): Use size_t for memory.
8416         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8418 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8420         * pretty-print.h (output_buffer::output_buffer): Declare.
8421         (pretty_printer::pretty_printer): Likewise.
8422         (pp_construct): Remove.
8423         * pretty-print.c (output_buffer::output_buffer): Define.
8424         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8425         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8426         (print_gimple_expr): Likewise.
8427         (print_gimple_seq): Likewise.
8428         (gimple_dump_bb): Likewise.
8429         * sched-vis.c (dump_value_slim): Likewise.
8430         (dump_insn_slim): Likewise.
8431         (dump_rtl_slim): Likewise.
8432         (str_pattern_slim): Likewise.
8433         * tree-mudflap.c (mf_varname_tree): Likewise.
8434         * graph.c (print_graph_cfg): Likewise.
8435         (start_graph_dump): Likewise.
8436         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8437         placement-new.
8438         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8439         pretty printer initialization.
8440         * coretypes.h (diagnostic_context): Remove superflous type alias
8441         declaration.
8442         (pretty_printer): Likewise.  Declare directly as a class.
8443         (pretty_print_info): Remove declaration as class.
8444         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8445         and pp_clear_output_area.
8446         (asan_add_global): Likewise.
8448 2013-08-22  Jan Hubicka  <jh@suse.cz>
8450         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8451         * ipa-utils.h (update_type_inheritance_graph): Declare.
8452         (possible_polymorphic_call_target_p): Declare.
8453         (possible_polymorphic_call_target_p): New.
8454         * ipa-devirt.c: Update toplevel comments.
8455         (cached_polymorphic_call_targets): Move up.
8456         (odr_type_d): Move ID down.
8457         (polymorphic_type_binfo_p): Update comment.
8458         (odr_hasher::remove): Likewise;
8459         (get_odr_type): Set anonymous_namespace.
8460         (dump_odr_type): Dump it.
8461         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8462         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8463         (record_binfo): Add comment.
8464         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8465         built.
8466         (devirt_node_removal_hook): Do not iCE when cache is freed.
8467         (possible_polymorphic_call_target_p): New predicate.
8468         (update_type_inheritance_graph): New function.
8470 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8471             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8472             Sergey Lega  <sergey.s.lega@intel.com>
8473             Anna Tikhonova  <anna.tikhonova@intel.com>
8474             Ilya Tocar  <ilya.tocar@intel.com>
8475             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8476             Ilya Verbin  <ilya.verbin@intel.com>
8477             Kirill Yukhin  <kirill.yukhin@intel.com>
8478             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8480         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8481         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8482         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8483         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8484         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8485         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8486         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8487         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8488         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8489         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8490         OPT_mavx512pf, OPT_mavx512er cases.
8491         * config/i386/constraints.md (v): New constraint.
8492         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8493         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8494         (bit_AVX512CD): New.
8495         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8496         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8497         * config/i386/i386-c.c (ix86_target_macros_internal):
8498         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8499         __AVX512PF__.
8500         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8501         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8502         * config/i386/i386.c (regclass_map, dbx_register_map)
8503         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8504         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8505         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8506         -mavx512pf options.
8507         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8508         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8509         -mavx512pf options.  Fix formatting.
8510         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8511         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8512         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8513         -mavx512cd, -mavx512pf options.
8514         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8515         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8516         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8517         ALL_SSE_REGS.
8518         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8519         (ix86_set_reg_reg_cost): Ditto.
8520         (x86_order_regs_for_local_alloc): Ditto.
8521         (MAX_VECT_LEN): Extend to 64-byte.
8522         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8523         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8524         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8525         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8526         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8527         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8528         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8529         (SSE_REG_MODE_P): Support new modes.
8530         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8531         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8532         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8533         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8534         (REG_CLASS_CONTENTS): Add new registers.
8535         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8536         (EXT_REX_SSE_REGNO_P): New.
8537         (HI_REGISTER_NAMES): Add new registers.
8538         * config/i386/i386.md: Define constants for new registers.
8539         (mode): Add new 512-bit modes.
8540         (prefix): Support evex prefix.
8541         (isa): Support avx512f, noavx512f, fma_avx512f.
8542         (ssemodesuffix): Add new 512-bit modes.
8543         (movxi): New.
8544         (*movxi_internal_avx512f): Ditto.
8545         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8546         Support MODE_XI.
8547         (*movsi_internal): Likewise.
8548         (*movdf_internal): Likewise.
8549         (*movsf_internal): Likewise.
8550         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8551         (<code><mode>3): Likewise.
8552         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8553         New.
8554         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8555         with the new constraint "v".
8556         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8557         modes.
8558         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8559         with the new constraint "v".
8560         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8561         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8562         (*<plusminus_insn><mode>3): Likewise.
8563         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8564         (*mul<mode>3): Likewise.
8565         (<sse>_vmmul<mode>3): Likewise.
8566         (<sse>_div<mode>3): Likewise.
8567         (<sse>_vmdiv<mode>3): Likewise.
8568         (<sse>_sqrt<mode>2): Likewise.
8569         (<sse>_vmsqrt<mode>2): Likewise.
8570         (*<code><mode>3_finite): Likewise.
8571         (*<code><mode>3) <smaxmin>: Likewise.
8572         (<sse>_vm<code><mode>3): Likewise.
8573         (*<code><mode>3) <any_logic>: Likewise.
8574         (*fma_fmadd_<mode>): Likewise.
8575         (*fma_fmsub_<mode>): Likewise.
8576         (*fma_fnmadd_<mode>): Likewise.
8577         (*fma_fnmsub_<mode>): Likewise.
8578         (*fma_fmaddsub_<mode>): Likewise.
8579         (*fma_fmsubadd_<mode>): Likewise.
8580         (*fmai_fmadd_<mode>): Likewise.
8581         (*fmai_fmsub_<mode>): Likewise.
8582         (*fmai_fnmadd_<mode>): Likewise.
8583         (*fmai_fnmsub_<mode>): Likewise.
8584         (sse_cvtsi2ss): Likewise.
8585         (sse_cvtsi2ssq): Likewise.
8586         (sse_cvtss2si): Likewise.
8587         (sse_cvtss2si_2): Likewise.
8588         (sse_cvtss2siq): Likewise.
8589         (sse_cvtss2siq_2): Likewise.
8590         (sse_cvttss2si): Likewise.
8591         (sse_cvtss2siq_2): Likewise.
8592         (float<sseintvecmodelower><mode>2): Likewise.
8593         (sse2_cvtsd2si_2): Likewise.
8594         (sse2_cvtsd2siq_2): Likewise.
8595         (*<plusminus_insn><mode>3): Likewise.
8596         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8597         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8598         (ashr<mode>3): Likewise.
8599         (<shift_insn><mode>3): Likewise.
8600         (avx2_<code><mode>3): Likewise.
8601         (*avx2_<code><mode>3): Likewise.
8602         (*andnot<mode>3): Likewise.
8603         (*<code><mode>3) <any_logic>: Likewise.
8604         (abs<mode>2): Likewise.
8605         (avx2_permvar<mode>): Likewise.
8606         (avx2_perm<mode>_1): Likewise.
8607         (*avx_vpermilp<mode>): Likewise.
8608         (avx_vpermilvar<mode>3): Likewise.
8609         (avx2_ashrv<mode>): Likewise.
8610         (avx2_<shift_insn>v<mode>): Likewise.
8611         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8612         -mavx512cd.
8613         * doc/rtl.texi: Document XImode.
8615 2013-08-21  Jeff Law  <law@redhat.com>
8617         * tree-flow.h (register_jump_thread): Pass vector of edges
8618         instead of each important edge.
8619         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8620         thread path into a vector and pass that to register_jump_thread.
8621         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8622         passed in edge vector to the current 3-edge form.
8624         Revert:
8625         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8627         * dce.c (fini_dce): Call df_analyze again just in case
8628         delete_unmarked_insns removed anything.
8630 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8632         * reload.h (struct reg_equivs): Rename to ..
8633         (struct reg_equivs_s): .. this.
8635 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8637         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8639 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8641         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8643 2013-08-21  Jeff Law  <law@redhat.com>
8645         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8646         simplify assignments too.  If the RHS collapses to a singleton
8647         range, then return the value for the range.
8649 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8651         * config/i386/sse.md (V16): Rename to...
8652         (VMOVE): this.
8653         (mov<mode>): Update iterator name.
8654         (*mov<mode>_internal): Ditto.
8655         (push<mode>1): Ditto.
8656         (movmisalign<mode>): Ditto.
8658 2013-08-20  Jan Hubicka  <jh@suse.cz>
8660         PR bootstrap/58186
8661         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8662         entry for direct edges.
8663         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8665 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8667         Revert my last two changes, r201865 and r201864:
8669         Revert r201865:
8670         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8672         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8673         instances can own GC refs.
8675         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8676         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8677         (gcc::context::gt_pch_nx): Likewise.
8678         (gcc::context::gt_pch_nx):  Likewise.
8679         * ggc.h (gt_ggc_mx <T>): New.
8680         (gt_pch_nx_with_op <T>): New.
8681         (gt_pch_nx <T>): New.
8682         * passes.c (opt_pass::gt_ggc_mx): New.
8683         (opt_pass::gt_pch_nx): New.
8684         (opt_pass::gt_pch_nx_with_op): New.
8685         (pass_manager::gt_ggc_mx): New.
8686         (pass_manager::gt_pch_nx): New.
8687         (pass_manager::gt_pch_nx_with_op): New.
8688         (pass_manager::operator new): Use
8689         ggc_internal_cleared_alloc_stat rather than xcalloc.
8690         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8691         (pass_manager::gt_ggc_mx): New.
8692         (pass_manager::gt_pch_nx): New.
8693         (pass_manager::gt_pch_nx_with_op): New.
8694         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8695         (opt_pass::operator new): New.
8696         (opt_pass::gt_ggc_mx): New.
8697         (opt_pass::gt_pch_nx): New.
8698         (opt_pass::gt_pch_nx_with_op): New.
8700         Revert r201864:
8701         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8703         * Makefile.in (GTFILES): Add context.h.
8704         * context.c (gcc::context::operator new): New.
8705         (gcc::context::gt_ggc_mx): New.
8706         (gcc::context::gt_pch_nx): New.
8707         (gcc::context::gt_pch_nx): New.
8708         * context.h (gcc::context): Add GTY((user)) marking.
8709         (gcc::context::operator new): New.
8710         (gcc::context::gt_ggc_mx): New.
8711         (gcc::context::gt_pch_nx): New.
8712         (gcc::context::gt_pch_nx): New.
8713         (g): Add GTY marking.
8714         (gt_ggc_mx (gcc::context *)): New.
8715         (gt_pch_nx (gcc::context *)): New.
8716         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8717         void *cookie)): New.
8718         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8720 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8722         * dce.c (fini_dce): Call df_analyze again just in case
8723         delete_unmarked_insns removed anything.
8725 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8727         PR rtl-optimizations/57451
8728         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8729         from crossing split section boundaries.
8731 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8733         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8734         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8735         with MULTLIB_DEFAULTS.
8737 2013-08-20  Nick Clifton  <nickc@redhat.com>
8739         * target.def (narrow_volatile_bitfield): Note that the default
8740         value is false, not !TARGET_STRICT_ALIGN.
8741         * doc/tm.texi: Regenerate.
8743 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8745         Fix LIB_SPEC for systems without libpthread.
8747         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8748         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8749         for Android.
8750         * config/i386/linux-common.h: Likewise.
8751         * config/mips/linux-common.h: Likewise.
8753 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8755         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8756         checks.
8758 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8760         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8761         instances can own GC refs.
8763         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8764         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8765         (gcc::context::gt_pch_nx): Likewise.
8766         (gcc::context::gt_pch_nx):  Likewise.
8767         * ggc.h (gt_ggc_mx <T>): New.
8768         (gt_pch_nx_with_op <T>): New.
8769         (gt_pch_nx <T>): New.
8770         * passes.c (opt_pass::gt_ggc_mx): New.
8771         (opt_pass::gt_pch_nx): New.
8772         (opt_pass::gt_pch_nx_with_op): New.
8773         (pass_manager::gt_ggc_mx): New.
8774         (pass_manager::gt_pch_nx): New.
8775         (pass_manager::gt_pch_nx_with_op): New.
8776         (pass_manager::operator new): Use
8777         ggc_internal_cleared_alloc_stat rather than xcalloc.
8778         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8779         (pass_manager::gt_ggc_mx): New.
8780         (pass_manager::gt_pch_nx): New.
8781         (pass_manager::gt_pch_nx_with_op): New.
8782         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8783         (opt_pass::operator new): New.
8784         (opt_pass::gt_ggc_mx): New.
8785         (opt_pass::gt_pch_nx): New.
8786         (opt_pass::gt_pch_nx_with_op): New.
8788 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8790         * Makefile.in (GTFILES): Add context.h.
8791         * context.c (gcc::context::operator new): New.
8792         (gcc::context::gt_ggc_mx): New.
8793         (gcc::context::gt_pch_nx): New.
8794         (gcc::context::gt_pch_nx): New.
8795         * context.h (gcc::context): Add GTY((user)) marking.
8796         (gcc::context::operator new): New.
8797         (gcc::context::gt_ggc_mx): New.
8798         (gcc::context::gt_pch_nx): New.
8799         (gcc::context::gt_pch_nx): New.
8800         (g): Add GTY marking.
8801         (gt_ggc_mx (gcc::context *)): New.
8802         (gt_pch_nx (gcc::context *)): New.
8803         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8804         void *cookie)): New.
8805         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8807 2013-08-20  Alan Modra  <amodra@gmail.com>
8809         PR target/57865
8810         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8811         (rs6000_emit_epilogue): Likewise.
8813 2013-08-19  Dehao Chen  <dehao@google.com>
8815         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8817 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8818             Jakub Jelinek  <jakub@redhat.com>
8820         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8821         (BUILT_IN_FABSD64): Likewise.
8822         (BUILT_IN_FABSD128): Likewise.
8823         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8824         (fold_builtin_1): Likewise.
8825         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8826         destination and source operands.
8827         (*abstd2_fpr): Likewise.
8828         (*nabstd2_fpr): Likewise.
8830 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8832         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8833         JUMP_P and INSN_P.
8835 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8837         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8838         incomplete.
8840 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8842         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8843         * builtins.c (default_libc_has_function): New.
8844         (gnu_libc_has_function): Ditto.
8845         (no_c99_libc_has_function): Ditto.
8846         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8847         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8848         (fold_builtin_sincos): Likewise.
8849         (fold_builtin_cexp): Likewise.
8850         * builtins.def (DEF_C94_BUILTIN): Likewise.
8851         (DEF_C99_BUILTIN): Likewise.
8852         (DEF_C99_C90RES_BUILTIN): Likewise.
8853         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8854         definitions to using this define.
8855         * config/darwin-protos.h (darwin_libc_has_function): New.
8856         * config/darwin.c (darwin_libc_has_function): Ditto.
8857         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8858         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8859         * config/darwin.h: Ditto.
8860         * config/elfos.h: Ditto.
8861         * config/freebsd.h: Ditto.
8862         * config/i386/cygming.h: Ditto.
8863         * config/i386/djgpp.h: Ditto.
8864         * config/i386/i386-interix.h: Ditto.
8865         * config/microblaze/microblaze.h: Ditto.
8866         * config/mmix/mmix.h: Ditto.
8867         * config/gnu-user.h: Ditto.
8868         * config/ia64/hpux.h: Ditto.
8869         * config/pa/pa-hpux.h: Ditto.
8870         * config/pdp11/pdp11.h: Ditto.
8871         * config/picochip/picochip.h: Ditto.
8872         * config/linux.h: Ditto.
8873         * config/netbsd.h: Ditto.
8874         * config/openbsd.h: Ditto.
8875         * config/rs6000/aix43.h: Ditto.
8876         * config/rs6000/aix51.h: Ditto.
8877         * config/rs6000/aix52.h: Ditto.
8878         * config/rs6000/aix53.h: Ditto.
8879         * config/rs6000/aix61.h: Ditto.
8880         * config/rs6000/darwin.h: Ditto.
8881         * config/rs6000/linux.h: Ditto.
8882         * config/rs6000/linux64.h: Ditto.
8883         * config/s390/tpf.h: Ditto.
8884         * config/sol2-10.h: Ditto.
8885         * config/sol2.h: Ditto.
8886         * config/vms/vms.h: Ditto.
8887         * config/vxworks.h: Ditto.
8888         * config/linux-android.c (linux_android_libc_has_function):
8889         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8890         * config/linux-protos.h (linux_android_libc_has_function):
8891         New declaration.
8892         * config/i386/i386.c (ix86_libc_has_function): New.
8893         * config/i386/i386-protos.h
8894         (ix86_libc_has_function): New declaration.
8895         * config/i386/i386.md
8896         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8897         ("isinf<mode>2): Likewise.
8898         * convert.c (convert_to_integer): Using new target hook
8899         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8900         TARGET_C99_FUNCTIONS.
8901         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8902         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8903         * coretypes.h (function_class): New enum for different
8904         classes of functions.
8905         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8906         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8907         (TARGET_HAS_SINCOS): Likewise.
8908         (TARGET_LIBC_HAS_FUNCTION): New.
8909         * doc/tm.texi: Regenerated.
8910         * targhooks.h (default_libc_has_function): New declaration.
8911         (no_c99_libc_has_function): Ditto.
8912         (gnu_libc_has_function): Ditto.
8913         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8914         and TARGET_HAS_SINCOS.
8916 2013-08-18  Jan Hubicka  <jh@suse.cz>
8918         * Makeifle-in (ipa-devirt.o): New.
8919         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8920         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8921         (analyze_functions): Look into possible targets of polymorphic call.
8922         * dumpfile.c (dump_files): Add type-inheritance dump.
8923         * dumpfile.h (TDI_inheritance): New.
8924         * ipa-devirt.c: New file.
8925         * ipa-utils.h (odr_type_d): Forward declare.
8926         (odr_type): New type.
8927         (build_type_inheritance_graph): Declare.
8928         (possible_polymorphic_call_targets): Declare and introduce inline
8929         variant when only edge is pased.
8930         (dump_possible_polymorphic_call_targets): Likewise.
8931         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8932         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8933         (types_same_for_odr): ... here.
8934         * tree.h (type_in_anonymous_namespace_p): Declare.
8936 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8938         PR tree-optimization/58006
8939         * tree-parloops.c (take_address_of): Don't ICE if get_name
8940         returns NULL.
8941         (eliminate_local_variables_stmt): Remove clobber stmts.
8943 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8945         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8946         error message is issued for an alias to undefined symbol.
8948 2013-08-18  Jan Hubicka  <jh@suse.cz>
8950         * cgraph.c (cgraph_create_indirect_edge): Discover
8951         polymorphic calls and record basic info into indirect_info.
8952         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8953         devirtualization, ignore objc function calls.
8954         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
8955         call with no parm index info.
8956         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
8957         * tree.c (virtual_method_call_p): New function.
8958         * tree.h (virtual_method_call_p): Declare.
8960 2013-08-16  Jan Hubicka  <jh@suse.cz>
8962         PR middle-end/58179
8963         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
8965 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
8967         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
8968         attribute.
8970 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8972         * gengtype.c (type_for_name): Add special-case support for
8973         locating types within the "gcc::" namespace.
8974         (open_base_files): Emit a "using namespace gcc" directive.
8976 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8978         PR target/58160
8979         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
8980         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
8982         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
8983         array instead of each individual operand as a separate argument.
8984         (emit_fusion_gpr_load): Likewise.
8985         (expand_fusion_gpr_load): Add new function declaration.
8987         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
8988         signature to have the operands passed as an array, instead of as
8989         separate arguments.  Allow ZERO_EXTEND to be in the memory
8990         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
8991         depend on the register live/dead flags when peepholes are run.
8992         (expand_fusion_gpr_load): New function to be called from the
8993         peephole2 pass, to change the register that addis sets to be the
8994         target register.
8995         (emit_fusion_gpr_load): Change the calling signature to have the
8996         operands passed as an array, instead of as separate arguments.
8997         Allow ZERO_EXTEND to be in the memory address, and also
8998         SIGN_EXTEND if -mpower8-fusion-sign.
9000         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9001         unspec enumeration.
9002         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9003         adjust the register addis loads up in the peephole2 pass.  Do not
9004         depend on the register live/dead state when the peephole pass is done.
9006 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9008         * gengtype.c (create_user_defined_type): Ensure that the kind
9009         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9010         declaration is seen before the GTY((user)) marking.
9012 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9014         PR target/58105
9015         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9017 2013-08-16  Jan Hubicka  <jh@suse.cz>
9019         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9020         arugment expected_type.
9021         (gimple_fold_call): Use it.
9022         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9023         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9024         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9025         (try_make_edge_direct_virtual_call): Likewise.
9026         * tree.c (obj_type_ref_class): New.
9027         * tree.h (obj_type_ref_class): Use it.
9029 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9031         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9032         (rtl_slim_pp): Likewise.
9033         (init_rtl_slim_pretty_print): Likewise.
9034         (dump_value_slim):  Don't call it.  Use local pretty printer.
9035         (dump_insn_slim): Likewise.
9036         (dump_rtl_slim): Likewise.
9037         (str_pattern_slim): Likewise.
9038         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9039         Simplify.
9041 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9043         PR tree-optimization/58164
9044         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9045         walk gimple_goto_dest of GIMPLE_GOTO.
9047         PR tree-optimization/58165
9048         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9049         bi_call must be the last stmt in a bb, don't split_block, instead
9050         use fallthru edge from it and give up if there is none.
9051         Release conds vector when returning early.
9053 2013-08-14  Xinliang David Li  <davidxl@google.com>
9055         * config/i386/i386.c (ix86_option_override_internal):
9056         Remove unused variable and field.
9058 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9060         PR target/57949
9061         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9062         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9063         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9064         and Linux, correct BLKmode alignment when 128-bit alignment is
9065         required and compatibility flag is not set.
9066         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9067         for zero-size arguments when compatibility flag is not set.
9069 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9071         PR tree-optimization/58145
9072         * tree-sra.c (build_ref_for_offset): If prev_base has
9073         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9075 2013-08-14  Xinliang David Li  <davidxl@google.com>
9077         * config/i386/i386.c (ix86_option_override_internal):
9078         Fix uninitialized variable error.
9080 2013-08-14  Xinliang David Li  <davidxl@google.com>
9082         * config/i386/i386.opt: Define two new options.
9083         * config/i386/x86-tune.def: Add arch selector field in macros.
9084         * config/i386/i386.h: Adjust macro definition.
9085         * config/i386/i386.c (ix86_option_override_internal):
9086         Refactor the code.
9087         (parse_mtune_ctrl_str): New function.
9088         (set_ix86_tune_features): New function.
9089         (ix86_function_specific_restore): Call the new helper function.
9091 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9093         PR rtl-optimization/57662
9094         * sel-sched.c (code_motion_process_successors): When the current insn
9095         is removed after the recursive traversal, break from the loop.
9096         Add comments and debug printouts.
9098 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9099             Alexandre Oliva  <aoliva@redhat.com>
9101         PR target/58067
9102         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9103         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9104         there also UNSPEC_PLTOFF.
9106 2013-08-14  Marek Polacek  <polacek@redhat.com>
9108         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9109         NUM_CONDITIONS bit positions.
9111 2013-08-13  Cary Coutant  <ccoutant@google.com>
9113         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9114         (attr_checksum): Hash vector contents instead of pointer.
9115         (attr_checksum_ordered): Likewise.
9117 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9119         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9120         when Pmode != word_mode.  Add length_address attribute.
9121         (sse3_monitor_<mode>): Merge from sse3_monitor and
9122         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9123         Pmode != word_mode.  Update insn length attribute.
9124         * config/i386/i386.c (ix86_option_override_internal): Update
9125         ix86_gen_monitor selection for merged sse3_monitor insn.
9127 2013-08-13  Julian Brown  <julian@codesourcery.com>
9129         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9130         perform invalid legitimization on greater-than-word-size modes for
9131         TARGET_E500_DOUBLE.
9133 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9135         * ira.c (setup_class_translate_array): Use aclass instead of cl
9136         for classes not fully covered by allocno classes.
9138 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9140         PR tree-optimization/57661
9141         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9142         * tree-inline.c (tree_function_versioning): Initialize it.
9143         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9144         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9145         that is not being copied.
9147         PR sanitizer/56417
9148         * asan.c (instrument_strlen_call): Fix typo in comment.
9149         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9151 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9153         * config/mips/mips.md (prefetch): Use lw instead of ld on
9154         loongson in 32bit mode.
9156 2013-08-13  Nick Clifton  <nickc@redhat.com>
9158         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9160 2013-08-13  Jan Hubicka  <jh@suse.cz>
9162         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9163         introduced edge; fix typo in sanity check.
9164         (cgraph_resolve_speculation): Export; improve diagnostic.
9165         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9166         speculation at type mismatch.
9167         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9168         (cgraph_resolve_speculation): Declare.
9169         (symtab_can_be_discarded): New function.
9170         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9171         * ipa-inline-transform.c (speculation_removed): New global var.
9172         (clone_inlined_nodes): See if speculation can be removed.
9173         (inline_call): If speculations was removed, we growths may not match.
9174         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9175         (speculation_useful_p): New function.
9176         (resolve_noninline_speculation): New function.
9177         (inline_small_functions): Resolve useless speculations.
9178         * ipa-inline.h (speculation_useful_p): Declare
9179         * ipa.c (can_replace_by_local_alias): Simplify.
9180         (ipa_profile): Produce speculative calls in non-lto, too;
9181         add simple cost model; produce local aliases.
9183 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9185         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9186         PASS_MANAGER_H.
9188 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9190         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9192 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9194         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9195         iterator instead of X87MODEF.
9197 2013-08-12  Perez Read  <netfirewall@gmail.com>
9199         PR target/58132
9200         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9201         operand 0 for intel asm alternative.
9202         (*movabs<mode>_2): Ditto for operand 1.
9204 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9206         * config/aarch64/arm_none.h
9207         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9209 2013-08-12  Nick Clifton  <nickc@redhat.com>
9211         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9213 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9215         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9216         expand for QI/HImode operand to produce more effictive code for
9217         unsigned char(short) --> float(double) conversion.
9219 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9221         * doc/invoke.texi: Mention that -ftls-model does not force the final
9222         model.
9224 2013-08-12  Marek Polacek  <polacek@redhat.com>
9225             Marc Glisse  <marc.glisse@inria.fr>
9227         PR tree-optimization/57980
9228         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9229         when creating -1 constant.
9231 2013-08-10  Jan Hubicka  <jh@suse.cz>
9233         Workaround binutils PR14342.
9234         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9235         (gimple_init_edge_profiler): Likewise.
9236         (gimple_gen_ic_func_profiler): Likewise.
9238 2013-08-09  Jan Hubicka  <jh@suse.cz>
9240         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9242 2013-08-09  Xinliang David Li  <davidxl@google.com>
9244         * config/i386/stringop.def: New file.
9245         * config/i386/stringop.opt: New file.
9246         * config/i386/i386-opts.h: Include stringopt.def.
9247         * config/i386/i386.opt: Include stringopt.opt.
9248         * config/i386/i386.c (ix86_option_override_internal):
9249         Override default size based stringop inline strategies with options.
9250         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9251         New function.
9253 2013-08-09  Jan Hubicka  <jh@suse.cz>
9255         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9257 2013-08-09  Jan Hubicka  <jh@suse.cz>
9259         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9260         CGRAPH_FREQ_MAX.
9261         (dump_cgraph_node): Dump profile-id.
9262         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9263         and common_target_probability.
9264         * lto-cgraph.c (lto_output_edge): Stream common targets.
9265         (lto_output_node): Stream profile ids.
9266         (input_node): Stream profile ids.
9267         (input_edge): Stream common targets.
9268         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9269         * ipa.c: Include value-prof.h
9270         (ipa_profile_generate_summary): Turn indirect call statement histograms
9271         into common targets.
9272         (ipa_profile): Turn common targets into speculative edges.
9274 2013-08-09  Jan Hubicka  <jh@suse.cz>
9276         * cgraph.h (cgraph_node): Add profile_id.
9277         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9278         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9279         (del_node_map): Update.
9280         (find_func_by_funcdef_no): Replace by ...
9281         (find_func_by_profile_id): ... this one.
9282         (gimple_ic_transform): Do not remove useful histograms when
9283         speculation is not done; dump info when indirect call removal
9284         can happen at LTO.
9285         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9286         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9287         (__gcov_indirect_call_profiler_v2): .. this one.
9288         * profile.h (init_node_map): Update.
9289         * coverage.c (coverage_compute_profile_id): New function.
9290         * coverage.h (coverage_compute_profile_id): Declare.
9291         * tree-profile.c (init_ic_make_global_vars): Make
9292         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9293         (gimple_init_edge_profiler): Update prototype of
9294         __gcov_indirect_call_profiler.
9295         (gimple_gen_ic_func_profiler): Simplify.
9296         (tree_profiling): Use init_node_map
9298 2013-08-09  Jan Hubicka  <jh@suse.cz>
9300         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9301         non-speculative refs.
9302         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9303         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9304         (cgraph_set_call_stmt): Likewise.
9305         (cgraph_create_edge_1): Fix release checking compilatoin;
9306         clear lto_stmt_uid.
9307         (cgraph_free_edge): Free indirect info.
9308         (cgraph_turn_edge_to_speculative): New function.
9309         (cgraph_speculative_call_info): New function.
9310         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9311         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9312         (dump_cgraph_node): Dump speculation.
9313         (verify_edge_count_and_frequency): Accept speculative edges.
9314         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9315         (verify_cgraph_node): Handle speculation.
9316         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9317         (cgraph_set_call_stmt): Update prototype.
9318         (cgraph_make_edge_direct): Update prototype.
9319         (cgraph_speculative_call_info): Declare.
9320         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9321         to change; update call of ipa_find_references.
9322         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9323         lto_stmt_uid and speculative flags.
9324         (ipa_dump_references): Dump speculation.
9325         (ipa_clone_references): Clone speculative flag.
9326         (ipa_clone_referring): Likewise.
9327         (ipa_clone_ref): New function.
9328         (ipa_find_reference): Look into lto_stmt_uids
9329         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9330         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9331         (ipa_find_reference): Update declaration.
9332         (ipa_clone_ref): Declare.
9333         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9334         stream speculative flag.
9335         (lto_output_ref): Stream statements uids and speculation.
9336         (input_ref): Likewise.
9337         (input_edge): Stream speuclation.
9338         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9339         (cgraph_set_call_stmt_including_clones): Handle speculation.
9340         * ipa-inline.c (heap_edge_removal_hook): New function.
9341         (inline_small_functions): Register it.
9342         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9343         also initialize refs.
9344         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9345         edge to change.
9346         (try_make_edge_direct_simple_call): Likewise.
9347         (try_make_edge_direct_simple_call): Likewise.
9348         (update_indirect_edges_after_inlining): Likewise.
9349         (remove_described_reference): Look proper lto_stmt_uid.
9350         (propagate_controlled_uses): Likewise.
9351         (propagate_controlled_uses): Liekwise.
9352         * tree-inline.c (copy_bb): Copy speculative edges.
9353         (redirect_all_calls): New function.
9354         (copy_cfg_body): Do redirection after loop info is updated.
9355         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9357 2013-08-09  Jan Hubicka  <jh@suse.cz>
9359         * lto-streamer-out.c (output_function): Renumber PHIs.
9360         * lto-streamer-in.c (input_function): Likewise.
9362 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9364         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9365         (get_lane_unsigned): Likewise.
9366         (dup_lane_scalar): Likewise.
9367         (get_lane): enable for VALL.
9368         * config/aarch64/aarch64-simd.md
9369         (aarch64_dup_lane_scalar<mode>): Remove.
9370         (aarch64_get_lane_signed<mode>): Likewise.
9371         (aarch64_get_lane_unsigned<mode>): Likewise.
9372         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9373         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9374         (aarch64_get_lane<mode>): Enable for all vector modes.
9375         (aarch64_get_lanedi): Remove misleading constraints.
9376         * config/aarch64/arm_neon.h
9377         (__aarch64_vget_lane_any): Define.
9378         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9379         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9380         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9381         * config/aarch64/iterators.md (VDQQH): New.
9382         (VDQQHS): Likewise.
9383         (vwcore): Likewise.
9385 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9387         * configure.ac: Add GAS check for LEON instructions on SPARC.
9388         * configure: Regenerate.
9389         * config.in: Likewise.
9390         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9391         sparc*-*-* block.
9392         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9393         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9394         for LEON or LEON3.
9395         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9396         (AS_LEON_FLAG): New macro.
9397         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9398         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9399         Deal with LEON and LEON3 for the memory model.
9400         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9401         (atomic_compare_and_swap<mode>_1): Likewise.
9402         (*atomic_compare_and_swap<mode>_1): Likewise.
9404 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9406         * config/arm/neon.md (vcond): Fix floating-point vector
9407         comparisons against 0.
9409 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9411         * lra-constraints.c (emit_spill_move): Remove assert.
9412         (process_alt_operands): Add more debugging
9413         output.  Increase reject for spilling into memory.  Decrease
9414         reject for reloading scratch.
9415         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9417 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9419         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9420         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9421         micromips incompatible.  Add nan2008.
9422         (MULTILIB_DIRNAMES): Add nan2008.
9423         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9424         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9425         and micromips incompatible.  Add nan2008.
9426         (MULTILIB_DIRNAMES): Add nan2008.
9427         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9429 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9431         PR rtl-optimization/58079
9432         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9433         simplify_comparison has widened a comparison with an integer.
9435 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9437         * config/arm/neon.md (movmisalign<mode>): Disable when we
9438         don't allow unaligned accesses.
9439         (*movmisalign<mode>_neon_store): Likewise.
9440         (*movmisalign<mode>_neon_load): Likewise.
9441         (*movmisalign<mode>_neon_store): Likewise.
9442         (*movmisalign<mode>_neon_load): Likewise.
9444 2013-08-08  Jan Hubicka  <jh@suse.cz>
9446         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9447         (make_pass_rebuild_cgraph_edges): Also clear references.
9448         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9449         * ipa-inline-transform.c (inline_transform): Remove all references
9450         after inlining.
9451         * cgraphunit.c (expand_function): Remove all references after
9452         expansion.
9453         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9454         (ipa_find_reference): Rewrite to iterator.
9455         (remove_stmt_references): Likewise.
9456         (ipa_clear_stmts_in_references): New function.
9457         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9458         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9459         clear references.
9460         * ipa-split.c (split_function): Remove references in split function.
9462 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9464         PR target/57431
9465         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9466         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9468 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9470         PR target/56979
9471         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9472         suggested mode for the assignment isn't compatible with the
9473         registers required.
9475 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9477         PR target/58065
9478         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9480 2013-08-07  Xinliang David Li  <davidxl@google.com>
9482         * config/i386/i386.opt: New option -mtune-ctrl=.
9483         * config/i386/x86-tune.def: New file.
9484         * config/i386/i386.h: include x86-tune.def.
9485         * config/i386/i386.c (ix86_option_override_internal):
9486         Parsing -mtune-ctrl= option and set tune features.
9488 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9490         PR other/12081
9491         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9492         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9493         to gen_2arg_fn_t.
9495 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9497         * rtl.h (update_alignments): Declare.
9498         * final.c (grow_label_align): New function extracted from...
9499         (shorten_branches): ...here.  Call it.
9500         (update_alignments): New function.
9501         * reorg.c (sibling_labels): New variable.
9502         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9503         the new label along with it onto the sibling_labels vector.
9504         (fill_simple_delay_slots): Adjust call to get_label_before.
9505         (fill_slots_from_thread): Likewise.
9506         (relax_delay_slots): Likewise.
9507         (make_return_insns): Likewise.
9508         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9510 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9512         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9513         document its semantics.
9514         (diagnostic_report_diagnostic): Adjust accordingly.
9516 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9518         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9519         (sparc_option_override): ...and port to new C++ pass API.
9520         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9522 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9524         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9526 2013-08-06  Caroline Tice  <cmtice@google.com>
9528         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9529         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9530         * tree-pass.h: Add pass_vtable_verify.
9531         * varasm.c (assemble_variable): Add code to properly set the comdat
9532         section and name for the .vtable_map_vars section.
9533         (assemble_vtyv_preinit_initializer): New function.
9534         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9535         LINK_ONCE flag.
9536         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9537         * vtable-verify.c: New file.
9538         * vtable-verify.h: New file.
9539         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9540         initialiation levels.
9541         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9542         * passes.def: Insert pass_vtable_verify.
9543         * aclocal.m4: Reorder includes.
9544         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9545         -fvtv-counts options.
9546         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9547         as appropriate, if -fvtable-verify=... is used.
9548         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9549         -fvtable-verify=... is used.
9550         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9551         (vtable-verify.o): Add new build rule.
9552         (GTFILES): Add vtable-verify.c to list.
9553         * common.opt (fvtable-verify=): New flag.
9554         (vtv_priority): Values for fvtable-verify= flag.
9555         (fvtv-counts): New flag.
9556         (fvtv-debug): New flag.
9557         * tree.h (save_vtable_map_decl): New extern function decl.
9559 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9561         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9562         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9563         (pass_data_rl78_devirt): ...new pass_data instance and...
9564         (make_pass_rl78_devirt): ...new function.
9565         (rl78_asm_file_start): Port pass registration to new C++ API.
9567 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9569         * coretypes.h (rtl_opt_pass): Add.
9570         (gcc::context): Add.
9571         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9572         (epiphany_init): Port to new C++ pass API.
9573         (epiphany_optimize_mode_switching): Likewise.
9574         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9575         (pass_manager::get_pass_mode_switching): New.
9576         (pass_manager::get_pass_peephole2): New.
9577         * mode-switching.c (pass_mode_switching): Add clone method.
9578         * recog.c (pass_peephole2): Add clone method.
9579         (pass_split_all_insns): Add clone method.
9581 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9583         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9584         (mips_option_override): ...here, porting to new C++ API for passes.
9586 2013-08-06  Jan Hubicka  <jh@suse.cz>
9588         * cgraph.c (cgraph_get_body): New function based on lto.c
9589         implementation.
9590         * cgraph.h (cgraph_get_body): Declare.
9591         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9592         LTO paths.
9593         * cgraphunit.c (expand_function): Get body prior expanding.
9594         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9595         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9596         really need.
9597         * passes.c (do_per_function_toporder): Get body.
9598         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9599         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9601 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9603         PR fortran/57987
9604         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9605         is not re-finalized.  Rename second parameter to no_collect.
9607 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9609         PR middle-end/58041
9610         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9611         MEM_REF has proper alignment information.
9613 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9615         PR other/12081
9616         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9617         class insn_gen_fn.
9618         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9619         rtx (*) (rtx, ...) with insn_gen_fn.
9620         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9621         insn_gen_fn::stored_funcptr.  Add initializer braces.
9623 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9625         Rewrite how instances of passes are cloned to remove assumptions
9626         about their sizes (thus allowing pass subclasses to have
9627         additional data fields, albeit non-GC-managed ones at this point).
9629         * passes.c (make_pass_instance): Now that passes have clone
9630         methods, rewrite this function to eliminate XNEW and memcpy
9631         calls that used hardcoded sizes.  Since this function no longer
9632         creates pass instances, rename it to...
9633         (add_pass_instance): ...this.  Document the old way that passes were
9634         numbered and flagged, and rework this function to continue using it.
9635         (next_pass_1): Add an initial_pass argument for use by
9636         add_pass_instance.
9637         (position_pass): When adding multiple instances of a pass, use
9638         the pass's clone method, rather than relying on the XNEW/memcpy
9639         within the former make_pass_instance (now add_pass_instance).
9640         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9641         the initial instance of the current pass within the pass manager.
9643 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9645         This is the automated part of the conversion of passes from C
9646         structs to C++ classes.
9648         Patch autogenerated by refactor_passes.py from
9649         https://github.com/davidmalcolm/gcc-refactoring-scripts
9650         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9652         * asan.c (pass_asan): Convert from a global struct to a subclass of
9653         gimple_opt_pass along with...
9654         (pass_data_asan): ...new pass_data instance and...
9655         (make_pass_asan): ...new function.
9656         (pass_asan_O0): Convert from a global struct to a subclass of
9657         gimple_opt_pass along with...
9658         (pass_data_asan_O0): ...new pass_data instance and...
9659         (make_pass_asan_O0): ...new function.
9660         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9661         subclass of rtl_opt_pass along with...
9662         (pass_data_inc_dec): ...new pass_data instance and...
9663         (make_pass_inc_dec): ...new function.
9664         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9665         a subclass of rtl_opt_pass along with...
9666         (pass_data_reorder_blocks): ...new pass_data instance and...
9667         (make_pass_reorder_blocks): ...new function.
9668         (pass_duplicate_computed_gotos): Convert from a global struct to a
9669         subclass of rtl_opt_pass along with...
9670         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9671         (make_pass_duplicate_computed_gotos): ...new function.
9672         (pass_partition_blocks): Convert from a global struct to a subclass of
9673         rtl_opt_pass along with...
9674         (pass_data_partition_blocks): ...new pass_data instance and...
9675         (make_pass_partition_blocks): ...new function.
9676         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9677         struct to a subclass of rtl_opt_pass along with...
9678         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9679         and...
9680         (make_pass_branch_target_load_optimize1): ...new function.
9681         (pass_branch_target_load_optimize2): Convert from a global struct to a
9682         subclass of rtl_opt_pass along with...
9683         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9684         and...
9685         (make_pass_branch_target_load_optimize2): ...new function.
9686         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9687         of rtl_opt_pass along with...
9688         (pass_data_jump): ...new pass_data instance and...
9689         (make_pass_jump): ...new function.
9690         (pass_jump2): Convert from a global struct to a subclass of
9691         rtl_opt_pass along with...
9692         (pass_data_jump2): ...new pass_data instance and...
9693         (make_pass_jump2): ...new function.
9694         * cfgexpand.c (pass_expand): Convert from a global struct to a
9695         subclass of rtl_opt_pass along with...
9696         (pass_data_expand): ...new pass_data instance and...
9697         (make_pass_expand): ...new function.
9698         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9699         of rtl_opt_pass along with...
9700         (pass_data_free_cfg): ...new pass_data instance and...
9701         (make_pass_free_cfg): ...new function.
9702         (pass_into_cfg_layout_mode): Convert from a global struct to a
9703         subclass of rtl_opt_pass along with...
9704         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9705         (make_pass_into_cfg_layout_mode): ...new function.
9706         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9707         subclass of rtl_opt_pass along with...
9708         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9709         (make_pass_outof_cfg_layout_mode): ...new function.
9710         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9711         struct to a subclass of gimple_opt_pass along with...
9712         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9713         (make_pass_build_cgraph_edges): ...new function.
9714         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9715         subclass of gimple_opt_pass along with...
9716         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9717         (make_pass_rebuild_cgraph_edges): ...new function.
9718         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9719         subclass of gimple_opt_pass along with...
9720         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9721         and...
9722         (make_pass_remove_cgraph_callee_edges): ...new function.
9723         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9724         struct to a subclass of rtl_opt_pass along with...
9725         (pass_data_stack_adjustments): ...new pass_data instance and...
9726         (make_pass_stack_adjustments): ...new function.
9727         * combine.c (pass_combine): Convert from a global struct to a subclass
9728         of rtl_opt_pass along with...
9729         (pass_data_combine): ...new pass_data instance and...
9730         (make_pass_combine): ...new function.
9731         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9732         global struct to a subclass of rtl_opt_pass along with...
9733         (pass_data_compare_elim_after_reload): ...new pass_data instance
9734         and...
9735         (make_pass_compare_elim_after_reload): ...new function.
9736         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9737         of rtl_opt_pass along with...
9738         (pass_data_rtl_cprop): ...new pass_data instance and...
9739         (make_pass_rtl_cprop): ...new function.
9740         * cse.c (pass_cse): Convert from a global struct to a subclass of
9741         rtl_opt_pass along with...
9742         (pass_data_cse): ...new pass_data instance and...
9743         (make_pass_cse): ...new function.
9744         (pass_cse2): Convert from a global struct to a subclass of
9745         rtl_opt_pass along with...
9746         (pass_data_cse2): ...new pass_data instance and...
9747         (make_pass_cse2): ...new function.
9748         (pass_cse_after_global_opts): Convert from a global struct to a
9749         subclass of rtl_opt_pass along with...
9750         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9751         (make_pass_cse_after_global_opts): ...new function.
9752         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9753         of rtl_opt_pass along with...
9754         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9755         (make_pass_ud_rtl_dce): ...new function.
9756         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9757         rtl_opt_pass along with...
9758         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9759         (make_pass_fast_rtl_dce): ...new function.
9760         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9761         a subclass of rtl_opt_pass along with...
9762         (pass_data_df_initialize_opt): ...new pass_data instance and...
9763         (make_pass_df_initialize_opt): ...new function.
9764         (pass_df_initialize_no_opt): Convert from a global struct to a
9765         subclass of rtl_opt_pass along with...
9766         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9767         (make_pass_df_initialize_no_opt): ...new function.
9768         (pass_df_finish): Convert from a global struct to a subclass of
9769         rtl_opt_pass along with...
9770         (pass_data_df_finish): ...new pass_data instance and...
9771         (make_pass_df_finish): ...new function.
9772         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9773         rtl_opt_pass along with...
9774         (pass_data_rtl_dse1): ...new pass_data instance and...
9775         (make_pass_rtl_dse1): ...new function.
9776         (pass_rtl_dse2): Convert from a global struct to a subclass of
9777         rtl_opt_pass along with...
9778         (pass_data_rtl_dse2): ...new pass_data instance and...
9779         (make_pass_rtl_dse2): ...new function.
9780         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9781         subclass of rtl_opt_pass along with...
9782         (pass_data_dwarf2_frame): ...new pass_data instance and...
9783         (make_pass_dwarf2_frame): ...new function.
9784         * except.c (pass_set_nothrow_function_flags): Convert from a global
9785         struct to a subclass of rtl_opt_pass along with...
9786         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9787         and...
9788         (make_pass_set_nothrow_function_flags): ...new function.
9789         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9790         subclass of rtl_opt_pass along with...
9791         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9792         and...
9793         (make_pass_convert_to_eh_region_ranges): ...new function.
9794         * final.c (pass_compute_alignments): Convert from a global struct to a
9795         subclass of rtl_opt_pass along with...
9796         (pass_data_compute_alignments): ...new pass_data instance and...
9797         (make_pass_compute_alignments): ...new function.
9798         (pass_final): Convert from a global struct to a subclass of
9799         rtl_opt_pass along with...
9800         (pass_data_final): ...new pass_data instance and...
9801         (make_pass_final): ...new function.
9802         (pass_shorten_branches): Convert from a global struct to a subclass of
9803         rtl_opt_pass along with...
9804         (pass_data_shorten_branches): ...new pass_data instance and...
9805         (make_pass_shorten_branches): ...new function.
9806         (pass_clean_state): Convert from a global struct to a subclass of
9807         rtl_opt_pass along with...
9808         (pass_data_clean_state): ...new pass_data instance and...
9809         (make_pass_clean_state): ...new function.
9810         * function.c (pass_instantiate_virtual_regs): Convert from a global
9811         struct to a subclass of rtl_opt_pass along with...
9812         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9813         (make_pass_instantiate_virtual_regs): ...new function.
9814         (pass_leaf_regs): Convert from a global struct to a subclass of
9815         rtl_opt_pass along with...
9816         (pass_data_leaf_regs): ...new pass_data instance and...
9817         (make_pass_leaf_regs): ...new function.
9818         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9819         subclass of rtl_opt_pass along with...
9820         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9821         and...
9822         (make_pass_thread_prologue_and_epilogue): ...new function.
9823         (pass_match_asm_constraints): Convert from a global struct to a
9824         subclass of rtl_opt_pass along with...
9825         (pass_data_match_asm_constraints): ...new pass_data instance and...
9826         (make_pass_match_asm_constraints): ...new function.
9827         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9828         subclass of rtl_opt_pass along with...
9829         (pass_data_rtl_fwprop): ...new pass_data instance and...
9830         (make_pass_rtl_fwprop): ...new function.
9831         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9832         rtl_opt_pass along with...
9833         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9834         (make_pass_rtl_fwprop_addr): ...new function.
9835         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9836         rtl_opt_pass along with...
9837         (pass_data_rtl_pre): ...new pass_data instance and...
9838         (make_pass_rtl_pre): ...new function.
9839         (pass_rtl_hoist): Convert from a global struct to a subclass of
9840         rtl_opt_pass along with...
9841         (pass_data_rtl_hoist): ...new pass_data instance and...
9842         (make_pass_rtl_hoist): ...new function.
9843         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9844         subclass of gimple_opt_pass along with...
9845         (pass_data_lower_cf): ...new pass_data instance and...
9846         (make_pass_lower_cf): ...new function.
9847         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9848         from a global struct to a subclass of gimple_opt_pass along with...
9849         (pass_data_strength_reduction): ...new pass_data instance and...
9850         (make_pass_strength_reduction): ...new function.
9851         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9852         of rtl_opt_pass along with...
9853         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9854         (make_pass_rtl_ifcvt): ...new function.
9855         (pass_if_after_combine): Convert from a global struct to a subclass of
9856         rtl_opt_pass along with...
9857         (pass_data_if_after_combine): ...new pass_data instance and...
9858         (make_pass_if_after_combine): ...new function.
9859         (pass_if_after_reload): Convert from a global struct to a subclass of
9860         rtl_opt_pass along with...
9861         (pass_data_if_after_reload): ...new pass_data instance and...
9862         (make_pass_if_after_reload): ...new function.
9863         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9864         a subclass of rtl_opt_pass along with...
9865         (pass_data_initialize_regs): ...new pass_data instance and...
9866         (make_pass_initialize_regs): ...new function.
9867         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9868         of ipa_opt_pass_d along with...
9869         (pass_data_ipa_cp): ...new pass_data instance and...
9870         (make_pass_ipa_cp): ...new function.
9871         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9872         global struct to a subclass of gimple_opt_pass along with...
9873         (pass_data_inline_parameters): ...new pass_data instance and...
9874         (make_pass_inline_parameters): ...new function.
9875         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9876         subclass of gimple_opt_pass along with...
9877         (pass_data_early_inline): ...new pass_data instance and...
9878         (make_pass_early_inline): ...new function.
9879         (pass_ipa_inline): Convert from a global struct to a subclass of
9880         ipa_opt_pass_d along with...
9881         (pass_data_ipa_inline): ...new pass_data instance and...
9882         (make_pass_ipa_inline): ...new function.
9883         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9884         struct to a subclass of gimple_opt_pass along with...
9885         (pass_data_local_pure_const): ...new pass_data instance and...
9886         (make_pass_local_pure_const): ...new function.
9887         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9888         ipa_opt_pass_d along with...
9889         (pass_data_ipa_pure_const): ...new pass_data instance and...
9890         (make_pass_ipa_pure_const): ...new function.
9891         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9892         to a subclass of ipa_opt_pass_d along with...
9893         (pass_data_ipa_reference): ...new pass_data instance and...
9894         (make_pass_ipa_reference): ...new function.
9895         * ipa-split.c (pass_split_functions): Convert from a global struct to
9896         a subclass of gimple_opt_pass along with...
9897         (pass_data_split_functions): ...new pass_data instance and...
9898         (make_pass_split_functions): ...new function.
9899         (pass_feedback_split_functions): Convert from a global struct to a
9900         subclass of gimple_opt_pass along with...
9901         (pass_data_feedback_split_functions): ...new pass_data instance and...
9902         (make_pass_feedback_split_functions): ...new function.
9903         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9904         global struct to a subclass of simple_ipa_opt_pass along with...
9905         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9906         instance and...
9907         (make_pass_ipa_function_and_variable_visibility): ...new function.
9908         (pass_ipa_free_inline_summary): Convert from a global struct to a
9909         subclass of simple_ipa_opt_pass along with...
9910         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9911         (make_pass_ipa_free_inline_summary): ...new function.
9912         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9913         subclass of ipa_opt_pass_d along with...
9914         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9915         and...
9916         (make_pass_ipa_whole_program_visibility): ...new function.
9917         (pass_ipa_profile): Convert from a global struct to a subclass of
9918         ipa_opt_pass_d along with...
9919         (pass_data_ipa_profile): ...new pass_data instance and...
9920         (make_pass_ipa_profile): ...new function.
9921         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9922         ipa_opt_pass_d along with...
9923         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9924         (make_pass_ipa_cdtor_merge): ...new function.
9925         * ira.c (pass_ira): Convert from a global struct to a subclass of
9926         rtl_opt_pass along with...
9927         (pass_data_ira): ...new pass_data instance and...
9928         (make_pass_ira): ...new function.
9929         (pass_reload): Convert from a global struct to a subclass of
9930         rtl_opt_pass along with...
9931         (pass_data_reload): ...new pass_data instance and...
9932         (make_pass_reload): ...new function.
9933         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9934         subclass of rtl_opt_pass along with...
9935         (pass_data_cleanup_barriers): ...new pass_data instance and...
9936         (make_pass_cleanup_barriers): ...new function.
9937         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9938         of rtl_opt_pass along with...
9939         (pass_data_loop2): ...new pass_data instance and...
9940         (make_pass_loop2): ...new function.
9941         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9942         rtl_opt_pass along with...
9943         (pass_data_rtl_loop_init): ...new pass_data instance and...
9944         (make_pass_rtl_loop_init): ...new function.
9945         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9946         rtl_opt_pass along with...
9947         (pass_data_rtl_loop_done): ...new pass_data instance and...
9948         (make_pass_rtl_loop_done): ...new function.
9949         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9950         subclass of rtl_opt_pass along with...
9951         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9952         (make_pass_rtl_move_loop_invariants): ...new function.
9953         (pass_rtl_unswitch): Convert from a global struct to a subclass of
9954         rtl_opt_pass along with...
9955         (pass_data_rtl_unswitch): ...new pass_data instance and...
9956         (make_pass_rtl_unswitch): ...new function.
9957         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
9958         subclass of rtl_opt_pass along with...
9959         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
9960         and...
9961         (make_pass_rtl_unroll_and_peel_loops): ...new function.
9962         (pass_rtl_doloop): Convert from a global struct to a subclass of
9963         rtl_opt_pass along with...
9964         (pass_data_rtl_doloop): ...new pass_data instance and...
9965         (make_pass_rtl_doloop): ...new function.
9966         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
9967         a subclass of rtl_opt_pass along with...
9968         (pass_data_lower_subreg): ...new pass_data instance and...
9969         (make_pass_lower_subreg): ...new function.
9970         (pass_lower_subreg2): Convert from a global struct to a subclass of
9971         rtl_opt_pass along with...
9972         (pass_data_lower_subreg2): ...new pass_data instance and...
9973         (make_pass_lower_subreg2): ...new function.
9974         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
9975         struct to a subclass of ipa_opt_pass_d along with...
9976         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
9977         (make_pass_ipa_lto_gimple_out): ...new function.
9978         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
9979         of ipa_opt_pass_d along with...
9980         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
9981         (make_pass_ipa_lto_finish_out): ...new function.
9982         * mode-switching.c (pass_mode_switching): Convert from a global struct
9983         to a subclass of rtl_opt_pass along with...
9984         (pass_data_mode_switching): ...new pass_data instance and...
9985         (make_pass_mode_switching): ...new function.
9986         * modulo-sched.c (pass_sms): Convert from a global struct to a
9987         subclass of rtl_opt_pass along with...
9988         (pass_data_sms): ...new pass_data instance and...
9989         (make_pass_sms): ...new function.
9990         * omp-low.c (pass_expand_omp): Convert from a global struct to a
9991         subclass of gimple_opt_pass along with...
9992         (pass_data_expand_omp): ...new pass_data instance and...
9993         (make_pass_expand_omp): ...new function.
9994         (pass_lower_omp): Convert from a global struct to a subclass of
9995         gimple_opt_pass along with...
9996         (pass_data_lower_omp): ...new pass_data instance and...
9997         (make_pass_lower_omp): ...new function.
9998         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
9999         of gimple_opt_pass along with...
10000         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10001         (make_pass_diagnose_omp_blocks): ...new function.
10002         * passes.c (pass_early_local_passes): Convert from a global struct to
10003         a subclass of simple_ipa_opt_pass along with...
10004         (pass_data_early_local_passes): ...new pass_data instance and...
10005         (make_pass_early_local_passes): ...new function.
10006         (pass_all_early_optimizations): Convert from a global struct to a
10007         subclass of gimple_opt_pass along with...
10008         (pass_data_all_early_optimizations): ...new pass_data instance and...
10009         (make_pass_all_early_optimizations): ...new function.
10010         (pass_all_optimizations): Convert from a global struct to a subclass
10011         of gimple_opt_pass along with...
10012         (pass_data_all_optimizations): ...new pass_data instance and...
10013         (make_pass_all_optimizations): ...new function.
10014         (pass_all_optimizations_g): Convert from a global struct to a subclass
10015         of gimple_opt_pass along with...
10016         (pass_data_all_optimizations_g): ...new pass_data instance and...
10017         (make_pass_all_optimizations_g): ...new function.
10018         (pass_rest_of_compilation): Convert from a global struct to a subclass
10019         of rtl_opt_pass along with...
10020         (pass_data_rest_of_compilation): ...new pass_data instance and...
10021         (make_pass_rest_of_compilation): ...new function.
10022         (pass_postreload): Convert from a global struct to a subclass of
10023         rtl_opt_pass along with...
10024         (pass_data_postreload): ...new pass_data instance and...
10025         (make_pass_postreload): ...new function.
10026         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10027         subclass of rtl_opt_pass along with...
10028         (pass_data_gcse2): ...new pass_data instance and...
10029         (make_pass_gcse2): ...new function.
10030         * postreload.c (pass_postreload_cse): Convert from a global struct to
10031         a subclass of rtl_opt_pass along with...
10032         (pass_data_postreload_cse): ...new pass_data instance and...
10033         (make_pass_postreload_cse): ...new function.
10034         * predict.c (pass_profile): Convert from a global struct to a subclass
10035         of gimple_opt_pass along with...
10036         (pass_data_profile): ...new pass_data instance and...
10037         (make_pass_profile): ...new function.
10038         (pass_strip_predict_hints): Convert from a global struct to a subclass
10039         of gimple_opt_pass along with...
10040         (pass_data_strip_predict_hints): ...new pass_data instance and...
10041         (make_pass_strip_predict_hints): ...new function.
10042         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10043         of rtl_opt_pass along with...
10044         (pass_data_peephole2): ...new pass_data instance and...
10045         (make_pass_peephole2): ...new function.
10046         (pass_split_all_insns): Convert from a global struct to a subclass of
10047         rtl_opt_pass along with...
10048         (pass_data_split_all_insns): ...new pass_data instance and...
10049         (make_pass_split_all_insns): ...new function.
10050         (pass_split_after_reload): Convert from a global struct to a subclass
10051         of rtl_opt_pass along with...
10052         (pass_data_split_after_reload): ...new pass_data instance and...
10053         (make_pass_split_after_reload): ...new function.
10054         (pass_split_before_regstack): Convert from a global struct to a
10055         subclass of rtl_opt_pass along with...
10056         (pass_data_split_before_regstack): ...new pass_data instance and...
10057         (make_pass_split_before_regstack): ...new function.
10058         (pass_split_before_sched2): Convert from a global struct to a subclass
10059         of rtl_opt_pass along with...
10060         (pass_data_split_before_sched2): ...new pass_data instance and...
10061         (make_pass_split_before_sched2): ...new function.
10062         (pass_split_for_shorten_branches): Convert from a global struct to a
10063         subclass of rtl_opt_pass along with...
10064         (pass_data_split_for_shorten_branches): ...new pass_data instance
10065         and...
10066         (make_pass_split_for_shorten_branches): ...new function.
10067         * ree.c (pass_ree): Convert from a global struct to a subclass of
10068         rtl_opt_pass along with...
10069         (pass_data_ree): ...new pass_data instance and...
10070         (make_pass_ree): ...new function.
10071         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10072         subclass of rtl_opt_pass along with...
10073         (pass_data_stack_regs): ...new pass_data instance and...
10074         (make_pass_stack_regs): ...new function.
10075         (pass_stack_regs_run): Convert from a global struct to a subclass of
10076         rtl_opt_pass along with...
10077         (pass_data_stack_regs_run): ...new pass_data instance and...
10078         (make_pass_stack_regs_run): ...new function.
10079         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10080         subclass of rtl_opt_pass along with...
10081         (pass_data_cprop_hardreg): ...new pass_data instance and...
10082         (make_pass_cprop_hardreg): ...new function.
10083         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10084         subclass of rtl_opt_pass along with...
10085         (pass_data_reginfo_init): ...new pass_data instance and...
10086         (make_pass_reginfo_init): ...new function.
10087         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10088         of rtl_opt_pass along with...
10089         (pass_data_regmove): ...new pass_data instance and...
10090         (make_pass_regmove): ...new function.
10091         * regrename.c (pass_regrename): Convert from a global struct to a
10092         subclass of rtl_opt_pass along with...
10093         (pass_data_regrename): ...new pass_data instance and...
10094         (make_pass_regrename): ...new function.
10095         * reorg.c (pass_delay_slots): Convert from a global struct to a
10096         subclass of rtl_opt_pass along with...
10097         (pass_data_delay_slots): ...new pass_data instance and...
10098         (make_pass_delay_slots): ...new function.
10099         (pass_machine_reorg): Convert from a global struct to a subclass of
10100         rtl_opt_pass along with...
10101         (pass_data_machine_reorg): ...new pass_data instance and...
10102         (make_pass_machine_reorg): ...new function.
10103         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10104         of rtl_opt_pass along with...
10105         (pass_data_sched): ...new pass_data instance and...
10106         (make_pass_sched): ...new function.
10107         (pass_sched2): Convert from a global struct to a subclass of
10108         rtl_opt_pass along with...
10109         (pass_data_sched2): ...new pass_data instance and...
10110         (make_pass_sched2): ...new function.
10111         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10112         to a subclass of rtl_opt_pass along with...
10113         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10114         (make_pass_stack_ptr_mod): ...new function.
10115         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10116         to a subclass of rtl_opt_pass along with...
10117         (pass_data_rtl_store_motion): ...new pass_data instance and...
10118         (make_pass_rtl_store_motion): ...new function.
10119         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10120         of gimple_opt_pass along with...
10121         (pass_data_tracer): ...new pass_data instance and...
10122         (make_pass_tracer): ...new function.
10123         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10124         to a subclass of gimple_opt_pass along with...
10125         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10126         (make_pass_diagnose_tm_blocks): ...new function.
10127         (pass_lower_tm): Convert from a global struct to a subclass of
10128         gimple_opt_pass along with...
10129         (pass_data_lower_tm): ...new pass_data instance and...
10130         (make_pass_lower_tm): ...new function.
10131         (pass_tm_init): Convert from a global struct to a subclass of
10132         gimple_opt_pass along with...
10133         (pass_data_tm_init): ...new pass_data instance and...
10134         (make_pass_tm_init): ...new function.
10135         (pass_tm_mark): Convert from a global struct to a subclass of
10136         gimple_opt_pass along with...
10137         (pass_data_tm_mark): ...new pass_data instance and...
10138         (make_pass_tm_mark): ...new function.
10139         (pass_tm_edges): Convert from a global struct to a subclass of
10140         gimple_opt_pass along with...
10141         (pass_data_tm_edges): ...new pass_data instance and...
10142         (make_pass_tm_edges): ...new function.
10143         (pass_tm_memopt): Convert from a global struct to a subclass of
10144         gimple_opt_pass along with...
10145         (pass_data_tm_memopt): ...new pass_data instance and...
10146         (make_pass_tm_memopt): ...new function.
10147         (pass_ipa_tm): Convert from a global struct to a subclass of
10148         simple_ipa_opt_pass along with...
10149         (pass_data_ipa_tm): ...new pass_data instance and...
10150         (make_pass_ipa_tm): ...new function.
10151         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10152         subclass of gimple_opt_pass along with...
10153         (pass_data_call_cdce): ...new pass_data instance and...
10154         (make_pass_call_cdce): ...new function.
10155         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10156         subclass of gimple_opt_pass along with...
10157         (pass_data_build_cfg): ...new pass_data instance and...
10158         (make_pass_build_cfg): ...new function.
10159         (pass_split_crit_edges): Convert from a global struct to a subclass of
10160         gimple_opt_pass along with...
10161         (pass_data_split_crit_edges): ...new pass_data instance and...
10162         (make_pass_split_crit_edges): ...new function.
10163         (pass_warn_function_return): Convert from a global struct to a
10164         subclass of gimple_opt_pass along with...
10165         (pass_data_warn_function_return): ...new pass_data instance and...
10166         (make_pass_warn_function_return): ...new function.
10167         (pass_warn_function_noreturn): Convert from a global struct to a
10168         subclass of gimple_opt_pass along with...
10169         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10170         (make_pass_warn_function_noreturn): ...new function.
10171         (pass_warn_unused_result): Convert from a global struct to a subclass
10172         of gimple_opt_pass along with...
10173         (pass_data_warn_unused_result): ...new pass_data instance and...
10174         (make_pass_warn_unused_result): ...new function.
10175         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10176         a subclass of gimple_opt_pass along with...
10177         (pass_data_merge_phi): ...new pass_data instance and...
10178         (make_pass_merge_phi): ...new function.
10179         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10180         a subclass of gimple_opt_pass along with...
10181         (pass_data_lower_complex): ...new pass_data instance and...
10182         (make_pass_lower_complex): ...new function.
10183         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10184         gimple_opt_pass along with...
10185         (pass_data_lower_complex_O0): ...new pass_data instance and...
10186         (make_pass_lower_complex_O0): ...new function.
10187         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10188         subclass of gimple_opt_pass along with...
10189         (pass_data_lower_eh): ...new pass_data instance and...
10190         (make_pass_lower_eh): ...new function.
10191         (pass_refactor_eh): Convert from a global struct to a subclass of
10192         gimple_opt_pass along with...
10193         (pass_data_refactor_eh): ...new pass_data instance and...
10194         (make_pass_refactor_eh): ...new function.
10195         (pass_lower_resx): Convert from a global struct to a subclass of
10196         gimple_opt_pass along with...
10197         (pass_data_lower_resx): ...new pass_data instance and...
10198         (make_pass_lower_resx): ...new function.
10199         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10200         of gimple_opt_pass along with...
10201         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10202         (make_pass_lower_eh_dispatch): ...new function.
10203         (pass_cleanup_eh): Convert from a global struct to a subclass of
10204         gimple_opt_pass along with...
10205         (pass_data_cleanup_eh): ...new pass_data instance and...
10206         (make_pass_cleanup_eh): ...new function.
10207         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10208         to a subclass of simple_ipa_opt_pass along with...
10209         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10210         (make_pass_ipa_lower_emutls): ...new function.
10211         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10212         a subclass of gimple_opt_pass along with...
10213         (pass_data_if_conversion): ...new pass_data instance and...
10214         (make_pass_if_conversion): ...new function.
10215         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10216         subclass of gimple_opt_pass along with...
10217         (pass_data_build_ssa): ...new pass_data instance and...
10218         (make_pass_build_ssa): ...new function.
10219         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10220         global struct to a subclass of gimple_opt_pass along with...
10221         (pass_data_loop_distribution): ...new pass_data instance and...
10222         (make_pass_loop_distribution): ...new function.
10223         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10224         subclass of gimple_opt_pass along with...
10225         (pass_data_mudflap_1): ...new pass_data instance and...
10226         (make_pass_mudflap_1): ...new function.
10227         (pass_mudflap_2): Convert from a global struct to a subclass of
10228         gimple_opt_pass along with...
10229         (pass_data_mudflap_2): ...new pass_data instance and...
10230         (make_pass_mudflap_2): ...new function.
10231         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10232         subclass of gimple_opt_pass along with...
10233         (pass_data_mudflap_1): ...new pass_data instance and...
10234         (make_pass_mudflap_1): ...new function.
10235         (pass_mudflap_2): Convert from a global struct to a subclass of
10236         gimple_opt_pass along with...
10237         (pass_data_mudflap_2): ...new pass_data instance and...
10238         (make_pass_mudflap_2): ...new function.
10239         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10240         gimple_opt_pass along with...
10241         (pass_data_nrv): ...new pass_data instance and...
10242         (make_pass_nrv): ...new function.
10243         (pass_return_slot): Convert from a global struct to a subclass of
10244         gimple_opt_pass along with...
10245         (pass_data_return_slot): ...new pass_data instance and...
10246         (make_pass_return_slot): ...new function.
10247         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10248         to a subclass of gimple_opt_pass along with...
10249         (pass_data_object_sizes): ...new pass_data instance and...
10250         (make_pass_object_sizes): ...new function.
10251         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10252         global struct to a subclass of gimple_opt_pass along with...
10253         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10254         and...
10255         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10256         (pass_fixup_cfg): Convert from a global struct to a subclass of
10257         gimple_opt_pass along with...
10258         (pass_data_fixup_cfg): ...new pass_data instance and...
10259         (make_pass_fixup_cfg): ...new function.
10260         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10261         (make_pass_mudflap_1): ...new function.
10262         (pass_mudflap_2): Replace declaration with that of...
10263         (make_pass_mudflap_2): ...new function.
10264         (pass_asan): Replace declaration with that of...
10265         (make_pass_asan): ...new function.
10266         (pass_asan_O0): Replace declaration with that of...
10267         (make_pass_asan_O0): ...new function.
10268         (pass_tsan): Replace declaration with that of...
10269         (make_pass_tsan): ...new function.
10270         (pass_tsan_O0): Replace declaration with that of...
10271         (make_pass_tsan_O0): ...new function.
10272         (pass_lower_cf): Replace declaration with that of...
10273         (make_pass_lower_cf): ...new function.
10274         (pass_refactor_eh): Replace declaration with that of...
10275         (make_pass_refactor_eh): ...new function.
10276         (pass_lower_eh): Replace declaration with that of...
10277         (make_pass_lower_eh): ...new function.
10278         (pass_lower_eh_dispatch): Replace declaration with that of...
10279         (make_pass_lower_eh_dispatch): ...new function.
10280         (pass_lower_resx): Replace declaration with that of...
10281         (make_pass_lower_resx): ...new function.
10282         (pass_build_cfg): Replace declaration with that of...
10283         (make_pass_build_cfg): ...new function.
10284         (pass_early_tree_profile): Replace declaration with that of...
10285         (make_pass_early_tree_profile): ...new function.
10286         (pass_cleanup_eh): Replace declaration with that of...
10287         (make_pass_cleanup_eh): ...new function.
10288         (pass_sra): Replace declaration with that of...
10289         (make_pass_sra): ...new function.
10290         (pass_sra_early): Replace declaration with that of...
10291         (make_pass_sra_early): ...new function.
10292         (pass_early_ipa_sra): Replace declaration with that of...
10293         (make_pass_early_ipa_sra): ...new function.
10294         (pass_tail_recursion): Replace declaration with that of...
10295         (make_pass_tail_recursion): ...new function.
10296         (pass_tail_calls): Replace declaration with that of...
10297         (make_pass_tail_calls): ...new function.
10298         (pass_tree_loop): Replace declaration with that of...
10299         (make_pass_tree_loop): ...new function.
10300         (pass_tree_loop_init): Replace declaration with that of...
10301         (make_pass_tree_loop_init): ...new function.
10302         (pass_lim): Replace declaration with that of...
10303         (make_pass_lim): ...new function.
10304         (pass_tree_unswitch): Replace declaration with that of...
10305         (make_pass_tree_unswitch): ...new function.
10306         (pass_predcom): Replace declaration with that of...
10307         (make_pass_predcom): ...new function.
10308         (pass_iv_canon): Replace declaration with that of...
10309         (make_pass_iv_canon): ...new function.
10310         (pass_scev_cprop): Replace declaration with that of...
10311         (make_pass_scev_cprop): ...new function.
10312         (pass_empty_loop): Replace declaration with that of...
10313         (make_pass_empty_loop): ...new function.
10314         (pass_record_bounds): Replace declaration with that of...
10315         (make_pass_record_bounds): ...new function.
10316         (pass_graphite): Replace declaration with that of...
10317         (make_pass_graphite): ...new function.
10318         (pass_graphite_transforms): Replace declaration with that of...
10319         (make_pass_graphite_transforms): ...new function.
10320         (pass_if_conversion): Replace declaration with that of...
10321         (make_pass_if_conversion): ...new function.
10322         (pass_loop_distribution): Replace declaration with that of...
10323         (make_pass_loop_distribution): ...new function.
10324         (pass_vectorize): Replace declaration with that of...
10325         (make_pass_vectorize): ...new function.
10326         (pass_slp_vectorize): Replace declaration with that of...
10327         (make_pass_slp_vectorize): ...new function.
10328         (pass_complete_unroll): Replace declaration with that of...
10329         (make_pass_complete_unroll): ...new function.
10330         (pass_complete_unrolli): Replace declaration with that of...
10331         (make_pass_complete_unrolli): ...new function.
10332         (pass_parallelize_loops): Replace declaration with that of...
10333         (make_pass_parallelize_loops): ...new function.
10334         (pass_loop_prefetch): Replace declaration with that of...
10335         (make_pass_loop_prefetch): ...new function.
10336         (pass_iv_optimize): Replace declaration with that of...
10337         (make_pass_iv_optimize): ...new function.
10338         (pass_tree_loop_done): Replace declaration with that of...
10339         (make_pass_tree_loop_done): ...new function.
10340         (pass_ch): Replace declaration with that of...
10341         (make_pass_ch): ...new function.
10342         (pass_ccp): Replace declaration with that of...
10343         (make_pass_ccp): ...new function.
10344         (pass_phi_only_cprop): Replace declaration with that of...
10345         (make_pass_phi_only_cprop): ...new function.
10346         (pass_build_ssa): Replace declaration with that of...
10347         (make_pass_build_ssa): ...new function.
10348         (pass_build_alias): Replace declaration with that of...
10349         (make_pass_build_alias): ...new function.
10350         (pass_build_ealias): Replace declaration with that of...
10351         (make_pass_build_ealias): ...new function.
10352         (pass_dominator): Replace declaration with that of...
10353         (make_pass_dominator): ...new function.
10354         (pass_dce): Replace declaration with that of...
10355         (make_pass_dce): ...new function.
10356         (pass_dce_loop): Replace declaration with that of...
10357         (make_pass_dce_loop): ...new function.
10358         (pass_cd_dce): Replace declaration with that of...
10359         (make_pass_cd_dce): ...new function.
10360         (pass_call_cdce): Replace declaration with that of...
10361         (make_pass_call_cdce): ...new function.
10362         (pass_merge_phi): Replace declaration with that of...
10363         (make_pass_merge_phi): ...new function.
10364         (pass_split_crit_edges): Replace declaration with that of...
10365         (make_pass_split_crit_edges): ...new function.
10366         (pass_pre): Replace declaration with that of...
10367         (make_pass_pre): ...new function.
10368         (pass_profile): Replace declaration with that of...
10369         (make_pass_profile): ...new function.
10370         (pass_strip_predict_hints): Replace declaration with that of...
10371         (make_pass_strip_predict_hints): ...new function.
10372         (pass_lower_complex_O0): Replace declaration with that of...
10373         (make_pass_lower_complex_O0): ...new function.
10374         (pass_lower_complex): Replace declaration with that of...
10375         (make_pass_lower_complex): ...new function.
10376         (pass_lower_vector): Replace declaration with that of...
10377         (make_pass_lower_vector): ...new function.
10378         (pass_lower_vector_ssa): Replace declaration with that of...
10379         (make_pass_lower_vector_ssa): ...new function.
10380         (pass_lower_omp): Replace declaration with that of...
10381         (make_pass_lower_omp): ...new function.
10382         (pass_diagnose_omp_blocks): Replace declaration with that of...
10383         (make_pass_diagnose_omp_blocks): ...new function.
10384         (pass_expand_omp): Replace declaration with that of...
10385         (make_pass_expand_omp): ...new function.
10386         (pass_expand_omp_ssa): Replace declaration with that of...
10387         (make_pass_expand_omp_ssa): ...new function.
10388         (pass_object_sizes): Replace declaration with that of...
10389         (make_pass_object_sizes): ...new function.
10390         (pass_strlen): Replace declaration with that of...
10391         (make_pass_strlen): ...new function.
10392         (pass_fold_builtins): Replace declaration with that of...
10393         (make_pass_fold_builtins): ...new function.
10394         (pass_stdarg): Replace declaration with that of...
10395         (make_pass_stdarg): ...new function.
10396         (pass_early_warn_uninitialized): Replace declaration with that of...
10397         (make_pass_early_warn_uninitialized): ...new function.
10398         (pass_late_warn_uninitialized): Replace declaration with that of...
10399         (make_pass_late_warn_uninitialized): ...new function.
10400         (pass_cse_reciprocals): Replace declaration with that of...
10401         (make_pass_cse_reciprocals): ...new function.
10402         (pass_cse_sincos): Replace declaration with that of...
10403         (make_pass_cse_sincos): ...new function.
10404         (pass_optimize_bswap): Replace declaration with that of...
10405         (make_pass_optimize_bswap): ...new function.
10406         (pass_optimize_widening_mul): Replace declaration with that of...
10407         (make_pass_optimize_widening_mul): ...new function.
10408         (pass_warn_function_return): Replace declaration with that of...
10409         (make_pass_warn_function_return): ...new function.
10410         (pass_warn_function_noreturn): Replace declaration with that of...
10411         (make_pass_warn_function_noreturn): ...new function.
10412         (pass_cselim): Replace declaration with that of...
10413         (make_pass_cselim): ...new function.
10414         (pass_phiopt): Replace declaration with that of...
10415         (make_pass_phiopt): ...new function.
10416         (pass_forwprop): Replace declaration with that of...
10417         (make_pass_forwprop): ...new function.
10418         (pass_phiprop): Replace declaration with that of...
10419         (make_pass_phiprop): ...new function.
10420         (pass_tree_ifcombine): Replace declaration with that of...
10421         (make_pass_tree_ifcombine): ...new function.
10422         (pass_dse): Replace declaration with that of...
10423         (make_pass_dse): ...new function.
10424         (pass_nrv): Replace declaration with that of...
10425         (make_pass_nrv): ...new function.
10426         (pass_rename_ssa_copies): Replace declaration with that of...
10427         (make_pass_rename_ssa_copies): ...new function.
10428         (pass_sink_code): Replace declaration with that of...
10429         (make_pass_sink_code): ...new function.
10430         (pass_fre): Replace declaration with that of...
10431         (make_pass_fre): ...new function.
10432         (pass_check_data_deps): Replace declaration with that of...
10433         (make_pass_check_data_deps): ...new function.
10434         (pass_copy_prop): Replace declaration with that of...
10435         (make_pass_copy_prop): ...new function.
10436         (pass_vrp): Replace declaration with that of...
10437         (make_pass_vrp): ...new function.
10438         (pass_uncprop): Replace declaration with that of...
10439         (make_pass_uncprop): ...new function.
10440         (pass_return_slot): Replace declaration with that of...
10441         (make_pass_return_slot): ...new function.
10442         (pass_reassoc): Replace declaration with that of...
10443         (make_pass_reassoc): ...new function.
10444         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10445         (make_pass_rebuild_cgraph_edges): ...new function.
10446         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10447         (make_pass_remove_cgraph_callee_edges): ...new function.
10448         (pass_build_cgraph_edges): Replace declaration with that of...
10449         (make_pass_build_cgraph_edges): ...new function.
10450         (pass_local_pure_const): Replace declaration with that of...
10451         (make_pass_local_pure_const): ...new function.
10452         (pass_tracer): Replace declaration with that of...
10453         (make_pass_tracer): ...new function.
10454         (pass_warn_unused_result): Replace declaration with that of...
10455         (make_pass_warn_unused_result): ...new function.
10456         (pass_diagnose_tm_blocks): Replace declaration with that of...
10457         (make_pass_diagnose_tm_blocks): ...new function.
10458         (pass_lower_tm): Replace declaration with that of...
10459         (make_pass_lower_tm): ...new function.
10460         (pass_tm_init): Replace declaration with that of...
10461         (make_pass_tm_init): ...new function.
10462         (pass_tm_mark): Replace declaration with that of...
10463         (make_pass_tm_mark): ...new function.
10464         (pass_tm_memopt): Replace declaration with that of...
10465         (make_pass_tm_memopt): ...new function.
10466         (pass_tm_edges): Replace declaration with that of...
10467         (make_pass_tm_edges): ...new function.
10468         (pass_split_functions): Replace declaration with that of...
10469         (make_pass_split_functions): ...new function.
10470         (pass_feedback_split_functions): Replace declaration with that of...
10471         (make_pass_feedback_split_functions): ...new function.
10472         (pass_strength_reduction): Replace declaration with that of...
10473         (make_pass_strength_reduction): ...new function.
10474         (pass_ipa_lower_emutls): Replace declaration with that of...
10475         (make_pass_ipa_lower_emutls): ...new function.
10476         (pass_ipa_function_and_variable_visibility): Replace declaration with
10477         that of...
10478         (make_pass_ipa_function_and_variable_visibility): ...new function.
10479         (pass_ipa_tree_profile): Replace declaration with that of...
10480         (make_pass_ipa_tree_profile): ...new function.
10481         (pass_early_local_passes): Replace declaration with that of...
10482         (make_pass_early_local_passes): ...new function.
10483         (pass_ipa_whole_program_visibility): Replace declaration with that
10484         of...
10485         (make_pass_ipa_whole_program_visibility): ...new function.
10486         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10487         (make_pass_ipa_lto_gimple_out): ...new function.
10488         (pass_ipa_increase_alignment): Replace declaration with that of...
10489         (make_pass_ipa_increase_alignment): ...new function.
10490         (pass_ipa_inline): Replace declaration with that of...
10491         (make_pass_ipa_inline): ...new function.
10492         (pass_ipa_free_lang_data): Replace declaration with that of...
10493         (make_pass_ipa_free_lang_data): ...new function.
10494         (pass_ipa_free_inline_summary): Replace declaration with that of...
10495         (make_pass_ipa_free_inline_summary): ...new function.
10496         (pass_ipa_cp): Replace declaration with that of...
10497         (make_pass_ipa_cp): ...new function.
10498         (pass_ipa_reference): Replace declaration with that of...
10499         (make_pass_ipa_reference): ...new function.
10500         (pass_ipa_pure_const): Replace declaration with that of...
10501         (make_pass_ipa_pure_const): ...new function.
10502         (pass_ipa_pta): Replace declaration with that of...
10503         (make_pass_ipa_pta): ...new function.
10504         (pass_ipa_lto_finish_out): Replace declaration with that of...
10505         (make_pass_ipa_lto_finish_out): ...new function.
10506         (pass_ipa_tm): Replace declaration with that of...
10507         (make_pass_ipa_tm): ...new function.
10508         (pass_ipa_profile): Replace declaration with that of...
10509         (make_pass_ipa_profile): ...new function.
10510         (pass_ipa_cdtor_merge): Replace declaration with that of...
10511         (make_pass_ipa_cdtor_merge): ...new function.
10512         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10513         of...
10514         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10515         (pass_init_datastructures): Replace declaration with that of...
10516         (make_pass_init_datastructures): ...new function.
10517         (pass_fixup_cfg): Replace declaration with that of...
10518         (make_pass_fixup_cfg): ...new function.
10519         (pass_expand): Replace declaration with that of...
10520         (make_pass_expand): ...new function.
10521         (pass_instantiate_virtual_regs): Replace declaration with that of...
10522         (make_pass_instantiate_virtual_regs): ...new function.
10523         (pass_rtl_fwprop): Replace declaration with that of...
10524         (make_pass_rtl_fwprop): ...new function.
10525         (pass_rtl_fwprop_addr): Replace declaration with that of...
10526         (make_pass_rtl_fwprop_addr): ...new function.
10527         (pass_jump): Replace declaration with that of...
10528         (make_pass_jump): ...new function.
10529         (pass_jump2): Replace declaration with that of...
10530         (make_pass_jump2): ...new function.
10531         (pass_lower_subreg): Replace declaration with that of...
10532         (make_pass_lower_subreg): ...new function.
10533         (pass_cse): Replace declaration with that of...
10534         (make_pass_cse): ...new function.
10535         (pass_fast_rtl_dce): Replace declaration with that of...
10536         (make_pass_fast_rtl_dce): ...new function.
10537         (pass_ud_rtl_dce): Replace declaration with that of...
10538         (make_pass_ud_rtl_dce): ...new function.
10539         (pass_rtl_dce): Replace declaration with that of...
10540         (make_pass_rtl_dce): ...new function.
10541         (pass_rtl_dse1): Replace declaration with that of...
10542         (make_pass_rtl_dse1): ...new function.
10543         (pass_rtl_dse2): Replace declaration with that of...
10544         (make_pass_rtl_dse2): ...new function.
10545         (pass_rtl_dse3): Replace declaration with that of...
10546         (make_pass_rtl_dse3): ...new function.
10547         (pass_rtl_cprop): Replace declaration with that of...
10548         (make_pass_rtl_cprop): ...new function.
10549         (pass_rtl_pre): Replace declaration with that of...
10550         (make_pass_rtl_pre): ...new function.
10551         (pass_rtl_hoist): Replace declaration with that of...
10552         (make_pass_rtl_hoist): ...new function.
10553         (pass_rtl_store_motion): Replace declaration with that of...
10554         (make_pass_rtl_store_motion): ...new function.
10555         (pass_cse_after_global_opts): Replace declaration with that of...
10556         (make_pass_cse_after_global_opts): ...new function.
10557         (pass_rtl_ifcvt): Replace declaration with that of...
10558         (make_pass_rtl_ifcvt): ...new function.
10559         (pass_into_cfg_layout_mode): Replace declaration with that of...
10560         (make_pass_into_cfg_layout_mode): ...new function.
10561         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10562         (make_pass_outof_cfg_layout_mode): ...new function.
10563         (pass_loop2): Replace declaration with that of...
10564         (make_pass_loop2): ...new function.
10565         (pass_rtl_loop_init): Replace declaration with that of...
10566         (make_pass_rtl_loop_init): ...new function.
10567         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10568         (make_pass_rtl_move_loop_invariants): ...new function.
10569         (pass_rtl_unswitch): Replace declaration with that of...
10570         (make_pass_rtl_unswitch): ...new function.
10571         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10572         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10573         (pass_rtl_doloop): Replace declaration with that of...
10574         (make_pass_rtl_doloop): ...new function.
10575         (pass_rtl_loop_done): Replace declaration with that of...
10576         (make_pass_rtl_loop_done): ...new function.
10577         (pass_web): Replace declaration with that of...
10578         (make_pass_web): ...new function.
10579         (pass_cse2): Replace declaration with that of...
10580         (make_pass_cse2): ...new function.
10581         (pass_df_initialize_opt): Replace declaration with that of...
10582         (make_pass_df_initialize_opt): ...new function.
10583         (pass_df_initialize_no_opt): Replace declaration with that of...
10584         (make_pass_df_initialize_no_opt): ...new function.
10585         (pass_reginfo_init): Replace declaration with that of...
10586         (make_pass_reginfo_init): ...new function.
10587         (pass_inc_dec): Replace declaration with that of...
10588         (make_pass_inc_dec): ...new function.
10589         (pass_stack_ptr_mod): Replace declaration with that of...
10590         (make_pass_stack_ptr_mod): ...new function.
10591         (pass_initialize_regs): Replace declaration with that of...
10592         (make_pass_initialize_regs): ...new function.
10593         (pass_combine): Replace declaration with that of...
10594         (make_pass_combine): ...new function.
10595         (pass_if_after_combine): Replace declaration with that of...
10596         (make_pass_if_after_combine): ...new function.
10597         (pass_ree): Replace declaration with that of...
10598         (make_pass_ree): ...new function.
10599         (pass_partition_blocks): Replace declaration with that of...
10600         (make_pass_partition_blocks): ...new function.
10601         (pass_match_asm_constraints): Replace declaration with that of...
10602         (make_pass_match_asm_constraints): ...new function.
10603         (pass_regmove): Replace declaration with that of...
10604         (make_pass_regmove): ...new function.
10605         (pass_split_all_insns): Replace declaration with that of...
10606         (make_pass_split_all_insns): ...new function.
10607         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10608         (make_pass_fast_rtl_byte_dce): ...new function.
10609         (pass_lower_subreg2): Replace declaration with that of...
10610         (make_pass_lower_subreg2): ...new function.
10611         (pass_mode_switching): Replace declaration with that of...
10612         (make_pass_mode_switching): ...new function.
10613         (pass_sms): Replace declaration with that of...
10614         (make_pass_sms): ...new function.
10615         (pass_sched): Replace declaration with that of...
10616         (make_pass_sched): ...new function.
10617         (pass_ira): Replace declaration with that of...
10618         (make_pass_ira): ...new function.
10619         (pass_reload): Replace declaration with that of...
10620         (make_pass_reload): ...new function.
10621         (pass_clean_state): Replace declaration with that of...
10622         (make_pass_clean_state): ...new function.
10623         (pass_branch_prob): Replace declaration with that of...
10624         (make_pass_branch_prob): ...new function.
10625         (pass_value_profile_transformations): Replace declaration with that
10626         of...
10627         (make_pass_value_profile_transformations): ...new function.
10628         (pass_postreload_cse): Replace declaration with that of...
10629         (make_pass_postreload_cse): ...new function.
10630         (pass_gcse2): Replace declaration with that of...
10631         (make_pass_gcse2): ...new function.
10632         (pass_split_after_reload): Replace declaration with that of...
10633         (make_pass_split_after_reload): ...new function.
10634         (pass_branch_target_load_optimize1): Replace declaration with that
10635         of...
10636         (make_pass_branch_target_load_optimize1): ...new function.
10637         (pass_thread_prologue_and_epilogue): Replace declaration with that
10638         of...
10639         (make_pass_thread_prologue_and_epilogue): ...new function.
10640         (pass_stack_adjustments): Replace declaration with that of...
10641         (make_pass_stack_adjustments): ...new function.
10642         (pass_peephole2): Replace declaration with that of...
10643         (make_pass_peephole2): ...new function.
10644         (pass_if_after_reload): Replace declaration with that of...
10645         (make_pass_if_after_reload): ...new function.
10646         (pass_regrename): Replace declaration with that of...
10647         (make_pass_regrename): ...new function.
10648         (pass_cprop_hardreg): Replace declaration with that of...
10649         (make_pass_cprop_hardreg): ...new function.
10650         (pass_reorder_blocks): Replace declaration with that of...
10651         (make_pass_reorder_blocks): ...new function.
10652         (pass_branch_target_load_optimize2): Replace declaration with that
10653         of...
10654         (make_pass_branch_target_load_optimize2): ...new function.
10655         (pass_leaf_regs): Replace declaration with that of...
10656         (make_pass_leaf_regs): ...new function.
10657         (pass_split_before_sched2): Replace declaration with that of...
10658         (make_pass_split_before_sched2): ...new function.
10659         (pass_compare_elim_after_reload): Replace declaration with that of...
10660         (make_pass_compare_elim_after_reload): ...new function.
10661         (pass_sched2): Replace declaration with that of...
10662         (make_pass_sched2): ...new function.
10663         (pass_stack_regs): Replace declaration with that of...
10664         (make_pass_stack_regs): ...new function.
10665         (pass_stack_regs_run): Replace declaration with that of...
10666         (make_pass_stack_regs_run): ...new function.
10667         (pass_df_finish): Replace declaration with that of...
10668         (make_pass_df_finish): ...new function.
10669         (pass_compute_alignments): Replace declaration with that of...
10670         (make_pass_compute_alignments): ...new function.
10671         (pass_duplicate_computed_gotos): Replace declaration with that of...
10672         (make_pass_duplicate_computed_gotos): ...new function.
10673         (pass_variable_tracking): Replace declaration with that of...
10674         (make_pass_variable_tracking): ...new function.
10675         (pass_free_cfg): Replace declaration with that of...
10676         (make_pass_free_cfg): ...new function.
10677         (pass_machine_reorg): Replace declaration with that of...
10678         (make_pass_machine_reorg): ...new function.
10679         (pass_cleanup_barriers): Replace declaration with that of...
10680         (make_pass_cleanup_barriers): ...new function.
10681         (pass_delay_slots): Replace declaration with that of...
10682         (make_pass_delay_slots): ...new function.
10683         (pass_split_for_shorten_branches): Replace declaration with that of...
10684         (make_pass_split_for_shorten_branches): ...new function.
10685         (pass_split_before_regstack): Replace declaration with that of...
10686         (make_pass_split_before_regstack): ...new function.
10687         (pass_convert_to_eh_region_ranges): Replace declaration with that
10688         of...
10689         (make_pass_convert_to_eh_region_ranges): ...new function.
10690         (pass_shorten_branches): Replace declaration with that of...
10691         (make_pass_shorten_branches): ...new function.
10692         (pass_set_nothrow_function_flags): Replace declaration with that of...
10693         (make_pass_set_nothrow_function_flags): ...new function.
10694         (pass_dwarf2_frame): Replace declaration with that of...
10695         (make_pass_dwarf2_frame): ...new function.
10696         (pass_final): Replace declaration with that of...
10697         (make_pass_final): ...new function.
10698         (pass_rtl_seqabstr): Replace declaration with that of...
10699         (make_pass_rtl_seqabstr): ...new function.
10700         (pass_release_ssa_names): Replace declaration with that of...
10701         (make_pass_release_ssa_names): ...new function.
10702         (pass_early_inline): Replace declaration with that of...
10703         (make_pass_early_inline): ...new function.
10704         (pass_inline_parameters): Replace declaration with that of...
10705         (make_pass_inline_parameters): ...new function.
10706         (pass_update_address_taken): Replace declaration with that of...
10707         (make_pass_update_address_taken): ...new function.
10708         (pass_convert_switch): Replace declaration with that of...
10709         (make_pass_convert_switch): ...new function.
10710         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10711         to a subclass of simple_ipa_opt_pass along with...
10712         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10713         (make_pass_ipa_tree_profile): ...new function.
10714         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10715         subclass of gimple_opt_pass along with...
10716         (pass_data_sra_early): ...new pass_data instance and...
10717         (make_pass_sra_early): ...new function.
10718         (pass_sra): Convert from a global struct to a subclass of
10719         gimple_opt_pass along with...
10720         (pass_data_sra): ...new pass_data instance and...
10721         (make_pass_sra): ...new function.
10722         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10723         gimple_opt_pass along with...
10724         (pass_data_early_ipa_sra): ...new pass_data instance and...
10725         (make_pass_early_ipa_sra): ...new function.
10726         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10727         subclass of gimple_opt_pass along with...
10728         (pass_data_ccp): ...new pass_data instance and...
10729         (make_pass_ccp): ...new function.
10730         (pass_fold_builtins): Convert from a global struct to a subclass of
10731         gimple_opt_pass along with...
10732         (pass_data_fold_builtins): ...new pass_data instance and...
10733         (make_pass_fold_builtins): ...new function.
10734         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10735         subclass of gimple_opt_pass along with...
10736         (pass_data_copy_prop): ...new pass_data instance and...
10737         (make_pass_copy_prop): ...new function.
10738         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10739         global struct to a subclass of gimple_opt_pass along with...
10740         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10741         (make_pass_rename_ssa_copies): ...new function.
10742         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10743         subclass of gimple_opt_pass along with...
10744         (pass_data_dce): ...new pass_data instance and...
10745         (make_pass_dce): ...new function.
10746         (pass_dce_loop): Convert from a global struct to a subclass of
10747         gimple_opt_pass along with...
10748         (pass_data_dce_loop): ...new pass_data instance and...
10749         (make_pass_dce_loop): ...new function.
10750         (pass_cd_dce): Convert from a global struct to a subclass of
10751         gimple_opt_pass along with...
10752         (pass_data_cd_dce): ...new pass_data instance and...
10753         (make_pass_cd_dce): ...new function.
10754         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10755         subclass of gimple_opt_pass along with...
10756         (pass_data_dominator): ...new pass_data instance and...
10757         (make_pass_dominator): ...new function.
10758         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10759         gimple_opt_pass along with...
10760         (pass_data_phi_only_cprop): ...new pass_data instance and...
10761         (make_pass_phi_only_cprop): ...new function.
10762         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10763         subclass of gimple_opt_pass along with...
10764         (pass_data_dse): ...new pass_data instance and...
10765         (make_pass_dse): ...new function.
10766         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10767         a subclass of gimple_opt_pass along with...
10768         (pass_data_forwprop): ...new pass_data instance and...
10769         (make_pass_forwprop): ...new function.
10770         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10771         struct to a subclass of gimple_opt_pass along with...
10772         (pass_data_tree_ifcombine): ...new pass_data instance and...
10773         (make_pass_tree_ifcombine): ...new function.
10774         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10775         subclass of gimple_opt_pass along with...
10776         (pass_data_ch): ...new pass_data instance and...
10777         (make_pass_ch): ...new function.
10778         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10779         subclass of gimple_opt_pass along with...
10780         (pass_data_tree_loop): ...new pass_data instance and...
10781         (make_pass_tree_loop): ...new function.
10782         (pass_tree_loop_init): Convert from a global struct to a subclass of
10783         gimple_opt_pass along with...
10784         (pass_data_tree_loop_init): ...new pass_data instance and...
10785         (make_pass_tree_loop_init): ...new function.
10786         (pass_lim): Convert from a global struct to a subclass of
10787         gimple_opt_pass along with...
10788         (pass_data_lim): ...new pass_data instance and...
10789         (make_pass_lim): ...new function.
10790         (pass_tree_unswitch): Convert from a global struct to a subclass of
10791         gimple_opt_pass along with...
10792         (pass_data_tree_unswitch): ...new pass_data instance and...
10793         (make_pass_tree_unswitch): ...new function.
10794         (pass_predcom): Convert from a global struct to a subclass of
10795         gimple_opt_pass along with...
10796         (pass_data_predcom): ...new pass_data instance and...
10797         (make_pass_predcom): ...new function.
10798         (pass_vectorize): Convert from a global struct to a subclass of
10799         gimple_opt_pass along with...
10800         (pass_data_vectorize): ...new pass_data instance and...
10801         (make_pass_vectorize): ...new function.
10802         (pass_graphite): Convert from a global struct to a subclass of
10803         gimple_opt_pass along with...
10804         (pass_data_graphite): ...new pass_data instance and...
10805         (make_pass_graphite): ...new function.
10806         (pass_graphite_transforms): Convert from a global struct to a subclass
10807         of gimple_opt_pass along with...
10808         (pass_data_graphite_transforms): ...new pass_data instance and...
10809         (make_pass_graphite_transforms): ...new function.
10810         (pass_check_data_deps): Convert from a global struct to a subclass of
10811         gimple_opt_pass along with...
10812         (pass_data_check_data_deps): ...new pass_data instance and...
10813         (make_pass_check_data_deps): ...new function.
10814         (pass_iv_canon): Convert from a global struct to a subclass of
10815         gimple_opt_pass along with...
10816         (pass_data_iv_canon): ...new pass_data instance and...
10817         (make_pass_iv_canon): ...new function.
10818         (pass_scev_cprop): Convert from a global struct to a subclass of
10819         gimple_opt_pass along with...
10820         (pass_data_scev_cprop): ...new pass_data instance and...
10821         (make_pass_scev_cprop): ...new function.
10822         (pass_record_bounds): Convert from a global struct to a subclass of
10823         gimple_opt_pass along with...
10824         (pass_data_record_bounds): ...new pass_data instance and...
10825         (make_pass_record_bounds): ...new function.
10826         (pass_complete_unroll): Convert from a global struct to a subclass of
10827         gimple_opt_pass along with...
10828         (pass_data_complete_unroll): ...new pass_data instance and...
10829         (make_pass_complete_unroll): ...new function.
10830         (pass_complete_unrolli): Convert from a global struct to a subclass of
10831         gimple_opt_pass along with...
10832         (pass_data_complete_unrolli): ...new pass_data instance and...
10833         (make_pass_complete_unrolli): ...new function.
10834         (pass_parallelize_loops): Convert from a global struct to a subclass
10835         of gimple_opt_pass along with...
10836         (pass_data_parallelize_loops): ...new pass_data instance and...
10837         (make_pass_parallelize_loops): ...new function.
10838         (pass_loop_prefetch): Convert from a global struct to a subclass of
10839         gimple_opt_pass along with...
10840         (pass_data_loop_prefetch): ...new pass_data instance and...
10841         (make_pass_loop_prefetch): ...new function.
10842         (pass_iv_optimize): Convert from a global struct to a subclass of
10843         gimple_opt_pass along with...
10844         (pass_data_iv_optimize): ...new pass_data instance and...
10845         (make_pass_iv_optimize): ...new function.
10846         (pass_tree_loop_done): Convert from a global struct to a subclass of
10847         gimple_opt_pass along with...
10848         (pass_data_tree_loop_done): ...new pass_data instance and...
10849         (make_pass_tree_loop_done): ...new function.
10850         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10851         struct to a subclass of gimple_opt_pass along with...
10852         (pass_data_cse_reciprocals): ...new pass_data instance and...
10853         (make_pass_cse_reciprocals): ...new function.
10854         (pass_cse_sincos): Convert from a global struct to a subclass of
10855         gimple_opt_pass along with...
10856         (pass_data_cse_sincos): ...new pass_data instance and...
10857         (make_pass_cse_sincos): ...new function.
10858         (pass_optimize_bswap): Convert from a global struct to a subclass of
10859         gimple_opt_pass along with...
10860         (pass_data_optimize_bswap): ...new pass_data instance and...
10861         (make_pass_optimize_bswap): ...new function.
10862         (pass_optimize_widening_mul): Convert from a global struct to a
10863         subclass of gimple_opt_pass along with...
10864         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10865         (make_pass_optimize_widening_mul): ...new function.
10866         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10867         subclass of gimple_opt_pass along with...
10868         (pass_data_phiopt): ...new pass_data instance and...
10869         (make_pass_phiopt): ...new function.
10870         (pass_cselim): Convert from a global struct to a subclass of
10871         gimple_opt_pass along with...
10872         (pass_data_cselim): ...new pass_data instance and...
10873         (make_pass_cselim): ...new function.
10874         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10875         subclass of gimple_opt_pass along with...
10876         (pass_data_phiprop): ...new pass_data instance and...
10877         (make_pass_phiprop): ...new function.
10878         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10879         subclass of gimple_opt_pass along with...
10880         (pass_data_pre): ...new pass_data instance and...
10881         (make_pass_pre): ...new function.
10882         (pass_fre): Convert from a global struct to a subclass of
10883         gimple_opt_pass along with...
10884         (pass_data_fre): ...new pass_data instance and...
10885         (make_pass_fre): ...new function.
10886         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10887         subclass of gimple_opt_pass along with...
10888         (pass_data_reassoc): ...new pass_data instance and...
10889         (make_pass_reassoc): ...new function.
10890         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10891         subclass of gimple_opt_pass along with...
10892         (pass_data_sink_code): ...new pass_data instance and...
10893         (make_pass_sink_code): ...new function.
10894         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10895         subclass of gimple_opt_pass along with...
10896         (pass_data_strlen): ...new pass_data instance and...
10897         (make_pass_strlen): ...new function.
10898         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10899         struct to a subclass of gimple_opt_pass along with...
10900         (pass_data_build_alias): ...new pass_data instance and...
10901         (make_pass_build_alias): ...new function.
10902         (pass_build_ealias): Convert from a global struct to a subclass of
10903         gimple_opt_pass along with...
10904         (pass_data_build_ealias): ...new pass_data instance and...
10905         (make_pass_build_ealias): ...new function.
10906         (pass_ipa_pta): Convert from a global struct to a subclass of
10907         simple_ipa_opt_pass along with...
10908         (pass_data_ipa_pta): ...new pass_data instance and...
10909         (make_pass_ipa_pta): ...new function.
10910         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10911         subclass of gimple_opt_pass along with...
10912         (pass_data_uncprop): ...new pass_data instance and...
10913         (make_pass_uncprop): ...new function.
10914         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10915         global struct to a subclass of gimple_opt_pass along with...
10916         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10917         (make_pass_late_warn_uninitialized): ...new function.
10918         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10919         to a subclass of gimple_opt_pass along with...
10920         (pass_data_init_datastructures): ...new pass_data instance and...
10921         (make_pass_init_datastructures): ...new function.
10922         (pass_early_warn_uninitialized): Convert from a global struct to a
10923         subclass of gimple_opt_pass along with...
10924         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10925         (make_pass_early_warn_uninitialized): ...new function.
10926         (pass_update_address_taken): Convert from a global struct to a
10927         subclass of gimple_opt_pass along with...
10928         (pass_data_update_address_taken): ...new pass_data instance and...
10929         (make_pass_update_address_taken): ...new function.
10930         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10931         struct to a subclass of gimple_opt_pass along with...
10932         (pass_data_release_ssa_names): ...new pass_data instance and...
10933         (make_pass_release_ssa_names): ...new function.
10934         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10935         subclass of gimple_opt_pass along with...
10936         (pass_data_stdarg): ...new pass_data instance and...
10937         (make_pass_stdarg): ...new function.
10938         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10939         global struct to a subclass of gimple_opt_pass along with...
10940         (pass_data_convert_switch): ...new pass_data instance and...
10941         (make_pass_convert_switch): ...new function.
10942         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10943         to a subclass of gimple_opt_pass along with...
10944         (pass_data_tail_recursion): ...new pass_data instance and...
10945         (make_pass_tail_recursion): ...new function.
10946         (pass_tail_calls): Convert from a global struct to a subclass of
10947         gimple_opt_pass along with...
10948         (pass_data_tail_calls): ...new pass_data instance and...
10949         (make_pass_tail_calls): ...new function.
10950         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10951         struct to a subclass of gimple_opt_pass along with...
10952         (pass_data_lower_vector): ...new pass_data instance and...
10953         (make_pass_lower_vector): ...new function.
10954         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
10955         gimple_opt_pass along with...
10956         (pass_data_lower_vector_ssa): ...new pass_data instance and...
10957         (make_pass_lower_vector_ssa): ...new function.
10958         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
10959         to a subclass of gimple_opt_pass along with...
10960         (pass_data_slp_vectorize): ...new pass_data instance and...
10961         (make_pass_slp_vectorize): ...new function.
10962         (pass_ipa_increase_alignment): Convert from a global struct to a
10963         subclass of simple_ipa_opt_pass along with...
10964         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
10965         (make_pass_ipa_increase_alignment): ...new function.
10966         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
10967         gimple_opt_pass along with...
10968         (pass_data_vrp): ...new pass_data instance and...
10969         (make_pass_vrp): ...new function.
10970         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
10971         subclass of simple_ipa_opt_pass along with...
10972         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
10973         (make_pass_ipa_free_lang_data): ...new function.
10974         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
10975         gimple_opt_pass along with...
10976         (pass_data_tsan): ...new pass_data instance and...
10977         (make_pass_tsan): ...new function.
10978         (pass_tsan_O0): Convert from a global struct to a subclass of
10979         gimple_opt_pass along with...
10980         (pass_data_tsan_O0): ...new pass_data instance and...
10981         (make_pass_tsan_O0): ...new function.
10982         * var-tracking.c (pass_variable_tracking): Convert from a global
10983         struct to a subclass of rtl_opt_pass along with...
10984         (pass_data_variable_tracking): ...new pass_data instance and...
10985         (make_pass_variable_tracking): ...new function.
10986         * web.c (pass_web): Convert from a global struct to a subclass of
10987         rtl_opt_pass along with...
10988         (pass_data_web): ...new pass_data instance and...
10989         (make_pass_web): ...new function.
10990         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
10991         declaration with that of...
10992         (make_pass_mode_switch_use): ...new function.
10993         (pass_resolve_sw_modes): Replace declaration with that of...
10994         (make_pass_resolve_sw_modes): ...new function.
10995         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
10996         from a global struct to a subclass of rtl_opt_pass along with...
10997         (pass_data_mode_switch_use): ...new pass_data instance and...
10998         (make_pass_mode_switch_use): ...new function.
10999         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11000         from a global struct to a subclass of rtl_opt_pass along with...
11001         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11002         (make_pass_resolve_sw_modes): ...new function.
11003         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11004         struct to a subclass of rtl_opt_pass along with...
11005         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11006         (make_pass_insert_vzeroupper): ...new function.
11007         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11008         global struct to a subclass of rtl_opt_pass along with...
11009         (pass_data_work_around_errata): ...new pass_data instance and...
11010         (make_pass_work_around_errata): ...new function.
11011         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11012         struct to a subclass of rtl_opt_pass along with...
11013         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11014         (make_pass_mips_machine_reorg2): ...new function.
11016 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11018         * passes.c (pass_manager::operator new): New.
11020 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11022         Handwritten part of conversion of passes to C++ classes.
11024         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11025         (toplev.o): Add dep on PASS_MANAGER_H.
11026         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11027         of early local pases to reflect this moving from a global to a
11028         member of gcc::pass_manager.
11029         (cgraph_add_new_function): Likewise.
11030         * lto-cgraph.c (lto_output_node): Update for conversion of
11031         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11032         * passes.c (opt_pass::clone): New.
11033         (opt_pass::gate): New.
11034         (opt_pass::execute): New.
11035         (opt_pass::opt_pass): New.
11036         (pass_manager::execute_early_local_passes): New.
11037         (pass_manager::execute_pass_mode_switching): new.
11038         (finish_optimization_passes): Convert to...
11039         (pass_manager::finish_optimization_passes): ...this.
11040         (finish_optimization_passes): Update for conversion of passes to
11041         C++ classes.
11042         (register_dump_files_1): Use has_gate since we cannot portably
11043         check a vtable entry against NULL.
11044         (dump_one_pass): Likewise.
11045         (ipa_write_summaries_2): Likewise.
11046         (ipa_write_optimization_summaries_1): Likewise.
11047         (ipa_read_summaries_1): Likewise.
11048         (ipa_read_optimization_summaries_1): Likewise.
11049         (execute_ipa_stmt_fixups): Likewise.
11050         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11051         pass-creation functions rather than wiring up globals, and
11052         storing the results in fields of pass_manager generated using
11053         pass-instances.def.
11054         (pass_manager::dump_profile_report): Update for conversion of
11055         passes to C++ classes.
11056         (pass_manager::execute_ipa_summary_passes): Likewise.
11057         (execute_one_ipa_transform_pass): Likewise.
11058         (execute_one_pass): Use has_gate and has_execute since we cannot
11059         portably check a vtable entry against NULL.
11060         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11061         (pass_manager): Use pass-instances.def to add fields for the
11062         various pass instances.
11063         * toplev.c (finalize): Update for move of
11064         finish_optimization_passes to a method of gcc::pass_manager.
11065         * toplev.h (finish_optimization_passes): Move to method of class
11066         pass_manager.
11067         * tree-pass.h (struct pass_data): New.
11068         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11069         (opt_pass::gate): Convert to virtual function.
11070         (opt_pass::~opt_pass): New.
11071         (opt_pass::clone): New.
11072         (opt_pass::execute): Convert to virtual function.
11073         (opt_pass::opt_pass): New.
11074         (opt_pass::ctxt_): new.
11075         (gimple_opt_pass): Convert to subclass of opt_pass.
11076         (gimple_opt_pass::gimple_opt_pass): New.
11077         (rtl_opt_pass): Convert to subclass of opt_pass.
11078         (rtl_opt_pass::rtl_opt_pass): New.
11079         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11080         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11081         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11082         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11083         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11084         invocation of pass_mode_switching to reflect this moving from a
11085         global to a member of gcc::pass_manager.
11086         (ix86_option_override): Rework how pass_insert_vzeroupper is
11087         added to the pass_manager to reflect autogenerated changes.
11088         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11090 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11092         PR rtl-optimization/57708
11093         * recog.c (peep2_find_free_register): Validate all regs in a
11094         multi-reg mode.
11096 2013-08-05  Jan Hubicka  <jh@suse.cz>
11098         PR lto/57602
11099         * cgraph.c (verify_cgraph_node): Accept local flags from other
11100         partitions.
11101         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11102         (function_and_variable_visibility): Likewise.
11103         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11105 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11107         * graph.c (init_graph_slim_pretty_print): Remove.
11108         (print_graph_cfg): Do not call it.  Use local pretty printer.
11109         (start_graph_dump): Likewise.
11111 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11113         * gimple-pretty-print.c (buffer): Remove.
11114         (initialized): Likewise.
11115         (maybe_init_pretty_print): Likewise.
11116         (print_gimple_stmt): Do not call it.  Use non-static local
11117         pretty_printer variable.
11118         (print_gimple_expr): Likewise.
11119         (print_gimple_seq): Likewise.
11120         (gimple_dump_bb): Likewise.
11122 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11124         * asan.c (asan_pp): Remove.
11125         (asan_pp_initialized): Likewise.
11126         (asan_pp_initialize): Likewise.
11127         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11128         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11129         (asan_add_global): Likewise.
11131 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11133         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11134         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11135         * pretty-print.c (pp_formatted_text_data): Likewise.
11136         (pp_indent): Rename from pp_base_indent.
11137         (pp_format): Rename from pp_base_format.
11138         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11139         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11140         (pp_flush): Rename from pp_base_flush.
11141         (pp_set_line_maximum_length): Rename from
11142         pp_base_set_line_maximum_length.
11143         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11144         (pp_set_prefix): Rename from pp_base_set_prefix.
11145         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11146         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11147         (pp_append_text): Rename from pp_base_append_text.
11148         (pp_formatted_text): Rename from pp_base_formatted_text.
11149         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11150         (pp_remaining_character_count_for_line): Rename from
11151         pp_base_remaining_character_count_for_line.
11152         (pp_newline): Rename from pp_base_newline.
11153         (pp_character): Rename from pp_base_character.
11154         (pp_string): Rename from pp_base_string.
11155         (pp_maybe_space): Rename from pp_base_maybe_space.
11156         * asan.c (asan_pp_string): Adjust.
11157         (asan_emit_stack_protection): Likewise.
11158         (asan_add_global): Likewise.
11159         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11160         * tree-mudflap.c (mf_varname_tree): Likewise.
11161         * tree-pretty-print.c (pp_tree_identifier): Rename from
11162         pp_base_tree_identifier.
11163         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11164         Declare as function.
11166 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11168         * pretty-print.h (pp_bar_bar): New.
11169         (pp_ampersand_ampersand): Likewise.
11170         (pp_less_equal): Likewise.
11171         (pp_greater_equal): Likewise.
11172         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11173         printer functions instead of pp_string or operators and punctuators.
11174         (dump_gimple_call): Likewise.
11175         (dump_gimple_omp_for): Likewise.
11176         (dump_gimple_transaction): Likewise.
11177         (dump_gimple_phi): Likewise.
11178         (pp_gimple_stmt_1): Likewise.
11179         * sched-vis.c (print_insn): Likewise.
11180         * tree-mudflap.c (mf_varname_tree): Likewise.
11181         * tree-pretty-print.c (dump_block_node): Likewise.
11182         (dump_generic_node): Likewise.
11184 2013-08-02  Jan Hubicka  <jh@suse.cz>
11186         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11187         boundaries.
11188         * lto-streamer-out.c (tree_is_indexable): Results decls and
11189         parm decls are not indexable.
11190         (DFS_write_tree_body): Do not follow args and results.
11191         (hash_tree): Likewise.
11192         (output_functions): Rearrange so struct function is needed
11193         only when real body is output; be able to also ouptut abstract
11194         functions; output DECL_ARGUMENTS and DECL_RESULT.
11195         (lto_output): When not in WPA, ale store abstract functions.
11196         (write_symbol): Do not care about RESULT_DECL.
11197         (output_symbol_p): Handle correctly sbtract decls.
11198         * lto-streamer-in.c (input_function): Rearrange so struct
11199         function can be NULL at entry; allow streaming of
11200         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11201         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11202         sanity check during LTO.
11203         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11204         RESULT_DECl and DECL_ARGUMENTS.
11205         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11206         Likewise.
11208 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11210         * pretty-print.h (pp_underscore): New.
11211         (pp_comma): Tidy.
11212         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11213         printer functions instead of pp_character.
11214         (dump_binary_rhs): Likewise.
11215         (dump_ternary_rhs): Likewise.
11216         (dump_gimple_call_args): Likewise.
11217         (pp_points_to_solution): Likewise.
11218         (dump_gimple_call): Likewise.
11219         (dump_gimple_switch): Likewise.
11220         (dump_gimple_cond): Likewise.
11221         (dump_gimple_bind): Likewise.
11222         (dump_gimple_try): Likewise.
11223         (dump_gimple_omp_for): Likewise.
11224         (dump_gimple_omp_continue): Likewise.
11225         (dump_gimple_omp_single): Likewise.
11226         (dump_gimple_omp_sections): Likewise.
11227         (dump_gimple_omp_block): Likewise.
11228         (dump_gimple_omp_critical): Likewise.
11229         (dump_gimple_transaction): Likewise.
11230         (dump_gimple_asm): Likewise.
11231         (dump_gimple_phi): Likewise.
11232         (dump_gimple_omp_parallel): Likewise.
11233         (dump_gimple_omp_task): Likewise.
11234         (dump_gimple_omp_atomic_load): Likewise.
11235         (dump_gimple_omp_atomic_store): Likewise.
11236         (dump_gimple_mem_ops): Likewise.
11237         (pp_gimple_stmt_1): Likewise.
11238         (pp_cfg_jump): Likewise.
11239         (dump_implicit_edges): Likewise.
11240         (gimple_dump_bb_for_graph): Likewise.
11241         * graph.c (draw_cfg_node): Likewise.
11242         * langhooks.c (lhd_print_error_function): Likewise.
11243         * sched-vis.c (print_exp): Likewise.
11244         (print_value): Likewise.
11245         (print_pattern): Likewise.
11246         (print_insn): Likewise.
11247         (rtl_dump_bb_for_graph): Likewise.
11248         * tree-pretty-print.c (dump_function_declaration): Likewise.
11249         (dump_array_domain): Likewise.
11250         (dump_omp_clause): Likewise.
11251         (dump_location): Likewise.
11252         (dump_generic_node): Likewise.
11253         (print_struct_decl): Likewise.
11254         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11256 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11258         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11259         candidate table when replacing a candidate statement.
11260         (replace_rhs_if_not_dup): Likewise.
11261         (replace_one_candidate): Likewise.
11263 2013-08-02  Jan Hubicka  <jh@suse.cz>
11264             Martin Liska  <marxin.liska@gmail.com>
11266         * cgraphunit.c (add_new_function): Fix logic when adding from
11267         late IPA pass.
11268         (assemble_thunk): Rename to ...
11269         (expand_thunk); .. this one; export; get it working with
11270         general functions; make produced gimple valid.
11271         * cgraph.h (expand_thunk): Declare.
11273 2013-08-02  Jan Hubicka  <jh@suse.cz>
11275         * ipa-cp.c (gather_context_independent_values): Use
11276         ipa_get_param_move_cost.
11277         (get_replacement_map): Remove PARAM; move parameter folding
11278         into tree-inline.c
11279         (create_specialized_node): Update.
11280         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11281         assert that we have gimple body; update move_cost.
11282         (count_formal_params): Assert that we have gimple body.
11283         (ipa_dump_param): New function.
11284         (ipa_alloc_node_params): Break out from ...
11285         (ipa_initialize_node_params): ... here.
11286         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11287         (ipa_write_node_info): Stream move costs.
11288         (ipa_read_node_info): Read move costs.
11289         (ipa_update_after_lto_read): Do not recompute node params.
11290         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11291         (ipa_get_param): Check we are not in WPA.
11292         (ipa_get_param_move_cost): New.
11293         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11294         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11295         parm numbers to be present.
11297 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11299         PR rtl-optimization/58048
11300         * lra-constraints.c (process_alt_operands): Don't check asm
11301         operand on register.
11303 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11305         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11306         the implied StoreLoad barrier for atomic operations if before.
11308 2013-08-02  Jan Hubicka  <jh@suse.cz>
11309             Martin Liska  <marxin.liska@gmail.com>
11311         * cgraph.c (cgraph_function_body_availability): Do not check
11312         cgraph flags.
11313         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11314         symtab_node_availability): Declare.
11315         * ipa.c (can_replace_by_local_alias): New.
11316         (function_and_variable_visibility): Use it.
11317         * symtab.c (symtab_for_node_and_aliases,
11318         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11320 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11322         PR rtl-optimization/57963
11323         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11324         (lra_constraints): Use them.
11326 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11328         * config/arm/types.md (define_attr "type"): Add "load_acq"
11329         and "store_rel".
11330         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11331         changes.
11332         (cortex_a53_store1): Likewise.
11334 2013-08-01  Jan Hubicka  <jh@suse.cz>
11336         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11337         partitions are not needed.
11339 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11341         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11342         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11343         MAYBE_NON_Q_CLASS_P where appropriate.
11345 2013-08-01  Jan Hubicka  <jh@suse.cz>
11347         * cgraph.h (release_function_body): Declare.
11348         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11349         of unused delcarations.
11351 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11353         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11354         RTL form when subtracting a constant.
11356 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11358         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11359         Generate canonical plus rtx with negated immediate instead of minus
11360         where appropriate.
11361         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11363 2013-08-01  Jan Hubicka  <jh@suse.cz>
11365         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11366         (cgraph_release_function_body): Likewise.
11367         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11368         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11369         to used_as_abstract_origin.
11370         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11371         symbols used as abstract origins.
11372         * cgraphunit.c (analyze_functions): Update.
11373         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11374         used_as_abstract_origin.
11375         * tree-inline.c (tree_function_versioning): Update
11376         used_as_abstract_origin; be ready for DECL_RESULT and
11377         DECL_ARGUMENTS to be NULL.
11379         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11380         for abstract functions.
11381         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11382         real symbols.
11384 2013-08-01  Jan Hubicka  <jh@suse.cz>
11386         * profile.c (compute_value_histograms): Fix thinko.
11388 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11390         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11391         aarch-common-protos.h to extra_headers.
11392         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11393         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11394         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11396 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11398         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11399         Include "../arm/types.md".  Define "type" attribute for all patterns.
11400         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11401         attribute changes.
11403 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11405         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11406         to support power8 load fusion.
11407         (fusion_gpr_mem_load): Likewise.
11409         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11411         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11412         declarations for power8 load fusion.
11413         (emit_fusion_gpr_load): Likewise.
11415         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11416         tuning for power8, turn on fusion mode by default.  Turn on sign
11417         extending fusion mode if normal fusion mode is on, and we are at
11418         -O2 or -O3.
11419         (fusion_gpr_load_p): New function, return true if we can fuse an
11420         addis instruction with a dependent load to a GPR.
11421         (emit_fusion_gpr_load): Emit the instructions for power8 load
11422         fusion to GPRs.
11424         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11425         (VSX load fusion peepholes): New peepholes to fuse together an
11426         addi instruction with a VSX load instruction.
11428         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11429         peepholes to fuse an addis instruction with a load to a GPR base
11430         register.  If we are supporting sign extending fusions, convert
11431         sign extending loads to zero extending loads and add an explicit
11432         sign extension.
11434 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11436         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11437         aarch-common-protos.h to extra_headers.
11438         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11439         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11440         (arm_early_store_addr_dep): Likewise.
11441         (arm_no_early_alu_shift_dep): Likewise.
11442         (arm_no_early_alu_shift_value_dep): Likewise.
11443         (arm_no_early_mul_dep): Likewise.
11444         (arm_no_early_store_addr_dep): Likewise.
11445         (arm_mac_accumulator_is_mul_result): Likewise.
11446         (arm_mac_accumulator_is_result): Likewise.
11447         * config/arm/aarch-common.c: ... here.  New file.
11448         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11449         here to ...
11450         (arm_early_store_addr_dep): Likewise.
11451         (arm_no_early_alu_shift_dep): Likewise.
11452         (arm_no_early_alu_shift_value_dep): Likewise.
11453         (arm_no_early_mul_dep): Likewise.
11454         (arm_no_early_store_addr_dep): Likewise.
11455         (arm_mac_accumulator_is_mul_result): Likewise.
11456         (arm_mac_accumulator_is_result): Likewise.
11457         * config/arm/aarch-common-protos.h: ... here.  New file.
11458         * config/arm/t-arm (aarch-common.o): Define.
11460 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11462         * config/arm/arm.md: Include new file "types.md".
11463         (define_attr "type"): Move from here to ...
11464         (define_attr "mul32"): Likewise.
11465         (define_attr "mul64"): Likewise.
11466         * config/arm/types.md: ... here.  New file.
11468 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11470         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11471         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11473 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11475         * gen-pass-instances.awk: Fix offset of substr().
11477 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11479         * Makefile.in (pass-instances.def): New.
11480         (passes.o): Replace dependency on passes.def with one on
11481         pass-instances.def
11483         * gen-pass-instances.awk: New.
11485         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11486         rather than passes.def, updating local definition of NEXT_PASS
11487         macro to add an extra NUM parameter (currently unused).
11489 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11491         * Makefile.in (PASS_MANAGER_H): New.
11492         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11493         (passes.o): Likewise.
11494         (statistics.o): Likewise.
11495         (cgraphunit.o): Likewise.
11496         (context.o): Depend on PASS_MANAGER_H.
11498         * pass_manager.h: New.
11500         * cgraphunit.c (cgraph_add_new_function): Update for moves
11501         of globals to fields of pass_manager.
11502         (analyze_function): Likewise.
11503         (expand_function): Likewise.
11504         (ipa_passes): Likewise.
11505         (compile): Likewise.
11507         * context.c (context::context): New.
11508         * context.h  (context::context): New.
11509         (context::get_passes): New.
11510         (context::passes_): New.
11512         * lto-cgraph.c (input_node): Update for moves of globals to
11513         fields of pass_manager.
11515         * passes.c (all_passes): Remove, in favor of a field of the
11516         same name within the new class pass_manager.
11517         (all_small_ipa_passes): Likewise.
11518         (all_lowering_passes): Likewise.
11519         (all_regular_ipa_passes): Likewise.
11520         (all_late_ipa_passes): Likewise.
11521         (all_lto_gen_passes): Likewise.
11522         (passes_by_id): Likewise.
11523         (passes_by_id_size): Likewise.
11524         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11525         the new class pass_manager.
11526         (set_pass_for_id): Convert to...
11527         (pass_manager::set_pass_for_id): ...method.
11528         (get_pass_for_id): Convert to...
11529         (pass_manager::get_pass_for_id): ...method.
11530         (register_one_dump_file): Move body of implementation into...
11531         (pass_manager::register_one_dump_file): ...here.
11532         (register_dump_files_1): Convert to...
11533         (pass_manager::register_dump_files_1): ...method.
11534         (register_dump_files): Convert to...
11535         (pass_manager::register_dump_files): ...method.
11536         (create_pass_tab): Update for moves of globals to fields of
11537         pass_manager.
11538         (dump_passes): Move body of implementation into...
11539         (pass_manager::dump_passes): ...here.
11540         (register_pass): Move body of implementation into...
11541         (pass_manager::register_pass): ...here.
11542         (init_optimization_passes): Convert into...
11543         (pass_manager::pass_manager): ...constructor for new
11544         pass_manager class, and initialize the pass_lists array.
11545         (check_profile_consistency): Update for moves of globals to
11546         fields of pass_manager.
11547         (dump_profile_report): Move body of implementation into...
11548         (pass_manager::dump_profile_report): ...here.
11549         (ipa_write_summaries_1): Update for moves of pass lists from
11550         being globals to fields of pass_manager.
11551         (ipa_write_optimization_summaries): Likewise.
11552         (ipa_read_summaries):  Likewise.
11553         (ipa_read_optimization_summaries): Likewise.
11554         (execute_all_ipa_stmt_fixups): Likewise.
11556         * statistics.c (statistics_fini): Update for moves of globals to
11557         fields of pass_manager.
11559         * toplev.c (general_init): Replace call to
11560         init_optimization_passes with construction of the pass_manager
11561         instance.
11563         * tree-pass.h (all_passes): Remove, in favor of a field of the
11564         same name within the new class pass_manager.
11565         (all_small_ipa_passes): Likewise.
11566         (all_lowering_passes): Likewise.
11567         (all_regular_ipa_passes): Likewise.
11568         (all_lto_gen_passes): Likewise.
11569         (all_late_ipa_passes): Likewise.
11570         (passes_by_id): Likewise.
11571         (passes_by_id_size): Likewise.
11572         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11573         the new class pass_manager.
11574         (get_pass_for_id): Remove.
11576 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11578         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11579         configs.
11581 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11583         * arm.md (mulhi3): New expand pattern.
11585 2013-07-30  Jan Hubicka  <jh@suse.cz>
11586             Martin Liska  <marxin.liska@gmail.com>
11588         * profile.c (compute_value_histograms): Do not ICE when
11589         there is mismatch only on some counters.
11591 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11593         PR rtl-optimization/57637
11594         * function.c (move_insn_for_shrink_wrap): Also check the
11595         GEN set of the LIVE problem for the liveness analysis
11596         if it exists, otherwise give up.
11598 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11600         PR tree-optimization/57993
11601         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11602         replaced statement in the candidate table.
11603         (phi_add_costs): Return infinite cost when the hidden basis does
11604         not dominate all phis on which the candidate is dependent.
11605         (replace_one_candidate): Record replaced statement in the
11606         candidate table.
11608 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11610         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11611         (ashlv2si3): New expander.
11612         (*ashlv2si3_i): New define_insn_and_split.
11613         * predicates.md (float_operation): Allow patterns with three
11614         basic sub-patterns.
11616         PR rtl-optimization/58021
11617         * mode-switching.c (create_pre_exit): Always split off preceding
11618         insns if we are not at the basic block head.
11620 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11622         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11623         (UCLIBC_DYNAMIC_LINKER): New macro.
11624         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11625         `-mnan=2008'.
11626         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11627         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11628         `-mnan=2008'.
11629         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11630         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11631         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11632         for SF and DF modes.  Use ieee_quad_format for TF mode.
11633         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11634         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11635         (mips_option_override): Handle `-mnan=legacy'.
11636         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11637         `-mabs=2008' and `-mnan=2008'.
11638         (OPTION_DEFAULT_SPECS): Add "nan" default.
11639         (ASM_SPEC): Handle `-mnan='.
11640         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11641         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11642         comment accordingly.
11643         (neg<mode>2): Likewise.
11644         * config/mips/mips.opt (mabs, mnan): New options.
11645         * doc/install.texi (Configuration): Document `--with-nan=' option.
11646         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11647         `-mnan=' options.
11648         (MIPS Options): Document them.
11649         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11650         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11651         * configure: Regenerate.
11652         * config.in: Regenerate.
11654 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11656         * config/i386/i386.md (float post-reload splitters): Do not check
11657         for subregs of SSE registers.
11659 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11660             H.J. Lu  <hongjiu.lu@intel.com>
11662         PR target/57954
11663         PR target/57988
11664         * config/i386/i386.md (post-reload splitter
11665         to avoid partial SSE reg dependency stalls): New pattern.
11667 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11669         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11671 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11673         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11674         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11676 2013-07-26  Cary Coutant  <ccoutant@google.com>
11678         * dwarf2out.c (die_checksum_ordered): Don't include template
11679         instantiations in signature.
11680         (is_template_parameter): New function.
11681         (is_template_instantiation): New function.
11682         (generate_skeleton_bottom_up): Don't include template instantiations
11683         in type unit DIE.
11684         (generate_skeleton): Likewise.
11685         (break_out_comdat_types): Move recursive call to break out nested
11686         types earlier.
11687         (prune_unused_types_mark_generic_parms_dies): Call
11688         is_template_parameter.
11690 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11692         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11693         uses vector registers.
11694         * config/aarch64/iterators.md: Add attributes rtn and vas.
11696 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11697             Richard Earnshaw  <richard.earnshaw@arm.com>
11699         * combine.c (simplify_comparison): Re-canonicalize operands
11700         where appropriate.
11701         * config/arm/arm.md (movcond_addsi): New splitter.
11703 2013-07-25  Sterling Augustine  <saugustine@google.com>
11705         * dwarf2out.c (size_of_pubnames): Move code to...
11706         (include_pubname_in_output): ...here.  New.
11707         (want_pubnames): Rearrange.
11708         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11710 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11712         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11714 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11716         PR target/38836
11717         * doc/extend.texi: Remove obsolete builtins. Fix
11718         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11720 2013-07-25  Jan Hubicka  <jh@suse.cz>
11722         * cgraph.c (release_function_body): Break out from ...
11723         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11724         and DECL_ARGUMENTS.
11725         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11726         old_tree in the map.
11727         (create_specialized_node): Update.
11728         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11729         into index.
11730         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11731         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11732         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11733         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11734         DECL_RESULT.
11736 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11738         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11739         addsi3_carryin_alt2_<optab>): Correct output template.
11741 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11743         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11744         Adjust for arm_restrict_it.
11745         Remove trailing whitespace.
11747 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11749         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11750         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11752         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11754 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11756         PR rtl-optimization/57960
11757         * lra-constraints.c (process_alt_operands): Use the right mode
11758         when checking strict_low.
11760 2013-07-25  Jan Hubicka  <jh@suse.cz>
11762         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11763         * cgraph.c (cgraph_remove_node): Do not release function body
11764         when in cgraph streaming.
11765         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11766         in other partitions are not considered reachable; fix handling of
11767         clones.
11769 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11771         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11773 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11775         PR target/19599
11776         PR target/57731
11777         PR target/57837
11778         * config/arm/arm.md ("*sibcall_insn): Replace use of
11779         Ss with US. Adjust output for v5 and v4t.
11780         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11782         * config/arm/constraints.md ("Ss"): Rename to US.
11784 2013-07-25  Terry Guo  <terry.guo@arm.com>
11786         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11787         shift_add/shift_sub0/shift_sub1 RTXs.
11789 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11790             Anton Blanchard  <anton@au1.ibm.com>
11792         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11793         (altivec_vpks<VI_char>ss): Likewise.
11794         (altivec_vpks<VI_char>us): Likewise.
11795         (altivec_vpku<VI_char>us): Likewise.
11796         (altivec_vpku<VI_char>um): Likewise.
11798 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11800         Introduce context class.
11802         * Makefile.in (CONTEXT_H): New.
11803         (OBJS): Add context.o.
11804         (toplev.o): Add CONTEXT_H to dependencies.
11805         (context.o): New.
11807         * toplev.c (general_init): Create the singleton gcc::context instance.
11809         * context.c: New.
11811         * context.h: New.
11813 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11815         PR rtl-optimization/57968
11816         * mode-switching.c (create_pre_exit): Allow instructions that
11817         don't set a return register to need a non-exit mode.
11819 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11820             Anton Blanchard  <anton@au1.ibm.com>
11822         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11823         operands to vperm for little endian.
11824         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11825         of lvsl to create the control mask for a vperm for little endian.
11827 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11828             Anton Blanchard  <anton@au1.ibm.com>
11830         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11831         two operands for little-endian.
11833 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11835         * config/mips/mips.c (mips_case_values_threshold): New.
11836         (TARGET_CASE_VALUES_THRESHOLD): Define.
11838 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11839             Anton Blanchard  <anton@au1.ibm.com>
11841         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11842         selection of field for vector splat in little endian mode.
11844 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11846         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11847         expanders to rs6000.md.
11848         (ior<mode>3): Likewise.
11849         (and<mode>3): Likewise.
11850         (one_cmpl<mode>2): Likewise.
11851         (nor<mode>3): Likewise.
11852         (andc<mode>3): Likewise.
11853         (eqv<mode>3): Likewise.
11854         (nand<mode>3): Likewise.
11855         (orc<mode>3): Likewise.
11857         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11858         declaration.
11860         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11861         to split multi-word logical operations.
11862         (rs6000_split_logical_di): Likewise.
11863         (rs6000_split_logical): Likewise.
11865         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11866         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11867         and allow TImode operations in 32-bit.
11868         (vsx_and<mode>3_64bit): Likewise.
11869         (vsx_ior<mode>3_32bit): Likewise.
11870         (vsx_ior<mode>3_64bit): Likewise.
11871         (vsx_xor<mode>3_32bit): Likewise.
11872         (vsx_xor<mode>3_64bit): Likewise.
11873         (vsx_one_cmpl<mode>2_32bit): Likewise.
11874         (vsx_one_cmpl<mode>2_64bit): Likewise.
11875         (vsx_nor<mode>3_32bit): Likewise.
11876         (vsx_nor<mode>3_64bit): Likewise.
11877         (vsx_andc<mode>3_32bit): Likewise.
11878         (vsx_andc<mode>3_64bit): Likewise.
11879         (vsx_eqv<mode>3_32bit): Likewise.
11880         (vsx_eqv<mode>3_64bit): Likewise.
11881         (vsx_nand<mode>3_32bit): Likewise.
11882         (vsx_nand<mode>3_64bit): Likewise.
11883         (vsx_orc<mode>3_32bit): Likewise.
11884         (vsx_orc<mode>3_64bit): Likewise.
11886         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11887         logical types in GPRs.
11889         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11890         logical insns to rs6000.md, and allow TImode operations in
11891         32-bit.
11892         (altivec_ior<mode>3): Likewise.
11893         (altivec_xor<mode>3): Likewise.
11894         (altivec_one_cmpl<mode>2): Likewise.
11895         (altivec_nor<mode>3): Likewise.
11896         (altivec_andc<mode>3): Likewise.
11898         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11899         attributes for moving the 128-bit logical operations into
11900         rs6000.md.
11901         (BOOL_REGS_OUTPUT): Likewise.
11902         (BOOL_REGS_OP1): Likewise.
11903         (BOOL_REGS_OP2): Likewise.
11904         (BOOL_REGS_UNARY): Likewise.
11905         (BOOL_REGS_AND_CR0): Likewise.
11906         (one_cmpl<mode>2): Add support for DI logical operations on
11907         32-bit, splitting the operations to 32-bit.
11908         (anddi3): Likewise.
11909         (iordi3): Likewise.
11910         (xordi3): Likewise.
11911         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11912         changes to combine the 32/64-bit code, allow logical operations on
11913         TI mode in 32-bit, and to use similar match_operator patterns like
11914         scalar mode uses.  Combine the Altivec and VSX code for logical
11915         operations, and move it here.
11916         (ior<mode>3, 128-bit types): Likewise.
11917         (xor<mode>3, 128-bit types): Likewise.
11918         (one_cmpl<mode>3, 128-bit types): Likewise.
11919         (nor<mode>3, 128-bit types): Likewise.
11920         (andc<mode>3, 128-bit types): Likewise.
11921         (eqv<mode>3, 128-bit types): Likewise.
11922         (nand<mode>3, 128-bit types): Likewise.
11923         (orc<mode>3, 128-bit types): Likewise.
11924         (and<mode>3_internal): Likewise.
11925         (bool<mode>3_internal): Likewise.
11926         (boolc<mode>3_internal1): Likewise.
11927         (boolc<mode>3_internal2): Likewise.
11928         (boolcc<mode>3_internal1): Likewise.
11929         (boolcc<mode>3_internal2): Likewise.
11930         (eqv<mode>3_internal1): Likewise.
11931         (eqv<mode>3_internal2): Likewise.
11932         (one_cmpl1<mode>3_internal): Likewise.
11934 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11936         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11937         Rename flag_stack_usage to flag_stack_usage_info.
11939 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11941         * config/microblaze/sync.md: New file.
11942         * config/microblaze/microblaze.md: Include sync.md
11943         * config/microblaze/microblaze.c: Add print_operand 'y'.
11944         * config/microblaze/constraints.md: Add memory_contraint
11945         'Q' which is a single register.
11947 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11949         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11951 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11953         * reload.c (find_reloads): Exit loop once we find this operand
11954         cannot be reloaded somehow for this alternative.
11956         * reload.c (find_reloads): Exit loop once we find a hard register.
11958         * rtlanal.c (computed_jump_p): Exit loop once we find label
11959         reference is used.
11961         * i386.c (ix86_pad_returns): Exit loop after setting replace.
11963         * cfgloopmanip.c (remove_path): Exit loop after setting
11964         irred_invalidated.
11966         * gensupport.c (subst_dup): Avoid loop if code is not
11967         MATCH_DUP nor MATCH_OP_DUP.
11969 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
11971         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
11973 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11975         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
11976         true for SP_REGNUM if mode == ptr_mode.
11977         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
11978         with value R0_REGNUM + 31.
11980 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11982         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
11983         pad pointer-typed argument downward.
11985 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11987         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
11988         and __ILP32__ when the ILP32 model is in use.
11990 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11992         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
11993         (aarch64_load_symref_appropriately): In the case of
11994         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
11995         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
11996         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
11997         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
11998         if 'mode' doesn't equal to 'ptr_mode'.
11999         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12000         'vcall_offset'; change to call aarch64_emit_move differently depending
12001         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12002         to calculate the upper bound of 'vcall_offset'.
12003         (aarch64_cannot_force_const_mem): Change to also return true if
12004         mode != ptr_mode.
12005         (aarch64_legitimize_reload_address): In the case of large
12006         displacements, add new local variable 'xmode' and an assertion
12007         based on it; change to use 'xmode' to generate the new rtx and
12008         reload.
12009         (aarch64_asm_trampoline_template): Change to generate the template
12010         differently depending on TARGET_ILP32 or not; change to use
12011         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12012         (aarch64_trampoline_size): Removed.
12013         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12014         and replace immediate literals with it.  Change to use 'ptr_mode'
12015         instead of 'DImode' and call convert_memory_address if the mode
12016         of 'fnaddr' doesn't equal to 'ptr_mode'.
12017         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12018         to output symbol.
12019         (aarch64_elf_asm_destructor): Likewise.
12020         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12021         on TARGET_ILP32 instead of aarch64_trampoline_size.
12022         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12023         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12024         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12025         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12026         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12027         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12028         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12029         depending on the value of 'mode'.
12030         (add_losym_<mode>): New.
12031         (ldr_got_small_<mode>): New, based on ldr_got_small.
12032         (ldr_got_small): Remove.
12033         (ldr_got_small_sidi): New.
12034         * config/aarch64/iterators.md (P): New.
12035         (PTR): Change to 'ptr_mode' in the condition.
12037 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12039         * config.gcc (aarch64*-*-*): Support --with-abi.
12040         (aarch64*-*-elf): Support --with-multilib-list.
12041         (aarch64*-*-linux*): Likewise.
12042         (supported_defaults): Add abi to aarch64*-*-*.
12043         * configure.ac: Mention AArch64 for --with-multilib-list.
12044         * configure: Re-generated.
12045         * config/aarch64/biarchilp32.h: New file.
12046         * config/aarch64/biarchlp64.h: New file.
12047         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12048         (ABI_SPEC): Ditto.
12049         (MULTILIB_DEFAULTS): Ditto.
12050         (DRIVER_SELF_SPECS): Ditto.
12051         (ASM_SPEC): Update to also substitute -mabi.
12052         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12053         file whose name depends on -mabi= and -mbig-endian.
12054         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12055         TARGET_ILP32.
12056         (POINTER_SIZE): New define.
12057         (POINTERS_EXTEND_UNSIGNED): Ditto.
12058         (enum aarch64_abi_type): New enumeration tag.
12059         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12060         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12061         (TARGET_ILP32): New define.
12062         * config/aarch64/aarch64.opt (mabi): New.
12063         (aarch64_abi): New.
12064         (ilp32, lp64): New values for -mabi.
12065         * config/aarch64/t-aarch64 (comma): New define.
12066         (MULTILIB_OPTIONS): Ditto.
12067         (MULTILIB_DIRNAMES): Ditto.
12068         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12069         * doc/invoke.texi: Document -mabi for AArch64.
12071 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12073         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12075 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12076             Anton Blanchard  <anton@au1.ibm.com>
12078         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12079         endianness when selecting field to splat.
12081 2013-07-22  Eric Christopher  <echristo@gmail.com>
12083         * dwarf2out.c (die_odr_checksum): New function to use
12084         CHECKSUM_ macros and ULEB128 for DIE tag.
12085         (generate_type_signature): Use.
12087 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12089         * config.gcc (sparc*-*-*): Accept leon3 processor.
12090         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12091         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12092         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12093         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12094         (mfix-ut699): Adjust comment.
12095         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12096         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12097         (CPP_CPU_SPEC): Likewise.
12098         (ASM_CPU_SPEC): Likewise.
12099         * config/sparc/sparc.c (leon3_cost): New constant.
12100         (sparc_option_override): Add leon3 support.
12101         (mem_ref): New function.
12102         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12103         (sparc_do_work_around_errata): Look into the instruction in the delay
12104         slot and adjust accordingly.  Add fix for the data cache nullify issues
12105         of the UT699.  Change insertion position for the NOP.
12106         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12107         (leon3_load): New reservation.
12108         (leon_store): Bump latency to 2.
12109         (grfpu): New automaton.
12110         (grfpu_alu): New unit.
12111         (grfpu_ds): Likewise.
12112         (leon_fp_alu): Adjust.
12113         (leon_fp_mult): Delete.
12114         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12115         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12116         * config/sparc/sparc.md (cpu): Add leon3.
12117         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12118         (swapsi): Likewise.
12119         (atomic_test_and_set): Likewise.
12120         (ldstub): Likewise.
12122 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12124         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12125         default for R5900 targets.
12126         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12127         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12128         * config/mips/mips.c (mips_option_override): Report an error for
12129         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12130         for -march=r5900 -mhard-float.
12132 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12134         * df-problems.c (can_move_insns_across): Exit loop once we
12135         find a non-fixed, non-global register.
12137         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12138         setting can_throw.
12140         * omega.c (omega_eliminate_red): Break after setting red_found.
12141         (omega_problem_has_red_equations): Similarly after setting found.
12142         (omega_query_variable): Similarly after setting coupled.
12144 2013-07-22  Marek Polacek  <polacek@redhat.com>
12146         * gimplify.c: Don't include gimple.h twice.
12148 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12150         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12151         instead of TARGET_THUMB1.
12152         (Pz): New constraint.
12153         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12154         encodings.
12155         (compare_negsi_si): Likewise.
12156         (compare_addsi2_op0): Likewise.
12157         (compare_addsi2_op1): Likewise.
12158         (addsi3_carryin_<optab>): Likewise.
12159         (addsi3_carryin_alt2_<optab>): Likewise.
12160         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12161         for arm_restrict_it.
12162         (subsi3_carryin): Likewise.
12163         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12164         (minmax_arithsi): Disable for arm_restrict_it.
12165         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12166         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12167         (satsi_<SAT:code>_shift): Likewise.
12168         (arm_shiftsi3): Add alternative for 16-bit encoding.
12169         (arm32_movhf): Disable for arm_restrict_it.
12170         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12171         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12173 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12175         * config/arm/arm.md (attribute "insn"): Delete.
12176         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12177         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12178         (not_shiftsi): Update for attribute change.
12179         (not_shiftsi_compare0): Likewise.
12180         (not_shiftsi_compare0_scratch): Likewise.
12181         (arm_one_cmplsi2): Likewise.
12182         (thumb1_one_cmplsi2): Likewise.
12183         (notsi_compare0): Likewise.
12184         (notsi_compare0_scratch): Likewise.
12185         (thumb1_movdi_insn): Likewise.
12186         (arm_movsi_insn): Likewise.
12187         (movhi_insn_arch4): Likewise.
12188         (movhi_bytes): Likewise.
12189         (arm_movqi_insn): Likewise.
12190         (thumb1_movqi_insn): Likewise.
12191         (arm32_movhf): Likewise.
12192         (thumb1_movhf): Likewise.
12193         (arm_movsf_soft_insn): Likewise.
12194         (thumb1_movsf_insn): Likewise.
12195         (thumb_movdf_insn): Likewise.
12196         (movsicc_insn): Likewise.
12197         (movsfcc_soft_insn): Likewise.
12198         (and_scc): Likewise.
12199         (cond_move): Likewise.
12200         (if_move_not): Likewise.
12201         (if_not_move): Likewise.
12202         (if_shift_move): Likewise.
12203         (if_move_shift): Likewise.
12204         (if_shift_shift): Likewise.
12205         (if_not_arith): Likewise.
12206         (if_arith_not): Likewise.
12207         (cond_move_not): Likewise.
12208         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12209         (neon_mov<mode>): Likewise.
12210         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12211         (thumb2_movsi_vfp): Likewise.
12212         (movsf_vfp): Likewise.
12213         (thumb2_movsf_vfp): Likewise.
12214         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12215         change.
12216         (cortexa7_older_only): Likewise.
12217         (cortexa7_younger): Likewise.
12218         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12219         (1020alu_shift_op): Likewise.
12220         (1020alu_shift_reg_op): Likewise.
12221         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12222         (alu_shift_op): Likewise.
12223         (alu_shift_reg_op): Likewise.
12224         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12225         (11_alu_shift_op): Likewise.
12226         (11_alu_shift_reg_op): Likewise.
12227         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12228         (9_alu_shift_reg_op): Likewise.
12229         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12230         change.
12231         (cortex_a15_alu_shift): Likewise.
12232         (cortex_a15_alu_shift_reg): Likewise.
12233         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12234         (cortex_a5_alu_shift): Likewise.
12235         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12236         change.
12237         (cortex_a53_alu_shift): Likewise.
12238         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12239         change.
12240         (cortex_a7_alu_reg): Likewise.
12241         (cortex_a7_alu_shift): Likewise.
12242         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12243         (cortex_a8_alu_shift): Likewise.
12244         (cortex_a8_alu_shift_reg): Likewise.
12245         (cortex_a8_mov): Likewise.
12246         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12247         (cortex_a9_dp_shift): Likewise.
12248         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12249         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12250         (cortex_r4_mov): Likewise.
12251         (cortex_r4_alu_shift): Likewise.
12252         (cortex_r4_alu_shift_reg): Likewise.
12253         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12254         (526_alu_shift_op): Likewise.
12255         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12256         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12257         (626te_alu_shift_op): Likewise.
12258         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12259         (726te_alu_op): Likewise.
12260         (726te_alu_shift_op): Likewise.
12261         (726te_alu_shift_reg_op): Likewise.
12262         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12263         (mp626_alu_shift_op): Likewise.
12264         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12265         (pj4_alu_e1_conds): Likewise.
12266         (pj4_alu): Likewise.
12267         (pj4_alu_conds): Likewise.
12268         (pj4_shift): Likewise.
12269         (pj4_shift_conds): Likewise.
12270         (pj4_alu_shift): Likewise.
12271         (pj4_alu_shift_conds): Likewise.
12273 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12275         * config/arm/predicates.md (shiftable_operator_strict_it):
12276         New predicate.
12277         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12278         Disable cond_exec version for arm_restrict_it.
12279         (thumb2_smaxsi3): Convert to generate cond_exec.
12280         (thumb2_sminsi3): Likewise.
12281         (thumb32_umaxsi3): Likewise.
12282         (thumb2_uminsi3): Likewise.
12283         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12284         (thumb2_neg_abssi2): Likewise.
12285         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12286         (thumb2_movsicc_insn): Adjust alternatives.
12287         (thumb2_mov_negscc): Disable for arm_restrict_it.
12288         (thumb2_mov_negscc_strict_it): New pattern.
12289         (thumb2_mov_notscc_strict_it): New pattern.
12290         (thumb2_mov_notscc): Disable for arm_restrict_it.
12291         (thumb2_ior_scc): Likewise.
12292         (thumb2_ior_scc_strict_it): New pattern.
12293         (thumb2_cond_move): Adjust for arm_restrict_it.
12294         (thumb2_cond_arith): Disable for arm_restrict_it.
12295         (thumb2_cond_arith_strict_it): New pattern.
12296         (thumb2_cond_sub): Adjust for arm_restrict_it.
12297         (thumb2_movcond): Likewise.
12298         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12299         (thumb2_zero_extendhisi2_v6): Likewise.
12300         (thumb2_zero_extendqisi2_v6): Likewise.
12301         (orsi_notsi_si): Likewise.
12302         (orsi_not_shiftsi_si): Likewise.
12304 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12306         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12307         instruction sequence is 1 byte shorter.
12309 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12311         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12312         it is not needed after split.
12314 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12316         PR target/51784
12317         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12318         second label for nonlocal goto receivers. Don't output pic base labels
12319         unless we're producing PIC; mark that action unreachable().
12320         (ix86_save_reg): If the function contains a nonlocal label, save the
12321         PIC base reg.
12322         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12323         * config/darwin.c (emitted_pic_label_num): New GTY.
12324         (update_pic_label_number_if_needed): New.
12325         (machopic_output_function_base_name): Adjust for nonlocal receiver
12326         case.
12327         (machopic_should_output_picbase_label): New.
12328         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12329         (nonlocal_goto_receiver): New insn and split.
12331 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12333         * config/aarch64/aarch64-builtins.c
12334         (aarch64_fold_builtin): Fold abs in all modes.
12335         * config/aarch64/aarch64-simd-builtins.def
12336         (abs): Enable for all modes.
12337         * config/aarch64/arm_neon.h
12338         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12339         (vabs_f64): Add missing intrinsic.
12341 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12343         * config/aarch64/arm_neon.h (vabs_s64): New function
12345 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12347         PR target/57516
12348         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12349         * config/avr/avr.md (adjust_len): Add `round'.
12350         * config/avr/avr-protos.h (avr_out_round): New prototype.
12351         (avr_out_plus): Add `out_label' argument.
12352         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12353         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12354         Handle the case where `insn' is just a pattern.
12355         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12356         (avr_out_round): New function.
12357         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12359 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12361         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12362         Add check for flag_stack_usage to handle -fstack-usage support
12364 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12366         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12367         interaction for new Power8 flags and VSX.
12369 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12371         PR middle-end/57698
12372         * tree-inline.c (expand_call_inline): Emit errors during
12373         early_inlining only if optimization is not turned on.
12375 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12377         * passes.def: New.
12379         * passes.c (init_optimization_passes): Move the construction of
12380         the pass hierarchy into a new passes.def file.
12382         * Makefile.in (passes.o): Add dependency on passes.def.
12384 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12386         * passes.c (init_optimization_passes): Introduce macros for
12387         constructing the tree of passes (INSERT_PASSES_AFTER,
12388         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12389         TERMINATE_PASS_LIST).
12391 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12392             Wei Mi  <wmi@google.com>
12394         PR rtl-optimization/57878
12395         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12396         top.
12397         (reload_pseudo_compare_func): Check nregs first for reload
12398         pseudos.
12400 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12402         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12404 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12406         * read-rtl.c (validate_const_int): Once an invalid character is
12407         seen, quit the loop.
12409         * gengtype.c (write_roots): Similarly once we find the "deletable"
12410         or "if_marked" option.
12412 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12414         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12415         "xtab" and "sat".  Move value "clz" from here to ...
12416         (attriubte "type"): ... here.
12417         (satsi_<SAT:code>): Delete "insn" attribute.
12418         (satsi_<SAT:code>_shift): Likewise.
12419         (arm_zero_extendqisi2addsi): Likewise.
12420         (arm_extendqisi2addsi): Likewise.
12421         (clzsi2): Update for attribute changes.
12422         (rbitsi2): Likewise.
12423         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12424         attribute.
12425         (arm_usatsihi): Likewise.
12426         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12428 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12430         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12431         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12432         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12433         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12434         in alphabetical order.
12435         (attribute "core_cycles"): Update for attribute changes.
12436         (arm_addsi3): Likewise.
12437         (addsi3_compare0): Likewise.
12438         (addsi3_compare0_scratch): Likewise.
12439         (addsi3_compare_op1): Likewise.
12440         (addsi3_compare_op2): Likewise.
12441         (compare_addsi2_op0): Likewise.
12442         (compare_addsi2_op1): Likewise.
12443         (addsi3_carryin_shift_<optab>): Likewise.
12444         (subsi3_carryin_shift): Likewise.
12445         (rsbsi3_carryin_shift): Likewise.
12446         (arm_subsi3_insn): Likewise.
12447         (subsi3_compare0): Likewise.
12448         (subsi3_compare): Likewise.
12449         (arm_andsi3_insn): Likewise.
12450         (thumb1_andsi3_insn): Likewise.
12451         (andsi3_compare0): Likewise.
12452         (andsi3_compare0_scratch): Likewise.
12453         (zeroextractsi_compare0_scratch
12454         (andsi_not_shiftsi_si): Likewise.
12455         (iorsi3_insn): Likewise.
12456         (iorsi3_compare0): Likewise.
12457         (iorsi3_compare0_scratch): Likewise.
12458         (arm_xorsi3): Likewise.
12459         (thumb1_xorsi3_insn): Likewise.
12460         (xorsi3_compare0): Likewise.
12461         (xorsi3_compare0_scratch): Likewise.
12462         (satsi_<SAT:code>_shift): Likewise.
12463         (rrx): Likewise.
12464         (arm_shiftsi3): Likewise.
12465         (shiftsi3_compare0): Likewise.
12466         (not_shiftsi): Likewise.
12467         (not_shiftsi_compare0): Likewise.
12468         (not_shiftsi_compare0_scratch): Likewise.
12469         (arm_one_cmplsi2): Likewise.
12470         (thumb_one_complsi2): Likewise.
12471         (notsi_compare0): Likewise.
12472         (notsi_compare0_scratch): Likewise.
12473         (thumb1_zero_extendhisi2): Likewise.
12474         (arm_zero_extendhisi2): Likewise.
12475         (arm_zero_extendhisi2_v6): Likewise.
12476         (arm_zero_extendhisi2addsi): Likewise.
12477         (thumb1_zero_extendqisi2): Likewise.
12478         (thumb1_zero_extendqisi2_v6): Likewise.
12479         (arm_zero_extendqisi2): Likewise.
12480         (arm_zero_extendqisi2_v6): Likewise.
12481         (arm_zero_extendqisi2addsi): Likewise.
12482         (thumb1_extendhisi2): Likewise.
12483         (arm_extendhisi2): Likewise.
12484         (arm_extendhisi2_v6): Likewise.
12485         (arm_extendqisi): Likewise.
12486         (arm_extendqisi_v6): Likewise.
12487         (arm_extendqisi2addsi): Likewise.
12488         (thumb1_extendqisi2): Likewise.
12489         (thumb1_movdi_insn): Likewise.
12490         (arm_movsi_insn): Likewise.
12491         (movsi_compare0): Likewise.
12492         (movhi_insn_arch4): Likewise.
12493         (movhi_bytes): Likewise.
12494         (arm_movqi_insn): Likewise.
12495         (thumb1_movqi_insn): Likewise.
12496         (arm32_movhf): Likewise.
12497         (thumb1_movhf): Likewise.
12498         (arm_movsf_soft_insn): Likewise.
12499         (thumb1_movsf_insn): Likewise.
12500         (movdf_soft_insn): Likewise.
12501         (thumb_movdf_insn): Likewise.
12502         (arm_cmpsi_insn): Likewise.
12503         (cmpsi_shiftsi): Likewise.
12504         (cmpsi_shiftsi_swp): Likewise.
12505         (arm_cmpsi_negshiftsi_si): Likewise.
12506         (movsicc_insn): Likewise.
12507         (movsfcc_soft_insn): Likewise.
12508         (arith_shiftsi): Likewise.
12509         (arith_shiftsi_compare0
12510         (arith_shiftsi_compare0_scratch
12511         (sub_shiftsi): Likewise.
12512         (sub_shiftsi_compare0
12513         (sub_shiftsi_compare0_scratch
12514         (and_scc): Likewise.
12515         (cond_move): Likewise.
12516         (if_plus_move): Likewise.
12517         (if_move_plus): Likewise.
12518         (if_move_not): Likewise.
12519         (if_not_move): Likewise.
12520         (if_shift_move): Likewise.
12521         (if_move_shift): Likewise.
12522         (if_shift_shift): Likewise.
12523         (if_not_arith): Likewise.
12524         (if_arith_not): Likewise.
12525         (cond_move_not): Likewise.
12526         (thumb1_ashlsi3): Set type attribute.
12527         (thumb1_ashrsi3): Likewise.
12528         (thumb1_lshrsi3): Likewise.
12529         (thumb1_rotrsi3): Likewise.
12530         (shiftsi3_compare0_scratch): Likewise.
12531         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12532         (neon_mov<mode>): Likewise.
12533         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12534         attribute changes.
12535         (thumb2_movsi_insn): Likewise.
12536         (thumb2_cmpsi_neg_shiftsi): Likewise.
12537         (thumb2_extendqisi_v6): Likewise.
12538         (thumb2_zero_extendhisi2_v6): Likewise.
12539         (thumb2_zero_extendqisi2_v6): Likewise.
12540         (thumb2_shiftsi3_short): Likewise.
12541         (thumb2_addsi3_compare0_scratch): Likewise.
12542         (orsi_not_shiftsi_si): Likewise.
12543         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12544         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12545         changes.
12546         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12547         (1020alu_shift_op): Likewise.
12548         (1020alu_shift_reg_op): Likewise.
12549         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12550         (alu_shift_op): Likewise.
12551         (alu_shift_reg_op): Likewise.
12552         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12553         (11_alu_shift_op): Likewise.
12554         (11_alu_shift_reg_op): Likewise.
12555         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12556         (9_alu_shift_reg_op): Likewise.
12557         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12558         changes.
12559         (cortex_a15_alu_shift): Likewise.
12560         (cortex_a15_alu_shift_reg): Likewise.
12561         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12562         changes.
12563         (cortex_a5_alu_shift): Likewise.
12564         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12565         changes.
12566         (cortex_a53_alu_shift): Likewise.
12567         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12568         changes.
12569         (cortex_a7_alu_reg): Likewise.
12570         (cortex_a7_alu_shift): Likewise.
12571         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12572         changes.
12573         (cortex_a8_alu_shift): Likewise.
12574         (cortex_a8_alu_shift_reg): Likewise.
12575         (cortex_a8_mov): Likewise.
12576         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12577         (cortex_a9_dp_shift): Likewise.
12578         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12579         changes.
12580         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12581         changes.
12582         (cortex_r4_mov): Likewise.
12583         (cortex_r4_alu_shift): Likewise.
12584         (cortex_r4_alu_shift_reg): Likewise.
12585         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12586         (526_alu_shift_op): Likewise.
12587         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12588         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12589         (626te_alu_shift_op): Likewise.
12590         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12591         (726te_alu_op): Likewise.
12592         (726te_alu_shift_op): Likewise.
12593         (726te_alu_shift_reg_op): Likewise.
12594         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12595         (mp626_alu_shift_op): Likewise.
12596         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12597         (pj4_alu_e1_conds): Likewise.
12598         (pj4_alu): Likewise.
12599         (pj4_alu_conds): Likewise.
12600         (pj4_shift): Likewise.
12601         (pj4_shift_conds): Likewise.
12602         (pj4_alu_shift): Likewise.
12603         (pj4_alu_shift_conds): Likewise.
12604         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12605         changes.
12606         (cortexa7_older_only): Likewise.
12607         (cortexa7_younger): Likewise.
12609 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12611         * ipa-pure-const.c (generate_summary): Rename to...
12612         (pure_const_generate_summary): ... this.
12614 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12616         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12618 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12620         PR target/57909
12621         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12622         usage in HI mode.
12624 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12626         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12627         enabled without -march=zEC12.
12628         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12629         flags to be set.
12631 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12633         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12634         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12635         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12636         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12637         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12638         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12639         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12640         ISA_HAS_FP4.
12641         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12642         and ISA_HAS_NMADD3_NMSUB3.
12643         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12644         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12645         (nmsub4<mode>, nmsub3<mode>): Likewise.
12646         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12648 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12650         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12651         TARGET_MIPS5400 checking.
12653 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12654             Peter Bergner  <bergner@vnet.ibm.com>
12656         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12657         registers in the comment.
12658         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12659         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12660         rather than FIRST_PSEUDO_REGISTERS.
12662 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12664         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12665         enable extra ISA flags with TARGET_HTM.
12667 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12669         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12670         Fix comment typos.
12672 2013-07-15  Cong Hou  <congh@google.com>
12674         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12675         in compare function for sorting.
12677 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12679         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12680         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12681         * config/rs6000/rs6000.opt: Add -mhtm option.
12682         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12683         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12684         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12685         __HTM__ if the HTM instructions are available.
12686         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12687         htm_spr_reg_operand): New define_predicates.
12688         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12689         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12690         Include htm.md.
12691         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12692         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12693         HTM builtin functions.
12694         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12695         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12696         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12697         (rs6000_builtin_mask_calculate): Likewise.
12698         (rs6000_option_override_internal): Likewise.
12699         (bdesc_htm): Add new HTM builtin support.
12700         (htm_spr_num): New function.
12701         (htm_spr_regno): Likewise.
12702         (rs6000_htm_spr_icode): Likewise.
12703         (htm_expand_builtin): Likewise.
12704         (htm_init_builtins): Likewise.
12705         (rs6000_expand_builtin): Add support for HTM builtin functions.
12706         (rs6000_init_builtins): Likewise.
12707         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12708         option.
12709         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12710         (TARGET_HTM, MASK_HTM): Define macros.
12711         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12712         (FIXED_REGISTERS): Likewise.
12713         (CALL_USED_REGISTERS): Likewise.
12714         (CALL_REALLY_USED_REGISTERS): Likewise.
12715         (REG_ALLOC_ORDER): Likewise.
12716         (enum reg_class): Likewise.
12717         (REG_CLASS_NAMES): Likewise.
12718         (REG_CLASS_CONTENTS): Likewise.
12719         (REGISTER_NAMES): Likewise.
12720         (ADDITIONAL_REGISTER_NAMES): Likewise.
12721         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12722         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12723         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12724         * config/rs6000/htm.md: New file.
12725         * config/rs6000/htmintrin.h: New file.
12726         * config/rs6000/htmxlintrin.h: New file.
12728 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12730         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12731         Define SYMBOL_TINY_GOT, update comment.
12732         * config/aarch64/aarch64.c
12733         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12734         (aarch64_expand_mov_immediate): Likewise.
12735         (aarch64_print_operand): Likewise.
12736         (aarch64_classify_symbol): Likewise.
12737         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12738         (ldr_got_tiny): Define.
12740 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12742         PR tree-optimization/54094
12743         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12744           scheduling dimension for the parallelism check from the polyhedral
12745           information in the AST.
12746         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12747           in 2D + 1 form.
12749 2013-07-13  Jason Merrill  <jason@redhat.com>
12751         * print-tree.c (debug_vec_tree): Use debug_raw.
12752         (debug_raw (vec<tree, va_gc> &)): New.
12753         (debug_raw (vec<tree, va_gc> *)): New.
12754         * tree.h: Declare them.
12756 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12758         * ifcvt.c (ifcvt_after_combine): New static variable.
12759         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12760         for size.
12761         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12762         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12763         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12765 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12767         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12769 2013-07-12  Michael Matz  <matz@suse.de>
12771         PR middle-end/55771
12772         * convert.c (convert_to_real): Reject non-float inner types.
12774 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12776         * config/aarch64/aarch64-protos.h
12777         (aarch64_simd_immediate_valid_for_move): Remove.
12778         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12779         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12780         cases.
12781         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12783 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12785         * config/mips/mips.c (mips_conditional_register_usage): Do not
12786         use t[0-7] registers in MIPS16 mode when optimizing for size.
12788 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12790         * config/i386/i386.c (dispatch_function_versions): Fix array
12791         indexing of function_version_info to match actual_versions.
12793 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12795         * vec.h (struct va_gc): Move release out-of-line.
12796         (va_gc::release): Call ggc_free on released vec.
12798 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12800         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12801         Require GOT register as additional operand in UNSPEC.
12802         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12803         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12804         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12805         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12806         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12807         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12808         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12810 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12812         PR target/57631
12813         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12814         name seen by assembler/linker rather if available.
12816 2013-07-11  Andreas Schwab  <schwab@suse.de>
12818         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12820 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12822         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12824 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12826         * doc/tm.texi.in: Move hook documentation to ....
12827         * target.def: ... here.
12829         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12830         text on @hook line.
12831         * doc/tm.texi: Regenerate.
12833 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12835         PR c++/57869
12836         * doc/invoke.texi: Document -Wconditionally-supported.
12838 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12840         PR target/57844
12841         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12842         of my_fp.
12844 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12846         PR target/57506
12847         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12848         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12849         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12850         Remove duplicate devices.
12851         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12852         * config/avr/t-multilib: Regenerate.
12853         * config/avr/avr-tables.opt: Regenerate.
12854         * doc/avr-mmcu.texi: Regenerate.
12856 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12858         PR target/56987
12859         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12861 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12863         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12864         the cost of MULT when optimizing for size.
12866 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12868         * config/cr16/cr16-protos.h: Don't include target.h.
12870 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12872         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12873         adjust register size for TDmode and TFmode for VSX registers.
12875 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12877         PR target/56892
12878         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12879         hook_bool_const_tree_true.
12881 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12883         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12884         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12885         * config/s390/s390.md: Define FPR*_REGNUM constants.
12886         Fix FPR2_REGNUM constant (18 -> 17).
12887         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12888         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12889         ("trunc<BFP:mode><DFP_ALL:mode>2")
12890         ("trunc<DFP_ALL:mode><BFP:mode>2")
12891         ("*extend<BFP:mode><DFP_ALL:mode>2")
12892         ("*extend<DFP_ALL:mode><BFP:mode>2")
12893         ("extend<BFP:mode><DFP_ALL:mode>2")
12894         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12895         FPR4_REGNUM.
12897 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12899         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12901 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12903         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12904         and cfun_fpr_bit_p to cfun_fpr_save_p.
12905         (s390_frame_area, s390_register_info, s390_frame_info)
12906         (s390_emit_prologue, s390_emit_epilogue)
12907         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12908         register numbers.
12909         * config/s390/s390.h: Define *_REGNUM macros for floating point
12910         register numbers.
12912 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12914         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12916 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12918         PR rtl-optimization/57786
12919         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12920         and break out of the loop when it is set to false.
12922 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12924         PR target/57819
12925         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12926         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12927         (const_int 63)) 0)).
12928         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12929         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12930         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12932         PR rtl-optimization/57829
12933         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12934         mask bits outside of mode are just sign-extension from mode to HWI.
12936 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12938         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12939         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12940         adjust_address instead of change_address to keep info about alignment.
12941         (emit_strmov): Remove.
12942         (emit_memmov): New function.
12943         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12944         (expand_movmem_epilogue): Likewise and return updated rtx for
12945         destination.
12946         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12947         destination and source.
12948         (decide_alignment): Refactor, handle vector_loop.
12949         (ix86_expand_movmem): Likewise.
12950         (ix86_expand_setmem): Likewise.
12951         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12953 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
12955         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
12956         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
12958 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12960         * config/i386/sse.md (sse_movlhps): Change alternative 3
12961         of operand 2 to "m".
12963 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12965         PR target/57807
12966         * config/i386/sse.md (iptr): New mode attribute.
12967         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
12968         (<sse>_vm<plusminus_insn><mode>3): Ditto.
12969         (<sse>_vmmul<mode>3): Ditto.
12970         (<sse>_vmdiv<mode>3): Ditto.
12971         (sse_vmrcpv4sf2): Ditto.
12972         (<sse>_vmsqrt<mode>2): Ditto.
12973         (sse_vmrsqrtv4sf2): Ditto.
12974         (<sse>_vm<code><mode>3): Ditto.
12975         (avx_vmcmp<mode>3): Ditto.
12976         (<sse>_vmmaskcmp<mode>3): Ditto.
12977         (<sse>_comi): Ditto.
12978         (<sse>_ucomi): Ditto.
12979         (*xop_vmfrcz_<mode>): Ditto.
12980         (*fmai_fmadd_<mode>): Ditto.
12981         (*fmai_fmsub_<mode>): Ditto.
12982         (*fmai_fnmadd_<mode>): Ditto.
12983         (*fmai_fnmsub_<mode>): Ditto.
12984         (*fma4i_vmfmadd_<mode>): Ditto.
12985         (*fma4i_vmfmsub_<mode>): Ditto.
12986         (*fma4i_vmfnmadd_<mode>): Ditto.
12987         (*fma4i_vmfnmsub_<mode>): Ditto.
12988         (*xop_vmfrcz_<mode>): Ditto.
12989         (sse_cvtps2pi): Ditto.
12990         (sse_cvttps2pi): Ditto.
12991         (sse_cvtss2si): Ditto.
12992         (sse_cvtss2si_2): Ditto.
12993         (sse_cvtss2siq_2): Ditto.
12994         (sse_cvttss2si): Ditto.
12995         (sse_cvttss2siq): Ditto.
12996         (sse_cvtsd2si): Ditto.
12997         (sse_cvtsd2si_2): Ditto.
12998         (sse_cvtsd2siq_2): Ditto.
12999         (sse_cvttsd2si): Ditto.
13000         (sse_cvttsd2siq): Ditto.
13001         (sse_cvtsd2ss): Ditto.
13002         (sse_cvtss2sd): Ditto.
13003         (avx2_pbroadcast<mode>): Ditto.
13004         (avx2_pbroadcast<mode>_1): Ditto.
13005         (*avx_vperm_broadcast_v4sf): Ditto.
13007         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13008         (sse_movlhps): Ditto.
13009         (sse_storehps): Ditto.
13010         (sse_loadhps): Ditto.
13011         (sse_storelps): Ditto.
13012         (sse_loadlps): Ditto.
13013         (*vec_concatv4sf): Ditto.
13014         (*vec_interleave_highv2df): Ditto.
13015         (*vec_interleave_lowv2df): Ditto.
13016         (*vec_extractv2df_1_sse): Ditto.
13017         (*vec_extractv2df_0_sse): Ditto.
13018         (sse2_storelpd): Ditto.
13019         (sse2_loadlpd): Ditto.
13020         (sse2_movsd): Ditto.
13021         (*vec_concatv4si): Ditto.
13022         (vec_concatv2di): Ditto.
13024         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13025         for Intel asm dialect.
13026         (mmx_punpcklwd): Ditto.
13027         (mmx_punpckldq): Ditto.
13029         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13030         for intel assembler dialect.
13032 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13034         PR target/29776
13035         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13036         for BUILT_IN_C{LZ,LRSB}*.
13037         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13038         * tree-vrp.c (extract_range_basic): Handle
13039         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13040         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13041         fall thru to code calling set_value*.
13042         * builtins.c (expand_builtin): Remove *IMAX cases.
13043         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13044         if width is bigger than 2*HWI.
13046 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13048         PR rtl-optimization/55342
13049         * lra-int.h (lra_subreg_reload_pseudos): New.
13050         * lra.c: Add undoing optional reloads to the block diagram.
13051         (lra_subreg_reload_pseudos): New.
13052         (lra_optional_reload_pseudos): Change comments.
13053         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13054         lra_optional_reload_pseudos after undo transformations.
13055         * lra-assigns.c (pseudo_prefix_title): New.
13056         (lra_setup_reg_renumber): Use it.
13057         (spill_for): Ditto.  Check subreg reload pseudos too.
13058         (assign_by_spills): Consider subreg reload pseudos too.
13059         * lra-constraints.c (simplify_operand_subreg): Use
13060         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13061         (curr_insn_transform): Recognize and do optional reloads.
13062         (undo_optional_reloads): New.
13063         (lra_undo_inheritance): Call undo_optional_reloads.
13065 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13067         * tree-complex.c (expand_complex_operations_1): Fix typo.
13069 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13071         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13072         (tune_params): New member 'const vec_costs'.
13073         * config/aarch64/aarch64.c (generic_vector_cost): New.
13074         (generic_tunings): New member 'generic_vector_cost'.
13075         (aarch64_builtin_vectorization_cost): New.
13076         (aarch64_add_stmt_cost): New.
13077         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13078         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13080 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13082         PR target/57777
13083         * config/i386/predicates.md (vsib_address_operand): Disallow
13084         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13086 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13088         PR middle-end/55030
13089         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13090         expand_builtin_setjmp_receiver.
13091         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13092         with NULL for the label parameter.
13093         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13094         the frame-pointer.  Adjust comments.
13095         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13096         only if LABEL is non-NULL.
13098 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13100         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13101         (ARM_ABI_AAPCS64): Ditto.
13102         (arm_abi): Ditto.
13103         (ARM_DEFAULT_ABI): Ditto.
13105 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13107         * config/aarch64/aarch64-builtins.c
13108         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13109         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13110         (st1): Likewise.
13111         * config/aarch64/aarch64-simd.md
13112         (aarch64_ld1<VALL:mode>): New.
13113         (aarch64_st1<VALL:mode>): Likewise.
13114         * config/aarch64/arm_neon.h
13115         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13117 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13119         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13120         target ISA is AVX.
13121         (ix86_option_override_internal):Turn on all -mavx target flags by
13122         default as they are dependent on AVX anyway.
13124 2013-07-02  Cary Coutant  <ccoutant@google.com>
13126         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13127         deterministic hash.
13128         (loc_checksum_ordered): Likewise.
13129         (hash_loc_operands): Remove inline keyword.
13131 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13133         PR tree-optimization/57741
13134         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13135         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13136         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13137         Allow REAL_CST step_exprs if flag_associative_math.
13138         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13140 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13142         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13144 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13146         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13148 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13150         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13151         encoding.
13152         (iorsi3_insn): Likewise.
13153         (arm_xorsi3): Likewise.
13155 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13157         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13158         to ...
13159         (attribute "type"): ... here, and prefix with "wmmx_".
13160         (attribute "core_cycles"): Update for attribute changes.
13161         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13162         (tbcstv4hi): Likewise.
13163         (tbcstv2si): Likewise.
13164         (iwmmxt_iordi3): Likewise.
13165         (iwmmxt_xordi3): Likewise.
13166         (iwmmxt_anddi3): Likewise.
13167         (iwmmxt_nanddi3): Likewise.
13168         (iwmmxt_arm_movdi): Likewise.
13169         (iwmmxt_movsi_insn): Likewise.
13170         (mov<mode>_internal): Likewise.
13171         (and<mode>3_iwmmxt): Likewise.
13172         (ior<mode>3_iwmmxt): Likewise.
13173         (xor<mode>3_iwmmxt): Likewise.
13174         (add<mode>3_iwmmxt): Likewise.
13175         (ssaddv8qi3): Likewise.
13176         (ssaddv4hi3): Likewise.
13177         (ssaddv2si3): Likewise.
13178         (usaddv8qi3): Likewise.
13179         (usaddv4hi3): Likewise.
13180         (usaddv2si3): Likewise.
13181         (sub<mode>3_iwmmxt): Likewise.
13182         (sssubv8qi3): Likewise.
13183         (sssubv4hi3): Likewise.
13184         (sssubv2si3): Likewise.
13185         (ussubv8qi3): Likewise.
13186         (ussubv4hi3): Likewise.
13187         (ussubv2si3): Likewise.
13188         (mulv4hi3_iwmmxt): Likewise.
13189         (smulv4hi3_highpart): Likewise.
13190         (umulv4hi3_highpart): Likewise.
13191         (iwmmxt_wmacs): Likewise.
13192         (iwmmxt_wmacsz): Likewise.
13193         (iwmmxt_wmacu): Likewise.
13194         (iwmmxt_wmacuz): Likewise.
13195         (iwmmxt_clrdi): Likewise.
13196         (iwmmxt_clrv8qi): Likewise.
13197         (iwmmxt_clr4hi): Likewise.
13198         (iwmmxt_clr2si): Likewise.
13199         (iwmmxt_uavgrndv8qi3): Likewise.
13200         (iwmmxt_uavgrndv4hi3): Likewise.
13201         (iwmmxt_uavgv8qi3): Likewise.
13202         (iwmmxt_uavgv4hi3): Likewise.
13203         (iwmmxt_tinsrb): Likewise.
13204         (iwmmxt_tinsrh): Likewise.
13205         (iwmmxt_tinsrw): Likewise.
13206         (iwmmxt_textrmub): Likewise.
13207         (iwmmxt_textrmsb): Likewise.
13208         (iwmmxt_textrmuh): Likewise.
13209         (iwmmxt_textrmsh): Likewise.
13210         (iwmmxt_textrmw): Likewise.
13211         (iwmxxt_wshufh): Likewise.
13212         (eqv8qi3): Likewise.
13213         (eqv4hi3): Likewise.
13214         (eqv2si3): Likewise.
13215         (gtuv8qi3): Likewise.
13216         (gtuv4hi3): Likewise.
13217         (gtuv2si3): Likewise.
13218         (gtv8qi3): Likewise.
13219         (gtv4hi3): Likewise.
13220         (gtv2si3): Likewise.
13221         (smax<mode>3_iwmmxt): Likewise.
13222         (umax<mode>3_iwmmxt): Likewise.
13223         (smin<mode>3_iwmmxt): Likewise.
13224         (umin<mode>3_iwmmxt): Likewise.
13225         (iwmmxt_wpackhss): Likewise.
13226         (iwmmxt_wpackwss): Likewise.
13227         (iwmmxt_wpackdss): Likewise.
13228         (iwmmxt_wpackhus): Likewise.
13229         (iwmmxt_wpackwus): Likewise.
13230         (iwmmxt_wpackdus): Likewise.
13231         (iwmmxt_wunpckihb): Likewise.
13232         (iwmmxt_wunpckihh): Likewise.
13233         (iwmmxt_wunpckihw): Likewise.
13234         (iwmmxt_wunpckilb): Likewise.
13235         (iwmmxt_wunpckilh): Likewise.
13236         (iwmmxt_wunpckilw): Likewise.
13237         (iwmmxt_wunpckehub): Likewise.
13238         (iwmmxt_wunpckehuh): Likewise.
13239         (iwmmxt_wunpckehuw): Likewise.
13240         (iwmmxt_wunpckehsb): Likewise.
13241         (iwmmxt_wunpckehsh): Likewise.
13242         (iwmmxt_wunpckehsw): Likewise.
13243         (iwmmxt_wunpckelub): Likewise.
13244         (iwmmxt_wunpckeluh): Likewise.
13245         (iwmmxt_wunpckeluw): Likewise.
13246         (iwmmxt_wunpckelsb): Likewise.
13247         (iwmmxt_wunpckelsh): Likewise.
13248         (iwmmxt_wunpckelsw): Likewise.
13249         (ror<mode>3): Likewise.
13250         (ashr<mode>3_iwmmxt): Likewise.
13251         (lshr<mode>3_iwmmxt): Likewise.
13252         (ashl<mode>3_iwmmxt): Likewise.
13253         (ror<mode>3_di): Likewise.
13254         (ashr<mode>3_di): Likewise.
13255         (lshr<mode>3_di): Likewise.
13256         (ashl<mode>3_di): Likewise.
13257         (iwmmxt_wmadds): Likewise.
13258         (iwmmxt_wmaddu): Likewise.
13259         (iwmmxt_tmia): Likewise.
13260         (iwmmxt_tmiaph): Likewise.
13261         (iwmmxt_tmiabb): Likewise.
13262         (iwmmxt_tmiatb): Likewise.
13263         (iwmmxt_tmiabt): Likewise.
13264         (iwmmxt_tmiatt): Likewise.
13265         (iwmmxt_tmovmskb): Likewise.
13266         (iwmmxt_tmovmskh): Likewise.
13267         (iwmmxt_tmovmskw): Likewise.
13268         (iwmmxt_waccb): Likewise.
13269         (iwmmxt_wacch): Likewise.
13270         (iwmmxt_waccw): Likewise.
13271         (iwmmxt_waligni): Likewise.
13272         (iwmmxt_walignr): Likewise.
13273         (iwmmxt_walignr0): Likewise.
13274         (iwmmxt_walignr1): Likewise.
13275         (iwmmxt_walignr2): Likewise.
13276         (iwmmxt_walignr3): Likewise.
13277         (iwmmxt_wsadb): Likewise.
13278         (iwmmxt_wsadh): Likewise.
13279         (iwmmxt_wsadbz): Likewise.
13280         (iwmmxt_wsadhz): Likewise.
13281         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13282         (iwmmxt_wabsdiffb): Likewise.
13283         (iwmmxt_wabsdiffh): Likewise.
13284         (iwmmxt_wabsdiffw): Likewise.
13285         (iwmmxt_waddsubhx): Likewise
13286         (iwmmxt_wsubaddhx): Likewise.
13287         (addc<mode>3): Likewise.
13288         (iwmmxt_avg4): Likewise.
13289         (iwmmxt_avg4r): Likewise.
13290         (iwmmxt_wmaddsx): Likewise.
13291         (iwmmxt_wmaddux): Likewise.
13292         (iwmmxt_wmaddsn): Likewise.
13293         (iwmmxt_wmaddun): Likewise.
13294         (iwmmxt_wmulwsm): Likewise.
13295         (iwmmxt_wmulwum): Likewise.
13296         (iwmmxt_wmulsmr): Likewise.
13297         (iwmmxt_wmulumr): Likewise.
13298         (iwmmxt_wmulwsmr): Likewise.
13299         (iwmmxt_wmulwumr): Likewise.
13300         (iwmmxt_wmulwl): Likewise.
13301         (iwmmxt_wqmulm): Likewise.
13302         (iwmmxt_wqmulwm): Likewise.
13303         (iwmmxt_wqmulmr): Likewise.
13304         (iwmmxt_wqmulwmr): Likewise.
13305         (iwmmxt_waddbhusm): Likewise.
13306         (iwmmxt_waddbhusl): Likewise.
13307         (iwmmxt_wqmiabb): Likewise.
13308         (iwmmxt_wqmiabt): Likewise.
13309         (iwmmxt_wqmiatb): Likewise.
13310         (iwmmxt_wqmiatt): Likewise.
13311         (iwmmxt_wqmiabbn): Likewise.
13312         (iwmmxt_wqmiabtn): Likewise.
13313         (iwmmxt_wqmiatbn): Likewise.
13314         (iwmmxt_wqmiattn): Likewise.
13315         (iwmmxt_wmiabb): Likewise.
13316         (iwmmxt_wmiabt): Likewise.
13317         (iwmmxt_wmiatb): Likewise.
13318         (iwmmxt_wmiatt): Likewise.
13319         (iwmmxt_wmiabbn): Likewise.
13320         (iwmmxt_wmiabtn): Likewise.
13321         (iwmmxt_wmiatbn): Likewise.
13322         (iwmmxt_wmiattn): Likewise.
13323         (iwmmxt_wmiawbb): Likewise.
13324         (iwmmxt_wmiawbt): Likewise.
13325         (iwmmxt_wmiawtb): Likewise.
13326         (iwmmxt_wmiawtt): Likewise.
13327         (iwmmxt_wmiawbbn): Likewise.
13328         (iwmmxt_wmiawbtn): Likewise.
13329         (iwmmxt_wmiawtbn): Likewise.
13330         (iwmmxt_wmiawttn): Likewise.
13331         (iwmmxt_wmerge): Likewise.
13332         (iwmmxt_tandc<mode>3): Likewise.
13333         (iwmmxt_torc<mode>3): Likewise.
13334         (iwmmxt_torvsc<mode>3): Likewise.
13335         (iwmmxt_textrc<mode>3): Likewise.
13336         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13337         (wmmxt_pack): Likewise.
13338         (wmmxt_mult_c1): Likewise.
13339         (wmmxt_mult_c2): Likewise.
13340         (wmmxt_alu_c1): Likewise.
13341         (wmmxt_alu_c2): Likewise.
13342         (wmmxt_alu_c3): Likewise.
13343         (wmmxt_transfer_c1): Likewise.
13344         (wmmxt_transfer_c2): Likewise.
13345         (wmmxt_transfer_c3): Likewise.
13346         (marvell_f_iwmmxt_wstr): Likewise.
13347         (marvell_f_iwmmxt_wldr): Likewise.
13349 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13351         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13353 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13355         Revert:
13356         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13357         * lra-constraints.c (need_for_split_p): Check call used hard regs
13358         living through calls.
13360         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13361         call used regs for call insn.
13363 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13365         PR target/57736
13366         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13367         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13369 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13371         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13373 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13375         * lra-constraints.c (need_for_split_p): Check call used hard regs
13376         living through calls.
13378 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13380         PR target/57744
13381         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13382         to tie with any other modes.  Eliminate Altivec vector mode tests,
13383         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13384         code, to return 0 if testing MODE2 for a condition, if we've
13385         already tested MODE1 for the same condition.
13387 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13389         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13390         layout.
13392 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13394         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13395         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13397 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13399         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13400         Define.
13401         (aarch64_symbolic_constant_p): Remove.
13402         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13403         static.  Fix line length and white space.
13404         (aarch64_symbolic_constant_p): Remove.
13405         * config/aarch64/predicates.md (aarch64_valid_symref):
13406         Use aarch64_classify_symbol_expression.
13408 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13410         * config/arm/constraints.md (Ts): New constraint.
13411         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13412         16-bit encodings.
13413         (compare_scc): Use "Ts" constraint for operand 0.
13414         (ior_scc_scc): Likewise.
13415         (and_scc_scc): Likewise.
13416         (and_scc_scc_nodom): Likewise.
13417         (ior_scc_scc_cmp): Likewise for operand 7.
13418         (and_scc_scc_cmp): Likewise.
13419         * config/arm/thumb2.md (thumb2_movsi_insn):
13420         Add alternatives for 16-bit encodings.
13421         (thumb2_movhi_insn): Likewise.
13422         (thumb2_movsicc_insn): Likewise.
13423         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13424         (thumb2_negscc): Use "Ts" constraint.
13425         Move mvn instruction outside cond_exec block.
13426         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13427         for 16-bit encodings.
13429 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13431         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13432         encoding.
13433         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13434         (mulsi3subsi): Likewise.
13435         (mulsidi3adddi): Likewise.
13436         (mulsidi3_v6): Likewise.
13437         (umulsidi3_v6): Likewise.
13438         (umulsidi3adddi_v6): Likewise.
13439         (smulsi3_highpart_v6): Likewise.
13440         (umulsi3_highpart_v6): Likewise.
13441         (mulhisi3tb): Likewise.
13442         (mulhisi3bt): Likewise.
13443         (mulhisi3tt): Likewise.
13444         (maddhisi4): Likewise.
13445         (maddhisi4tb): Likewise.
13446         (maddhisi4tt): Likewise.
13447         (maddhidi4): Likewise.
13448         (maddhidi4tb): Likewise.
13449         (maddhidi4tt): Likewise.
13450         (zeroextractsi_compare0_scratch): Likewise.
13451         (insv_zero): Likewise.
13452         (insv_t2): Likewise.
13453         (anddi_notzesidi_di): Likewise.
13454         (anddi_notsesidi_di): Likewise.
13455         (andsi_notsi_si): Likewise.
13456         (iordi_zesidi_di): Likewise.
13457         (xordi_zesidi_di): Likewise.
13458         (andsi_iorsi3_notsi): Likewise.
13459         (smax_0): Likewise.
13460         (smax_m1): Likewise.
13461         (smin_0): Likewise.
13462         (not_shiftsi): Likewise.
13463         (unaligned_loadsi): Likewise.
13464         (unaligned_loadhis): Likewise.
13465         (unaligned_loadhiu): Likewise.
13466         (unaligned_storesi): Likewise.
13467         (unaligned_storehi): Likewise.
13468         (extv_reg): Likewise.
13469         (extzv_t2): Likewise.
13470         (divsi3): Likewise.
13471         (udivsi3): Likewise.
13472         (arm_zero_extendhisi2addsi): Likewise.
13473         (arm_zero_extendqisi2addsi): Likewise.
13474         (compareqi_eq0): Likewise.
13475         (arm_extendhisi2_v6): Likewise.
13476         (arm_extendqisi2addsi): Likewise.
13477         (arm_movt): Likewise.
13478         (thumb2_ldrd): Likewise.
13479         (thumb2_ldrd_base): Likewise.
13480         (thumb2_ldrd_base_neg): Likewise.
13481         (thumb2_strd): Likewise.
13482         (thumb2_strd_base): Likewise.
13483         (thumb2_strd_base_neg): Likewise.
13484         (arm_negsi2): Add alternative for 16-bit encoding.
13485         (arm_one_cmplsi2): Likewise.
13487 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13489         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13490         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13491         (movdfcc): Likewise.
13492         * config/arm/vfp.md (*thumb2_movsf_vfp):
13493         Disable predication for arm_restrict_it.
13494         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13495         (*thumb2_movdfcc_vfp): Likewise.
13496         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13497         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13498         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13499         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13500         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13501         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13502         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13503         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13504         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13505         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13506         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13507         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13508         Disable predication for arm_restrict_it.
13510 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13512         * config/i386/bmiintrin.h (_bextr_u32): New.
13513         (_bextr_u64): Ditto.
13515 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13517         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13518         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13519         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13520         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13521         * config/mips/n32-elf.h: ...this new file.
13523 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13525         PR target/57224
13526         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13527         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13529 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13531         * config/mips/mips-tables.opt: Regenerate.
13532         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13533         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13534         * doc/invoke.texi: Add -m14kc.
13536 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13538         PR target/57623
13539         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13540         constraints of operand 1 and 2.
13542         PR target/57623
13543         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13544         to match RTL canonicalization.  Swap predicates and
13545         constraints of operand 1 and 2.
13547 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13549         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13550         Process OP_INOUT regs for splitting too.
13552 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13554         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13555         decl before the loop, initialize to NULL.
13556         (vectorizable_load): Initialize ptr_incr to NULL.
13558 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13560         PR lto/57208
13561         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13562         * ipa-ref.c (ipa_maybe_record_reference): New function.
13563         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13564         * ipa-cp.c (create_specialized_node): Record potential references from
13565         aggvals.
13566         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13568 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13570         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13571         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13572         'mode' to force_reg.
13573         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13574         (aarch64_expand_mov_immediate): Likewise.
13576 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13578         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13579         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13581 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13583         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13584         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13585         (struct machine_function): Add tbegin_p.
13586         (s390_canonicalize_comparison): Fold CC mode compares to
13587         conditional jump if possible.
13588         (s390_emit_jump): Return the emitted jump.
13589         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13590         Handle CCRAWmode compares.
13591         (s390_option_override): Default to -mhtm if available.
13592         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13593         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13594         FPRs instead of df_regs_ever_live_p.
13595         (s390_optimize_nonescaping_tx): New function.
13596         (s390_init_frame_layout): Extend clobbered_regs array to cover
13597         FPRs as well.
13598         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13599         (s390_expand_tbegin): New function.
13600         (enum s390_builtin): New enum definition.
13601         (code_for_builtin): New array definition.
13602         (s390_init_builtins): New function.
13603         (s390_expand_builtin): New function.
13604         (TARGET_INIT_BUILTINS): Define.
13605         (TARGET_EXPAND_BUILTIN): Define.
13606         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13607         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13608         (s390_alc_comparison): Likewise.
13609         * config/s390/s390-modes.def: Add CCRAWmode.
13610         * config/s390/s390.h (processor_flags): Add PF_TX.
13611         (TARGET_CPU_HTM): Define macro.
13612         (TARGET_HTM): Define macro.
13613         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13614         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13615         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13616         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13617         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13618         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13619         ("*cc_to_int"): Move up.
13620         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13621         constants other than 0.
13622         ("*ccraw_to_int"): New insn and splitter definition.
13623         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13624         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13625         ("tx_assist"): New expander.
13626         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13627         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13628         * config/s390/s390.opt: Add -mhtm option.
13629         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13630         * config/s390/htmxlintrin.h: New file.
13631         * config/s390/htmintrin.h: New file.
13632         * config/s390/s390intrin.h: New file.
13633         * doc/extend.texi: Document htm builtins.
13634         * config.gcc: Add the new header files to extra_headers.
13636 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13638         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13639         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13641 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13642             Pat Haugen  <pthaugen@us.ibm.com>
13643             Peter Bergner  <bergner@vnet.ibm.com>
13645         * config/rs6000/power8.md: New.
13646         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13647         setting for power8 entry.
13648         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13649         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13650         test for Power4/Power5 only.
13651         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13652         support.
13653         (force_new_group): Adjust comment.
13654         * config/rs6000/rs6000.md: Include power8.md.
13656 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13658         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13659         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13660         declaration.
13661         (tune_params): Update comment.
13662         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13663         (arm_max_conditional_execute): New function.
13664         (thumb2_final_prescan_insn): Use max_insn_skipped and
13665         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13667 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13669         PR tree-optimization/57705
13670         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13671         SSA_NAME step, provided that it is not defined inside the loop.
13672         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13673         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13675 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13677         PR middle-end/57670
13678         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13679         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13680         calls in the dump.
13681         (ipa_note_param_call): Initialize member_ptr flag.
13682         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13683         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13684         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13685         (ipa_read_indirect_edge_info): Likewise.
13687 2013-06-25  Richard Biener  <rguenther@suse.de>
13689         PR middle-end/56977
13690         * passes.c (init_optimization_passes): Move pass_fold_builtins
13691         and pass_dce earlier with -Og.
13693 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13695         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13696         <BIT_FIELD_REF>: Remove trailing TAB.
13697         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13698         remove blank line.
13700 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13702         PR tree-optimization/57358
13703         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13704         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13705         (ipa_analyze_params_uses): Generate pessimistic info when true.
13707 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13709         PR tree-optimization/57539
13710         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13711         global.inlined_to of the new node to it.  All callers changed.
13712         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13713         inlining_into, pass it to cgraph_clone_node.
13714         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13715         ipa_free_edge_args_substructures.
13716         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13717         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13718         Assert we have found an rdesc in the rdesc list.
13720 2013-06-24  Richard Biener  <rguenther@suse.de>
13722         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13723         (pointer_set_lookup): Declare.
13724         (class pointer_map): New template class implementing a
13725         generic pointer to T map.
13726         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13727         pointer_map<T>::contains, pointer_map<T>::insert,
13728         pointer_map<T>::traverse): New functions.
13729         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13730         (pointer_set_lookup): New function.
13731         (pointer_set_contains): Use pointer_set_lookup.
13732         (pointer_set_insert): Likewise.
13733         (insert_aux): Remove.
13734         (struct pointer_map_t): Embed a pointer_set_t.
13735         (pointer_map_create): Adjust.
13736         (pointer_map_destroy): Likewise.
13737         (pointer_map_contains): Likewise.
13738         (pointer_map_insert): Likewise.
13739         (pointer_map_traverse): Likewise.
13740         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13741         pointer_map<unsigned> instead of a pointer_map_t.
13742         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13743         (streamer_tree_cache_lookup): Likewise.
13744         (streamer_tree_cache_create): Likewise.
13745         (streamer_tree_cache_delete): Likewise.
13746         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13747         pointer_map<unsigned> instead of a pointer_map_t.
13748         (lto_init_tree_ref_encoder): Adjust.
13749         (lto_destroy_tree_ref_encoder): Likewise.
13750         * lto-section-out.c (lto_output_decl_index): Likewise.
13751         (lto_record_function_out_decl_state): Likewise.
13752         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13754 2013-06-24  Richard Biener  <rguenther@suse.de>
13756         PR tree-optimization/57488
13757         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13759 2013-06-24  Alan Modra  <amodra@gmail.com>
13761         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13762         (gen_easy_altivec_constant): Likewise.
13763         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13764         easy_vector_constant_msb): Likewise.
13766 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13768         PR target/57688
13769         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13770         add missing return true.
13772 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13774         PR target/52483
13775         * config/sh/predicates.md (general_extend_operand): Invoke
13776         general_movsrc_operand for memory operands.
13777         (general_movsrc_operand): Allow reg+reg addressing, do not use
13778         general_operand for memory operands.
13780 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13782         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13783         when current target options does not apply.
13784         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13785         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13786         * config/i386/bmiintrin.h: Pass appropriate target
13787         attributes to header.
13788         * config/i386/mmintrin.h: Ditto.
13789         * config/i386/nmmintrin.h: Ditto.
13790         * config/i386/avx2intrin.h: Ditto.
13791         * config/i386/fxsrintrin.h: Ditto.
13792         * config/i386/tbmintrin.h: Ditto.
13793         * config/i386/xsaveintrin.h: Ditto.
13794         * config/i386/f16cintrin.h: Ditto.
13795         * config/i386/xtestintrin.h: Ditto.
13796         * config/i386/xsaveoptintrin.h: Ditto.
13797         * config/i386/bmi2intrin.h: Ditto.
13798         * config/i386/lzcntintrin.h: Ditto.
13799         * config/i386/smmintrin.h: Ditto.
13800         * config/i386/wmmintrin.h: Ditto.
13801         * config/i386/x86intrin.h: Remove all header include guards.
13802         * config/i386/prfchwintrin.h: Ditto.
13803         * config/i386/pmmintrin.h: Ditto.
13804         * config/i386/tmmintrin.h: Ditto.
13805         * config/i386/xmmintrin.h: Ditto.
13806         * config/i386/popcntintrin.h: Ditto.
13807         * config/i386/rdseedintrin.h: Ditto.
13808         * config/i386/ammintrin.h: Ditto.
13809         * config/i386/emmintrin.h: Ditto.
13810         * config/i386/immintrin.h: Remove all header include guards.
13811         * config/i386/fma4intrin.h: Ditto.
13812         * config/i386/lwpintrin.h: Ditto.
13813         * config/i386/xopintrin.h: Ditto.
13814         * config/i386/ia32intrin.h: Ditto.
13815         * config/i386/avxintrin.h: Ditto.
13816         * config/i386/rtmintrin.h: Ditto.
13817         * config/i386/fmaintrin.h: Ditto.
13818         * config/i386/mm3dnow.h: Ditto.
13820 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13822         * common/config/i386/i386-common.c: Handle LZCNT.
13824 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13826         * doc/extend.texi: Use __atomic_store_n instead of
13827         __atomic_store in HLE example.
13829 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13831         * config/sh/sh.c: Remove <cstdlib> workaround.
13833 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13835         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13837 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13839         * doc/extend.texi: Document that __atomic_clear and
13840           __atomic_test_and_set should only be used with bool.
13842 2013-06-20  Jan Hubicka  <jh@suse.cz>
13844         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13845         types_same_for_odr.
13846         * tree.c (decls_same_for_odr): New function.
13847         (same_for_edr): New function.
13848         (types_same_for_odr): New function.
13849         (get_binfo_at_offset): Use it.
13850         * tree.h (types_same_for_odr): Declare.
13852 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13853             Jason Merrill  <jason@redhat.com>
13855         * system.h: Include <cstdlib> as well as <stdlib.h>.
13857 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13859         PR target/57655
13860         * config/i386/i386.c (construct_container): Report error if
13861         long double is used with disabled x87 float returns.
13863 2013-06-20  Jan Hubicka  <jh@suse.cz>
13865         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13867 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13869         PR rtl-optimization/57425
13870         PR rtl-optimization/57569
13871         * alias.c (write_dependence_p): Remove parameters mem_mode and
13872         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13873         Changed all callers.
13874         (canon_anti_dependence): Get comments and semantics in sync.
13875         Add parameter mem_canonicalized.  Changed all callers.
13876         * rtl.h (canon_anti_dependence): Update prototype.
13878 2013-06-20  Richard Biener  <rguenther@suse.de>
13880         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13881         byte case, inline streamer_read_uchar and defer section
13882         overrun check.
13884 2013-06-20  Richard Biener  <rguenther@suse.de>
13886         PR tree-optimization/57584
13887         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13888         SSA names into the expanded expression that take part in
13889         abnormal coalescing.
13891 2013-06-19  Sharad Singhai  <singhai@google.com>
13893         * gcov.c (print_usage): Handle new option.
13894         (process_args): Ditto.
13895         (get_gcov_intermediate_filename): New function.
13896         (output_intermediate_file): New function.
13897         (output_gcov_file): New function
13898         (generate_results): Handle new option.
13899         (release_function): Relase demangled name.
13900         (read_graph_file): Handle demangled name.
13901         (output_lines): Ditto.
13902         * doc/gcov.texi: Document gcov intermediate format.
13904 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13906         PR bootstrap/57604
13907         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13908         (lra_emit_add): Use the functions.  Add comment about Y as an
13909         address segment.
13911 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13913         PR driver/57652
13914         * collect2.c (collect_atexit): New.
13915         (collect_exit): Delete.
13916         (main): Register collect_atexit with atexit.
13917         (collect_wait): Change collect_exit to exit.
13918         (do_wait): Same.
13919         * collect2.h (collect_exit): Delete.
13920         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13922 2013-06-19  Wei Mi  <wmi@google.com>
13924         PR rtl-optimization/57518
13925         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13926         if regno is used in paradoxical subreg.
13927         (update_equiv_regs): Check pdx_subregs[regno] before
13928         set a reg to be equivalent with a mem.
13930 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13932         PR driver/57651
13933         * file-find.h (find_a_file): Add a mode parameter.
13934         * file-find.c (find_a_file): Likewise.
13935         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13936         with X_OK for the executables.
13937         * collect2.c (main): Call find_a_file with X_OK.
13939 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13941         PR target/56942
13942         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13943         Use NEXT_INSN instead of next_real_insn.
13945 2013-06-19  Jan Hubicka  <jh@suse.cz>
13947         * cgraph.h (const_value_known_p): Replace by ...
13948         (ctor_for_folding): .. this one.
13949         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13950         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13951         * expr.c (expand_expr_real_1): Likewise.
13952         (string_constant): Likewise.
13953         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
13954         * ipa.c (process_references): Likewise.
13955         (symtab_remove_unreachable_nodes): Likewise.
13956         * ipa-inline-analysis.c (param_change_prob): Likewise.
13957         * gimple-fold.c (canonicalize_constructor_val): Likewise.
13958         (get_base_constructor): Likwise.
13959         * varpool.c (varpool_remove_node): Likewise.
13960         (varpool_remove_initializer): LIkewise.
13961         (dump_varpool_node): LIkwise.
13962         (const_value_known_p): Rewrite to ...
13963         (ctor_for_folding): ... this one.
13965 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
13967         PR driver/57651
13968         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
13969         PERSONALITY in $PATH derived prefixes.
13971 2013-06-19  Jeff Law  <law@redhat.com>
13973         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
13974         in comment.
13976         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
13977         (simplify_bitwise_binary): Use it to simpify certain binary ops on
13978         booleans.
13980 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
13982         * config/arm/vfp.md: Move VFP instruction classification documentation
13983         to ...
13984         * config/arm/arm.md: ... here.  Update instruction classification
13985         documentation.
13987 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
13989         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
13990         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
13991         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
13993 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
13995         PR target/57609
13996         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13997         with NEXT_INSN.  Use tablejump_p to check for jump table data
13998         insns.
14000 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14002         PR c++/56544
14003         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14004         that now in C++ the value is correct per the C++ standards.
14006 2013-06-19  Richard Biener  <rguenther@suse.de>
14008         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14009         for global context.
14011 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14013         Revert:
14014         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14016         PR target/57609
14017         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14018         with next_active_insn.
14020 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14022         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14023         functions are inlined during failures to flag an error.
14024         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14025         in early inline pass.
14027 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14029         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14030         in comments.
14032 2013-06-18  Julian Brown  <julian@codesourcery.com>
14034         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14035         Permit virtual register pre-reload if !strict.
14036         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14037         change.
14038         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14039         prototype.
14040         * config/arm/neon.md (movmisalign<mode>): Use
14041         neon_perm_struct_or_reg_operand instead of
14042         neon_struct_or_register_operand.
14043         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14044         neon_permissive_struct_operand instead of neon_struct_operand.
14045         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14046         neon_vector_mem_operand.
14047         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14048         neon_vector_mem_operand.
14049         (neon_permissive_struct_operand): New.
14050         (neon_struct_or_register_operand): Rename to...
14051         (neon_perm_struct_or_reg_operand): This. Adjust call to
14052         neon_vector_mem_operand.
14054 2013-06-18  Richard Biener  <rguenther@suse.de>
14056         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14057         * lto-streamer.h: Include pointer-set.h.
14058         (struct lto_decl_slot): Remove.
14059         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14060         Remove next_index entry.
14061         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14062         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14063         (lto_init_tree_ref_encoder): Adjust.
14064         (lto_destroy_tree_ref_encoder): Likewise.
14065         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14066         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14067         (lto_output_decl_index): Adjust.
14068         (lto_new_out_decl_state): Likewise.
14069         (lto_record_function_out_decl_state): Likewise.
14070         * lto-streamer-out.c (copy_function): Likewise.
14072 2013-06-18  Richard Biener  <rguenther@suse.de>
14074         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14075         * cgraphunit.c: Include cfgloop.h.
14076         (init_lowered_empty_function): Initialize the loop tree.
14077         (assemble_thunk): Insert new BBs into loops.
14079 2013-06-18  Richard Biener  <rguenther@suse.de>
14081         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14082         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14083         the map from cache entry to cache index optional.
14084         (streamer_tree_cache_replace_tree): Adjust accordingly.
14085         (streamer_tree_cache_append): Likewise.
14086         (streamer_tree_cache_delete): Likewise.
14087         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14088         streamer cache map from cache entry to cache index.
14089         * lto-streamer-out.c (create_output_block): Adjust.
14091 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14093         * config/arm/arm.md (attribute "insn"): Move multiplication and
14094         division attributes to...
14095         (attribute "type"): ... here.  Remove mult.
14096         (attribute "mul32"): New attribute.
14097         (attribute "mul64"): Add umaal.
14098         (*arm_mulsi3): Update attributes.
14099         (*arm_mulsi3_v6): Likewise.
14100         (*thumb_mulsi3): Likewise.
14101         (*thumb_mulsi3_v6): Likewise.
14102         (*mulsi3_compare0): Likewise.
14103         (*mulsi3_compare0_v6): Likewise.
14104         (*mulsi_compare0_scratch): Likewise.
14105         (*mulsi_compare0_scratch_v6): Likewise.
14106         (*mulsi3addsi): Likewise.
14107         (*mulsi3addsi_v6): Likewise.
14108         (*mulsi3addsi_compare0): Likewise.
14109         (*mulsi3addsi_compare0_v6): Likewise.
14110         (*mulsi3addsi_compare0_scratch): Likewise.
14111         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14112         (*mulsi3subsi): Likewise.
14113         (*mulsidi3adddi): Likewise.
14114         (*mulsi3addsi_v6): Likewise.
14115         (*mulsidi3adddi_v6): Likewise.
14116         (*mulsidi3_nov6): Likewise.
14117         (*mulsidi3_v6): Likewise.
14118         (*umulsidi3_nov6): Likewise.
14119         (*umulsidi3_v6): Likewise.
14120         (*umulsidi3adddi): Likewise.
14121         (*umulsidi3adddi_v6): Likewise.
14122         (*smulsi3_highpart_nov6): Likewise.
14123         (*smulsi3_highpart_v6): Likewise.
14124         (*umulsi3_highpart_nov6): Likewise.
14125         (*umulsi3_highpart_v6): Likewise.
14126         (mulhisi3): Likewise.
14127         (*mulhisi3tb): Likewise.
14128         (*mulhisi3bt): Likewise.
14129         (*mulhisi3tt): Likewise.
14130         (maddhisi4): Likewise.
14131         (*maddhisi4tb): Likewise.
14132         (*maddhisi4tt): Likewise.
14133         (maddhidi4): Likewise.
14134         (*maddhidi4tb): Likewise.
14135         (*maddhidi4tt): Likewise.
14136         (divsi3): Likewise.
14137         (udivsi3): Likewise.
14138         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14139         (thumb2_mulsi_short_compare0): Likewise.
14140         (thumb2_mulsi_short_compare0_scratch): Likewise.
14141         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14142         (1020mult2): Likewise.
14143         (1020mult3): Likewise.
14144         (1020mult4): Likewise.
14145         (1020mult5): Likewise.
14146         (1020mult6): Likewise.
14147         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14148         change.
14149         (cortex_a15_mult64): Likewise.
14150         (cortex_a15_sdiv): Likewise.
14151         (cortex_a15_udiv): Likewise.
14152         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14153         (mult2): Likewise.
14154         (mult3): Likewise.
14155         (mult4): Likewise.
14156         (mult5): Likewise.
14157         (mult6): Likewise.
14158         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14159         (pj4_ir_div): Likewise.
14160         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14161         (11_mult2): Likewise.
14162         (11_mult3): Likewise.
14163         (11_mult4): Likewise.
14164         (11_mult5): Likewise.
14165         (11_mult6): Likewise.
14166         (11_mult7): Likewise.
14167         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14168         (cortex_a8_mla): Likewise.
14169         (cortex_a8_mull): Likewise.
14170         (cortex_a8_smulwy): Likewise.
14171         (cortex_a8_smlald): Likewise.
14172         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14173         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14174         (cortex_r4_mul_3): Likewise.
14175         (cortex_r4_mla_4): Likewise.
14176         (cortex_r4_mla_3): Likewise.
14177         (cortex_r4_smlald): Likewise.
14178         (cortex_r4_mull): Likewise.
14179         (cortex_r4_sdiv): Likewise.
14180         (cortex_r4_udiv): Likewise.
14181         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14182         (cortex_a7_idiv): Likewise.
14183         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14184         (9_mult2): Likewise.
14185         (9_mult3): Likewise.
14186         (9_mult4): Likewise.
14187         (9_mult5): Likewise.
14188         (9_mult6): Likewise.
14189         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14190         (cortex_a53_sdiv): Likewise.
14191         (cortex_a53_udiv): Likewise.
14192         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14193         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14194         (mp626_mult2): Likewise.
14195         (mp626_mult3): Likewise.
14196         (mp626_mult4): Likewise.
14197         * config/arm/fa526.md (526_mult1): Update attribute change.
14198         (526_mult2): Likewise.
14199         * config/arm/arm-generic.md (mult): Update attribute change.
14200         (mult_ldsched_strongarm): Likewise.
14201         (mult_ldsched): Likewise.
14202         (multi_cycle): Likewise.
14203         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14204         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14205         (606te_mult2): Likewise.
14206         (606te_mult3): Likewise.
14207         (606te_mult4): Likewise.
14208         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14209         (cortex_a9_mac16): Likewise.
14210         (cortex_a9_multiply): Likewise.
14211         (cortex_a9_mac): Likewise.
14212         (cortex_a9_multiply_long): Likewise.
14213         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14214         (626te_mult2): Likewise.
14215         (626te_mult3): Likewise.
14216         (626te_mult4): Likewise.
14218 2013-06-18  Richard Biener  <rguenther@suse.de>
14220         PR lto/57334
14221         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14223 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14225         PR target/57609
14226         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14227         with next_active_insn.
14229 2013-06-18  Alan Modra  <amodra@gmail.com>
14231         * config/rs6000/rs6000.h (enum data_align): New.
14232         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14233         (DATA_ABI_ALIGNMENT): Define.
14234         (CONSTANT_ALIGNMENT): Correct comment.
14235         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14236         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14238 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14240         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14241         ATTRIBUTE_UNUSED marking.
14243 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14245         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14246         alternative and update.
14247         (aarch64_dup_lanedi): Delete.
14248         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14249         * config/aarch64/aarch64-simd-builtins.def: Update.
14251 2013-06-17  Richard Biener  <rguenther@suse.de>
14253         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14254         (lto_input_scc): Declare.
14255         (lto_input_tree_1): Likewise.
14256         (struct lto_stats_d): Add num_tree_bodies_output and
14257         num_pickle_refs_output.
14258         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14259         (lto_read_tree_1): Split out from ...
14260         (lto_read_tree): ... this.
14261         (lto_input_scc): New function.
14262         (lto_input_tree_1): Split out from ...
14263         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14264         (lto_data_in_create): Create the streamer cache without hashes.
14265         * lto-streamer-out.c (create_output_block): Create the streamer
14266         cache with hashes when not doing WPA.
14267         (lto_write_tree_1): Split out from ...
14268         (lto_write_tree): ... this.
14269         (get_symbol_initial_value): New function.
14270         (lto_output_tree_1): Split out from ...
14271         (lto_output_tree): ... this.  Write trees as series of SCCs
14272         using a DFS walk via DFS_write_tree.
14273         (struct sccs, struct scc_entry): New types.
14274         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14275         (DFS_write_tree_body): New function.
14276         (DFS_write_tree): Likewise.
14277         (hash_tree): Likewise.
14278         (scc_entry_compare): Likewise.
14279         (hash_scc): Likewise.
14280         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14281         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14282         TREE_CHAIN as regular reference.
14283         (streamer_read_integer_cst): Remove.
14284         (streamer_get_pickled_tree): Adjust.
14285         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14286         of DECL_EXTERNALs in BLOCK_VARS for now.
14287         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14288         reference.
14289         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14290         Add hash value argument and record that if hashes are recorded
14291         in the cache.
14292         (streamer_tree_cache_insert_1): Adjust.
14293         (streamer_tree_cache_insert): Likewise.
14294         (streamer_tree_cache_insert_at): Rename to ...
14295         (streamer_tree_cache_replace_tree): ... this and adjust.
14296         (streamer_tree_cache_append): Adjust.
14297         (record_common_node): Likewise.
14298         (streamer_tree_cache_create): Add argument whether to
14299         record hash values together with trees.
14300         (streamer_tree_cache_delete): Adjust.
14301         * tree-streamer.h (struct streamer_tree_cache_d): Add
14302         vector of hashes.
14303         (streamer_read_integer_cst): Remove.
14304         (streamer_tree_cache_insert): Adjust.
14305         (streamer_tree_cache_append): Likewise.
14306         (streamer_tree_cache_insert_at): Rename to ...
14307         (streamer_tree_cache_replace_tree): ... this and adjust.
14308         (streamer_tree_cache_create): Add argument whether to record hashes.
14309         (streamer_tree_cache_get): Rename to ...
14310         (streamer_tree_cache_get_tree): ... this.
14311         (streamer_tree_cache_get_hash): New function.
14312         * tree.c (cache_integer_cst): New function.
14313         * tree.h (cache_integer_cst): Declare.
14314         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14315         * lto-symtab.c (lto_varpool_replace_node): Only release
14316         DECL_INITIAL of non-prevailing decls.
14317         * varpool.c (varpool_remove_initializer): Do not release
14318         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14320 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14322         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14323         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14324         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14325         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14326         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14327         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14328         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14329         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14330         instead of TARGET_64BIT.
14331         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14332         Require ISA_HAS_<D>DIV.
14334 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14336         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14337         (mips*-*-linux*): Move default with_llsc setting to where other
14338         defaults are set.
14339         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14340         with_arch block.
14341         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14342         Likewise.  Remove default with_tune setting.  Move default float
14343         setting to its own block.  Handle with_llsc in the same block as above.
14345 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14347         PR rtl-optimization/57425
14348         PR rtl-optimization/57569
14349         * alias.c (write_dependence_p): Add new parameters mem_mode,
14350         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14351         Changed all callers.
14352         (canon_anti_dependence): New function.
14353         * cse.c (check_dependence): Use canon_anti_dependence.
14354         * cselib.c (cselib_invalidate_mem): Likewise.
14355         * rtl.h (canon_anti_dependence): Declare.
14357 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14359         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14360         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14361         ".set mips3" for 64-bit targets.
14363 2013-06-15  Dehao Chen  <dehao@google.com>
14365         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14366         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14367         (gimple_check_call_args): Likewise.
14368         * value-prof.c (check_ic_target): Likewise.
14369         * ipa-inline.c (early_inliner): Likewise.
14370         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14371         * cgraph.c (cgraph_create_edge_1): Likewise.
14372         (cgraph_make_edge_direct): Likewise.
14374 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14376         PR target/57615
14377         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14378         rs6000_output_move_128bit to handle emitting quad memory
14379         operations.  Set attribute length to 8 bytes.
14381 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14383         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14384         New pattern.
14385         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14386         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14387         (aarch64_<su>mlsl<mode>): Likewise.
14389 2013-06-14  Mike Stump  <mikestump@comcast.net>
14391         * Makefile.in (TARGET_H): Add insn-codes.h.
14393 2013-06-14  Alan Modra  <amodra@gmail.com>
14395         PR middle-end/57134
14396         PR middle-end/57586
14397         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14398         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14399         bitfield expansion when EXPAND_MEMORY.
14400         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14402 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14404         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14405         test for clearing quad memory on 32-bit later.
14407 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14409         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14410         (fold_negate_expr): Likewise.
14411         (fold_real_zero_addition_p): Handle vectors.
14412         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14414 2013-06-14  Alan Modra  <amodra@gmail.com>
14416         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14418 2013-06-13  Jan Hubicka  <jh@suse.cz>
14420         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14421         Local comdats are not externally visible.
14422         * symtab.c (dump_symtab_base): Dump externally visible.
14423         (verify_symtab_base): Verify back links in the symtab hash.
14425 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14427         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14428         CONVERT_EXPR as equal nodes.
14430 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14432         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14434 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14436         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14437         Generalize to complex and vector.
14438         * tree.c (build_all_ones_cst): New function.
14439         * tree.h (build_all_ones_cst): Declare it.
14441 2013-06-13  Alan Modra  <amodra@gmail.com>
14443         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14444         * config/rs6000/rs6000.md (signbittf2): New insn.
14445         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14446         (abstf2_internal, cmptf_internal2): Likewise.
14447         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14449 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14450             Pat Haugen  <pthaugen@us.ibm.com>
14451             Peter Bergner  <bergner@vnet.ibm.com>
14453         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14454         power8 byte, half-word, and quad-word atomic instructions.
14455         (emit_store_conditional): Likewise.
14456         (rs6000_expand_atomic_compare_and_swap): Likewise.
14457         (rs6000_expand_atomic_op): Likewise.
14459         * config/rs6000/sync.md (larx): Add new modes for power8.
14460         (stcx): Likewise.
14461         (AINT): New mode iterator to include TImode as well as normal
14462         integer modes on power8.
14463         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14464         that VSX registers are not considered.  Use AINT mode iterator
14465         instead of INT1 to allow inclusion of quad word atomic operations
14466         on power8.
14467         (load_locked<mode>): Likewise.
14468         (store_conditional<mode>): Likewise.
14469         (atomic_compare_and_swap<mode>): Likewise.
14470         (atomic_exchange<mode>): Likewise.
14471         (atomic_nand<mode>): Likewise.
14472         (atomic_fetch_<fetchop_name><mode>): Likewise.
14473         (atomic_nand_fetch<mode>): Likewise.
14474         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14475         each type.
14476         (ATOMIC): On power8, add QImode, HImode modes.
14477         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14478         modes that promote to SImode.
14479         (load_lockedti): Convert TImode arguments to PTImode, so that we
14480         get a guaranteed even/odd register pair.
14481         (load_lockedpti): Likewise.
14482         (store_conditionalti): Likewise.
14483         (store_conditionalpti): Likewise.
14485         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14486         atomic load/store instructions.
14487         (HSI): Likewise.
14489 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14491         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14492         loads.
14493         (insn_count): New attribute, with most cases extracted from...
14494         (length): ...here.  Redefine most cases in terms of insn_count.
14495         (single_insn): Delete.
14496         (can_delay): Use insn_count to check for single instructions.
14497         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14498         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14499         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14500         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14501         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14502         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14503         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14504         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14505         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14506         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14507         rather than "length".
14508         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14509         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14510         Use "insn_count" rather than "length".
14511         * config/mips/mips-dsp.md
14512         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14513         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14514         length attributes.
14516 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14518         PR tree-optimization/57361
14519         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14521 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14523         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14524         to split.
14525         (aarch64_simd_combine<mode>): New instruction expansion.
14526         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14527         function prototype.
14528         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14529         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14531 2013-06-12  Jan Hubicka  <jh@suse.cz>
14533         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14534         decl has when in streaming stage.
14535         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14536         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14538 2013-06-12  Roland Stigge  <stigge@antcom.de>
14540         PR target/57578
14541         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14543 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14545         PR tree-optimization/57537
14546         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14547         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14549 2013-06-12  Richard Biener  <rguenther@suse.de>
14551         * data-streamer.h (streamer_write_char_stream): CSE
14552         obs->current_pointer.
14553         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14554         streamer_write_char_stream manually and optimize the resulting loop.
14555         (streamer_write_hwi_stream): Likewise.
14557 2013-06-12  Jan Hubicka  <jh@suse.cz>
14559         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14560         * cgraph.h (varpool_create_empty_node): Declare.
14561         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14562         duplicated nodes.
14563         * symtab.c (symtab_unregister_node): Be lax about missin entries
14564         in node hash.
14565         (symtab_get_node): Update comment.
14566         * varpool.c (varpool_create_empty_node): Break out from ...
14567         (varpool_node_for_decl): ... here.
14568         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14570 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14572         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14573         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14574         part.  Use straight-line flow at the end.
14575         <COMPONENT_REF>: Remove superfluous else.
14576         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14578 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14580         PR target/56564
14581         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14582         target hook even for !TREE_PUBLIC decls.  If no resolution info
14583         is available, return false for common and external decls.
14585 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14587         * config/rl78/constraints.md (U): New constraint.
14588         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14589         valloc attribute.
14591 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14593         PR target/57589
14594         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14595         to allow returning address to AT_PLATFORM name.
14597 2013-06-11  Jan Hubicka  <jh@suse.cz>
14599         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14600         * cgraph.h (symtab_node_base): Add weakref flag.
14601         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14602         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14603         (output_weakrefs): Use weakref flag.
14604         * fold-const.c (simple_operand_p): Handle WEAK.
14605         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14606         * ipa.c (varpool_externally_visible_p): Drop weakref.
14607         (function_and_variable_visibility): Update comment; fix weakref
14608         sanity checks; do not clear DECL_WEAK on them.
14609         * lto-cgraph.c (lto_output_node): update.
14610         (lto_output_varpool_node): Update.
14611         (input_overwrite_node): Update.
14612         (input_node): Update.
14613         (input_varpool_node): Update.
14614         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14615         (lto_symtab_merge_symbols): Add sanity check.
14616         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14617         * passes.c (rest_of_decl_compilation): Set static flag, too.
14618         * symtab.c (dump_symtab_base): Dump weakref.
14619         (verify_symtab_base): Sanity check weakrefs.
14620         (symtab_make_decl_local): Remove duplicated code.
14621         (symtab_alias_ultimate_target): Simplify.
14622         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14624 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14626         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14627         sequence_vect == NULL.
14629 2013-06-11  DJ Delorie  <dj@redhat.com>
14631         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14632         (rl78_unwind_word_mode): New.
14634 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14636         * final.c (debug_prefix_maps): Make static.
14638 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14640         * function.c (initial_trampoline): Remove stray copy.
14642 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14644         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14646 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14648         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14649         within bounds at the beginning of the function.
14651 2013-06-11  Alan Modra  <amodra@gmail.com>
14653         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14654         reporting.
14655         (get_named_section): Don't NULL !DECL_P decl.
14657 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14659         * doc/invoke.texi (core-avx2): Document.
14660         (slm): Likewise.
14661         (atom): Updated with MOVBE.
14663 2013-06-11  Richard Biener  <rguenther@suse.de>
14665         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14667 2013-06-11  Anton Blanchard  <anton@samba.org>
14669         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14670         correct shift value in little-endian mode.
14672 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14674         PR target/56564
14675         * varasm.c (get_variable_align): Move #endif to the right place.
14677 2013-06-10  Cary Coutant  <ccoutant@google.com>
14679         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14680         for hash so that hash table traversal order is deterministic.
14682 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14683             Pat Haugen  <pthaugen@us.ibm.com>
14684             Peter Bergner  <bergner@vnet.ibm.com>
14686         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14687         of vectors in GPRS if they are direct moves or quad word load or
14688         store moves.
14690         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14691         declaration.
14692         (direct_move_p): Likewise.
14693         (quad_load_store_p): Likewise.
14695         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14696         classes into bins based on the physical register type.
14697         (reg_class_to_reg_type): Likewise.
14698         (IS_STD_REG_TYPE): Likewise.
14699         (IS_FP_VECT_REG_TYPE): Likewise.
14700         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14701         use direct move instructions.
14702         (reload_gpr_vsx): Likewise.
14703         (reload_vsx_gpr): Likewise.
14704         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14705         information that is a simplification of register classes.  Also
14706         precalculate direct move reload helpers.
14707         (direct_move_p): New function to return true if the operation can
14708         be done as a direct move instruciton.
14709         (quad_load_store_p): New function to return true if the operation
14710         is a quad memory operation.
14711         (rs6000_legitimize_address): If quad memory, only allow register
14712         indirect for TImode addresses.
14713         (rs6000_legitimate_address_p): Likewise.
14714         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14715         (rs6000_reload_register_type): Likewise.
14716         (register_to_reg_type): Return register type.
14717         (rs6000_secondary_reload_simple_move): New helper function for
14718         secondary reload and secondary memory needed to identify anything
14719         that is a simple move, and does not need reloading.
14720         (rs6000_secondary_reload_direct_move): New helper function for
14721         secondary reload to identify cases that can be done with several
14722         instructions via the direct move instructions.
14723         (rs6000_secondary_reload_move): New helper function for secondary
14724         reload to identify moves between register types that can be done.
14725         (rs6000_secondary_reload): Add support for quad memory operations
14726         and for direct move.
14727         (rs6000_secondary_memory_needed): Likewise.
14728         (rs6000_debug_secondary_memory_needed): Change argument names.
14729         (rs6000_output_move_128bit): New function to return the move to
14730         use for 128-bit moves, including knowing about the various
14731         limitations of quad memory operations.
14733         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14734         memory operations.  call rs6000_output_move_128bit for the actual
14735         instruciton(s) to generate.
14736         (vsx_movti_64bit): Likewise.
14738         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14739         (UNSPEC_P8V_MTVSRWZ): Likewise.
14740         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14741         (UNSPEC_P8V_MTVSRD): Likewise.
14742         (UNSPEC_P8V_XXPERMDI): Likewise.
14743         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14744         (UNSPEC_FUSION_GPR): Likewise.
14745         (FMOVE128_GPR): New iterator for direct move.
14746         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14747         (f32_sv): Likewise.
14748         (f32_dm): Likewise.
14749         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14750         loads and direct move instructions.
14751         (zero_extendsidi2_lfiwzx): Likewise.
14752         (extendsidi2_lfiwax): Likewise.
14753         (extendsidi2_nocell): Likewise.
14754         (floatsi<mode>2_lfiwax): Likewise.
14755         (lfiwax): Likewise.
14756         (floatunssi<mode>2_lfiwzx): Likewise.
14757         (lfiwzx): Likewise.
14758         (fix_trunc<mode>_stfiwx): Likewise.
14759         (fixuns_trunc<mode>_stfiwx): Likewise.
14760         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14761         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14762         (parity<mode>2_cmpb): Set length/type attr.
14763         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14764         for 'mr.' to fast_compare.
14765         (bpermd_<mode>): Change type attr to popcnt.
14766         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14767         (p8_mtvsrwz_1): Likewise.
14768         (p8_mtvsrwz_2): Likewise.
14769         (reload_fpr_from_gpr<mode>): Likewise.
14770         (p8_mtvsrd_1): Likewise.
14771         (p8_mtvsrd_2): Likewise.
14772         (p8_xxpermdi_<mode>): Likewise.
14773         (reload_vsx_from_gpr<mode>): Likewise.
14774         (reload_vsx_from_gprsf): Likewise.
14775         (p8_mfvsrd_3_<mode>): LIkewise.
14776         (reload_gpr_from_vsx<mode>): Likewise.
14777         (reload_gpr_from_vsxsf): Likewise.
14778         (p8_mfvsrd_4_disf): Likewise.
14779         (multi-word GPR splits): Do not split direct moves or quad memory
14780         operations.
14782 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14784         * tree-into-ssa.c (interesting_blocks): Make static.
14786 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14788         PR target/56564
14789         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14790         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14791         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14792         (get_variable_align): New function.
14793         (get_variable_section, emit_bss, emit_common,
14794         assemble_variable_contents, place_block_symbol): Use
14795         get_variable_align instead of DECL_ALIGN.
14796         (assemble_noswitch_variable): Add align argument, use it
14797         instead of DECL_ALIGN.
14798         (assemble_variable): Adjust caller.  Use get_variable_align
14799         instead of DECL_ALIGN.
14800         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14801         caller.
14802         (DATA_ABI_ALIGNMENT): Define.
14803         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14804         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14805         opt is false, only return the psABI mandated alignment increase.
14806         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14807         (DATA_ABI_ALIGNMENT): ... this.
14808         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14809         (DATA_ABI_ALIGNMENT): ... this.
14810         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14811         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14812         (DATA_ABI_ALIGNMENT): ... this.
14813         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14814         * doc/tm.texi: Regenerated.
14816 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14818         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14819         cmp_code to construct REG_EQUAL note.
14821 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14823         PR target/57568
14824         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14825         that operands[2] doesn't overlap with operands[0].
14827 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14828             Jan Hubicka  <jh@suse.cz>
14830         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14831         hack to mark symbols as used.
14833 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14835         PR rtl-optimization/57559
14836         * lra-constraints.c (process_alt_operands): Don't discourage
14837         memory with known offset for offsetable memory constraint.
14838         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14840 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14842         * varasm.c (struct oc_local_state): Reorder fields.
14843         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14844         and adjust accordingly.
14845         (output_constructor): Reorder initialization code and adjust call to
14846         output_constructor_bitfield.
14848 2013-06-07  Jan Hubicka  <jh@suse.cz>
14850         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14852 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14854         * tree-object-size.c (unknown): Make const.
14856 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14858         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14859         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14860         for last alternative in the cpu_facility attribute.
14862 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14864         PR target/56315
14865         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14866         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14867         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14868         * config/arm/constraints.md (Dg): New constraint.
14869         * config/arm/neon.md (xordi3_neon): Remove.
14870         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14871         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14873 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14875         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14876         Clean up alternatives.
14878 2013-06-07  Alan Modra  <amodra@gmail.com>
14880         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14881         va_list_gpr_size.
14883 2013-06-07  Alan Modra  <amodra@gmail.com>
14885         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14887 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14889         * config/arm/constraints.md (Df): New constraint.
14890         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14891         Correct length attribute for last two alternatives.
14893 2013-06-07  Alan Modra  <amodra@gmail.com>
14895         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14896         override user -mfp-in-toc.
14897         (offsettable_ok_by_alignment): Consider just the current access
14898         rather than the whole object, unless BLKmode.  Handle
14899         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14900         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14901         for -mcmodel=medium.
14902         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14903         override user -mfp-in-toc or -msum-in-toc.  Default to
14904         -mno-fp-in-toc for -mcmodel=medium.
14906 2013-06-06  DJ Delorie  <dj@redhat.com>
14908         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14909         TARGET_VALID_POINTER_MODE.
14911 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14912             Pat Haugen  <pthaugen@us.ibm.com>
14913             Peter Bergner  <bergner@vnet.ibm.com>
14915         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14916         Document new power8 builtins.
14918         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14919         condition code register, to allow 128-bit logical operations to be
14920         done in the VSX or GPR registers.
14921         (nor<mode>3): Use the canonical form for nor.
14922         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14923         vclz*, and vpopcnt* vector instructions.
14924         (nand<mode>3): Likewise.
14925         (orc<mode>3): Likewise.
14926         (clz<mode>2): LIkewise.
14927         (popcount<mode>2): Likewise.
14929         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14930         that only the GPRs are recognized.
14932         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14933         support for new power8 builtins.
14935         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14936         builtin functions.
14937         (xscvdpspn): Likewise.
14938         (vclz): Likewise.
14939         (vclzb): Likewise.
14940         (vclzh): Likewise.
14941         (vclzw): Likewise.
14942         (vclzd): Likewise.
14943         (vpopcnt): Likewise.
14944         (vpopcntb): Likewise.
14945         (vpopcnth): Likewise.
14946         (vpopcntw): Likewise.
14947         (vpopcntd): Likewise.
14948         (vgbbd): Likewise.
14949         (vmrgew): Likewise.
14950         (vmrgow): Likewise.
14951         (eqv): Likewise.
14952         (eqv_v16qi3): Likewise.
14953         (eqv_v8hi3): Likewise.
14954         (eqv_v4si3): Likewise.
14955         (eqv_v2di3): Likewise.
14956         (eqv_v4sf3): Likewise.
14957         (eqv_v2df3): Likewise.
14958         (nand): Likewise.
14959         (nand_v16qi3): Likewise.
14960         (nand_v8hi3): Likewise.
14961         (nand_v4si3): Likewise.
14962         (nand_v2di3): Likewise.
14963         (nand_v4sf3): Likewise.
14964         (nand_v2df3): Likewise.
14965         (orc): Likewise.
14966         (orc_v16qi3): Likewise.
14967         (orc_v8hi3): Likewise.
14968         (orc_v4si3): Likewise.
14969         (orc_v2di3): Likewise.
14970         (orc_v4sf3): Likewise.
14971         (orc_v2df3): Likewise.
14973         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
14974         allow power8 quad mode in 64-bit.
14975         (rs6000_builtin_vectorized_function): Add support to vectorize
14976         ISA 2.07 count leading zeros, population count builtins.
14977         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
14978         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
14979         (builtin_function_type): Add vgbbd builtin function which takes an
14980         unsigned argument.
14981         (altivec_expand_vec_perm_const): Add support for new power8 merge
14982         instructions.
14984         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
14985         that does not include TImdoe for use with 32-bit.
14986         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
14987         instructions.
14988         (UNSPEC_VSX_CVDPSPN): Likewise.
14989         (vsx_xscvdpspn): Likewise.
14990         (vsx_xscvspdpn): Likewise.
14991         (vsx_xscvdpspn_scalar): Likewise.
14992         (vsx_xscvspdpn_directmove): Likewise.
14993         (vsx_and<mode>3): Split logical operations into 32-bit and
14994         64-bit. Add support to do logical operations on TImode as well as
14995         VSX vector types.  Allow logical operations to be done in either
14996         VSX registers or in general purpose registers in 64-bit mode.  Add
14997         splitters if GPRs were used. For AND, add clobber of CCmode to
14998         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
14999         encoding.
15000         (vsx_and<mode>3_32bit): Likewise.
15001         (vsx_and<mode>3_64bit): Likewise.
15002         (vsx_ior<mode>3): Likewise.
15003         (vsx_ior<mode>3_32bit): Likewise.
15004         (vsx_ior<mode>3_64bit): Likewise.
15005         (vsx_xor<mode>3): Likewise.
15006         (vsx_xor<mode>3_32bit): Likewise.
15007         (vsx_xor<mode>3_64bit): Likewise.
15008         (vsx_one_cmpl<mode>2): Likewise.
15009         (vsx_one_cmpl<mode>2_32bit): Likewise.
15010         (vsx_one_cmpl<mode>2_64bit): Likewise.
15011         (vsx_nor<mode>3): Likewise.
15012         (vsx_nor<mode>3_32bit): Likewise.
15013         (vsx_nor<mode>3_64bit): Likewise.
15014         (vsx_andc<mode>3): Likewise.
15015         (vsx_andc<mode>3_32bit): Likewise.
15016         (vsx_andc<mode>3_64bit): Likewise.
15017         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15018         and xxlorc instructions.
15019         (vsx_eqv<mode>3_64bit): Likewise.
15020         (vsx_nand<mode>3_32bit): Likewise.
15021         (vsx_nand<mode>3_64bit): Likewise.
15022         (vsx_orc<mode>3_32bit): Likewise.
15023         (vsx_orc<mode>3_64bit): Likewise.
15025         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15027         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15028         instruction.
15029         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15030         (p8_vmrgow): Likewise.
15031         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15032         GPRs to be split under VSX.
15033         (p8v_clz<mode>2): Add power8 count leading zero support.
15034         (p8v_popcount<mode>2): Add power8 population count support.
15035         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15036         support.
15038         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15039         instruction.
15041         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15042         builtin functions.
15043         (vec_nand): Likewise.
15044         (vec_vclz): Likewise.
15045         (vec_vclzb): Likewise.
15046         (vec_vclzd): Likewise.
15047         (vec_vclzh): Likewise.
15048         (vec_vclzw): Likewise.
15049         (vec_vgbbd): Likewise.
15050         (vec_vmrgew): Likewise.
15051         (vec_vmrgow): Likewise.
15052         (vec_vpopcnt): Likewise.
15053         (vec_vpopcntb): Likewise.
15054         (vec_vpopcntd): Likewise.
15055         (vec_vpopcnth): Likewise.
15056         (vec_vpopcntw): Likewise.
15058 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15060         PR rtl-optimization/57468
15061         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15062         spilled pseudos.
15064 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15066         PR rtl-optimization/57459
15067         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15068         type when setting live regs.
15070 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15072         * config/s390/s390.opt (mlra): New option.
15073         * config/s390/s390.c (s390_decompose_address): Check displacement
15074         for all registers for LRA.
15075         (s390_secondary_reload): Don't used secondary reloads for LRA.
15076         (s390_lra_p): New function.
15077         (TARGET_LRA_P): Define.
15078         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15079         of attribute cpu_facility to zarch for the last alternative.
15080         (*cmpmem_short): Ditto.
15082 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15084         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15085         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15086         (arm_expand_prologue): Likewise.
15088 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15090         PR c++/53743
15091         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15092         as this is now done by redirect_edge_and_branch_force.
15093         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15094         barriers, and fix interaction with splitting.
15095         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15096         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15097         reflect changes made in the routine.
15098         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15099         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15100         since this is called in cfglayout mode, and replace partition fixup
15101         with assert as that is now done by force_nonfallthru_and_redirect.
15102         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15103         already be marked with region crossing note.
15104         (insert_section_boundary_note): Make non-static, gate on flag
15105         has_bb_partition, rewrite to also check for multiple partitions.
15106         (rest_of_handle_reorder_blocks): Remove call to
15107         insert_section_boundary_note, now done later during free_cfg.
15108         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15109         * bb-reorder.h (insert_section_boundary_note): Declare.
15110         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15111         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15112         invoke insert_section_boundary_note.
15113         (try_redirect_by_replacing_jump): Remove unnecessary
15114         check for region crossing note.
15115         (fixup_partition_crossing): New function.
15116         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15117         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15118         in non-cfglayout mode.
15119         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15120         remove old code that tried to do this. Emit barrier correctly
15121         when we are in cfglayout mode.
15122         (last_bb_in_partition): New function.
15123         (rtl_split_edge): Correctly fixup partition boundaries.
15124         (commit_one_edge_insertion): Remove old code that tried to
15125         fixup region crossing edge since this is now handled in
15126         split_block, and set up insertion point correctly since
15127         block may now end in a jump.
15128         (verify_hot_cold_block_grouping): Guard against checking when not in
15129         linearized RTL mode.
15130         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15131         notes.
15132         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15133         rtl_verify_flow_info, so not called in cfglayout mode.
15134         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15135         (fixup_reorder_chain): Remove old code that attempted to fixup region
15136         crossing note as this is now handled in force_nonfallthru_and_redirect.
15137         (duplicate_insn_chain): Don't duplicate switch section notes.
15138         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15139         note.
15140         * basic-block.h (emit_barrier_after_bb): Declare.
15142 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15144         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15145         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15146         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15148 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15150         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15151         where appropriate.
15152         * config/arm/ldmstm.md: Regenerate.
15154 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15156         * config/arm/sync.md (atomic_loaddi_1):
15157         Disable predication for arm_restrict_it.
15158         (arm_load_exclusive<mode>): Likewise.
15159         (arm_load_exclusivesi): Likewise.
15160         (arm_load_exclusivedi): Likewise.
15161         (arm_load_acquire_exclusive<mode>): Likewise.
15162         (arm_load_acquire_exclusivesi): Likewise.
15163         (arm_load_acquire_exclusivedi): Likewise.
15164         (arm_store_exclusive<mode>): Likewise.
15165         (arm_store_exclusive<mode>): Likewise.
15166         (arm_store_release_exclusivedi): Likewise.
15167         (arm_store_release_exclusive<mode>): Likewise.
15169 2013-06-06  Richard Biener  <rguenther@suse.de>
15171         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15172         after LTO_null.
15173         (lto_tag_is_tree_code_p): Adjust.
15174         (lto_tag_is_gimple_code_p): Likewise.
15175         (lto_gimple_code_to_tag): Likewise.
15176         (lto_tag_to_gimple_code): Likewise.
15177         (lto_tree_code_to_tag): Likewise.
15178         (lto_tag_to_tree_code): Likewise.
15179         * data-streamer.h (streamer_write_hwi_in_range): Use
15180         uhwi streaming to stream the normalized range.
15181         (streamer_read_hwi_in_range): Likewise.
15183 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15185         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15186         (predicable_short_it): Likewise.
15187         (predicated): Likewise.
15188         (enabled): Handle above.
15189         (define_cond_exec): Set predicated attribute to yes.
15191 2013-06-05  Mike Stump  <mikestump@comcast.net>
15193         * gdbinit.in (__FUNCTION__): Add.
15195 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15197         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15198         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15200 2013-06-05  Jan Hubicka  <jh@suse.cz>
15202         * varasm.c (mark_decl_referenced): Revert the removal until targets
15203         are fixed.
15205 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15207         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15208         instead of mark_decl_referenced.
15210 2013-06-05  Jan Hubicka  <jh@suse.cz>
15212         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15213         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15214         and symtab_used_from_object_file_p.
15215         (cgraph_make_node_local_1): Clear forced_by_abi.
15216         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15217         * cgraph.h (symtab_node_base): Add forced_by_abi.
15218         (decide_is_variable_needed): Remove.
15219         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15220         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15221         (decide_is_symbol_needed): ... this one; handle symbols in general;
15222         always analyze virtuals; honnor forced_by_abi.
15223         (cgraph_finalize_function): Update.
15224         (varpool_finalize_decl): Update.
15225         (symbol_defined_and_needed): Remove.
15226         (analyze_functions): Update.
15227         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15228         output_refs, input_overwrite_node): Handle forced_by_abi.
15229         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15230         (address_taken_from_non_vtable_p): ... this one.
15231         (comdat_can_be_unshared_p_1): New function.
15232         (cgraph_comdat_can_be_unshared_p): Rename to ...
15233         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15234         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15235         (function_and_variable_visibility): Clear forced_by_abi as needed.
15236         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15237         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15238         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15239         * varpool.c (decide_is_variable_needed): Remove.
15241 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15243         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15244         (arm_option_override): Override arm_restrict_it where appropriate.
15245         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15246         * config/arm/arm.opt (mrestrict-it): New command-line option.
15247         * doc/invoke.texi: Document -mrestrict-it.
15249 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15251         * tsan.c (tsan_atomic_table): Make const.
15253 2013-06-05  Richard Biener  <rguenther@suse.de>
15255         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15256         index associated with the tree we are supposed to replace.
15257         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15258         TREE_ASM_WRITTEN as zero for everything but SSA names.
15260 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15262         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15264 2013-06-04  Jan Hubicka  <jh@suse.cz>
15266         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15267         (input_node, input_varpool_node): Handle correctly external same
15268         body aliases.
15269         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15270         nodes at ltrans stage.
15272 2013-06-04  Jan Hubicka  <jh@suse.cz>
15274         * ipa-inline.c (update_caller_keys): Fix availability test.
15275         (update_callee_keys): Likewise.
15276         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15277         to follow ELF standard.
15279 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15281         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15282         (mips64r5900el-*-elf*): New configurations.
15283         * config/mips/mips-cpus.def (r5900): New processor.
15284         * config/mips/mips-tables.opt: Regenerate.
15285         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15286         (mips_issue_rate): Handle PROCESSOR_R5900.
15287         (mips_reorg_process_insns): Force reorder mode for the R5900.
15288         * config/mips/mips.h (TARGET_MIPS5900): Define.
15289         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15290         TARGET_MIPS5900.
15291         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15292         TARGET_MIPS5900.
15293         * config/mips/mips.md (processor): Add r5900.
15294         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15296 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15298         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15299         into function to generate MOVI instruction.
15300         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15301         (aarch64_preferred_simd_mode): Turn into wrapper.
15302         (aarch64_output_scalar_simd_mov_immediate): New function.
15303         * config/aarch64/aarch64-protos.h: Add prototype for above.
15305 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15307         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15308         element_char member.
15309         (sizetochar): Return signed char.
15310         (aarch64_simd_valid_immediate): Remove elchar and other
15311         unnecessary variables.
15312         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15313         Calculate element_char as required.
15314         * config/aarch64/aarch64-protos.h: Update and move prototype
15315         for aarch64_output_simd_mov_immediate.
15316         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15317         Update arguments.
15319 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15321         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15322         information completed by aarch64_simd_valid_immediate.
15323         (aarch64_legitimate_constant_p): Update arguments.
15324         (aarch64_simd_valid_immediate): Work with struct rather than many
15325         pointers.
15326         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15327         (aarch64_simd_make_constant): Update arguments.
15328         (aarch64_output_simd_mov_immediate): Work with struct rather than
15329         many pointers.  Output immediate directly rather than as operand.
15330         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15331         Update prototype.
15332         * config/aarch64/constraints.md (Dn): Update arguments.
15334 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15336         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15337         longer static.
15338         (aarch64_simd_immediate_valid_for_move): Remove.
15339         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15340         (aarch64_simd_make_constant): Update call.
15341         (aarch64_output_simd_mov_immediate): Update call.
15342         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15343         Add prototype.
15344         * config/aarch64/constraints.md (Dn): Update call.
15346 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15348         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15349         return type to bool for prototype.
15350         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15351         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15352         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15354 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15356         * config/mips/mips.opt (meva): New.
15357         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15358         (ASM_SPEC): Handle -meva.
15359         * doc/invoke.texi (meva):  Document.
15361 2013-06-04  Alan Modra  <amodra@gmail.com>
15363         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15364         constant output.
15366 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15368         * rtl.def: Add extra fourth optional field to define_cond_exec.
15369         * gensupport.c (process_one_cond_exec): Process attributes from
15370         define_cond_exec.
15371         * doc/md.texi: Document fourth field in define_cond_exec.
15373 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15375         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15376         out the processing order as in store_bit_field_1.
15378 2013-06-04  Jan Hubicka  <jh@suse.cz>
15380         PR middle-end/57500
15381         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15382         non-VAR_DECL node if it does not exist yet.
15384 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15386         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15387         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15388         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15389         target_cpu_default setting.
15391 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15393         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15394         default to -fopt-info=optimized instead of all.
15395         * doc/invoke.texi: Ditto.
15396         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15397         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15398         (execute_vect_slp): Emit BB vectorization success under
15399         MSG_OPTIMIZED_LOCATIONS.
15400         * tree-vect-slp.c (vect_slp_transform_bb): Change
15401         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15402         * tree-vect-loop.c (vect_transform_loop): Ditto.
15404 2013-06-03  Jason Merrill  <jason@redhat.com>
15406         PR c++/57415
15407         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15408         Use TARGET_EXPR for C++.
15410 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15412         PR rtl-optimization/57268
15413         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15414         if DEBUG_INSN_P (insn).
15416         Reapply
15417         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15419         PR rtl-optimization/57268
15420         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15421         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15423 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15425         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15426         (ix86_avoid_lea_for_addr): Likewise.
15427         (exact_dependency_1): Likewise.
15428         (ix86_adjust_cost): Likewise.
15429         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15430         removed.
15431         (do_reorder_for_imul): Fix typo, formatting and
15432         !reload_completed check removed.
15433         (ix86_sched_reorder): Fix typo and formatting.
15434         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15435         list.
15437 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15439         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15441 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15443         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15444         <STRING_CST>: Likewise.
15445         <VECTOR_CST>: Likewise.
15447 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15448             Mikael Morin  <mikael@gcc.gnu.org>
15450         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15451         * config.in: Regenerated.
15452         * configure: Regenerated.
15454 2013-06-01  Jan Hubicka  <jh@suse.cz>
15456         PR middle-end/57366
15457         * cgraphunit.c (compile): When weakref is not supported,
15458         set up transparent aliases before final output pass.
15459         * varasm.c (assemble_alias): Do not try to do it here.
15461 2013-06-01  Jan Hubicka  <jh@suse.cz>
15463         PR middle-end/57467
15464         * passes.c (for_per_function): Skip unanalyzed functions.
15466 2013-06-01  Jan Hubicka  <jh@suse.cz>
15468         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15469         (lto_symtab_merge_symbols_1): ... this one.
15470         (lto_symtab_merge_cgraph_nodes): Rename to ...
15471         (lto_symtab_merge_symbols): ... this one; simplify.
15472         * cgraph.c (same_body_aliases_done): Rename to ...
15473         (cpp_implicit_aliases_done): ... this one.
15474         (cgraph_create_function_alias): Update.
15475         (cgraph_same_body_alias): Update.
15476         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15477         (verify_edge_corresponds_to_fndecl): Simplify.
15478         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15479         (cgraph_node): Remove same_body_alias.
15480         (varpool_node): Remove alias_of and extra_name_alias.
15481         (same_body_aliases_done): Rename to ..
15482         (cpp_implicit_aliases_done): ... this one.
15483         (symtab_alias_ultimate_target): Add default parameter.
15484         (symtab_resolve_alias): New function.
15485         (fixup_same_cpp_alias_visibility): Declare.
15486         (cgraph_function_node): Add default parameter.
15487         (cgraph_node_asm_name): Likewise.
15488         (cgraph_function_or_thunk_node): Add default parameter; do
15489         not ICE when it is NULL.
15490         (varpool_variable_node): Likewise.
15491         * tree-emutls.c (create_emultls_var): Update.
15492         (ipa_lower_emutls): Update.
15493         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15494         (cgraph_reset_node): Reset alias info.
15495         (cgraph_finalize_function): Update.
15496         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15497         (analyze_function): Simplify.
15498         (cgraph_process_same_body_aliases): Simplify.
15499         (analyze_functions): Fixup same body aliases.
15500         (handle_alias_pairs): Simplify.
15501         (assemble_thunk): Update.
15502         (assemble_thunks_and_aliases): Update.
15503         (output_weakrefs): Rewrite.
15504         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15505         (lto_output_varpool_node): Likewise.
15506         (compute_ltrans_boundary): Remve assert.
15507         (get_alias_symbol): New functoin.
15508         (input_node): Rewrite alias handling.
15509         (input_varpool_node): Likewise.
15510         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15511         * ipa.c (process_references): Handle weakrefs correctly.
15512         (symtab_remove_unreachable_nodes): Likewise.
15513         * trans-mem.c (get_cg_data): Update.
15514         (ipa_tm_create_version_alias): Update.
15515         (ipa_tm_execute): Update.
15516         * symtab.c (dump_symtab_base): Dump aliases.
15517         (verify_symtab_base): Verify aliases.
15518         (symtab_node_availability): New function.
15519         (symtab_alias_ultimate_target): Simplify.
15520         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15521         handle all the fixup cases.
15522         (symtab_resolve_alias): New function.
15523         * passes.c (ipa_write_summaries): Handle weakrefs.
15524         * varpool.c (varpool_analyze_node): Simplify.
15525         (assemble_aliases): Update.
15526         (varpool_create_variable_alias): Simplify.
15527         (varpool_extra_name_alias): Simplify.
15528         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15529         (lto_symtab_merge_symbols): ... this one.
15531 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15533         Revert
15534         PR rtl-optimization/57268
15535         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15536         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15538 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15540         Partially reverted:
15541         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15543         PR middle-end/57073
15544         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15545         further up.
15547 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15549         PR rtl-optimization/57268
15550         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15551         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15553 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15555         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15556         unordered comparison operators when -fno-trapping-math is in effect
15557         on the e500.
15558         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15559         and implement unordered comparison operators properly on the e500.
15561 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15563         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15564         for constant scalar integers.
15565         (simplify_relational_operation_1): Likewise.
15567 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15569         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15570         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15571         Fix comment.
15573 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15574             Igor Zamyatin  <igor.zamyatin@intel.com>
15576         Silvermont (SLM) architecture performance tuning.
15577         * config/i386/i386.h (enum ix86_tune_indices): Add
15578         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15579         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15581         * config/i386/i386.c (initial_ix86_tune_features)
15582         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15583         (ix86_lea_outperforms): Handle Silvermont tuning.
15584         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15585         call.
15586         (ix86_use_lea_for_mov): Likewise.
15587         (ix86_avoid_lea_for_addr): Likewise.
15588         (ix86_lea_for_add_ok): Likewise.
15589         (exact_dependency_1): New function.
15590         (exact_store_load_dependency): Likewise.
15591         (ix86_adjust_cost): Handle Silvermont tuning.
15592         (do_reoder_for_imul): Likewise.
15593         (swap_top_of_ready_list): New function.
15594         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15596         * config/i386/i386.md (peepholes that split memory operand in fp
15597         converts): New.
15599 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15601         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15602         Remove un-necessary braces.
15604 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15606         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15607         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15609 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15611         PR middle-end/57073
15612         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15614 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15616         PR target/56315
15617         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15618         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15619         * config/arm/neon.md (iordi3_neon): Remove.
15620         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15621         * config/arm/predicates.md (imm_for_neon_logic_operand):
15622         Move to earlier in the file.
15623         (neon_logic_op2): Likewise.
15624         (arm_iordi_operand_neon): New predicate.
15626 2013-05-31  Richard Biener  <rguenther@suse.de>
15628         PR tree-optimization/57478
15629         PR tree-optimization/57453
15630         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15631         are life as well.
15633 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15635         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15636         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15638 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15639             Thomas Koenig  <tkoenig@gcc.gnu.org>
15641         PR middle-end/57073
15642         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15643         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15645 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15647         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15649 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15651         * target.def (register_usage_leveling_p): New hook.
15652         * targhooks.c (default_register_usage_leveling_p): New.
15653         * targhooks.h (default_register_usage_leveling_p): New prototype.
15654         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15655         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15656         * doc/tm.texi: Update.
15657         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15659 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15661         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15662         (*insv_reg<mode>): New define_insn.
15664 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15666         PR rtl-optimization/57439
15667         * postreload.c (move2add_valid_value_p): Check that we have
15668         a zero subreg_regno_offset when accessing the register in
15669         the requested mode.
15671 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15672             Igor Zamyatin  <igor.zamyatin@intel.com>
15674         Silvermont (SLM) architecture pipeline model, tuning and
15675         insn selection.
15676         * config.gcc: Add slm config options and target.
15678         * config/i386/slm.md: New.
15680         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15682         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15683         PROCESSOR_SLM.
15684         (ix86_target_macros_internal): Likewise.
15686         * config/i386/i386.c (slm_cost): New cost.
15687         (m_SLM): New macro flag.
15688         (initial_ix86_tune_features): Set m_SLM.
15689         (x86_accumulate_outgoing_args): Likewise.
15690         (x86_arch_always_fancy_math_387): Likewise.
15691         (processor_target_table): Add slm cost.
15692         (cpu_names): Add slm cpu name.
15693         (x86_option_override_internal): Set SLM ISA.
15694         (ix86_issue_rate): New case PROCESSOR_SLM.
15695         (ia32_multipass_dfa_lookahead): Likewise.
15696         (fold_builtin_cpu): Add slm.
15698         * config/i386/i386.h (TARGET_SLM): New target macro.
15699         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15700         (processor_type): Add PROCESSOR_SLM.
15702         * config/i386/i386.md (cpu): Add new value "slm".
15703         (slm.md): Include slm.md.
15705 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15706             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15708         * config/arm/arm-protos.h: Add and update function protos.
15709         * config/arm/arm.c (use_simple_return_p): New added.
15710         (thumb2_expand_return): Check simple_return flag.
15711         * config/arm/arm.md: Add simple_return and conditional simple_return.
15712         * config/arm/iterators.md: Add iterator for return and simple_return.
15714 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15716         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15717         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15718         (arm_emit_vfp_multi_reg_pop): Likewise.
15719         (thumb2_emit_ldrd_pop): Likewise.
15720         (arm_expand_epilogue): Add misc REG_CFA notes.
15721         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15723 2013-05-29  Lawrence Crowl  <crowl@google.com>
15725         * config/arm/t-arm: Update for below.
15727         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15728         Change type to hash_table.  Update dependent calls and types.
15730         * config/i386/t-cygming: Update for below.
15732         * config/i386/t-interix: Update for below.
15734         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15735         Change type to hash_table.  Update dependent calls and types.
15736         (i386_find_on_wrapper_list::wrappers): Likewise.
15738         * config/ia64/t-ia64: Update for below.
15740         * config/ia64/ia64.c (bundle_state_table):
15741         Change type to hash_table.  Update dependent calls and types.
15743         * config/mips/mips.c (mips_reorg_process_insns::htab):
15744         Change type to hash_table.  Update dependent calls and types.
15746         * config/sol2.c (solaris_comdat_htab):
15747         Change type to hash_table.  Update dependent calls and types.
15749         * config/t-sol2: Update for above.
15751 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15753         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15754         functions are not yet marked as defined.
15756 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15757             Pat Haugen  <pthaugen@us.ibm.com>
15758             Peter Bergner  <bergner@vnet.ibm.com>
15760         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15761         instructions.
15762         (VEC_A): Likewise.
15763         (VEC_C): Likewise.
15764         (vrotl<mode>3): Likewise.
15765         (vashl<mode>3): Likewise.
15766         (vlshr<mode>3): Likewise.
15767         (vashr<mode>3): Likewise.
15769         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15770         support for power8 V2DI builtins.
15772         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15773         power8 V2DI builtins.
15774         (vupkhsw): Likewise.
15775         (vupklsw): Likewise.
15776         (vaddudm): Likewise.
15777         (vminsd): Likewise.
15778         (vmaxsd): Likewise.
15779         (vminud): Likewise.
15780         (vmaxud): Likewise.
15781         (vpkudum): Likewise.
15782         (vpksdss): Likewise.
15783         (vpkudus): Likewise.
15784         (vpksdus): Likewise.
15785         (vrld): Likewise.
15786         (vsld): Likewise.
15787         (vsrd): Likewise.
15788         (vsrad): Likewise.
15789         (vsubudm): Likewise.
15790         (vcmpequd): Likewise.
15791         (vcmpgtsd): Likewise.
15792         (vcmpgtud): Likewise.
15793         (vcmpequd_p): Likewise.
15794         (vcmpgtsd_p): Likewise.
15795         (vcmpgtud_p): Likewise.
15796         (vupkhsw): Likewise.
15797         (vupklsw): Likewise.
15798         (vaddudm): Likewise.
15799         (vmaxsd): Likewise.
15800         (vmaxud): Likewise.
15801         (vminsd): Likewise.
15802         (vminud): Likewise.
15803         (vpksdss): Likewise.
15804         (vpksdus): Likewise.
15805         (vpkudum): Likewise.
15806         (vpkudus): Likewise.
15807         (vrld): Likewise.
15808         (vsld): Likewise.
15809         (vsrad): Likewise.
15810         (vsrd): Likewise.
15811         (vsubudm): Likewise.
15813         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15814         support for power8 V2DI instructions.
15816         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15817         power8 V2DI instructions.  Combine pack and unpack insns to use an
15818         iterator for each mode.  Check whether a particular mode supports
15819         Altivec instructions instead of just checking TARGET_ALTIVEC.
15820         (UNSPEC_VPKUWUM): Likewise.
15821         (UNSPEC_VPKSHSS): Likewise.
15822         (UNSPEC_VPKSWSS): Likewise.
15823         (UNSPEC_VPKUHUS): Likewise.
15824         (UNSPEC_VPKSHUS): Likewise.
15825         (UNSPEC_VPKUWUS): Likewise.
15826         (UNSPEC_VPKSWUS): Likewise.
15827         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15828         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15829         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15830         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15831         (UNSPEC_VUPKHSB): Likewise.
15832         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15833         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15834         (UNSPEC_VUPKHSH): Likewise.
15835         (UNSPEC_VUPKLSB): Likewise.
15836         (UNSPEC_VUPKLSH): Likewise.
15837         (VI2): Likewise.
15838         (VI_char): Likewise.
15839         (VI_scalar): Likewise.
15840         (VI_unit): Likewise.
15841         (VP): Likewise.
15842         (VP_small): Likewise.
15843         (VP_small_lc): Likewise.
15844         (VU_char): Likewise.
15845         (add<mode>3): Likewise.
15846         (altivec_vaddcuw): Likewise.
15847         (altivec_vaddu<VI_char>s): Likewise.
15848         (altivec_vadds<VI_char>s): Likewise.
15849         (sub<mode>3): Likewise.
15850         (altivec_vsubcuw): Likewise.
15851         (altivec_vsubu<VI_char>s): Likewise.
15852         (altivec_vsubs<VI_char>s): Likewise.
15853         (altivec_vavgs<VI_char>): Likewise.
15854         (altivec_vcmpbfp): Likewise.
15855         (altivec_eq<mode>): Likewise.
15856         (altivec_gt<mode>): Likewise.
15857         (altivec_gtu<mode>): Likewise.
15858         (umax<mode>3): Likewise.
15859         (smax<mode>3): Likewise.
15860         (umin<mode>3): Likewise.
15861         (smin<mode>3): Likewise.
15862         (altivec_vpkuhum): Likewise.
15863         (altivec_vpkuwum): Likewise.
15864         (altivec_vpkshss): Likewise.
15865         (altivec_vpkswss): Likewise.
15866         (altivec_vpkuhus): Likewise.
15867         (altivec_vpkshus): Likewise.
15868         (altivec_vpkuwus): Likewise.
15869         (altivec_vpkswus): Likewise.
15870         (altivec_vpks<VI_char>ss): Likewise.
15871         (altivec_vpks<VI_char>us): Likewise.
15872         (altivec_vpku<VI_char>us): Likewise.
15873         (altivec_vpku<VI_char>um): Likewise.
15874         (altivec_vrl<VI_char>): Likewise.
15875         (altivec_vsl<VI_char>): Likewise.
15876         (altivec_vsr<VI_char>): Likewise.
15877         (altivec_vsra<VI_char>): Likewise.
15878         (altivec_vsldoi_<mode>): Likewise.
15879         (altivec_vupkhsb): Likewise.
15880         (altivec_vupkhs<VU_char>): Likewise.
15881         (altivec_vupkls<VU_char>): Likewise.
15882         (altivec_vupkhsh): Likewise.
15883         (altivec_vupklsb): Likewise.
15884         (altivec_vupklsh): Likewise.
15885         (altivec_vcmpequ<VI_char>_p): Likewise.
15886         (altivec_vcmpgts<VI_char>_p): Likewise.
15887         (altivec_vcmpgtu<VI_char>_p): Likewise.
15888         (abs<mode>2): Likewise.
15889         (vec_unpacks_hi_v16qi): Likewise.
15890         (vec_unpacks_hi_v8hi): Likewise.
15891         (vec_unpacks_lo_v16qi): Likewise.
15892         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15893         (vec_unpacks_lo_v8hi): Likewise.
15894         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15895         (vec_pack_trunc_v8h): Likewise.
15896         (vec_pack_trunc_v4si): Likewise.
15897         (vec_pack_trunc_<mode>): Likewise.
15899         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15900         V2DI builtins.
15901         (vec_vmaxsd): Likewise.
15902         (vec_vmaxud): Likewise.
15903         (vec_vminsd): Likewise.
15904         (vec_vminud): Likewise.
15905         (vec_vpksdss): Likewise.
15906         (vec_vpksdus): Likewise.
15907         (vec_vpkudum): Likewise.
15908         (vec_vpkudus): Likewise.
15909         (vec_vrld): Likewise.
15910         (vec_vsld): Likewise.
15911         (vec_vsrad): Likewise.
15912         (vec_vsrd): Likewise.
15913         (vec_vsubudm): Likewise.
15914         (vec_vupkhsw): Likewise.
15915         (vec_vupklsw): Likewise.
15917 2013-05-29  Jan Hubicka  <jh@suse.cz>
15919         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15920         flags; reorder rest of fields in more consistent way.
15921         (varpool_node): Remove analyzed, finalized and alias.
15922         (cgraph_ndoe): Likewise.
15923         (symtab_alias_ultimate_target): New function.
15924         (cgraph_function_node): Move offline.
15925         (cgraph_reset_node): Declare.
15926         (cgraph_comdat_can_be_unshared_p): Remove.
15927         (varpool_remove_initializer): Declare.
15928         (varpool_first_defined_variable, varpool_next_defined_variable
15929         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15930         (cgraph_function_with_gimple_body_p): Update.
15931         (varpool_all_refs_explicit_p): Update.
15932         (symtab_alias_target): New function.
15933         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15934         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15935         (cgraph_function_or_thunk_node): Simplify using
15936         symtab_alias_ultimate_target.
15937         (varpool_variable_node): Likewise.
15938         * cgraph.c (cgraph_create_function_alias): Update.
15939         (cgraph_add_thunk): Update.
15940         (cgraph_remove_node): Update.
15941         (dump_cgraph_node): Do not dump removed flags.
15942         (cgraph_function_body_availability): Update.
15943         (cgraph_propagate_frequency): Update.
15944         (verify_cgraph_node): Check sanity of local flag.
15945         (cgraph_function_node): Move here from cgraph.h; revamp for
15946         cgraph_function_or_thunk_node.
15947         * lto-symtab.c (lto_varpool_replace_node): Update.
15948         (lto_symtab_resolve_can_prevail_p): Update.
15949         (lto_symtab_merge_cgraph_nodes): Update.
15950         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15951         propagate_constants_accross_call, devirtualization_time_bonus,
15952         ipcp_propagate_stage): Update.
15953         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
15954         * ipa-inline-transform.c (clone_inlined_nodes,
15955         preserve_function_body_p): Update.
15956         * ipa-reference.c (propagate): Update.
15957         (write_node_summary_p): Update.
15958         * toplev.c (wrapup_global_declaration_2): Update.
15959         * cgraphunit.c (cgraph_analyze_function): Rename to ...
15960         (analyze_function) ... this one.
15961         (cgraph_process_new_functions): Update.
15962         (cgraph_reset_node): Export.
15963         (cgraph_finalize_function): Update.
15964         (cgraph_add_new_function): Update.
15965         (process_function_and_variable_attributes): Update.
15966         (varpool_finalize_decl): Update.
15967         (symbol_finalized): Remove.
15968         (symbol_finalized_and_needed): Rename to ...
15969         (symbol_defined_and_needed): ... update.
15970         (cgraph_analyze_functions): Update.
15971         (handle_alias_pairs): Update.
15972         (mark_functions_to_output): Update.
15973         (assemble_thunk): Update.
15974         (output_in_order): Update.
15975         (output_weakrefs): Update.
15976         (finalize_compilation_unit): Update.
15977         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
15978         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
15979         input_node, input_varpool_node): Update.
15980         * dbxout.c (dbxout_expand_expr): Update.
15981         * cgraphclones.c (cgraph_clone_node): Update.
15982         (cgraph_copy_node_for_versioning): Update.
15983         (cgraph_materialize_clone): Update.
15984         (cgraph_materialize_all_clones): Update.
15985         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
15986         propagate_pure_const, propagate_nothrow): Update.
15987         * lto-streamer-out.c (lto_output, write_symbol): Update.
15988         * ipa-utils.c (ipa_reverse_postorder): Update.
15989         * ipa-inline.c (can_inline_edge_p): Update.
15990         (update_caller_keys, ipa_inline): Update.
15991         * dwarf2out.c (reference_to_unused,
15992         premark_types_used_by_global_vars_helper): Update.
15993         * tree-eh.c (tree_could_trap_p): Update.
15994         * ipa-split.c (consider_split, execute_split_functions): Update.
15995         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
15996         has_addr_references_p): Update; move ahead in file for better
15997         readability.
15998         (process_references): Simplify.
15999         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16000         bodies are removed.
16001         (cgraph_comdat_can_be_unshared_p): Make static.
16002         (cgraph_externally_visible_p): Update.
16003         (varpool_externally_visible_p): Update.
16004         (function_and_variable_visibility): Update.
16005         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16006         ipa_tm_mark_force_output_node): Update.
16007         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16008         estimate_edge_devirt_benefit, inline_generate_summary,
16009         inline_write_summary): Update.
16010         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16011         * ipa-prop.c (ipa_compute_jump_functions): Update.
16012         (ipa_print_node_params, ipa_prop_read_section,
16013         ipa_update_after_lto_read, read_replacements_section): Update.
16014         * varasm.c (mark_decl_referenced): Update.
16015         (assemble_alias, dump_tm_clone_pairs): Update.
16016         * tree-inline.c (copy_bb): Update.
16017         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16018         Update.
16019         * symtab.c (dump_symtab_base): Print new flags.
16020         (verify_symtab_base): Verify new flags.
16021         (symtab_alias_ultimate_target): New function.
16022         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16023         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16024         Update.
16025         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16026         Update.
16027         * i386.c (ix86_get_function_versions_dispatcher,
16028         ix86_generate_version_dispatcher_body): Update.
16029         (fold_builtin_cpu): Use varpool_add_new_variable.
16030         * varpool.c (varpool_remove_initializer): Break out from ...
16031         (varpool_remove_node): ... this one.
16032         (dump_varpool_node, varpool_node_for_asm,
16033         cgraph_variable_initializer_availability, varpool_analyze_node,
16034         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16035         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16036         Update.
16038 2013-05-29  Jan Hubicka  <jh@suse.cz>
16040         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16042 2013-05-29  Easwaran Raman  <eraman@google.com>
16044         PR tree-optimization/57442
16045         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16046         when control exits the main loop.
16048 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16050         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16051         and RX600.
16052         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16053         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16054         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16055         option.
16057 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16059         PR tree-optimization/57441
16060         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16061         Don't limit size of incr_vec to number of candidates.
16063 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16065         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16066         and mips16 directories.
16067         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16068         (MULTILIB_DIRNAMES): Ditto.
16069         (MULTILIB_EXCEPTIONS): Add new exceptions.
16070         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16071         (MULTILIB_DIRNAMES): Ditto.
16072         (MULTILIB_EXCEPTIONS): Add new exceptions.
16074 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16075             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16077         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16078         SYMBOL_TINY_ABSOLUTE.
16079         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16080         SYMBOL_TINY_ABSOLUTE.
16081         (aarch64_expand_mov_immediate): Likewise.
16082         (aarch64_classify_symbol): Likewise.
16083         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16084         Permit SYMBOL_TINY_ABSOLUTE.
16085         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16087 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16088             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16090         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16091         Refactor if/switch.  Replace gcc_assert with if.
16093 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16095         * config/i386/i386.c (initial_ix86_tune_features): Enable
16096         FP Reassociation for AMD bdver1 and bdver2.
16098 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16100         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16101         and IMAGPART_EXPR do not occur within other handled_components.
16103 2013-05-29  Richard Biener  <rguenther@suse.de>
16105         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16106         access on whether the use is in the BB we currently try to
16107         vectorize.
16108         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16109         vectorize to vect_bb_slp_scalar_cost.
16111 2013-05-29  Richard Biener  <rguenther@suse.de>
16113         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16114         computing scalar cost offsetted by stmts that are kept live
16115         by scalar uses.
16116         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16117         for computation of scalar cost.
16119 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16121         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16123 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16125         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16126         array notation built-in reduction functions.
16127         * doc/passes.texi (Passes): Added documentation about changes done
16128         for Cilk Plus.
16129         * doc/invoke.texi (C Dialect Options): Added documentation about
16130         the -fcilkplus flag.
16131         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16132         (BUILTINS_DEF): Depend on cilkplus.def.
16133         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16134         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16135         * cilkplus.def: New file.
16137 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16139         PR rtl-optimization/57439
16140         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16142 2013-05-28  Easwaran Raman  <eraman@google.com>
16144         PR tree-optimization/57337
16145         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16146         (find_insert_point): Correctly identify the insertion point
16147         when two statements with the same UID is compared.
16149 2013-05-28  Richard Biener  <rguenther@suse.de>
16151         PR tree-optimization/56787
16152         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16153         from the list of data references.
16154         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16155         clobbers.
16156         (vect_analyze_loop_operations): Likewise.
16157         (vect_transform_loop): Remove clobbers.
16159 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16161         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16162         and REALPART_EXPRs have scalar type.
16164 2013-05-28  Richard Biener  <rguenther@suse.de>
16166         PR tree-optimization/57411
16167         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16168         virtual operands.
16169         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16170         virtual operand propagation.
16172 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16174         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16175         destination register for bmasksi_vis.
16176         (vector_init_bshuffle): Likewise.
16177         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16179 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16181         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16182         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16183         mode if the instruction isn't available in the original mode.
16184         * config/sparc/sparc.opt (mfix-ut699): New option.
16185         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16186         (divdf3): Turn into expander.
16187         (divdf3_nofix): New insn.
16188         (divdf3_fix): Likewise.
16189         (divsf3): Disable if -mfix-ut699.
16190         (sqrtdf2): Turn into expander.
16191         (sqrtdf2_nofix): New insn.
16192         (sqrtdf2_fix): Likewise.
16193         (sqrtsf2): Disable if -mfix-ut699.
16195 2013-05-27  Richard Biener  <rguenther@suse.de>
16197         PR middle-end/57412
16198         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16199         block for the new loop.
16201 2013-05-27  Richard Biener  <rguenther@suse.de>
16203         PR tree-optimization/57343
16204         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16205         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16206         (number_of_iterations_cond): Do not build the folded tree.
16208 2013-05-27  Richard Biener  <rguenther@suse.de>
16210         Revert
16211         PR middle-end/57381
16212         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16213         OEP_CONSTANT_ADDRESS_OF retained.
16215         PR tree-optimization/57417
16216         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16217         for unchanged base.
16218         (set_ssa_val_to): Compare addresses using
16219         get_addr_base_and_unit_offset.
16221 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16223         PR rtl-optimization/56833
16224         * postreload.c (move2add_record_mode): New function.
16225         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16226         (move2add_use_add2_insn): Use move2add_record_sym_value.
16227         (move2add_use_add3_insn): Likewise.
16228         (reload_cse_move2add): Use move2add_valid_value_p and
16229         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16230         affected regs by setting reg_mode to VOIDmode.
16231         (move2add_note_store): Don't pretend the inside of a SUBREG is
16232         the actual destination.  Invalidate single/leading registers by
16233         setting reg_mode to VOIDmode.
16234         Use move2add_record_sym_value, move2add_valid_value_p and
16235         move2add_record_mode.
16237 2013-05-27  Richard Biener  <rguenther@suse.de>
16239         PR tree-optimization/57396
16240         * tree-affine.c (double_int_constant_multiple_p): Properly
16241         return false for val == 0 and div != 0.
16243 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16245         * config/mips/mips.h: Use #elif in preprocessor conditions.
16247 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16249         PR target/53916
16250         * config/mips/constraints.md (kl): New constraint.
16251         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16252         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16253         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16254         from LO for MIPS16.
16255         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16257 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16259         PR target/55777
16260         * config/mips/mips.c (mips_can_inline_p): New function.
16261         (TARGET_CAN_INLINE_P): Define.
16263 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16265         * sched-int.h (ds_t, dw_t): Make unsigned int.
16266         Fix documentation that describes how all the ds_t bits are used.
16267         Reserve the last bit for delayed-branch scheduling.
16268         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16269         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16270         (gen_dep_weak_1): Remove prototype.
16271         * sched-deps.c (get_dep_weak_1): Make static.
16272         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16273         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16274         * doc/tm.texi: Regenerate.
16275         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16277 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16279         PR debug/56950
16280         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16282 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16283             Sandra Loosemore  <sandra@codesourcery.com>
16285         * config.gcc (powerpc-*): Allow native for with-cpu.
16287 2013-05-24  Jeff Law  <law@redhat.com>
16289         PR tree-optimization/57124
16290         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16291         conversion feeding a condition if the range has an overflow
16292         if -fstrict-overflow.  Add warnings for when we do make the
16293         transformation.
16295 2013-05-24  Dehao Chen  <dehao@google.com>
16297         * tree-cfg.c (locus_discrim_map): Fix the typo.
16298         (locus_discrim_hasher): Likewise.
16299         (locus_discrim_hasher::hash): Likewise.
16300         (locus_discrim_hasher::equal): Likewise.
16302 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16304         PR tree-optimization/57294
16305         * cgraph.h (ipa_record_stmt_references): Declare.
16306         * cgraphbuild.c (ipa_record_stmt_references): New function.
16307         (build_cgraph_edges): Use ipa_record_stmt_references.
16308         (rebuild_cgraph_edges): Likewise.
16309         (cgraph_rebuild_references): Likewise.
16310         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16311         associated with the old statement and build references from the
16312         newly built statements.
16313         * ipa-ref.c (ipa_remove_stmt_references): New function.
16314         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16316 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16318         * lra-constraints.c (emit_spill_move): Use smaller mode for
16319         mem-mem moves.
16320         (check_and_process_move): Consider mem-reg moves for secondary
16321         too.
16322         (curr_insn_transform): Don't lose insns emitted before for
16323         secondary memory moves.
16324         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16325         reg set up in the current insn.
16327 2013-05-24  Dehao Chen  <dehao@google.com>
16329         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16330         hash function.
16331         (locus_descrim_hasher::equal): Likewise.
16332         (build_gimple_cfg): New discriminator assignment algorithm.
16333         (make_edges): Likewise.
16334         (next_discriminator_for_locus): Likewise.
16335         (same_line_p): Likewise.
16336         (assign_discriminators): Likewise.
16337         (make_cond_expr_edges): Likewise.
16338         (make_gimple_switch_edges): Likewise.
16339         (make_goto_expr_edges): Likewise.
16340         (make_gimple_asm_edges): Likewise.
16342 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16344         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16345         X format specifier to only display bottom 16 bits.
16346         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16347         immediate to match for operand 2, since it will be masked.
16349 2013-05-24  Richard Biener  <rguenther@suse.de>
16351         PR tree-optimization/57287
16352         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16353         all SSA names that occur in abnormal PHIs.
16355 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16357         PR tree-ssa/57385
16358         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16359         that index is not negative.
16361 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16363         PR rtl-optimization/55177
16364         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16365         (simplify_byte_swapping_operation): New.
16366         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16367         (simplify_relational_operation_1): Deal with BSWAP.
16369 2013-05-23  Richard Henderson  <rth@redhat.com>
16371         PR target/56742
16372         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16373         (ix86_reorg): Call it.
16375 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16377         PR target/57379
16378         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16379         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16380         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16382 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16384         PR debug/57351
16385         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16387 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16388             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16390         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16391         * config/aarch64/constraints.md (Usa): Remove.
16392         * doc/md.texi (AArch64 Usa): Remove.
16394 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16395             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16397         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16398         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16399         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16400         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16402 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16404         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16405         instruction (AdvSIMD).
16406         * config/aarch64/aarch64-builtins.c
16407         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16408         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16410 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16412         PR middle-end/57347
16413         * tree.h (contains_bitfld_component_ref_p): Declare.
16414         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16415         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16416         caller.
16417         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16418         not access a bit-field.  Assert all final offsets are byte-aligned.
16420 2013-05-23  Richard Biener  <rguenther@suse.de>
16422         PR tree-optimization/57380
16423         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16424         least one invariant or re-used load.
16425         * passes.c (init_optimization_passes): Move pass_phiprop before
16426         pass_forwprop.
16428 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16430         * config/aarch64/aarch64-simd.md
16431         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16433 2013-05-23  Richard Biener  <rguenther@suse.de>
16435         PR middle-end/57381
16436         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16437         OEP_CONSTANT_ADDRESS_OF retained.
16439 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16441         PR middle-end/57344
16442         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16443         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16445 2013-05-23  Richard Biener  <rguenther@suse.de>
16447         PR rtl-optimization/57341
16448         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16449         instead of true_dependence.
16451 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16453         * bb-reorder.c (branch_threshold): Make const.
16454         (exec_threshold): Ditto.
16456 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16457             Pat Haugen  <pthaugen@us.ibm.com>
16458             Peter Bergner  <bergner@vnet.ibm.com>
16460         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16461         documentation for the power8 crypto builtins.
16463         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16465         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16466         macros for defining power8 builtin functions.
16467         (BU_P8V_AV_2): Likewise.
16468         (BU_P8V_AV_P): Likewise.
16469         (BU_P8V_VSX_1): Likewise.
16470         (BU_P8V_OVERLOAD_1): Likewise.
16471         (BU_P8V_OVERLOAD_2): Likewise.
16472         (BU_CRYPTO_1): Likewise.
16473         (BU_CRYPTO_2): Likewise.
16474         (BU_CRYPTO_3): Likewise.
16475         (BU_CRYPTO_OVERLOAD_1): Likewise.
16476         (BU_CRYPTO_OVERLOAD_2): Likewise.
16477         (XSCVSPDP): Fix typo, point to the correct instruction.
16478         (VCIPHER): Add power8 crypto builtins.
16479         (VCIPHERLAST): Likewise.
16480         (VNCIPHER): Likewise.
16481         (VNCIPHERLAST): Likewise.
16482         (VPMSUMB): Likewise.
16483         (VPMSUMH): Likewise.
16484         (VPMSUMW): Likewise.
16485         (VPERMXOR_V2DI): Likewise.
16486         (VPERMXOR_V4SI: Likewise.
16487         (VPERMXOR_V8HI: Likewise.
16488         (VPERMXOR_V16QI: Likewise.
16489         (VSHASIGMAW): Likewise.
16490         (VSHASIGMAD): Likewise.
16491         (VPMSUM): Likewise.
16492         (VPERMXOR): Likewise.
16493         (VSHASIGMA): Likewise.
16495         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16496         __CRYPTO__ if the crypto instructions are available.
16497         (altivec_overloaded_builtins): Add support for overloaded power8
16498         builtins.
16500         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16501         support for power8 crypto builtins.
16502         (builtin_function_type): Likewise.
16503         (altivec_init_builtins): Add support for builtins that take vector
16504         long long (V2DI) arguments.
16506         * config/rs6000/crypto.md: New file, define power8 crypto
16507         instructions.
16509 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16510             Pat Haugen  <pthaugen@us.ibm.com>
16511             Peter Bergner  <bergner@vnet.ibm.com>
16513         * doc/invoke.texi (Option Summary): Add power8 options.
16514         (RS/6000 and PowerPC Options): Likewise.
16516         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16517         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16518         wm, wn, wr documentation.
16520         * config/rs6000/constraints.md (wm): New constraint for VSX
16521         registers if direct move instructions are enabled.
16522         (wn): New constraint for no registers.
16523         (wq): New constraint for quad word even GPR registers.
16524         (wr): New constraint if 64-bit instructions are enabled.
16525         (wv): New constraint if power8 vector instructions are enabled.
16526         (wQ): New constraint for quad word memory locations.
16528         * config/rs6000/predicates.md (const_0_to_15_operand): New
16529         constraint for 0..15 for crypto instructions.
16530         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16531         as GPR and floating point registers.
16532         (int_reg_operand): New predicate to match only GPR registers.
16533         (base_reg_operand): New predicate to match base registers.
16534         (quad_int_reg_operand): New predicate to match even GPR registers
16535         for quad memory operations.
16536         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16537         operations in both GPR and VSX registers.
16538         (quad_memory_operand): New predicate for quad memory operations.
16539         (reg_or_indexed_operand): New predicate for direct move support.
16541         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16542         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16543         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16544         (POWERPC_MASKS): Add power8 options.
16545         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16546         various options.
16548         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16549         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16551         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16552         (-mpower8-fusion): New power8 options.
16553         (-mpower8-fusion-sign): Likewise.
16554         (-mpower8-vector): Likewise.
16555         (-mcrypto): Likewise.
16556         (-mdirect-move): Likewise.
16557         (-mquad-memory): Likewise.
16559         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16560         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16561         registers.
16562         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16563         (rs6000_debug_vector_unit): Add p8_vector.
16564         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16565         definitions.  Also print fusion state.
16566         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16567         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16568         (rs6000_option_override_internal): Add support for power8.
16569         (rs6000_common_init_builtins): Add debugging for skipped builtins
16570         if -mdebug=builtin.
16571         (rs6000_adjust_cost): Add power8 support.
16572         (rs6000_issue_rate): Likewise.
16573         (insn_must_be_first_in_group): Likewise.
16574         (insn_must_be_last_in_group): Likewise.
16575         (force_new_group): Likewise.
16576         (rs6000_register_move_cost): Likewise.
16577         (rs6000_opt_masks): Likewise.
16579         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16580         power8 capable assembler, default to power7 options.
16581         (TARGET_DIRECT_MOVE): Likewise.
16582         (TARGET_CRYPTO): Likewise.
16583         (TARGET_P8_VECTOR): Likewise.
16584         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16585         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16586         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16587         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16588         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16589         (TARGET_XSCVDPSPN): Likewise.
16590         (TARGET_XSCVSPDPN): Likewsie.
16591         (TARGET_SYNC_HI_QI): Likewise.
16592         (TARGET_SYNC_TI): Likewise.
16593         (MASK_CRYPTO): Likewise.
16594         (MASK_DIRECT_MOVE): Likewise.
16595         (MASK_P8_FUSION): Likewise.
16596         (MASK_P8_VECTOR): Likewise.
16597         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16598         temporary used by some of the direct move instructions to get two FP
16599         temporary registers does not force creation of a stack frame.
16600         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16601         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16602         that any VSX registers are tieable, even if they are also an
16603         Altivec vector mode.
16604         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16605         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16606         (RS6000_BTM_CRYPTO): Likewise.
16607         (RS6000_BTM_COMMON): Likewise.
16609         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16610         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16611         (enum rs6000_vector): Add power8 vector support.
16613 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16615         PR target/19599
16616         PR target/57340
16617         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16618         (any_sibcall_could_use_r3): this and handle indirect calls.
16619         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16621 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16623         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16625 2013-05-22  Richard Biener  <rguenther@suse.de>
16627         PR middle-end/57349
16628         * profile.c (branch_prob): Do not split blocks that are
16629         abnormally receiving from ECF_RETURNS_TWICE functions.
16631 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16633         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16634         address mode.  Move pointer mode initialization to the same place.
16636 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16638         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16639         while it has any effect.
16641 2013-05-21  Easwaran Raman  <eraman@google.com>
16643         PR tree-optimization/57322
16644         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16645         UID of the statement added to the BB to be 1.
16647 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16649         PR tree-optimization/57331
16650         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16651         of conversion from pointer type to integral type with integer.
16653 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16655         PR lto/57289
16656         * ipa-prop.c (ipa_read_node_info): Process param_used and
16657         controlled_uses in the same order as when writing.
16659 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16661         PR plugins/56754
16662         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16664 2013-05-21  Richard Biener  <rguenther@suse.de>
16666         PR tree-optimization/57318
16667         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16668         estimate stmts with side-effects as likely eliminated.
16670 2013-05-21  Richard Biener  <rguenther@suse.de>
16672         PR tree-optimization/57330
16673         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16674         preserve the call stmts fntype.
16676 2013-05-21  Richard Biener  <rguenther@suse.de>
16678         PR tree-optimization/57303
16679         * tree-ssa-sink.c (statement_sink_location): Improve killing
16680         stmt detection and properly handle self-assignments.
16682 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16684         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16685         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16686         registers. Set register size out of the PARALLEL loop.
16688 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16690         PR target/56547
16691         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16692         (*fmasf4, *fmasf4_media): New insns.
16694 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16696         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16697         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16698         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16699         (mips_idiv_insns): Update the comments to say that the returned
16700         instruction counts are in units of BASE_INSN_LENGTH.
16701         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16702         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16703         using 2 rather than 4 as the length of indirect MIPS16 and
16704         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16705         length of a NOP.  Don't divide MIPS16 lengths by 2.
16706         (mips16_split_long_branches): Assume a branch is long if the
16707         length is greater than 4 rather than 8.
16708         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16709         rather than multiplying them by 2.  Multiply instruction counts
16710         by BASE_INSN_LENGTH rather than 4.
16711         (*jump_mips16, tls_get_tp_mips16_<mode>)
16712         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16714 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16716         * config/mips/mips.md (extended_mips16): Remove branch case.
16717         (length): Remove duplicated extended_mips16 test.
16719 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16721         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16723 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16725         * recog.h (Recog_data): Rename to...
16726         (recog_data_d): ...this.
16727         (recog_data): Update accordingly.
16728         * recog.c (recog_data): Likewise.
16729         * reload.c (save_recog_data): Likewise.
16730         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16731         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16733 2013-05-17  Julian Brown  <julian@codesourcery.com>
16735         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16736         found in a REG_EQUAL note, invalidate it.
16738 2013-05-17  Easwaran Raman  <eraman@google.com>
16740         * tree-ssa-reassoc.c (find_insert_point): New function.
16741         (insert_stmt_after): Likewise.
16742         (get_def_stmt): Likewise.
16743         (ensure_ops_are_available): Likewise.
16744         (not_dominated_by): Likewise.
16745         (rewrite_expr_tree): Do not move statements beyond what is
16746         necessary. Remove call to swap_ops_for_binary_stmt...
16747         (reassociate_bb): ... and move it here.
16748         (build_and_add_sum): Assign UIDs for new statements.
16749         (linearize_expr): Likewise.
16750         (do_reassoc): Renumber gimple statement UIDs.
16752 2013-05-17  Jan Hubicka  <jh@suse.cz>
16754         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16755         weakrefs.
16756         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16757         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16758         weakrefs.
16759         (output_weakrefs): Update.
16761 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16762             Martin Jambor  <mjambor@suse.cz>
16764         PR middle-end/57276
16765         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16766         value that corresponds to the given aggval is found in values vector.
16768 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16770         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16771         sse, sse2, sse3, ssse3 and sse4a flags to options.
16773 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16775         * gengtype-state.c: (s_expr_writer): New class, to handle
16776         prettifying of output layout of s-expressions.
16777         (state_writer): New class, to write out gtype.state.
16778         (state_written_type_count): Move this variable into member data of
16779         state_writer.
16780         (s_expr_writer::s_expr_writer): New code: constructor for new class
16781         (state_writer::state_writer(): ditto
16782         (s_expr_writer::write_new_line): New function
16783         (s_expr_writer::write_any_indent): ditto
16784         (s_expr_writer::begin_s_expr): ditto
16785         (s_expr_writer::end_s_expr): ditto
16786         (write_state_fileloc): convert to method of state_writer...
16787         (state_writer:: write_state_fileloc): ...and use methods of
16788         s_expr_writer to write indentation into the gtype.state output file
16789         to visually represent the hierarchical structure of the list
16790         structures
16791         (write_state_fields): ditto, renaming to...
16792         (state_writer::write_state_fields)
16793         (write_state_a_string): ditto, renaming to...
16794         (state_writer::write_state_a_string)
16795         (write_state_string_option): ditto, renaming to...
16796         (state_writer::write_state_string_option)
16797         (write_state_type_option): ditto, renaming to...
16798         (state_writer::write_state_type_option)
16799         (write_state_nested_option): ditto, renaming to...
16800         (state_writer::write_state_nested_option)
16801         (write_state_option): ditto, renaming to...
16802         (state_writer::write_state_option)
16803         (write_state_options): ditto, renaming to...
16804         (state_writer::write_state_options)
16805         (write_state_lang_bitmap): ditto, renaming to...
16806         (state_writer::write_state_lang_bitmap)
16807         (write_state_version): ditto, renaming to...
16808         (state_writer::write_state_version)
16809         (write_state_scalar_type): ditto, renaming to...
16810         (state_writer::write_state_scalar_type)
16811         (write_state_string_type): ditto, renaming to...
16812         (state_writer::write_state_string_type)
16813         (write_state_undefined_type): ditto, renaming to...
16814         (state_writer::write_state_undefined_type)
16815         (write_state_struct_union_type): ditto, renaming to...
16816         (state_writer::write_state_struct_union_type)
16817         (write_state_struct_type): ditto, renaming to...
16818         (state_writer::write_state_struct_type)
16819         (write_state_user_struct_type): ditto, renaming to...
16820         (state_writer::write_state_user_struct_type)
16821         (write_state_lang_struct_type): ditto, renaming to...
16822         (state_writer::write_state_lang_struct_type)
16823         (write_state_param_struct_type): ditto, renaming to...
16824         (state_writer::write_state_param_struct_type)
16825         (write_state_pointer_type): ditto, renaming to...
16826         (state_writer::write_state_pointer_type)
16827         (write_state_array_type): ditto, renaming to...
16828         (state_writer::write_state_array_type)
16829         (write_state_gc_used): ditto, renaming to...
16830         (state_writer::write_state_gc_used)
16831         (write_state_common_type_content): ditto, renaming to...
16832         (state_writer::write_state_common_type_content)
16833         (write_state_type): ditto, renaming to...
16834         (state_writer::write_state_type)
16835         (write_state_pair_list): ditto, renaming to...
16836         (state_writer::write_state_pair_list)
16837         (write_state_pair): ditto, renaming to...
16838         (state_writer::write_state_pair)
16839         (write_state_typedefs): ditto, renaming to...
16840         (state_writer::write_state_typedefs)
16841         (write_state_structures): ditto, renaming to...
16842         (state_writer::write_state_structures)
16843         (write_state_param_structs): ditto, renaming to...
16844         (state_writer::write_state_param_structs)
16845         (write_state_variables): ditto, renaming to...
16846         (state_writer::write_state_variables)
16847         (write_state_srcdir): ditto, renaming to...
16848         (state_writer::write_state_srcdir)
16849         (write_state_files_list): ditto, renaming to...
16850         (state_writer::write_state_files_list)
16851         (write_state_languages): ditto, renaming to...
16852         (state_writer::write_state_languages)
16853         (write_state): create a state_writer instance and use it when
16854         writing out the state file
16856 2013-05-17  Mike Stump  <mikestump@comcast.net>
16858         PR rtl-optimization/57304
16859         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16860         accessing DF_REF_REAL_LOC.
16862 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16864         PR rtl-optimization/57281
16865         PR rtl-optimization/57300
16866         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16867         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16868         what the other splitter did if the registers are dead.
16870 2013-05-17  Richard Biener  <rguenther@suse.de>
16872         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16873         MEM_REF offsets.
16875 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16877         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16878         linking.
16880 2013-05-17  Marek Polacek  <polacek@redhat.com>
16882         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16883         length when doing non-zero store of storing '\0' to '\0'.
16885 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16887         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16888         vect_external_def oprnd1 with loop_vinfo, try to emit
16889         optional cast, negation and and stmts on the loop preheader
16890         edge instead of into the pattern def seq.
16892         PR tree-optimization/57051
16893         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16894         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16896 2013-05-16  Nick Clifton  <nickc@redhat.com>
16898         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16899         (rl78_is_naked_func): New function.
16900         (rl78_expand_prologue): Skip prologue generation for naked functions.
16901         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16902         * doc/extend.texi (naked): Add RL78 to the list of processors
16903         that supports this attribute.
16905 2013-05-16  Jeff Law  <law@redhat.com>
16907         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16909 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16911         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16912         cache parameters using detect_caches_amd also for CYRIX,
16913         NSC and TM2 signatures.
16915 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16916             Dzianis Kahanovich  <mahatma@eu.by>
16918         PR target/45359
16919         PR target/46396
16920         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16921         VIA/Centaur processors and determine their cache parameters
16922         using detect_caches_amd.
16924 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16926         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16927         (rtl_verify_edges): New function.
16928         (rtl_verify_bb_insns): Ditto.
16929         (rtl_verify_bb_pointers): Ditto.
16930         (rtl_verify_bb_insn_chain): Ditto.
16931         (rtl_verify_fallthru): Ditto.
16932         (rtl_verify_bb_layout): Ditto.
16933         (rtl_verify_flow_info_1): Outline checks into new functions.
16934         (rtl_verify_flow_info): Ditto.
16936 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16938         * cfghooks.c (copy_bbs): Add update_dominance argument.
16939         * cfghooks.h (copy_bbs): Update prototype.
16940         * tree-cfg.c (gimple_duplicate_sese_region):
16941         Add update_dominance argument.
16942         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16943         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16944         gimple_duplicate_sese_region call.
16945         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16946         Update copy_bbs call.
16947         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16948         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16950 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16952         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16953         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
16954         vect_recog_rotate_pattern.
16955         (vect_recog_rotate_pattern): New function.
16957 2013-05-16  Jason Merrill  <jason@redhat.com>
16959         * Makefile.in (LLINKER): New variable.
16960         (mostlyclean): Remove link mutex.
16961         * configure.ac: Handle --enable-link-mutex.
16962         * lock-and-run.sh: New script.
16964 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16966         PR target/19599
16967         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
16968         for NULL decl.
16970 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16972         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
16974 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16976         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
16977         * config/arm/arm.c (next_consecutive_mem): New function.
16978         (gen_movmem_ldrd_strd): Likewise.
16979         * config/arm/arm.md (movmemqi): Update condition and code.
16980         (unaligned_loaddi, unaligned_storedi): New patterns.
16982 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16984         * config.gcc: Obsolete *-*-solaris2.9*.
16985         * doc/install.texi (Specific, *-*-solaris2*): Document it.
16987 2013-05-16  Richard Biener  <rguenther@suse.de>
16989         * passes.c (init_optimization_passes): Move pass_parallelize_loops
16990         earlier, after GRAPHITE transforms and IV canonicalization.
16992 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16994         * omp-low.c (extract_omp_for_data): For collapsed loops,
16995         if at least one of the loops is known at compile time to
16996         iterate zero times, set count to 0.
16997         (expand_omp_regimplify_p): New function.
16998         (expand_omp_for_generic): For collapsed loops, if at least
16999         one of the loops isn't known to iterate at least once,
17000         add runtime check with setting count to 0.
17001         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17002         For unsigned types if it isn't known at compile time that
17003         the loop will iterate at least once, add runtime check to bypass
17004         the whole loop if initial condition isn't true.
17006 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17008         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17010 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17012         PR middle-end/57286
17013         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17014         transformations to avoid an infinite loop.
17016 2013-05-16  Marek Polacek  <polacek@redhat.com>
17018         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17020 2013-05-15  Leif Ekblad  <leif@rdos.net>
17022         * config/i386/i386.c (ix86_decompose_address): Use
17023         DEFAULT_TLS_SEG_REG to access TLS segment register.
17024         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17025         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17026         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17028 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17030         PR target/57260
17031         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17032         sibling calls to functions that would normally be lazily bound,
17033         unless $gp is call-clobbered.
17035 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17037         * config/i386/i386.c (ix86_option_override_internal): Update
17038         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17039         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17040         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17041         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17042         of TARGET_3DNOW.
17043         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17045 2013-05-15  Andreas Schwab  <schwab@suse.de>
17047         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17048         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17049         third operand.
17051 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17053         * loop-unroll.c (report_unroll_peel): Check decision before
17054         emitting unroll/peel message.
17056 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17058         * function.h (has_bb_partition): New rtl_data flag.
17059         (bb_reorder_complete): Ditto.
17060         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17061         instead of flag_reorder_blocks_and_partition.
17062         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17063         with some enhancements.
17064         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17065         * bb-reorder.c (connect_traces): Check for has_bb_partition
17066         instead of flag_reorder_blocks_and_partition.
17067         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17068         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17069         verify_hot_cold_block_grouping.
17070         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17072 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17074         PR target/19599
17075         * config/arm/predicates.md (call_insn_operand): New predicate.
17076         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17077         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17078         if insn is not a tail call.
17079         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17080         registers.
17081         * config/arm/arm.h (enum reg_class): New caller save register class.
17082         (REG_CLASS_NAMES): Likewise.
17083         (REG_CLASS_CONTENTS): Likewise.
17084         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17085         without decls.
17087 2013-05-15  Richard Biener  <rguenther@suse.de>
17089         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17090         of MSG_OPTIMIZED_LOCATIONS.
17091         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17092         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17093         message.
17094         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17095         of MSG_OPTIMIZED_LOCATIONS.
17096         (execute_vect_slp): Likewise.
17097         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17098         (vect_create_cond_for_alias_checks): Likewise.
17099         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17100         (vect_recog_widen_mult_pattern): Likewise.
17101         (vect_recog_widen_sum_pattern): Likewise.
17102         (vect_recog_over_widening_pattern): Likewise.
17103         (vect_recog_widen_shift_pattern): Likewise.
17104         (vect_recog_vector_vector_shift_pattern): Likewise.
17105         (vect_recog_divmod_pattern): Likewise.
17106         (vect_recog_mixed_size_cond_pattern): Likewise.
17107         (vect_recog_bool_pattern): Likewise.
17108         (vect_pattern_recog_1): Likewise.
17110 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17112         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17113         non-functions to builtin_unreachable.
17114         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17115         correct when new direct edges were discovered.
17117 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17119         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17120         header, print symbol order instead of node uid, print more information
17121         about indirect edge targets.
17122         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17123         uids.
17124         (ipa_make_edge_direct_to_target): Likewise.
17125         (remove_described_reference): Likewise.
17126         (propagate_controlled_uses): Likewise.
17127         (ipa_print_node_params): Also print symbol order.
17128         (ipcp_transform_function): Print symbol order instead of node uids.
17129         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17130         (cgraph_get_create_real_symbol_node): Likewise.
17131         * ipa-cp.c (print_lattice): Likewise.
17132         (print_all_lattices): Likewise.
17133         (determine_versionability): Likewise.
17134         (initialize_node_lattices): Likewise.
17135         (estimate_local_effects): Likewise.
17136         (update_profiling_info): Likewise.
17137         (create_specialized_node): Likewise.
17138         (perhaps_add_new_callers): Likewise.
17139         (decide_about_value): Likewise.
17140         (decide_whether_version_node): Likewise.
17141         (identify_dead_nodes): Likewise.
17142         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17143         (dump_inline_summary): Likewise.
17144         (estimate_node_size_and_time): Likewise.
17145         (inline_analyze_function): Likewise.
17146         * ipa-inline.c (report_inline_failed_reason): Likewise.
17147         (want_early_inline_function_p): Likewise.
17148         (edge_badness): Likewise.
17149         (update_edge_key): Likewise.
17150         (inline_small_functions): Likewise.  Add dumping of order to two other
17151         dumps.
17152         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17153         instead of node uids.
17154         (propagate_pure_const): Likewise.
17155         (propagate_pure_const): Likewise.
17156         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17157         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17158         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17159         of node uids.
17160         * tree-pretty-print.c (dump_function_header): Likewise.
17161         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17162         Print symbol order instead of node uids.
17164 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17166         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17167         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17169 2013-05-15  Richard Biener  <rguenther@suse.de>
17171         PR tree-optimization/57275
17172         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17173         return value for fail to do runtime alias checks for gather loads.
17175 2013-05-15  Jan Hubicka  <jh@suse.cz>
17177         PR lto/57038
17178         PR lto/47375
17179         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17180         weakrefs are not external.
17181         (lto_symtab_merge_decls): Fix thinko when dealing with
17182         non-lto_symtab decls.
17183         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17184         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17185         * varpool.c (dump_varpool_node): Dump more flags.
17187 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17189         * config/i386/i386.c (processor_alias_table): Add instruction
17190         FSGSBASE for AMD bdver3 architecture.
17192 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17194         * tree.c (warn_deprecated_use): Print file:line using locus color.
17195         * diagnostic.c (diagnostic_report_current_module): Print file:line
17196         and file:line:column using locus color.
17198 2013-05-14  Mike Stump  <mikestump@comcast.net>
17200         * gdbinit.in: Add __null.
17202 2013-05-14  Mike Stump  <mikestump@comcast.net>
17204         * recog.h: Rename struct recog_data to Recog_data.
17205         * recog.c: Likewise.
17206         * reload.c (can_reload_into): Likewise.
17207         * config/picochip/picochip.c: Likewise.
17209 2013-05-14  Mike Stump  <mikestump@comcast.net>
17211         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17213 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17215         * resource.h (struct resources): Remove unch_memory member.
17216         (CLEAR_RESOURCE): Don't clear unch_memory.
17217         * resource.c (mark_referenced_resources): Don't set it.
17218         (mark_set_resources): Likewise.
17219         (mark_target_live_regs): Don't clear it.
17220         (init_resource_info): Likewise.
17221         * reorg.c (resource_conflicts_p): Don't compare it.
17222         (redundant_insn): Don't set it.
17224         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17225         Remove prototypes.
17226         * emit-rtl.c (next_label): Remove unused function.
17227         (skip_consecutive_labels, link_cc0_insns): Move to ...
17228         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17229         only place where these functions are used, and make them static.
17231 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17233         * fold-const.c (fold_negate_expr): Handle vectors.
17234         (fold_truth_not_expr): Make it static.
17235         (fold_invert_truthvalue): New static function.
17236         (invert_truthvalue_loc): Handle vectors. Do not call
17237         fold_truth_not_expr directly.
17238         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17239         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17240         (fold_comparison): Handle vector constants.
17241         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17242         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17243         * tree.h (fold_truth_not_expr): Remove declaration.
17245 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17247         * config/aarch64/aarch64-simd.md
17248         (aarch64_vcond_internal<mode>): Rename to...
17249         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17250         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17251         float modes. Clarify all iterator modes.
17252         (vcond<mode><mode>): Use new name for vcond expanders.
17253         (vcond<v_cmp_result><mode>): Likewise.
17254         (vcondu<mode><mode>: Likewise.
17255         * config/aarch64/iterators.md (VDQF_COND): New.
17257 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17259         PR bootstrap/57266
17260         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17261         variable for the shift amount. Check that we shift by non-negative
17262         amounts.
17264 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17266         PR target/42017
17267         * config/arm/arm.h (EPILOGUE_USES): Only return true
17268         for LR_REGNUM after epilogue_completed.
17270 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17272         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17273         is error_mark_node.
17275 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17277         PR target/57261
17278         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17279         and Solaris 11+/x86 with gld.
17280         * configure: Regenerate.
17282 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17284         * expmed.c (expand_shift_1): Canonicalize rotates by
17285         constant bitsize / 2 to bitsize - 1.
17286         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17287         case ROTATERT>: Likewise.
17289         Revert:
17290         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17292         * config/i386/i386.md (rotateinv): New code attr.
17293         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17294         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17295         roll $31, %eax, etc.
17297 2013-05-14  Richard Biener  <rguenther@suse.de>
17299         PR middle-end/57235
17300         * tree-eh.c (sink_clobbers): Give up for successors with
17301         multiple predecessors and no virtual uses.
17303 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17305         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17306         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17308 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17310         PR middle-end/57251
17311         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17312         the case when both op0 and op1 have VOIDmode.
17314 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17316         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17317         in multiply-accumulate mode.
17319 2013-05-13  Guozhi Wei  <carrot@google.com>
17321         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17323 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17325         PR target/56975
17326         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17327         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17328         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17329         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17330         (ix86_expand_prologue): Likewise.
17331         (ix86_expand_split_stack_prologue): Likewise.
17332         (legitimate_pic_address_disp_p): Likewise.
17333         (legitimize_pic_address): Likewise.
17334         (legitimize_tls_address): Likewise.
17335         (legitimize_pe_coff_symbol): Likewise.
17336         (output_pic_addr_const): Likewise.
17337         (construct_plt_address): Likewise.
17338         (ix86_expand_call): Likewise.
17339         (x86_output_mi_thunk): Likewise.
17340         (x86_function_profiler): Likewise.
17342 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17344         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17345         similar switch cases.
17346         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17347         (aarch64_simd_mov_to_<mode>low): Delete.
17348         (aarch64_simd_mov_to_<mode>high): Delete.
17349         (move_lo_quad_<mode>): Add w<-r alternative.
17350         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17351         (aarch64_simd_mov_from_*): Update type attribute.
17352         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17353         statement.
17355 2013-05-13  Jan Hubicka  <jh@suse.cz>
17357         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17358         * config/i386/i386.c (ix86_compute_frame_layout,
17359         ix86_expand_epilogue, emit_i387_cw_initialization,
17360         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17361         ix86_local_alignment): Fix use of size/speed predicates.
17363 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17365         PR tree-optimization/45216
17366         PR tree-optimization/57157
17367         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17368         the (-Y) & (B - 1) variant if OP is |.
17369         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17370         return shifted.  Use (-op1) & (prec - 1) as other_amount
17371         instead of prec - op1.
17373 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17375         PR middle-end/42371
17376         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17377         (ipa_constant_data): New type.
17378         (ipa_jump_func): Use ipa_constant_data to hold information about
17379         constant jump functions.
17380         (ipa_get_jf_constant): Adjust to jump function type changes.
17381         (ipa_get_jf_constant_rdesc): New function.
17382         (ipa_param_descriptor): New field controlled_uses.
17383         (ipa_get_controlled_uses): New function.
17384         (ipa_set_controlled_uses): Likewise.
17385         * ipa-ref.h (ipa_find_reference): Declare.
17386         * ipa-prop.c (ipa_cst_ref_desc): New type.
17387         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17388         changes.
17389         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17390         New parameter cs.  Adjust all callers.
17391         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17392         (remove_described_reference): New function.
17393         (jfunc_rdesc_usable): Likewise.
17394         (try_make_edge_direct_simple_call): Decrement controlled use count,
17395         attempt to remove reference if it hits zero.
17396         (combine_controlled_uses_counters): New function.
17397         (propagate_controlled_uses): Likewise.
17398         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17399         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17400         (ipa_print_node_params): Print described use counter.
17401         (ipa_write_jump_function): Adjust to jump function type changes.
17402         (ipa_read_jump_function): New parameter CS, pass it to
17403         ipa_set_jf_constant.  Adjust caller.
17404         (ipa_write_node_info): Stream controlled use count
17405         (ipa_read_node_info): Likewise.
17406         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17407         asserting.
17408         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17409         count.  Remove cloning-added reference if it reaches zero.
17410         * ipa-ref.c (ipa_find_reference): New function.
17412 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17414         * config/i386/i386.c (processor_target_table): Modified default
17415         alignment values for AMD BD and BT architectures.
17417 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17419         * tree-vect-generic.c (uniform_vector_p): Move ...
17420         * tree.c (uniform_vector_p): ... here.
17421         * tree.h (uniform_vector_p): Declare it.
17422         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17423         into a scalar.
17425 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17427         PR tree-optimization/57230
17428         * tree-ssa-strlen.c (handle_char_store): Record length for
17429         array store from STRING_CST.
17431         PR tree-optimization/57230
17432         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17433         check.
17435 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17437         * config/epiphany/epiphany.c (epiphany_init): Check size of
17438         NUM_MODES_FOR_MODE_SWITCHING.
17439         (epiphany_expand_prologue):
17440         Remove CONFIG_REGNUM initial value handling code.
17441         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17442         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17443         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17444         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17445         Don't return 1 for FP_MODE_NONE.
17446         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17447         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17448         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17449         * config/epiphany/epiphany.md (save_config): New pattern.
17451 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17453         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17455 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17457         * config/i386/i386.md (memory): Handle sseishft1.
17458         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17459         (*vec_extractv2di_1): Ditto.
17461 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17463         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17464         saved registers.
17466 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17468         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17469         Add mthumb/march=armv7-a multilib.
17470         Add mthumb/march=armv7-r multilib.
17471         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17473 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17475         * config/v850/t-rtems: Add more multilibs.
17477 2013-05-10  Richard Biener  <rguenther@suse.de>
17479         PR tree-optimization/57214
17480         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17481         not propagate from SSA names that occur in abnormal PHI nodes.
17483 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17485         * stor-layout.c (element_precision): New function.
17486         * machmode.h (element_precision): Declare it.
17487         * tree.c (build_minus_one_cst): New function.
17488         (element_precision): Likewise.
17489         * tree.h (build_minus_one_cst): Declare new function.
17490         (element_precision): Likewise.
17491         * fold-const.c (operand_equal_p): Use element_precision.
17492         (fold_binary_loc): Handle vector types.
17493         * convert.c (convert_to_integer): Use element_precision.
17494         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17495         separately.
17497 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17499         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17500         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17501         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17502         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17503         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17504         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17505         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17506         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17507         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17508         (Uuw8): New constraints.
17509         (Usb4): Move into alphabetical order.
17510         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17511         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17512         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17513         previously unnamed patterns.
17514         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17515         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17516         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17517         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17518         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17519         of specific lengths.
17521 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17523         * config/i386/i386.md (rotateinv): New code attr.
17524         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17525         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17526         roll $31, %eax, etc.
17528         PR tree-optimization/45216
17529         PR tree-optimization/57157
17530         * tree-ssa-forwprop.c (simplify_rotate): New function.
17531         (ssa_forward_propagate_and_combine): Call it.
17533 2013-05-10  Richard Biener  <rguenther@suse.de>
17535         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17536         disable peeling when we version for aliasing.
17537         (vector_alignment_reachable_p): Honor explicit user alignment.
17538         (vect_supportable_dr_alignment): Likewise.
17539         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17540         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17541         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17542         then peeling to arrange for the cost-model check to come first.
17544 2013-05-10  Alan Modra  <amodra@gmail.com>
17546         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17547         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17548         * configure: Regenerate.
17550 2013-05-10  Alan Modra  <amodra@gmail.com>
17552         PR target/55033
17553         * varasm.c (default_elf_select_section): Move !DECL_P check..
17554         (get_named_section): ..to here before calling get_section_name.
17555         Adjust assertion.
17556         (default_section_type_flags): Add DECL_P check.
17557         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17558         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17560 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17562         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17563         When using gen_stack_adjust_str with a register offset, add a
17564         REG_FRAME_RELATED_EXPR note.
17566 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17568         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17569         (*vec_extractv4si_zext_mem): Ditto.
17570         (*vec_extractv2di): Add 0->x and x->x alternatives.
17571         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17572         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17574 2013-05-09  Jason Merrill  <jason@redhat.com>
17576         N3639 C++1y VLA support
17577         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17578         DECL_VALUE_EXPR.
17580         * tree.c (build_constructor_va): New.
17581         * tree.h: Declare it.
17583 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17585         PR lto/57084
17586         * gimple-fold.c (canonicalize_constructor_val): Call
17587         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17589 2013-05-09  Jan Hubicka  <jh@suse.cz>
17590             Richard Biener  <rguenther@suse.de>
17592         PR lto/54095
17593         * symtab.c (symtab_make_decl_local): Do not add private names.
17595 2013-05-09  Jan Hubicka  <jh@suse.cz>
17597         PR lto/54095
17598         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17599         (unlink_from_assembler_name_hash): Likewise.
17600         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17601         symtab_unregister_node, symtab_initialize_asm_name_hash,
17602         change_decl_assembler_name): Update.
17604 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17606         * config/aarch64/aarch64.md: New movtf split.
17607         (*movtf_aarch64): Update.
17608         (aarch64_movdi_tilow): Handle TF modes and rename to
17609         aarch64_movdi_<mode>low.
17610         (aarch64_movdi_tihigh): Handle TF modes and rename to
17611         aarch64_movdi_<mode>high
17612         (aarch64_movtihigh_di): Handle TF modes and rename to
17613         aarch64_mov<mode>high_di
17614         (aarch64_movtilow_di): Handle TF modes and rename to
17615         aarch64_mov<mode>low_di
17616         (aarch64_movtilow_tilow): Remove spurious whitespace.
17617         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17618         splits.
17619         (aarch64_print_operand): Update.
17621 2013-05-09  Alan Modra  <amodra@gmail.com>
17623         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17624         powerpc64le.
17625         * configure: Regenerate.
17627 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17629         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17630         splitter preparation statements.
17631         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17632         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17633         adjust_address_nv.
17635 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17637         * gimple-ssa-strength-reduction.c (count_candidates): Change
17638         return value to int.
17639         (analyze_candidates_and_replace): Change type of length to int.
17641 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17643         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17644         (*vec_extract<mode>): Use VI12_128 mode iterator.
17645         (*vec_extract<mode>_mem): Ditto.
17646         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17647         attribute.
17649 2013-05-08  Diego Novillo  <dnovillo@google.com>
17651         PR bootstrap/54659
17653         Revert:
17654         2012-08-17  Diego Novillo  <dnovillo@google.com>
17656         PR bootstrap/54281
17657         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17658         * config.in: Regenerate.
17659         * configure: Regenerate.
17660         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17662 2013-05-08  Jan Hubicka  <jh@suse.cz>
17664         PR lto/54095
17665         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17666         * cgraph.h (symtab_node_base): Add unique_name.
17667         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17668         input_overwrite_node, input_varpool_node): Stream unique_name.
17669         * cgraphclones.c (cgraph_create_virtual_clone,
17670         cgraph_function_versioning): Set unique_name.
17671         * ipa.c (function_and_variable_visibility): Set unique_name.
17673 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17675         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17676         (alloc_cand_and_find_basis): Restrict conditional candidate
17677         processing to CAND_MULTs.
17679 2013-05-08  Jan Hubicka  <jh@suse.cz>
17681         PR lto/54095
17682         lto-symtab.c (lto_symtab_symbol_p): New function.
17683         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17684         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17685         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17686         Skip static symbols.
17688 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17690         PR tree-optimization/57200
17691         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17692         Only call inform if the preceding warning_at returns true.
17694 2013-05-07  Han Shen  <shenhan@google.com>
17696         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17697         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17698         * common.opt (fstack-protector-strong): New option.
17699         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17700         * doc/invoke.texi (Optimization Options): Document
17701         "-fstack-protector-strong".
17702         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17704 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17706         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17708 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17710         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17711         it is not lhs of the insn.
17712         (setup_reg_equiv): Remove insn having equiv memory which it is not
17713         lhs of the insn.
17714         * lra-constraints.c (process_address): Try to improve generation
17715         code for address base + disp.
17716         (lra_constraints): Make correct the code for checking insn setting
17717         up backward equivalence.  Remove insn only if it is in the init
17718         insn list.
17719         * lra-eliminations.c (update_reg_eliminate): Change return value.
17720         (lra_eliminate): Use the result.
17722 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17724         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17725         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17726         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17727         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17728         register target operands.
17729         (*vec_extractv8hi_sse2): New pattern.
17730         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17731         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17732         (*vec_extract<mode>_mem): New insn and split pattern.
17734 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17736         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17737         (TARGET_ASAN_SHADOW_OFFSET): Define.
17738         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17739         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17741 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17743         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17744         (incr_vec_index): Return -1 if increment not found.
17745         (create_add_on_incoming_edge): Assert if increment not found.
17746         (record_increment): Limit number of increments recorded.
17747         (all_phi_incrs_profitable): Return false if an increment not found.
17748         (replace_profitable_candidates): Don't process increments that were
17749         not recorded.
17750         (analyze_candidates_and_replace): Limit size of incr_vec.
17752 2013-05-07  Richard Biener  <rguenther@suse.de>
17754         * calls.c (special_function_p): setjmp-like functions are leaf.
17755         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17756         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17758 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17760         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17761         (aarch64_simd_mov<mode>): New expander.
17762         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17763         (aarch64_simd_mov_to_<mode>high): Likewise.
17764         (aarch64_simd_mov_from_<mode>low): Likewise.
17765         (aarch64_simd_mov_from_<mode>high): Likewise.
17766         (aarch64_dup_lane<mode>): Update.
17767         (aarch64_dup_lanedi): New instruction pattern.
17768         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17769         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17771 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17773         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17774         (replace_mult_candidate): Remove unnecessary argument; remove
17775         unnecessary parameter from call to introduce_cast_before_cand.
17776         (replace_unconditional_candidate): Remove unnecessary parameter
17777         from call to replace_mult_candidate.
17778         (replace_conditional_candidate): Likewise.
17779         (insert_initializers): Use make_temp_ssa_name.
17780         (introduce_cast_before_cand): Remove unnecessary argument; use
17781         make_temp_ssa_name.
17782         (replace_one_candidate): Remove unnecessary argument; remove
17783         unnecessary parameter from calls to introduce_cast_before_cand.
17784         (replace_profitable_candidates): Remove unnecessary parameters
17785         from calls to replace_one_candidate.
17787 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17789         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17790         phi def as possibly hiding a basis for a CAND_ADD whose operands
17791         have been commuted in the analysis.
17792         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17794 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17796         * config/aarch64/aarch64.md
17797         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17798         shift value between 0-4.
17800 2013-05-07  Richard Biener  <rguenther@suse.de>
17802         * double-int.h (rshift): New overload.
17803         * double-int.c (rshift): New function.
17804         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17805         (create_reference_ops_from_ref): Remove.
17806         (vn_reference_insert): Use shared ops for constructing the
17807         reference and copy it.
17809 2013-05-07  Richard Biener  <rguenther@suse.de>
17811         PR middle-end/57190
17812         * tree-eh.c (sink_clobbers): Properly propagate
17813         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17815 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17817         PR tree-optimization/57149
17818         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17819         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17820         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17821         uninit_undefined_value_p instead of ssa_undefined_value_p.
17823         PR debug/57184
17824         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17825         for modifier == EXPAND_INITIALIZER.
17827 2013-05-07  Anton Blanchard  <anton@samba.org>
17829         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17830         for powerpc64 little endian.
17831         * configure: Regenerate.
17833 2013-05-06  Graham Stott  <grahams@btinternet.com>
17835         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17836         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17837         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17838         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17840 2013-05-06  Graham Stott  <grahams@btinternet.com>
17842         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17843         codes which allow non-lvalues.
17845 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17847         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17848         components are all 1s.
17849         (integer_minus_onep): New function.
17850         * tree.h (integer_minus_onep): Declare it.
17851         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17852         integer_minus_onep instead of integer_all_onesp.
17854 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17856         PR target/52933
17857         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17858         variations of these patterns.
17860 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17862         * config/i386/i386.md (isa): Add x64_sse4 member.
17863         (enabled): Handle x64_sse4.
17864         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17865         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17866         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17867         instruction for SSE4_1 targets.  Update insn attributes.
17868         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17869         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17870         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17871         const_1 selector.
17872         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17873         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17874         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17876 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17878         PR target/57108
17879         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17881 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17883         * final.c (do_assembler_dialects): Don't handle curly braces and
17884         vertical bar escaped by % as dialect delimiters.
17885         (output_asm_insn): Print curly braces and vertical bar if escaped
17886         by % and ASSEMBLER_DIALECT defined.
17887         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17888         * doc/tm.texi: Regenerated.
17890 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17892         * config/mips/mips.c: Include tree-pass.h.
17893         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17894         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17895         (pass_mips_machine_reorg2): New machine specific pass.
17896         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17897         (mips_option_override): Register the new pass.
17898         * rtl.h (cleanup_barriers): Remove prototype.
17899         (dbr_schedule): Likewise.
17900         * jump.c (cleanup_barriers): Make static.
17901         * reorg.c (dbr_schedule): Likewise.
17903 2013-05-06  Richard Biener  <rguenther@suse.de>
17905         PR tree-optimization/57185
17906         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17907         SSA names properly.
17909 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17911         PR target/57106
17912         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17913         between "first_arg" and "insn", not "last" and "insn".
17915 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17917         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17918         (find_candidates_in_block): Re-enable slsr_process_phi.
17919         (create_phi_basis): Fix double counting of candidate adjustment.
17921 2013-05-06  Richard Biener  <rguenther@suse.de>
17923         PR middle-end/57147
17924         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17925         the edge is also fallthru, preserve it and just clear the
17926         abnormal flag.
17927         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17928         also complex, preserve that and just clear the fallthru flag.
17929         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17930         update virtual operands.
17932 2013-05-06  Alan Modra  <amodra@gmail.com>
17934         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17935         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17936         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17937         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17938         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17939         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17941 2013-05-06  Alan Modra  <amodra@gmail.com>
17943         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17944         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17945         (DEFAULT_ASM_ENDIAN): Define.
17946         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17947         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17948         Update -K PIC clause from sysv4.h.
17949         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17950         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17952 2013-05-06  Alan Modra  <amodra@gmail.com>
17954         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
17955         twice for little-endian.
17956         (ashrdi3_no_power, ashrdi3): Support little-endian.
17958 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17960         PR target/55303
17961         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
17962         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
17963         related expanders.
17964         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
17965         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
17966         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
17967         New predicates.
17969 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
17970             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17972         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
17973         * config/pa/pa.opt: Make mbig-switch a no-op.
17974         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
17975         (CASE_VECTOR_MODE): Always return SImode.
17976         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
17977         for the !TARGET_BIG_SWITCH case.
17978         * config/pa/pa-linux.h: Likewise.
17979         * config/pa/pa-openbsd.h: Likewise.
17980         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
17981         * config/pa/pa.md (short_jump): Remove define_insn.
17982         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
17983         (casesi0): Remove define_insn.
17984         (type): Remove btable_branch.
17985         (pa_combine_type): Likewise.
17986         (in_nullified_branch_delay): Likewise.
17987         (in_call_delay): Likewise.
17988         (define_delay): Likewise.
17989         (define_insn_reservation "Z3"): Likewise.
17990         (define_insn_reservation "Z4"): Likewise.
17991         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
17992         (pa_adjust_insn_length): Remove adjustment for btable branches.
17993         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
17994         and mno-big-switch
17996 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
17998         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17999         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18000         Add m->r,x alternatives.
18001         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18002         splitters using SWI48x mode iterator.
18003         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18004         TARGET_64BIT.  Add m->x alternative.
18005         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18006         Add o->x alternative.  Enable for TARGET_SSE.
18007         (sse_storeq): Remove expander.
18008         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18009         with memory input operand.
18010         (*vec_extractv2di_1 splitter): New.
18011         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18012         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18014 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18016         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18017         (INT_LOWPART): Delete.
18018         (extract_MB): Adjust.
18019         (extract_ME): Adjust.
18020         (print_operand): Adjust.
18022 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18024         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18025         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18026         (reg_or_logical_cint_operand, easy_fp_constant,
18027         logical_const_operand): Delete "CONST_DOUBLE" case.
18028         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18029         "HOST_BITS_PER_WIDE_INT == 64" test.
18030         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18031         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18032         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18033         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18034         test.
18035         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18036         CONST_DOUBLE DImode/VOIDmode case.
18037         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18038         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18039         CONST_DOUBLE VOIDmode case.
18040         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18041         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18042         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18043         Delete CONST_DOUBLE case.
18044         (splitters for mov FMOVE64 const_double): Delete
18045         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18046         "HOST_BITS_PER_WIDE_INT >= 64" test.
18047         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18048         case.
18049         (mov DI const_double): Delete.
18051 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18053         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18054         on op shows all bits zero in mode of a lowpart subreg, return zero.
18056 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18058         PR target/57150
18059         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18060         to save TFmode registers and DImode to save TImode registers for
18061         caller save operations.
18062         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18063         mark being partially clobbered since they only use the first
18064         double word.
18066         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18067         and TDmode only use the upper 64-bits of each VSX register.
18069 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18071         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18072         (find_candidates_in_block): Disable slsr_process_phi.
18074 2013-05-03  Guozhi Wei  <carrot@google.com>
18076         * coverage.c (coverage_obj_init): Move the construction of gcov
18077         constructor to ...
18078         (build_init_ctor): ... here.
18080 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18082         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18083         (slsr_cand_d): Redefine def_phi.
18084         (stride_status, phi_adjust_status, count_phis_status): New enums.
18085         (find_phi_def): New.
18086         (find_basis_for_base_expr): New.
18087         (find_basis_for_candidate): Handle hidden bases.
18088         (alloc_cand_and_find_basis): Handle phi candidates.
18089         (slsr_process_phi): New.
18090         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18091         (create_mul_imm_cand): Likewise.
18092         (create_add_ssa_cand): Exclude phi base candidates.
18093         (create_add_imm_cand): Likewise.
18094         (slsr_process_cast): Likewise.
18095         (slsr_process_copy): Likewise.
18096         (find_candidates_in_block): Handle phi candidates.
18097         (dump_candidate): Likewise.
18098         (unconditional_cands): Delete.
18099         (unconditional_cands_with_known_stride_p): Delete.
18100         (phi_dependent_cand_p): New.
18101         (cand_increment): Handle phi-dependent candidates.
18102         (replace_dependent): Delete.
18103         (replace_mult_candidate): New.
18104         (replace_unconditional_candidate): New.
18105         (incr_vec_index): Move to avoid forward reference.
18106         (create_add_on_incoming_edge): New.
18107         (create_phi_basis): New.
18108         (replace_dependents): Delete.
18109         (replace_conditional_candidate): New.
18110         (phi_add_costs): New.
18111         (replace_uncond_cands_and_profitable_phis): New.
18112         (record_increment): Handle phi adjustments.
18113         (record_phi_increments): New.
18114         (record_increments): Handle phi adjustments.
18115         (phi_incr_cost): New.
18116         (lowest_cost_path): Handle phis.
18117         (total_savings): Likewise.
18118         (analyze_increments): Likewise.
18119         (ncd_with_phi): New.
18120         (ncd_of_cand_and_phis): New.
18121         (nearest_common_dominator_for_cands): Handle phi increments.
18122         (all_phi_incrs_profitable): New.
18123         (replace_profitable_candidates): Handle phi-dependent candidates.
18124         (analyze_candidates_and_replace): Likewise.
18126 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18128         PR bootstrap/57154
18129         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18130         do not exceed REG_BR_PROB_BASE.
18132 2013-05-03  Jeff Law  <law@redhat.com>
18134         PR tree-optimization/57144
18135         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18136         operand of the condition will bit into the new type when eliminating
18137         a cast feeding a condition.
18139 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18141         PR rtl-optimization/57130
18142         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18143         of COMPARE as in_code to the recursive call if needed.
18145 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18147         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18148         (enabled): Handle new members.
18149         * config/i386/sse.md (*vec_concatv2si): Merge from
18150         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18151         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18153 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18155         PR tree-optimization/57027
18156         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18157         for fnms opportunity, check we got the prerequisite kind
18158         of tree / gimple before using accessor functions.
18160 2013-05-03  Richard Biener  <rguenther@suse.de>
18162         * double-int.h (lshift): New overload without precision
18163         and arith argument.
18164         (operator *=, operator +=, operator -=): Move ...
18165         * double-int.c (operator *=, operator +=, operator -=): ... here
18166         and implement more efficiently.
18167         (mul_double_with_sign): Remove.
18168         (lshift_double): Adjust to take unsinged shift argument, push
18169         dispatching code to callers.
18170         (mul_double_wide_with_sign): Add early out for callers that
18171         are not interested in high parts or overflow.
18172         (lshift): New function.
18173         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18174         dispatch code here.
18175         (lrotate, rrotate): Use logical shifts.
18176         * expr.c (get_inner_reference): Use lshift.
18177         * fixed-value.c (do_fixed_divide): Likewise.
18178         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18179         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18180         (indirect_refs_may_alias_p): Likewise.
18181         (stmt_kills_ref_p_1): Likewise.
18183 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18185         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18187 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18189         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18190         scalar form of FABD instruction.
18192 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18194         * lra-constraints.c (process_alt_operands): Add checking alt
18195         number to choose the best alternative.
18197 2013-05-02  Richard Biener  <rguenther@suse.de>
18199         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18200         bitmap and its handling.
18201         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18203 2013-05-02  Richard Biener  <rguenther@suse.de>
18205         PR middle-end/57140
18206         * tree-inline.c (copy_loops): Properly handle removed loops.
18207         (copy_cfg_body): Mark destination loops for fixup if source
18208         loops needed fixup.
18210 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18212         PR target/56732
18213         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18214         generating simple_return for naked functions.
18216 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18218         PR middle-end/56988
18219         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18220         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18221         flags match.
18222         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18223         ipa_agg_replacement_value structures.
18224         (known_aggs_to_agg_replacement_list): Likewise.
18225         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18226         (read_agg_replacement_chain): Likewise.
18227         (ipcp_transform_function): Also check that by_ref flags match.
18229 2013-05-02  Richard Biener  <rguenther@suse.de>
18231         * graphds.h (struct graph): Add obstack member.
18232         * graphds.c (new_graph): Initialize obstack and allocate
18233         vertices from it.
18234         (add_edge): Allocate edge from the obstack.
18235         (free_graph): Free the obstack instead of all edges and vertices.
18237 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18239         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18240         divides.
18241         * cfg.c (update_bb_profile_for_threading): Ditto.
18242         * tree-inline.c (copy_bb): Ditto.
18243         (copy_edges_for_bb): Ditto.
18244         (initialize_cfun): Ditto.
18245         (copy_cfg_body): Ditto.
18246         (expand_call_inline): Ditto.
18247         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18248         (estimate_node_size_and_time): Ditto.
18249         (inline_merge_summary): Ditto.
18250         * cgraphclones.c (cgraph_clone_edge): Ditto.
18251         (cgraph_clone_node): Ditto.
18252         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18253         (compute_trg_info): Ditto.
18255 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18257         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18258         S reg when fp attribute set.
18259         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18261 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18263         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18264         New pattern.
18265         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18266         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18267         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18269 2013-05-02  Richard Biener  <rguenther@suse.de>
18271         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18272         (struct instantiate_cache_entry): New type.
18273         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18274         (struct instantiate_cache_type): New type.
18275         (set_instantiated_value, get_instantiated_value): Remove.
18276         (get_instantiated_value_entry): New function.
18277         (instantiate_scev_name): Use the new cache and adjust.
18278         (instantiate_scev_poly): Adjust.
18279         (instantiate_scev_binary): Likewise.
18280         (instantiate_array_ref): Likewise.
18281         (instantiate_scev_convert): Likewise.
18282         (instantiate_scev_not): Likewise.
18283         (instantiate_scev_3): Likewise.
18284         (instantiate_scev_2): Likewise.
18285         (instantiate_scev_r): Likewise.
18286         (instantiate_scev): Likewise.
18287         (resolve_mixers): Likewise.
18289 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18291         PR target/57091
18292         * lra-constraints.c (best_small_class_operands_num): Remove.
18293         (process_alt_operands): Remove small_class_operands_num.  Take
18294         small classes operands into losers and only if the operand is not
18295         matched.  Modify debugging output.
18296         (curr_insn_transform): Remove best_small_class_operands_num.
18297         Print insn name.
18299 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18301         * config/aarch64/aarch64-builtins.c
18302         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18303         * config/aarch64/aarch64-simd-builtins.def
18304         (reduc_splus_): Add new modes.
18305         (reduc_uplus_): New.
18306         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18307         (reduc_uplus_v4sf): Likewise.
18308         (reduc_splus_v4sf): Likewise.
18309         (aarch64_addv<mode>): Likewise.
18310         (reduc_uplus_<mode>): Likewise.
18311         (reduc_splus_<mode>): Likewise.
18312         (aarch64_addvv2di): Likewise.
18313         (reduc_uplus_v2di): Likewise.
18314         (reduc_splus_v2di): Likewise.
18315         (aarch64_addvv2si): Likewise.
18316         (reduc_uplus_v2si): Likewise.
18317         (reduc_splus_v2si): Likewise.
18318         (reduc_<sur>plus_<mode>): New.
18319         (reduc_<sur>plus_v2di): Likewise.
18320         (reduc_<sur>plus_v2si): Likewise.
18321         (reduc_<sur>plus_v4sf): Likewise.
18322         (aarch64_addpv4sf): Likewise.
18323         * config/aarch64/arm_neon.h
18324         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18325         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18326         add UNSPEC_SADDV, UNSPEC_UADDV.
18327         (SUADDV): New.
18328         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18330 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18332         * config/aarch64/arm_neon.h
18333         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18335 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18337         * config/aarch64/aarch64-builtins
18338         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18340 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18342         * config/aarch64/aarch64-simd-builtins.def
18343         (reduc_smax_): New.
18344         (reduc_smin_): Likewise.
18345         (reduc_umax_): Likewise.
18346         (reduc_umin_): Likewise.
18347         (reduc_smax_nan_): Likewise.
18348         (reduc_smin_nan_): Likewise.
18349         (fmax): Remove.
18350         (fmin): Likewise.
18351         (smax): Update for V2SF, V4SF and V2DF modes.
18352         (smin): Likewise.
18353         (smax_nan): New.
18354         (smin_nan): Likewise.
18355         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18356         (<su><maxmin><mode>3): ...This, refactor.
18357         (s<maxmin><mode>3): New.
18358         (<maxmin_uns><mode>3): Likewise.
18359         (reduc_<maxmin_uns>_<mode>): Refactor.
18360         (reduc_<maxmin_uns>_v4sf): Likewise.
18361         (reduc_<maxmin_uns>_v2si): Likewise.
18362         (aarch64_<fmaxmin><mode>: Remove.
18363         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18364         new builtin names.
18365         (vmin<q>_f<32,64>): Likewise.
18366         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18367         (FMAXMIN): New.
18368         (su): Add mappings for smax, smin, umax, umin.
18369         (maxmin): New.
18370         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18371         (FMAXMIN): Rename as...
18372         (FMAXMIN_UNS): ...This.
18373         (maxminv): Remove.
18374         (fmaxminv): Likewise.
18375         (fmaxmin): Likewise.
18376         (maxmin_uns): New.
18377         (maxmin_uns_op): Likewise.
18379 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18381         * config/aarch64/arm_neon.h
18382         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18383         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18384         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18386 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18388         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18389         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18391 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18393         * config/aarch64/aarch64-simd.md
18394         (vcond<mode>_internal): Handle special cases for constant masks.
18395         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18396         (vcondu<mode><mode>): Likewise.
18397         (vcond<v_cmp_result><mode>): New.
18399 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18401         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18402         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18403         * config/aarch64/aarch64-simd-builtins.def
18404         (cmeq): Update to BUILTIN_VALLDI.
18405         (cmgt): Likewise.
18406         (cmge): Likewise.
18407         (cmle): Likewise.
18408         (cmlt): Likewise.
18409         * config/aarch64/arm_neon.h
18410         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18411         to builtins or C as appropriate.
18413 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18415         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18416         (cmgeu): ...This.
18417         (cmhi): Rename to...
18418         (cmgtu): ...This.
18419         * config/aarch64/aarch64-simd.md
18420         (simd_mode): Add SF.
18421         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18422         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18423         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18424         (cstore<mode>_neg): ...This.
18425         * config/aarch64/iterators.md
18426         (VALLF): new.
18427         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18428         (COMPARISONS): New.
18429         (UCOMPARISONS): Likewise.
18430         (optab): Add missing comparisons.
18431         (n_optab): New.
18432         (cmp_1): Likewise.
18433         (cmp_2): Likewise.
18434         (CMP): Likewise.
18435         (cmp): Remove.
18436         (VCMP_S): Likewise.
18437         (VCMP_U): Likewise.
18438         (V_cmp_result): Add DF, SF modes.
18439         (v_cmp_result): Likewise.
18440         (v): Likewise.
18441         (vmtype): Likewise.
18442         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18444 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18446         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18447         define_insn to define_insn_and_split.
18448         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18449         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18450         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18451         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18452         (thumb2_negscc): Likewise.
18454 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18456         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18458 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18460         * config/arm/thumb2.md: Remove trailing whitespaces.
18462 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18464         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18465         Use gen_int_mode rather than GEN_INT.
18467 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18469         * value-prof.c (stream_in_histogram_value): Remove the strayed
18470         debug_gimple_stmt.
18472 2013-04-30  Richard Biener  <rguenther@suse.de>
18474         PR middle-end/57122
18475         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18477 2013-04-30  Richard Biener  <rguenther@suse.de>
18479         PR middle-end/57107
18480         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18482 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18484         PR rtl-optimization/56957
18485         PR rtl-optimization/57105
18486         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18487         variable.  Use just INSN_UID for determining whether an insn
18488         should be only disconnected from the insn stream.
18489         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18491 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18493         PR tree-optimization/57104
18494         * tsan.c (instrument_expr): Don't instrument accesses to
18495         DECL_HARD_REGISTER VAR_DECLs.
18497 2013-04-30  Richard Biener  <rguenther@suse.de>
18499         * function.h (loops_for_fn): New inline function.
18500         (set_loops_for_fn): Likewise.
18501         * cfgloop.h (place_new_loop): Add struct function parameter.
18502         (get_loop): Likewise.
18503         (get_loops): Likewise.
18504         (number_of_loops): Likewise.
18505         (fel_next): Adjust.
18506         (fel_init): Likewise.
18507         * cfg.c (get_loop_copy): Adjust.
18508         * cfgloop.c (flow_loops_dump): Likewise.
18509         (record_loop_exits): Likewise.
18510         (verify_loop_structure): Likewise.
18511         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18512         (estimate_reg_pressure_cost): Likewise.
18513         (mark_loop_exit_edges): Likewise.
18514         * cfgloopmanip.c (place_new_loop): Likewise.
18515         (add_loop): Likewise.
18516         (duplicate_loop): Likewise.
18517         * graph.c (draw_cfg_nodes): Likewise.
18518         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18519         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18520         (extract_affine_chrec): Likewise.
18521         (build_scop_iteration_domain): Likewise.
18522         * graphite.c (graphite_initialize): Likewise.
18523         * ira-build.c (create_loop_tree_nodes): Likewise.
18524         (more_one_region_p): Likewise.
18525         (rebuild_regno_allocno_maps): Likewise.
18526         (mark_loops_for_removal): Likewise.
18527         (mark_all_loops_for_removal): Likewise.
18528         (remove_unnecessary_regions): Likewise.
18529         (ira_build): Likewise.
18530         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18531         * loop-init.c (fix_loop_structure): Likewise.
18532         (gate_rtl_move_loop_invariants): Likewise.
18533         (gate_rtl_unswitch): Likewise.
18534         (gate_rtl_unroll_and_peel_loops): Likewise.
18535         (rtl_doloop): Likewise.
18536         * lto-streamer-in.c (input_cfg): Likewise.
18537         * lto-streamer-out.c (output_cfg): Likewise.
18538         * modulo-sched.c (sms_schedule): Likewise.
18539         * predict.c (tree_estimate_probability): Likewise.
18540         (tree_estimate_probability_driver): Likewise.
18541         (estimate_loops): Likewise.
18542         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18543         (move_sese_region_to_fn): Likewise.
18544         (debug_loop_num): Likewise.
18545         * tree-chrec.c (chrec_evaluate): Likewise.
18546         (hide_evolution_in_other_loops_than_loop): Likewise.
18547         (chrec_component_in_loop_num): Likewise.
18548         (reset_evolution_in_loop): Likewise.
18549         (evolution_function_is_invariant_rec_p): Likewise.
18550         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18551         * tree-inline.c (copy_loops): Likewise.
18552         (copy_cfg_body): Likewise.
18553         (tree_function_versioning): Likewise.
18554         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18555         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18556         Likewise.
18557         (add_to_evolution_1): Likewise.
18558         (scev_const_prop): Likewise.
18559         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18560         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18561         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18562         (tree_ssa_lim_initialize): Likewise.
18563         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18564         (verify_loop_closed_ssa): Likewise.
18565         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18566         (tree_ssa_loop_im): Likewise.
18567         (tree_ssa_loop_unswitch): Likewise.
18568         (tree_vectorize): Likewise.
18569         (check_data_deps): Likewise.
18570         (tree_ssa_loop_ivcanon): Likewise.
18571         (tree_ssa_loop_bounds): Likewise.
18572         (tree_complete_unroll): Likewise.
18573         (tree_complete_unroll_inner): Likewise.
18574         (tree_parallelize_loops): Likewise.
18575         (tree_ssa_loop_prefetch): Likewise.
18576         (tree_ssa_loop_ivopts): Likewise.
18577         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18578         * tree-vectorizer.c (vectorize_loops): Likewise.
18580 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18582         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18583         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18584         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18585         with EABI_LINK_SPEC.
18587 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18589         PR target/44578
18590         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18591         alternative.
18593 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18595         PR target/57097
18596         * lra-constraints.c (process_alt_operands): Discourage a bit more
18597         using memory for pseudos.  Print cost dump for alternatives.
18598         Modify cost values for conflicts with early clobbers.
18599         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18601 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18603         PR target/57098
18604         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18606 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18608         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18609         from/to S register.
18610         (movdi_aarch64): Support LDR/STR from/to D register.
18612 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18614         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18615         or higher by default.
18617 2013-04-29  Richard Biener  <rguenther@suse.de>
18619         PR middle-end/57075
18620         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18621         even if not adding abnormal edges for calls that can make
18622         abnormal gotos.
18624 2013-04-29  Richard Biener  <rguenther@suse.de>
18626         PR middle-end/57103
18627         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18628         TREE_BLOCK.
18629         (move_stmt_r): Remove redundant checking.
18631 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18633         PR bootstrap/57077
18634         * basic-block.h (apply_scale): New function.
18635         (apply_probability): Use apply_scale.
18636         * gimple-streamer-in.c (input_bb): Ditto.
18637         * lto-streamer-in.c (input_cfg): Ditto.
18638         * lto-cgraph.c (merge_profile_summaries): Ditto.
18639         * tree-optimize.c (execute_fixup_cfg): Ditto.
18640         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18641         (copy_edges_for_bb): Ditto.
18642         (copy_cfg_body): Ditto.
18644 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18646         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18647         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18648         (tail_merge_optimize): Handle current_loops == NULL.
18650 2013-04-26  Jeff Law  <law@redhat.com>
18652         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18653         (simplify_cond_using_ranges): Generalize code to simplify
18654         COND_EXPRs where one argument is a constant and the other
18655         is an SSA_NAME created by an integral type conversion.
18657 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18659         * config/arm/arm.md (store_minmaxsi): Use only when
18660         optimize_insn_for_size_p.
18662 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18664         PR target/57108
18665         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18667 2013-04-29  Richard Biener  <rguenther@suse.de>
18669         PR middle-end/57089
18670         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18671         loop tree make sure to schedule a fixup for the child as well.
18672         (expand_omp_for_generic): Properly add loops.
18673         (expand_omp_for_static_nochunk): Likewise.
18674         (expand_omp_for_static_chunk): Likewise.
18675         (expand_omp_for): For the degenerate case fixup loops.
18676         (expand_omp_sections): Fix default bb placement in loops.
18677         (expand_omp_atomic_pipeline): Properly add loops.
18679 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18681         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18683 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18685         * tree-ssa-tail-merge.c: Update header comment.
18687 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18689         * config/aarch64/arm_neon.h
18690         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18691         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18692         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18693         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18694         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18695         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18696         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18697         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18699 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18701         * config/aarch64/aarch64-simd.md
18702         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18703         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18704         fix_trunc, fixuns_trunc.
18705         (ftrunc<VDQF:mode>2): New.
18706         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18707         (fix_trunc_optab): New.
18709 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18711         * config/aarch64/aarch64-builtins.c
18712         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18713         iceilf, lround, iroundf.
18715 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18717         PR target/54349
18718         * config/i386/i386.h (enum ix86_tune_indices)
18719         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18720         New, split from X86_TUNE_INTER_UNIT_MOVES.
18721         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18722         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18723         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18724         (TARGET_INTER_UNIT_MOVES): Remove.
18725         * config/i386/i386.c (initial_ix86_tune_features): Update.
18726         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18727         (ix86_expand_convert_uns_didf_sse): Use
18728         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18729         (ix86_expand_vector_init_one_nonzero): Ditto.
18730         (ix86_expand_vector_init_interleave): Ditto.
18731         (inline_secondary_memory_needed): Return true for moves from SSE class
18732         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18733         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18734         * config/i386/constraints.md (Yi, Ym): Depend on
18735         TARGET_INTER_UNIT_MOVES_TO_VEC.
18736         (Yj, Yn): New constraints.
18737         * config/i386/i386.md (*movdi_internal): Change constraints of
18738         operand 1 from Yi to Yj and from Ym to Yn.
18739         (*movsi_internal): Ditto.
18740         (*movdf_internal): Ditto.
18741         (*movsf_internal): Ditto.
18742         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18743         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18744         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18745         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18746         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18747         * config/i386/sse.md (movdi_to_sse): Ditto.
18748         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18749         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18750         TARGET_INTER_UNIT_MOVES.
18751         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18752         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18753         instead of TARGET_INTER_UNIT_MOVES.
18754         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18755         operand 1 from Yi to Yj and from Ym to Yn.
18757 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18759         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18760         (float_truncate_hi_): Likewise.
18761         (float_extend_lo_): Likewise.
18762         (float_truncate_lo_): Likewise.
18763         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18764         (aarch64_float_extend_lo_v2df): Likewise.
18765         (vec_unpacks_hi_v4sf): Likewise.
18766         (aarch64_float_truncate_lo_v2sf): Likewise.
18767         (aarch64_float_truncate_hi_v4sf): Likewise.
18768         (vec_pack_trunc_v2df): Likewise.
18769         (vec_pack_trunc_df): Likewise.
18771 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18773         * config/aarch64/aarch64-builtins.c
18774         (aarch64_fold_builtin): Fold float conversions.
18775         * config/aarch64/aarch64-simd-builtins.def
18776         (floatv2si, floatv4si, floatv2di): New.
18777         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18778         * config/aarch64/aarch64-simd.md
18779         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18780         * config/aarch64/iterators.md (FLOATUORS): New.
18781         (optab): Add float, floatuns.
18782         (su_optab): Likewise.
18784 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18786         * config/aarch64/aarch64-builtins.c
18787         (aarch64_builtin_vectorized_function): Use new names for
18788         fcvt builtins.
18789         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18790         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18791         (fcvtzu): Split as...
18792         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18793         (fcvtas): Split as...
18794         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18795         (fcvtau): Split as...
18796         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18797         (fcvtps): Split as...
18798         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18799         (fcvtpu): Split as...
18800         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18801         (fcvtms): Split as...
18802         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18803         (fcvtmu): Split as...
18804         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18805         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18806         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18807         (lfrintnusf, lfrintnudf): Likewise.
18808         * config/aarch64/aarch64-simd.md
18809         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18810         define_insn.
18811         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18812         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18813         (fcvt_pattern): Likewise.
18815 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18817         * config/aarch64/aarch64-simd.md
18818         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18819         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18821 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18823         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18824         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18825         (vrnd<a,m,n,p>_f32): Implement using builtins.
18826         (vrnd<i,x><q>_f<32, 64>): New.
18828 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18830         * config/aarch64/aarch64-builtins.c
18831         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18832         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18833         (frintz): Rename to...
18834         (btrunc): ...this.
18835         (frintp): Rename to...
18836         (ceil): ...this.
18837         (frintm): Rename to...
18838         (floor): ...this.
18839         (frinti): Rename to...
18840         (nearbyint): ...this.
18841         (frintx): Rename to...
18842         (rint): ...this.
18843         (frinta): Rename to...
18844         (round): ...this.
18845         * config/aarch64/aarch64-simd.md
18846         (aarch64_frint<frint_suffix><mode>): Delete.
18847         (<frint_pattern><mode>2): Convert to insn.
18848         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18849         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18850         (frint_pattern): Likewise.
18851         (frint_suffix): Likewise.
18853 2013-04-29  Richard Biener  <rguenther@suse.de>
18855         PR tree-optimization/57081
18856         * loop-init.c: Include tree-flow.h.
18857         (loop_optimizer_finalize): Free number of iteration estimates.
18858         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18860 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18862         PR tree-optimization/57083
18863         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18864         non-singleton shift count range, zero extend low_bound for uns case.
18866         * config/i386/predicates.md (general_vector_operand): New predicate.
18867         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18868         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18869         if they aren't nonimmediate operands.  If their original values
18870         satisfy const_vector_equal_evenodd_p, don't shift them.
18871         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18872         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18873         if not nonimmediate_operand.
18874         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18875         instead of register_operand.
18876         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18878 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18880         * stor-layout.c (finalize_size_functions): Allocate a structure and
18881         reset cfun before dumping the functions.
18883 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18885         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18887         PR target/56866
18888         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18889         use xop_pmacsdqh if uns_p.
18890         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18891         the immediate rotate count.
18893 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18895         * rtl.h (struct rtx_def): Add comment for field jump.
18896         (LRA_SUBREG_P): New macro.
18897         * recog.c (register_operand): Check LRA_SUBREG_P.
18898         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18899         stack frame.
18900         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18901         (lra_final_code_change): Skip subreg change for operators.
18902         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18903         if there are no operand changes.
18904         * lra-constraints.c (curr_insn_set): New.
18905         (match_reload): Set LRA_SUBREG_P.
18906         (emit_spill_move): Ditto.
18907         (check_and_process_move): Use curr_insn_set. Process only single
18908         set insns.  Don't initialize sec_mem_p and change_p.
18909         (simplify_operand_subreg): Use LRA_SUBREG_P.
18910         (reg_in_class_p): New function.
18911         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18912         of #ifdef.  Add code to remove cycling.
18913         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18914         non-null disp.  Reload inner instead of disp when base and index
18915         are null.  Try to put lo_sum into register.
18916         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18917         (check_and_process_move): Move code for move cost check to
18918         simple_move_p.  Remove equiv_substitution.
18919         (simple_move_p): New function.
18920         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18921         curr_insn_set.  Call check_and_process_move only for single set
18922         insns.  Use the new function.  Move call of check_and_process_move
18923         after operand equiv substitution and address process.
18925 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18927         PR go/57045
18928         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18929         with nonlocal goto receivers or returns twice calls, ignore
18930         unininitialized values from abnormal edges to nl goto receiver
18931         or returns twice call.
18933 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18935         PR tree-optimization/57051
18936         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18937         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18938         bitsize.
18940 2013-04-26  Richard Biener  <rguenther@suse.de>
18942         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18943         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18944         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18945         (fixup_loop_arrays_after_move): New function.
18946         (move_sese_region_to_fn): Properly outline the loop tree parts
18947         of the SESE region.
18949 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18951         * config/i386/i386.md (type, unit): Fix long lines.
18953 2013-04-26  Richard Biener  <rguenther@suse.de>
18955         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
18956         (lto-streamer-out.o): Likewise.
18957         * cfgloop.c (init_loops_structure): Export, add struct function
18958         argument and adjust.
18959         (flow_loops_find): Adjust.
18960         * cfgloop.h (enum loop_estimation): Add EST_LAST.
18961         (init_loops_structure): Declare.
18962         * lto-streamer-in.c: Include cfgloop.h.
18963         (input_cfg): Input the loop tree.
18964         * lto-streamer-out.c: Include cfgloop.h.
18965         (output_cfg): Output the loop tree.
18966         (output_struct_function_base): Do not drop PROP_loops.
18968 2013-03-26  Richard Biener  <rguenther@suse.de>
18970         * tree-cfg.c (execute_build_cfg): Build the loop tree.
18971         (pass_build_cfg): Provide PROP_loops.
18972         (move_sese_region_to_fn): Remove loops that are outlined into fn
18973         for now.
18974         * tree-inline.c: Include cfgloop.h.
18975         (initialize_cfun): Do not drop PROP_loops.
18976         (copy_loops): New function.
18977         (copy_cfg_body): Copy loop structure.
18978         (tree_function_versioning): Initialize destination loop tree.
18979         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
18980         (pass_parallelize_loops): Do IL verification.
18981         * loop-init.c (loop_optimizer_init): Fixup loops if required.
18982         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
18983         the CFG make sure we fixup loops as well.
18984         * tree-ssa-tail-merge.c: Include cfgloop.h.
18985         (replace_block_by): When merging loop latches mark loops for fixup.
18986         * lto-streamer-out.c (output_struct_function_base): Drop
18987         PROP_loops for now.
18988         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
18989         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
18990         * ipa-split.c: Include cfgloop.h.
18991         (split_function): Add the new return block to the loop tree root.
18992         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
18993         whether we have removed the forwarder block.
18994         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
18995         * cfgloop.h (place_new_loop): Declare.
18996         * cfgloopmanip.c (place_new_loop): Export.
18997         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
18998         (tree-switch-conversion.o): Likewise.
18999         (tree-complex.o): Likewise.
19000         (tree-inline.o): Likewise.
19001         (tree-ssa-tailmerge.o): Likewise.
19002         (ipa-split.o): Likewise.
19003         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19004         (tree-ssa-copy.o): Likewise.
19005         * tree-switch-conversion.c: Include cfgloop.h
19006         (process_switch): If we emit a bit-test cascade, schedule loops
19007         for fixup.
19008         * tree-complex.c: Include cfgloop.h.
19009         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19010         * asan.c: Include cfgloop.h.
19011         (create_cond_insert_point): Properly add new basic-blocks to
19012         loops, schedule loop fixup.
19013         * cfgloop.c (verify_loop_structure): Check that looks are not
19014         marked for fixup.
19015         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19016         to loops.
19017         (expand_omp_for_generic): Likewise.
19018         (expand_omp_sections): Likewise.
19019         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19020         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19021         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19022         is initialized, not when loops are present.
19023         * tree-parloops.c (parallelize_loops): Remove checking here.
19024         * passes.c (init_optimization_passes): Schedule a copy-propagation
19025         pass before complete unrolling of inner loops.
19027 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19029         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19030         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19031         (colorize_init): Add argument to _WIN32 version.
19032         * toplev.c: Include diagnostic-color.h.
19033         (process_options): Default to -fdiagnostics-color=auto if
19034         GCC_COLORS env var is in the environment.
19035         * common.opt (fdiagnostics-color=): Add Var and Init.
19036         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19037         env var is in the environment, the default is auto rather than never.
19039         * diagnostic.h (file_name_as_prefix): Add context argument.
19040         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19041         the string as locus.
19042         * langhooks.c (lhd_print_error_function): Adjust caller.
19044 2013-04-25  Lawrence Crowl  <crowl@google.com>
19046         * var-tracking.c (shared_hash_def::htab):
19047         Change type to hash_table.  Update dependent calls and types.
19049 2013-04-25  Lawrence Crowl  <crowl@google.com>
19051         * Makefile.in: Update as needed below.
19053         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19054         Move declaration to after the type's method definitons.
19056         * attribs.c (htab_t scoped_attributes::attribute_hash):
19057         Change type to hash_table.  Update dependent calls and types.
19059         * bitmap.c (htab_t bitmap_desc_hash):
19060         Change type to hash_table.  Update dependent calls and types.
19062         * cselib.c (htab_t cselib_hash_table):
19063         Change type to hash_table.  Update dependent calls and types.
19065         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19066         (hash_string_slot_node): Move implementation into lto-streamer.h
19067         struct string_slot_hasher.
19068         (eq_string_slot_node): Likewise.
19070         * data-streamer-out.c: Update output_block::string_hash_table
19071         dependent calls and types.
19073         * dwarf2cfi.c (htab_t trace_index):
19074         Change type to hash_table.  Update dependent calls and types.
19076         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19077         Change type to hash_table.  Update dependent calls and types.
19078         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19079         (htab_t optimize_external_refs::map): Likewise.
19080         (htab_t output_comp_unit::extern_map): Likewise.
19081         (htab_t output_comdat_type_unit::extern_map): Likewise.
19082         (htab_t output_macinfo::macinfo_htab): Likewise.
19083         (htab_t optimize_location_lists::htab): Likewise.
19084         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19086         * except.c (htab_t ehspec_hash_type):
19087         Change type to hash_table.  Update dependent calls and types.
19088         (assign_filter_values::ttypes): Likewise.
19089         (assign_filter_values::ehspec): Likewise.
19090         (sjlj_assign_call_site_values::ar_hash): Likewise.
19091         (convert_to_eh_region_ranges::ar_hash): Likewise.
19093         * gcse.c (htab_t pre_ldst_table):
19094         Change type to hash_table.  Update dependent calls and types.
19096         * ggc-common.c (htab_t saving_htab):
19097         Change type to hash_table.  Update dependent calls and types.
19098         (htab_t loc_hash): Likewise.
19099         (htab_t ptr_hash): Likewise.
19100         (call_count): Rename ggc_call_count.
19101         (call_alloc): Rename ggc_call_alloc.
19102         (loc_descriptor): Rename make_loc_descriptor.
19103         (add_statistics): Rename ggc_add_statistics.
19105         * ggc-common.c (saving_htab):
19106         Change type to hash_table.  Update dependent calls and types.
19108         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19109         (push_gimplify_context): Likewise.
19110         (pop_gimplify_context): Likewise.
19111         (struct gimple_temp_hash_elt): Added.
19112         (struct gimplify_hasher): Likewise.
19113         (struct gimplify_ctx.temp_htab):
19114         Change type to hash_table.  Update dependent calls and types.
19116         * gimple-fold.c: Include gimplify-ctx.h.
19118         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19119         Change type to hash_table.  Update dependent calls and types.
19120         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19121         avoid potential global name collision.
19123         * gimplify.c: Include gimplify-ctx.h.
19124         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19125         (htab_t gimplify_ctx::temp_htab):
19126         Update dependent calls and types for new type hash_table.
19127         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19128         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19130         * gimplify-ctx.h: New.
19131         (struct gimple_temp_hash_elt): Move from gimplify.c.
19132         (class gimplify_hasher): New.
19133         (struct gimplify_ctx): Move from gimple.h.
19134         (htab_t gimplify_ctx::temp_htab):
19135         Change type to hash_table.  Update dependent calls and types.
19137         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19138         (htab_t ivs_params::newivs_index):
19139         Change type to hash_table.  Update dependent calls and types.
19140         (htab_t ivs_params::params_index): Likewise.
19141         (htab_t print_generated_program::params_index): Likewise.
19142         (htab_t gloog::newivs_index): Likewise.
19143         (htab_t gloog::params_index): Likewise.
19145         * graphite.c: Include graphite-htab.h.
19146         4htab_t graphite_transform_loops::bb_pbb_mapping):
19147         Change type to hash_table.  Update dependent calls and types.
19149         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19150         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19151         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19153         * graphite-dependences.c: Include graphite-htab.h.
19154         (loop_is_parallel_p): Change hash table type of parameter.
19156         * graphite-htab.h: New.
19157         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19158         (extern find_pbb_via_hash): Move from graphite-poly.h.
19159         (extern loop_is_parallel_p): Move from graphite-poly.h.
19160         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19162         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19163         (extern loop_is_parallel_p): Move to graphite-htab.h.
19164         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19166         * haifa-sched.c (htab_t delay_htab):
19167         Change type to hash_table.  Update dependent calls and types.
19168         (htab_t delay_htab_i2): Likewise.
19170         * ira-color.c (htab_t allocno_hard_regs_htab):
19171         Change type to hash_table.  Update dependent calls and types.
19173         * ira-costs.c (htab_t cost_classes_htab):
19174         Change type to hash_table.  Update dependent calls and types.
19176         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19177         Change type to hash_table.  Update dependent calls and types.
19179         * loop-iv.c (htab_t bivs):
19180         Change type to hash_table.  Update dependent calls and types.
19182         * loop-unroll.c (htab_t opt_info::insns_to_split):
19183         Change type to hash_table.  Update dependent calls and types.
19184         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19186         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19187         (struct string_slot_hasher): New.
19188         (htab_t output_block::string_hash_table):
19189         Change type to hash_table.  Update dependent calls and types.
19191         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19192         (htab_t file_name_hash_table):
19193         Change type to hash_table.  Update dependent calls and types.
19195         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19196         calls and types.
19198         * lto-streamer.c (htab_t tree_htab):
19199         Change type to hash_table.  Update dependent calls and types.
19201         * omp-low.c: Include gimplify-ctx.h.
19203         * passes.c (htab_t name_to_pass_map):
19204         Change type to hash_table.  Update dependent calls and types.
19205         (pass_traverse): Rename to passes_pass_traverse.
19207         * plugin.c (htab_t event_tab):
19208         Change type to hash_table.  Update dependent calls and types.
19210         * postreload-gcse.c (htab_t expr_table):
19211         Change type to hash_table.  Update dependent calls and types.
19212         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19214         * sese.c (debug_rename_map_1): Make extern.
19215         (htab_t copy_bb_and_scalar_dependences::rename_map):
19216         Change type to hash_table.  Update dependent calls and types.
19218         * sese.h (extern debug_rename_map): Move to .c file.
19220         * store-motion.c (htab_t store_motion_mems_table):
19221         Change type to hash_table.  Update dependent calls and types.
19223         * trans-mem.c (htab_t tm_new_mem_hash):
19224         Change type to hash_table.  Update dependent calls and types.
19226         * tree-browser.c (htab_t TB_up_ht):
19227         Change type to hash_table.  Update dependent calls and types.
19229         * tree-cfg.c (htab_t discriminator_per_locus):
19230         Change type to hash_table.  Update dependent calls and types.
19232         * tree-complex.c: Include tree-hasher.h
19233         (htab_t complex_variable_components):
19234         Change type to hash_table.  Update dependent calls and types.
19236         * tree-eh.c (htab_t finally_tree):
19237         Change type to hash_table.  Update dependent calls and types.
19239         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19240         struct int_tree_hasher.
19241         (extern int_tree_map_eq): Likewise.
19242         (uid_decl_map_hash): Removed.
19243         (extern decl_tree_map_eq): Likewise.
19245         * tree-hasher.h: New.
19246         (struct int_tree_hasher): New.
19247         (typedef int_tree_htab_type): New.
19249         * tree-inline.c: Include gimplify-ctx.h.
19251         * tree-mudflap.c: Include gimplify-ctx.h.
19253         * tree-parloops.c: Include tree-hasher.h.
19254         (htab_t eliminate_local_variables_stmt::decl_address):
19255         Change type to hash_table.  Update dependent calls and types.
19256         (htab_t separate_decls_in_region::decl_copies): Likewise.
19258         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19259         Change type to hash_table.  Update dependent calls and types.
19261         * tree-sra.c (candidates):
19262         Change type to hash_table.  Update dependent calls and types.
19264         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19265         in tree-flow.h.
19266         (int_tree_map_hash): Likewise.
19268         * tree-ssa-dom.c (htab_t avail_exprs):
19269         Change type to hash_table.  Update dependent calls and types.
19271         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19272         Change type to hash_table.  Update dependent calls and types.
19274         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19275         Change type to hash_table.  Update dependent calls and types.
19277         * tree-ssa-phiopt.c (seen_ssa_names):
19278         Change type to hash_table.  Update dependent calls and types.
19280         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19281         Change type to hash_table.  Update dependent calls and types.
19283         * tree-ssa-uncprop.c (equiv):
19284         Change type to hash_table.  Update dependent calls and types.
19286 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19288         PR rtl-optimization/57003
19289         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19290         call note_stores with kill_clobbered_value callback again after
19291         killing regs_invalidated_by_call.
19293 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19295         * config/aarch64/aarch64-simd.md
19296         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19297         (aarch64_simd_bsl<mode>): Likewise.
19298         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19300 2013-04-25  Marek Polacek  <polacek@redhat.com>
19302         PR tree-optimization/57066
19303         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19305 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19307         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19309 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19311         * config/aarch64/aarch64-builtins.c
19312         (aarch64_fold_builtin): New.
19313         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19314         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19315         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19316         * config/aarch64/arm_neon.h
19317         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19319 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19320             Tejas Belagod  <tejas.belagod@arm.com>
19322         * config/aarch64/aarch64-builtins.c
19323         (aarch64_gimple_fold_builtin): New.
19324         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19325         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19326         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19327         (addvv4sf): Update.
19328         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19330 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19332         * config/aarch64/aarch64.md
19333         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19335 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19337         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19338         (*ngcsi_uxtw): New pattern.
19340 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19341             Julian Brown  <julian@codesourcery.com>
19343         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19344         (TB_DREG): Add T_V4HF.
19345         (v4hf_UP): New macro.
19346         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19347         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19348         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19349         built-ins.
19350         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19351         (arm_vector_mode_supported_p): Handle V4HF.
19352         (arm_mangle_map): Handle V4HFmode.
19353         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19354         * config/arm/arm_neon_builtins.def: Add entries for
19355         vcvtv4hfv4sf, vcvtv4sfv4hf.
19356         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19357         (neon_vcvtv4hfv4sf): Likewise.
19358         * config/arm/neon-gen.ml: Handle half-precision floating point
19359         features.
19360         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19361         * config/arm/arm_neon.h: Regenerate.
19362         * config/arm/neon.ml (type elts): Add F16.
19363         (type vectype): Add T_float16x4, T_floatHF.
19364         (type vecmode): Add V4HF.
19365         (type features): Add Requires_FP_bit feature.
19366         (elt_width): Handle F16.
19367         (elt_class): Likewise.
19368         (elt_of_class_width): Likewise.
19369         (mode_of_elt): Refactor.
19370         (type_for_elt): Handle F16, fix error messages.
19371         (vectype_size): Handle T_float16x4.
19372         (vcvt_sh): New function.
19373         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19374         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19375         (string_of_mode): Handle V4HF.
19376         * doc/arm-neon-intrinsics.texi: Regenerate.
19378 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19380         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19381         format specifier in 'X' case.
19383 2013-04-25  Alan Modra  <amodra@gmail.com>
19385         PR target/57052
19386         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19387         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19388         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19389         Repeat for many other rotate/shift and mask patterns using subregs.
19390         Name lshiftrt insns.
19391         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19392         on WORDS_BIG_ENDIAN.
19394 2013-04-25  Alan Modra  <amodra@gmail.com>
19396         * config.gcc: Support little-endian powerpc-linux targets.
19397         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19398         (LINK_OS_LINUX_SPEC): Define.
19399         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19400         Preserve MASK_LITTLE_ENDIAN.
19401         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19402         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19403         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19404         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19405         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19406         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19407         Correct fp word order for little-endian.  Don't shift toc entries
19408         smaller than a word for little-endian.
19409         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19410         (bswapdi2 splits): Correct low-part subreg for little-endian.
19411         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19412         low/high where such is correct only for be.
19413         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19414         little-endian for -mcall-aixdesc.
19416 2013-04-25  Alan Modra  <amodra@gmail.com>
19418         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19419         replace_equiv_address_nv.
19421 2013-04-25  Alan Modra  <amodra@gmail.com>
19423         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19425 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19427         Revert:
19428         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19429         * rtl.h (struct rtx_def): ...
19431 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19433         PR rtl-optimizations/57046
19434         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19435         for multi-reg splits.
19437 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19439         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19441 2013-04-24  Sterling Augustine  <saugustine@google.com>
19443         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19444         (comp_dir_string, debug_str_dwo_section): New.
19445         (DEBUG_STR_DWO_SECTION): Rename to ...
19446         (DEBUG_DWO_STR_SECTION): ... this.
19447         (DEBUG_NORM_STR_SECTION): Delete.
19448         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19449         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19450         (find_AT_string): Move most logic to ...
19451         (find_AT_string_in_table): ... here.  New.
19452         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19453         add_skeleton_AT_string.  Delete logic.
19454         (output_skeleton_debug_sections): Remove call to
19455         add_top_level_skeleton_die_attrs.
19456         (add_comp_dir_attribute): Move logic to comp_dir_string.
19457         (dwarf2out_init): Initialize debug_str_dwo_section.
19458         (output_indirect_string): Call find_string_form.
19459         (output_indirect_strings): Rewrite.
19460         (prune_unused_types): Empty skeleton_debug_str_hash.
19461         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19462         (dwarf2out_finish):  Call output_indirect_strings.
19464 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19466         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19468 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19470         * rtl.h (struct rtx_def): Add comment for field jump.
19471         (LRA_SUBREG_P): New macro.
19472         * recog.c (register_operand): Check LRA_SUBREG_P.
19473         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19474         stack frame.
19475         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19476         (lra_final_code_change): Skip subreg change for operators.
19477         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19478         if there are no operand changes.
19479         * lra-constraints.c (curr_insn_set): New.
19480         (match_reload): Set LRA_SUBREG_P.
19481         (emit_spill_move): Ditto.
19482         (check_and_process_move): Use curr_insn_set. Process only single
19483         set insns.  Don't initialize sec_mem_p and change_p.
19484         (simplify_operand_subreg): Use LRA_SUBREG_P.
19485         (reg_in_class_p): New function.
19486         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19487         of #ifdef.  Add code to remove cycling.
19488         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19489         non-null disp.  Reload inner instead of disp when base and index
19490         are null.  Try to put lo_sum into register.
19491         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19492         (check_and_process_move): Move code for move cost check to
19493         simple_move_p.  Remove equiv_substitution.
19494         (simple_move_p): New function.
19495         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19496         curr_insn_set.  Call check_and_process_move only for single set
19497         insns.  Use the new function.  Move call of check_and_process_move
19498         after operand equiv substitution and address process.
19500 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19502         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19503         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19504         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19506 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19508         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19510 2013-04-24  Marek Polacek  <polacek@redhat.com>
19512         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19513         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19514         (select_loops_exit_conditions): Likewise.
19515         (number_of_iterations_for_all_loops): Likewise.
19516         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19517         (scev_analysis): Likewise.
19519 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19520             Chao-ying Fu  <fu@mips.com>
19522         * config/mips/micromips.md (jraddiusp): New pattern.
19523         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19524         instruction if possible.
19526 2013-04-24  Alan Modra  <amodra@gmail.com>
19528         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19530 2013-04-24  Julian Brown  <julian@codesourcery.com>
19531             Chung-Lin Tang  <cltang@codesourcery.com>
19533         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19534         dependency behavior in enumeration type DIE generation. Add TODO note
19535         to comments about future DW_FORM_sdata/udata re-work of related code.
19537 2013-04-23  Lawrence Crowl  <crowl@google.com>
19539         * Makefile.in: Update as needed below.
19541         * hash-table.h (class hash_table):
19542         Correct many methods with parameter types compare_type to the correct
19543         value_type.  (Correct code was unlikely to notice the change.)
19544         (hash_table::elements_with_deleted) New.
19545         (class hashtable::iterator): New.
19546         (hashtable::begin()): New.
19547         (hashtable::end()): New.
19548         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19550         * statistics.c (statistics_hashes):
19551         Change type to hash_table.  Update dependent calls and types.
19553         * tree-into-ssa.c (var_infos):
19554         Change type to hash_table.  Update dependent calls and types.
19556         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19557         Change type to hash_table.  Update dependent calls and types.
19559         * tree-ssa-loop-im.c (struct mem_ref.refs):
19560         Change type to hash_table.  Update dependent calls and types.
19562         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19563         Change type to hash_table.  Update dependent calls and types.
19565         * tree-ssa-sccvn.c (vn_tables_s::nary):
19566         Change type to hash_table.  Update dependent calls and types.
19567         (vn_tables_s::phis): Likewise.
19568         (vn_tables_s::references): Likewise.
19570         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19571         (vn_reference_eq): Update parameter and return types.
19573         * tree-ssa-structalias.c (pointer_equiv_class_table):
19574         Change type to hash_table.  Update dependent calls and types.
19575         (location_equiv_class_table): Likewise.
19577         * tree-vect-data-refs.c: Consequential changes for making
19578         peeling a hash_table.
19580         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19581         (destroy_loop_vec_info): Dependent hash_table update.
19583         * tree-vectorizer.h (peeling_htab):
19584         Change type to hash_table.  Update dependent calls and types.
19586 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19588         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19589         to check the register content is equal or not.
19590         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19591         to assign register content record.
19592         * lra-eliminations.c (update_reg_eliminate): Use
19593         lra_update_reg_val_offset to update register content offset.
19594         * lra-int.h (struct lra_reg): Add offset member.
19595         (lra_reg_val_equal_p): New static inline function.
19596         (lra_update_reg_val_offset): New static inline function.
19597         (lra_assign_reg_val): New static inline function.
19598         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19599         to assign register content record.
19600         (initialize_lra_reg_info_element): Initial offset to zero.
19602 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19604         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19605         operands.  Record compression.
19607 2013-04-23  Xinliang David Li  <davidxl@google.com>
19609         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19611 2013-04-23  Richard Biener  <rguenther@suse.de>
19613         PR middle-end/57036
19614         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19615         parameter, only add abnormal goto edges from the copied body
19616         if the call could perform abnormal gotos.
19617         (copy_cfg_body): Adjust.
19619 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19621         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19623 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19625         * coretypes.h (gimple_stmt_iterator): Add struct to make
19626         compatible with C.
19628 2013-04-23  Richard Biener  <rguenther@suse.de>
19630         PR tree-optimization/57026
19631         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19632         from SSA names occuring in abnormal PHI nodes.
19634 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19636         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19638 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19640         * lto/lto.c (print_lto_report_1): Declare early.
19641         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19643 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19645         * common.opt (-flto-report-wpa): Add.
19646         * doc/invoke.texi (-flto-report-wpa): Add.
19647         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19648         (lto_main): dito.
19650 2013-04-22  Xinliang David Li  <davidxl@google.com>
19652         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19653         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19654         * Makefile.in: New dependency
19656             David Daney <ddaney.cavm@gmail.com>
19658         * configure.ac (gcc_cv_as_micromips_support):  Use the
19659         --fatal-warnings option.
19660         * configure: Regenerate.
19662 2013-04-22  Marek Polacek  <polacek@redhat.com>
19664         PR sanitizer/56990
19665         * tsan.c (instrument_expr): Don't instrument expression
19666         in case its size is zero.
19668 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19670         PR target/57032
19671         Revert:
19672         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19674         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19676 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19678         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19679         (gimple_stmt_iterator): New typedef.
19680         * gimple.h (gimple_stmt_iterator): Rename to...
19681         (gimple_stmt_iterator_d): ... This.
19682         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19683         trees be valid for GIMPLE and GENERIC.
19684         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19685         * gimple-fold.c (gimple_fold_call): Call target hook
19686         gimple_fold_builtin.
19687         * hooks.c (hook_bool_gsiptr_false): New.
19688         * hooks.h (hook_bool_gsiptr_false): New.
19689         * target.def (fold_stmt): New.
19690         * doc/tm.texi: Regenerate.
19692 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19694         PR target/57018
19695         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19696         a set sp if no stack realignment.
19698 2013-04-22  Nick Clifton  <nickc@redhat.com>
19700         * config.gcc (tilegx-linux): Extend extra_objs rather than
19701         overwriting it.
19702         (tilepro-linux): Likewise.
19704 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19706         * config/aarch64/aarch64-builtins.c
19707         (CF): Remove.
19708         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19709         (VAR<1-12>): Add MAP parameter.
19710         (BUILTIN_*): Likewise.
19711         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19712         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19713         (aarch64_ushl_n<mode>): Likewise.
19714         (aarch64_sshr_n<mode>): Likewise.
19715         (aarch64_ushr_n<mode>): Likewise.
19716         (aarch64_<maxmin><mode>): Likewise.
19717         (aarch64_sqrt<mode>): Likewise.
19718         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19719         (vshr<q>_n_*): Likewise.
19721 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19723         * config/aarch64/aarch64-builtins.c
19724         (aarch64_simd_builtin_type_mode): Handle SF types.
19725         (sf_UP): Define.
19726         (BUILTIN_GPF): Define.
19727         (aarch64_init_simd_builtins): Handle SF types.
19728         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19729         (frecps): Likewise.
19730         (frecpx): Likewise.
19731         * config/aarch64/aarch64-simd.md
19732         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19733         (aarch64_frecpe<mode>): New.
19734         (aarch64_frecps<mode>): Likewise.
19735         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19736         (v8type): Add frecp<esx>.
19737         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19738         (aarch64_frecps<mode>): Likewise.
19739         * config/aarch64/iterators.md (FRECP): New.
19740         (frecp_suffix): Likewise.
19741         * config/aarch64/arm_neon.h
19742         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19744 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19746         PR target/56995
19747         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19748         (REG_CLASS_NAMES): Idem.
19749         (REG_CLASS_CONTENTS): Idem.
19750         (REGCLASS_HAS_FP_REG): Idem.
19751         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19752         (sh_conditional_register_usage): Idem.
19754 2013-04-21  Jeff Law  <law@redhat.com>
19756         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19757         (ssa_forward_propagate_and_combine): Use it.
19759 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19761         * lra.c: Update the flow chart diagram.
19763 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19765         PR rtl-optimization/56847
19766         * lra-constraints.c (process_alt_operands): Discourage alternative
19767         with non-matche doffsettable memory constraint fro memory with
19768         known offset.
19770 2013-04-19  Richard Biener  <rguenther@suse.de>
19772         PR tree-optimization/56982
19773         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19774         function.
19775         * gimplify.c (gimplify_call_expr): Notice special calls.
19776         (gimplify_modify_expr): Likewise.
19777         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19778         abnormal control flow receivers.
19779         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19780         in the same way as cfun->has_nonlocal_labels.
19781         (gimple_purge_dead_abnormal_call_edges): Likewise.
19782         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19783         receivers start a basic-block.
19785 2013-04-19  Richard Biener  <rguenther@suse.de>
19787         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19788         member ...
19789         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19790         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19791         (SLP_TREE_LOAD_PERMUTATION): Add.
19792         (vect_transform_slp_perm_load): Adjust prototype.
19793         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19794         (vect_free_slp_instance): Likewise.
19795         (vect_create_new_slp_node): Likewise.
19796         (vect_supported_slp_permutation_p): Remove.
19797         (vect_slp_rearrange_stmts): Adjust.
19798         (vect_supported_load_permutation_p): Likewise.  Inline
19799         vect_supported_slp_permutation_p here.
19800         (vect_analyze_slp_instance): Compute load permutations per
19801         slp node instead of per instance.
19802         (vect_get_slp_defs): Adjust.
19803         (vect_transform_slp_perm_load): Likewise.
19804         (vect_schedule_slp_instance): Remove redundant code.
19805         (vect_schedule_slp): Remove hack for PR56270, add it ...
19806         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19807         CSE loads for SLP.  Adjust.
19809 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19811         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19812         spelling in two comments.
19814 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19816         PR target/56797
19817         * config/arm/arm.c (load_multiple_sequence): Require SP
19818         as base register for loads if SP is in the register list.
19820 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19822         PR tree-optimization/56718
19823         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19824         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19825         and made public.  Adjusted all callers.
19826         (ipa_intraprocedural_devirtualization): New function.
19827         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19828         (ipa_intraprocedural_devirtualization): Likewise.
19829         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19831 2013-04-19  Richard Biener  <rguenther@suse.de>
19833         PR tree-optimization/57000
19834         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19836 2013-04-19  Terry Guo  <terry.guo@arm.com>
19838         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19839         Replace with ...
19840         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19841         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19842         (cortex_m4_fmacs): Use new reservations.
19843         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19845 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19847         PR rtl-optimization/56999
19848         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19849         related code.
19850         (lra_coalesce): Remove split_origin_bitmap and related code.
19851         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19852         ranges if necessary.
19854 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19856         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19857         New array.
19858         (ix86_expand_call): Remove clobbered_registers array and use
19859         x86_64_ms_sysv_extra_clobbered_registers instead.
19860         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19861         Declare here.
19862         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19863         predicate.
19864         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19865         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19866         (*call_value_rex64_ms_sysv): Ditto.
19868 2013-04-18  Cary Coutant  <ccoutant@google.com>
19870         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19871         parent instead of die_mark.
19873 2013-04-18  Diego Novillo  <dnovillo@google.com>
19875         * gimple.c (create_gimple_tmp): New.
19876         (get_expr_type): New.
19877         (build_assign): New.
19878         (build_type_cast): New.
19879         * gimple.h (enum ssa_mode): Define.
19880         (gimple_seq_set_location): New.
19881         * asan.c (build_check_stmt): Change some gimple_build_* calls
19882         to use build_assign and build_type_cast.
19884 2013-04-18  Richard Biener  <rguenther@suse.de>
19886         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19887         handle negative step.  Remove redundant checks.
19888         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19889         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19890         for negative step and grouped loads fail to vectorize.
19892 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19894         * emit-rtl.c (reset_insn_used_flags): New function.
19895         (reset_all_used_flags): Use it.
19896         (verify_insn_sharing): New function.
19897         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19899 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19901         PR tree-optimization/56984
19902         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19903         and (x >> M) >= N don't register any assertion if N << M is the
19904         minimum value.
19906 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19908         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19909         do not delete_insn insns that have not yet been emitted, only
19910         unlink them with remove_insn.
19911         * df-scan.c (df_insn_delete): Revert r197492.
19913 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19915         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19916         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19918 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19920         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19921         define_insn_and_split.
19922         (and_scc,ior_scc,negscc): Likewise.
19923         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19925 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19927         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19928         can benefit from using a sequence of LDRD instructions in epilogue
19929         instead of a single LDM instruction.
19931 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19933         PR 45688
19934         * doc/extend.texi: Fix typo.
19936 2013-04-17  Richard Biener  <rguenther@suse.de>
19938         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19939         (vect_build_slp_tree): ... here.
19940         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19941         match.  Remove special-casing of mismatched complex loads.
19942         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19943         re-try the match with swapped commutative operands.
19944         (vect_supported_load_permutation_p): Remove special-casing of
19945         mismatched complex loads.
19946         (vect_analyze_slp_instance): Adjust.
19948 2013-04-17  Richard Biener  <rguenther@suse.de>
19950         PR rtl-optimization/56921
19951         * cfgloop.h (struct loop): Add simple_loop_desc member.
19952         (struct niter_desc): Mark with GTY(()).
19953         (simple_loop_desc): Do not use aux field but simple_loop_desc.
19954         * loop-iv.c (get_simple_loop_desc): Likewise.
19955         (free_simple_loop_desc): Likewise.
19957         Revert
19958         2013-04-16  Richard Biener  <rguenther@suse.de>
19960         PR rtl-optimization/56921
19961         * loop-init.c (pass_rtl_move_loop_invariants): Add
19962         TODO_do_not_ggc_collect to todo_flags_finish.
19963         (pass_rtl_unswitch): Same.
19964         (pass_rtl_unroll_and_peel_loops): Same.
19965         (pass_rtl_doloop): Same.
19967 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
19969         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
19970         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
19971         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
19972         references.
19973         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
19974         * tree-streamer.c (record_common_node): Adjust reference in comment.
19976 2013-04-17  Terry Guo  <terry.guo@arm.com>
19978         * config/arm/cortex-m4.md: Add a new bypass.
19980 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19982         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
19983         New pattern.
19984         (*subs_<optab><mode>_multp2): New pattern.
19985         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19986         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19988 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19990         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
19991         (*subs_mul_imm_<mode>): New pattern.
19993 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
19995         PR target/56948
19996         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
19997         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
19998         (vsx_movti_32bit): Change j->wa to O->wa.
20000 2013-04-16  Richard Biener  <rguenther@suse.de>
20002         PR rtl-optimization/56921
20003         * loop-init.c (pass_rtl_move_loop_invariants): Add
20004         TODO_do_not_ggc_collect to todo_flags_finish.
20005         (pass_rtl_unswitch): Same.
20006         (pass_rtl_unroll_and_peel_loops): Same.
20007         (pass_rtl_doloop): Same.
20009 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20011         * config/arm/arm.c (emit_multi_reg_push): New declaration
20012         for an existing function.
20013         (arm_emit_strd_push): New function.
20014         (arm_expand_prologue): Used here.
20015         (arm_emit_ldrd_pop): New function.
20016         (arm_expand_epilogue): Used here.
20017         (arm_get_frame_offsets): Update condition.
20018         (arm_emit_multi_reg_pop): Add a special case for load of a single
20019         register with writeback.
20021 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20023         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20024         description.
20026 2013-04-16  Richard Biener  <rguenther@suse.de>
20028         PR tree-optimization/56756
20029         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20030         (first_mem_ref_loc): New.
20031         (execute_sm): Place the load temporarily before a previous
20032         access instead of in the latch edge to ensure its SSA dependencies
20033         are defined at points dominating the load.
20035 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20037         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20038         correct fix by moving header and footer insn to the footer of
20039         the merged basic block.  Clear BB_END of the merged-away block.
20041         PR middle-end/43631
20042         * emit-rtl.c (make_note_raw): New function.
20043         (link_insn_into_chain): New static inline function.
20044         (add_insn): Use it.
20045         (add_insn_before, add_insn_after): Factor insn chain linking code...
20046         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20047         using link_insn_into_chain.
20048         (note_outside_basic_block_p): New helper function for emit_note_after
20049         and emit_note_before.
20050         (emit_note_after): Use nobb variant of add_insn_after if the note
20051         should not be contained in a basic block.
20052         (emit_note_before): Use nobb variant of add_insn_before if the note
20053         should not be contained in a basic block.
20054         (emit_note_copy): Use make_note_raw.
20055         (emit_note): Likewise.
20056         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20057         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20058         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20059         the moved barrier the tail of the basic block it follows.
20060         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20062 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20064         PR tree-optimization/56962
20065         * gimple-ssa-strength-reduction.c (record_increment): Only set
20066         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20067         either rhs1 or rhs2 is equal to c->base_expr.
20069 2013-04-15  Richard Biener  <rguenther@suse.de>
20071         PR tree-optimization/56933
20072         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20073         member.
20074         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20075         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20076         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20077         dependence check ...
20078         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20079         ... here.
20080         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20081         GROUP_READ_WRITE_DEPENDENCE.
20083 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20085         * emit-rtl.c (reset_all_used_flags): New function.
20086         (verify_rtl_sharing): Call reset_all_used_flags before and after
20087         performing the checks.
20089 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20091         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20092         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20093         * config/arm/constraints.md (De): New constraint.
20094         * config/arm/neon.md (anddi3_neon): Delete.
20095         (neon_vand<mode>): Expand to standard anddi3 pattern.
20096         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20097         Move earlier in the file.
20098         (neon_inv_logic_op2): Likewise.
20099         (arm_anddi_operand_neon): New predicate.
20101 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20103         * configure.ac (gcc_cv_ld_as_needed): Set
20104         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20105         Use -z ignore, -z record on *-*-solaris2*.
20106         (HAVE_LD_AS_NEEDED): Update comment.
20107         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20108         * configure: Regenerate.
20109         * config.in: Regenerate.
20110         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20111         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20112         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20113         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20114         equivalents.  Fix markup.
20115         * doc/tm.texi: Regenerate.
20117 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20119         * config/i386/i386.opt: New option mstack-protector-guard=.
20120         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20121         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20122         TARGET_SSP_TLS_GUARD.
20123         * config/i386/i386.c (ix86_option_override_internal): Set
20124         ix86_stack_protector_guard.
20125         * config/i386/i386.md (stack_protect_set): Enable for
20126         TARGET_SSP_TLS_GUARD only.
20127         (stack_protect_set_<mode>): Ditto.
20128         (stack_protect_test): Ditto.
20129         (stack_protect_test_<mode>): Ditto.
20130         * doc/invoke.texi (i386 Option): Document.
20132 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20134         PR target/56890
20135         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20136         (S_MODES): Set H_MODE bit.
20137         (SF_MODES): Set only S_MODE and SF_MODE bits.
20138         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20139         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20140         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20141         <MODE_FLOAT>: Likewise.
20143 2013-04-15  Joey Ye  <joey.ye@arm.com>
20145         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20147 2013-04-15  Joey Ye  <joey.ye@arm.com>
20149         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20150         for real far jump.
20151         (thumb_far_jump_used_p): Count instruction size and set
20152         far_jump_used.
20154 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20156         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20157         * resource.c (mark_target_live_regs): Reformat conditional block.
20159 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20161         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20162         notes, they are emitted only just before final.
20163         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20165 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20167         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20168         * cfgrtl.c (delete_insn): Call it here instead.
20169         * lra-spills.c (lra_final_code_change): Use delete_insn.
20170         * haifa-sched.c (sched_remove_insn): Likewise.
20171         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20172         returning to the nop pool.
20173         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20174         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20176 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20178         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20179         * doc/tm.texi: Regenerated.
20181 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20183         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20184         QImode checks.
20186 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20188         * df-core.c (df_find_def): Compare register numbers.
20189         (df_find_use): Likewise.
20191 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20193         PR target/56903
20194         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20195         lra_in_progress for return.
20197 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20199         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20200         define_insn into define_insn_and_split and emit movsicc patterns.
20202 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20204         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20206 2013-04-12  Richard Biener  <rguenther@suse.de>
20208         * tree-pass.h (TODO_do_not_ggc_collect): New.
20209         * passes.c (execute_one_ipa_transform_pass): Honor
20210         TODO_do_not_ggc_collect.
20211         (execute_one_pass): Likewise.
20213         Revert
20214         2013-04-10  Richard Biener  <rguenther@suse.de>
20216         * passes.c (init_optimization_passes): Remove reload pass.
20217         * ira.c (do_reload): Merge into ...
20218         (ira): ... this.
20219         (rest_of_handle_reload): Remove.
20220         (pass_reload): Likewise.
20221         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20222         of reload for vzeroupper pass placement.
20224 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20226         PR tree-optimization/56918
20227         PR tree-optimization/56920
20228         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20229         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20230         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20231         use wide_mul_with_sign method.
20233 2013-04-12  Richard Biener  <rguenther@suse.de>
20235         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20236         not be considered a gimple constant.
20238 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20240         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20241         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20242         operations.
20244 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20245             Jakub Jelinek  <jakub@redhat.com>
20247         * opts.c: Include diagnostic-color.h.
20248         (common_handle_option): Handle OPT_fdiagnostics_color_.
20249         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20250         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20251         (diagnostic-color.o): New.
20252         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20253         (diagnostic_color_rule): New enum.
20254         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20255         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20256         the location string.
20257         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20258         either NULL, or color kind.
20259         * diagnostic-color.c: New file.
20260         * diagnostic-color.h: New file.
20261         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20262         arguments.
20263         * doc/invoke.texi (-fdiagnostics-color): Document.
20264         * pretty-print.h (pp_show_color): Define.
20265         (struct pretty_print_info): Add show_color field.
20266         * diagnostic.c: Include diagnostic-color.h.
20267         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20268         macros.  Colorize error:, warning: etc. strings and also the location
20269         string.
20270         (diagnostic_show_locus): Colorize the caret line.
20271         * pretty-print.c: Include diagnostic-color.h.
20272         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20273         inside of %< %> quotes or quoted through q format modifier.
20275 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20277         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20279 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20281         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20282         code in CC_NZ mode.
20283         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20284         pattern.
20286 2013-04-11  Marek Polacek  <polacek@redhat.com>
20288         PR tree-optimization/48184
20289         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20291 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20293         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20294         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20295         (skip_simple_arithmetic): Tidy up.
20296         * tree.h (skip_simple_constant_arithmetic): Declare.
20298 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20300         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20302 2013-04-11  Richard Biener  <rguenther@suse.de>
20304         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20305         generate vector constants.
20307 2013-04-11  Richard Biener  <rguenther@suse.de>
20309         PR tree-optimization/56878
20310         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20311         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20312         New function.
20313         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20314         Prefer to align the DR with the most invariant base address.
20316 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20318         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20319         comment.
20321 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20323         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20324         floating-point vector comparisons against 0.
20326 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20328         PR tree-optimization/56899
20329         * fold-const.c (extract_muldiv_1): Apply distributive law
20330         only if TYPE_OVERFLOW_WRAPS (ctype).
20332 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20334         PR target/56124
20335         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20336         loading has side effect.
20338 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20340         * config/sparc/sparc.c: Include tree-pass.h.
20341         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20342         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20343         head of file.  Change return type.  Split off gate function.
20344         (sparc_gate_work_around_errata): New function.
20345         (pass_work_around_errata): New pass definition.
20346         (insert_pass_work_around_errata) New pass insert definition to
20347         insert pass_work_around_errata just after delayed-branch scheduling.
20348         (sparc_option_override): Insert the pass.
20349         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20351 2013-04-10  David S. Miller  <davem@davemloft.net>
20353         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20354         or -mcpu=hypersparc.
20356         * target.def (cstore_mode): New hook.
20357         * target.h: Include insn-codes.h
20358         * targhooks.c: Likewise.
20359         (default_cstore_mode): New function.
20360         * targhooks.h: Declare it.
20361         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20362         * doc/tm.texi: Rebuild.
20363         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20364         target hook, rather than inspecting the insn_data.
20365         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20366         (TARGET_CSTORE_MODE): Redefine.
20367         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20368         result patterns.
20369         * config/sparc/predicates.md (cstore_result_operand): New special
20370         predicate.
20371         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20372         Use it for operand 0.
20373         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20374         (*snesi_special): Likewise.
20375         (*snesi_zero): Likewise.
20376         (*seqsi_zero): Likewise.
20377         (*sltu_insn): Likewise.
20378         (*sgeu_insn): Likewise.
20379         (*seqdi_special): Make operand 0 and comparison operation be of
20380         DImode.
20381         (*snedi_special): Likewise.
20382         (*snedi_special_vis3): Likewise.
20383         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20384         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20385         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20386         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20387         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20388         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20389         (*sltu_extend_sp64): Likewise.
20390         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20391         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20392         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20393         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20394         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20396 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20398         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20399         (aarch64_start_file): Use the new function.
20401 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20402             Jason Merrill  <jason@redhat.com>
20404         * common.opt: Add -gdwarf.
20405         * opts.c (common_handle_option): Handle it.
20406         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20408 2013-04-10  Richard Biener  <rguenther@suse.de>
20410         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20411         (execute_one_ipa_transform_pass): But unconditionally here.
20412         (execute_one_pass): And here.
20413         (init_optimization_passes): Remove reload pass.
20414         * tree-pass.h (TODO_ggc_collect): Remove.
20415         (pass_reload): Likewise.
20416         * ira.c (do_reload): Merge into ...
20417         (ira): ... this.
20418         (rest_of_handle_reload): Remove.
20419         (pass_reload): Likewise.
20420         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20421         of reload for vzeroupper pass placement.
20422         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20423         and todo_flags_finish of all passes.
20425 2013-04-10  Richard Biener  <rguenther@suse.de>
20427         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20428         first_const_oprnd field, rename first_def_type to first_op_type.
20429         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20430         (vect_get_and_check_slp_defs): Always use the type of the
20431         operand.  Allow mixed vect_external_def, vect_constant_def types.
20432         (vect_get_constant_vectors): Handle mixed vect_external_def,
20433         vect_constant_def types.
20435 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20437         PR tree-optimization/55524
20438         * tree-ssa-math-opts.c
20439         (convert_mult_to_fma): Don't use an fms construct
20440         when we don't have an fms operation, but fnma, and it looks
20441         likely that we'll be able to use the latter.
20443 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20445         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20446         function.
20447         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20448         inline fail caused by overwritable functions.
20450 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20452         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20453         unnecessary bits in the constant power of two case.
20455 2013-04-10  Richard Biener  <rguenther@suse.de>
20457         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20458         broken code swapping operands.
20459         (vect_build_slp_tree): Do not compute load permutations here.
20460         (vect_analyze_slp_instance): Compute load permutations here,
20461         after building the SLP tree.
20463 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20465         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20466         of next/prev_real_insn.
20468 2013-04-09  Jan Hubicka  <jh@suse.cz>
20470         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20471         Drop aliased parameter.
20472         (function_and_variable_visibility): Do not handle alias pairs.
20473         * cgraph.c (varpool_externally_visible_p): Update prototype.
20474         * varpool.c (varpool_add_new_variable): Update.
20476 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20478         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20480 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20482         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20484         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20486 2013-04-09  Marek Polacek  <polacek@redhat.com>
20488         PR tree-optimization/48762
20489         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20491 2013-04-09  Richard Biener  <rguenther@suse.de>
20493         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20494         dealing with cost.
20495         (vect_build_slp_tree): Likewise.
20496         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20497         calculating the cost of a SLP instance.
20498         (vect_analyze_slp_instance): Use it from here, after building
20499         the SLP tree.
20501 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20503         PR middle-end/56883
20504         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20505         expand_omp_for_static_chunk): Use simple_p = true in
20506         force_gimple_operand_gsi calls when assigning to addressable decls.
20508 2013-04-09  Jeff Law  <law@redhat.com>
20510         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20511         when the boolean was created by converting a wider object which
20512         had a boolean range.
20514 2013-04-09  Richard Biener  <rguenther@suse.de>
20516         * tree-vectorizer.h (slp_void_p): Remove.
20517         (slp_tree): Typedef before _slp_tree declaration.
20518         (struct _slp_tree): Use a vector of slp_tree as children.
20519         (vect_get_place_in_interleaving_chain): Remove.
20520         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20521         Move ...
20522         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20523         and make static.
20524         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20525         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20526         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20527         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20528         Use slp_node instead of slp_void_p and adjust.
20530 2013-04-09  Richard Biener  <rguenther@suse.de>
20532         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20533         work that is not necessary.
20535 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20537         PR tree-optimization/56854
20538         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20539         forward into clobber stmts if it would change MEM_REF lhs into
20540         non-MEM_REF.
20542 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20544         * tree.c (type_hash_lookup, type_hash_add): Make static.
20545         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20547 2013-04-09  Richard Biener  <rguenther@suse.de>
20549         * tree.h (unsave_expr_now): Remove.
20550         * tree-inline.c (mark_local_for_remap_r): Remove.
20551         (unsave_expr_1): Likewise.
20552         (unsave_r): Likewise.
20553         (unsave_expr_now): Likewise.
20554         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20555         (propagate_tree_value): Likewise.
20557 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20559         * doc/rtl.texi (sequence): Rewrite documentation to match the
20560         current use of SEQUENCE rtl objects.
20561         * rtl.def (SEQUENCE): Likewise.
20563         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20564         Update documentation.
20565         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20566         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20568         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20570 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20572         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20573         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20574         methods.
20575         (estimate_edge_size_and_time): Add comment to suggest using rounding
20576         methods.
20577         (estimate_node_size_and_time): Ditto.
20578         (remap_edge_change_prob): Use helper rounding divide methods.
20579         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20580         (gimple_mod_pow2_value_transform): Ditto.
20581         (gimple_mod_subtract_transform): Ditto.
20582         (gimple_ic_transform): Ditto.
20583         (gimple_stringops_transform): Ditto.
20584         * stmt.c (conditional_probability): Ditto.
20585         (emit_case_dispatch_table): Ditto.
20586         * lto-cgraph.c (merge_profile_summaries): Ditto.
20587         * tree-optimize.c (execute_fixup_cfg): Ditto.
20588         * cfgcleanup.c (try_forward_edges): Ditto.
20589         * cfgloopmanip.c (scale_loop_profile): Ditto.
20590         (loopify): Ditto.
20591         (duplicate_loop_to_header_edge): Ditto.
20592         (lv_adjust_loop_entry_edge): Ditto.
20593         * tree-vect-loop.c (vect_transform_loop): Ditto.
20594         * profile.c (compute_branch_probabilities): Ditto.
20595         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20596         * lto-streamer-in.c (input_cfg): Ditto.
20597         * gimple-streamer-in.c (input_bb): Ditto.
20598         * ipa-cp.c (update_profiling_info): Ditto.
20599         (update_specialized_profile): Ditto.
20600         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20601         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20602         rounding methods.
20603         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20604         (compute_trg_info): Ditto.
20605         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20606         (purge_dead_edges): Ditto.
20607         * loop-unswitch.c (unswitch_loop): Ditto.
20608         * cgraphclones.c (cgraph_clone_edge): Ditto.
20609         (cgraph_clone_node): Ditto.
20610         * tree-inline.c (copy_bb): Ditto.
20611         (copy_edges_for_bb): Ditto.
20612         (initialize_cfun): Ditto.
20613         (copy_cfg_body): Ditto.
20614         (expand_call_inline): Ditto.
20616 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20618         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20619         TARGET_CYGWIN64 by TARGET_64BIT.
20621 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20623         * config/epiphany/epiphany.md (GPR_1): New constant.
20624         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20625         * config/epiphany/epiphany.c (gen_compare_reg):
20626         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20627         is already in place.
20628         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20629         Don't require being called during rtl expansion; If y operlaps r0,
20630         return 0.
20631         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20632         (epiphany_expand_epilogue): Likewise.
20634         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20635         Don't use CC_FPmode for ORDERED / UNORDERED.
20636         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20638         * config/epiphany/constraints.md (CnL): New constraint.
20639         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20640         * config/epiphany/predicates.md (add_operand): Allow 1024.
20642         * config/epiphany/epiphany.md (logical_op): New code iterator.
20643         (op_mnc): New code attribute.
20644         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20645         (mov_f+1, mov_f+2): New peephole2 patterns.
20647         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20648         (cstoresi4): Also allow re-use of zero result when doing a NE
20649         comparison to a non-zero operand.
20650         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20652         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20653         Use gen_addsi3_i / gen_subsi3_i.
20655 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20657         PR c++/34949
20658         PR c++/50243
20659         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20660         contain anything but clobbers, at most one __builtin_stack_restore,
20661         optionally debug stmts and final resx, and if it has at least one
20662         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20663         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20664         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20665         which isn't defaut definition, remove them.
20666         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20667         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20668         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20669         with MEM_REF LHS with SSA_NAME address.
20671 2013-04-08  Jeff Law  <law@redhat.com>
20673         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20675 2013-04-08  Richard Biener  <rguenther@suse.de>
20677         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20678         extra newline.
20679         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20680         determined vector type.
20681         (vect_analyze_data_refs): Likewise.
20682         (vect_get_new_vect_var): Adjust.
20683         (vect_create_destination_var): Preserve SSA name versions.
20684         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20685         not dump anything here.
20687 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20689         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20690         Add member lr_slot_known.
20691         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20692         if necessary.
20693         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20694         Remove code that sets lr_slot_offset according to what a previous
20695         version of epiphany_emit_save_restore used to do.
20696         (epiphany_emit_save_restore): When doing an lr save or restore,
20697         set/verify lr_slot_known and lr_slot_offset.
20699 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20701         PR target/54338
20702         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20703         in ALL_REGS.
20705 2013-04-08  Richard Biener  <rguenther@suse.de>
20707         * alias.c (find_base_term): Fix thinko in previous change.
20709 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20711         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20712         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20713         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20714         if possible to compute val.
20715         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20716         For QImode integers don't require anything about precision.  Use
20717         const_with_all_bytes_same to find out if the constant doesn't have
20718         repeated bytes in it.
20720 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20722         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20723         within mode size.
20725 2013-04-08  Marek Polacek  <polacek@redhat.com>
20727         PR rtl-optimization/48182
20728         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20729         value to 1.
20731 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20733         PR target/55487
20734         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20735         nuses, make sure we have a label.
20737 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20739         PR target/56843
20740         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20741         (rs6000_emit_swdiv_low_precision): Remove.
20742         (rs6000_emit_swdiv): Rewrite to handle between one and four
20743         iterations of Newton-Raphson generally; modify required number of
20744         iterations for some cases.
20745         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20747 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20749         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20750         set-but-unused variable.
20752         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20753         basic blocks of released function bodies garbage-collectable.
20755         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20756         (struct rtl_opt_pass): Add TODO_df_finish.
20758         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20760 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20762         * config/arm/constraints.md (q): New constraint.
20763         * config/arm/ldrdstrd.md: New file.
20764         * config/arm/arm.md (ldrdstrd.md) New include.
20765         (arm_movdi): Use "q" instead of "r" constraint
20766         for double-word memory access.
20767         (movdf_soft_insn): Likewise.
20768         * config/arm/vfp.md (movdi_vfp): Likewise.
20769         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20770         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20771         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20772         (mem_ok_for_ldrd_strd): Likewise.
20773         (output_move_double): Update assertion.
20775 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20777         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20779 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20781         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20782         define_insn_and_split.
20783         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20785 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20787         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20788         define_insn_and_split.
20789         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20790         (shiftsi3_compare): New pattern.
20791         (rrx): New pattern.
20792         * config/arm/unspecs.md (UNSPEC_RRX): New.
20794 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20796         * config/arm/arm.md (negdi_extendsidi): New pattern.
20797         (negdi_zero_extendsidi): Likewise.
20799 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20801         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20802         define_insn_and_split.
20803         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20804         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20806 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20808         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20809         define_insn_and_split.
20810         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20811         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20813 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20815         * config/arm/arm.md (subsi3_carryin): New pattern.
20816         (subsi3_carryin_const): Likewise.
20817         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20818         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20820 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20822         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20824 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20826         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20827         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20829 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20831         * config/arm/arm.c (arm_expand_builtin): Change fcode
20832         type to unsigned int.
20834 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20836         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20838 2013-04-04  Ian Lance Taylor  <iant@google.com>
20840         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20841         language standard.
20843 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20845         PR middle-end/56729
20846         * df-scan.c (df_insn_delete): Disable failing assert.
20848 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20850         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20851         New function prototype.
20852         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20853         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20854         (arm_builtin_vectorized_function): New function.
20856 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20858         * config/arm/arm_neon_builtins.def: New file.
20859         * config/arm/arm.c (neon_builtin_data): Move contents to
20860         arm_neon_builtins.def.
20861         (enum arm_builtins): Include neon builtin definitions.
20862         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20863         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20865 2013-04-04  Marek Polacek  <polacek@redhat.com>
20867         PR tree-optimization/48186
20868         * predict.c (maybe_hot_frequency_p): Return false if
20869         HOT_BB_FREQUENCY_FRACTION is 0.
20870         (cgraph_maybe_hot_edge_p): Likewise.
20872 2013-04-04  Richard Biener  <rguenther@suse.de>
20874         PR tree-optimization/56826
20875         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20876         more accurately.
20878 2013-04-04  Richard Biener  <rguenther@suse.de>
20880         PR tree-optimization/56213
20881         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20882         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20884 2013-04-04  Richard Biener  <rguenther@suse.de>
20886         PR tree-optimization/56837
20887         * tree-loop-distribution.c (classify_partition): For non-zero
20888         values require that the value has the same precision as its
20889         mode to be useful as memset value.
20891 2013-04-03  Nick Clifton  <nickc@redhat.com>
20893         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20894         (fmssf4): Use fmsf.s on E3V5 architectures.
20895         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20896         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20898 2013-04-03  Jeff Law  <law@redhat.com>
20900         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20901         (lra-eliminations.o): Likewise.
20903 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20905         * gcov-io.c (compute_working_sets): Moved most of body of old
20906         compute_working_sets here from profile.c.
20907         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20908         (gcov_working_set_t): Moved typedef here from basic-block.h
20909         (compute_working_set): Declare.
20910         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20911         (get_working_sets): Renamed from compute_working_set,
20912         replace most of body with call to new compute_working_sets.
20913         (get_exec_counts): Replace call to compute_working_sets
20914         to get_working_sets.
20915         * profile.h (get_working_sets): Renamed from compute_working_set.
20916         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20917         to get_working_sets.
20918         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20919         * gcov-dump.c (dump_working_sets): New function.
20921 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20923         * hwint.c (sext_hwi, zext_hwi): New functions.
20924         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20925         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20926         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20927         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20928         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20929         (sext_hwi, zext_hwi): New functions.
20931 2013-04-03  Jeff Law  <law@redhat.com>
20933         PR tree-optimization/56799
20934         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20935         back test for widening conversion erroneously dropped in prior change.
20937 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20939         PR target/56809
20940         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20941         instead of next_real_insn.
20943 2013-04-03  Marek Polacek  <polacek@redhat.com>
20945         PR sanitizer/55702
20946         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20948 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20950         PR target/56809
20951         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20952         next_real_insn.
20953         (thumb1_output_casesi): Likewise.
20954         (thumb2_output_casesi): Likewise.
20956 2013-04-03  Richard Biener  <rguenther@suse.de>
20958         PR tree-optimization/56817
20959         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
20960         Split out ...
20961         (tree_unroll_loops_completely_1): ... new function to manually
20962         walk the loop tree, properly defering outer loops of unrolled
20963         loops to later iterations.
20965 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20967         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
20968         (vectorizable_load): Likewise.
20969         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
20970         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
20972 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20974         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
20975         BIT_FIELD_REF.
20977 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
20979         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
20981 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
20983         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
20985 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20987         PR tree-optimization/56790
20988         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
20989         folding.
20991 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20993         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
20994         Handle VEC_MERGE.
20995         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
20996         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
20997         equal arguments.
20999 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21001         PR c/19449
21002         * tree.h (force_folding_builtin_constant_p): New decl.
21003         * builtins.c (force_folding_builtin_constant_p): New variable.
21004         (fold_builtin_constant_p): Fold immediately also if
21005         force_folding_builtin_constant_p.
21007 2013-04-03  Richard Biener  <rguenther@suse.de>
21009         PR tree-optimization/56812
21010         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21011         DRs of the same interleaving chain are independent.
21013 2013-04-02  Jason Merrill  <jason@redhat.com>
21015         * gdbinit.in (pbb): Use debug fn.
21017 2013-04-02  Lawrence Crowl  <crowl@google.com>
21019         * sese.h (struct ivtype_map_elt_s): Remove unused.
21020         (extern debug_ivtype_map): Remove unused.
21021         (extern eq_ivtype_map_elts): Remove unused.
21022         * sese.c (debug_ivtype_map): Removed unused.
21023         (debug_ivtype_map_1): Removed unused.
21024         (debug_ivtype_elt): Remove unused.
21025         (eq_ivtype_map_elts): Remove unused.
21028 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21030         PR target/52790
21031         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21032         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21033         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21034         function.
21035         (legitimize_pe_coff_symbol): Likewise.
21036         (is_imported_p): New helper-function.
21037         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21038         for Windows x64 targets.
21039         (ix86_expand_prologue): Optimize for pe-coff targets.
21040         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21041         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21042         medium/large code-model.
21043         (legitimize_pic_address): Likewise.
21044         (legitimize_tls_address): Likewise.
21045         (ix86_expand_call): Likewise.
21046         (x86_output_mi_thunk): Likewise.
21047         (get_dllimport_decl): Add new beimport argument.
21048         (construct_plt_address): Don't assert for x64 pe-coff targets.
21049         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21050         targets.
21051         (SYMBOL_FLAG_STUBVAR): New macro.
21052         (SYMBOL_REF_STUBVAR_P): Likewise.
21053         * config/i386/winnt.c (stub_list): New structure.
21054         (stub_head): New local variable.
21055         (i386_pe_record_stub): New function.
21056         (i386_pe_file_end): Emit refptr-stubs.
21058 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21060         PR rtl-optimization/56745
21061         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21062         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21064         PR c++/34949
21065         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21066         and both of them are MEM_REFs, just compare first argument for
21067         equality and attempt to deal even with differing offsets.
21069         PR c++/34949
21070         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21071         of gimple_clobber_p to be MEM_REF.
21072         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21073         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21074         after gimplification.
21075         * asan.c (get_mem_ref_of_assignment): Don't instrument
21076         gimple_clobber_p stmts.
21077         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21078         gimple_clobber_p stmt if they have MEM_REF lhs and
21079         are dead because of another gimple_clobber_p stmt.
21080         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21081         gimple_clobber_p stmts like debug stmts.
21082         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21083         that refer to unused VAR_DECLs or uninitialized values.
21084         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21085         gimple_clobber_p stmts if they refer to removed parameters.
21086         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21087         formatting.
21089 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21091         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21092         using SWI48 mode attribute.
21094 2013-04-02  Wei Mi  <wmi@google.com>
21096         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21097         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21098         *<rotate_insn><mode>3_mask in i386.md.
21100 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21102         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21104 2013-04-02  Richard Biener  <rguenther@suse.de>
21106         PR tree-optimization/56778
21107         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21108         Runtime alias tests are not supported for gather loads.
21109         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21110         stmts referenced from SSA operands before updating SSA form.
21112 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21113             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21115         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21116         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21117         * config/arm/cortex-a53.md: New file.
21118         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21119         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21120         * config/arm/arm.c (arm_issue_rate): Likewise.
21121         * config/arm/arm-tune.md: Regenerate
21122         * config/arm/arm-tables.opt: Regenerate.
21123         * config/arm/arm-cores.def: Add cortex-a53.
21125 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21127         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21128         non-static link.
21130 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21132         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21133         scalar load/store operations using B/H registers.
21134         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21136 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21138         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21139         scalar move.
21140         * config/aarch64/aarch64.c
21141         (aarch64_simd_scalar_immediate_valid_for_move): New.
21142         * config/aarch64/aarch64-protos.h
21143         (aarch64_simd_scalar_immediate_valid_for_move): New.
21144         * config/aarch64/constraints.md (Dh, Dq): New.
21145         * config/aarch64/iterators.md (hq): New.
21147 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21149         * reorg.c (get_branch_condition): Deal with conditional returns.
21150         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21152 2013-04-01  Wei Mi  <wmi@google.com>
21154         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21155         Truncate operand 2 using %b asm operand modifier.
21156         (*<shift_insn><mode>3_mask): Ditto.
21157         (*<rotate_insn><mode>3_mask): Ditto.
21159 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21161         PR middle-end/56798
21162         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21164 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21166         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21167         of next_real_insn.
21168         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21170 2013-03-30  Lawrence Crowl  <crowl@google.com>
21172         * dse.c (clear_alias_sets): Remove never set.
21173         (disqualified_clear_alias_sets): Remove never set.
21174         (clear_alias_mode_pool): Remove never set.
21175         (dse_step0): Remove condition that is never true.
21176         (canon_address): Remove condition that is never true.
21177         (dse_step7): Remove condition that is never true.
21178         (rest_of_handle_dse): Remove condition that is never true.
21179         (rest_of_handle_dse::did_global): Remove never read from above.
21180         (dse_step2_spill): Remove never called from above.
21181         (dse_step5_spill): Remove never called from above.
21183 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21185         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21186         JUMP_TABLE_DATA changes.
21187         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21188         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21189         (Insns) <jump_table_data>: New entry.
21190         * doc/tm.texi: Regenerate.
21192         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21194         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21195         for table jump at the end of a basic block using tablejump_p.
21196         * targhooks.c (default_invalid_within_doloop): Likewise.
21197         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21198         target hook implementation that is identical to the default hook.
21199         (rs6000_invalid_within_doloop): Remove.
21201         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21202         unused variable from tablejump_p call.
21204         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21205         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21206         (INSN_DELETED_P): Likewise.
21207         (emit_jump_table_data): New prototype.
21208         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21209         after 4th as unused.
21210         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21211         * sched-vis.c (print_insn): Likewise.
21212         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21213         insn for compatibility with back ends that use next_active_insn to
21214         identify jump table data.
21215         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21216         (remove_insn): Likewise.
21217         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21218         to be emitted.
21219         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21220         (emit_jump_table_data): New function.
21222         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21223         basic block, a JUMP_TABLE_DATA never is.
21224         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21225         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21226         off from code handling real insns.
21227         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21228         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21229         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21230         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21231         is not a NONDEBUG_INSN_P.
21232         * ira-costs.c (scan_one_insn): Likewise.
21233         * jump.c (mark_all_labels): Likewise.
21234         (mark_jump_label_1): Likewise.
21235         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21236         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21237         a basic block.
21238         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21239         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21240         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21241         JUMP_TABLE_DATA_P insns.
21242         (calculate_elim_costs_all_insns): Likewise.
21243         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21244         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21245         (delete_output_reload): Code style fixups.
21246         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21247         insn flags on this non-insn.
21248         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21249         as scheduling barriers, for pre-change compatibility.
21250         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21251         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21253         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21254         redundant JUMP_TABLE_DATA_P test.
21255         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21256         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21257         (frv_for_each_packet): Likewise.
21258         * config/i386/i386.c (min_insn_size): Likewise.
21259         (ix86_avoid_jump_mispredicts): Likewise.
21260         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21261         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21262         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21263         (mips16_insn_length): Robustify.
21264         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21265         (mips16_split_long_branches): Likewise.
21266         * config/pa/pa.c (pa_combine_instructions): Likewise.
21267         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21268         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21269         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21270         as contributing to pool range lengths.
21271         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21272         Remove redundant JUMP_TABLE_DATA_P test.
21273         (sh_loop_align): Likewise.
21274         (split_branches): Likewise.
21275         (sh_insn_length_adjustment): Likewise.
21276         * config/spu/spu.c (get_branch_target): Likewise.
21278 2013-03-29  Jan Hubicka  <jh@suse.cz>
21280         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21281         gcov streaming; stream hot bb threshold to ltrans.
21282         * predict.c (get_hot_bb_threshold): Break out from ....
21283         (maybe_hot_count_p): ... here.
21284         (set_hot_bb_threshold): New function.
21285         * lto-section-in.c (lto_section_name): Add profile.
21286         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21287         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21288         and data-streamer.h
21289         (histogram_entry): New structure.
21290         (histogram, histogram_pool): New global vars.
21291         (histogram_hash): New structure.
21292         (histogram_hash::hash): New method.
21293         (histogram_hash::equal): Likewise.
21294         (account_time_size): New function.
21295         (cmp_counts): New function.
21296         (dump_histogram): New function.
21297         (ipa_profile_generate_summary): New function.
21298         (ipa_profile_write_summary): New function.
21299         (ipa_profile_read_summary): New function.
21300         (ipa_profile): Decide on threshold.
21301         (pass_ipa_profile): Add ipa_profile_write_summary and
21302         ipa_profile_read_summary.
21303         * Makefile.in (ipa.o): Update dependencies.
21304         * lto-streamer.h (LTO_section_ipa_profile): New section.
21306 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21308         * tree.h (VAR_P): New.
21310 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21312         PR lto/56777
21313         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21315 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21317         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21318         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21319         (control_flow_insn_p): Likewise.
21320         * cfgrtl.c (duplicate_insn_chain): Likewise.
21321         * final.c (get_attr_length_1): Likewise.
21322         (shorten_branches): Likewise.
21323         (final_scan_insn): Likewise.
21324         * function.c (instantiate_virtual_regs): Likewise.
21325         * gcse.c (insert_insn_end_basic_block): Likewise.
21326         * ira-costs.c (scan_one_insn): Likewise.
21327         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21328         * lra.c (check_rtl): Likewise.
21329         * reload1.c (elimination_costs_in_insn): Likewise.
21330         * reorg.c (follow_jumps): Likewise.
21332         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21333         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21334         (thumb_far_jump_used_p): Likewise.
21335         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21336         (workaround_speculation): Likewise.
21337         (add_sched_insns_for_speculation): Likewise.
21338         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21339         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21340         (frv_for_each_packet): Likewise.
21341         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21342         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21343         (final_emit_insn_group_barriers): Likewise.
21344         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21345         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21346         (mips16_insn_length): Likewise.
21347         * config/pa/pa.c (pa_reorg): Likewise.
21348         (pa_combine_instructions): Likewise.
21349         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21350         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21351         (sh_reorg): Likewise.
21352         (split_branches): Likewise.
21353         * config/spu/spu.c (get_branch_target): Likewise.
21355         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21356         JUMP_TABLE_DATA_P.
21358 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21360         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21361         Fix declaration name.
21363 2013-03-28  Lawrence Crowl  <crowl@google.com>
21365         * graphds.h (struct graph.indicies): Remove unused.
21366         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21367         (SCOP_ORIGINAL_PDDRS): Remove unused.
21368         * sese.h (extern insert_loop_close_phis): Removed unused.
21369         (extern insert_guard_phis): Removed unused.
21370         (extern ivtype_map_elt_info): Removed unused.
21371         (new_ivtype_map_elt): Removed unused.
21372         * sese.c (ivtype_map_elt_info): Removed unused.
21374 2013-03-28  Lawrence Crowl  <crowl@google.com>
21376         * Makefile.in: Add several missing include dependences.
21377         (DUMPFILE_H): New.
21378         (test-dump.o): New.  This object is not added to any executable,
21379         but is present for ad-hoc testing.
21380         * bitmap.c
21381         (debug (const bitmap_head_def &)): New.
21382         (debug (const bitmap_head_def *)): New.
21383         * bitmap.h
21384         (extern debug (const bitmap_head_def &)): New.
21385         (extern debug (const bitmap_head_def *)): New.
21386         * cfg.c
21387         (debug (edge_def &)): New.
21388         (debug (edge_def *)): New.
21389         * cfghooks.c
21390         (debug (basic_block_def &)): New.
21391         (debug (basic_block_def *)): New.
21392         * dumpfile.h
21393         (dump_node (const_tree, int, FILE *)): Correct source file.
21394         * dwarf2out.c
21395         (debug (die_struct &)): New.
21396         (debug (die_struct *)): New.
21397         * dwarf2out.h
21398         (extern debug (die_struct &)): New.
21399         (extern debug (die_struct *)): New.
21400         * gimple-pretty-print.c
21401         (debug (gimple_statement_d &)): New.
21402         (debug (gimple_statement_d *)): New.
21403         * gimple-pretty-print.h
21404         (extern debug (gimple_statement_d &)): New.
21405         (extern debug (gimple_statement_d *)): New.
21406         * ira-build.c
21407         (debug (ira_allocno_copy &)): New.
21408         (debug (ira_allocno_copy *)): New.
21409         (debug (ira_allocno &)): New.
21410         (debug (ira_allocno *)): New.
21411         * ira-int.h
21412         (extern debug (ira_allocno_copy &)): New.
21413         (extern debug (ira_allocno_copy *)): New.
21414         (extern debug (ira_allocno &)): New.
21415         (extern debug (ira_allocno *)): New.
21416         * ira-lives.c
21417         (debug (live_range &)): New.
21418         (debug (live_range *)): New.
21419         * lra-int.h
21420         (debug (lra_live_range &)): New.
21421         (debug (lra_live_range *)): New.
21422         * lra-lives.c
21423         (debug (lra_live_range &)): New.
21424         (debug (lra_live_range *)): New.
21425         * omega.c
21426         (debug (omega_pb_d &)): New.
21427         (debug (omega_pb_d *)): New.
21428         * omega.h
21429         (extern debug (omega_pb_d &)): New.
21430         (extern debug (omega_pb_d *)): New.
21431         * print-rtl.c
21432         (debug (const rtx_def &)): New.
21433         (debug (const rtx_def *)): New.
21434         * print-tree.c
21435         (debug_tree (tree): Move within file.
21436         (debug_raw (const tree_node &)): New.
21437         (debug_raw (const tree_node *)): New.
21438         (dump_tree_via_hooks (const tree_node *, int)): New.
21439         (debug (const tree_node &)): New.
21440         (debug (const tree_node *)): New.
21441         (debug_verbose (const tree_node &)): New.
21442         (debug_verbose (const tree_node *)): New.
21443         (debug_head (const tree_node &)): New.
21444         (debug_head (const tree_node *)): New.
21445         (debug_body (const tree_node &)): New.
21446         (debug_body (const tree_node *)): New.
21447         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21448         (debug (vec<tree, va_gc> &)): New.
21449         (debug (vec<tree, va_gc> *)): New.
21450         * rtl.h
21451         (extern debug (const rtx_def &)): New.
21452         (extern debug (const rtx_def *)): New.
21453         * sbitmap.c
21454         (debug_raw (simple_bitmap_def &)): New.
21455         (debug_raw (simple_bitmap_def *)): New.
21456         (debug (simple_bitmap_def &)): New.
21457         (debug (simple_bitmap_def *)): New.
21458         * sbitmap.h
21459         (extern debug (simple_bitmap_def &)): New.
21460         (extern debug (simple_bitmap_def *)): New.
21461         (extern debug_raw (simple_bitmap_def &)): New.
21462         (extern debug_raw (simple_bitmap_def *)): New.
21463         * sel-sched-dump.c
21464         (debug (vinsn_def &)): New.
21465         (debug (vinsn_def *)): New.
21466         (debug_verbose (vinsn_def &)): New.
21467         (debug_verbose (vinsn_def *)): New.
21468         (debug (expr_def &)): New.
21469         (debug (expr_def *)): New.
21470         (debug_verbose (expr_def &)): New.
21471         (debug_verbose (expr_def *)): New.
21472         (debug (vec<rtx> &)): New.
21473         (debug (vec<rtx> *)): New.
21474         * sel-sched-dump.h
21475         (extern debug (vinsn_def &)): New.
21476         (extern debug (vinsn_def *)): New.
21477         (extern debug_verbose (vinsn_def &)): New.
21478         (extern debug_verbose (vinsn_def *)): New.
21479         (extern debug (expr_def &)): New.
21480         (extern debug (expr_def *)): New.
21481         (extern debug_verbose (expr_def &)): New.
21482         (extern debug_verbose (expr_def *)): New.
21483         (extern debug (vec<rtx> &)): New.
21484         (extern debug (vec<rtx> *)): New.
21485         * sel-sched-ir.h
21486         (_list_iter_cond_expr): Make inline instead of static.
21487         * sreal.c
21488         (debug (sreal &)): New.
21489         (debug (sreal *)): New.
21490         * sreal.h
21491         (extern debug (sreal &)): New.
21492         (extern debug (sreal *)): New.
21493         * tree.h
21494         (extern debug_raw (const tree_node &)): New.
21495         (extern debug_raw (const tree_node *)): New.
21496         (extern debug (const tree_node &)): New.
21497         (extern debug (const tree_node *)): New.
21498         (extern debug_verbose (const tree_node &)): New.
21499         (extern debug_verbose (const tree_node *)): New.
21500         (extern debug_head (const tree_node &)): New.
21501         (extern debug_head (const tree_node *)): New.
21502         (extern debug_body (const tree_node &)): New.
21503         (extern debug_body (const tree_node *)): New.
21504         (extern debug (vec<tree, va_gc> &)): New.
21505         (extern debug (vec<tree, va_gc> *)): New.
21506         * tree-cfg.c
21507         (debug (struct loop &)): New.
21508         (debug (struct loop *)): New.
21509         (debug_verbose (struct loop &)): New.
21510         (debug_verbose (struct loop *)): New.
21511         * tree-dump.c: Add header dependence.
21512         * tree-flow.h
21513         (extern debug (struct loop &)): New.
21514         (extern debug (struct loop *)): New.
21515         (extern debug_verbose (struct loop &)): New.
21516         (extern debug_verbose (struct loop *)): New.
21517         * tree-data-ref.c
21518         (debug (data_reference &)): New.
21519         (debug (data_reference *)): New.
21520         (debug (vec<data_reference_p> &)): New.
21521         (debug (vec<data_reference_p> *)): New.
21522         (debug (vec<ddr_p> &)): New.
21523         (debug (vec<ddr_p> *)): New.
21524         * tree-data-ref.h
21525         (extern debug (data_reference &)): New.
21526         (extern debug (data_reference *)): New.
21527         (extern debug (vec<data_reference_p> &)): New.
21528         (extern debug (vec<data_reference_p> *)): New.
21529         (extern debug (vec<ddr_p> &)): New.
21530         (extern debug (vec<ddr_p> *)): New.
21531         * tree-ssa-alias.c
21532         (debug (pt_solution &)): New.
21533         (debug (pt_solution *)): New.
21534         * tree-ssa-alias.h
21535         (extern debug (pt_solution &)): New.
21536         (extern debug (pt_solution *)): New.
21537         * tree-ssa-alias.c
21538         (debug (_var_map &)): New.
21539         (debug (_var_map *)): New.
21540         (debug (tree_live_info_d &)): New.
21541         (debug (tree_live_info_d *)): New.
21542         * tree-ssa-alias.h
21543         (extern debug (_var_map &)): New.
21544         (extern debug (_var_map *)): New.
21545         (extern debug (tree_live_info_d &)): New.
21546         (extern debug (tree_live_info_d *)): New.
21548 2013-03-28  Jan Hubicka  <jh@suse.cz>
21550         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21552 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21554         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21555         record only when desired or required.
21557 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21559         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21560         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21562 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21564         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21565         (*andsi3_compare0_uxtw): New pattern.
21566         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21567         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21569 2013-03-28  Jan Hubicka  <jh@suse.cz>
21571         * data-streamer-in.c (streamer_read_gcov_count): New function.
21572         * gimple-streamer-out.c: Include value-prof.h.
21573         (output_gimple_stmt): Output histogram.
21574         (output_bb): Use streamer_write_gcov_count.
21575         * value-prof.c: Include data-streamer.h
21576         (dump_histogram_value): Add HIST_TYPE_MAX.
21577         (stream_out_histogram_value): New function.
21578         (stream_in_histogram_value): New function.
21579         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21580         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21581         * data-streamer-out.c (streamer_write_gcov_count): New function.
21582         (streamer_write_gcov_count_stream): New function.
21583         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21584         (lto_output_node): Likewise.
21585         (input_node, input_edge): Likewise.
21586         * lto-streamer-out.c (output_cfg): Update streaming.
21587         * lto-streamer-in.c (input_cfg): Likewise.
21588         * data-streamer.h (streamer_write_gcov_count,
21589         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21590         * gimple-streamer-in.c: Include value-prof.h
21591         (input_gimple_stmt): Input histograms.
21592         (input_bb): Update profile streaming.
21594 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21596         * genmodes.c (emit_max_int): New function.
21597         (emit_insn_modes_h): Added call to emit_max_function.
21598         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21599         Added doc.
21600         * machmode.def: Fixed comment.
21602 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21604         * combine.c (try_combine): Removed useless assert.
21605         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21607 2013-03-28  Marek Polacek  <polacek@redhat.com>
21608             Richard Biener  <rguenther@suse.de>
21610         PR tree-optimization/56695
21611         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21612         build signed result of a vector comparison.
21613         * tree-cfg.c (verify_gimple_comparison): Check that a result
21614         of a vector comparison has signed type.
21616 2013-03-28  Richard Biener  <rguenther@suse.de>
21618         PR tree-optimization/37021
21619         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21620         do not restrict gaps between groups.
21621         * tree-vect-stmts.c (vectorizable_load): Properly account for
21622         a gap between groups.
21624 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21626         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21627         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21628         is not enabled.
21630 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21632         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21633         * doc/extend.texi (Named Address Spaces): Ditto.
21634         (Variable Attributes): Ditto.
21636 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21638         * config.build: Add support for cygwin x64 target.
21639         * config.gcc: Likewise.
21640         * config.host: Likewise.
21641         * configure.ac: Likewise
21642         * configure: Regenerated.
21644 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21646         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21647         * config/i386/t-cygwin-w64: New file.
21648         * config/i386/cygwin-w64.h: New file.
21649         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21650         and add support for x64-cygwin target.
21651         (CPP_SPEC): Likewise.
21652         (CXX_WRAP_SPEC_LIST): Undefine before define.
21653         (LIBGCJ_SONAME): Use 15 as version.
21655 2013-03-27  Richard Biener  <rguenther@suse.de>
21657         PR tree-optimization/56716
21658         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21659         dumping for ref nodes.
21661 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21663         PR tree-optimization/55334
21664         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21665         restricted pointers to arrays.
21667 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21669         * Makefile.in (.SUFFIXES): Add .cc.
21670         (.c.o): Apply same recipe for implicit rule .cc.o.
21672 2013-03-27  Richard Biener  <rguenther@suse.de>
21674         PR tree-optimization/37021
21675         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21676         REALPART/IMAGPART_EXPRs around the supported refs.
21677         * tree-ssa-structalias.c (find_func_aliases): Assume that
21678         floating-point values are not used to transfer pointers.
21680 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21682         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21683         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21684         * doc/tm.texi: Regenerate.
21685         * targhooks.h (default_has_ifunc_p): New.
21686         * targhooks.c (default_has_ifunc_p): Ditto.
21687         * config/linux-protos.h: New file.
21688         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21689         hook for linux which disables support of indirect functions in android.
21690         * config/linux-android.c: New file.
21691         * config/t-linux-android.c: Ditto.
21692         * config.gcc: Added new object file linux-android.o.
21693         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21694         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21695         * varasm.c (do_assemble_alias): Likewise.
21696         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21697         doesn't support indirect functions.
21698         * configure: Regenerate.
21700 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21702         PR target/56102
21703         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21704         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21705         mult-word mode.
21707 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21709         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21711 2013-03-27  Terry Guo  <terry.guo@arm.com>
21713         * config/arm/arm-cores.def: Added core cortex-r7.
21714         * config/arm/arm-tune.md: Regenerated.
21715         * config/arm/arm-tables.opt: Regenerated.
21716         * doc/invoke.texi: Added entry for core cortex-r7.
21718 2013-03-27  Walter Lee  <walt@tilera.com>
21720         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21721         double-decrement of next_scratch_regno.
21723 2013-03-27  Walter Lee  <walt@tilera.com>
21725         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21726         input operands.
21727         (insn_v1mulus): Ditto.
21728         (insn_v2muls): Ditto.
21730 2013-03-27  Walter Lee  <walt@tilera.com>
21732         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21733         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21735 2013-03-27  Walter Lee  <walt@tilera.com>
21737         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21738         (*sibcall_value): Ditto.
21740 2013-03-27  Walter Lee  <walt@tilera.com>
21742         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21743         (insn_mnz_v8qi): ... this ...
21744         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21745         vector equivalent.
21746         (insn_v<n>mnz): Replaced by ...
21747         (insn_v1mnz): ... this ...
21748         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21749         equivalent.
21750         (insn_mz_<mode>): Replaced by ...
21751         (insn_mz_v8qi): ... this ...
21752         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21753         vector equivalent.
21754         (insn_v<n>mz): Replaced by ...
21755         (insn_v1mz): ... this ...
21756         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21757         equivalent.
21759 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21761         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21763 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21765         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21766         than fprintf with a non-constant, non-format string.
21768 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21770         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21771         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21772         operand 0 predicate.
21773         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21774         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21775         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21776         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21777         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21778         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21779         (mov<mode>_insv_1): Remove expander.  Merge insn with
21780         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21781         Use general_x64nomem_operand as operand 1 predicate.
21782         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21783         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21784         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21785         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21786         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21787         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21788         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21789         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21790         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21791         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21792         (general_x64nomem_operand): Ditto.
21794 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21796         * config/rtems.opt: Add -pthread option.
21798 2013-03-26  Richard Biener  <rguenther@suse.de>
21800         * alias.c (find_base_term): Avoid redundant and not used recursion.
21801         (base_alias_check): Get the initial base term from the caller.
21802         (true_dependence_1): Compute and pass base terms to base_alias_check.
21803         (write_dependence_p): Likewise.
21804         (may_alias_p): Likewise.
21806 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21808         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21809         PC-relative load in SI modes and above only.
21811 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21813         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21814         * config/arm/iwmmxt.md (WCGR0): Update.
21815         (WCGR1, WCGR2, WCGR3): Likewise.
21817 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21819         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21820         Use x64 and nox64 isa attributes.
21822 2013-03-26  Richard Biener  <rguenther@suse.de>
21824         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21825         alignment computations and rely on get_object_alignment_1
21826         for the !TYPE_P case.
21827         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21829 2013-03-26  Walter Lee  <walt@tilera.com>
21831         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21832         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21834 2013-03-25  Jeff Law  <law@redhat.com>
21836         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21837         check for INTEGRAL_TYPE_P that was missing due to checking in
21838         wrong version of prior patch.
21840 2013-03-25  Walter Lee  <walt@tilera.com>
21842         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21843         TILEGX_INSN_SHUFFLEBYTES1.
21844         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21845         shufflebytes1.
21846         (tilegx_builtins): Ditto.
21847         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21849 2013-03-25  Walter Lee  <walt@tilera.com>
21851         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21852         (floatunssisf2): New pattern.
21853         (floatsidf2): New pattern.
21854         (floatunssidf2): New pattern.
21856 2013-03-25  Walter Lee  <walt@tilera.com>
21858         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21859         tests for constraint J, K, N, P.
21861 2013-03-25  Walter Lee  <walt@tilera.com>
21863         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21864         Use indirect/pcrel encoding.
21865         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21866         Ditto.
21868 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21870         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21871         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21872         * config/mips/mips.c (mips_option_override): Set IMADD default.
21873         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21874         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21875         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21876         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21877         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21879 2013-03-25  Jeff Law  <law@redhat.com>
21881         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21882         slightly to avoid creating and folding useless trees.  Simplify
21883         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21885 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21887         * config/i386/i386.md (*zero_extendsidi2): Merge with
21888         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21889         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21890         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21891         targets.  Clarify comment.
21893 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21895         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21896         pass-through jump functions differently.
21897         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21898         up jump functions.
21900 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21902         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21903         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21904         process it.
21905         (ipa_get_indirect_edge_target): New function.
21906         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21907         ipa_get_indirect_edge_target.  Update all callers.
21908         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21909         ipa_get_indirect_edge_target_1 instead of calling
21910         ipa_get_indirect_edge_target.
21911         (create_specialized_node): Pass aggvlas to
21912         ipcp_discover_new_direct_edges.
21914 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21916         * config/arm/arm.md (f_sels, f_seld): New types.
21917         (*cmov<mode>): New pattern.
21918         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21919         predicate.
21921 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21923         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21924         POSIX-printf for mingw-hosted  builds.
21926 2013-03-25  Richard Biener  <rguenther@suse.de>
21928         PR middle-end/56694
21929         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21930         must-not-throw stmt location.
21932 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21934         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21935         Emit load-acquire versions when acq is true.
21936         (arm_emit_store_exclusive): Add rel parameter.
21937         Emit store-release versions when rel is true.
21938         (arm_split_compare_and_swap): Use acquire-release instructions
21939         instead.
21940         of barriers when appropriate.
21941         (arm_split_atomic_op): Likewise.
21942         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21943         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21944         (VUNSPEC_SLX): Likewise.
21945         (VUNSPEC_LDA): Likewise.
21946         (VUNSPEC_STL): Likewise.
21947         * config/arm/sync.md (atomic_load<mode>): New pattern.
21948         (atomic_store<mode>): Likewise.
21949         (arm_load_acquire_exclusive<mode>): Likewise.
21950         (arm_load_acquire_exclusivesi): Likewise.
21951         (arm_load_acquire_exclusivedi): Likewise.
21952         (arm_store_release_exclusive<mode>): Likewise.
21954 2013-03-25  Catherine Moore  <clm@codesourcery.com>
21956         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
21957         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
21958         * config/mip/predicates.md (lwsp_swsp_operand,
21959         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
21960         sb16_operand, db4_operand, db7_operand, ib3_operand,
21961         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
21962         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
21963         andi16_operand): New predicates.
21964         * config/mips/mips.md (compression): New attribute.
21965         (enabled): New attribute.
21966         (length): Consider compression in computing length.
21967         (shift_compression): New code attribute.
21968         (*add<mode>3): New operands. Record compression.
21969         (sub<mode>3): Likewise.
21970         (one_cmpl<mode>2): Likewise.
21971         (*and<mode>3): Likewise.
21972         (*ior<mode>3): Likewise.
21973         (unnamed pattern for xor): Likewise.
21974         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
21975         (*<optab><mode>3): Likewise.
21976         (*mov<mode>_internal: Likewise.
21977         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
21978         (mips_unsigned_immediate_p): New.
21979         (umips_lwsp_swsp_address_p): New.
21980         (m16_based_address_p): New.
21981         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
21982         (mips_unsigned_immediate_p): New prototype.
21983         (lwsp_swsp_address_p): New prototype.
21984         (m16_based_address_p): New prototype.
21985         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
21986         (mips_signed_immediate_p): New function.
21987         (m16_based_address_p): New function.
21988         (lwsp_swsp_address_p): New function.
21989         (mips_print_operand_punctuation): Recognize short delay slot insns
21990         for microMIPS.add<mode>3"
21992 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21994         PR target/56720
21995         * config/arm/iterators.md (v_cmp_result): New mode attribute.
21996         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
21998 2013-03-25  Richard Biener  <rguenther@suse.de>
22000         PR tree-optimization/56689
22001         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22002         any edge.
22004 2013-03-25  Richard Biener  <rguenther@suse.de>
22006         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22007         of bitmap.
22008         (memory_references): Likewise.
22009         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22010         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22011         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22012         (gather_mem_refs_in_loops): Fold into ...
22013         (analyze_memory_references): ... this.  Move initialization
22014         to tree_ssa_lim_initialize.
22015         (fill_always_executed_in): Rename to ...
22016         (fill_always_executed_in_1): ... this.
22017         (fill_always_executed_in): Move contains_call computation to
22018         this new function from ...
22019         (tree_ssa_lim_initialize): ... here.
22020         (tree_ssa_lim): Call fill_always_executed_in.
22022 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22024         * postreload.c (reload_combine): Fix code detecting returns.
22026 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22028         * function.c (emit_use_return_register_into_block): On cc0 targets,
22029         do not emit the sequence between cc0 setter and user.
22031 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22033         * config/i386/predicates.md (local_symbolic_operand): Interpret
22034         dll-imported symbols as none-local.
22036 2013-03-25  Richard Biener  <rguenther@suse.de>
22038         * tree-ssa-loop-im.c (struct depend): Remove.
22039         (struct lim_aux_data): Make depends a vec of gimples.
22040         (free_lim_aux_data): Adjust.
22041         (add_dependency): Likewise.
22042         (set_level): Likewise.
22044 2013-03-25  Richard Biener  <rguenther@suse.de>
22046         PR middle-end/56434
22047         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22048         the pointer returned by calls with ECF_MALLOC set.
22050 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22052         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22054 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22056         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22057         using MMXMODE mode iterator.
22058         (*move<mode>_internal): Merge with *movv2sf_internal and
22059         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22061 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22063         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22064         (record_last_mem_set_info): Likewise.
22066         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22067         of XNEWVEC followed by memset.
22068         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22070 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22072         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22073         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22074         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22075         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22076         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22077         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22078         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22079         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22080         BARRIER_P instead of GET_CODE.
22082 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22084         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22085         inaccuracy in the probing code.
22087         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22088         (ctrapdi4): Likewise.
22090 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22092         * calls.c (expand_call): Add missing guard to code handling return
22093         of non-BLKmode structures in MSB.
22094         * function.c (expand_function_end): Likewise.
22096 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22098         * combine.c (try_combine): Adjust comment.  Do not add the set of
22099         insn #0 if the destination indirectly is set or dies in insn #2.
22100         Tidy up code to distribute a new note.
22102 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22104         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22105         also for alternatives 16 and 17.
22107 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22109         * config/i386/sse.md (*mov<mode>_internal): Merge with
22110         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22111         Emit insn template depending on type attribute.  Use
22112         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22113         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22114         attribute calculation.  Remove unit attribute calculation.
22115         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22116         Set prefix_data16 attribute for DImode ssemov types.
22117         Use Ym instead of y for SSE-MMX conversion alternatives.
22118         Reorder operand constraints.
22120 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22122         * df.h (df_insn_delete): Adjust prototype.
22123         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22124         and let it decide whether mark the basic block dirty.
22125         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22126         * df-scan.c (df_insn_info_delete): New helper function, split
22127         off from df_insn_delete.
22128         (df_scan_free_bb_info): Use it.
22129         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22130         Likewise.
22131         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22132         that the insn is actually an insn and it has a non-NULL basic block.
22133         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22135 2013-03-22  Richard Biener  <rguenther@suse.de>
22137         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22138         dep_ref members.
22139         (mem_ref_alloc): Do not allocate them.
22140         (refs_independent_p): Do not query or maintain a cache.
22142 2013-03-22  Richard Biener  <rguenther@suse.de>
22144         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22145         (gather_mem_refs_in_loops): Do not compute it.
22146         (analyze_memory_references): Do not allocate it.
22147         (tree_ssa_lim_finalize): Do not free it.
22148         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22150 2013-03-22  Richard Biener  <rguenther@suse.de>
22152         * is-a.h (as_a): Use gcc_checking_assert.
22154 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22156         * config/aarch64/aarch64.c (aarch64_print_operand): New
22157         format specifier for printing a constant in hex.
22158         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22159         format specifier for printing second operand.
22161 2013-03-22  Richard Biener  <rguenther@suse.de>
22163         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22164         bitmaps.
22165         (gather_mem_refs_in_loops): Perform store accumulation here.
22166         (create_vop_ref_mapping_loop): Remove.
22167         (create_vop_ref_mapping): Likewise.
22168         (analyze_memory_references): Initialize refs_stored_in_loop.
22169         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22170         (record_indep_loop): Remove.
22171         (record_dep_loop): New function.
22172         (ref_indep_loop_p_1): Adjust to only walk over references
22173         in the loop, not its subloops.
22174         (ref_indep_loop_p): Rename to ...
22175         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22176         maintaining a more fine-grained cache.
22177         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22178         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22180 2013-03-22  Richard Biener  <rguenther@suse.de>
22182         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22183         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22184         aggregate mem_ref_loc.
22185         (free_mem_ref_locs): Inline into ...
22186         (memref_free): ... this and adjust.
22187         (mem_ref_alloc): Adjust.
22188         (mem_ref_locs_alloc): Remove.
22189         (record_mem_ref_loc): Adjust.
22190         (get_all_locs_in_loop): Rewrite into ...
22191         (for_all_locs_in_loop): ... this iterator.
22192         (rewrite_mem_ref_loc): New functor.
22193         (rewrite_mem_refs): Use for_all_locs_in_loop.
22194         (sm_set_flag_if_changed): New functor.
22195         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22196         (ref_always_accessed): New functor.
22197         (ref_always_accessed_p): Use for_all_locs_in_loop.
22199 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22201         * tree-pass.h (PROP_gimple_lvec): New.
22202         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22203         (init_optimization_passes): Move pass_lower_vector.
22204         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22205         PROP_gimple_lvec.
22206         (pass_lower_vector): Provide PROP_gimple_lvec.
22207         (pass_lower_vector_ssa): Likewise.
22208         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22210 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22212         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22214 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22216         * config/i386/i386.md (*movdi_internal): Disparage slightly
22217         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22218         conversion alternatives.
22220 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22222         PR middle-end/48087
22223         * diagnostic.def (DK_WERROR): New kind.
22224         * diagnostic.h (werrorcount): Define.
22225         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22226         promoted to DK_ERROR, increment DK_WERROR counter instead of
22227         DK_ERROR counter.
22228         * toplev.c (toplev_main): Call print_ignored_options even if
22229         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22230         even if just werrorcount is non-zero.
22232         PR debug/55608
22233         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22234         on failure.
22235         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22236         (string_cst_pool_decl): New function.
22237         (optimize_one_addr_into_implicit_ptr): New function.
22238         (resolve_addr_in_expr): Optimize DWARF location expression
22239         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22240         which doesn't live in memory, but has DW_AT_location or
22241         DW_AT_const_value, or refers to a string literal, into
22242         DW_OP_GNU_implicit_pointer.
22243         (optimize_location_into_implicit_ptr): New function.
22244         (resolve_addr): If removing DW_AT_location of a variable because
22245         it was DW_OP_addr of address of the variable, but the variable doesn't
22246         live in memory, try to emit const value attribute for the initializer.
22248 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22250         * tree.h (VECTOR_TYPE_P): New macro.
22251         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22252         TYPE_MODE): Use it.
22253         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22254         VEC_COND_EXPR cannot be lvalues.
22255         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22257 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22259         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22260         Restrict the transformation to equal modes.
22262 2013-03-21  Richard Biener  <rguenther@suse.de>
22264         PR tree-optimization/39326
22265         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22266         (MEM_ANALYZABLE): Adjust.
22267         (record_mem_ref_loc): Move bitmap ops ...
22268         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22269         unanalyzable refs, do not record locations for it.
22270         (analyze_memory_references): Allocate ref zero as shared
22271         unanalyzable ref.
22272         (refs_independent_p): Do not test for unanalyzed mems here.
22273         (ref_indep_loop_p_1): Special-case disambiguation against
22274         the unanalyzed ref.
22275         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22277 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22279         * config/arm/arm-protos.h (tune_params): Add
22280         prefer_neon_for_64bits field.
22281         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22282         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22283         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22284         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22285         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22286         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22287         (arm_option_override): Handle -mneon-for-64bits new option.
22288         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22289         (prefer_neon_for_64bits): Declare new variable.
22290         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22291         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22292         (arch_enabled): Handle new arch types. Remove support for onlya8
22293         and nota8.
22294         (one_cmpldi2): Use new arch names.
22295         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22296         * config/arm/arm.opt (mneon-for-64bits): Add option.
22297         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22298         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22299         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22300         of onlya8.
22301         * doc/invoke.texi (-mneon-for-64bits): Document.
22303 2013-03-21  Richard Biener  <rguenther@suse.de>
22305         PR tree-optimization/39326
22306         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22307         (sort_bbs_in_loop_postorder_cmp): New function.
22308         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22310 2013-03-21  Richard Biener  <rguenther@suse.de>
22312         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22313         (vect_insert_into_interleaving_chain): Likewise.
22314         (vect_drs_dependent_in_basic_block): Inline ...
22315         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22316         split out from ...
22317         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22318         (vect_check_interleaving): Simplify.
22319         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22320         (vect_slp_analyze_data_ref_dependences): ... this new function.
22321         (dr_group_sort_cmp): New function.
22322         (vect_analyze_data_ref_accesses): Compute data-reference groups
22323         here instead of in vect_analyze_data_ref_dependence.  Use
22324         a more efficient algorithm.
22325         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22326         vect_slp_analyze_data_ref_dependences.  Call
22327         vect_analyze_data_ref_accesses earlier.
22328         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22329         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22330         (vect_slp_analyze_data_ref_dependences): New prototype.
22332 2013-03-21  Richard Biener  <rguenther@suse.de>
22334         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22335         ref is stored in the loop.
22336         (find_refs_for_sm): Walk only over all stores.
22337         (store_motion_loop): Allocate from lim_bitmap_obstack.
22338         (store_motion): Likewise.
22340 2013-03-21  Richard Biener  <rguenther@suse.de>
22342         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22343         Update virtual SSA form.
22345 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22347         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22348         * configure: Regenerate.
22349         * config.in: Regenerate.
22350         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22351         if !HAVE_LD_EH_FRAME_CIEV3.
22353 2013-03-21  Richard Biener  <rguenther@suse.de>
22355         * tree-cfg.c (verify_expr_no_block): New function.
22356         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22357         nor DECL_VALUE_EXPR have locations with associated blocks.
22358         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22359         (clear_unused_block_pointer): Remove code dealing with
22360         blocks in DECL_DEBUG_EXPR locations.
22362 2013-03-21  Richard Biener  <rguenther@suse.de>
22364         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22365         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22366         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22367         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22368         instead of DECL_DEBUG_EXPR_IS_FROM.
22369         * gimplify.c (gimplify_modify_expr): Likewise.
22370         * tree-cfg.c (verify_expr_location_1): Likewise.
22371         * tree-complex.c (create_one_component_var): Likewise.
22372         * tree-sra.c (create_access_replacement): Likewise.
22373         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22374         (clear_unused_block_pointer): Likewise.
22375         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22376         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22377         * var-tracking.c (var_debug_decl): Likewise.
22378         (track_expr_p): Likewise.
22379         * tree-inline.c (add_local_variables): Likewise.  Set
22380         DECL_HAS_DEBUG_EXPR_P after copying it.
22381         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22382         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22384 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22386         PR bootstrap/56656
22387         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22388         * configure: Regenerate.
22389         * config.in: Regenerate.
22390         * config/i386/i386.md (*movdf_internal): Use
22391         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22392         movd instead of movq mnemonic for interunit moves.
22393         (*movdi_internal): Ditto.
22395 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22397         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22398         (abd<mode>_3): New pattern.
22399         (aba<mode>_3): New pattern.
22400         (fabd<mode>_3): New pattern.
22402 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22404         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22405         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22406         occurrence of REGISTER_PREFIX as its empty string.
22408 2013-03-20  Jeff Law  <law@redhat.com>
22410         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22411         addititional equivalences for equality comparisons between an SSA_NAME
22412         and a constant where the SSA_NAME was set from a widening conversion.
22414 2013-03-20  Walter Lee  <walt@tilera.com>
22416         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22418 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22420         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22421         depending on type attribute.
22422         (*movti_internal): Ditto.
22423         (*movtf_internal): Ditto.
22424         (*movxf_internal): Ditto.
22425         (*movdf_internal): Ditto.
22426         (*movsf_internal): Ditto.
22428 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22430         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22431         maybe_vex for sselog1 and ssemov types.
22432         (*movdi_internal): Reorder operand constraints.
22433         (*movsi_internal): Ditto.  Set prefix attribute to
22434         maybe_vex for sselog1 and ssemov types.
22435         (*movtf_internal): Set prefix attribute to maybe_vex
22436         for sselog1 and ssemov types.
22437         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22438         DImode ssemov types.  Reorder operand constraints.
22439         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22440         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22441         attribute for SImode ssemov types.  Reorder operand constraints.
22443 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22445         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22446         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22448 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22450         * config/rs6000/predicates.md (indexed_address, update_address_mem
22451         update_indexed_address_mem): New predicates.
22452         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22453         attribute for load/store instructions.
22454         * config/rs6000/dfp.md (movsd_store): Likewise.
22455         (movsd_load): Likewise.
22456         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22457         (unnamed HI->DI extend define_insn): Likewise.
22458         (unnamed SI->DI extend define_insn): Likewise.
22459         (unnamed QI->SI extend define_insn): Likewise.
22460         (unnamed QI->HI extend define_insn): Likewise.
22461         (unnamed HI->SI extend define_insn): Likewise.
22462         (unnamed HI->SI extend define_insn): Likewise.
22463         (extendsfdf2_fpr): Likewise.
22464         (movsi_internal1): Likewise.
22465         (movsi_internal1_single): Likewise.
22466         (movhi_internal): Likewise.
22467         (movqi_internal): Likewise.
22468         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22469         attribute for load/store instructions.
22470         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22471         instructions.
22472         (mov<mode>_softfloat): Likewise.
22473         (mov<mode>_hardfloat32): Likewise.
22474         (mov<mode>_hardfloat64): Likewise.
22475         (mov<mode>_softfloat64): Likewise.
22476         (movdi_internal32): Likewise.
22477         (movdi_internal64): Likewise.
22478         (probe_stack_<mode>): Likewise.
22480 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22482         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22483         floating point, and decimal floating point to reload iterator.
22485         * config/rs6000/constraints.md (wl constraint): New constraints to
22486         return FLOAT_REGS if certain options are used to reduce the number
22487         of separate patterns that exist in the file.
22488         (wx constraint): Likewise.
22489         (wz constraint): Likewise.
22491         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22492         -mdebug=reg, print wg, wl, wx, and wz constraints.
22493         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22494         Initialize the reload functions for 64-bit binary/decimal floating
22495         point types.
22496         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22497         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22498         create the buffer on the stack to overcome not having a 32-bit
22499         load and store.
22500         (rs6000_emit_move): Likewise.
22501         (rs6000_secondary_memory_needed_rtx): Likewise.
22502         (rs6000_alloc_sdmode_stack_slot): Likewise.
22503         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22504         via xxlxor, just like DFmode 0.0.
22506         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22507         define as 1 if we are running on a power7 or newer.
22508         (enum r6000_reg_class_enum): Add new constraints.
22510         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22511         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22512         with other moves by using conditional constraits (wg).  Use LFIWZX
22513         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22514         (movsd splitter): Likewise.
22515         (movsd_hardfloat): Likewise.
22516         (movsd_softfloat): Likewise.
22518         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22519         binary and decimal floating point moves.
22520         (fmove_ok): New attributes to combine binary and decimal floating
22521         point moves, and to combine power6x (mfpgpr) moves along normal
22522         floating moves.
22523         (real_value_to_target): Likewise.
22524         (f32_lr): Likewise.
22525         (f32_lm): Likewise.
22526         (f32_li): Likewise.
22527         (f32_sr): Likewise.
22528         (f32_sm): Likewise.
22529         (f32_si): Likewise.
22530         (movsf): Combine binary and decimal floating point moves.  Combine
22531         power6x (mfpgpr) moves with other moves by using conditional
22532         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22533         (mov<mode> for SFmode/SDmode); Likewise.
22534         (SFmode/SDmode splitters): Likewise.
22535         (movsf_hardfloat): Likewise.
22536         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22537         (movsf_softfloat): Likewise.
22538         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22540         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22541         wx and wz constraints.
22543         * config/rs6000/constraints.md (wg constraint): New constraint to
22544         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22546         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22547         constraint.
22549         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22550         -mdebug=reg, print wg, wl, wx, and wz constraints.
22551         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22552         Initialize the reload functions for 64-bit binary/decimal floating
22553         point types.
22554         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22555         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22556         create the buffer on the stack to overcome not having a 32-bit
22557         load and store.
22558         (rs6000_emit_move): Likewise.
22559         (rs6000_secondary_memory_needed_rtx): Likewise.
22560         (rs6000_alloc_sdmode_stack_slot): Likewise.
22561         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22562         via xxlxor, just like DFmode 0.0.
22564         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22565         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22566         with other moves by using conditional constraits (wg).  Use LFIWZX
22567         and STFIWX for loading SDmode on power7.
22568         (movdd splitters): Likewise.
22569         (movdd_hardfloat32): Likewise.
22570         (movdd_softfloat32): Likewise.
22571         (movdd_hardfloat64_mfpgpr): Likewise.
22572         (movdd_hardfloat64): Likewise.
22573         (movdd_softfloat64): Likewise.
22575         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22576         64-bit binary and decimal floating point moves.
22577         (FMOVE64X): Likewise.
22578         (movdf): Combine 64-bit binary and decimal floating point moves.
22579         Combine power6x (mfpgpr) moves with other moves by using
22580         conditional constraits (wg).
22581         (mov<mode> for DFmode/DDmode): Likewise.
22582         (DFmode/DDmode splitters): Likewise.
22583         (movdf_hardfloat32): Likewise.
22584         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22585         (movdf_softfloat32): Likewise.
22586         (movdf_hardfloat64_mfpgpr): Likewise.
22587         (movdf_hardfloat64): Likewise.
22588         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22589         (movdf_softfloat64): Likewise.
22590         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22591         (reload_<mode>_load): Move to later in the file so they aren't in
22592         the middle of the floating point move insns.
22593         (reload_<mode>_store): Likewise.
22595         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22596         constraint.
22598         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22599         constraint if -mdebug=reg.
22600         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22601         Enable using dd reload support if needed.
22603         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22604         binary and decimal floating point moves in rs6000.md.
22605         (movtd_internal): Likewise.
22607         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22608         decimal floating point moves.
22609         (movtf): Likewise.
22610         (movtf_internal): Likewise.
22611         (mov<mode>_internal, TDmode/TFmode): Likewise.
22612         (movtf_softfloat): Likewise.
22613         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22615         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22616         movdi_internal64, using wg constraint for move direct operations.
22617         (movdi_internal64): Likewise.
22619         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22620         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22621         the various virtual registers. Use GPR/FPR first/last values,
22622         instead of hard coding the register numbers.  Print which modes
22623         have reload functions registered.
22624         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22625         settings before/after setting cpu, target and subtarget settings.
22626         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22627         and for secondary reload failures in rs6000_secondary_reload_inner.
22628         (rs6000_secondary_reload_fail): Likewise.
22629         (rs6000_secondary_reload_inner): Likewise.
22631         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22632         macros for first/last GPR and FPR registers.
22633         (LAST_GPR_REGNO): Likewise.
22634         (FIRST_FPR_REGNO): Likewise.
22635         (LAST_FPR_REGNO): Likewise.
22637         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22638         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22639         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22640         (vcond<mode><mode>): Likewise.
22641         (vcondu<mode><mode>): Likewise.
22642         (vector_gtu<mode>): Likewise.
22643         (vector_gte<mode>): Likewise.
22644         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22645         to prevent the compiler from converting DImode operations to TImode.
22646         (ior<mode>3): Likewise.
22647         (and<mode>3): Likewise.
22648         (one_cmpl<mode>2): Likewise.
22649         (nor<mode>3): Likewise.
22650         (andc<mode>3): Likewise.
22652         * config/rs6000/constraints.md (wt constraint): New constraint
22653         that returns VSX_REGS if TImode is allowed in VSX registers.
22655         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22656         constant under VSX.
22658         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22659         similar to TImode, but it is restricted to being in the GPRs.
22661         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22662         TImode to occupy a single VSX register.
22664         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22665         -mvsx-timode for power7/power8.
22666         (power7 cpu): Likewise.
22667         (power8 cpu): Likewise.
22669         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22670         sure that TFmode/TDmode take up two registers if they are ever
22671         allowed in the upper VSX registers.
22672         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22673         registers.
22674         (rs6000_init_hard_regno_mode_ok): Likewise.
22675         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22676         constraint.  Print if LRA is turned on.
22677         (rs6000_option_override_internal): Give an error if -mvsx-timode
22678         and VSX is not enabled.
22679         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22680         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22681         to reg+offset addressing.  Use PTImode when checking offset
22682         addresses for validity.
22683         (reg_offset_addressing_ok_p): Likewise.
22684         (rs6000_legitimate_offset_address_p): Likewise.
22685         (rs6000_legitimize_address): Likewise.
22686         (rs6000_legitimize_reload_address): Likewise.
22687         (rs6000_legitimate_address_p): Likewise.
22688         (rs6000_eliminate_indexed_memrefs): Likewise.
22689         (rs6000_emit_move): Likewise.
22690         (rs6000_secondary_reload): Likewise.
22691         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22692         reloads to fpr registers to continue to use reg+offset addressing,
22693         but 64-bit reloads to altivec registers need reg+reg addressing.
22694         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22695         it.  Treat LO_SUM like a PLUS operation.
22696         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22697         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22698         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22699         registers to share a register with a smaller sized type, since VSX
22700         puts scalars in the upper 64-bits.
22701         (print_operand): Add support for PTImode.
22702         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22703         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22704         registers, but don't have arithmetic support.
22705         (rs6000_memory_move_cost): Add test for VSX.
22706         (rs6000_opt_masks): Add -mvsx-timode.
22708         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22709         for TImode.
22710         (VSs): Likewise.
22711         (VSr): Use wt constraint for TImode.
22712         (VSv): Drop TImode support.
22713         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22714         (vsx_movti_64bit): Likewise.
22715         (vsx_movti_32bit): Likewise.
22716         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22717         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22718         one '?' on the appropriate output constraint.  Do not allow TImode
22719         logical operations on 32-bit systems.
22720         (vsx_ior<mode>3): Likewise.
22721         (vsx_xor<mode>3): Likewise.
22722         (vsx_one_cmpl<mode>2): Likewise.
22723         (vsx_nor<mode>3): Likewise.
22724         (vsx_andc<mode>3): Likewise.
22725         (vsx_concat_<mode>): Likewise.
22726         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22728         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22729         OPTION_MASK_VSX_TIMODE.
22730         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22731         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22733         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22734         (TI2 iterator): New iterator for TImode, PTImode.
22735         (wd mode attribute): Add values for vector types.
22736         (movti_string): Replace TI move operations with operations for TImode
22737         and PTImode.  Add support for TImode being allowed in VSX registers.
22738         (mov<mode>_string, TImode/PTImode): Likewise.
22739         (movti_ppc64): Likewise.
22740         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22741         (TI mode splitters): Likewise.
22743         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22744         constraint.
22746 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22748         PR tree-optimization/56355
22749         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22750         Also handle integers with undefined overflow.
22752 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22753             Maciej W. Rozycki  <macro@codesourcery.com>
22754             Tom de Vries  <tom@codesourcery.com>
22755             Nathan Sidwell  <nathan@codesourcery.com>
22756             Iain Sandoe  <iain@codesourcery.com>
22757             Nathan Froyd  <froydnj@codesourcery.com>
22758             Chao-ying Fu  <fu@mips.com>
22760         * doc/extend.texi: (micromips, nomicromips, nocompression):
22761         Document new function attributes.
22762         * doc/invoke.texi (minterlink-compressed, mmicromips,
22763         m14k, m14ke, m14kec): Document new options.
22764         (minterlink-mips16): Update documentation.
22765         * doc/md.texi (ZC, ZD): Document new constraints.
22766         * configure.ac (gcc_cv_as_micromips): Check if linker
22767         supports the .set micromips directive.
22768         * configure: Regenerate.
22769         * config.in: Regenerate.
22770         * config/mips/mips-tables.opt: Regenerate.
22771         * config/mips/micromips.md: New file.
22772         * constraints.md (ZC, ZD): New constraints.
22773         * config/mips/predicates.md (movep_src_register): New predicate.
22774         (movep_src_operand): New predicate.
22775         (non_volatile_mem_operand): New predicate.
22776         * config/mips/mips.md (multimem): New type.
22777         (length): Differentiate between 17-bit and 18-bit branch offsets.
22778         (MOVEP1, MOVEP2): New mode iterator.
22779         (mov_<load>l): Use ZC constraint.
22780         (mov_<load>r): Likewise.
22781         (mov_<store>l): Likewise.
22782         (mov_<store>r): Likewise.
22783         (*branch_equality<mode>_inverted): Add microMIPS support.
22784         (*branch_equality<mode>): Likewise.
22785         (*jump_absolute): Likewise.
22786         (indirect_jump_<mode>): Likewise.
22787         (tablejump_<mode>): Likewise.
22788         (<optab>_internal): Likewise.
22789         (sibcall_internal): Likewise.
22790         (sibcall_value_internal): Likewise.
22791         (prefetch): Use constraint ZD.
22792         * config/mips/mips.opt (minterlink-compressed): New option.
22793         (minterlink-mips16): Now an alias for minterlink-compressed.
22794         (mmicromips): New option.
22795         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22796         (compare_and_swap_12): Likewise.
22797         (sync_add<mode>): Likewise.
22798         (sync_<optab>_12): Likewise.
22799         (sync_old_<optab>_12): Likewise.
22800         (sync_new_<optab>_12): Likewise.
22801         (sync_nand_12): Likewise.
22802         (sync_old_nand_12): Likewise.
22803         (sync_new_nand_12): Likewise.
22804         (sync_sub<mode>): Likewise.
22805         (sync_old_add<mode>): Likewise.
22806         (sync_old_sub<mode>): Likewise.
22807         (sync_new_add<mode>): Likewise.
22808         (sync_new_sub<mode>): Likewise.
22809         (sync_<optab><mode>): Likewise.
22810         (sync_old_<optab><mode>): Likewise.
22811         (sync_new_<optab><mode>): Likewise.
22812         (sync_nand<mode>): Likewise.
22813         (sync_old_nand<mode>): Likewise.
22814         (sync_new_nand<mode>): Likewise.
22815         (sync_lock_test_and_set<mode>): Likewise.
22816         (test_and_set_12): Likewise.
22817         (atomic_compare_and_swap<mode>): Likewise.
22818         (atomic_exchange<mode>_llsc): Likewise.
22819         (atomic_fetch_add<mode>_llsc): Likewise.
22820         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22821         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22822         (umips_save_restore_pattern_p): Likewise.
22823         (umips_load_store_pair_p): Likewise.
22824         (umips_output_load_store_pair): Likewise.
22825         (umips_movep_target_p): Likewise.
22826         (umips_12bit_offset_address_p): Likewise.
22827         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22828         (mips_base_mips16): Rename this...
22829         (mips_base_compression_flags): ...to this. Update all uses.
22830         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22831         (mips_mips16_decl_p): Delete.
22832         (mips_nomips16_decl_p): Delete.
22833         (mips_get_compress_on_flags): New function.
22834         (mips_get_compress_off_flags): New function.
22835         (mips_get_compress_mode): New function.
22836         (mips_get_compress_on_name): New function.
22837         (mips_get_compress_off_name): New function.
22838         (mips_insert_attributes): Support multiple compression types.
22839         (mips_merge_decl_attributes): Likewise.
22840         (umips_12bit_offset_address_p): New function.
22841         (mips_start_function_definition): Emit .set micromips directive.
22842         (mips_call_may_need_jalx_p): New function.
22843         (mips_function_ok_for_sibcall): Add microMIPS support.
22844         (mips_print_operand_punctuation): Support short delay slots and
22845         compact jumps.
22846         (umips_swm_mask, umips_swm_encoding): New.
22847         (umips_build_save_restore): New function.
22848         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22849         (was_mips16_p): Remove.
22850         (old_compression_mode): New.
22851         (mips_set_compression_mode): New function.
22852         (mips_set_current_function): Add microMIPS support.
22853         (mips_option_override): Likewise.
22854         (umips_save_restore_pattern_p): New function.
22855         (umips_output_save_restore): New function.
22856         (umips_load_store_pair_p_1): New function.
22857         (umips_load_store_pair_p): New function.
22858         (umips_output_load_store_pair_1): New function.
22859         (umips_output_load_store_pair): New function.
22860         (umips_movep_target_p) New function.
22861         (mips_prepare_pch_save): Add microMIPS support.
22862         * config/mips/mips.h (TARGET_COMPRESSION): New.
22863         (TARGET_CPU_CPP_BUILTINS): Update macro
22864         to use new compression flags and to support microMIPS.
22865         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22866         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22867         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22868         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22869         (ASM_SPEC): Support mmicromips and mno-micromips.
22870         (M16STORE_REG_P): New macro.
22871         (MIPS_CALL): Support TARGET_MICROMIPS.
22872         (MICROMIPS_J): New macro.
22873         (mips_base_mips16): Rename this...
22874         (mips_base_compression_flags): ...to this.
22875         (UMIPS_12BIT_OFFSET_P): New macro.
22876         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22877         (MULTILIB_DIRNAMES): Likewise.
22878 2013-03-20  Richard Biener  <rguenther@suse.de>
22880         PR tree-optimization/56661
22881         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22882         the result does not have to be distinct.
22884 2013-03-20  Richard Biener  <rguenther@suse.de>
22886         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22887         remap_gimple_op_r.
22889 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22890             Steven Bosscher  <steven@gcc.gnu.org>
22892         PR rtl-optimization/56605
22893         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22895 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22897         PR bootstrap/56656
22898         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22899         that require movd instead of movq.
22901 2013-03-20  Richard Biener  <rguenther@suse.de>
22903         * tree-ssa-structalias.c (struct variable_info): Add pointer
22904         to the first field of an aggregate with sub-vars.  Make
22905         this and the pointer to the next subfield its ID.
22906         (vi_next): New function.
22907         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22908         storedanything_id, integer_id): Increment by one.
22909         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22910         get_call_clobber_vi): Adjust.
22911         (solution_set_expand): Simplify and speedup.
22912         (solution_set_add): Inline into ...
22913         (set_union_with_increment): ... this.  Adjust accordingly.
22914         (do_sd_constraint): Likewise.
22915         (do_ds_constraint): Likewise.
22916         (do_complex_constraint): Simplify.
22917         (build_pred_graph): Adjust.
22918         (solve_graph): Likewise.  Simplify and speedup.
22919         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22920         get_constraint_for_component_ref, get_constraint_for_1,
22921         first_vi_for_offset, first_or_preceding_vi_for_offset,
22922         create_function_info_for, create_variable_info_for_1,
22923         create_variable_info_for, intra_create_variable_infos): Adjust.
22924         (init_base_vars): Push NULL for ID zero.
22925         (compute_points_to_sets): Adjust.
22927 2013-03-20  Richard Biener  <rguenther@suse.de>
22929         * cfgloop.c (verify_loop_structure): Streamline and avoid
22930         ICEing on corrupt loop tree.
22931         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22932         loop tree.
22934 2013-03-20  Richard Biener  <rguenther@suse.de>
22936         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22937         check whether an SSA update is needed.
22939 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22941         * config/mips/constraints.md (T): Rename to...
22942         (Yf): ...this.
22943         (U): Rename to...
22944         (Yd): ...this.
22945         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22946         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22948 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22950         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22951         (*subsi3_carryin_uxtw): Likewise.
22953 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22955         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
22956         (*rorsi3_insn_uxtw): Likewise.
22958 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22960         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
22961         (*extrsi5_insn_uxtw): Likewise.
22963 2013-03-19  Richard Biener  <rguenther@suse.de>
22965         PR tree-optimization/56273
22966         * passes.c (init_optimization_passes): Move second VRP after DOM.
22968 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
22970         * config/i386/i386.md (*movti_internal): Merge from
22971         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
22972         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
22973         nox64 isa attributes.
22975 2013-03-18  Richard Biener  <rguenther@suse.de>
22977         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
22978         (unite): Likewise.
22979         (merge_node_constraints): Likewise.
22980         (build_succ_graph): Likewise.
22981         (valid_graph_edge): Inline into single caller.
22982         (unify_nodes): Likewise.  Use bitmap_set_bit return value
22983         and cache varinfo.
22984         (scc_visit): Fix formatting and variable use.
22985         (do_sd_constraint): Use gcc_checking_assert.
22986         (do_ds_constraint): Likewise.
22987         (do_complex_constraint): Likewise.
22988         (condense_visit): Likewise.  Cleanup.
22989         (dump_pred_graph): New function.
22990         (perform_var_substitution): Dump the pred-graph before
22991         variable substitution.
22992         (find_equivalent_node): Use gcc_checking_assert.
22993         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
22995 2013-03-18  Richard Biener  <rguenther@suse.de>
22997         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
22998         Remove cond_expr_stmt_list argument and do not gimplify the
22999         built expression.
23000         (vect_loop_versioning): Adjust.
23001         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23002         Cleanup to use less temporaries.
23003         (vect_create_data_ref_ptr): Cleanup.
23005 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23007         PR tree-optimization/56635
23008         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23009         require types_compatible_p types.
23011 2013-03-18  Nick Clifton  <nickc@redhat.com>
23013         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23014         spurious backslash.
23016         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23017         Add missing line to comment describing function.
23019 2013-03-18  Richard Biener  <rguenther@suse.de>
23021         PR tree-optimization/56210
23022         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23023         Handle string / character search functions.
23024         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23026 2013-03-18  Richard Biener  <rguenther@suse.de>
23028         PR middle-end/56483
23029         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23030         and implement properly.
23031         * gimple.h (gimple_cond_single_var_p): Remove.
23033 2013-03-18  Richard Biener  <rguenther@suse.de>
23035         * tree-data-ref.h (find_data_references_in_loop): Declare.
23036         * tree-data-ref.c (get_references_in_stmt): Use a stack
23037         vector pre-allocated in the callers.
23038         (find_data_references_in_stmt): Adjust.
23039         (graphite_find_data_references_in_stmt): Likewise.
23040         (create_rdg_vertices): Likewise.
23041         (find_data_references_in_loop): Export.
23042         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23043         Compute dependences here...
23044         (vect_analyze_data_refs): ...not here.  When we encounter
23045         a non-vectorizable data reference in basic-block vectorization
23046         truncate the data reference vector.  Do not bother to
23047         fixup data-dependence information for gather loads.
23048         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23049         of data references, as reported.
23051 2013-03-18  Richard Biener  <rguenther@suse.de>
23053         PR tree-optimization/3713
23054         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23055         has_constants and expr.
23056         (stmt_has_constants): Properly valueize SSA names when deciding
23057         whether the stmt has constants.
23059 2013-03-18  Richard Biener  <rguenther@suse.de>
23061         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23062         whole function when there is nothing to do.
23063         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23064         * tree-vectorizer.c (vectorize_loops): Update virtual and
23065         loop-closed SSA once.
23066         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23068 2013-03-18  Richard Biener  <rguenther@suse.de>
23070         PR middle-end/56113
23071         * domwalk.c (bb_postorder): New global static.
23072         (cmp_bb_postorder): New function.
23073         (walk_dominator_tree): Replace scheme imposing an order for
23074         visiting dominator sons by one sorting them at the time they
23075         are pushed on the stack.
23077 2013-03-18  Richard Biener  <rguenther@suse.de>
23079         PR tree-optimization/39326
23080         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23081         (struct mem_ref): Replace mem member with ao_ref typed member.
23082         (MEM_ANALYZABLE): Adjust.
23083         (memref_eq): Likewise.
23084         (mem_ref_alloc): Likewise.
23085         (gather_mem_refs_stmt): Likewise.
23086         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23087         (execute_sm_if_changed_flag_set): Adjust.
23088         (execute_sm): Likewise.
23089         (ref_always_accessed_p): Likewise.
23090         (refs_independent_p): Likewise.
23091         (can_sm_ref_p): Likewise.
23093 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23095         PR c/56566
23096         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23097         return 1 even for !unsignedp.
23099 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23101         * config/i386/i386.md (isa): Add x64 and nox64.
23102         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23103         (*pushtf): Enable *roF alternative for x64 isa only.
23104         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23105         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23106         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23107         (*movtf_internal): Merge from *movtf_internal_rex64 and
23108         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23109         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23110         nox64 isa attributes.
23111         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23112         nox64 isa attributes.
23113         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23115 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23117         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23119 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23121         PR target/56640
23122         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23123         class names.  Remove trailing comma after "ALL_REGS".
23125 2013-03-16  Jan Hubicka  <jh@suse.cz>
23127         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23128         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23129         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23130         of cgraph_get_create_node.
23131         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23133 2013-03-16  Jason Merrill  <jason@redhat.com>
23135         PR debug/49090
23136         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23137         with DW_AT_default_value.
23139 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23141         * BASE-VER: Set to 4.9.0.
23143 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23145         PR target/56619
23146         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23147         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23148         Document _x* TSX intrinsics.
23150 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23151             David Holsgrove  <david.holsgrove@xilinx.com>
23153         * configure.ac: Add MicroBlaze TLS support detection.
23154         * configure: Regenerate.
23155         * config/microblaze/microblaze-protos.h
23156         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23157         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23158         * config/microblaze/microblaze.c (microblaze_address_type): Add
23159         ADDRESS_TLS and tls_reloc address types.
23160         (microblaze_address_info): Add tls_reloc.
23161         (TARGET_HAVE_TLS): Define.
23162         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23163          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23164          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23165         load_tls_operand,  microblaze_call_tls_get_addr,
23166         microblaze_legitimize_tls_address): New functions.
23167         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23168         (get_base_reg): Use microblaze_tls_symbol_p.
23169         (microblaze_classify_address): Handle TLS.
23170         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23171         label_mentioned_p and microblaze_tls_referenced_p.
23172         (microblaze_legitimize_address): Handle TLS.
23173         (microblaze_address_insns): Handle ADDRESS_TLS.
23174         (pic_address_needs_scratch): Handle TLS.
23175         (print_operand_address): Handle TLS.
23176         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23177         (microblaze_expand_move): Handle TLS.
23178         (microblaze_legitimate_constant_p): Check
23179         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23180         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23181         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23182         (PIC_OFFSET_TABLE_REGNUM): Set.
23183         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23184         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23185         (addsi3, movsi_internal2, movdf_internal): Update constraints
23186         * config/microblaze/predicates.md (arith_plus_operand): Define
23187         (move_operand): Redefine as move_src_operand,
23188         check microblaze_tls_referenced_p.
23190 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23192         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23193         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23195 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23197         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23198         CC mode for AND.
23200 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23202         PR tree-optimization/53265
23203         * common.opt (Waggressive-loop-optimizations): New option.
23204         * tree-ssa-loop-niter.c: Include tree-pass.h.
23205         (do_warn_aggressive_loop_optimizations): New function.
23206         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23207         if number_of_latch_executions returned constant.
23208         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23209         early.  If number_of_latch_executions returned constant, set
23210         nb_iterations_upper_bound back to it.
23211         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23212         field.
23213         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23214         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23216         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23217         (MULTILIB_OSDIRNAMES): Set.
23218         * genmultilib: If defaultosdirname doesn't start with :: , set
23219         defaultosdirname2 instead, clear it and emit two . multilib_raw
23220         entries instead of just one.
23222 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23224         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23225         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23226         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23227         (SUBTARGET_OVERRIDE_OPTIONS): New.
23229 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23231         PR target/49880
23232         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23233         (musermode): Convert to Var(TARGET_USERMODE).
23234         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23235         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23236         * config/sh/sh.c (sh_option_override): Use
23237         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23238         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23239         condition.
23240         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23241         TARGET_SH4.
23242         (udivsi3_i4_single, divsi3_i4_single): Use
23243         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23245 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23247         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23248         default setting.
23250 2013-03-13  Richard Biener  <rguenther@suse.de>
23252         PR tree-optimization/56608
23253         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23254         calls when vectorizing basic-blocks.
23256 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23258         PR plugins/45078
23259         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23260         tm_file.
23262 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23264         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23266 2013-03-11  Jan Hubicka  <jh@suse.cz>
23268         PR lto/56557
23269         * lto-streamer-out.c (output_symbol_p): Skip references from
23270         constructors of external variables.
23272 2013-03-11  Jan Hubicka  <jh@suse.cz>
23274         PR middle-end/56571
23275         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23276         from pseudos.
23277         * emit-rtl.c (verify_rtx_sharing): Likewise.
23278         (copy_insn_1): Likewise.
23279         * rtl.c (copy_rtx): Likewise.
23281 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23283         PR target/56591
23284         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23285         output_operand_lossage message.
23287 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23289         PR target/56470
23290         * arm.c (shift_op): Validate RTL pattern on the fly.
23291         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23292         the RTL.
23294 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23296         PR target/56347
23297         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23298         new call patterns that clobber %fr12.
23299         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23300         split and postreload patterns.
23301         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23302         registers %fr12 and %fr12R as call used.
23304 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23306         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23307         (canon_address, record_store, replace_read, check_mem_read_rtx,
23308         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23309         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23310         rest_of_handle_dse): Likewise.
23312 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23314         PR middle-end/56524
23315         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23316         Add base_optabs.
23317         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23318         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23319         (save_optabs_if_changed): Replace with...
23320         (init_tree_optimization_optabs): ...this.
23321         * optabs.c (save_optabs_if_changed): Rename to...
23322         (init_tree_optimization_optabs): ...this.  Take the optimization node
23323         as argument.  Do nothing if the base optabs are already correct.
23324         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23325         to recompute optabs.
23326         * function.h (function): Remove optabs field.
23327         * function.c (invoke_set_current_function_hook): Call
23328         init_tree_optimization_optabs.  Use the result to initialize
23329         this_fn_optabs.
23331 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23333         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23334         if GTMA_HAS_NO_INSTRUMENTATION.
23335         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23336         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23337         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23338         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23339         GTMA_HAS_NO_INSTRUMENTATION.
23341 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23343         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23344         libasan_preinit.o.
23346 2013-03-08  Marek Polacek  <polacek@redhat.com>
23347             Jakub Jelinek  <jakub@redhat.com>
23349         PR tree-optimization/56478
23350         * predict.c (is_comparison_with_loop_invariant_p): Change the
23351         type of loop_step to tree.
23352         (predict_loops): Adjust.
23353         (predict_iv_comparison): Perform the computations on double_ints.
23355 2013-03-08  Richard Biener  <rguenther@suse.de>
23357         PR tree-optimization/56570
23358         * tree-cfg.c (verify_expr_location_1): Verify locations for
23359         DECL_DEBUG_EXPR.
23360         * tree-sra.c (create_access_replacement): Strip locations
23361         from DECL_DEBUG_EXPRs.
23363 2013-03-08  Richard Biener  <rguenther@suse.de>
23365         * tree-inline.c (expand_call_inline): Do not associate
23366         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23367         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23369 2013-03-08  Richard Biener  <rguenther@suse.de>
23371         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23372         or block changes with -Og.  Fix for location / block encoding
23373         changes and PHI arguments with locations.
23375 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23377         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23378         for all counters.
23379         (struct output_info): Likewise.
23380         (register_overhead): Remove bad gcc_assert.
23381         (bitmap_find_bit): If there is only a single bitmap element, do not
23382         count a miss as a search.
23383         (print_statistics): Update for counter type changes.
23384         (dump_bitmap_statistics): Likewise.  Print headers such that they
23385         are properly lined up with the printed counters.
23387 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23389         PR tree-optimization/56559
23390         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23391         check that it has only a single use.
23393 2013-03-07  Richard Biener  <rguenther@suse.de>
23395         * doc/invoke.texi (fwhole-program): Discourage use in combination
23396         with -flto.
23398 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23400         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23402         PR tree-optimization/56539
23403         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23404         instead of GSI_CONTINUE_LINKING as last argument to
23405         force_gimple_operand_gsi.  Adjust function comment.
23407         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23408         aarch64-cores.def.
23410         PR middle-end/56548
23411         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23412         promoted mode, convert the result back to the original mode.
23414 2013-03-06  Richard Biener  <rguenther@suse.de>
23416         PR middle-end/56294
23417         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23418         (insert_updated_phi_nodes_compare_uids): New function.
23419         (update_ssa): Sort symbols_to_rename after UID before
23420         traversing it to insert PHI nodes.
23422 2013-03-06  Richard Biener  <rguenther@suse.de>
23424         PR middle-end/50494
23425         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23426         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23428         Revert
23429         2013-02-13  Richard Biener  <rguenther@suse.de>
23431         PR lto/50494
23432         * varasm.c (output_constant_def_1): Get the decl representing
23433         the constant as argument.
23434         (output_constant_def): Wrap output_constant_def_1.
23435         (make_decl_rtl): Use output_constant_def_1 with the decl
23436         representing the constant.
23437         (build_constant_desc): Optionally re-use a decl already
23438         representing the constant.
23439         (tree_output_constant_def): Adjust.
23441 2013-03-06  Joey Ye  <joey.ye@arm.com>
23443         PR lto/50293
23444         * gcc.c (convert_white_space): New function.
23445         (main): Handles white space in function name.
23447 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23449         PR target/56529
23450         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23451         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23452         to SH_DIV_CALL_TABLE for TARGET_SH2.
23453         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23454         list.
23455         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23456         call-table options.
23458 2013-03-05  Sterling Augustine  <saugustine@google.com>
23459             Cary Coutant  <ccoutant@google.com>
23461         PR debug/55364
23462         * dwarf2out.c (resolve_addr): Don't call
23463         remove_loc_list_addr_table_entries a second time for the same
23464         expression.
23466 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23468         PR debug/56510
23469         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23470         (avoid_complex_debug_insns): New function.
23471         (expand_debug_locations): Call it.
23473         PR rtl-optimization/56484
23474         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23475         lifetimes of hard registers on small register class machines.
23477 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23479         * config/microblaze/microblaze-protos.h: Rename
23480         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23481         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23482         fast_interrupt.
23483         (microblaze_fast_interrupt_function_p): New function.
23484         (microblaze_is_interrupt_handler): Rename to
23485         microblaze_is_interrupt_variant and add fast_interrupt check.
23486         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23487         (save_restore_insns): Likewise.
23488         (compute_frame_size): Likewise.
23489         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23490         (microblaze_globalize_label): Likewise.
23491         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23492         * config/microblaze/microblaze.md: Use wrapper
23493         microblaze_is_interrupt_variant.
23495 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23497         * sdbout.c (sdbout_one_type): Switch to current function's section
23498         supporting cold/hot.
23500 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23502         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23503         -mxl-reorder.
23505 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23507         PR middle-end/56461
23508         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23509         if VALGRIND_GET_VBITS is defined, temporarily make object
23510         memory all defined, and restore previous valgrind addressability
23511         and definability afterwards.  Free this_object at the end.
23513         PR middle-end/56461
23514         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23515         right before calling lra_create_live_ranges, also call it
23516         when clearing live_p.  Only call lra_clear_live_ranges
23517         at the end if live_p.
23519         PR middle-end/56461
23520         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23522 2013-03-05  Richard Biener  <rguenther@suse.de>
23524         PR tree-optimization/56521
23525         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23526         value-id.
23528 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23530         PR c++/55135
23531         * except.h (remove_unreachable_eh_regions): New prototype.
23532         * except.c (remove_eh_handler_splicer): New function, split out
23533         of remove_eh_handler.
23534         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23535         warning about running it on many EH regions one at a time.
23536         (remove_unreachable_eh_regions_worker): New function, walk the
23537         EH tree in depth-first order and remove non-marked regions.
23538         (remove_unreachable_eh_regions): New function.
23539         * tree-eh.c (mark_reachable_handlers): New function, split out
23540         from remove_unreachable_handlers.
23541         (remove_unreachable_handlers): Use mark_reachable_handlers and
23542         remove_unreachable_eh_regions.
23543         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23544         and remove_unreachable_eh_regions.
23546 2013-03-05  Richard Biener  <rguenther@suse.de>
23548         PR middle-end/56525
23549         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23550         not freeing them until the end.
23552 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23554         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23556 2013-03-05  Richard Biener  <rguenther@suse.de>
23558         PR tree-optimization/56270
23559         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23560         of loads after scheduling an SLP instance.
23562 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23564         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23565         tic6x.exp.
23566         (check_gcc_parallelize): Run guality.exp as a separate job from
23567         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23568         struct-layout-1.exp with stackalign.exp.
23570         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23572         PR middle-end/56461
23573         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23574         load_index sbitmap even if some bit in it isn't set.
23576         PR middle-end/56461
23577         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23578         (discover_iteration_bound_by_body_walk): Change queues to
23579         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23580         spelling in comment.  Call safe_push on queues[bound_index] directly.
23581         Release queues[queue_index] in every iteration unconditionally.
23582         Release bounds vector.
23584         PR middle-end/56461
23585         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23586         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23587         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23588         inner_phis vector.
23590 2013-03-05  Richard Biener  <rguenther@suse.de>
23592         PR lto/56515
23593         * tree-inline.c (remap_blocks_to_null): New function.
23594         (expand_call_inline): When expanding a call stmt without
23595         an associated block inline remap all callee blocks to NULL.
23597 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23599         PR rtl-optimization/56494
23600         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23601         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23602         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23604         PR middle-end/56461
23605         * sel-sched-ir.c (free_sched_pools): Release
23606         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23607         if succs_info_pool.max_top isn't -1.
23609         PR bootstrap/56509
23610         * opts.c (opts_obstack, opts_concat): Moved to...
23611         * opts-common.c (opts_obstack, opts_concat): ... here.
23613 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23615         PR middle-end/56461
23616         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23618 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23620         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23621         all appropriate places.
23623 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23625         PR tree-optimization/56424
23626         * ipa-split.c (split_function): Do not set the RSO flag if result is
23627         not by reference and its type is a register type.
23629 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23631         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23632         (microblaze_legitimate_pic_operand): Likewise
23633         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23634         new function microblaze_legitimate_pic_operand
23635         * config/microblaze/microblaze-protos.h
23636         (microblaze_legitimate_pic_operand): Declare.
23638 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23640         * config/microblaze/predicates.md (call_insn_simple_operand):
23641         New predicate for supported rtx code types.
23642         * config/microblaze/microblaze.md (call_internal1): Use
23643         call_insn_simple_operand predicate.
23645 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23647         PR middle-end/56461
23648         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23649         partitions.ordered_remove.
23651         PR middle-end/56461
23652         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23653         vec_oprnds0.create (1) for modifier == NONE.
23655         PR middle-end/56461
23656         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23657         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23658         vec_oprnds1 right before pushing anything to it for
23659         scalar_shift_arg.
23661         PR middle-end/56461
23662         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23663         set nbbs to 0 instead of having separate code path.
23664         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23665         instead of false as last argument if returning NULL.
23667 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23669         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23670         the attribute is now called "target" instead of "option".
23671         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23672         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23673         attribute/pragma name for TARGET_OPTION_VALID_P and
23674         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23675         * doc/tm.texi: Regenerated.
23677 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23679         * config/microblaze/microblaze.c:
23680         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23681         * config/microblaze/microblaze.h: Add -mxl-reorder to
23682         DRIVER_SELF_SPECS.
23683         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23684         instructions emitted if TARGET_REORDER.
23685         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23686         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23687         separately.
23689 2013-03-01  Xinliang David Li  <davidxl@google.com>
23691         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23692         walk length.
23694 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23696         PR middle-end/56461
23697         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23698         vector even when returning true.  Fix up function comment formatting.
23700         PR middle-end/56461
23701         * ira-build.c (ira_loop_nodes_count): New variable.
23702         (create_loop_tree_nodes): Initialize it.
23703         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23705         PR middle-end/56461
23706         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23707         method on dr_chain and result_chain.
23708         * tree-vect-stmts.c (vectorizable_store): Only call
23709         result_chain.create if j == 0.
23711         PR middle-end/56461
23712         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23713         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23714         before overwriting it.
23716 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23718         * doc/extended.texi (C Extensions): Change order in @menu
23719         to match @node.
23720         (Other MIPS Built-in Functions): Move last MIPS entry before
23721         "picoChip Built-in Functions".
23722         (SH Built-in Functions): Move after RX Built-in Functions.
23723         * doc/gcc.texi (Introduction): Change order in @menu
23724         to match @node.
23725         * doc/md.texi (Constraints): Ditto.
23726         * gty.texi (Type Information): Ditto.
23727         (User-provided marking routines for template types): Make
23728         subsection.
23729         * doc/invoke.texi (AArch64 Options): Move before
23730         "Adapteva Epiphany Options".
23732 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23733             Jakub Jelinek  <jakub@redhat.com>
23735         PR sanitizer/56454
23736         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23737         no_address_safety_analysis attribute.
23738         * doc/extend.texi (no_address_safety_attribute): Rename to
23739         no_sanitize_address attribute, mention no_address_safety_analysis
23740         attribute as deprecated alias.
23742 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23744         PR middle-end/56461
23745         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23746         type to vec<vec<tree> > *.
23747         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23748         to be vec<tree> instead of vec<tree> *, set vec_defs
23749         to vNULL and call vec_defs.create (number_of_vects), adjust other
23750         uses of vec_defs.
23751         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23752         vectorizable_condition): Adjust vect_get_slp_defs callers.
23754 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23756         * config/aarch64/aarch64.c
23757         (aarch64_float_const_representable): Remove unused variable.
23759 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23761         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23763 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23765         * config/aarch64/aarch64-builtins.c
23766         (aarch64_init_simd_builtins): Make static.
23768 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23770         * config/aarch64/aarch64.c
23771         (aarch64_simd_make_constant): Make static.
23773 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23775         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23776         with no initialization to the RHS of debug statements.
23778 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23780         PR tree-optimization/56294
23781         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23782         Adjust dumping.
23783         (get_access_replacement): Do not call create_access_replacement.
23784         Assert a replacement exists.
23785         (get_repl_default_def_ssa_name): Create the replacement declaration
23786         itself.
23788 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23790         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23791         final_end_function.
23793 2013-02-28  Marek Polacek  <polacek@redhat.com>
23795         PR rtl-optimization/56466
23796         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23797         if we're changing a loop.
23798         (peel_loops_completely): Likewise.
23800 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23802         PR c++/55813
23803         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23805 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23807         PR target/56445
23808         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23809         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23810         INTX_FTYPE_FX, FX_FTYPE_INTX.
23811         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23813 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23815         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23816         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23817         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23818         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23819         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23820         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23821         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23822         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23823         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23824         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23825         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23826         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23827         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23828         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23829         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23830         (avrxmega6): Increase max flash segments from 5 to 6.
23831         * config/avr/t-multilib: Regenerate.
23832         * config/avr/avr-tables.opt: Regenerate.
23833         * doc/avr-mmcu.texi: Regenerate.
23835 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23837         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23838         (avr_device_to_arch): Rename to avr_device_to_ld.
23839         (avr_device_to_as): New prototype.
23840         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23841         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23842         * config/avr/driver-avr.c (avr_device_to_as): New.
23843         (avr_device_to_arch): Rename to avr_device_to_ld.
23845 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23847         PR middle-end/56461
23848         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23849         method on dr_chain and result_chain.
23851         PR middle-end/56461
23852         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23853         pointer_set_destroy on not_executed_last_iteration.
23855         PR middle-end/56461
23856         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23858         PR middle-end/56461
23859         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23860         FOR_EACH_DEFINED_FUNCTION when freeing state.
23862         PR middle-end/56461
23863         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23864         pool_free.
23865         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23866         overwriting it.
23868         PR middle-end/56461
23869         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23870         known_aggs[i].items and release known_aggs vector.
23872         PR middle-end/56461
23873         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23875 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23877         * config/microblaze/microblaze.c (microblaze_emit_compare):
23878         Use xor for EQ/NE comparisions.
23879         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23880         (cbranchsf4): Adjust operator to comparison_operator.
23882 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23884         PR middle-end/56461
23885         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23886         vector.
23887         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23888         vec_safe_push, always update *slot.
23889         (redirect_edge_var_map_clear): Use vec_free.
23890         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23891         (free_var_map_entry): Use vec_free.
23892         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23893         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23895 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23897         PR middle-end/45472
23898         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23899         when the may_trap_p bit of the exprs being merged differs.
23900         Reorder tests for speculativeness in the logical and operator.
23902 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23904         * incpath.c (add_standard_paths): Use reconcat instead of concat
23905         where appropriate and avoid leaking memory.
23907         * opts.h: Include obstack.h.
23908         (opts_concat): New prototype.
23909         (opts_obstack): New declaration.
23910         * opts.c (opts_concat): New function.
23911         (opts_obstack): New variable.
23912         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23913         (finish_options): Use opts_concat instead of concat
23914         and XOBNEWVEC instead of XNEWVEC.
23915         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23916         generate_option): Likewise.
23917         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23918         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23920         PR target/56455
23921         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23922         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23924 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23926         PR middle-end/56461
23927         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23929 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23931         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23932         (arm_block_move_unaligned_straight): Likewise.
23933         (arm_adjust_block_mem): Likewise.
23935 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23937         PR target/48901
23938         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23939         temp, cond and label.
23940         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23942         PR target/52500
23943         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23944         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23946         PR target/52501
23947         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23948         of prologue/epilogue functions.
23950         PR target/52550
23951         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23952         Remove unused variable cfa_offset.
23953         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
23955         PR target/54639
23956         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
23957         type promotion to unsigned.
23959         PR target/54640
23960         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
23961         for HOST_WIDE_INT of 32 bit / same size as int.
23962         (arm_block_move_unaligned_straight): Likewise.
23963         (arm_adjust_block_mem): Likewise.
23965         PR target/54662
23966         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
23967         ALL_CFLAGS.
23969 2013-02-26  Marek Polacek  <polacek@redhat.com>
23971         PR tree-optimization/56426
23972         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
23974 2013-02-26  Richard Biener  <rguenther@suse.de>
23976         PR target/56444
23977         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
23978         unused variable loops.
23980 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23982         PR tree-optimization/56448
23983         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
23984         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
23985         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
23986         later operands of the references, or even first operand for
23987         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
23989         PR tree-optimization/56443
23990         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
23991         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
23992         to type_for_mode langhook.
23994 2013-02-25  Matt Turner  <mattst88@gmail.com>
23996         * doc/invoke.texi: Document r4700.
23998 2013-02-25  Richard Biener  <rguenther@suse.de>
24000         PR tree-optimization/56175
24001         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24002         split out from ...
24003         (simplify_bitwise_binary): ... here.  Also guard the conversion
24004         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24006 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24008         Revert:
24009         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24010             Maciej W. Rozycki  <macro@codesourcery.com>
24011             Tom de Vries  <tom@codesourcery.com>
24012             Nathan Sidwell  <nathan@codesourcery.com>
24013             Iain Sandoe  <iain@codesourcery.com>
24014             Nathan Froyd  <froydnj@codesourcery.com>
24015             Chao-ying Fu  <fu@mips.com>
24017         * doc/extend.texi: (micromips, nomicromips, nocompression):
24018         Document new function attributes.
24019         * doc/invoke.texi (minterlink-compressed, mmicromips,
24020         m14k, m14ke, m14kec): Document new options.
24021         (minterlink-mips16): Update documentation.
24022         * doc/md.texi (ZC, ZD): Document new constraints.
24023         * configure.ac (gcc_cv_as_micromips): Check if linker
24024         supports the .set micromips directive.
24025         * configure: Regenerate.
24026         * config.in: Regenerate.
24027         * config/mips/mips-tables.opt: Regenerate.
24028         * config/mips/micromips.md: New file.
24029         * constraints.md (ZC, AD): New constraints.
24030         * config/mips/predicates.md (movep_src_register): New predicate.
24031         (movep_src_operand): New predicate.
24032         (non_volatile_mem_operand): New predicate.
24033         * config/mips/mips.md (multimem): New type.
24034         (length): Differentiate between 17-bit and 18-bit branch offsets.
24035         (MOVEP1, MOVEP2): New mode iterator.
24036         (mov_<load>l): Use ZC constraint.
24037         (mov_<load>r): Likewise.
24038         (mov_<store>l): Likewise.
24039         (mov_<store>r): Likewise.
24040         (*branch_equality<mode>_inverted): Add microMIPS support.
24041         (*branch_equality<mode>): Likewise.
24042         (*jump_absolute): Likewise.
24043         (indirect_jump_<mode>): Likewise.
24044         (tablejump_<mode>): Likewise.
24045         (<optab>_internal): Likewise.
24046         (sibcall_internal): Likewise.
24047         (sibcall_value_internal): Likewise.
24048         (prefetch): Use constraint ZD.
24049         * config/mips/mips.opt (minterlink-compressed): New option.
24050         (minterlink-mips16): Now an alias for minterlink-compressed.
24051         (mmicromips): New option.
24052         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24053         (compare_and_swap_12): Likewise.
24054         (sync_add<mode>): Likewise.
24055         (sync_<optab>_12): Likewise.
24056         (sync_old_<optab>_12): Likewise.
24057         (sync_new_<optab>_12): Likewise.
24058         (sync_nand_12): Likewise.
24059         (sync_old_nand_12): Likewise.
24060         (sync_new_nand_12): Likewise.
24061         (sync_sub<mode>): Likewise.
24062         (sync_old_add<mode>): Likewise.
24063         (sync_old_sub<mode>): Likewise.
24064         (sync_new_add<mode>): Likewise.
24065         (sync_new_sub<mode>): Likewise.
24066         (sync_<optab><mode>): Likewise.
24067         (sync_old_<optab><mode>): Likewise.
24068         (sync_new_<optab><mode>): Likewise.
24069         (sync_nand<mode>): Likewise.
24070         (sync_old_nand<mode>): Likewise.
24071         (sync_new_nand<mode>): Likewise.
24072         (sync_lock_test_and_set<mode>): Likewise.
24073         (test_and_set_12): Likewise.
24074         (atomic_compare_and_swap<mode>): Likewise.
24075         (atomic_exchange<mode>_llsc): Likewise.
24076         (atomic_fetch_add<mode>_llsc): Likewise.
24077         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24078         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24079         (umips_save_restore_pattern_p): Likewise.
24080         (umips_load_store_pair_p): Likewise.
24081         (umips_output_load_store_pair): Likewise.
24082         (umips_movep_target_p): Likewise.
24083         (umips_12bit_offset_address_p): Likewise.
24084         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24085         (mips_base_mips16): Rename this...
24086         (mips_base_compression_flags): ...to this. Update all uses.
24087         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24088         (mips_mips16_decl_p): Delete.
24089         (mips_nomips16_decl_p): Delete.
24090         (mips_get_compress_on_flags): New function.
24091         (mips_get_compress_off_flags): New function.
24092         (mips_get_compress_mode): New function.
24093         (mips_get_compress_on_name): New function.
24094         (mips_get_compress_off_name): New function.
24095         (mips_insert_attributes): Support multiple compression types.
24096         (mips_merge_decl_attributes): Likewise.
24097         (umips_12bit_offset_address_p): New function.
24098         (mips_start_function_definition): Emit .set micromips directive.
24099         (mips_call_may_need_jalx_p): New function.
24100         (mips_function_ok_for_sibcall): Add microMIPS support.
24101         (mips_print_operand_punctuation): Support short delay slots and
24102         compact jumps.
24103         (umips_swm_mask, umips_swm_encoding): New.
24104         (umips_build_save_restore): New function.
24105         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24106         (was_mips16_p): Remove.
24107         (old_compression_mode): New.
24108         (mips_set_compression_mode): New function.
24109         (mips_set_current_function): Add microMIPS support.
24110         (mips_option_override): Likewise.
24111         (umips_save_restore_pattern_p): New function.
24112         (umips_output_save_restore): New function.
24113         (umips_load_store_pair_p_1): New function.
24114         (umips_load_store_pair_p): New function.
24115         (umips_output_load_store_pair_1): New function.
24116         (umips_output_load_store_pair): New function.
24117         (umips_movep_target_p) New function.
24118         (mips_prepare_pch_save): Add microMIPS support.
24119         * config/mips/mips.h (TARGET_COMPRESSION): New.
24120         (TARGET_CPU_CPP_BUILTINS): Update macro
24121         to use new compression flags and to support microMIPS.
24122         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24123         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24124         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24125         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24126         (ASM_SPEC): Support mmicromips and mno-micromips.
24127         (M16STORE_REG_P): New macro.
24128         (MIPS_CALL): Support TARGET_MICROMIPS.
24129         (MICROMIPS_J): New macro.
24130         (mips_base_mips16): Rename this...
24131         (mips_base_compression_flags): ...to this.
24132         (UMIPS_12BIT_OFFSET_P): New macro.
24133         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24134         (MULTILIB_DIRNAMES): Likewise.
24136 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24138         PR rtl-optimization/56131
24139         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24140         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24141         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24143 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24145         * doc/invoke.texi (-fsanitize=): Move from optimization
24146         to debugging options.
24148 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24150         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24152 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24153             Alexander Monakov  <amonakov@ispras.ru>
24155         PR middle-end/56077
24156         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24157         flush pending lists also on non-jumps.  Adjust comment.
24159 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24160             Maciej W. Rozycki  <macro@codesourcery.com>
24161             Tom de Vries  <tom@codesourcery.com>
24162             Nathan Sidwell  <nathan@codesourcery.com>
24163             Iain Sandoe  <iain@codesourcery.com>
24164             Nathan Froyd  <froydnj@codesourcery.com>
24165             Chao-ying Fu  <fu@mips.com>
24167         * doc/extend.texi: (micromips, nomicromips, nocompression):
24168         Document new function attributes.
24169         * doc/invoke.texi (minterlink-compressed, mmicromips,
24170         m14k, m14ke, m14kec): Document new options.
24171         (minterlink-mips16): Update documentation.
24172         * doc/md.texi (ZC, ZD): Document new constraints.
24173         * configure.ac (gcc_cv_as_micromips): Check if linker
24174         supports the .set micromips directive.
24175         * configure: Regenerate.
24176         * config.in: Regenerate.
24177         * config/mips/mips-tables.opt: Regenerate.
24178         * config/mips/micromips.md: New file.
24179         * constraints.md (ZC, AD): New constraints.
24180         * config/mips/predicates.md (movep_src_register): New predicate.
24181         (movep_src_operand): New predicate.
24182         (non_volatile_mem_operand): New predicate.
24183         * config/mips/mips.md (multimem): New type.
24184         (length): Differentiate between 17-bit and 18-bit branch offsets.
24185         (MOVEP1, MOVEP2): New mode iterator.
24186         (mov_<load>l): Use ZC constraint.
24187         (mov_<load>r): Likewise.
24188         (mov_<store>l): Likewise.
24189         (mov_<store>r): Likewise.
24190         (*branch_equality<mode>_inverted): Add microMIPS support.
24191         (*branch_equality<mode>): Likewise.
24192         (*jump_absolute): Likewise.
24193         (indirect_jump_<mode>): Likewise.
24194         (tablejump_<mode>): Likewise.
24195         (<optab>_internal): Likewise.
24196         (sibcall_internal): Likewise.
24197         (sibcall_value_internal): Likewise.
24198         (prefetch): Use constraint ZD.
24199         * config/mips/mips.opt (minterlink-compressed): New option.
24200         (minterlink-mips16): Now an alias for minterlink-compressed.
24201         (mmicromips): New option.
24202         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24203         (compare_and_swap_12): Likewise.
24204         (sync_add<mode>): Likewise.
24205         (sync_<optab>_12): Likewise.
24206         (sync_old_<optab>_12): Likewise.
24207         (sync_new_<optab>_12): Likewise.
24208         (sync_nand_12): Likewise.
24209         (sync_old_nand_12): Likewise.
24210         (sync_new_nand_12): Likewise.
24211         (sync_sub<mode>): Likewise.
24212         (sync_old_add<mode>): Likewise.
24213         (sync_old_sub<mode>): Likewise.
24214         (sync_new_add<mode>): Likewise.
24215         (sync_new_sub<mode>): Likewise.
24216         (sync_<optab><mode>): Likewise.
24217         (sync_old_<optab><mode>): Likewise.
24218         (sync_new_<optab><mode>): Likewise.
24219         (sync_nand<mode>): Likewise.
24220         (sync_old_nand<mode>): Likewise.
24221         (sync_new_nand<mode>): Likewise.
24222         (sync_lock_test_and_set<mode>): Likewise.
24223         (test_and_set_12): Likewise.
24224         (atomic_compare_and_swap<mode>): Likewise.
24225         (atomic_exchange<mode>_llsc): Likewise.
24226         (atomic_fetch_add<mode>_llsc): Likewise.
24227         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24228         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24229         (umips_save_restore_pattern_p): Likewise.
24230         (umips_load_store_pair_p): Likewise.
24231         (umips_output_load_store_pair): Likewise.
24232         (umips_movep_target_p): Likewise.
24233         (umips_12bit_offset_address_p): Likewise.
24234         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24235         (mips_base_mips16): Rename this...
24236         (mips_base_compression_flags): ...to this. Update all uses.
24237         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24238         (mips_mips16_decl_p): Delete.
24239         (mips_nomips16_decl_p): Delete.
24240         (mips_get_compress_on_flags): New function.
24241         (mips_get_compress_off_flags): New function.
24242         (mips_get_compress_mode): New function.
24243         (mips_get_compress_on_name): New function.
24244         (mips_get_compress_off_name): New function.
24245         (mips_insert_attributes): Support multiple compression types.
24246         (mips_merge_decl_attributes): Likewise.
24247         (umips_12bit_offset_address_p): New function.
24248         (mips_start_function_definition): Emit .set micromips directive.
24249         (mips_call_may_need_jalx_p): New function.
24250         (mips_function_ok_for_sibcall): Add microMIPS support.
24251         (mips_print_operand_punctuation): Support short delay slots and
24252         compact jumps.
24253         (umips_swm_mask, umips_swm_encoding): New.
24254         (umips_build_save_restore): New function.
24255         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24256         (was_mips16_p): Remove.
24257         (old_compression_mode): New.
24258         (mips_set_compression_mode): New function.
24259         (mips_set_current_function): Add microMIPS support.
24260         (mips_option_override): Likewise.
24261         (umips_save_restore_pattern_p): New function.
24262         (umips_output_save_restore): New function.
24263         (umips_load_store_pair_p_1): New function.
24264         (umips_load_store_pair_p): New function.
24265         (umips_output_load_store_pair_1): New function.
24266         (umips_output_load_store_pair): New function.
24267         (umips_movep_target_p) New function.
24268         (mips_prepare_pch_save): Add microMIPS support.
24269         * config/mips/mips.h (TARGET_COMPRESSION): New.
24270         (TARGET_CPU_CPP_BUILTINS): Update macro
24271         to use new compression flags and to support microMIPS.
24272         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24273         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24274         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24275         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24276         (ASM_SPEC): Support mmicromips and mno-micromips.
24277         (M16STORE_REG_P): New macro.
24278         (MIPS_CALL): Support TARGET_MICROMIPS.
24279         (MICROMIPS_J): New macro.
24280         (mips_base_mips16): Rename this...
24281         (mips_base_compression_flags): ...to this.
24282         (UMIPS_12BIT_OFFSET_P): New macro.
24283         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24284         (MULTILIB_DIRNAMES): Likewise.
24286 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24288         PR target/52555
24289         * target-globals.c (save_target_globals): For init_reg_sets and
24290         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24292 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24294         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24295         * config/aarch64/t-aarch64
24296         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24298 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24300         PR inline-asm/56148
24301         * lra-constraints.c (process_alt_operands): Reload operand
24302         conflicting with earlier clobber only if no more other conflicting
24303         operands.
24305 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24307         PR sanitizer/56393
24308         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24309         if not linking a shared library.
24311 2013-02-22  Seth LaForge  <sethml@google.com>
24313         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24315 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24317         * config/arm/arm.md (split for extendsidi): Update condition.
24318         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24319         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24320         (qhs_zextenddi_cstr): Likewise.
24322 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24324         PR middle-end/56420
24325         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24326         avoid signed wrapping.
24327         (expand_mult): Handle properly multiplication by
24328         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24329         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24330         in the compiler if coeff is HOST_WIDE_INT_MIN.
24331         (expand_divmod): Don't make ext_op1 static, change it's type to
24332         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24334         PR rtl-optimization/50339
24335         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24336         field.
24337         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24338         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24339         into splitting_ashiftrt field.
24340         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24341         ASHIFTRT.
24342         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24343         choices.
24345 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24347         PR middle-end/56108
24348         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24349         are sure to go irrevocable.
24351 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24353         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24354         scalars are valid operands.
24356 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24358         PR tree-optimization/56310
24359         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24360         only matching indices and non-negative final offsets.
24361         (intersect_aggregates_with_edge): Pass src_idx to
24362         agg_replacements_to_vector.  Pass src_idx insstead of index to
24363         intersect_with_agg_replacements.
24365 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24367         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24368         instead of hard-wired defaults.
24370 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24372         * doc/invoke.texi (MIPS Options): Update documentation of the
24373         floating-point multiply-accumulate instruction restrictions.
24375 2013-02-21  Kostya Serebryany  <kcc@google.com>
24377         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24378         asan_shadow_offset on x86_64 linux.
24380 2013-02-21  Richard Biener  <rguenther@suse.de>
24382         PR tree-optimization/56415
24383         Revert
24384         2013-02-11  Richard Biener  <rguenther@suse.de>
24386         PR tree-optimization/56273
24387         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24388         first VRP run.
24390 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24392         PR bootstrap/56258
24393         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24394         instead of @itemx.
24396         PR inline-asm/56405
24397         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24398         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24400 2013-02-20  Jan Hubicka  <jh@suse.cz>
24402         PR tree-optimization/56265
24403         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24404         when target is referenced for first time.
24406 2013-02-20  Richard Biener  <rguenther@suse.de>
24408         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24409         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24410         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24411         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24412         not return anything.
24413         (rename_ssa_copies): Do not remove unused locals.
24414         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24415         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24416         * passes.c (execute_function_todo): Do not schedule unused locals
24417         removal if cleanup_tree_cfg did something.
24418         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24419         about the number of removed locals.
24421 2013-02-20  Richard Biener  <rguenther@suse.de>
24423         PR tree-optimization/56398
24424         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24426 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24428         PR tree-optimization/55334
24429         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24430         restricted pointers to arrays.
24432 2013-02-20  Richard Biener  <rguenther@suse.de>
24433             Jakub Jelinek  <jakub@redhat.com>
24435         PR tree-optimization/56396
24436         * tree-ssa-ccp.c (n_const_val): New static variable.
24437         (get_value): Return NULL for SSA names we don't have a lattice
24438         entry for.
24439         (ccp_initialize): Initialize n_const_val.
24440         * tree-ssa-copy.c (n_copy_of): New static variable.
24441         (init_copy_prop): Initialize n_copy_of.
24442         (get_value): Return NULL_TREE for SSA names we don't have a
24443         lattice entry for.
24445 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24447         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24449 2013-02-20  Richard Biener  <rguenther@suse.de>
24451         * genpreds.c (write_lookup_constraint): Do not compare first
24452         letter of the constraint again.
24454 2013-02-20  Richard Biener  <rguenther@suse.de>
24456         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24457         and ceil_log2.
24458         (get_use_iv_cost): Terminate hashtable walk when coming across
24459         an empty entry.
24461 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24463         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24464         reassociation for avx2 targets.
24466 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24468         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24469         Add version check for v8.10.a to enable microblaze_has_clz
24470         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24471         version and TARGET_PATTERN_COMPARE check
24472         * config/microblaze/microblaze.md: New clzsi2 instruction
24474 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24476         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24477         function before branching.
24479 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24481         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24482         DUMP_INSN_RTX_UID.
24483         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24485 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24487         PR middle-end/55889
24488         * sel-sched.c: Include ira.h.
24489         (implicit_clobber_conflict_p): New function.
24490         (moveup_expr): Use it.
24491         * Makefile.in (sel-sched.o): Depend on ira.h.
24493 2013-02-19  Richard Biener  <rguenther@suse.de>
24495         PR tree-optimization/56384
24496         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24497         (vn_hash_type): Split out from ...
24498         (vn_hash_constant_with_type): ... here.
24499         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24500         (vn_phi_eq): Compare types from vn_phi_s structure.
24501         (vn_phi_lookup): Populate vn_phi_s type.
24502         (vn_phi_insert): Likewise.
24504 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24506         PR tree-optimization/56350
24507         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24508         if haven't found reduction or nested cycle operand, rather than
24509         asserting we must find it.
24511         PR tree-optimization/56381
24512         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24513         to fold_build3.
24515 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24516             Jakub Jelinek  <jakub@redhat.com>
24518         PR target/52555
24519         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24520         (swap_optab_enable): Same.
24521         (init_all_optabs): Use argument instead of global.
24522         * tree.h (struct tree_optimization_option): New field target_optabs.
24523         * expr.h (init_all_optabs): Add argument to prototype.
24524         (TREE_OPTIMIZATION_OPTABS): New.
24525         (save_optabs_if_changed): Protoize.
24526         * optabs.h: Declare this_fn_optabs.
24527         * optabs.c (save_optabs_if_changed): New.
24528         Declare this_fn_optabs.
24529         (init_optabs): Add argument to init_all_optabs() call.
24530         * function.c (invoke_set_current_function_hook): Handle per
24531         function optabs.
24532         * function.h (struct function): New field optabs.
24533         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24534         optimization_current_node has changed.
24535         * target-globals.h (save_target_globals_default_opts): Protoize.
24536         * target-globals.c (save_target_globals_default_opts): New.
24538 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24540         PR target/56347
24541         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24542         registers %fr12 and %fr12R as call used.
24544         PR target/56214
24545         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24546         and HImode, require all displacements to be an integer multiple of
24547         their mode size.
24548         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24549         only allow QImode and HImode when reload is in progress and strict is
24550         true.  Likewise for symbolic addresses.  Use base14_operand to check
24551         displacements in REG+BASE addresses.
24553 2013-02-18  Richard Biener  <rguenther@suse.de>
24555         PR tree-optimization/56366
24556         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24557         handle sign-conversion of outer-loop initial induction value.
24559 2013-02-18  Richard Biener  <rguenther@suse.de>
24561         PR middle-end/56349
24562         * cfghooks.c (merge_blocks): If we merge a latch into another
24563         block adjust references to it.
24564         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24565         (verify_loop_structure): Verify that a recorded latch is in fact
24566         a latch.
24568 2013-02-18  Richard Biener  <rguenther@suse.de>
24570         PR tree-optimization/56321
24571         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24572         order SSA name release and virtual operand unlinking.
24574 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24576         * config/microblaze/microblaze.md (save_stack_block): Define.
24577         (restore_stack_block): Likewise.
24579 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24581         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24582         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24583         * config/microblaze/microblaze.c (microblaze_option_override):
24584         Bail out early for PIC modes when target does not support PIC.
24586 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24588         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24589         Replace with a microblaze version.
24590         (microblaze_trampoline_init): Adapt for microblaze.
24591         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24592         microblaze.
24594 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24595             Dodji Seketeli  <dodji@redhat.com>
24597         PR asan/56330
24598         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24599         (instrument_mem_region_access): Do not forget to always put
24600         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24601         0) statement, even for cases where either 'base' or 'base + len'
24602         are not instrumented -- because they have been previously
24603         instrumented.  Simplify the logic by putting all the statements
24604         instrument 'base + len' inside a sequence, and then insert that
24605         sequence right before the current insertion point.  Then, to
24606         instrument 'base + len', just get an iterator on that statement.
24607         And do not forget to update the pointer to iterator the function
24608         received as argument.
24610 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24612         PR rtl-optimization/56348
24613         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24615 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24617         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24618         (clean_graph_dump_file): Pass base to start_graph_dump.
24620 2013-02-14  Richard Henderson  <rth@redhat.com>
24622         PR target/55941
24623         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24625 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24627         * collect2-aix.h: Define F_LOADONLY.
24629 2013-02-14  Richard Biener  <rguenther@suse.de>
24631         PR lto/50494
24632         * varasm.c (output_constant_def_1): Get the decl representing
24633         the constant as argument.
24634         (output_constant_def): Wrap output_constant_def_1.
24635         (make_decl_rtl): Use output_constant_def_1 with the decl
24636         representing the constant.
24637         (build_constant_desc): Optionally re-use a decl already
24638         representing the constant.
24639         (tree_output_constant_def): Adjust.
24641 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24643         Fix an asan crash
24644         * asan.c (instrument_builtin_call):  Really put the length of the
24645         second source argument into src1_len.
24647 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24649         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24650         argument.  If it is false, don't create edge from then_bb to
24651         fallthru_bb.
24652         (insert_if_then_before_iter): Pass true to it.
24653         (build_check_stmt): Pass false to it.
24654         (transform_statements): Flush hash table only on extended basic
24655         block boundaries, rather than at the beginning of every bb.
24656         Don't flush hash table on nonfreeing_call_p calls.
24657         * tree-flow.h (nonfreeing_call_p): New prototype.
24658         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24660 2013-02-13  David S. Miller  <davem@davemloft.net>
24662         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24664 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24666         PR target/56184
24667         * ira.c (max_regno_before_ira): Move from ...
24668         (ira): ... here.
24669         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24670         vec_safe_length.
24672 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24674         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24676 2013-02-13  Richard Biener  <rguenther@suse.de>
24678         PR lto/56295
24679         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24680         globals in MEM_REFs.
24682 2013-02-13  Richard Biener  <rguenther@suse.de>
24684         * loop-init.c (loop_optimizer_init): Clear loop state when
24685         re-initializing preserved loops.
24686         * loop-unswitch.c (unswitch_single_loop): Return whether
24687         we unswitched the loop.  Do not verify loop state here.
24688         (unswitch_loops): When we unswitched a loop discover new loops.
24690 2013-02-13  Kostya Serebryany  <kcc@google.com>
24692         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24693         on x86_64 linux.
24694         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24696 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24698         Avoid instrumenting duplicated memory access in the same basic block
24699         * Makefile.in (asan.o): Add new dependency on hash-table.h
24700         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24701         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24702         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24703         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24704         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24705         (get_mem_ref_of_assignment): New functions.
24706         (get_mem_refs_of_builtin_call): Extract from
24707         instrument_builtin_call and tweak a little bit to make it fit with
24708         the new signature.
24709         (instrument_builtin_call): Use the new
24710         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24711         of is_gimple_builtin_call.
24712         (instrument_derefs, instrument_mem_region_access): Insert the
24713         instrumented memory reference into the hash table.
24714         (maybe_instrument_assignment): Renamed instrument_assignment into
24715         this, and change it to advance the iterator when instrumentation
24716         actually happened and return true in that case.  This makes it
24717         homogeneous with maybe_instrument_assignment, and thus give a
24718         chance to callers to be more 'regular'.
24719         (transform_statements): Clear the memory reference hash table
24720         whenever we enter a new BB, when we cross a function call, or when
24721         we are done transforming statements.  Use
24722         maybe_instrument_assignment instead of instrumentation.  No more
24723         need to special case maybe_instrument_assignment and advance the
24724         iterator after calling it; it's now handled just like
24725         maybe_instrument_call.  Update comment.
24727 2013-02-13  Richard Biener  <rguenther@suse.de>
24729         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24730         Fix loop discovery code.
24732 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24734         PR inline-asm/56148
24735         * lra-constraints.c (process_alt_operands): Match early clobber
24736         operand with itself.  Check conflicts with earlyclobber only if
24737         the operand is not reloaded.  Prefer to reload conflicting operand
24738         if earlyclobber and matching operands are the same.
24740 2013-02-12  Richard Biener  <rguenther@suse.de>
24742         PR lto/56297
24743         * lto-streamer-out.c (write_symbol): Do not output symbols
24744         for hard register variables.
24746 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24748         PR target/54222
24749         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24750         (umulsidi3_insn, mulsidi3_insn): New insns.
24752 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24754         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24755         (struct tune_params): Add vec_costs field.
24756         * config/arm/arm.c (arm_builtin_vectorization_cost)
24757         (arm_add_stmt_cost): New functions.
24758         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24759         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24760         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24761         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24762         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24763         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24764         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24766 2013-02-12  Richard Biener  <rguenther@suse.de>
24768         PR lto/56295
24769         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24770         decls again if possible.
24772 2013-02-12  Richard Biener  <rguenther@suse.de>
24774         PR middle-end/56288
24775         * tree-ssa.c (verify_ssa_name): Fix check, move
24776         SSA_NAME_IN_FREE_LIST check up.
24778 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24779             Steven Bosscher   <steven@gcc.gnu.org>
24781         PR rtl-optimization/56151
24782         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24783         equal to op0 or op1, and last_insn pattern is CODE operation
24784         with MEM dest and one of the operands matches that MEM.
24786 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24788         * doc/extend.texi: Document Function Multiversioning and "default"
24789         parameter string to target attribute.
24790         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24791         target attribute parameter is "default".
24792         (ix86_compare_version_priority): Remove checks for target attribute.
24793         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24794         Remove check for target attribute equal to NULL. Add assert.
24795         (ix86_generate_version_dispatcher_body): Change error to sorry.
24797 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24798             Jack Howarth  <howarth@bromo.med.uc.edu>
24799             Patrick Marlier  <patrick.marlier@gmail.com>
24801         PR libitm/55693
24802         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24803         define ENDFILE_SPEC as TM_DESTRUCTOR.
24804         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24806 2013-02-11  Alexander Potapenko  <glider@google.com>
24807             Jack Howarth  <howarth@bromo.med.uc.edu>
24808             Jakub Jelinek  <jakub@redhat.com>
24810         PR sanitizer/55617
24811         * config/darwin.c (cdtor_record): Rename ctor_record.
24812         (sort_cdtor_records): Rename sort_ctor_records.
24813         (finalize_dtors): New routine to sort destructors by
24814         priority before use in assemble_integer.
24815         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24817 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24819         PR rtl-optimization/56275
24820         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24821         offset is non-negative and less than cmode size before
24822         calling simplify_subreg.
24824 2013-02-11  Richard Biener  <rguenther@suse.de>
24826         PR tree-optimization/56264
24827         * cfgloop.h (fix_loop_structure): Adjust prototype.
24828         * loop-init.c (fix_loop_structure): Return the number of
24829         newly discovered loops.
24830         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24831         are discovered, do a full loop-closed SSA rewrite.
24833 2013-02-11  Richard Biener  <rguenther@suse.de>
24835         PR tree-optimization/56273
24836         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24837         first VRP run.
24838         (check_array_ref): Fix missing newline in dumps.
24839         (search_for_addr_array): Likewise.
24841 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24843         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24845 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24847         PR target/56256
24848         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24850 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24852         PR rtl-optimization/56246
24853         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24854         reload pseudo.
24855         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24856         constraints are satisfied.
24858 2013-02-08  Jeff Law  <law@redhat.com>
24860         PR debug/53948
24861         * emit-rtl.c (reg_is_parm_p): New function.
24862         * regs.h (reg_is_parm_p): New prototype.
24863         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24864         callee-clobbered registers.
24866 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24868         PR target/56043
24869         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24870         If there is no implicit builtin declaration, just return NULL.
24872 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24874         * config/i386/sse.md (FMAMODEM): New mode iterator.
24875         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24876         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24878 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24880         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24881         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24882         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24884 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24886         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24887         (microblaze*-*-elf): Likewise.
24888         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24889         LINK_SPEC.
24890         * config/microblaze/microblaze-c.c: Add builtin defines for
24891         _LITTLE_ENDIAN and _BIG_ENDIAN.
24892         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24893         add to TARGET_DEFAULT flags.
24894         Expand ASM_SPEC and LINK_SPEC.
24895         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24896         * config/microblaze/microblaze.md: Update extendsidi2 and
24897         movdi_internal instructions to use low-order / high-order reg
24898         print_operands.
24899         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24900         options and inversemask / mask of LITTLE_ENDIAN.
24901         * config/microblaze/t-microblaze: Expand multilib options to
24902         include mlittle-endian (le) and update exceptions patterns.
24904 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24906         PR rtl-optimization/56195
24907         * lra-constraints.c (get_reload_reg): Don't reuse regs
24908         if they have smaller mode than requested, if they have
24909         wider mode than requested, try to return a SUBREG.
24911         PR tree-optimization/56250
24912         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24913         if type is unsigned and code isn't MULT_EXPR.
24915 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24917         PR tree-optimization/56064
24918         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24919         bits according to mode.
24920         * fixed-value.h (fixed_from_double_int)
24921         (const_fixed_from_double_int): Adjust comments.
24923 2013-02-08  Richard Biener  <rguenther@suse.de>
24925         PR lto/56231
24926         * lto-streamer.h (struct data_in): Remove current_file, current_line
24927         and current_col members.
24928         * lto-streamer-out.c (lto_output_location): Stream changed bits
24929         en-block for efficiency.
24930         * lto-streamer-in.c (clear_line_info): Remove.
24931         (lto_input_location): Cache current file, line and column
24932         globally via local statics.  Read changed bits en-block.
24933         (input_function): Do not call clear_line_info.
24934         (lto_read_body): Likewise.
24935         (lto_input_toplevel_asms): Likewise.
24937 2013-02-08  Michael Matz  <matz@suse.de>
24939         PR tree-optimization/52448
24940         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24941         (nt_call_phase): New static.
24942         (add_or_mark_expr): Only mark accesses with newer phase than any
24943         call seen.
24944         (nonfreeing_call_p): New.
24945         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24946         (nt_fini_block): Keep blocks marked as visited.
24947         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24949 2013-02-08  Richard Biener  <rguenther@suse.de>
24951         * ira.c (ira): Free broken dominator information.
24953 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24955         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
24957 2013-02-08  Marek Polacek  <polacek@redhat.com>
24959         * cfgloop.c (verify_loop_structure): Add more checking of headers.
24961 2013-02-08  Richard Biener  <rguenther@suse.de>
24963         PR middle-end/56181
24964         * cfgloop.h (flow_loops_find): Adjust.
24965         (bb_loop_header_p): Declare.
24966         * cfgloop.c (bb_loop_header_p): New function split out from ...
24967         (flow_loops_find): ... here.  Adjust function signature,
24968         support incremental loop structure update.
24969         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
24970         * cfgloopmanip.c (fix_loop_structure): Move ...
24971         * loop-init.c (fix_loop_structure): ... here.
24972         (apply_loop_flags): Split out from ...
24973         (loop_optimizer_init): ... here.
24974         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
24975         in incremental mode, only remove dead loops here.
24977 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24979         PR target/54222
24980         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
24981         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
24982         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
24983         (*round<mode>3.libgcc): New insns for fixed-modes.
24984         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
24985         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
24986         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
24987         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
24988         implementations.  Define to __builtin_avr_absFX,
24989         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
24990         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
24991         __builtin_avr_countlsFX, respectively.
24992         * config/avr/avr-c.c (target.h): Include it.
24993         (enum avr_builtin_id): New enum.
24994         (avr_resolve_overloaded_builtin): New static function.
24995         (avr_register_target_pragmas): Use it to set
24996         targetm.resolve_overloaded_builtin.
24997         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
24998         tree nodes used by DEF_BUILTIN.
24999         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25000         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25001         <AVR_BUILTIN_xxBITS>: Same.
25003 2013-02-08  Richard Biener  <rguenther@suse.de>
25005         * cfgloop.c (verify_loop_structure): Properly handle
25006         a loop exiting to another loop header.
25007         * ira-int.h (ira_loops): Remove.
25008         * ira.c (ira_loops): Remove.
25009         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25010         (do_reload): Use loop_optimizer_finalize.
25011         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25012         number_of_loops to access the loop tree.
25013         (more_one_region_p): Likewise.
25014         (finish_loop_tree_nodes): Likewise.
25015         (rebuild_regno_allocno_maps): Likewise.
25016         (mark_loops_for_removal): Likewise.
25017         (mark_all_loops_for_removal): Likewise.
25018         (remove_unnecessary_regions): Likewise.
25019         (ira_build): Likewise.
25020         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25022 2013-02-08  Richard Biener  <rguenther@suse.de>
25024         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25025         * ipa-pure-const.c (analyze_function): Avoid calling
25026         mark_irreducible_loops twice.
25027         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25029 2013-02-07  David S. Miller  <davem@davemloft.net>
25031         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25032         on 'reg'.
25033         * var-tracking.c (vt_add_function_parameter): Test the presence of
25034         HAVE_window_save properly and do not remap argument registers when
25035         we have a leaf function.
25037 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25039         PR bootstrap/56227
25040         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25041         instead of "ll".
25042         * config/i386/i386.c (ix86_print_operand): Ditto.
25044 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25046         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25048 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25050         PR rtl-optimization/56225
25051         * lra-constraints.c (process_alt_operands): Check that reload hard
25052         reg can hold value for strict_low_part.
25054 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25056         PR debug/56154
25057         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25058         dwarf2out_end_function.
25059         (in_first_function_p, maybe_at_text_label_p,
25060         first_loclabel_num_not_at_text_label): New variables.
25061         (dwarf2out_var_location): In the first function find out
25062         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25063         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25064         functions.
25066 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25068         PR rtl-optimization/56178
25069         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25070         SUBREG of a register.  Tidy up related block of code.
25071         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25072         note if the source is a register or a SUBREG of a register.
25074 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25076         PR target/56228
25077         * config/rs6000/rs6000.md (ptrm): New mode attr.
25078         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25079         call_value_indirect_aix<pttrsize>,
25080         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25081         m in constraints.
25083 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25085         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25086         if -bnortl. Convert to strcmp and strncmp.
25088 2013-02-07  Alan Modra  <amodra@gmail.com>
25090         PR target/54009
25091         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25092         addresses won't wrap when offsetting.
25093         (rs6000_secondary_reload): Provide secondary reloads needed for
25094         wrapping LO_SUM addresses.
25096 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25098         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25099         MACH, just __MACH__.
25101 2013-02-06  Richard Biener  <rguenther@suse.de>
25103         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25104         instead of calling fix_loop_structure.
25106 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25108         PR middle-end/56217
25109         * omp-low.c (use_pointer_for_field): Return false if
25110         lower_send_shared_vars doesn't generate any copy-out code.
25112 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25114         PR rtl-optimization/56131
25115         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25116         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25117         of the label is NULL.  Add comment.
25119 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25121         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25123         PR sanitizer/55374
25124         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25125         (STATIC_LIBTSAN_LIBS): Likewise.
25126         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25127         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25128         is defined, don't add anything else beyond that.
25129         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25130         (LINK_COMMAND_SPEC): Use them.
25132         PR tree-optimization/56205
25133         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25134         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25135         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25137 2013-02-05  Richard Biener  <rguenther@suse.de>
25139         PR tree-optimization/53342
25140         PR tree-optimization/53185
25141         * tree-vectorizer.h (vect_check_strided_load): Remove.
25142         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25143         not disallow peeling for vectorized strided loads.
25144         (vect_check_strided_load): Make static and simplify.
25145         (vect_analyze_data_refs): Adjust.
25146         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25147         correctly when vectorizing strided loads.
25149 2013-02-05  Richard Biener  <rguenther@suse.de>
25151         * doc/install.texi: Refer to ISL, not PPL.
25153 2013-02-05  Jan Hubicka  <jh@suse.cz>
25155         PR tree-optimization/55789
25156         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25158 2013-02-05  Jan Hubicka  <jh@suse.cz>
25160         PR tree-optimization/55789
25161         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25162         the dead call anyway.
25164 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25166         PR sanitizer/55374
25167         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25169 2013-02-04  Alexander Potapenko  <glider@google.com>
25170             Jack Howarth  <howarth@bromo.med.uc.edu>
25171             Jakub Jelinek  <jakub@redhat.com>
25173         PR sanitizer/55617
25174         * config/darwin.c (sort_ctor_records): Stabilized qsort
25175         on constructor priority by using original position.
25176         (finalize_ctors): New routine to sort constructors by
25177         priority before use in assemble_integer.
25178         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25180 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25182         PR libstdc++/54314
25183         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25184         about visibility on artificial decls.
25185         * config/sol2.c (solaris_assemble_visibility): Likewise.
25187 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25189         PR target/56186
25190         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25191         argument and improve checking of return-argument types for 16-byte
25192         modes.
25193         (ix86_function_value_1): Add additional valtype argument on call
25194         of function_value_64.
25195         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25196         handling infunction_value_64 function.
25198 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25200         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25202 2013-02-04  Richard Biener  <rguenther@suse.de>
25204         PR tree-optimization/56188
25205         * tree-ssa-structalias.c (label_visit): Consider case with
25206         initially non-empty points-to set.
25207         (perform_var_substitution): Dump node mapping and clean up.
25209 2013-02-04  Richard Guenther  <rguenther@suse.de>
25211         PR lto/56168
25212         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25213         node prevail as last resort.
25214         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25215         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25217 2013-02-04  Richard Biener  <rguenther@suse.de>
25219         PR tree-optimization/56113
25220         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25221         Merge into ...
25222         (equiv_class_lookup_or_add): ... this.
25223         (label_visit): Adjust and fix error in previous patch.
25224         (perform_var_substitution): Adjust.
25226 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25228         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25229         * config/sh/sh4-300.md: Likewise.
25230         * config/sh/sh4a.md: Likewise.
25231         * config/sh/constraints.md: Likewise.
25232         * config/sh/sh.md: Likewise.
25233         * config/sh/netbsd-elf.h: Likewise.
25234         * config/sh/predicates.md: Likewise.
25235         * config/sh/sh-protos.h: Likewise.
25236         * config/sh/ushmedia.h: Likewise.
25237         * config/sh/linux.h: Likewise.
25238         * config/sh/sh.c: Likewise.
25239         * config/sh/superh.h: Likewise.
25240         * config/sh/elf.h: Likewise.
25241         * config/sh/sh4.md: Likewise.
25242         * config/sh/sh.h: Likewise.
25244 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25246         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25247         constraint to match_test floating_point_store_memory_operand().
25248         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25249         (base14_operand): New.
25250         (floating_point_store_memory_operand): New.
25251         (integer_store_memory_operand): Revise to use base14_operand and
25252         reg_plus_base_memory_operand.
25253         (move_dest_operand): Allow symbolic_memory_operands.
25254         (symbolic_memory_operand): Check for LO_SOM.
25255         (symbolic_operand): Change default case to break.
25256         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25257         CONST_DOUBLE values to be reloaded by putting them into memory when
25258         the destination is a floating point register.
25259         (movdf): Remove code to handle CONST_DOUBLE.
25260         (movsf): Likewise.
25261         (reload_indf_r1): New.
25262         (reload_insf_r1): New.
25263         Consistently use "Q" and "T" constraints with integer and floating
25264         point move instructions, respectively.
25265         (movdi): Remove FAIL.
25266         Change predicate for source operand unamed DImode move from
25267         general_operand to move_src_operand.
25268         (umulsidi3): Change predicate for destination operand to
25269         register_operand.
25270         Likewise for similar unamed patterns.
25271         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25272         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25273         (hppa_legitimize_address): Simplify mask calculation.
25274         (pa_emit_move_sequence): Revised handling of secondary reloads from
25275         REG+D addresses for floating point loads and stores.  Directly handle
25276         loading CONST0_RTX (mode) to a floating point register.
25277         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25278         to floating point registers.  Don't restrict secondary reloads to
25279         floating point registers to integer modes.  Revise some comments and
25280         cleanup some code.
25281         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25282         (pa_legitimate_address_p): New.
25283         (pa_legitimize_reload_address): New.
25284         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25285         (STRICT_REG_OK_FOR_BASE_P): New.
25286         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25287         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25289 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25290             Andrew Dixie  <andrewd@gentrack.com>
25292         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25293         flag set.
25295 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25297         * expmed.c (extract_bit_field_1): Pass the full width of the
25298         structure to get_best_reg_extraction_insn.
25300 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25302         PR target/54601
25303         * configure.ac (use_cxa_atexit): Add AIX.
25304         * configure: Regenerate.
25306         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25308 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25310         PR debug/54793
25311         * final.c (need_profile_function): New variable.
25312         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25313         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25314         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25315         notes, targetm.asm_out.function_prologue doesn't emit anything,
25316         HAVE_prologue and profiler should be emitted before prologue,
25317         set need_profile_function instead of emitting it.
25318         (final_scan_insn): If need_profile_function, emit
25319         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25320         NOTE_INSN_FUNCTION_BEG note.
25322 2013-02-01  Richard Henderson  <rth@redhat.com>
25324         * config/rs6000/rs6000.md (smulditi3): New.
25325         (umulditi3): New.
25327         * config/alpha/alpha.md (umulditi3): New.
25329 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25331         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25332         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25334 2013-02-01  Richard Biener  <rguenther@suse.de>
25336         PR tree-optimization/56113
25337         * tree-ssa-structalias.c (label_visit): Reduce work for
25338         single-predecessor nodes.
25340 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25342         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25343         range isn't testing for zero.
25345 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25347         PR middle-end/56113
25348         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25350 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25351             Nick Clifton  <nickc@redhat.com>
25353         * config/v850/constraints.md (Q): Define as a memory constraint.
25354         * config/v850/predicates.md (label_ref_operand): New predicate.
25355         (e3v5_shift_operand): New predicate.
25356         (ior_operator): New predicate.
25357         * config/v850/t-v850: Add e3v5 multilib.
25358         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25359         (v850_gen_movdi): Prototype.
25360         * config/v850/v850.c: Add support for e3v5 architecture.
25361         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25362         TARGET_V850E_UP.
25363         (construct_save_jarl): Add e3v5 long JARL support.
25364         (v850_adjust_insn_length): New function.  Adjust length of call
25365         insns when using e3v5 instructions.
25366         (v850_gen_movdi): New function: Generate instructions to move a
25367         DImode value.
25368         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25369         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25370         (TARGET_USE_FPU): Enable for e3v5.
25371         (CONST_OK_FOR_W): New macro.
25372         (ADJUST_INSN_LENGTH): Define.
25373         * config/v850/v850.md (UNSPEC_LOOP): Define.
25374         (attr cpu): Add v850e3v5.
25375         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25376         (movdi): New pattern.
25377         (movdi_internal): New pattern.
25378         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25379         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25380         (cstoresf4): Likewise.
25381         (cstoredf4): Likewise.
25382         (insv): New pattern.
25383         (rotlso3_a): New pattern.
25384         (rotlsi3_b): New pattern
25385         (rotlsi3_v850e3v5): New pattern.
25386         (doloop_begin): New pattern.
25387         (fix_loop_counter): New pattern.
25388         (doloop_end): New pattern.
25389         (branch_normal): Add e3v5 long branch support.
25390         (branch_invert): Likewise.
25391         (branch_z_normal): Likewise.
25392         (branch_z_invert): Likewise.
25393         (branch_nz_normal): Likewise.
25394         (branch_nz_invert): Likewise.
25395         (call_internal_short): Add e3v5 register-indirect JARL support.
25396         (call_internal_long): Likewise.
25397         (call_value_internal_short): Likewise.
25398         (call_value_internal_long): Likewise.
25399         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25400         (mloop): New option.
25401         * config.gcc: Add support for configuring v840e3v5 target.
25402         * doc/invoke.texi: Document new v850 specific command line options.
25404 2013-01-31  Paul Koning  <ni1d@arrl.net>
25406         PR debug/55059
25407         PR debug/54508
25408         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25409         children if parent is a class.
25410         (prune_unused_types_prune): Don't add DW_AT_declaration.
25412 2013-01-31  Richard Biener  <rguenther@suse.de>
25414         PR tree-optimization/56157
25415         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25416         match up operand with SLP child.
25418 2013-01-31  Jason Merrill  <jason@redhat.com>
25420         PR debug/54410
25421         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25422         parameters the first time.
25423         (gen_scheduled_generic_parms_dies): Check completeness here.
25425 2013-01-31  Richard Biener  <rguenther@suse.de>
25427         PR middle-end/53073
25428         * common.opt (faggressive-loop-optimizations): New flag,
25429         enabled by default.
25430         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25431         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25432         infer_loop_bounds_from_undefined by it.
25434 2013-01-31  Richard Biener  <rguenther@suse.de>
25436         PR tree-optimization/56150
25437         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25438         visit virtual operands.
25439         (find_uses_to_rename_bb): Likewise.
25441 2013-01-31  Richard Biener  <rguenther@suse.de>
25443         PR tree-optimization/56150
25444         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25445         mixed store non-store stmts.
25447 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25449         PR sanitizer/55374
25450         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25451         LIBASAN_EARLY_SPEC is defined.
25452         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25453         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25454         before %o.
25455         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25457         PR c++/55742
25458         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25459         invalid args instead of ICEing on it.
25460         (ix86_valid_target_attribute_tree): Return error_mark_node if
25461         ix86_valid_target_attribute_inner_p failed.
25462         (ix86_valid_target_attribute_p): Return false only if
25463         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25464         target("default") attribute.
25465         (sorted_attr_string): Change argument from const char * to tree,
25466         merge in all target attribute arguments rather than just one.
25467         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25468         instead of free.  Avoid using strcat.
25469         (ix86_mangle_function_version_assembler_name): Mangle
25470         target("default") as if no target attribute is present.  Adjust
25471         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25472         instead of xmalloc and XDELETEVEC instead of free.
25473         (ix86_function_versions): Don't return true if one of the decls
25474         doesn't have target attribute.  If they don't and one of the decls
25475         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25476         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25477         (ix86_supports_function_versions): Remove.
25478         (make_name): Fix up formatting.
25479         (make_dispatcher_decl): Remove resolver_name and its initialization.
25480         Avoid leaking memory.
25481         (is_function_default_version): Return true if there is
25482         target("default") attribute rather than no target attribute at all.
25483         (make_resolver_func): Avoid leaking memory.
25484         (ix86_generate_version_dispatcher_body): Likewise.
25485         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25486         * target.def (supports_function_versions): Remove.
25487         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25488         * doc/tm.texi: Regenerated.
25490 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25492         PR rtl-optimization/56144
25493         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25494         for values with side effects.
25496 2013-01-30  Richard Biener  <rguenther@suse.de>
25498         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25499         (sparseset_pop): Likewise.
25500         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25501         to be able to use quick_push in the worker loop.
25503 2013-01-30  Marek Polacek  <polacek@redhat.com>
25505         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25507 2013-01-30  Richard Biener  <rguenther@suse.de>
25509         PR lto/56147
25510         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25512 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25514         PR tree-optimization/56064
25515         * fixed-value.c (fixed_from_double_int): New function.
25516         * fixed-value.h (fixed_from_double_int): New prototype.
25517         (const_fixed_from_double_int): New static inline function.
25518         * fold-const.c (native_interpret_fixed): New static function.
25519         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25520         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25521         (native_encode_fixed): New static function.
25522         (native_encode_expr) <FIXED_CST>: Use it.
25523         (native_interpret_int): Move double_int worker code to...
25524         * double-int.c (double_int::from_buffer): ...this new static method.
25525         * double-int.h (double_int::from_buffer): Prototype it.
25527 2013-01-30  Richard Biener  <rguenther@suse.de>
25529         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25530         New pointer-map and obstack.
25531         (init_alias_vars): Allocate pointer-map and obstack.
25532         (delete_points_to_sets): Free them.
25533         (find_what_var_points_to): Cache result.
25534         (find_what_p_points_to): Adjust for changed interface of
25535         find_what_var_points_to.
25536         (compute_points_to_sets): Likewise.
25537         (ipa_pta_execute): Likewise.
25539 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25541         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25542         * configure: Regenerate.
25543         * config.in: Regenerate.
25544         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25545         #nobits/#progbits if supported.
25547 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25549         PR target/56121
25550         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25551         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25552         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25554 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25556         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25557         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25559 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25561         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25562         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25564 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25566         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25567         declaration.
25568         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25569         * config/arm/cortex-a7.md: New bypasses using
25570         arm_mac_accumulator_is_result.
25572 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25574         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25575         (cortex_a7_neon_mla): Likewise.
25576         (cortex_a7_fpfmad): New reservation.
25577         (cortex_a7_fpmacs): Use ffmas and update required units.
25578         (cortex_a7_fpmuld): Update required units and latency.
25579         (cortex_a7_fpmacd): Likewise.
25580         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25581         (cortex_a7_neon). Likewise.
25582         (bypass) Update participating units.
25584 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25586         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25587         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25588         from fmac to ffma.
25589         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25590         (vfp_fmul): Use ffmad.
25591         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25592         (cortex_r4_fmacd): Use ffmad.
25593         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25594         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25595         (cortex_a9_fmacd): Use ffmad.
25596         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25597         (cortex_a8_vfp_macd): Use ffmad.
25598         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25599         (cortex_a5_fpmacd): Use ffmad.
25600         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25601         (cortex_a15_vfp_macd): Use ffmad.
25602         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25604 2013-01-29  Jason Merrill  <jason@redhat.com>
25606         PR libstdc++/54314
25607         * varasm.c (default_assemble_visibility): Don't warn about
25608         visibility on artificial decls.
25610 2013-01-29  Richard Biener  <rguenther@suse.de>
25612         PR tree-optimization/56113
25613         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25614         the bitmap leader.
25615         (label_visit): Free duplicate bitmaps and record the leader instead.
25616         (perform_var_substitution): Adjust.
25618 2013-01-29  Richard Biener  <rguenther@suse.de>
25620         PR tree-optimization/55270
25621         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25622         the CFG, schedule loops for fixup.
25624 2013-01-29  Nick Clifton  <nickc@redhat.com>
25626         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25627         SP_REG.
25629 2013-01-28  Leif Ekblad  <leif@rdos.net>
25631         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25632         * config/i386/i386.h (TARGET_RDOS): New macro.
25633         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25634         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25635         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25636         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25637         DEFAULT_LARGE_SECTION_THRESHOLD.
25638         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25639         * config/i386/rdos.h: New file.
25640         * config/i386/rdos64.h: New file.
25642 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25644         PR other/54814
25645         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25646         TEST_HARD_REG_BIT.
25648 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25650         PR rtl-optimization/56117
25651         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25652         call cselib_lookup_from_insn on the MEM before calling
25653         add_insn_mem_dependence.
25655 2013-01-28  Richard Biener  <rguenther@suse.de>
25657         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25658         to a stmt that didn't have one.
25659         (copy_phis_for_bb): Likewise for PHI arguments.
25660         (copy_debug_stmt): Likewise for debug stmts.
25662 2013-01-28  Richard Biener  <rguenther@suse.de>
25664         PR tree-optimization/56034
25665         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25666         (partition_builtin_p): Adjust.
25667         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25668         it is the last partition.
25669         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25670         up the vertex for the definition.
25671         (classify_partition): Classify whether a partition is a
25672         PKIND_REDUCTION, thus has uses outside of the loop.
25673         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25674         Merge all PKIND_REDUCTION partitions into the last partition.
25675         (tree_loop_distribution): Seed partitions from reductions as well.
25677 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25679         PR tree-optimization/56125
25680         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25681         pow(x,c) into sqrt(x) * powi(x, n/2) or
25682         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25683         optimizing for size.
25684         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25685         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25686         integer.
25688         PR tree-optimization/56094
25689         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25690         to UNKNOWN_LOCATION while gimplifying expr.
25692 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25694         PR target/56114
25695         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25696         operand 0 in movabs insn template for -masm=intel asm alternative.
25697         (*movabs<mode>_2): Ditto for operand 1.
25699 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25701         PR target/54663
25702         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25703         of microblaze-c.o
25705 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25707         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25708         tm_file.
25710 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25712         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25713         Undef to avoid warning.
25715 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25717         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25718         * configure: Regenerate.
25720 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25722         PR tree-optimization/56098
25723         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25724         for stmts with volatile ops.
25725         (cond_store_replacement): Don't optimize if assign has volatile ops.
25726         (cond_if_else_store_replacement_1): Don't optimize if either
25727         then_assign or else_assign have volatile ops.
25728         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25729         volatile ops.
25731 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25733         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25735 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25737         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25738         missing ':' in asm example.
25740 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25742         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25743         entries into lane and laneq entries.
25744         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25745         Remove AdvSIMD scalar modes.
25746         (aarch64_sq<r>dmulh_laneq<mode>): New.
25747         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25748         modes.
25749         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25750         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25751         * config/aarch64/iterators.md (VCOND): New.
25752         (VCONQ): New.
25754 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25756         PR target/54222
25757         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25758         Add NULL LIBNAME argument to existing definitions.
25759         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25760         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25761         * config/avr/avr.c (DEF_BUILTIN): Same.
25762         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25763         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25764         implementation is available (DECL_ASSEMBLER_NAME is set).
25765         (avr_fold_absfx): New static function.
25766         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25767         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25768         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25769         AVR_BUILTIN_ABSLLK.
25770         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25771         (abshk, absk, abslk, absllk): Provide as static inline functions.
25773 2013-01-25  Marek Polacek  <polacek@redhat.com>
25775         PR tree-optimization/56035
25776         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25778 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25780         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25781         (*movtf_internal_rex64): Add (!o,C) alternative
25782         (*movxf_internal_rex64): Ditto.
25783         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25785 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25787         * doc/invoke.texi: fix typo.
25788         * doc/objc.texi: fix typo.
25790 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25792         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25793         for the first two alternatives.
25795 2013-01-24  Diego Novillo  <dnovillo@google.com>
25797         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25798         (ggc-zone.o): Remove.
25799         * configure.ac: Remove option --with-gc.
25800         * configure: Re-generate.
25801         * doc/install.texi: Remove documentation for --with-gc.
25802         * gengtype.c (write_enum_defn): Remove.  Update all users.
25803         (write_Types_process_field): Remove generation of gt_e_* argument.
25804         (output_type_enum): Remove.  Update all users.
25805         (write_enum_defn): Remove.  Update all users.
25806         (enum alloc_zone): Remove.  Update all users.
25807         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25808         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25809         Update all callers.
25810         (struct ptr_data): Remove field TYPE.  Update all users.
25811         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25812         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25813         Update all users.
25814         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25815         (struct alloc_zone): Remove.
25816         (ggc_internal_alloc_zone_stat): Remove.
25817         (ggc_internal_cleared_alloc_zone_stat): Remove.
25818         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25819         (ggc_pch_count_object): Remove last argument.  Update all users.
25820         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25821         (struct alloc_zone): Remove.
25822         * ggc-zone.c: Remove.
25823         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25824         (struct alloc_zone): Remove.
25825         (ggc_alloc_typed_stat): Remove.
25826         (ggc_alloc_typed): Remove.
25827         (ggc_splay_alloc): Remove first argument.
25828         (rtl_zone): Remove.  Update all users.
25829         (tree_zone): Remove.  Update all users.
25830         (tree_id_zone): Remove.  Update all users.
25831         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25832         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25833         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25834         * tree-ssanames.c: Remove references to zone allocator in comments.
25836 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25838         * config/avr/avr.c (avr_out_fract): Make register numbers that
25839         might be outside of source operand signed.
25841 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25843         * config/i386/constraints.md (Yf): New constraint.
25844         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25845         of f constraint to conditionaly disable x87 register preferences.
25846         (*movdf_internal): Ditto.
25847         (*movsf_internal): Ditto.
25849 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25851         PR inline-asm/55934
25852         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25853         that have operands with impossible constraints.
25854         Add a FIXME for a speed-up opportunity.
25855         * lra-constraints.c (process_alt_operands): Verify that a class
25856         selected from constraints on asms is valid for the operand mode.
25857         (curr_insn_transform): Remove incorrect comment.
25859 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25861         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25862         TOC operand is a valid symbol ref in the constant pool.
25864 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25866         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25868 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25870         PR target/54222
25871         * config/avr/stdfix.h: New file.
25872         * t-avr (stdfix-gcc.h): New rule to build it.
25873         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25875 2013-01-23  Kostya Serebryany  <kcc@google.com>
25877         * config/darwin.h: remove dependency on
25878         CoreFoundation (asan on Mac OS).
25880 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25882         PR target/49069
25883         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25884         instead of cmpdi_operand for first comparison operand.
25885         Don't assert that comparison operands aren't both constants.
25887 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25889         * doc/install.texi (Downloading the Source): Update references to
25890         downloading separate components.
25892 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25894         * doc/extend.texi (__int128): Improve grammar.
25896 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25898         PR target/56028
25899         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25900         alternative to (o,r).
25901         (*movdi_internal_rex64): Remove (!o,n) alternative.
25902         (DImode immediate->memory splitter): Remove.
25903         (DImode immediate->memory peephole2): Remove.
25904         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25905         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25906         alternative to (!o,*r).
25907         (*movtf_internal_sse): New pattern.
25908         (*movxf_internal_rex64): New pattern.
25909         (*movxf_internal): Disable for TARGET_64BIT.
25910         (*movdf_internal_rex64): Remove (!o,F) alternative.
25912 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25914         PR middle-end/56074
25915         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25916         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25917         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25918         stmt locations where LOCATION_LOCUS of the stmt location is
25919         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25921         PR target/55686
25922         * config/i386/i386.md (UNSPEC_STOS): New.
25923         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25924         *strsetqi_1): Add UNSPEC_STOS.
25926 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25928         PR c++/56067
25929         * doc/invoke.texi: Remove left over -Wsynth example.
25931 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25933         PR tree-optimization/56051
25934         * fold-const.c (fold_binary_loc): Don't fold
25935         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25936         a narrowing conversion, or widening conversion from signed
25937         to unsigned.
25939 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25941         PR rtl-optimization/56023
25942         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25943         dependent on debug instruction.
25945 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25947         PR middle-end/56022
25948         * function.c (allocate_struct_function): Call
25949         invoke_set_current_function_hook earlier.
25951 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25953         * reload1.c (init_reload): Only initialize reload_obstack
25954         during the first call.
25956 2013-01-21  Marek Polacek  <polacek@redhat.com>
25958         * cfgloop.c (verify_loop_structure): Fix up grammar.
25960 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
25962         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
25963         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
25965 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25967         PR target/56058
25968         * config/arm/marvell-pj4.md: Update copyright year.
25969         Fix up use of alu to alu_reg and simple_alu_imm.
25971 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25973         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
25975 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
25977         PR target/55433
25978         * lra-constraints.c (curr_insn_transform): Don't reuse original
25979         insn for secondary memory move when memory mode should be different.
25981 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25983         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
25984         atomic_storedi_1): New patterns.
25986 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
25988         btver2 pipeline descriptions.
25989         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
25990         descriptions.
25991         * config/i386/i386.md (btver2_decode): New type attributes.
25992         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
25993         type attributes.
25994         * config/i386/btver2.md: New file describing btver2 pipelines.
25996 2013-01-19  Andrew Pinski  <apinski@cavium.com>
25998         PR tree-optimization/52631
25999         * tree-ssa-sccvn (visit_use): Before looking up the original
26000         statement, try looking up the simplified expression.
26002 2013-01-19  Anthony Green  <green@moxielogic.com>
26004         * config/moxie/moxie.c (moxie_expand_prologue): Set
26005         current_function_static_stack_size.
26007 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26009         PR tree-optimization/56029
26010         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26011         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26013 2013-01-18  Sharad Singhai  <singhai@google.com>
26015         PR tree-optimization/55995
26016         * dumpfile.c (dump_loc): Print location only if available.
26017         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26019 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26021         PR target/55433
26022         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26023         secondary memory move.
26024         (inherit_reload_reg): Use rclass instead of cl for
26025         check_secondary_memory_needed_p.
26027 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26029         PR middle-end/56015
26030         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26031         the case where writing real complex part of target modifies op1.
26033 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26035         * config/aarch64/aarch64-simd.md
26036         (aarch64_vcond_internal<mode>): Handle unordered cases.
26037         * config/aarch64/iterators.md (v_cmp_result): New.
26039 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26040             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26042         * config/arm/marvell-pj4.md: New file.
26043         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26044         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26045         (generic_vfp): Likewise.
26046         * config/arm/arm-cores.def: Add marvell-pj4.
26047         * config/arm/arm-tune.md: Regenerate.
26048         * config/arm/arm-tables.opt: Regenerate.
26049         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26050         * doc/invoke.texi: Document marvell-pj4.
26052 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26054         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26056 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26058         PR debug/54114
26059         PR debug/54402
26060         PR debug/49888
26061         * var-tracking.c (negative_power_of_two_p): New.
26062         (global_get_addr_cache, local_get_addr_cache): New.
26063         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26064         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26065         heading comment.
26066         (vt_stack_offset_p): Remove.
26067         (vt_canon_true_dep): Always canonicalize loc's address.
26068         (clobber_overlapping_mems): Make sure we have a MEM.
26069         (local_get_addr_clear_given_value): New.
26070         (val_reset): Clear local cached entries.
26071         (compute_bb_dataflow): Create and release the local cache.
26072         Disable duplicate MEMs clobbering.
26073         (emit_notes_in_bb): Clobber MEMs likewise.
26074         (vt_emit_notes): Create and release the local cache.
26075         (vt_initialize, vt_finalize): Create and release the global
26076         cache, respectively.
26077         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26079 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26081         PR libmudflap/53359
26082         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26083         not found in the symtab.
26085 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26087         PR debug/56006
26088         PR rtl-optimization/55547
26089         PR rtl-optimization/53827
26090         PR debug/53671
26091         PR debug/49888
26092         * alias.c (offset_overlap_p): New, factored out of...
26093         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26094         the conservative special case for symbolic constants.  Don't
26095         adjust zero sizes on alignment.
26097 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26099         PR rtl-optimization/52573
26100         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26101         REG_UNUSED for the same register.
26103 2013-01-17  Richard Biener  <rguenther@suse.de>
26104             Marek Polacek  <polacek@redhat.com>
26106         PR rtl-optimization/55833
26107         * loop-unswitch.c (unswitch_loops): Move loop verification...
26108         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26109         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26110         Set it to true when we're removing a loop from hierarchy tree in
26111         an irreducible region.
26112         (fix_bb_placements): Adjust caller.
26113         (fix_loop_placements): Likewise.
26115 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26117         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26118         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26119         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26120         Remove ID.  Adjust comments.
26121         * config/avr/avr-c.c (avr_builtin_name): Remove.
26122         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26123         * config/avr/avr.c (avr_tolower): New static function.
26124         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26125         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26126         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26127         default expansion.
26129 2013-01-17  Jan Hubicka  <jh@suse.cz>
26131         PR tree-optimization/55273
26132         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26134 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26136         PR target/55981
26137         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26138         store through atomic_store<mode>_1.
26139         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26141 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26143         PR tree-optimizations/55264
26144         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26145         for virtual methods.
26146         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26147         virtual methods before inlining is over.
26148         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26149         virtual functions.
26150         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26151         non-virtual.
26153 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26155         PR rtl-optimization/56005
26156         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26157         pending reads for prefetch.
26159 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26161         * config/aarch64/aarch64.md
26162         (*cstoresi_neg_uxtw): New pattern.
26163         (*cmovsi_insn_uxtw): New pattern.
26164         (*<optab>si3_uxtw): New pattern.
26165         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26166         (*<optab>si3_insn_uxtw): New pattern.
26167         (*bswapsi2_uxtw): New pattern.
26169 2013-01-16  Richard Biener  <rguenther@suse.de>
26171         * tree-inline.c (tree_function_versioning): Remove set but
26172         never used variable.
26174 2013-01-16  Richard Biener  <rguenther@suse.de>
26176         PR tree-optimization/55964
26177         * tree-flow.h (rename_variables_in_loop): Remove.
26178         (rename_variables_in_bb): Likewise.
26179         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26180         (copy_loop_before): Adjust and delete update-ssa status.
26181         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26182         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26183         (rename_variables_in_loop): Remove.
26184         (slpeel_update_phis_for_duplicate_loop): Likewise.
26185         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26186         use available cfg machinery instead of duplicating it.
26187         Update PHI nodes and perform poor-mans SSA update here.
26188         (slpeel_tree_peel_loop_to_edge): Adjust.
26190 2013-01-16  Richard Biener  <rguenther@suse.de>
26192         PR tree-optimization/54767
26193         PR tree-optimization/53465
26194         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26195         (vrp_visit_phi_node): For PHI arguments coming via backedges
26196         drop all symbolical range information.
26197         (execute_vrp): Compute backedges.
26199 2013-01-16  Richard Biener  <rguenther@suse.de>
26201         * doc/install.texi: Update CLooG and ISL requirements to
26202         0.18.0 and 0.11.1.
26204 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26206         PR target/55301
26207         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26208         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26209         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26211 2013-01-16  DJ Delorie  <dj@redhat.com>
26213         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26214         (UNSPECV_SP_SWITCH_E): New.
26215         (sp_switch_1): Change to an unspec.
26216         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26217         replace $r15.
26219 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26221         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26222         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26223         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26224         (expand_mem_thread_fence): Ditto.
26225         (expand_mem_signal_fence): Ditto.
26226         (expand_atomic_load): Ditto.
26227         (expand_atomic_store): Ditto.
26229 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26231         PR rtl-optimization/55547
26232         PR rtl-optimization/53827
26233         PR debug/53671
26234         PR debug/49888
26235         * alias.c (memrefs_conflict_p): Set sizes to negative after
26236         AND adjustments.
26238 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26240         PR target/55940
26241         * function.c (thread_prologue_and_epilogue_insns): Always
26242         add crtl->drap_reg to set_up_by_prologue.set, even if
26243         stack_realign_drap is false.
26245 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26247         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26248         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26249         *call): Fix indention.
26251 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26253         PR target/55876
26254         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26255         Update comment.
26257 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26259         PR rtl-optimization/55153
26260         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26262 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26264         PR tree-optimization/55920
26265         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26266         accesses as grp_to_be_debug_replaced.
26268 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26270         PR tree-optimization/55920
26271         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26272         there is non-useless type conversion needed from debug rhs to lhs,
26273         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26275 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26276             Mikael Pettersson  <mikpe@it.uu.se>
26278         PR target/43961
26279         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26280         Thumb.
26281         (ASM_OUTPUT_CASE_LABEL): Remove.
26282         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26283         * final.c (shorten_branches): Update alignment of labels before
26284         jump tables if CASE_VECTOR_SHORTEN_MODE.
26286 2013-01-15  Richard Biener  <rguenther@suse.de>
26288         PR bootstrap/55961
26289         * system.h: Do not include gmp.h for building host tools.
26291 2013-01-15  Richard Biener  <rguenther@suse.de>
26293         PR middle-end/55882
26294         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26295         account for bitpos when computing alignment.
26297 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26299         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26300         (ix86_target_macros_internal): Likewise.
26302         * config/i386/i386.c (m_CORE2I7): Removed.
26303         (m_CORE_HASWELL): New macro.
26304         (m_CORE_ALL): Likewise.
26305         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26306         (initial_ix86_arch_features): Likewise.
26307         (processor_target_table): Initializations for Core avx2.
26308         (cpu_names): New names "core-avx2".
26309         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26310         PROCESSOR_CORE_HASWELL.
26311         (ix86_issue_rate): New case.
26312         (ia32_multipass_dfa_lookahead): Likewise.
26313         (ix86_sched_init_global): Likewise.
26315         * config/i386/i386.h (TARGET_HASWELL): New macro.
26316         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26317         (processor_type): New PROCESSOR_HASWELL.
26319 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26321         PR tree-optimization/55955
26322         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26323         *SHIFT_EXPR and *ROTATE_EXPR codes.
26325         PR tree-optimization/48766
26326         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26327         -ftrapv disable -fwrapv.
26329 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26331         PR target/55974
26332         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26333         etc. to 1 and not to __flash.
26334         Use LL suffix for __INT24_MAX__ with -mint8.
26335         Use ULL suffix for __UINT24_MAX__ with -mint8.
26337 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26339         * config/avr/avr-arch.h
26340         (struct base_arch_s): Use typedef avr_arch_t instead.
26341         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26342         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26343         * config/avr/avr.c: Same.
26344         * config/avr/avr-devices.c: Same.
26345         * config/avr/driver-avr.c: Same.
26346         * config/avr/gen-avr-mmcu-texi.c: Same.
26347         * config/avr/avr-mcus.def: Adjust comment.
26349 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26351         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26352         * config/aarch64/iterators.md (VALLDI): New.
26354 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26355             Andi Kleen  <ak@linux.intel.com>
26357         PR target/55948
26358         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26359         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26360         memmodel flag.
26362 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26364         * config/avr/avr-stdint.h: Remove trailing blanks.
26365         * config/avr/avr-log.h: Same.
26366         * config/avr/avr-arch.h: Same.
26367         * config/avr/avr-devices.c: Same.
26368         * config/avr/avr-dimode.md: Same.
26369         * config/avr/predicates.md: Same.
26370         * config/avr/avr-c.c: Same.  And fix typo.
26372         * config/avr/avr-protos.h: Same.  And:
26373         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26374         (init_cumulative_args): Rename to avr_init_cumulative_args.
26375         (expand_prologue): Rename to avr_expand_prologue.
26376         (expand_epilogue): Rename to avr_expand_epilogue.
26377         (adjust_insn_length): Rename to avr_adjust_insn_length.
26378         (notice_update_cc): Rename to avr_notice_update_cc.
26379         (final_prescan_insn): Rename to avr_final_prescan_insn.
26380         * config/avr/avr.c: Same.
26381         * config/avr/avr.h: Same.
26382         * config/avr/avr.md: Remove trailing blanks.
26383         (prologue): Use avr_expand_prologue.
26384         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26386 2013-01-14  Richard Biener  <rguenther@suse.de>
26388         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26389         verify_location, collect_subblocks): New functions.
26390         (verify_gimple_in_cfg): Verify that locations only reference
26391         BLOCKs in the functions BLOCK tree.
26393 2013-01-14  Richard Biener  <rguenther@suse.de>
26395         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26396         PHI argument.
26397         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26398         unshare reference.
26399         (insert_out_of_ssa_copy_on_edge): Likewise.
26400         (rewrite_close_phi_out_of_ssa): Likewise.
26401         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26402         debug expressions.
26403         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26404         propagated constants.
26405         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26406         can not be shared.
26408 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26410         * config/avr/avr-modes.def: Add GPL copyright notice.
26412 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26414         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26415         MEMMODEL_MASK to determine memory model.
26416         (atomic_store<mode>): Ditto from operands[2].
26417         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26419 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26421         PR fortran/55935
26422         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26423         (fold_gimple_assign): Don't call unshare_expr here.
26424         (fold_ctor_reference): Call unshare_expr.
26426 2013-01-13  Terry Guo  <terry.guo@arm.com>
26428         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26429         * doc/fragments.texi: Document MULTILIB_REUSE.
26430         * gcc.c (multilib_reuse): New internal spec.
26431         (set_multilib_dir): Also search multilib from multilib_reuse.
26432         * genmultilib (tmpmultilib3): Refactor code.
26433         (tmpmultilib4): Ditto.
26434         (multilib_reuse): New multilib argument.
26436 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26438         * Makefile.in: Update copyright.
26440 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26442         PR middle-end/55890
26443         * calls.c (expand_call): Check if arg_nr is valid.
26445 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26447         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26448         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26449         documentation.  Add missing '__' in front of
26450         __builtin_ia32_packssdw256.
26452 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26454         PR target/55719
26455         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26456         NO_REGS for larl operands.
26457         (s390_reload_larl_operand): Use s390_load_address instead of
26458         emit_move_insn.
26460 2013-01-11  Richard Biener  <rguenther@suse.de>
26462         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26463         (verify_node_sharing): ... here.
26464         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26466 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26468         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26469         Substitute TREECHECKING.
26470         * configure: Regenerate.
26471         * Makefile.in (TREECHECKING): New.
26473 2013-01-11  Richard Guenther  <rguenther@suse.de>
26475         PR tree-optimization/44061
26476         * tree-vrp.c (extract_range_basic): Compute zero as
26477         value-range for __builtin_constant_p of function parameters.
26479 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26481         Update copyright years.
26483 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26485         PR rtl-optimization/55672
26486         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26487         const to be eliminable.
26489 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26491         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26492         * configure: Regenerate.
26494 2013-01-10  Richard Biener  <rguenther@suse.de>
26496         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26498 2013-01-10  Richard Biener  <rguenther@suse.de>
26500         PR bootstrap/55792
26501         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26502         locations for virtual PHI arguments.
26503         (rewrite_update_phi_arguments): Likewise.
26505 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26507         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26508         on to assembler.
26510 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26512         PR tree-optimization/55921
26513         * tree-complex.c (expand_complex_asm): New function.
26514         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26516 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26518         PR target/55718
26519         * config/s390/s390.c (s390_symref_operand_p)
26520         (s390_loadrelative_operand_p): Merge the two functions.
26521         (s390_check_qrst_address, print_operand_address): Add parameters
26522         to s390_loadrelative_operand_p invokation.
26523         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26524         (s390_reload_larl_operand, s390_secondary_reload): Use
26525         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26526         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26528 2013-01-09  Mike Stump  <mikestump@comcast.net>
26530         * dse.c (record_store): Remove unnecessary assert.
26532 2013-01-09  Jan Hubicka  <jh@suse.cz>
26534         PR tree-optimization/55569
26535         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26536         * cfgloop.h (scale_loop_profile): Likewise.
26538 2013-01-09  Jan Hubicka  <jh@suse.cz>
26540         PR lto/45375
26541         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26542         functions.
26543         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26545 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26547         PR middle-end/55114
26548         * expr.h (maybe_emit_group_store): Declare.
26549         * expr.c (maybe_emit_group_store): New function.
26550         * builtins.c (expand_builtin_int_roundingfn): Call it.
26551         (expand_builtin_int_roundingfn_2): Likewise.
26553 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26555         PR rtl-optimization/55829
26556         * lra-constraints.c (match_reload): Add code for absent output.
26557         (curr_insn_transform): Add code for reloads of matched inputs
26558         without output.
26560 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26562         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26563         attribute of movddup insn to DF.
26564         (*vec_interleave_lowv2df): Ditto.
26565         (vec_dupv2df): Ditto.
26567 2013-01-09  Jan Hubicka  <jh@suse.cz>
26569         PR tree-optimiation/55875
26570         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26571         EVERY_ITERATION parameter.
26572         (number_of_iterations_exit): Check if exit is executed every iteration.
26573         (idx_infer_loop_bounds): Similarly here.
26574         (n_of_executions_at_most): Simplify
26575         to only test for cases where statement is dominated by the
26576         particular bound; handle correctly the "postdominance" test.
26577         (scev_probably_wraps_p): Use max loop iterations info
26578         as a global bound first.
26580 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26581             Nick Clifton  <nickc@redhat.com>
26583         * config/v850/v850.md (cbranchsf4): New pattern.
26584         (cstoresf4): New pattern.
26585         (cbranchdf4): New pattern.
26586         (cstoredf4): New pattern.
26587         (movsicc): Disallow floating point comparisons.
26588         (cmpsf_le_insn): Fix order of operators.
26589         (cmpsf_lt_insn): Likewise.
26590         (cmpsf_eq_insn): Likewise.
26591         (cmpdf_le_insn): Likewise.
26592         (cmpdf_lt_insn): Likewise.
26593         (cmpdf_eq_insn): Likewise.
26594         (cmpsf_ge_insn): Use LE comparison.
26595         (cmpdf_ge_insn): Likewise.
26596         (cmpsf_gt_insn): Use LT comparison.
26597         (cmpdf_gt_insn): Likewise.
26598         (cmpsf_ne_insn): Delete pattern.
26599         (cmpdf_ne_insn): Delete pattern.
26600         * config/v850/v850.c (v850_gen_float_compare): Use
26601         gen_cmpdf_eq_insn for NE comparison.
26602         (v850_float_z_comparison_operator)
26603         (v850_float_nz_comparison_operator): Move from here ...
26604         * config/v850/predicates.md: ... to here.  Move GT and GE
26605         comparisons into v850_float_z_comparison_operator.
26606         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26607         Delete prototype.
26608         (v850_float_nz_comparison_operator): Likewise.
26610 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26612         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26613         with calls to gen_insvsi/gen_insvdi.
26615 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26617         * config/i386/i386.c (initial_ix86_tune_features): Set up
26618         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26620 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26621             Jakub Jelinek  <jakub@redhat.com>
26623         PR tree-optimization/48189
26624         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26625         If nitercst is 0, don't predict the exit edge.
26627 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26629         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26630         in asm_fprintf with reg_names.
26631         (aarch64_print_operand_address): Likewise.
26632         (aarch64_return_addr): Likewise.
26633         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26635 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26637         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26638         (INT_U6_BITS): Likewise.
26639         * config/pa/predicates.md (uint6_operand): New predicate.
26640         (shift5_operand, shift6_operand): Likewise.
26641         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26642         arith32_operand.
26643         (lshrdi3): Use shift6_operand.
26644         (shrpsi4, shrpdi4): New insn patterns.
26645         (extzv): Delete expander.
26646         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26647         predicates in unamed zero extract patterns.  Tighten common constraint.
26648         (extv): Delete expander.
26649         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26650         predicates in unamed sign extract patterns.  Tighten common constraint.
26651         (insv): Delete expander.
26652         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26653         predicates in unamed insert patterns.  Tighten common constraint.
26654         Change uint32_operand predicate to uint6_operand predicate in unamed
26655         DImode pattern to insert constant values of type 1...1xxxx.
26657 2013-01-04  Jan Hubicka  <jh@suse.cz>
26659         PR tree-optimization/55823
26660         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26661         issue.
26663 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26664             Uros Bizjak  <ubizjak@gmail.com>
26666         PR rtl-optimization/55845
26667         * df-problems.c (can_move_insns_across): Stop scanning at
26668         volatile_insn_p source instruction or give up if
26669         across_from .. across_to range contains any volatile_insn_p
26670         instructions.
26672 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26674         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26675         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26676         Declare.
26677         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26678         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26680 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26682         PR fortran/55341
26683         * asan.c (asan_clear_shadow): New function.
26684         (asan_emit_stack_protection): Use it.
26686 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26688         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26689         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26690         with tab instead of space.
26692 2013-01-08  Nick Clifton  <nickc@redhat.com>
26694         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26695         register bank 0 at the start of an interrupt handler.
26696         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26697         MDBH registers.
26699 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26701         * config/aarch64/aarch64-simd.md
26702         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26703         (aarch64_simd_bsl): Likewise.
26704         (aarch64_vcond_internal<mode>): Likewise.
26705         (vcond<mode><mode>): Likewise.
26706         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26707         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26709 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26711         * config/aarch64/aarch64-builtins.c
26712         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26714 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26716         PR debug/55579
26717         * tree-sra.c (analyze_access_subtree): Return true also after
26718         potentially creating a debug-only replacement.
26720 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26722         PR middle-end/55890
26723         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26725         PR tree-optimization/54120
26726         * tree-vrp.c (range_fits_type_p): Don't allow
26727         src_precision < precision from signed vr to unsigned_p
26728         if vr->min or vr->max is negative.
26729         (simplify_float_conversion_using_ranges): Test can_float_p
26730         against CODE_FOR_nothing.
26732 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26733             Richard Biener  <rguenther@suse.de>
26735         PR middle-end/55851
26736         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26737         types instead of just INTEGER_TYPE types.
26739 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26741         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26742         TF_SIZE): Define.
26744 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26746         PR target/42661
26747         * config/mips/mips.opt: Change mad to mmad to match documentation.
26749 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26751         PR target/55897
26752         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26753         .progmemx.data now.
26755 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26757         PR target/55897
26758         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26759         (avr_addrspace_t): Add .section_name field.
26760         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26761         array size.
26762         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26763         NULL entry.  Put __memx into .progmemx.data.
26764         (progmem_section_prefix): Remove.
26765         (avr_asm_init_sections): No need to initialize progmem_section.
26766         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26767         section name prefix.
26768         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26769         retrieve the progmem section.
26770         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26771         boundary to run over avr_addrspace[].
26772         (avr_register_target_pragmas): Ditto.
26774 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26776         * varasm.c (output_constant_def_contents): For asan_protect_global
26777         protected strings, adjust DECL_ALIGN if needed, before testing for
26778         anchored symbols.
26779         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26780         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26781         normal decls.
26782         (output_object_block): For asan protected decls, emit asan padding
26783         after their contents.
26784         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26785         (asan_finish_file): Test it here instead.
26787 2013-01-07  Nick Clifton  <nickc@redhat.com>
26788             Matthias Klose  <doko@debian.org>
26789             Doug Kwan  <dougkwan@google.com>
26790             H.J. Lu  <hongjiu.lu@intel.com>
26792         PR driver/55470
26793         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26795         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26797         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26799         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26801         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26803 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26805         PR target/54461
26806         * doc/install.texi (Cross-Compiler-Specific Options): Document
26807         --with-avrlibc.
26809 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26811         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26812         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26813         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26814         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26815         vqmovun_high_s64): Fix source operand number and update copyright.
26817 2013-01-07  Richard Biener  <rguenther@suse.de>
26819         PR middle-end/55890
26820         * gimple.h (gimple_call_builtin_p): New overload.
26821         * gimple.c (validate_call): New function.
26822         (gimple_call_builtin_p): Likewise.
26823         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26824         Use gimple_call_builtin_p.
26825         (find_func_clobbers): Likewise.
26826         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26827         (strlen_optimize_stmt): Likewise.
26829 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26831         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26832         (vld1q_dup_*): Likewise.
26833         (vld1_*): Likewise.
26834         (vld1q_*): Likewise.
26835         (vld1_lane_*): Likewise.
26836         (vld1q_lane_*): Likewise.
26838 2013-01-07  Richard Biener  <rguenther@suse.de>
26840         * lto-streamer.h (LTO_minor_version): Bump to 2.
26842 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26844         * config/aarch64/aarch64-protos.h
26845         (aarch64_const_double_zero_rtx_p): Rename to...
26846         (aarch64_float_const_zero_rtx_p): ...this.
26847         (aarch64_float_const_representable_p): New.
26848         (aarch64_output_simd_mov_immediate): Likewise.
26849         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26850         move immediate case.
26851         * config/aarch64/aarch64.c
26852         (aarch64_const_double_zero_rtx_p): Rename to...
26853         (aarch64_float_const_zero_rtx_p): ...this.
26854         (aarch64_print_operand): Allow printing of new constants.
26855         (aarch64_valid_floating_const): New.
26856         (aarch64_legitimate_constant_p): Check for valid floating-point
26857         constants.
26858         (aarch64_simd_valid_immediate): Likewise.
26859         (aarch64_vect_float_const_representable_p): New.
26860         (aarch64_float_const_representable_p): Likewise.
26861         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26862         (aarch64_output_simd_mov_immediate): New.
26863         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26864         (*movdf_aarch64): Likewise.
26865         * config/aarch64/constraints.md (Ufc): New.
26866         (Y): call aarch64_float_const_zero_rtx.
26867         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26869 2013-01-07  Richard Biener  <rguenther@suse.de>
26871         PR tree-optimization/55888
26872         PR tree-optimization/55862
26873         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26874         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26875         not if it is contained therein.
26877 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26879         * config/avr/t-avr: Typo.
26881 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26883         PR55243
26884         * config/avr/t-avr: Don't automatically rebuild
26885         $(srcdir)/config/avr/t-multilib
26886         $(srcdir)/config/avr/avr-tables.opt
26887         $(srcdir)/doc/avr-mmcu.texi
26888         (avr-mcus): New phony target to build them on request.
26889         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26890         * avr/avr-mcus.def: Adjust comments.
26892 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26894         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26896 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26898         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26900 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26902         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26904 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26906         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26907         to generate profiling.
26908         * config/rs6000/aix64.h (LIB_SPEC): Same.
26910 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26912         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26913         New function.
26914         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26916 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26918         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26919         unconditionally.
26920         (ix86_expand_move): Ditto.
26921         (ix86_zero_extend_to_Pmode): Ditto.
26922         (ix86_expand_call): Ditto.
26923         (ix86_expand_special_args_builtin): Ditto.
26924         (ix86_expand_builtin): Ditto.
26926 2013-01-04  Richard Biener  <rguenther@suse.de>
26928         PR tree-optimization/55862
26929         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26930         translating them through PHI nodes.
26932 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26934         PR tree-optimization/55755
26935         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26936         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26938 2013-01-04  Marek Polacek  <polacek@redhat.com>
26940         PR middle-end/55859
26941         * opts.c (default_options_optimization): Clarify error message.
26943 2013-01-04  Richard Biener  <rguenther@suse.de>
26945         PR middle-end/55863
26946         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26947         reassociation.
26949 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26951         PR target/53789
26952         * config/pa/pa.md (movsi): Revert previous change.
26953         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
26954         references.
26956 2013-01-03  Richard Henderson  <rth@redhat.com>
26958         * config/i386/i386.c (ix86_expand_move): Always assign to op1
26959         after eliminating TLS symbols.
26961 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
26963         PR bootstrap/50167
26964         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
26965         * graphite-poly.c (debug_gmp_value): Likewise.
26967 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
26969         PR target/55712
26970         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
26971         selected code model, define __code_mode_small__, __code_model_medium__,
26972         __code_model_large__, __code_model_32__ or __code_model_kernel__.
26973         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
26974         xchg temporary register with %k.  Declare temporary register as
26975         early clobbered.
26976         [__x86_64__]: For medium and large code models, preserve %rbx register.
26978 2013-01-03  Richard Biener  <rguenther@suse.de>
26980         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
26981         (dump_subscript): Adjust.
26982         (finalize_ddr_dependent): Do not dump redundant info.
26983         (analyze_siv_subscript): Adjust.
26984         (subscript_dependence_tester): Likewise.
26985         (compute_affine_dependence): Likewise.
26987 2013-01-03  Richard Biener  <rguenther@suse.de>
26989         Revert
26990         2013-01-03  Richard Biener  <rguenther@suse.de>
26992         PR tree-optimization/55857
26993         * tree-vect-stmts.c (vectorizable_load): Do not setup
26994         re-alignment for invariant loads.
26996         2013-01-02  Richard Biener  <rguenther@suse.de>
26998         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26999         invariant load do not generate a vector load from the scalar location.
27001 2013-01-03  Richard Biener  <rguenther@suse.de>
27003         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27004         for not vectorizing.
27005         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27006         not build INDIRECT_REFs, call get_name once only.
27007         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27008         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27010 2013-01-03  Richard Biener  <rguenther@suse.de>
27012         PR tree-optimization/55857
27013         * tree-vect-stmts.c (vectorizable_load): Do not setup
27014         re-alignment for invariant loads.
27016 2013-01-03  Richard Biener  <rguenther@suse.de>
27018         PR lto/55848
27019         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27020         prefer a built-in decl.
27022 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27024         * gcc.c (process_command): Update copyright notice dates.
27025         * gcov.c (print_version): Likewise.
27026         * gcov-dump.c (print_version): Likewise.
27028         PR rtl-optimization/55838
27029         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27030         iv0.step, iv1.step and step.
27032 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27033             Marc Glisse  <marc.glisse@inria.fr>
27035         PR tree-optimization/55832
27036         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27037         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27038         integer_{one,zero}_node.
27040 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27042         PR debug/54402
27043         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27044         * var-tracking.c (reverse_op): Don't add reverse ops to
27045         VALUEs that have already
27046         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27048 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27050         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27052 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27054         * dumpfile.c (dump_loc): Print filename with location.
27055         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27056         new location_t parameter to emit complete unroll message with
27057         new dump framework.
27058         (canonicalize_loop_induction_variables): Compute loops location
27059         and pass to try_unroll_loop_completely.
27060         * loop-unroll.c (report_unroll_peel): New function.
27061         (peel_loops_completely): Use new dump format with location
27062         for main dumpfile message, and invoke report_unroll_peel on success.
27063         (decide_unrolling_and_peeling): Ditto.
27064         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27065         by report_unroll_peel.
27066         (decide_peel_completely): Ditto.
27067         (decide_unroll_constant_iterations): Ditto.
27068         (decide_unroll_runtime_iterations): Ditto.
27069         (decide_peel_simple): Ditto.
27070         (decide_unroll_stupid): Ditto.
27071         * cfgloop.c (get_loop_location): New function.
27072         * cfgloop.h (get_loop_location): Declare.
27074 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27076         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27077         NULL.
27079 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27081         PR middle-end/55198
27082         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27083         BLKmode objects when EXPAND_MEMORY is specified.
27085 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27087         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27088         in loop predicate.
27089         (fold_builtin_cpu): Do not share cpu model decls across statements.
27091 2013-01-02  Jason Merrill  <jason@redhat.com>
27093         PR c++/55804
27094         * tree.c (build_array_type_1): Revert earlier change.
27096 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27098         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27099         "cortex-a57".
27100         * config/aarch64/aarch64-tune.md: Re-generate.
27102 2013-01-02  Richard Biener  <rguenther@suse.de>
27104         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27105         invariant load do not generate a vector load from the scalar location.
27107 2013-01-02  Richard Biener  <rguenther@suse.de>
27109         PR bootstrap/55784
27110         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27111         * configure: Regenerate.
27113 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27115         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27116         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27117         (expand_builtin_int_roundingfn_2): Keep the original target around
27118         for the fallback case.
27120 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27122         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27123         to be clear for sign changes.
27125 2013-01-01  Jan Hubicka  <jh@suse.cz>
27127         * ipa-inline-analysis.c: Fix formatting.
27129 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27131         PR tree-optimization/55831
27132         * tree-vect-loop.c (get_initial_def_for_induction): Use
27133         gsi_after_labels instead of gsi_start_bb.
27135 Copyright (C) 2013 Free Software Foundation, Inc.
27137 Copying and distribution of this file, with or without modification,
27138 are permitted in any medium without royalty provided the copyright
27139 notice and this notice are preserved.