2013-10-21 Richard Biener <rguenther@suse.de>
[official-gcc.git] / gcc / ChangeLog
blob389c9afd9d506bd3fc913030c54365046e4a63a5
1 2013-10-21  Richard Biener  <rguenther@suse.de>
3         PR tree-optimization/58794
4         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
5         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
7 2013-10-21  Richard Biener  <rguenther@suse.de>
9         PR middle-end/58742
10         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
11         to (T) X for sign-changing conversions (or no conversion).
13 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
15         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
17 2013-10-20  Jan Hubicka  <jh@suse.cz>
19         * config/i386/i386-tune.def: Add comment; organize into categories
21 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
23         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
24         argument.  Update function comment.
25         (expand_set_or_movmem_via_rep): New function combining
26         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
27         (expand_movmem_via_rep_mov): Remove.
28         expand_setmem_via_rep_stos): Remove.
29         (expand_movmem_epilogue): Update calls correspondingly.
30         (expand_setmem_epilogue_via_loop): Likewise.
31         (emit_memset): New.
32         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
33         (expand_set_or_movmem_prologue): New function combining
34         expand_movmem_prologue and expand_setmem_prologue.
35         (expand_movmem_prologue): Remove.
36         (expand_setmem_prologue): Remove.
37         (expand_set_or_movmem_constant_prologue): New function combining
38         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
39         (expand_constant_movmem_prologue): Remove.
40         (expand_constant_setmem_prologue): Remove.
41         (promote_duplicated_reg): Allow vector-const0 value.
42         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
43         and ix86_expand_setmem.
44         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
45         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
47 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
49         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
50         little endian.
51         (vec_unpacku_hi_v8hi): Likewise.
52         (vec_unpacku_lo_v16qi): Likewise.
53         (vec_unpacku_lo_v8hi): Likewise.
55 2013-10-20  Jan Hubicka  <jh@suse.cz>
57         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
58         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
59         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
61 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
63         * config/mips/mips.h (ISA_HAS_WSBH): Define.
64         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
65         constants.
66         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
68 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
70         PR target/58603
71         * system.h: Undef m_slot.
73 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
75         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
76         all elements for both endian flavors.
78 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
80         PR target/58792
81         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
82         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
83         and SI_REG for 64bit SYSV ABI targets.
85 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
87         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
88         to multi_reg_return.  Clarify that we are skipping USEs of multiple
89         return registers.  Use bool type where appropriate.
91 2013-10-18  Jan Hubicka  <jh@suse.cz>
93         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
94         for cold functions.
95         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
96         (X86_TUNE_PUSH_MEMORY): Likewise.
97         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
98         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
99         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
100         New.
101         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
102         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
103         Remove.
104         (ix86_option_override_internal): Update to use tune features instead
105         of variables.
107 2013-10-18  Cong Hou  <congh@google.com>
109         PR tree-optimization/58508
110         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
111         statement that contains data refs with zero-step.
113 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
115         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
116         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
117         * gimple-low.c (gimple_check_call_arg,
118         gimple_check_call_matching_types): Move to cgraph.c.
119         * gimple-low.h: Remove prototype.
120         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
121         Relocate from gimple-low.c.
122         * cgraph.h: Add prototype.  Don't include basic-block.h.
123         * gimplify.c: Add gimple-low to include list.
124         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
125         * tree-eh.c: Add gimple-low to include list.
126         * tree-nested.c: Likewise.
127         * cfgexpand.c: Add tree-ssa-address.h to include list.
128         * expr.c: Likewise.
129         * gimple-fold.c: Likewise.
130         * gimple-ssa-strength-reduction.c: Likewise.
131         * trans-mem.c: Likewise.
132         * tree-mudflap.c: Likewise.
133         * tree-ssa-loop-ivopts.c: Likewise.
134         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
135         (degenerate_phi_result): Move to tree-phinodes.c.
136         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
137         * tree-ssa-threadedge.c: Likewise.
138         * tree-vrp.c: Likewise.
139         * tree-phinodes.c (degenerate_phi_result): Relocate here.
140         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
141         * tree-phinodes.h (degenerate_phi_result): Add prototype.
142         * tree-ssa-copy.c: Include tree-ssa-dom.h.
143         * tree-ssa-forwprop.c: Likewise.
144         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
145         pass_data_cleanup_cfg_post_optimizing,
146         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
147         * tree-optimize.c: Delete File.
148         * graphite.c: Include tree-cfgcleanup.h.
149         * passes.c: Likewise.
150         * tree-cfg.c: Likewise.
151         * tree-profile.c: Likewise.
152         * tree-ssa-dse.c: Likewise.
153         * tree-ssa-loop-ivcanon.c: Likewise.
154         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
155         * tree-outof-ssa.c: Include sbitmap.h.
156         * tree-ssa-live.c: Likewise.
157         * tree-ssa-propagate.c: Likewise.
158         * tree-ssa-structalias.c: Likewise.
159         * tree-stdarg.c: Likewise.
160         * Makefile.in (OBJS): Delete tree-optimize.o.
161         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
162         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
163         * varasm.c: Include basic-block.h.
164         * cfgloop.h: Include function.h instead of basic-block.h
165         (bb_loop_depth): Move to cfgloop.c.
166         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
168 2013-10-18  Teresa Johnson  <tejohnson@google.com>
170         * predict.c (probably_never_executed): Compare frequency-based
171         count to number of training runs.
172         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
174 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
176         * config/arm/arm.c (cortexa9_extra_costs): New table.
177         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
179 2013-10-18  Jeff Law  <law@redhat.com>
181         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
183         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
184         (register_jump_thread): Add "registered_jump_thread" debug
185         counter support.
186         * dbgcnt.def (registered_jump_thread): New debug counter.
188 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
190         * config/rs6000/rs6000.c: Include cgraph.h.
192 2013-10-18  Teresa Johnson  <tejohnson@google.com>
194         * tree-ssa-tail-merge.c (replace_block_by): Update edge
195         weights during merging.
197 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
199         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
200         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
201         * tree-ssa-operands.h: Remove prototype.
202         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
203         * gimple.c (virtual_operand_p): Relocate from gimple.c.
204         * gimple.h: Add prototype.
205         * gimple-ssa.h: Include tree-ssa-operands.h.
206         * tree-dump.c: Add tree-cfg.h to include list.
207         * tree-ssa-alias.c: Add ipa-reference.h to include list.
208         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
209         * config/i386/i386.c: Don't include tree-flow.h.
210         * config/rs6000/rs6000.c: Likewise.
212 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
214         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
216 2013-10-18  Richard Biener  <rguenther@suse.de>
218         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
219         or TYPE_UNSIGNED of integral types.
220         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
221         NULL_TREE for zero-precision integral types.
223 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
225         * config/aarch64/arm_neon.h
226         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
228 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
230         * opth-gen.awk: Define target_flags_explicit.
232 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
234         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
235         fields to the reg_addr array that describes the valid addressing
236         mode for any register, general purpose registers, floating point
237         registers, and Altivec registers.
238         (FIRST_RELOAD_REG_CLASS): Likewise.
239         (LAST_RELOAD_REG_CLASS): Likewise.
240         (struct reload_reg_map_type): Likewise.
241         (reload_reg_map_type): Likewise.
242         (RELOAD_REG_VALID): Likewise.
243         (RELOAD_REG_MULTIPLE): Likewise.
244         (RELOAD_REG_INDEXED): Likewise.
245         (RELOAD_REG_OFFSET): Likewise.
246         (RELOAD_REG_PRE_INCDEC): Likewise.
247         (RELOAD_REG_PRE_MODIFY): Likewise.
248         (reg_addr): Likewise.
249         (mode_supports_pre_incdec_p): New helper functions to say whether
250         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
251         (mode_supports_pre_modify_p): Likewise.
252         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
253         print the valid address mode bits for each mode.
254         (rs6000_debug_print_mode): Likewise.
255         (rs6000_debug_reg_global): Likewise.
256         (rs6000_setup_reg_addr_masks): New function to set up the address
257         mask bits for each type.
258         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
259         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
260         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
261         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
262         PRE_MODIFY are supported.
263         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
264         registers, instead of {src,dest}_av_p.
265         (rs6000_print_options_internal): Tweak the debug output slightly.
267 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
269         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
270         isa attribute.
272 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
274         * tree-flow.h (struct omp_region): Move to omp-low.c.
275         Remove omp_ prototypes and variables.
276         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
277         (copy_var_decl): Relocate prototype from tree-flow.h.
278         * gimple.c (copy_var_decl): Relocate from omp-low.c.
279         * tree.h: Move prototype to omp-low.h.
280         * omp-low.h: New File.  Relocate prototypes here.
281         * omp-low.c (struct omp_region): Make local here.
282         (root_omp_region): Make static.
283         (copy_var_decl) Move to gimple.c.
284         (new_omp_region): Make static.
285         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
286         * tree-cfg.c: Include omp-low.h.
287         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
288         * gimplify.c: Include omp-low.h.
289         * tree-parloops.c: Likewise.
291 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
293         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
294         operands are in memory, prefer to force non-matched operand 1 to
295         the register.
297 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
299         PR target/58673
300         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
301         restrict TImode addresses to single indirect registers if both
302         -mquad-memory and -mvsx-timode are used.
303         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
304         we should emit load/store quad.  Remove using %y for quad memory
305         addresses.
307         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
308         constraints to allow load/store quad on machines where TImode is
309         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
310         for TImode to load integer constants.
312 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
314         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
316 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
318         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
319         handling of STACK_REG.
321 2013-10-17  Richard Biener  <rguenther@suse.de>
323         PR tree-optimization/58143
324         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
325         New function.
326         (rewrite_to_defined_overflow): Likewise.
327         (move_computations_dom_walker::before_dom): Rewrite stmts
328         with undefined signed overflow that are not always executed
329         into unsigned arithmetic.
331 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
333         PR target/57756
334         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
335         explicit isa flag to be an options variable, instead of using
336         global_options_set.  Remove define from rs6000.h.
337         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
339         * config/rs6000/rs6000.c (rs6000_option_override_internal):
340         Initialize rs6000_isa_flags_explicit.
341         (rs6000_function_specific_save): Add gcc_options* parameter, so
342         that the powerpc builds after the 2013-10-15 changes.
343         (rs6000_function_specific_restore): Likewise.
345 2013-10-16  DJ Delorie  <dj@redhat.com>
347         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
348         op is a REG before checking REGNO.
349         (rl78_alloc_physical_registers): Verify pattern is a SET before
350         checking SET_SRC.
352 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
354         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
355         endianness.
356         (vec_unpacks_lo_v4sf): Likewise.
357         (vec_unpacks_float_hi_v4si): Likewise.
358         (vec_unpacks_float_lo_v4si): Likewise.
359         (vec_unpacku_float_hi_v4si): Likewise.
360         (vec_unpacku_float_lo_v4si): Likewise.
362 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
364         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
365         (vsx_concat_v2sf): Likewise.
367 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
369         * config/aarch64/aarch64.md
370         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
372 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
374         PR tree-optimization/58697
375         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
376         estimated_loop_iterations_int.
377         (max_stmt_executions_int): Call get_max_loop_iterations_int.
378         (get_max_loop_iterations_int): New.  HWINT version of
379         get_max_loop_iterations.
380         * cfgloop.h: Add prototypes.
381         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
382         * loop-unroll.c (decide_peel_once_rolling): Call
383         get_estimated_loop_iterations_int.
384         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
385         * tree-ssa-loop-niter.h: Tweak prototypes.
387 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
389         * gengtype-parse.c (struct_field_seq): Ignore access-control
390         keywords ("public:" etc).
392 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
394         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
395         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
397 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
399         * config/arm/arm.opt (mlra): New option.
400         * config/arm/arm.c (arm_lra_p): New function.
401         (TARGET_LRA_P): Define.
403 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
405         * tree-core.h (tree_code_name): Remove.
406         * tree.h (get_tree_code_name): New prototype.
407         * tree.c (tree_code_name): Make static.
408         (get_tree_code_name): New function.
409         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
410         tree_class_check_failed, tree_range_check_failed,
411         tree_not_class_check_failed, omp_clause_check_failed,
412         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
413         wrapper get_tree_code_name instead of calling tree_code_name directly.
414         * tree-vrp.c (dump_asserts_for): Likewise.
415         * tree-dump.c (dequeue_and_dump): Likewise.
416         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
417         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
418         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
419         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
420         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
421         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
422         dump_gimple_omp_for): Likewise.
423         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
424         * tree-ssa-pre.c (print_pre_expr): Likewise.
425         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
426         * print-tree.c (print_node_brief, print_node): Likewise.
427         * gimple.c (gimple_check_failed): Likewise.
428         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
429         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
430         * config/mep/mep.c (mep_validate_vliw): Likewise.
431         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
432         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
434 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
436         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
437         for AMD bdver3.
439 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
441         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
442         (MULTILIB_MATCHES): Add multilib for -march=v8.
444 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
446         PR target/57756
447         * optc-save-gen.awk: Add extra parameter to the save and restore
448         target calls.
449         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
450         * tree.c (build_optimization_node): New parameter.  Add extra parameter
451         to call to cl_optimization_save.
452         (build_target_option_node): New parameter. Add extra parameter
453         to call to cl_target_option_save.
454         * tree.h (build_optimization_node): New parameter.
455         (build_target_option_node): New parameter.
456         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
457         build_optimization_node and build_target_option_node.
458         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
459         (handle_pragma_push_options): Ditto.
460         * toplev.c (process_options): Ditto.
461         * opts.c (init_options_struct): Check for opts_set non-null.
462         * target.def (target_option.save): New parameter.
463         (target_option.restore): New parameter.
464         * tm.texi: Generate.
465         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
466         (ix86_pragma_target_parse): Ditto.
467         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
468         parameters.
469         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
470         to  build_optimization_node and build_target_option_node.
471         (rs6000_valid_attribute_p): Ditto.
472         (rs6000_pragma_target_parse): Ditto.
473         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
474         data.
475         * config/i386/i386.h:
476         TARGET_64BIT_P: New Macro
477         TARGET_MMX_P: New Macro.
478         TARGET_3DNOW_P: New Macro.
479         TARGET_3DNOW_A_P: New Macro.
480         TARGET_SSE_P: New Macro.
481         TARGET_SSE2_P: New Macro.
482         TARGET_SSE3_P: New Macro.
483         TARGET_SSSE3_P: New Macro.
484         TARGET_SSE4_1_P: New Macro.
485         TARGET_SSE4_2_P: New Macro.
486         TARGET_AVX_P: New Macro.
487         TARGET_AVX2_P: New Macro.
488         TARGET_AVX512F_P: New Macro.
489         TARGET_AVX512PF_P: New Macro.
490         TARGET_AVX512ER_P: New Macro.
491         TARGET_AVX512CD_P: New Macro.
492         TARGET_FMA_P: New Macro.
493         TARGET_SSE4A_P: New Macro.
494         TARGET_FMA4_P: New Macro.
495         TARGET_XOP_P: New Macro.
496         TARGET_LWP_P: New Macro.
497         TARGET_ABM_P: New Macro.
498         TARGET_BMI_P: New Macro.
499         TARGET_BMI2_P: New Macro.
500         TARGET_LZCNT_P: New Macro.
501         TARGET_TBM_P: New Macro.
502         TARGET_POPCNT_P: New Macro.
503         TARGET_SAHF_P: New Macro.
504         TARGET_MOVBE_P: New Macro.
505         TARGET_CRC32_P: New Macro.
506         TARGET_AES_P: New Macro.
507         TARGET_PCLMUL_P: New Macro.
508         TARGET_CMPXCHG16B_P: New Macro.
509         TARGET_FSGSBASE_P: New Macro.
510         TARGET_RDRND_P: New Macro.
511         TARGET_F16C_P: New Macro.
512         TARGET_RTM_P: New Macro.
513         TARGET_HLE_P: New Macro.
514         TARGET_RDSEED_P: New Macro.
515         TARGET_PRFCHW_P: New Macro.
516         TARGET_ADX_P: New Macro.
517         TARGET_FXSR_P: New Macro.
518         TARGET_XSAVE_P: New Macro.
519         TARGET_XSAVEOPT_P: New Macro.
520         TARGET_LP64_P: New Macro.
521         TARGET_X32_P: New Macro.
522         TARGET_FPMATH_DEFAULT_P: New Macro.
523         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
524         * config/i386/i386.c (ix86_option_override_internal): New parameters.
525         opts and opts_set.
526         Change ix86_tune_string to access opts->x_ix86_tune_string.
527         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
528         Change ix86_arch_string to access opts->x_ix86_arch_string.
529         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
530         Change ix86_pmode to access opts->x_ix86_pmode.
531         Change ix86_abi to access opts->x_ix86_abi.
532         Change ix86_cmodel to access opts->x_ix86_cmodel.
533         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
534         Change ix86_isa_flags_explicit to access
535         opts->x_ix86_isa_flags_explicit.
536         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
537         Change ix86_regparm to access opts->x_ix86_regparm.
538         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
539         Change ix86_preferred_stack_boundary_arg to access
540         opts->x_ix86_preferred_stack_boundary_arg.
541         Change ix86_force_align_arg_pointer to access
542         opts->x_ix86_force_align_arg_pointer.
543         Change ix86_incoming_stack_boundar_arg to access
544         opts->x_ix86_incoming_stack_boundar_arg.
545         Change ix86_fpmath to access opts->x_ix86_fpmath.
546         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
547         Change ix86_recip_name to access opts->x_ix86_recip_name.
548         Change ix86_stack_protector_guard to access
549         opts->x_ix86_stack_protector_guard.
550         Change ix86_tune_memcpy_strategy to access
551         opts->x_ix86_tune_memcpy_strategy.
552         Change ix86_tune_memset_strategy to access
553         opts->x_ix86_tune_memset_strategy.
554         Change global_options to access opts.
555         Change global_options_set to access opts_set.
556         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
557         Change TARGET_MMX to TARGET_MMX_P (opts->...).
558         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
559         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
560         Change TARGET_SSE to TARGET_SSE_P (opts->...).
561         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
562         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
563         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
564         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
565         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
566         Change TARGET_AVX to TARGET_AVX_P (opts->...).
567         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
568         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
569         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
570         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
571         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
572         Change TARGET_FMA to TARGET_FMA_P (opts->...).
573         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
574         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
575         Change TARGET_XOP to TARGET_XOP_P (opts->...).
576         Change TARGET_LWP to TARGET_LWP_P (opts->...).
577         Change TARGET_ABM to TARGET_ABM_P (opts->...).
578         Change TARGET_BMI to TARGET_BMI_P (opts->...).
579         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
580         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
581         Change TARGET_TBM to TARGET_TBM_P (opts->...).
582         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
583         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
584         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
585         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
586         Change TARGET_AES to TARGET_AES_P (opts->...).
587         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
588         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
589         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
590         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
591         Change TARGET_F16C to TARGET_F16C_P (opts->...).
592         Change TARGET_RTM to TARGET_RTM_P (opts->...).
593         Change TARGET_HLE to TARGET_HLE_P (opts->...).
594         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
595         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
596         Change TARGET_ADX to TARGET_ADX_P (opts->...).
597         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
598         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
599         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
600         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
601         Change TARGET_X32 to TARGET_X32_P (opts->...).
602         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
603         Change TARGET_FLOAT_RETURNS_IN_80387 to
604         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
605         (ix86_function_specific_save): New parameter. Use opts-> fields
606         to replace global fields.
607         (ix86_function_specific_restore): Ditto.
608         (ix86_valid_target_attribute_inner_p): New parameters.
609         Fix recursive call.
610         Fix call to ix86_handle_option and set_option.
611         (ix86_valid_target_attribute_tree): New parameters.
612         Change global_options to access opts.
613         Change global_options_set to access opts_set.
614         Fix call to ix86_valid_target_attribute_inner_p.
615         Change ix86_tune_string to access opts->x_ix86_tune_string.
616         Change ix86_arch_string to access opts->x_ix86_arch_string.
617         Change ix86_fpmath to access opts->x_ix86_fpmath
618         Fix call to ix86_option_override_internal.
619         Fix call to ix86_add_new_builtins.
620         Fix calls to build_optimization_node and build_target_option_node.
621         (ix86_valid_target_attribute_p): Remove access to global_options.
622         Use new gcc_options structure func_options.
623         Fix call to ix86_valid_target_attribute_tree.
624         Fix call to  build_optimization_node.
625         (get_builtin_code_for_version): Fix call to
626         ix86_valid_target_attribute_tree.
628 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
630         * Makefile.in (PICFLAG): New.
631         (enable_host_shared): New.
632         (INTERNAL_CFLAGS): Use PICFLAG.
633         (LIBIBERTY): Use pic build of libiberty.a if configured with
634         --enable-host-shared.
635         * configure.ac: Add --enable-host-shared, setting up new
636         PICFLAG variable.
637         * configure: Regenerate.
638         * doc/install.texi (--enable-shared): Add note contrasting it with ...
639         (--enable-host-shared): New option.
641 2013-10-15  Richard Biener  <rguenther@suse.de>
643         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
644         for built-in functions.
646 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
648         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
649         (optimize_range_tests_1): New function,
650         extracted from optimize_range_tests.
651         (optimize_range_tests_xor): Similarly.
652         (optimize_range_tests_diff): New function.
653         (optimize_range_tests): Use optimize_range_tests_1.
655 2013-10-15  Cong Hou  <congh@google.com>
657         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
658         requirement of the reduction pattern so that one operand of the
659         reduction operation can come from outside of the loop.
661 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
663         * config/arm/neon-schedgen.ml: Remove.
664         * config/arm/cortex-a9-neon.md: Remove comment regarding
665         neon-schedgen.ml.
667 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
669         * config/arm/types: Remove old neon types.
671 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
673         * config/arm/cortex-a7.md
674         (cortex_a7_neon_type): New.
675         (cortex_a7_neon_mul): Update for new types.
676         (cortex_a7_neon_mla): Likewise.
677         (cortex_a7_neon): Likewise.
679 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
681         * config/arm/cortex-a15-neon.md
682         (cortex_a15_neon_type): New,
684         (cortex_a15_neon_int_1): Remove.
685         (cortex_a15_neon_int_2): Likewise.
686         (cortex_a15_neon_int_3): Likewise.
687         (cortex_a15_neon_int_4): Likewise.
688         (cortex_a15_neon_int_5): Likewise.
689         (cortex_a15_neon_vqneg_vqabs): Likewise.
690         (cortex_a15_neon_vmov): Likewise.
691         (cortex_a15_neon_vaba): Likewise.
692         (cortex_a15_neon_vaba_qqq): Likewise.
693         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
694         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
695         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
696         Likewise.
697         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
698         (cortex_a15_neon_mla_qqq_8_16): Likewise.
699         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
700         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
701         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
702         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
703         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
704         (cortex_a15_neon_shift_1): Likewise.
705         (cortex_a15_neon_shift_2): Likewise.
706         (cortex_a15_neon_shift_3): Likewise.
707         (cortex_a15_neon_vshl_ddd): Likewise.
708         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
709         (cortex_a15_neon_vsra_vrsra): Likewise.
710         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
711         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
712         (cortex_a15_neon_bp_3cycle): Likewise.
713         (cortex_a15_neon_ldm_2): Likewise.
714         (cortex_a15_neon_stm_2): Likewise.
715         (cortex_a15_neon_mcr): Likewise.
716         (cortex_a15_neon_mrc): Likewise.
717         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
718         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
719         (cortex_a15_neon_fp_vmul_ddd): Likewise.
720         (cortex_a15_neon_fp_vmul_qqd): Likewise.
721         (cortex_a15_neon_fp_vmla_ddd): Likewise.
722         (cortex_a15_neon_fp_vmla_qqq): Likewise.
723         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
724         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
725         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
726         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
727         (cortex_a15_neon_bp_simple): Likewise.
728         (cortex_a15_neon_bp_2cycle): Likewise.
729         (cortex_a15_neon_bp_3cycle): Likewise.
730         (cortex_a15_neon_vld1_1_2_regs): Likewise.
731         (cortex_a15_neon_vld1_3_4_regs): Likewise.
732         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
733         (cortex_a15_neon_vld2_4_regs): Likewise.
734         (cortex_a15_neon_vld3_vld4): Likewise.
735         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
736         (cortex_a15_neon_vst1_3_4_regs): Likewise.
737         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
738         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
739         (cortex_a15_neon_vst3_vst4): Rename to...
740         (cortex_a15_neon_vst4): This, update for new attributes.
741         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
742         (cortex_a15_neon_vld3_vld4_lane): Likewise.
743         (cortex_a15_neon_vst1_vst2_lane): Likewise.
744         (cortex_a15_neon_vst3_vst4_lane): Likewise.
745         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
746         (cortex_a15_neon_ldm_2): Likewise.
747         (cortex_a15_neon_stm_2): Likewise.
748         (cortex_a15_neon_mcr): Likewise.
749         (cortex_a15_neon_mcr_2_mcrr): Likewise.
750         (cortex_a15_neon_mrc): Likewise.
751         (cortex_a15_neon_mrrc): Likewise.
753         (cortex_a15_neon_abd): New.
754         (cortex_a15_neon_abd_q): Likewise.
755         (cortex_a15_neon_aba): Likewise.
756         (cortex_a15_neon_aba_q): Likewise.
757         (cortex_a15_neon_acc): Likewise.
758         (cortex_a15_neon_acc_q): Likewise.
759         (cortex_a15_neon_arith_basic): Likewise.
760         (cortex_a15_neon_arith_complex): Likewise.
761         (cortex_a15_neon_multiply): Likewise.
762         (cortex_a15_neon_multiply_q): Likewise.
763         (cortex_a15_neon_mla): Likewise.
764         (cortex_a15_neon_mla_q): Likewise.
765         (cortex_a15_neon_sat_mla_long): Likewise.
766         (cortex_a15_neon_shift_acc): Likewise.
767         (cortex_a15_neon_shift_imm_basic): Likewise.
768         (cortex_a15_neon_shift_imm_complex): Likewise.
769         (cortex_a15_neon_shift_reg_basic): Likewise.
770         (cortex_a15_neon_shift_reg_basic_q): Likewise.
771         (cortex_a15_neon_shift_reg_complex): Likewise.
772         (cortex_a15_neon_shift_reg_complex_q): Likewise.
773         (cortex_a15_neon_fp_negabs): Likewise
774         (cortex_a15_neon_fp_arith): Likewise
775         (cortex_a15_neon_fp_arith_q): Likewise
776         (cortex_a15_neon_fp_cvt_int): Likewise
777         (cortex_a15_neon_fp_cvt_int_q): Likewise
778         (cortex_a15_neon_fp_cvt_16): Likewise
779         (cortex_a15_neon_fp_mul): Likewise
780         (cortex_a15_neon_fp_mul_q): Likewise
781         (cortex_a15_neon_fp_mla): Likewise
782         (cortex_a15_neon_fp_mla_q): Likewise
783         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
784         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
785         (cortex_a15_neon_bitops): Likewise.
786         (cortex_a15_neon_bitops_q): Likewise.
787         (cortex_a15_neon_from_gp): Likewise.
788         (cortex_a15_neon_from_gp_q): Likewise.
789         (cortex_a15_neon_tbl3_tbl4): Likewise.
790         (cortex_a15_neon_zip_q): Likewise.
791         (cortex_a15_neon_to_gp): Likewise.
792         (cortex_a15_neon_load_a): Likewise.
793         (cortex_a15_neon_load_b): Likewise.
794         (cortex_a15_neon_load_c): Likewise.
795         (cortex_a15_neon_load_d): Likewise.
796         (cortex_a15_neon_load_e): Likewise.
797         (cortex_a15_neon_load_f): Likewise.
798         (cortex_a15_neon_store_a): Likewise.
799         (cortex_a15_neon_store_b): Likewise.
800         (cortex_a15_neon_store_c): Likewise.
801         (cortex_a15_neon_store_d): Likewise.
802         (cortex_a15_neon_store_e): Likewise.
803         (cortex_a15_neon_store_f): Likewise.
804         (cortex_a15_neon_store_g): Likewise.
805         (cortex_a15_neon_store_h): Likewise.
806         (cortex_a15_vfp_to_from_gp): Likewise.
808 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
810         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
812         (cortex_a9_neon_vshl_ddd): Remove.
813         (cortex_a9_neon_vst3_vst4): Likewise.
814         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
816         (cortex_a9_neon_bit_ops_q): New.
818         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
819         (cortex_a9_neon_int_2): Likewise.
820         (cortex_a9_neon_int_3): Likewise.
821         (cortex_a9_neon_int_4): Likewise.
822         (cortex_a9_neon_int_5): Likewise.
823         (cortex_a9_neon_vqneg_vqabs): Likewise.
824         (cortex_a9_neon_vmov): Likewise.
825         (cortex_a9_neon_vaba): Likewise.
826         (cortex_a9_neon_vaba_qqq): Likewise.
827         (cortex_a9_neon_shift_1): Likewise.
828         (cortex_a9_neon_shift_2): Likewise.
829         (cortex_a9_neon_shift_3): Likewise.
830         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
831         (cortex_a9_neon_vsra_vrsra): Likewise.
832         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
833         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
834         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
835         Likewise.
836         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
837         (cortex_a9_neon_mla_qqq_8_16): Likewise.
838         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
839         Likewise.
840         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
841         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
842         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
843         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
844         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
845         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
846         (cortex_a9_neon_fp_vsum): Likewise.
847         (cortex_a9_neon_fp_vmul_ddd): Likewise.
848         (cortex_a9_neon_fp_vmul_qqd): Likewise.
849         (cortex_a9_neon_fp_vmla_ddd): Likewise.
850         (cortex_a9_neon_fp_vmla_qqq): Likewise.
851         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
852         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
853         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
854         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
855         (cortex_a9_neon_bp_simple): Likewise.
856         (cortex_a9_neon_bp_2cycle): Likewise.
857         (cortex_a9_neon_bp_3cycle): Likewise.
858         (cortex_a9_neon_ldr): Likewise.
859         (cortex_a9_neon_str): Likewise.
860         (cortex_a9_neon_vld1_1_2_regs): Likewise.
861         (cortex_a9_neon_vld1_3_4_regs): Likewise.
862         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
863         (cortex_a9_neon_vld2_4_regs): Likewise.
864         (cortex_a9_neon_vld3_vld4): Likewise.
865         (cortex_a9_neon_vld1_vld2_lane): Likewise.
866         (cortex_a9_neon_vld3_vld4_lane): Likewise.
867         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
868         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
869         (cortex_a9_neon_vst1_3_4_regs): Likewise.
870         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
871         (cortex_a9_neon_vst1_vst2_lane): Likewise.
872         (cortex_a9_neon_vst3_vst4_lane): Likewise.
873         (cortex_a9_neon_mcr): Likewise.
874         (cortex_a9_neon_mcr_2_mcrr): Likewise.
875         (cortex_a9_neon_mrc): Likewise.
876         (cortex_a9_neon_mrrc): Likewise.
878 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
880         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
882         (cortex_a8_neon_vshl_ddd): Remove.
883         (cortex_a8_neon_vst3_vst4): Likewise.
884         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
886         (cortex_a8_neon_bit_ops_q): New.
888         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
889         (cortex_a8_neon_int_2): Likewise..
890         (cortex_a8_neon_int_3): Likewise.
891         (cortex_a8_neon_int_5): Likewise.
892         (cortex_a8_neon_vqneg_vqabs): Likewise.
893         (cortex_a8_neon_int_4): Likewise.
894         (cortex_a8_neon_vaba): Likewise.
895         (cortex_a8_neon_vaba_qqq): Likewise.
896         (cortex_a8_neon_shift_1): Likewise.
897         (cortex_a8_neon_shift_2): Likewise.
898         (cortex_a8_neon_shift_3): Likewise.
899         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
900         (cortex_a8_neon_vsra_vrsra): Likewise.
901         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
902         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
903         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
904         Likewise.
905         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
906         (cortex_a8_neon_mla_qqq_8_16): Likewise.
907         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
908         Likewise.
909         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
910         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
911         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
912         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
913         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
914         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
915         (cortex_a8_neon_fp_vsum): Likewise.
916         (cortex_a8_neon_fp_vmul_ddd): Likewise.
917         (cortex_a8_neon_fp_vmul_qqd): Likewise.
918         (cortex_a8_neon_fp_vmla_ddd): Likewise.
919         (cortex_a8_neon_fp_vmla_qqq): Likewise.
920         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
921         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
922         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
923         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
924         (cortex_a8_neon_bp_simple): Likewise.
925         (cortex_a8_neon_bp_2cycle): Likewise.
926         (cortex_a8_neon_bp_3cycle): Likewise.
927         (cortex_a8_neon_ldr): Likewise.
928         (cortex_a8_neon_str): Likewise.
929         (cortex_a8_neon_vld1_1_2_regs): Likewise.
930         (cortex_a8_neon_vld1_3_4_regs): Likewise.
931         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
932         (cortex_a8_neon_vld2_4_regs): Likewise.
933         (cortex_a8_neon_vld3_vld4): Likewise.
934         (cortex_a8_neon_vld1_vld2_lane): Likewise.
935         (cortex_a8_neon_vld3_vld4_lane): Likewise.
936         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
937         (cortex_a8_neon_vst1_3_4_regs): Likewise.
938         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
939         (cortex_a8_neon_vst1_vst2_lane): Likewise.
940         (cortex_a8_neon_vst3_vst4_lane): Likewise.
941         (cortex_a8_neon_mcr): Likewise.
942         (cortex_a8_neon_mcr_2_mcrr): Likewise.
943         (cortex_a8_neon_mrc): Likewise.
944         (cortex_a8_neon_mrrc): Likewise.
946 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
948         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
949         (fp): New.
950         * config/aarch64/aarch64-simd.md (neon_type): Remove.
951         (aarch64_simd_dup<mode>): Add "type" attribute.
952         (aarch64_dup_lane<mode>): Likewise.
953         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
954         (*aarch64_simd_mov<mode>): Likewise.
955         (aarch64_simd_mov_from_<mode>low): Likewise.
956         (aarch64_simd_mov_from_<mode>high): Likewise.
957         (orn<mode>3): Likewise.
958         (bic<mode>3): Likewise.
959         (add<mode>3): Likewise.
960         (sub<mode>3): Likewise.
961         (mul<mode>3): Likewise.
962         (*aarch64_mul3_elt<mode>): Likewise.
963         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
964         (*aarch64_mul3_elt_to_128df): Likewise.
965         (*aarch64_mul3_elt_to_64v2df): Likewise.
966         (neg<mode>2): Likewise.
967         (abs<mode>2): Likewise.
968         (abd<mode>_3): Likewise.
969         (aba<mode>_3): Likewise.
970         (fabd<mode>_3): Likewise.
971         (*fabd_scalar<mode>3): Likewise.
972         (and<mode>3): Likewise.
973         (ior<mode>3): Likewise.
974         (xor<mode>3): Likewise.
975         (one_cmpl<mode>2): Likewise.
976         (aarch64_simd_vec_set<mode>): Likewise.
977         (aarch64_simd_lshr<mode>): Likewise.
978         (aarch64_simd_ashr<mode>): Likewise.
979         (aarch64_simd_imm_shl<mode>): Likewise.
980         (aarch64_simd_reg_sshl<mode): Likewise.
981         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
982         (aarch64_simd_reg_shl<mode>_signed): Likewise.
983         (aarch64_simd_vec_setv2di): Likewise.
984         (aarch64_simd_vec_set<mode>): Likewise.
985         (aarch64_mla<mode>): Likewise.
986         (*aarch64_mla_elt<mode>): Likewise.
987         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
988         (aarch64_mls<mode>): Likewise.
989         (*aarch64_mls_elt<mode>): Likewise.
990         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
991         (<su><maxmin><mode>3): Likewise.
992         (move_lo_quad_<mode>): Likewise.
993         (aarch64_simd_move_hi_quad_<mode>): Likewise.
994         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
995         (vec_pack_trunc_<mode>): Likewise.
996         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
997         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
998         (*aarch64_<su>mlal_lo<mode>): Likewise.
999         (*aarch64_<su>mlal_hi<mode>): Likewise.
1000         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1001         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1002         (*aarch64_<su>mlal<mode>): Likewise.
1003         (*aarch64_<su>mlsl<mode>): Likewise.
1004         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1005         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1006         (add<mode>3): Likewise.
1007         (sub<mode>3): Likewise.
1008         (mul<mode>3): Likewise.
1009         (div<mode>3): Likewise.
1010         (neg<mode>2): Likewise.
1011         (abs<mode>2): Likewise.
1012         (fma<mode>4): Likewise.
1013         (*aarch64_fma4_elt<mode>): Likewise.
1014         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1015         (*aarch64_fma4_elt_to_128df): Likewise.
1016         (*aarch64_fma4_elt_to_64v2df): Likewise.
1017         (fnma<mode>4): Likewise.
1018         (*aarch64_fnma4_elt<mode>): Likewise.
1019         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1020         (*aarch64_fnma4_elt_to_128df): Likewise.
1021         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1022         (<frint_pattern><mode>2): Likewise.
1023         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1024         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1025         (vec_unpacks_lo_v4sf): Likewise.
1026         (aarch64_float_extend_lo_v2df): Likewise.
1027         (vec_unpacks_hi_v4sf): Likewise.
1028         (aarch64_float_truncate_lo_v2sf): Likewise.
1029         (aarch64_float_truncate_hi_v4sf): Likewise.
1030         (aarch64_vmls<mode>): Likewise.
1031         (<su><maxmin><mode>3): Likewise.
1032         (<maxmin_uns><mode>3): Likewise.
1033         (reduc_<sur>plus_<mode>): Likewise.
1034         (reduc_<sur>plus_v2di): Likewise.
1035         (reduc_<sur>plus_v2si): Likewise.
1036         (reduc_<sur>plus_<mode>): Likewise.
1037         (aarch64_addpv4sf): Likewise.
1038         (clz<mode>2): Likewise.
1039         (reduc_<maxmin_uns>_<mode>): Likewise.
1040         (reduc_<maxmin_uns>_v2di): Likewise.
1041         (reduc_<maxmin_uns>_v2si): Likewise.
1042         (reduc_<maxmin_uns>_<mode>): Likewise.
1043         (reduc_<maxmin_uns>_v4sf): Likewise.
1044         (aarch64_simd_bsl<mode>_internal): Likewise.
1045         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1046         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1047         (aarch64_get_lane<mode>): Likewise.
1048         (*aarch64_combinez<mode>): Likewise.
1049         (aarch64_combine<mode>): Likewise.
1050         (aarch64_simd_combine<mode>): Likewise.
1051         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1052         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1053         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1054         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1055         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1056         (aarch64_<sur>h<addsub><mode>): Likewise.
1057         (aarch64_<sur><addsub>hn<mode>): Likewise.
1058         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1059         (aarch64_pmul<mode>): Likewise.
1060         (aarch64_<su_optab><optab><mode>): Likewise.
1061         (aarch64_<sur>qadd<mode>): Likewise.
1062         (aarch64_sqmovun<mode>): Likewise.
1063         (aarch64_<sur>qmovn<mode>): Likewise.
1064         (aarch64_s<optab><mode>): Likewise.
1065         (aarch64_sq<r>dmulh<mode>): Likewise.
1066         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1067         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1068         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1069         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1070         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1071         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1072         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1073         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1074         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1075         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1076         (aarch64_sqdmull<mode>): Likewise.
1077         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1078         (aarch64_sqdmull_n<mode>): Likewise.
1079         (aarch64_sqdmull2<mode>_internal): Likewise.
1080         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1081         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1082         (aarch64_<sur>shl<mode>): Likewise.
1083         (aarch64_<sur>q<r>shl<mode>
1084         (aarch64_<sur>shll_n<mode>): Likewise.
1085         (aarch64_<sur>shll2_n<mode>): Likewise.
1086         (aarch64_<sur>shr_n<mode>): Likewise.
1087         (aarch64_<sur>sra_n<mode>): Likewise.
1088         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1089         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1090         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1091         (aarch64_cm<optab><mode>): Likewise.
1092         (aarch64_cm<optab>di): Likewise.
1093         (aarch64_cm<optab><mode>): Likewise.
1094         (aarch64_cm<optab>di): Likewise.
1095         (aarch64_cmtst<mode>): Likewise.
1096         (aarch64_cmtstdi): Likewise.
1097         (aarch64_cm<optab><mode>): Likewise.
1098         (*aarch64_fac<optab><mode>): Likewise.
1099         (aarch64_addp<mode>): Likewise.
1100         (aarch64_addpdi): Likewise.
1101         (sqrt<mode>2): Likewise.
1102         (vec_load_lanesoi<mode>): Likewise.
1103         (vec_store_lanesoi<mode>): Likewise.
1104         (vec_load_lanesci<mode>): Likewise.
1105         (vec_store_lanesci<mode>): Likewise.
1106         (vec_load_lanesxi<mode>): Likewise.
1107         (vec_store_lanesxi<mode>): Likewise.
1108         (*aarch64_mov<mode>): Likewise.
1109         (aarch64_ld2<mode>_dreg): Likewise.
1110         (aarch64_ld2<mode>_dreg): Likewise.
1111         (aarch64_ld3<mode>_dreg): Likewise.
1112         (aarch64_ld3<mode>_dreg): Likewise.
1113         (aarch64_ld4<mode>_dreg): Likewise.
1114         (aarch64_ld4<mode>_dreg): Likewise.
1115         (aarch64_tbl1<mode>): Likewise.
1116         (aarch64_tbl2v16qi): Likewise.
1117         (aarch64_combinev16qi): Likewise.
1118         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1119         (aarch64_st2<mode>_dreg): Likewise.
1120         (aarch64_st2<mode>_dreg): Likewise.
1121         (aarch64_st3<mode>_dreg): Likewise.
1122         (aarch64_st3<mode>_dreg): Likewise.
1123         (aarch64_st4<mode>_dreg): Likewise.
1124         (aarch64_st4<mode>_dreg): Likewise.
1125         (*aarch64_simd_ld1r<mode>): Likewise.
1126         (aarch64_frecpe<mode>): Likewise.
1127         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1128         (aarch64_frecps<mode>): Likewise.
1130 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1132         * config/arm/iterators.md (V_elem_ch): New.
1133         (q): Likewise.
1134         (VQH_type): Likewise.
1135         * config/arm/arm.md (is_neon_type): New.
1136         (conds): Use is_neon_type.
1137         (anddi3_insn): Update type attribute.
1138         (xordi3_insn): Likewise.
1139         (one_cmpldi2): Likewise.
1140         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1141         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1142         (*movmisalign<mode>_neon_store): Likewise.
1143         (*movmisalign<mode>_neon_load): Likewise.
1144         (vec_set<mode>_internal): Likewise.
1145         (vec_set<mode>_internal): Likewise.
1146         (vec_setv2di_internal): Likewise.
1147         (vec_extract<mode>): Likewise.
1148         (vec_extract<mode>): Likewise.
1149         (vec_extractv2di): Likewise.
1150         (*add<mode>3_neon): Likewise.
1151         (adddi3_neon): Likewise.
1152         (*sub<mode>3_neon): Likewise.
1153         (subdi3_neon): Likewise.
1154         (fma<VCVTF:mode>4): Likewise.
1155         (fma<VCVTF:mode>4_intrinsic): Likewise.
1156         (*fmsub<VCVTF:mode>4): Likewise.
1157         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1158         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1159         (ior<mode>3): Likewise.
1160         (and<mode>3): Likewise.
1161         (orn<mode>3_neon): Likewise.
1162         (orndi3_neon): Likewise.
1163         (bic<mode>3_neon): Likewise.
1164         (bicdi3_neon): Likewise.
1165         (xor<mode>3): Likewise.
1166         (one_cmpl<mode>2): Likewise.
1167         (abs<mode>2): Likewise.
1168         (neg<mode>2): Likewise.
1169         (negdi2_neon): Likewise.
1170         (*umin<mode>3_neon): Likewise.
1171         (*umax<mode>3_neon): Likewise.
1172         (*smin<mode>3_neon): Likewise.
1173         (*smax<mode>3_neon): Likewise.
1174         (vashl<mode>3): Likewise.
1175         (vashr<mode>3_imm): Likewise.
1176         (vlshr<mode>3_imm): Likewise.
1177         (ashl<mode>3_signed): Likewise.
1178         (ashl<mode>3_unsigned): Likewise.
1179         (neon_load_count): Likewise.
1180         (ashldi3_neon_noclobber): Likewise.
1181         (ashldi3_neon): Likewise.
1182         (signed_shift_di3_neon): Likewise.
1183         (unsigned_shift_di3_neon): Likewise.
1184         (ashrdi3_neon_imm_noclobber): Likewise.
1185         (lshrdi3_neon_imm_noclobber): Likewise.
1186         (<shift>di3_neon): Likewise.
1187         (widen_ssum<mode>3): Likewise.
1188         (widen_usum<mode>3): Likewise.
1189         (quad_halves_<code>v4si): Likewise.
1190         (quad_halves_<code>v4sf): Likewise.
1191         (quad_halves_<code>v8hi): Likewise.
1192         (quad_halves_<code>v16qi): Likewise.
1193         (reduc_splus_v2di): Likewise.
1194         (neon_vpadd_internal<mode>): Likewise.
1195         (neon_vpsmin<mode>): Likewise.
1196         (neon_vpsmax<mode>): Likewise.
1197         (neon_vpumin<mode>): Likewise.
1198         (neon_vpumax<mode>): Likewise.
1199         (*ss_add<mode>_neon): Likewise.
1200         (*us_add<mode>_neon): Likewise.
1201         (*ss_sub<mode>_neon): Likewise.
1202         (*us_sub<mode>_neon): Likewise.
1203         (neon_vadd<mode>_unspec): Likewise.
1204         (neon_vaddl<mode>): Likewise.
1205         (neon_vaddw<mode>): Likewise.
1206         (neon_vhadd<mode>): Likewise.
1207         (neon_vqadd<mode>): Likewise.
1208         (neon_vaddhn<mode>): Likewise.
1209         (neon_vmul<mode>): Likewise.
1210         (neon_vfms<VCVTF:mode>): Likewise.
1211         (neon_vmlal<mode>): Likewise.
1212         (neon_vmls<mode>): Likewise.
1213         (neon_vmlsl<mode>): Likewise.
1214         (neon_vqdmulh<mode>): Likewise.
1215         (neon_vqdmlal<mode>): Likewise.
1216         (neon_vqdmlsl<mode>): Likewise.
1217         (neon_vmull<mode>): Likewise.
1218         (neon_vqdmull<mode>): Likewise.
1219         (neon_vsub<mode>_unspec): Likewise.
1220         (neon_vsubl<mode>): Likewise.
1221         (neon_vsubw<mode>): Likewise.
1222         (neon_vqsub<mode>): Likewise.
1223         (neon_vhsub<mode>): Likewise.
1224         (neon_vsubhn<mode>): Likewise.
1225         (neon_vceq<mode>): Likewise.
1226         (neon_vcge<mode>): Likewise.
1227         (neon_vcgeu<mode>): Likewise.
1228         (neon_vcgt<mode>): Likewise.
1229         (neon_vcgtu<mode>): Likewise.
1230         (neon_vcle<mode>): Likewise.
1231         (neon_vclt<mode>): Likewise.
1232         (neon_vcage<mode>): Likewise.
1233         (neon_vcagt<mode>): Likewise.
1234         (neon_vtst<mode>): Likewise.
1235         (neon_vabd<mode>): Likewise.
1236         (neon_vabdl<mode>): Likewise.
1237         (neon_vaba<mode>): Likewise.
1238         (neon_vabal<mode>): Likewise.
1239         (neon_vmax<mode>): Likewise.
1240         (neon_vmin<mode>): Likewise.
1241         (neon_vpaddl<mode>): Likewise.
1242         (neon_vpadal<mode>): Likewise.
1243         (neon_vpmax<mode>): Likewise.
1244         (neon_vpmin<mode>): Likewise.
1245         (neon_vrecps<mode>): Likewise.
1246         (neon_vrsqrts<mode>): Likewise.
1247         (neon_vqabs<mode>): Likewise.
1248         (neon_vqneg<mode>): Likewise.
1249         (neon_vcls<mode>): Likewise.
1250         (clz<mode>2): Likewise.
1251         (popcount<mode>2): Likewise.
1252         (neon_vrecpe<mode>): Likewise.
1253         (neon_vrsqrte<mode>): Likewise.
1254         (neon_vget_lane<mode>_sext_internal): Likewise.
1255         (neon_vget_lane<mode>_zext_internal): Likewise.
1256         (neon_vdup_n<mode>): Likewise.
1257         (neon_vdup_n<mode>): Likewise.
1258         (neon_vdup_nv2di): Likewise.
1259         (neon_vdup_lane<mode>_interal): Likewise.
1260         (*neon_vswp<mode>): Likewise.
1261         (neon_vcombine<mode>): Likewise.
1262         (float<mode><V_cvtto>2): Likewise.
1263         (floatuns<mode><V_cvtto>2): Likewise.
1264         (fix_trunc<mode><V_cvtto>2): Likewise.
1265         (fixuns_trunc<mode><V_cvtto>2
1266         (neon_vcvt<mode>): Likewise.
1267         (neon_vcvt<mode>): Likewise.
1268         (neon_vcvtv4sfv4hf): Likewise.
1269         (neon_vcvtv4hfv4sf): Likewise.
1270         (neon_vcvt_n<mode>): Likewise.
1271         (neon_vcvt_n<mode>): Likewise.
1272         (neon_vmovn<mode>): Likewise.
1273         (neon_vqmovn<mode>): Likewise.
1274         (neon_vqmovun<mode>): Likewise.
1275         (neon_vmovl<mode>): Likewise.
1276         (neon_vmul_lane<mode>): Likewise.
1277         (neon_vmul_lane<mode>): Likewise.
1278         (neon_vmull_lane<mode>): Likewise.
1279         (neon_vqdmull_lane<mode>): Likewise.
1280         (neon_vqdmulh_lane<mode>): Likewise.
1281         (neon_vqdmulh_lane<mode>): Likewise.
1282         (neon_vmla_lane<mode>): Likewise.
1283         (neon_vmla_lane<mode>): Likewise.
1284         (neon_vmlal_lane<mode>): Likewise.
1285         (neon_vqdmlal_lane<mode>): Likewise.
1286         (neon_vmls_lane<mode>): Likewise.
1287         (neon_vmls_lane<mode>): Likewise.
1288         (neon_vmlsl_lane<mode>): Likewise.
1289         (neon_vqdmlsl_lane<mode>): Likewise.
1290         (neon_vext<mode>): Likewise.
1291         (neon_vrev64<mode>): Likewise.
1292         (neon_vrev32<mode>): Likewise.
1293         (neon_vrev16<mode>): Likewise.
1294         (neon_vbsl<mode>_internal): Likewise.
1295         (neon_vshl<mode>): Likewise.
1296         (neon_vqshl<mode>): Likewise.
1297         (neon_vshr_n<mode>): Likewise.
1298         (neon_vshrn_n<mode>): Likewise.
1299         (neon_vqshrn_n<mode>): Likewise.
1300         (neon_vqshrun_n<mode>): Likewise.
1301         (neon_vshl_n<mode>): Likewise.
1302         (neon_vqshl_n<mode>): Likewise.
1303         (neon_vqshlu_n<mode>): Likewise.
1304         (neon_vshll_n<mode>): Likewise.
1305         (neon_vsra_n<mode>): Likewise.
1306         (neon_vsri_n<mode>): Likewise.
1307         (neon_vsli_n<mode>): Likewise.
1308         (neon_vtbl1v8qi): Likewise.
1309         (neon_vtbl2v8qi): Likewise.
1310         (neon_vtbl3v8qi): Likewise.
1311         (neon_vtbl4v8qi): Likewise.
1312         (neon_vtbl1v16qi): Likewise.
1313         (neon_vtbl2v16qi): Likewise.
1314         (neon_vcombinev16qi): Likewise.
1315         (neon_vtbx1v8qi): Likewise.
1316         (neon_vtbx2v8qi): Likewise.
1317         (neon_vtbx3v8qi): Likewise.
1318         (neon_vtbx4v8qi): Likewise.
1319         (*neon_vtrn<mode>_insn): Likewise.
1320         (*neon_vzip<mode>_insn): Likewise.
1321         (*neon_vuzp<mode>_insn): Likewise.
1322         (neon_vld1<mode>): Likewise.
1323         (neon_vld1_lane<mode>): Likewise.
1324         (neon_vld1_lane<mode>): Likewise.
1325         (neon_vld1_dup<mode>): Likewise.
1326         (neon_vld1_dup<mode>): Likewise.
1327         (neon_vld1_dupv2di): Likewise.
1328         (neon_vst1<mode>): Likewise.
1329         (neon_vst1_lane<mode>): Likewise.
1330         (neon_vst1_lane<mode>): Likewise.
1331         (neon_vld2<mode>): Likewise.
1332         (neon_vld2<mode>): Likewise.
1333         (neon_vld2_lane<mode>): Likewise.
1334         (neon_vld2_lane<mode>): Likewise.
1335         (neon_vld2_dup<mode>): Likewise.
1336         (neon_vst2<mode>): Likewise.
1337         (neon_vst2<mode>): Likewise.
1338         (neon_vst2_lane<mode>): Likewise.
1339         (neon_vst2_lane<mode>): Likewise.
1340         (neon_vld3<mode>): Likewise.
1341         (neon_vld3qa<mode>): Likewise.
1342         (neon_vld3qb<mode>): Likewise.
1343         (neon_vld3_lane<mode>): Likewise.
1344         (neon_vld3_lane<mode>): Likewise.
1345         (neon_vld3_dup<mode>): Likewise.
1346         (neon_vst3<mode>): Likewise.
1347         (neon_vst3qa<mode>): Likewise.
1348         (neon_vst3qb<mode>): Likewise.
1349         (neon_vst3_lane<mode>): Likewise.
1350         (neon_vst3_lane<mode>): Likewise.
1351         (neon_vld4<mode>): Likewise.
1352         (neon_vld4qa<mode>): Likewise.
1353         (neon_vld4qb<mode>): Likewise.
1354         (neon_vld4_lane<mode>): Likewise.
1355         (neon_vld4_lane<mode>): Likewise.
1356         (neon_vld4_dup<mode>): Likewise.
1357         (neon_vst4<mode>): Likewise.
1358         (neon_vst4qa<mode>): Likewise.
1359         (neon_vst4qb<mode>): Likewise.
1360         (neon_vst4_lane<mode>): Likewise.
1361         (neon_vst4_lane<mode>): Likewise.
1362         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1363         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1364         (neon_vec_<US>mult_lo_<mode>): Likewise.
1365         (neon_vec_<US>mult_hi_<mode>): Likewise.
1366         (neon_vec_<US>shiftl_<mode>): Likewise.
1367         (neon_unpack<US>_<mode>): Likewise.
1368         (neon_vec_<US>mult_<mode>): Likewise.
1369         (vec_pack_trunc_<mode>): Likewise.
1370         (neon_vec_pack_trunc_<mode>): Likewise.
1371         (neon_vabd<mode>_2): Likewise.
1372         (neon_vabd<mode>_3): Likewise.
1374 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1376         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1377         (load_pair): Update type attribute.
1378         (store_pair): Update type attribute.
1379         * config/aarch64/iterators.md (q): New.
1381 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1383         * config/arm/types.md: Add new types for Neon insns.
1385 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1386             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1387             Sergey Lega  <sergey.s.lega@intel.com>
1388             Anna Tikhonova  <anna.tikhonova@intel.com>
1389             Ilya Tocar  <ilya.tocar@intel.com>
1390             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1391             Ilya Verbin  <ilya.verbin@intel.com>
1392             Kirill Yukhin  <kirill.yukhin@intel.com>
1393             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1395         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1396         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1397         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1398         (rcp14<mode>): New.
1399         (srcp14<mode>): Ditto.
1400         (rsqrt14<mode>): Ditto.
1401         (rsqrt14<mode>): Ditto.
1402         (avx512f_vmscalef<mode>): Ditto.
1403         (avx512f_scalef<mode>): Ditto.
1404         (avx512f_getexp<mode>): Ditto.
1405         (avx512f_sgetexp<mode>): Ditto.
1406         (avx512f_fixupimm<mode>): Ditto.
1407         (avx512f_sfixupimm<mode>): Ditto.
1408         (avx512f_rndscale<mode>): Ditto.
1409         (*avx512er_exp2<mode>): Ditto.
1410         (*avx512er_rcp28<mode>): Ditto.
1411         (avx512er_rsqrt28<mode>): Ditto.
1412         (avx512f_getmant<mode>): Ditto.
1413         (avx512f_getmant<mode>): Ditto.
1414         (avx512f_rndscale<mode>): Fix formatting.
1416 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1418         * ipa-utils.h (ipa_edge_within_scc): Declare.
1419         * ipa-cp.c (edge_within_scc): Moved...
1420         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1422 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1423             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1424             Sergey Lega  <sergey.s.lega@intel.com>
1425             Anna Tikhonova  <anna.tikhonova@intel.com>
1426             Ilya Tocar  <ilya.tocar@intel.com>
1427             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1428             Ilya Verbin  <ilya.verbin@intel.com>
1429             Kirill Yukhin  <kirill.yukhin@intel.com>
1430             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1432         * config/i386/predicates.md (const_8_to_15_operand): New.
1433         (const_16_to_31_operand): Ditto.
1434         * config/i386/sse.md (V8FI): New.
1435         (V16FI): Ditto.
1436         (reduc_splus_v8df): Ditto.
1437         (reduc_splus_v16sf): Ditto.
1438         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1439         (vec_extract_hi_<mode>): Ditto.
1440         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1441         (vec_set_lo_<mode>): Ditto.
1442         (vec_set_hi_<mode>): Ditto.
1443         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1444         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1445         (avx512f_pshufd_1): Ditto.
1446         (avx512f_broadcast<mode>): Ditto.
1447         (avx512f_broadcast<mode>): Ditto.
1448         (define_split): Split vec_extract_lo into move.
1449         (ssequartermode): Ditto.
1450         (ssedoublemode): Extened with wider modes.
1451         (vec_extract_lo_<mode>): Ditto.
1453 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1454             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1455             Sergey Lega  <sergey.s.lega@intel.com>
1456             Anna Tikhonova  <anna.tikhonova@intel.com>
1457             Ilya Tocar  <ilya.tocar@intel.com>
1458             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1459             Ilya Verbin  <ilya.verbin@intel.com>
1460             Kirill Yukhin  <kirill.yukhin@intel.com>
1461             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1463         * config/i386/predicates.md (register_or_constm1_operand): New.
1464         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1465         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1466         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1467         UNSPEC_SCATTER_PREFETCH
1468         (VI48_512): New.
1469         (avx512f_ucmp<mode>3): Ditto.
1470         (avx512f_vternlog<mode>): Ditto.
1471         (avx512f_align<mode>): Ditto.
1472         (<shift_insn><mode>3): Ditto.
1473         (avx512f_<rotate>v<mode>): Ditto.
1474         (avx512f_<rotate><mode>): Ditto.
1475         (avx512f_eq<mode>3): Ditto.
1476         (avx512f_eq<mode>3_1): Ditto.
1477         (avx512f_gt<mode>3): Ditto.
1478         (avx512f_testm<mode>3): Ditto.
1479         (avx512f_testnm<mode>3): Ditto.
1480         (avx512pf_gatherpf<mode>): Ditto.
1481         (*avx512pf_gatherpf<mode>_mask): Ditto.
1482         (*avx512pf_gatherpf<mode>): Ditto.
1483         (avx512pf_scatterpf<mode>): Ditto.
1484         (*avx512pf_scatterpf<mode>_mask): Ditto.
1485         (*avx512pf_scatterpf<mode>): Ditto.
1486         (avx512f_vec_dup_gpr<mode>): Ditto.
1487         (clz<mode>2): Ditto.
1488         (conflict<mode>): Ditto.
1489         (REDUC_SMINMAX_MODE): Extened with wider modes.
1490         (reduc_<code>_<mode>): Ditto.
1491         (vlshr<mode>3): Ditto.
1492         (vashl<mode>3): Ditto.
1494 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1495             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1496             Sergey Lega  <sergey.s.lega@intel.com>
1497             Anna Tikhonova  <anna.tikhonova@intel.com>
1498             Ilya Tocar  <ilya.tocar@intel.com>
1499             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1500             Ilya Verbin  <ilya.verbin@intel.com>
1501             Kirill Yukhin  <kirill.yukhin@intel.com>
1502             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1504         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1505         UNSPEC_SCATTER.
1506         (VI48F_512): New.
1507         (avx512fmaskmode): Ditto.
1508         (bcstscalarsuff): Ditto.
1509         (avx512f_blendm<mode>): Ditto.
1510         (cmp_imm_predicate): Ditto.
1511         (avx512f_cmp<mode>3): Ditto.
1512         (avx512f_vec_dup<mode>): Ditto.
1513         (avx512f_vec_dup_mem<mode>): Ditto.
1514         (avx512f_vpermi2var<mode>3): Ditto.
1515         (avx512f_vpermt2var<mode>3): Ditto.
1516         (vec_init<mode>): Ditto.
1517         (avx512f_gathersi<mode>): Ditto.
1518         (*avx512f_gathersi<mode>): Ditto.
1519         (*avx512f_gathersi<mode>_2): Ditto.
1520         (avx512f_gatherdi<mode>): Ditto.
1521         (*avx512f_gatherdi<mode>): Ditto.
1522         (*avx512f_gatherdi<mode>_2): Ditto.
1523         (avx512f_scattersi<mode>): Ditto.
1524         (*avx512f_scattersi<mode>): Ditto.
1525         (avx512f_scatterdi<mode>): Ditto.
1526         (*avx512f_scatterdi<mode>): Ditto.
1527         (sseintprefix): Extened with wider modes.
1528         (VEC_GATHER_IDXSI): Ditto.
1529         (VEC_GATHER_IDXDI): Ditto.
1530         (VEC_GATHER_SRCDI): Ditto.
1532 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1533             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1535         * config/arm/t-aprofile: New file.
1536         * config.gcc: Handle --with-multilib-list option.
1538 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1540         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1541         the input for r2.
1543 2013-10-15  Richard Biener  <rguenther@suse.de>
1545         * tree-loop-distribution.c (build_empty_rdg): Inline into
1546         single user.
1547         (rdg_flag_vertex): Inline into single user.
1548         (rdg_flag_vertex_and_dependent): Likewise.
1549         (build_rdg_partition_for_vertex): Remove processed bitmap.
1550         (rdg_build_partitions): Simplify.
1552 2013-10-15  Richard Biener  <rguenther@suse.de>
1554         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1555         Restructure forwarding through conversions and copies to
1556         avoid performing copy-propagation the wrong way.  Adjust
1557         recursion invocations.
1558         (forward_propagate_addr_expr): Add argument stating if we
1559         are recursing from a single-use.
1560         (ssa_forward_propagate_and_combine): Adjust.
1562 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1564         * dumpfile.h (gcc::dump_manager): New class, to hold state
1565         relating to dumpfile management.
1566         (get_dump_file_name): Remove in favor of method of dump_manager.
1567         (dump_initialized_p): Likewise.
1568         (dump_start): Likewise.
1569         (dump_finish): Likewise.
1570         (dump_switch_p): Likewise.
1571         (dump_register): Likewise.
1572         (get_dump_file_info): Likewise.
1573         * context.c (gcc::context::context): Construct the dump_manager
1574         instance.
1575         * context.h (gcc::context::get_dumps): New.
1576         (gcc::context::m_dumps): New.
1577         * coverage.c (coverage_init): Port to dump_manager API.
1578         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1579         (extra_dump_files_in_use): Likewise.
1580         (extra_dump_files_alloced): Likewise.
1581         (gcc::dump_manager::dump_manager): New.
1582         (dump_register): Convert to...
1583         (gcc::dump_manager::dump_register): ...method, replacing
1584         function-static next_dump with m_next_dump field.
1585         (get_dump_file_info): Convert to...
1586         (gcc::dump_manager::get_dump_file_info): ...method.
1587         (get_dump_file_name): Convert to...
1588         (gcc::dump_manager::get_dump_file_name): ...method.
1589         (dump_start): Convert to...
1590         (gcc::dump_manager::dump_start): ...method.
1591         (dump_finish): Convert to...
1592         (gcc::dump_manager::dump_finish): ...method.
1593         (dump_begin): Replace body with...
1594         (gcc::dump_manager::dump_begin): ...new method.
1595         (dump_phase_enabled_p): Convert to...
1596         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1597         (dump_phase_enabled_p): Convert to...
1598         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1599         (dump_initialized_p):  Convert to...
1600         (gcc::dump_manager::dump_initialized_p): ...method.
1601         (dump_flag_name): Replace body with...
1602         (gcc::dump_manager::dump_flag_name): ...new method.
1603         (dump_enable_all): Convert to...
1604         (gcc::dump_manager::dump_enable_all): ...new method.
1605         (opt_info_enable_passes): Convert to...
1606         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1607         (dump_switch_p_1): Convert to...
1608         (gcc::dump_manager::dump_switch_p_1): ...new method.
1609         (dump_switch_p):  Convert to...
1610         (gcc::dump_manager::dump_switch_p): ...new method.
1611         (opt_info_switch_p): Port to dump_manager API.
1612         (enable_rtl_dump_file): Likewise.
1613         * opts-global.c (handle_common_deferred_options): Port to new
1614         dump_manager API.
1615         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1616         (pass_manager::register_one_dump_file): Likewise.
1617         (pass_manager::register_pass): Likewise.
1618         (pass_init_dump_file): Likewise.
1619         (pass_fini_dump_file): Likewise.
1620         * statistics.c (statistics_early_init): Likewise.
1622 2013-10-14  Richard Biener  <rguenther@suse.de>
1624         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1625         iterative_hash_canonical_type, gimple_canonical_type_hash,
1626         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1627         gimple_register_canonical_type, print_gimple_types_stats,
1628         free_gimple_type_tables): Move to lto/lto.c
1629         (gt-gimple.h): Do not include.
1630         * gimple.h (gimple_register_canonical_type,
1631         print_gimple_types_stats, free_gimple_type_tables): Remove.
1632         * Makefile.in (GTFILES): Remove gimple.c.
1634 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1636         PR target/58716
1637         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1638         scanning for msp430x targets.
1640 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1642         PR bootstrap/58509
1643         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1644         (registers_ok_for_ldd_peep): Move around.
1645         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1646         * config/sparc/sparc.md (widening peepholes): Use it.
1648 2013-10-14  Richard Biener  <rguenther@suse.de>
1650         PR middle-end/58712
1651         PR middle-end/55358
1652         * gimple.c (iterative_hash_canonical_type): Make sure to
1653         record the hash into the correct hashtable slot.
1655 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1657         PR rtl-optimization/58662
1658         * combine.c (try_combine): Take into account death nodes on I2 when
1659         splitting a PARALLEL of two independent SETs.  Fix dump message.
1661 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1663         PR target/51244
1664         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1665         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1666         * config/sh/t-sh (sh_treg_combine.o): New entry.
1667         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1668         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1669         (register_sh_passes): New function.  Register sh_treg_combine pass.
1670         (sh_option_override): Invoke it.
1671         (sh_canonicalize_comparison): Handle op0_preserve_value.
1672         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1673         opportunities.  Canonicalize branch condition.
1674         (nott): Allow only if pseudos can be created for non-SH2A.
1676 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1678         PR target/58690
1679         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1680         (ix86_expand_movmem): Replace copy_addr_to_reg with
1681         ix86_copy_addr_to_reg.
1682         (ix86_expand_setmem): Likewise.
1684 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1686         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1687         provided by ix86_fp_compare_mode instead of CCFPUmode.
1689 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1691         * config/aarch64/arm_neon.h
1692         (vtbx<1,3>_<psu>8): Fix register constriants.
1694 2013-10-11  Jeff Law  <law@redhat.com>
1696         PR tree-optimization/58640
1697         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1698         threading paths that cross over two loop entry points.
1700 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1702         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1703         handle vector float as well.
1704         (*vsx_le_perm_load_v4si): Likewise.
1705         (*vsx_le_perm_store_v2di): Likewise.
1706         (*vsx_le_perm_store_v4si): Likewise.
1708 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1710         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1711         directly to circumvent subtract from splat{31} workaround.
1712         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1713         prototype.
1714         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1715         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1716         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1717         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1718         separate big and little endian logic.
1719         (*altivec_vperm_<mode>_internal): New define_insn.
1720         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1721         separate big and little endian logic.
1722         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1723         (vec_permv16qi): Add little endian logic.
1725 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1727         * doc/extend.texi (returns_nonnull): Remove arguments.
1729 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1730             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1731             Sergey Lega  <sergey.s.lega@intel.com>
1732             Anna Tikhonova  <anna.tikhonova@intel.com>
1733             Ilya Tocar  <ilya.tocar@intel.com>
1734             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1735             Ilya Verbin  <ilya.verbin@intel.com>
1736             Kirill Yukhin  <kirill.yukhin@intel.com>
1737             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1739         * config/i386/sse.md (VI48F_256_512): New.
1740         (avx2_permvar<mode>): Change to ...
1741         (<avx2_avx512f>_permvar<mode>): This.
1743 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1744             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1745             Sergey Lega  <sergey.s.lega@intel.com>
1746             Anna Tikhonova  <anna.tikhonova@intel.com>
1747             Ilya Tocar  <ilya.tocar@intel.com>
1748             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1749             Ilya Verbin  <ilya.verbin@intel.com>
1750             Kirill Yukhin  <kirill.yukhin@intel.com>
1751             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1753         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1754         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1755         * config/i386/sse.md (VI4_AVX): New.
1756         (sf2simodelower): Ditto.
1757         (sse2_cvtps2dq): Change to ...
1758         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1760 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1761             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1762             Sergey Lega  <sergey.s.lega@intel.com>
1763             Anna Tikhonova  <anna.tikhonova@intel.com>
1764             Ilya Tocar  <ilya.tocar@intel.com>
1765             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1766             Ilya Verbin  <ilya.verbin@intel.com>
1767             Kirill Yukhin  <kirill.yukhin@intel.com>
1768             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1770         * config/i386/sse.md (V_512): New.
1771         (VI_512): Ditto.
1772         (vcond<V_512:mode><VF_512:mode>): Ditto.
1773         (vcond<V_512:mode><VI_512:mode>): Ditto.
1774         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1776 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1777             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1778             Sergey Lega  <sergey.s.lega@intel.com>
1779             Anna Tikhonova  <anna.tikhonova@intel.com>
1780             Ilya Tocar  <ilya.tocar@intel.com>
1781             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1782             Ilya Verbin  <ilya.verbin@intel.com>
1783             Kirill Yukhin  <kirill.yukhin@intel.com>
1784             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1786         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1787         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1788         (FMAMODE): Ditto.
1789         (fma<mode>4): Removed condition.
1790         (fms<mode>4): Ditto.
1791         (fnma<mode>4): Ditto.
1792         (fnms<mode>4): Ditto.
1793         (fma4i_fmadd_<mode>): Ditto.
1794         (*fma_fmadd_<mode>): Ditto.
1795         (*fma_fmsub_<mode>): Ditto.
1796         (*fma_fnmadd_<mode>): Ditto.
1797         (*fma_fnmsub_<mode>): Ditto.
1798         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1799         (*fma_fmaddsub_<mode>): Ditto.
1800         (*fma_fmsubadd_<mode>): Ditto.
1801         (*fmai_fmadd_<mode>): Ditto.
1802         (*fmai_fmsub_<mode>): Ditto.
1803         (*fmai_fnmadd_<mode>): Ditto.
1804         (*fmai_fnmsub_<mode>): Ditto.
1806 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1807             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1808             Sergey Lega  <sergey.s.lega@intel.com>
1809             Anna Tikhonova  <anna.tikhonova@intel.com>
1810             Ilya Tocar  <ilya.tocar@intel.com>
1811             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1812             Ilya Verbin  <ilya.verbin@intel.com>
1813             Kirill Yukhin  <kirill.yukhin@intel.com>
1814             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1816         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1817         (VI124_256): Changed to ...
1818         (VI124_256_48_512): This.
1819         (ssepackmode): Extended with wider modes.
1820         (<code><mode>3): Changed iterator.
1821         (*avx2_<code><mode>3): Ditto.
1822         (vec_pack_trunc_<mode>): Ditto.
1824 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1825             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1826             Sergey Lega  <sergey.s.lega@intel.com>
1827             Anna Tikhonova  <anna.tikhonova@intel.com>
1828             Ilya Tocar  <ilya.tocar@intel.com>
1829             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1830             Ilya Verbin  <ilya.verbin@intel.com>
1831             Kirill Yukhin  <kirill.yukhin@intel.com>
1832             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1834         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1835         (VI8F_256_512): Ditto.
1836         (abs<mode>2): Changed iterator.
1837         (avx2_perm<mode>): Changed to ...
1838         (<avx2_avx512f>_perm<mode>): This.
1839         (avx2_perm<mode>_1): Changed to ...
1840         (<avx2_avx512f>_perm<mode>_1): This.
1842 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1843             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1844             Sergey Lega  <sergey.s.lega@intel.com>
1845             Anna Tikhonova  <anna.tikhonova@intel.com>
1846             Ilya Tocar  <ilya.tocar@intel.com>
1847             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1848             Ilya Verbin  <ilya.verbin@intel.com>
1849             Kirill Yukhin  <kirill.yukhin@intel.com>
1850             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1852         * config/i386/sse.md (VI48_AVX512F): New.
1853         (VI48_AVX2): Changed to ...
1854         (VI48_AVX2_48_AVX512F): This.
1855         (avx2_ashrv<mode>): Changed to ...
1856         (<avx2_avx512f>_ashrv<mode>): This.
1857         (avx2_<shift_insn>v<mode>): Changed to ...
1858         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1860 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1861             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1862             Sergey Lega  <sergey.s.lega@intel.com>
1863             Anna Tikhonova  <anna.tikhonova@intel.com>
1864             Ilya Tocar  <ilya.tocar@intel.com>
1865             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1866             Ilya Verbin  <ilya.verbin@intel.com>
1867             Kirill Yukhin  <kirill.yukhin@intel.com>
1868             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1870         * config/i386/sse.md (VI4_AVX512F): New.
1871         (VI8_AVX2_AVX512F): Ditto.
1872         (mul<mode>3): Extended with wider modes.
1873         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1874         (mul<mode>3): Ditto.
1875         (vec_widen_<s>mult_odd_<mode>): Ditto.
1877 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1878             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1879             Sergey Lega  <sergey.s.lega@intel.com>
1880             Anna Tikhonova  <anna.tikhonova@intel.com>
1881             Ilya Tocar  <ilya.tocar@intel.com>
1882             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1883             Ilya Verbin  <ilya.verbin@intel.com>
1884             Kirill Yukhin  <kirill.yukhin@intel.com>
1885             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1887         * config/i386/sse.md (VI2_AVX512F): New.
1888         (VI124_AVX512F): Ditto.
1889         (sseunpackmode): Extended with wider modes.
1890         (sseunpackfltmode): Ditto.
1891         (vec_unpacks_float_hi_<mode>): Ditto.
1892         (vec_unpacks_float_lo_<mode>): Ditto.
1893         (vec_unpacku_float_hi_<mode>): Ditto.
1894         (vec_unpacku_float_lo_<mode>): Ditto.
1895         (vec_unpacks_lo_<mode>): Ditto.
1896         (vec_unpacks_hi_<mode>): Ditto.
1897         (vec_unpacku_lo_<mode>): Ditto.
1898         (vec_unpacku_hi_<mode>): Ditto.
1900 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1901             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1902             Sergey Lega  <sergey.s.lega@intel.com>
1903             Anna Tikhonova  <anna.tikhonova@intel.com>
1904             Ilya Tocar  <ilya.tocar@intel.com>
1905             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1906             Ilya Verbin  <ilya.verbin@intel.com>
1907             Kirill Yukhin  <kirill.yukhin@intel.com>
1908             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1910         * config/i386/i386.md (multdiv): New.
1911         (multdiv_mnemonic): Ditto.
1912         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1913         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1914         (<sse>_vmdiv<mode>3): Removed.
1916 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1917             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1918             Sergey Lega  <sergey.s.lega@intel.com>
1919             Anna Tikhonova  <anna.tikhonova@intel.com>
1920             Ilya Tocar  <ilya.tocar@intel.com>
1921             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1922             Ilya Verbin  <ilya.verbin@intel.com>
1923             Kirill Yukhin  <kirill.yukhin@intel.com>
1924             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1926         * config/i386/sse.md (V): Extended with wider modes.
1927         (VF2): Ditto.
1928         (ssehalfvecmode): Ditto.
1929         (i128): Ditto.
1930         (ssepackfltmode): Ditto.
1931         (avx_vec_concat<mode>): Ditto.
1932         (V_256_512): New iterator.
1933         (VF2_512_256): Ditto.
1934         (si2dfmode): New attribute.
1935         (si2dfmodelower): Ditto.
1936         (sf2dfmode): Ditto.
1937         (concat_tg_mode): Ditto.
1938         (floatv4siv4df2): Changed to ...
1939         (float<si2dfmodelower><mode>2): This.
1940         (avx_cvtps2pd256): Changed to ...
1941         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1942         (vec_pack_trunc_v4df): Changed to ...
1943         (vec_pack_trunc_<mode>): This.
1944         (avx_vpermil<mode>): Changed to ...
1945         (<sse2_avx_avx512f>_vpermil<mode>): This.
1946         (<fixsuffix>fix_truncv8dfv8si2): New.
1947         (vec_pack_sfix_trunc_v8df): Ditto.
1948         (avx512f_rndscale<mode>): Ditto.
1949         (avx512f_roundpd512): Ditto.
1950         (vec_pack_ufix_trunc_<mode>): Updated iterator.
1952 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1953             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1954             Sergey Lega  <sergey.s.lega@intel.com>
1955             Anna Tikhonova  <anna.tikhonova@intel.com>
1956             Ilya Tocar  <ilya.tocar@intel.com>
1957             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1958             Ilya Verbin  <ilya.verbin@intel.com>
1959             Kirill Yukhin  <kirill.yukhin@intel.com>
1960             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1962         * config/i386/i386.md (any_fix): New iterator.
1963         (fixsuffix): New attribute.
1964         * config/i386/sse.md (VF1): Extened with wider modes.
1965         (VI): Ditto.
1966         (VI_AVX2): Ditto.
1967         (VI8): Ditto.
1968         (sseintvecmodelower): Ditto.
1969         (ssescalarmode): Ditto.
1970         (ssescalarnum): Ditto.
1971         (VF1_128_256): New.
1972         (ssexmmmode): Ditto.
1973         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
1974         (<sse>_rcp<mode>2): Change iterator.
1975         (rsqrt<mode>2): Ditto.
1976         (<sse>_rsqrt<mode>2): Ditto.
1977         (avx2_vec_dup<mode>): Ditto.
1978         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1979         (round<mode>2_sfix): Ditto.
1980         (avx2_pbroadcast<mode>): Ditto.
1981         (*andnot<mode>3): Handle XI mode.
1982         (*<code><mode>3): Ditto.
1983         (AVXTOSSEMODE): Removed.
1984         (avx_vpermil<mode>): Changed to ...
1985         (<sse2_avx_avx512f>_vpermil<mode>): This.
1987 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1988             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1989             Sergey Lega  <sergey.s.lega@intel.com>
1990             Anna Tikhonova  <anna.tikhonova@intel.com>
1991             Ilya Tocar  <ilya.tocar@intel.com>
1992             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1993             Ilya Verbin  <ilya.verbin@intel.com>
1994             Kirill Yukhin  <kirill.yukhin@intel.com>
1995             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1997         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
1998         (<sse>_comi): Ditto.
1999         (<sse>_ucomi): Ditto.
2000         (sse_cvtss2siq_2): Ditto.
2001         (sse2_cvtsd2si): Ditto.
2002         (sse2_cvtsd2siq): Ditto.
2003         (sse2_cvttsd2si): Ditto.
2004         (sse2_cvttsd2siq): Ditto.
2005         (<shift_insn><mode>3): Ditto.
2006         (sse2_cvtsi2sdq): Update constraint and prefix.
2007         (sse_cvtsi2ss): Update prefix.
2008         (sse_cvtsi2ssq): Ditto.
2010 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2012         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2013         ignore internal calls.
2015 2013-10-11  Richard Biener  <rguenther@suse.de>
2017         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2018         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2019         TREE_OVERFLOW is set.
2021 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2023         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2025         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2026         * gimple.h: Likewise.
2028         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2030         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2031         macros.
2033         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2035         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2036         * configure: Regenerate.
2038 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2040         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2041         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2042         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2043         similarly pp_right_paren.
2044         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2045         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2046         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2047         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2048         clauses.
2049         (omp_declare_simd_clauses_equal,
2050         omp_remove_redundant_declare_simd_attrs): New functions.
2051         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2052         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2053         * tree.h (OMP_LOOP_CHECK): Define.
2054         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2055         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2056         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2057         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2058         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2059         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2060         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2061         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2062         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2063         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2064         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2065         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2066         (omp_remove_redundant_declare_simd_attrs): New prototype.
2067         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2068         GIMPLE_OMP_TEAMS): New codes.
2069         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2070         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2071         fields.
2072         (target_nesting_level): New variable.
2073         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2074         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2075         for collapse > 1 static schedule unless ordered.
2076         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2077         (determine_parallel_type): Adjust get_ws_args_for caller.
2078         (install_var_field): Handle mask & 4 for double indirection.
2079         (scan_sharing_clauses): Ignore shared clause on teams construct.
2080         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2081         (create_omp_child_function): If inside target or declare target
2082         constructs, set "omp declare target" attribute on the child function.
2083         (find_combined_for): New function.
2084         (scan_omp_parallel): Handle combined loops.
2085         (scan_omp_target, scan_omp_teams): New functions.
2086         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2087         restrictions and set ctx->cancellable for cancellable constructs.
2088         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2089         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2090         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2091         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2092         (omp_clause_aligned_alignment): New function.
2093         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2094         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2095         on teams constructs.  Handle user defined reductions and new
2096         OpenMP 4.0 clauses.
2097         (lower_reduction_clauses): Don't set placeholder to address of ref
2098         if it has already the right type.
2099         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2100         (expand_parallel_call): Use the new non-_start suffixed builtins,
2101         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2102         and GOMP_parallel_end after the call.
2103         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2104         (expand_omp_for_init_counts): Handle combined loops.
2105         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2106         loops.
2107         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2108         end of cancellable loops.
2109         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2110         Likewise.  Handle collapse > 1 loops.
2111         (expand_omp_simd): Handle combined loops.
2112         (expand_omp_for): Add inner_stmt argument, adjust callers of
2113         expand_omp_for* functions, use expand_omp_for_static*chunk even
2114         for collapse > 1 unless ordered.
2115         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2116         of cancellable sections.
2117         (expand_omp_single): Remove need_barrier variable, just rely on
2118         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2119         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2120         (expand_omp_atomic_load, expand_omp_atomic_store,
2121         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2122         (expand_omp_target): New function.
2123         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2124         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2125         (build_omp_regions_1): Immediately close region for
2126         GF_OMP_TARGET_KIND_UPDATE.
2127         (maybe_add_implicit_barrier_cancel): New function.
2128         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2129         cancellation.
2130         (lower_omp_single): Likewise.  Add clobber after the barrier.
2131         (lower_omp_taskgroup): New function.
2132         (lower_omp_for): Handle combined loops.  Adjust
2133         lower_rec_input_clauses caller.  Handle cancellation.
2134         (lower_depend_clauses): New function.
2135         (lower_omp_taskreg): Lower depend clauses.  Adjust
2136         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2137         cancellation.
2138         (lower_omp_target, lower_omp_teams): New functions.
2139         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2140         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2141         and GOMP_cancellation_point calls.
2142         (lower_omp): Fold stmts inside of target region.
2143         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2144         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2145         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2146         (BT_FN_VOID_OMPFN_PTR_UINT,
2147         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2148         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2149         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2150         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2151         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2152         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2153         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2154         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2155         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2156         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2157         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2158         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2159         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2160         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2161         BUILT_IN_GOMP_PARALLEL_END.
2162         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2163         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2164         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2165         GF_OMP_FOR_KIND_DISTRIBUTE.
2166         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2167         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2168         (dump_gimple_omp_return): Print lhs if it has any.
2169         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2170         gimple_omp_atomic_seq_cst_p.
2171         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2172         and GIMPLE_OMP_TEAMS.
2173         * langhooks.c (lhd_omp_mappable_type): New function.
2174         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2175         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2176         hook.
2177         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2178         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2179         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2180         (struct gimplify_omp_ctx): Add combined_loop field.
2181         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2182         on stmts inside of target region.
2183         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2184         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2185         ORT_TARGET and ORT_TARGET_DATA.
2186         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2187         Handle GOVD_MAP.
2188         (omp_notice_threadprivate_variable): Complain about threadprivate
2189         variables in target region.
2190         (omp_notice_variable): Complain about vars with non-mappable type
2191         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2192         (omp_check_private): Ignore ORT_TARGET* regions.
2193         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2194         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2195         (find_combined_omp_for): New function.
2196         (gimplify_omp_for): Handle gimplification of combined loops.
2197         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2198         OMP_TEAMS.
2199         (gimplify_omp_target_update): New function.
2200         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2201         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2202         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2203         (gimplify_body): If fndecl has "omp declare target" attribute, add
2204         implicit ORT_TARGET context around it.
2205         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2206         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2207         * tree-nested.c (convert_nonlocal_reference_stmt,
2208         convert_local_reference_stmt, convert_gimple_call): Handle
2209         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2210         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2211         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2212         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2213         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2214         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2215         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2216         BUILT_IN_GOMP_LOOP_END_CANCEL,
2217         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2218         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2219         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2220         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2221         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2222         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2223         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2224         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2225         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2226         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2227         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2228         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2229         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2230         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2231         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2232         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2233         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2234         gimple_build_omp_teams): New functions.
2235         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2236         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2237         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2238         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2239         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2240         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2241         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2242         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2243         GF_OMP_ATOMIC_SEQ_CST): New.
2244         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2245         gimple_build_omp_teams): New prototypes.
2246         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2247         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2248         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2249         GIMPLE_OMP_SINGLE as end of range.
2250         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2251         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2252         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2253         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2254         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2255         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2256         gimple_omp_target_kind, gimple_omp_target_set_kind,
2257         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2258         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2259         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2260         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2261         gimple_omp_teams_set_clauses): New inlines.
2262         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2263         and GIMPLE_OMP_TASKGROUP.
2264         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2265         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2266         enum omp_clause_proc_bind_kind): New.
2267         (union omp_clause_subcode): Add depend_kind, map_kind and
2268         proc_bind_kind fields.
2269         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2270         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2271         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2272         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2273         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2275 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2277         * predict.c (tree_estimate_probability): Add new parameter
2278         for estimate_bb_frequencies.
2279         (estimate_bb_frequencies): Add new parameter to force estimation.
2280         (rebuild_frequencies): When max frequency in function is small,
2281         recompute counts from frequencies.
2282         * predict.h (estimate_bb_frequencies): New parameter.
2284 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2286         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2287         optimizations are disabled.
2289 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2291         * coverage.c (coverage_finish): Fix leak of da_file_name.
2293 2013-10-10  Jan Hubicka  <jh@suse.cz>
2295         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2296         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2297         Bobcat and generic.
2299 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2301         PR middle-end/58670
2302         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2303         if any labels are in FALLTHRU_BB, use a special label emitted
2304         immediately after the asm goto insn rather than label_rtx
2305         of the LABEL_DECL.
2306         (expand_asm_stmt): Adjust caller.
2307         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2308         edge if the last insn in predecessor is a jump with single successor,
2309         but it isn't simplejump_p.
2311 2013-10-10  Richard Biener  <rguenther@suse.de>
2313         PR tree-optimization/58656
2314         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2316 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2318         * gimplify.c: Include expr.h and tm_p.h for targets with special
2319         va-arg padding requirements.
2321 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2323         * tree-flow.h: Move some prototypes to gimple.h.
2324         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2325         * gimple.h: Relocate some prototypes from tree-flow.h
2326         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2327         Move to gimplify.c.
2328         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2329         (build_va_arg_indirect_ref): Relocate and make static.
2330         (std_gimplify_va_arg_expr): Relocate here.
2331         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2332         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2334 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2336         * doc/md.texi: Document the mnemonic attribute.
2338 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2340         PR target/57377
2341         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2342         (set_attr_alternative x ...) when searching for user defined
2343         mnemonic attribute.
2345 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2347         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2349 2013-10-09  Easwaran Raman <eraman@google.com>
2351         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2352         * cfgexpand.c (defer_stack_allocation): ...use here
2353         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2355 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2357         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2358         (operand_equal_for_value_replacement): New function, extracted from
2359         value_replacement and enhanced to catch more cases.
2360         (value_replacement): Use operand_equal_for_value_replacement.
2362 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2364         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2365         get_max_loop_iterations.
2367 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2369         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2370         Place comment above function.
2372 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2374         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2375         are not related to tree-cfg.c.
2376         * tree-ssa-address.h: New file.  Relocate prototypes.
2377         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2378         (addr_for_mem_ref): New.  Combine call to get_address_description and
2379         return addr_for_mem_ref.
2380         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2381         * tree-ssa-live.h: Adjust prototypes.
2382         * passes.c: Include tree-ssa-live.h.
2383         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2384         * graphite.c (graphite_transform_loops): Make static.
2385         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2386         make_pass_graphite, pass_data_graphite_transforms,
2387         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2388         * ipa-pure-const.c (warn_function_noreturn): Make static.
2389         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2390         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2391         Relocate from tree-cfg.c
2392         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2393         static.
2394         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2395         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2396         Move to ipa-pure-const.c.
2397         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2398         Relocate from tree-optimize.c.
2399         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2400         make_pass_fixup_cfg): Move to tree-cfg.c.
2401         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2402         Relocate some prototypes.
2403         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2404         * tree-dump.c (dump_function_to_file): Remove prototype.
2405         Add tree-flow.h to the include file.
2406         * tree-dump.h: Remove prototype.
2407         * tree-parloops.h: New File.  Add prototypes.
2408         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2409         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2410         from tree-ssa-loop.c.
2411         * tree-predcom.c (run_tree_predictive_commoning,
2412         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2413         Relocate here from tree-ssa-loop.c.
2414         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2415         ssa_name_values.release ().
2416         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2417         (ssa_name_values): Relocate from tree-flow.h.
2418         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2419         * tree-ssa-loop.c (run_tree_predictive_commoning,
2420         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2421         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2422         make_pass_graphite, pass_data_graphite_transforms,
2423         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2424         tree_parallelize_loops, pass_data_parallelize_loops,
2425         make_pass_parallelize_loops): Move to other files.
2426         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2427         moved here.
2428         * tree.h: Remove prototypes from tree-address.c.
2430 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2432         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2433         (struct int_tree_map): Move to tree-hasher.h
2434         (SCALE, LABEL, PERCENT): Move to gimple.h
2435         * tree-flow-inline.h: Delete.  Move functions to other files.
2436         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2437         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2438         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2439         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2440         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2441         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2442         * tree-hasher.h: Don't include tree-flow.h.
2443         (struct int_tree_map): Relocate from tree-flow.h.
2444         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2445         tree-flow-inline.h and make static.
2446         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2447         tree-flow-inline.h.
2448         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2449         tree-flow-inline.h and make static.
2450         * tree.h (is_global_var, may_be_aliased): Relocate from
2451         tree-flow-inline.h.
2452         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2453         * value-prof.c: No longer include tree-flow-inline.h.
2454         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2456 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2458         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2459         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2460         (enum move_pos): Move to tree-ssa-loop-im.h
2461         * cfgloop.h: Move some prototypes.
2462         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2463         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2464         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2465         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2466         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2467         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2468         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2469         (enum move_pos): Relocate here.
2470         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2471         tree-ssa-loop.c.
2472         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2473         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2474         make_pass_lim): Relocate here from tree-ssa-loop.c.
2475         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2476         tree-ssa-loop.c.
2477         (loop_edge_to_cancel, unloop_loops): Make static.
2478         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2479         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2480         (tree_complete_unroll, gate_tree_complete_unroll,
2481         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2482         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2483         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2484         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2485         (stmt_invariant_in_loop_p): Remove unused function.
2486         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2487         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2488         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2489         (gcov_type_to_double_int): Move to cfgloop.h.
2490         (double_int_cmp, bound_index,
2491         estimate_numbers_of_iterations_loop): Make static.
2492         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2493         (max_loop_iterations): Factor out get_max_loop_iterations.
2494         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2495         cfgloop.c.
2496         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2497         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2498         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2499         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2500         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2501         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2502         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2503         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2504         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2505         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2506         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2507         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2508         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2509         pass_data_complete_unroll, make_pass_complete_unroll,
2510         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2511         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2512         tree-ssa-loop-ivcanon.c.
2513         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2514         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2515         tree-ssa-loop-prefetch.c.
2516         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2517         tree-ssa-loop-im.c.
2518         (get_lsm_tmp_name): Relocate and add suffix parameter.
2519         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2520         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2521         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2522         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2523         (get_estimated_loop_iterations): Factor out accessor from
2524         estimated_loop_iterations in tree-ssa-loop-niter.c.
2525         (get_max_loop_iterations): Factor out accessor from
2526         _max_loop_iterations in tree-ssa-niter.c.
2527         * loop-unroll.c (decide_unroll_constant_iterations,
2528         decide_unroll_runtime_iterations, decide_peel_simple,
2529         decide_unroll_stupid): Use new get_* accessors.
2531 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2533         PR tree-optimization/20318
2534         * doc/extend.texi (returns_nonnull): New function attribute.
2535         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2536         attribute.
2537         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2538         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2540 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2542         PR middle-end/58570
2543         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2544         false if both components are bitfields.
2546 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2548         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2549         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2550         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2551         * config/aarch64/aarch64.h
2552         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2553         * config/aarch64/aarch64-simd-builtins.def
2554         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2556 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2558         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2560 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2562         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2564 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2566         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2567         (vneg_f64): New intrinsic.
2568         (vneg_s8): Asm replaced with C.
2569         (vneg_s16): Likewise.
2570         (vneg_s32): Likewise.
2571         (vneg_s64): New intrinsic.
2572         (vnegq_f32): Asm replaced with C.
2573         (vnegq_f64): Likewise.
2574         (vnegq_s8): Likewise.
2575         (vnegq_s16): Likewise.
2576         (vnegq_s32): Likewise.
2577         (vnegq_s64): Likewise.
2579 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2581         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2583 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2585         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2586         packed stack special handling.
2587         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2588         back to fixed stack slots for FPRs saved due to stdarg.
2590 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2592         * config/s390/s390.c (s390_frame_info): Restructure function.
2594 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2596         * config/s390/s390.c (struct s390_frame_layout): New field
2597         gpr_save_slots.
2598         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2599         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2600         regs_ever_clobbered to char*.
2601         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2602         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2603         (s390_register_info): Enable FPR save slots.  Move/Copy some
2604         functionality into ...
2605         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2606         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2607         function.
2608         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2609         not imply a stack frame.
2610         (s390_init_frame_layout): Remove variable clobbered_regs.
2611         (s390_update_register_info): Remove function.
2612         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2613         cannot be used for register renaming.
2614         (s390_hard_regno_scratch_ok): New function.
2615         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2616         (s390_initial_elimination_offset): Change offset calculation of
2617         the return address pointer.
2618         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2619         (restore_gprs): Set frame related flag.
2620         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2621         (s390_emit_prologue): Call s390_register_info instead of
2622         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2623         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2624         (s390_optimize_prologue): Call s390_optimize_register_info.
2625         Try to remove also FPR slot save/restore INSNs.  Remove frame
2626         related flags from restore INSNs.
2628 2013-10-08  DJ Delorie  <dj@redhat.com>
2630         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2631         (movhi): Likewise.
2633         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2634         avoid conflict with the MI use of %c.
2635         * config/rl78/rl78-real.md: change %c to %C throughout.
2636         * config/rl78/rl78-virt.md: Likewise.
2638 2013-10-08  Jan Hubicka  <jh@suse.cz>
2640         * config/i386/i386.c (ix86_option_override_internal): Switch
2641         to SSE math for -ffast-math when target ISA supports SSE2.
2643 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2645         * tree-flow.h: Remove some prototypes.
2646         * tree.h: Remove some protypes, add a couple.
2647         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2648         using_eh_for_cleanups_p): Add interface routines for front ends.
2649         * tree-eh.h: New file.  Add protoptyes.
2650         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2651         (add_stmt_to_eh_lp_fn): Make static.
2652         (lower_try_finally): Use new using_eh_for_cleanups_p.
2653         * emit-rtl.c: Include tree-eh.h.
2654         * gimple.h: Include tree-eh.h.
2656 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2658         PR tree-optimization/58480
2659         * tree-vrp.c (infer_nonnull_range): New function.
2660         (infer_value_range): Call infer_nonnull_range.
2662 2013-10-08  Dehao Chen  <dehao@google.com>
2664         PR tree-optimization/58619
2665         * tree-inline.c (copy_phis_for_bb): Combine location data
2666         only if non-null.
2668 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2670         PR target/58423
2671         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2672         RTX_FRAME_RELATED_P on INSN.
2674 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2676         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2677         (altivec_expand_vec_perm_const): Call it.
2679 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2681         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2682         sequences for LE VSX loads and stores at expand time.
2683         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2684         prototype.
2685         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2686         (rs6000_gen_le_vsx_permute): New.
2687         (rs6000_gen_le_vsx_load): New.
2688         (rs6000_gen_le_vsx_store): New.
2689         (rs6000_gen_le_vsx_move): New.
2690         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2691         (*vsx_le_perm_load_v4si): New.
2692         (*vsx_le_perm_load_v8hi): New.
2693         (*vsx_le_perm_load_v16qi): New.
2694         (*vsx_le_perm_store_v2di): New.
2695         (*vsx_le_perm_store_v4si): New.
2696         (*vsx_le_perm_store_v8hi): New.
2697         (*vsx_le_perm_store_v16qi): New.
2698         (*vsx_xxpermdi2_le_<mode>): New.
2699         (*vsx_xxpermdi4_le_<mode>): New.
2700         (*vsx_xxpermdi8_le_V8HI): New.
2701         (*vsx_xxpermdi16_le_V16QI): New.
2702         (*vsx_lxvd2x2_le_<mode>): New.
2703         (*vsx_lxvd2x4_le_<mode>): New.
2704         (*vsx_lxvd2x8_le_V8HI): New.
2705         (*vsx_lxvd2x16_le_V16QI): New.
2706         (*vsx_stxvd2x2_le_<mode>): New.
2707         (*vsx_stxvd2x4_le_<mode>): New.
2708         (*vsx_stxvd2x8_le_V8HI): New.
2709         (*vsx_stxvd2x16_le_V16QI): New.
2711 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2713         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2715 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2717         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2718         loop to work also for 31bit ABI.
2719         Save the stack pointer for frame_size > 0.
2721 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2723         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2724         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2725         constraint letters from expanders.
2726         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2727         retry count to general_operand.
2728         ("tabort"): Give operand 0 a mode.
2729         ("tabort_1"): Add mode and constraint letter for operand 0.
2730         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2732 2013-10-04  Jeff Law  <law@redhat.com>
2734         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2736         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2737         out of ...
2738         (thread_across_edge): Here.  Call it.
2740 2013-10-04  Cary Coutant  <ccoutant@google.com>
2742         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2743         discarding a location list expression (or a piece of one).
2745 2013-10-03  Jan Hubicka  <jh@suse.cz>
2747         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2748         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2749         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2751 2013-10-03  Jan Hubicka  <jh@suse.cz>
2753         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2754         accumulate-outgoing-args when producing unwind info.
2756 2013-10-03  Wei Mi  <wmi@google.com>
2758         * lra-constraints.c (insert_move_for_subreg): New function
2759         extracted from simplify_operand_subreg.
2760         (simplify_operand_subreg): Add reload for paradoxical subreg.
2762 2013-10-03  Rong Xu  <xur@google.com>
2764         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2765         the candidate of builtin_expect such that we should fix the
2766         size/time estimation.
2767         (estimate_function_body_sizes): Do the acutally size/time fix-up
2768         for builtin_expect.
2770 2013-10-03  Rong Xu  <xur@google.com>
2772         * predict.c (tree_predict_by_opcode): Get the probability
2773         for builtin_expect from param builtin_expect_probability.
2774         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2775         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2776         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2778 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2780         PR c++/19476
2781         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2782         * calls.c (alloca_call_p): Use get_callee_fndecl.
2783         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2784         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2785         Likewise.
2786         (vrp_visit_stmt): Remove duplicated code.
2788 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2790         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2791         ceildf2, btruncdf2, instead of vsx_* name.
2793         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2794         iterators to only do V2DF and V4SF here.  Move the DF code to
2795         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2796         just 'v' since only vector operations are handled with these insns
2797         after moving the DF support to rs6000.md.
2798         (vsx_sub<mode>3): Likewise.
2799         (vsx_mul<mode>3): Likewise.
2800         (vsx_div<mode>3): Likewise.
2801         (vsx_fre<mode>2): Likewise.
2802         (vsx_neg<mode>2): Likewise.
2803         (vsx_abs<mode>2): Likewise.
2804         (vsx_nabs<mode>2): Likewise.
2805         (vsx_smax<mode>3): Likewise.
2806         (vsx_smin<mode>3): Likewise.
2807         (vsx_sqrt<mode>2): Likewise.
2808         (vsx_rsqrte<mode>2): Likewise.
2809         (vsx_fms<mode>4): Likewise.
2810         (vsx_nfma<mode>4): Likewise.
2811         (vsx_copysign<mode>3): Likewise.
2812         (vsx_btrunc<mode>2): Likewise.
2813         (vsx_floor<mode>2): Likewise.
2814         (vsx_ceil<mode>2): Likewise.
2815         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2816         (vsx_sminsf3): Likewise.
2817         (vsx_fmadf4): Likewise.
2818         (vsx_fmsdf4): Likewise.
2819         (vsx_nfmadf4): Likewise.
2820         (vsx_nfmsdf4): Likewise.
2821         (vsx_cmpdf_internal1): Likewise.
2823         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2824         simpler to select whether a target has SPE or traditional floating
2825         point support in iterators.
2826         (TARGET_DF_SPE): Likewise.
2827         (TARGET_SF_FPR): Likewise.
2828         (TARGET_DF_FPR): Likewise.
2829         (TARGET_SF_INSN): Macros to say whether floating point support
2830         exists for a given operation for expanders.
2831         (TARGET_DF_INSN): Likewise.
2833         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2834         combining of SF/DF mode operations, using both traditional and VSX
2835         registers.
2836         (Fvsx): Likewise.
2837         (Ff): Likewise.
2838         (Fv): Likewise.
2839         (Fs): Likewise.
2840         (Ffre): Likewise.
2841         (FFRE): Likewise.
2842         (abs<mode>2): Combine SF/DF modes using traditional floating point
2843         instructions.  Add support for using the upper DF registers with
2844         VSX support, and SF registers with power8-vector support.  Update
2845         expanders for operations supported by both the SPE and traditional
2846         floating point units.
2847         (abs<mode>2_fpr): Likewise.
2848         (nabs<mode>2): Likewise.
2849         (nabs<mode>2_fpr): Likewise.
2850         (neg<mode>2): Likewise.
2851         (neg<mode>2_fpr): Likewise.
2852         (add<mode>3): Likewise.
2853         (add<mode>3_fpr): Likewise.
2854         (sub<mode>3): Likewise.
2855         (sub<mode>3_fpr): Likewise.
2856         (mul<mode>3): Likewise.
2857         (mul<mode>3_fpr): Likewise.
2858         (div<mode>3): Likewise.
2859         (div<mode>3_fpr): Likewise.
2860         (sqrt<mode>3): Likewise.
2861         (sqrt<mode>3_fpr): Likewise.
2862         (fre<Fs>): Likewise.
2863         (rsqrt<mode>2): Likewise.
2864         (cmp<mode>_fpr): Likewise.
2865         (smax<mode>3): Likewise.
2866         (smin<mode>3): Likewise.
2867         (smax<mode>3_vsx): Likewise.
2868         (smin<mode>3_vsx): Likewise.
2869         (negsf2): Delete SF operations that are merged with DF.
2870         (abssf2): Likewise.
2871         (addsf3): Likewise.
2872         (subsf3): Likewise.
2873         (mulsf3): Likewise.
2874         (divsf3): Likewise.
2875         (fres): Likewise.
2876         (fmasf4_fpr): Likewise.
2877         (fmssf4_fpr): Likewise.
2878         (nfmasf4_fpr): Likewise.
2879         (nfmssf4_fpr): Likewise.
2880         (sqrtsf2): Likewise.
2881         (rsqrtsf_internal1): Likewise.
2882         (smaxsf3): Likewise.
2883         (sminsf3): Likewise.
2884         (cmpsf_internal1): Likewise.
2885         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2886         (negdf2): Delete DF operations that are merged with SF.
2887         (absdf2): Likewise.
2888         (nabsdf2): Likewise.
2889         (adddf3): Likewise.
2890         (subdf3): Likewise.
2891         (muldf3): Likewise.
2892         (divdf3): Likewise.
2893         (fred): Likewise.
2894         (rsqrtdf_internal1): Likewise.
2895         (fmadf4_fpr): Likewise.
2896         (fmsdf4_fpr): Likewise.
2897         (nfmadf4_fpr): Likewise.
2898         (nfmsdf4_fpr): Likewise.
2899         (sqrtdf2): Likewise.
2900         (smaxdf3): Likewise.
2901         (smindf3): Likewise.
2902         (cmpdf_internal1): Likewise.
2903         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2904         (btrunc<mode>2): Delete separate expander, and combine with the
2905         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2906         (btrunc<mode>2_fpr): Likewise.
2907         (ceil<mode>2): Likewise.
2908         (ceil<mode>2_fpr): Likewise.
2909         (floor<mode>2): Likewise.
2910         (floor<mode>2_fpr): Likewise.
2911         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2912         Add support for using the upper registers with VSX and
2913         power8-vector.  Move insns to be closer to the define_expands. On
2914         VSX systems, prefer the traditional form of FMA over the VSX
2915         version, since the traditional form allows the target not to
2916         overlap with the inputs.
2917         (fms<mode>4_fpr): Likewise.
2918         (nfma<mode>4_fpr): Likewise.
2919         (nfms<mode>4_fpr): Likewise.
2921 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2922             Richard Earnshaw  <richard.earnshaw@arm.com>
2924         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2925         (struct mult_cost_table): Likewise.
2926         (struct mem_cost_table): Likewise.
2927         (struct fp_cost_table): Likewise.
2928         (struct vector_cost_table): Likewise.
2929         (cpu_cost_table): Likewise.
2930         * config/arm/arm.opt (mold-rts-costs): New option.
2931         (mnew-generic-costs): Likewise.
2932         * config/arm/arm.c (generic_extra_costs): New table.
2933         (cortexa15_extra_costs): Likewise.
2934         (arm_slowmul_tune): Use NULL as new costs.
2935         (arm_fastmul_tune): Likewise.
2936         (arm_strongarm_tune): Likewise.
2937         (arm_xscale_tune): Likewise.
2938         (arm_9e_tune): Likewise.
2939         (arm_v6t2_tune): Likewise.
2940         (arm_cortex_a5_tune): Likewise.
2941         (arm_cortex_a9_tune): Likewise.
2942         (arm_v6m_tune): Likewise.
2943         (arm_fa726te_tune): Likewise.
2944         (arm_cortex_a15_tune): Use cortex15_extra_costs.
2945         (arm_cortex_tune): Use generict_extra_costs.
2946         (shifter_op_p): New function.
2947         (arm_unspec_cost): Likewise.
2948         (LIBCALL_COST): Define.
2949         (arm_new_rtx_costs): New function.
2950         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
2951         table is available. Use old costs otherwise unless mnew-generic-costs
2952         is specified.
2953         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
2954         (cpu_cost_table): Declare.
2956 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2958         PR target/58460
2959         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
2960         (*subs_mul_imm_<mode>)
2961         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
2962         (*sub_<shift>_<mode>)
2963         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
2964         Remove k constraint.
2966 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
2968         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
2969         code.
2970         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
2972 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2974         * predict.c (probably_never_executed): New function.
2975         (probably_never_executed_bb_p): Invoke probably_never_executed.
2976         (probably_never_executed_edge_p): Ditto.
2977         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
2978         Treat profile insanities conservatively.
2980 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
2982         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
2984 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
2986         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
2987         use it.  Use smaller increase for scratch.  Don't increase reject
2988         for early clobber scratch.
2989         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
2990         setting eliminated regs except setting fp from hfp.
2991         (lra_eliminate): Check lra_insn_recog_data on NULL.
2993 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
2995         PR target/58587
2996         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
2997         setting -mvsx-timode by default until the underlying problem is fixed.
2998         (RS6000_CPU, power7 defaults): Likewise.
3000 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3002         * config/x-linux (host-linux.o): Remove header dependencies.
3003         Use $(COMPILE) and $(POSTCOMPILE).
3004         * config/t-linux-android (linux-android.o): Ditto.
3006 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3008         * Makefile.in (expmed.o-warn): Remove.
3010 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3012         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3013         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3015 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3017         * dojump.c (do_jump_1): Divide probability between
3018         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3020 2013-10-02  Tom Tromey  <tromey@redhat.com>
3022         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3024 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3026         * tree-flow.h: Remove some prototypes.
3027         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3028         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3029         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3030         mark_virtual_phi_result_for_renaming): Relocate here.
3031         * tree-into-ssa.h: Add prototypes.
3032         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3033         single_pred_before_succ_order.
3034         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3035         (nonfreeing_call_p) Move to gimple.c.
3036         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3037         tree-ssa-phiopt.c.
3038         * basic-block.h (single_pred_before_succ_order): Add prototype.
3039         * gimple.c (nonfreeing_call_p): Relocate here.
3040         * gimple.h: Add prototype.
3041         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3042         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3043         * tree-ssa.h: Include tree-ssa-dom.h.
3045 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3047         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3048         Use $(COMPILE) and $(POSTCOMPILE).
3050         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3052 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3054         * tree-flow.h: Remove some prototypes.
3055         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3056         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3057         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3058         propagate_tree_value*): Move from here to...
3059         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3060         propagate_tree_value*): Relocate here.
3061         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3062         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3063         * gimple-fold.c: Remove gimple-fold.h from include list.
3064         * tree-vrp.c: Remove gimple-fold.h from include list.
3065         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3066         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3067         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3068         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3069         * sese.c: Add tree-ssa-propagate.h to include list.
3071 2013-10-02  Richard Biener  <rguenther@suse.de>
3073         * tree-loop-distribution.c: Include tree-vectorizer.h for
3074         find_loop_location.
3075         (enum partition_kind): Remove PKIND_REDUCTION.
3076         (struct partition_s): Remove has_writes member, add reduction_p member.
3077         (partition_alloc): Adjust.
3078         (partition_builtin_p): Likewise.
3079         (partition_has_writes): Remove.
3080         (partition_reduction_p): New function.
3081         (partition_merge_into): Likewise.
3082         (generate_code_for_partition): Commonize builtin partition
3083         handling tail.
3084         (rdg_cannot_recompute_vertex_p): Remove.
3085         (already_processed_vertex_p): Likewise.
3086         (rdg_flag_vertex): Do not set has_writes.
3087         (classify_partition): Adjust.
3088         (rdg_build_partitions): Do not set has_writes, treat all
3089         partitions as useful.
3090         (distribute_loop): Record number of library calls generated.  Adjust.
3091         (tree_loop_distribution): Report number of loops and library
3092         calls generated as opt-info.
3094 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3096         * tree-flow.h: Include new .h files.  Move prototypes.
3097         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3098         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3099         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3100         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3101         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3102         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3103         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3104         * tree.h (get_ref_base_and_extent): Move prototype out.
3105         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3106         tree-dfa.h.
3107         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3108         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3109         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3110         * tree-scalar-evolution.c: Include tree.h.
3111         * sese.c: Include tree.h.
3112         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3113         * dwarf2out.c: Include tree-dfa.h.
3114         * tree-chrec.c: Include tree.h.
3115         * tree-data-ref.c: Include tree.h.
3117 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3119         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3120         Fix whitespace.
3122 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3124         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3125         Use $(COMPILE) and $(POSTCOMPILE).
3126         (sol2-cxx.o): Likewise.
3127         (sol2-stubs.o): Likewise.
3128         (sol2.o): Likewise.
3129         * config/x-solaris (host-solaris.o): Likewise.
3131         * config/sparc/t-sparc (sparc.o): Remove.
3132         (sparc-c.o): Remove header dependencies.
3133         Use $(COMPILE) and $(POSTCOMPILE).
3134         * config/sparc/x-sparc: Likewise.
3136 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3138         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3139         * config/arc/arc700.md: Likewise.
3140         * config/arc/arc-modes.def: Likewise.
3141         * config/arc/arc-simd.h: Likewise.
3142         * config/arc/t-arc-uClibc: Likewise.
3143         * config/arc/t-arc-newlib: Likewise.
3145 2013-10-02  Renlin Li  <renlin.li@arm.com>
3147         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3148         plus_constant.
3149         (aarch64_expand_epilogue): Likewise.
3151 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3152             Yufeng Zhang  <yufeng.zhang@arm.com>
3154         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3155         declaration.
3156         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3157         'base_in' represent a conversion and legal_cast_p_1 holds; set
3158         'base_in' with the returned value from get_unwidened.
3160 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3162         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3163         plus_constant is not used.
3165 2013-10-01  Wei Mi  <wmi@google.com>
3167         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3168         * config/i386/i386.md: Add define_peephole2 to
3169         break partial reg stall for cvtss2sd/cvtsd2ss.
3171 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3173         * config/arc/arc.c (pass_arc_ifcvt::clone):
3174         Update for ctxt_ -> m_ctxt change.
3176 2013-10-01  Jeff Law  <law@redhat.com>
3178         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3179         outgoing_edge and intermediate_edge fields.  Instead store the path.
3180         (redirection_data::hash): Hash on the last edge's destination index.
3181         (redirection_data::equal): Check the entire thread path.
3182         (lookup_redirectio_data): Corresponding changes.
3183         (create_edge_and_update_destination_phis): Likewise.
3184         (thread_single_edge): Likewise.
3186 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3187             Diego Novillo <dnovillo@google.com>
3189         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3190         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3191         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3192         versions of these patterns.
3194         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3195         (__builtin_arc_aligned): Likewise.
3197         * config/arc/arc.md: Expand adc_0 comment stating the intended
3198         purpose and why it isn't ready.
3199         Replace commented out call_value_via_label_mixed with a
3200         plain comment about bl_s.
3202         * config/arc/arc.c (stdio.h): Don't include directly.
3203         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3204         Remove if (1) condition.
3205         (arc_encode_section_info): Fix comment.
3207 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3209         * config/arc/arc.c (arc_conditional_register_usage):
3210         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3211         Also set reg_alloc_order for DMA config regs.
3213 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3214             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3216         * doc/install.texi (--with-cpu): Mention ARC.
3217         (arc-*-elf32): New paragraph.
3218         (arc-linux-uclibc): Likewise.
3219         * doc/md.texi (Machine Constraints): Add ARC part.
3220         * doc/invoke.texi: (menu): Add ARC Options.
3221         (Machine Dependent Options) <ARC Options>: Add synopsis.
3222         (node ARC Options): Add.
3223         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3224         (ARC Built-in Functions): New section defining
3225         generic ARC built-in functions.
3226         (ARC SIMD Built-in Functions): New section defining SIMD specific
3227         built-in functions.
3228         (Declaring Attributes of Functions): Extended
3229         description of short_call and long_call attributes for ARC and
3230         added index entries.
3232 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3233             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3234             Joern Rennecke  <joern.rennecke@embecosm.com>
3235             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3236             Brendan Kehoe  <brendan@zen.org>
3237             Michael Eager  <eager@eagercon.com>
3238             Simon Cook  <simon.cook@embecosm.com>
3239             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3241         * config/arc, common/config/arc: New directories.
3243 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3244             Brendan Kehoe  <brendan@zen.org>
3245             Simon Cook  <simon.cook@embecosm.com>
3247         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3249 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3251         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3252         * tree-ssa-coalesce.h: New. Move prototype to here.
3253         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3254         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3255         (gimple_can_coalesce_p): Move to...
3256         * gimple.c (gimple_can_coalesce_p): Here.
3258 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3260         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3261         (dump_decl_set): Move to gimple.c.
3262         * gimple.h: Don't include tree-ssa-operands.h.
3263         (dump_decl_set): Add prototype.
3264         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3265         Move to gimple-ssa.h.
3266         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3267         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3268         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3269         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3270         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3271         than PHI_ARG_DEF.
3272         (dump_decl_set): Relocate here.
3273         * gimple-ssa.h: New file.
3274         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3275         Relocate from gimple.h.
3276         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3277         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3278         swap_tree_operands and remove non-ssa path.
3279         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3280         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3281         swap_ssa_operands.
3282         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3283         vect_is_simple_reduction_1): Use swap_ssa_operands.
3284         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3285         (enum need_phi_state): Move to tree-into-ssa.c.
3286         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3287         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3288         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3289         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3290         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3291         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3292         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3293         num_imm_uses): Move to ssa-iterators.h.
3294         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3295         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3296         tree-phinodes.h.
3297         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3298         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3299         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3300         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3301         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3302         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3303         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3304         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3305         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3306         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3307         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3308         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3309         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3310         (set_phi_nodes): Move to tree-phinodes.h.
3311         * tree-ssa-operands.h (enum ssa_op_iter_type,
3312         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3313         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3314         (dump_decl_set): Remove prototype.
3315         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3316         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3317         (set_phi_nodes): Relocate from tree-flow-inline.h.
3318         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3319         tree-flow-inline.h
3320         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3321         include list.  Temporarily add gimple.h to include list.
3322         * ssa-iterators.h: New file.
3323         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3324         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3325         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3326         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3327         Relocate from tree-ssa-operands.h.
3328         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3329         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3330         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3331         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3332         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3333         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3334         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3335         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3336         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3337         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3338         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3339         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3340         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3341         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3342         Relocate from tree-flow-inline.h.
3343         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3345 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3347         * aarch64-simd.md
3348         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3349         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3350         Insert '\t' to output template.
3351         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3352         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3353         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3354         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3356 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3358         * doc/install.texi (Host/target specific installation notes for GCC):
3359         Put @anchor before @heading.
3360         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3361         Use @email for email addresses.
3363 2013-10-01  Jeff Law  <law@redhat.com>
3365         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3366         a vec.  Only delete the path if we create one without successfully
3367         registering a jump thread.
3368         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3369         as a pointer.
3370         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3371         (paths): New vector of jump threading paths.
3372         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3373         (THREAD_PATH): New accessor macro for the entire thread path.
3374         (lookup_redirection_data): Get intermediate and final outgoing edge
3375         from the thread path.
3376         (create_edge_and_update_destination_phis): Copy the threading path.
3377         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3378         jump threading path.
3379         (ssa_redirect_edges): Get edges and block types from the jump threading
3380         path.  Free the path vector.
3381         (thread_block): Get edges from the jump threading path.  Look at the
3382         entire path to see if we thread to a loop exit.  If we cancel a jump
3383         thread request, then free the path vector.
3384         (thread_single_edge): Get edges and block types from the jump threading
3385         path.  Free the path vector.
3386         (thread_through_loop_header): Get edges and block types from the jump
3387         threading path.  Free the path vector.
3388         (mark_threaded_blocks): Iterate over the vector of paths and store
3389         the path on the appropriate edge.  Get edges and block types from the
3390         jump threading path.
3391         (mark_threaded_blocks): Get edges and block types from the jump
3392         threading path.  Free the path vector.
3393         (thread_through_all_blocks): Use the vector of paths rather than
3394         a vector of 3-edge sets.
3395         (register_jump_thread): Accept pointer to a path vector rather
3396         than the path vector itself.  Store the path vector for later use.
3397         Simplify.
3399 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3400             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3402         PR target/58574
3403         * config/s390/s390.c (s390_split_branches): Modify check for table
3404         jump insns.
3405         (s390_chunkify_start): Rearrange table jump insn check in order to
3406         deal with compare and branch insns correctly.
3408 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3410         PR target/58578
3411         Revert
3412         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3413         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3414         define_insn_and_split.
3415         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3416         (shiftsi3_compare): New pattern.
3417         (rrx): New pattern.
3418         * config/arm/unspecs.md (UNSPEC_RRX): New.
3420 2013-10-01  Alan Modra  <amodra@gmail.com>
3422         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3423         casing inout operands.
3425 2013-10-01  Richard Biener  <rguenther@suse.de>
3427         PR tree-optimization/58553
3428         * tree-loop-distribution.c (struct partition_s): Add niter member.
3429         (classify_partition): Populate niter member for the partition
3430         and properly identify whether the relevant store happens before
3431         or after the loop exit.
3432         (generate_memset_builtin): Use niter member from the partition.
3433         (generate_memcpy_builtin): Likewise.
3435 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3437         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3438         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3440 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3442         * basic-block.h (edge_list): Prefix member names with "m_".
3443         * context.h (context): Likewise.
3444         * domwalk.h (dom_walker): Likewise.
3445         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3446         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3447         * hash-table.h (hash_table): Likewise.
3448         * machmode.h (bit_field_mode_iterator): Likewise.
3449         * pass_manager.h (pass_list): Likewise.
3450         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3451         * tree-pass.h (pass_data): Likewise.
3452         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3453         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3454         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3455         * asan.c (pass_data_asan): Update accordingly.
3456         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3457         (control_dependences::control_dependences): Likewise.
3458         (control_dependences::~control_dependences): Likewise.
3459         (control_dependences::~control_dependences): Likewise.
3460         (control_dependences::get_edges_dependent_on): Likewise.
3461         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3462         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3463         * context.c (gcc::context::context): Likewise.
3464         * cprop.c (pass_rtl_cprop::clone): Likewise.
3465         * domwalk.c (dom_walker::walk): Likewise.
3466         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3467         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3468         * mode-switching.c (pass_mode_switching::clone): Likewise.
3469         * passes.c (opt_pass::opt_pass): Likewise.
3470         (pass_manager::pass_manager): Likewise.
3471         * predict.c (pass_strip_predict_hints::clone): Likewise.
3472         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3473         (pass_split_all_insns::clone): Likewise.
3474         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3475         Likewise.
3476         (bit_field_mode_iterator::next_mode): Likewise.
3477         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3478         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3479         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3480         * tree-complex.c (pass_lower_complex::clone): Likewise.
3481         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3482         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3483         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3484         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3485         (pass_fold_builtins::clone): Likewise.
3486         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3487         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3488         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3489         (pass_cd_dce::clone): Likewise.
3490         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3491         (pass_phi_only_cprop::clone): Likewise.
3492         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3493         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3494         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3495         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3496         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3497         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3498         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3499         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3500         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3501         * tree-vrp.c (pass_vrp::clone): Likewise.
3502         * tsan.c (pass_tsan::clone): Likewise.
3504 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3506         PR middle-end/58564
3507         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3508         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3510         PR middle-end/58564
3511         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3512         optimization, punt if sign_bit_p looked through any zero extension.
3514 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3516         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3517         Update redirected out edge count in joiner case.
3518         (ssa_redirect_edges): Common the joiner and non-joiner cases
3519         so that joiner case gets profile updates.
3521 2013-09-30  Richard Biener  <rguenther@suse.de>
3523         PR tree-optimization/58554
3524         * tree-loop-distribution.c (classify_partition): Require
3525         unconditionally executed stores for memcpy and memset recognition.
3526         (tree_loop_distribution): Calculate dominance info.
3528 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3530         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3531         (NO_PROFILE_COUNTERS): Likewise.
3532         (PROFILE_HOOK): Likewise.
3533         (FUNCTION_PROFILER): Likewise.
3534         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3536 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3538         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3539         calls and defines in TARGET_MACHO conditional.
3540         (load_macho_picbase_di): Likewise.
3541         (reload_macho_picbase): Likewise.
3542         (reload_macho_picbase_si): Likewise.
3543         (reload_macho_picbase_di): Likewise.
3544         (nonlocal_goto_receiver): Likewise.
3546 2013-09-30  Nick Clifton  <nickc@redhat.com>
3548         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3549         that use the MSP430X ISA.
3550         (msp430_option_override): Scan -mmcu command line option for any
3551         MCU name that supports the MSP430X ISA.
3552         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3553         -mmcu options which enable the MSP430X ISA.
3555 2013-09-30  Richard Biener  <rguenther@suse.de>
3557         PR middle-end/58532
3558         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3559         before looking for setjmp-like calls.
3561 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3563         PR target/10901
3564         * config/darwin-protos.h (machopic_get_function_picbase): New.
3565         * config/darwin.c (machopic_get_function_picbase): New.
3566         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3567         label for a new func.  (load_macho_picbase_di): Likewise.
3568         (reload_macho_picbase): New expand.
3569         (reload_macho_picbase_si): New insn.
3570         (reload_macho_picbase_di): New insn.
3571         (nonlocal_goto_receiver): New define and split.
3572         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3573         (unspecv enum): Add UNSPECV_NLGR.
3575 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3577         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3578         that altivec registers are correctly sized on Darwin.
3580 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3582         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3583         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3584         * config/x-darwin (host-darwin.o): Likewise.
3585         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3586         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3587         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3589 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3591         * doc/invoke.texi: Fix usage of @tie{} command.
3593 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3595         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3597 2013-09-28  Jan Hubicka  <jh@suse.cz>
3599         * config/i386/x86-tune.def: Add documentation for each of the options;
3600         add whitespace.
3602 2013-09-28  Jan Hubicka  <jh@suse.cz>
3604         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3605         generic.
3606         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3607         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3608         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3609         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3610         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3612 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3614         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3615         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3616         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3617         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3618         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3619         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3620         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3621         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3622         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3623         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3624         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3625         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3626         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3627         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3628         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3629         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3630         gimple.h, godump.c, graphite-clast-to-gimple.c,
3631         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3632         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3633         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3634         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3635         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3636         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3637         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3638         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3639         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3640         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3641         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3642         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3643         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3644         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3645         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3646         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3647         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3648         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3649         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3650         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3651         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3652         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3653         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3654         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3655         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3656         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3657         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3658         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3659         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3660         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3661         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3662         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3663         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3664         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3665         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3666         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3667         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3668         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3669         whitespace before "(".
3671 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3673         * expr.h (extract_bit_field): Remove packedp parameter.
3674         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3675         from forward declaration.
3676         (store_split_bit_field): Remove packedp arg from calls to
3677         extract_fixed_bit_field.
3678         (extract_bit_field_1): Remove packedp parameter and packedp
3679         argument from recursive calls and calls to extract_fixed_bit_field.
3680         (extract_bit_field): Remove packedp parameter and corresponding
3681         arg to extract_bit_field_1.
3682         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3683         to issue warnings.
3684         (extract_split_bit_field): Remove packedp arg from call to
3685         extract_fixed_bit_field.
3686         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3687         (copy_blkmode_from_reg): Likewise.
3688         (copy_blkmode_to_reg): Likewise.
3689         (read_complex_part): Likewise.
3690         (store_field): Likewise.
3691         (expand_expr_real_1): Likewise.
3692         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3693         to extract_bit_field.
3694         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3695         call to extract_bit_field.
3696         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3697         call to extract_bit_field.
3698         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3699         and special packedp behavior from -fstrict-volatile-bitfields
3700         documentation.
3702 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3704         * lra-eliminations.c (init_elim_table): Guard value_p.
3706 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3708         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3709         DFmode, DImode, and SFmode in the upper VSX registers based on the
3710         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3711         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3712         of the VSX register handling.
3714         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3715         (f32_sv): Likewise.
3716         (zero_extendsidi2_lfiwzx): Add support for loading into the
3717         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3718         only do VSX memory options on Altivec registers.
3719         (extendsidi2_lfiwax): Likewise.
3720         (extendsfdf2_fpr): Likewise.
3721         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3722         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3723         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3724         (movdi_internal64): Likewise.
3726 2013-09-27  Xinliang David Li  <davidxl@google.com>
3728         * opts.c (finish_options): Adjust parameters
3729         according to vect cost model.
3730         (common_handle_option): Set dynamic vect cost
3731         model for FDO.
3732         targhooks.c (default_add_stmt_cost): Compute stmt cost
3733         unconditionally.
3734         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3735         Use helper function.
3736         * tree-vectorizer.h (unlimited_cost_model): New function.
3737         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3738         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3739         function.
3740         (vect_enhance_data_refs_alignment): Ditto.
3741         * flag-types.h: New enum.
3742         * common/config/i386/i386-common.c (ix86_option_init_struct):
3743         No need to initialize vect_cost_model flag.
3744         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3745         unconditionally.
3747 2013-09-27  Diego Novillo  <dnovillo@google.com>
3749         * gimple.h (enum ssa_mode): Remove.
3751 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3753         * cfgloop.h (number_of_loops): Fix typo in check for null.
3755 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3757         PR middle-end/58551
3758         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3759         are children of outermost saved_cfun's loop, and set it up to
3760         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3761         if loop != loop0 and SESE region contains bbs that belong to loop0.
3763 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3765         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3766         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3767         (set_address_segment, set_address_base, set_address_index)
3768         (set_address_disp): Accept the argument unconditionally.
3769         (baseness): Remove must_be_base_p and must_be_index_p checks.
3770         (decompose_normal_address): Classify as much as possible in the
3771         main loop.
3773 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3775         * cse.c (count_reg_usage): Handle INT_LIST.
3776         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3777         * reginfo.c (reg_scan_mark_refs): Likewise.
3778         * reload1.c (eliminate_regs_1): Likewise.
3780 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3782         PR middle-end/58547
3783         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3784         signed.
3786 2013-09-27  Richard Biener  <rguenther@suse.de>
3788         PR tree-optimization/58459
3789         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3790         restriction not propagating into loops.
3792 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3794         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3795         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3796         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3798 2013-09-26  Richard Biener  <rguenther@suse.de>
3800         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3802 2013-09-26  Richard Biener  <rguenther@suse.de>
3804         * alias.h (component_uses_parent_alias_set): Rename to ...
3805         (component_uses_parent_alias_set_from): ... this.
3806         * alias.c (component_uses_parent_alias_set): Rename to ...
3807         (component_uses_parent_alias_set_from): ... this and return
3808         the desired parent.
3809         (reference_alias_ptr_type_1): Use the result from
3810         component_uses_parent_alias_set_from instead of stripping
3811         components one at a time.
3812         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3814 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3816         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3817         Move prototypes to...
3818         * tree-ssa-ter.h: New File.  Move prototypes here.
3819         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3820         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3821         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3822         from is_replaceable_p.
3823         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3824         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3825         (process_replaceable): Use ter_is_replaceable_p.
3826         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3827         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3828         newly refactored ssa_is_replaceable_p.
3829         * cfgexpand.c: Include tree-outof-ssa.h.
3830         * ssaexpand.h: Delete.
3832 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3834         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3835         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3836         tree-ssa.c
3837         (create_gimple_tmp): Delete.
3838         (get_expr_type, build_assign, build_type_cast): Move to...
3839         * gimple-builder.c: New File.
3840         (get_expr_type): Relocate from gimple.c.
3841         (build_assign, build_type_cast): Change to only create ssanames.
3842         * gimple.h: Move prototypes to...
3843         * gimple-builder.h: New File. Here.
3844         * tree-ssa.h: And here.
3845         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3846         count_uses_and_derefs): Relocate from gimple.c.
3847         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3848         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3849         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3850         gimple_replace_ssa_lhs.
3851         * asan.c: Include gimple-builder.h.
3852         * Makefile.in: Add gimple-builder.o.
3854 2013-09-26  Richard Biener  <rguenther@suse.de>
3856         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3857         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3858         (loe_visit_block): Use gcc_checking_assert.
3859         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3860         gimple_assign_ssa_name_copy_p.
3861         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3862         change.
3864 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3866         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3867         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3869 2013-09-26  Richard Biener  <rguenther@suse.de>
3871         PR tree-optimization/58539
3872         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3873         the fact that debug statements are not taking part in loop-closed
3874         SSA construction.
3876 2013-09-26  Nick Clifton  <nickc@redhat.com>
3878         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3879         time warning message.
3880         (msp430_print_operand_raw): Delete unused letter parameter.
3881         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3882         (msp430_print_operand_address): New function.
3883         (msp430_print_operand): Move address printing code from here to
3884         new function.
3885         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3886         assembler.
3887         (zero_extendpsisi2): Likewise.
3888         (extendpsisi2): New pattern.
3889         (andneghi3): New pattern.
3891 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3893         * config/aarch64/aarch64.opt (mlra): New option.
3894         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3895         (TARGET_LRA_P): Define.
3897 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3899         * expr.c (expand_assignment): Remove obsolete comment.
3901 2013-09-25  Jeff Law  <law@redhat.com>
3903         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3904         tree-ssa-threadupdate.h.
3905         (register_jump_thread): Similarly.
3906         * tree-ssa-threadupdate.h: New header file.
3907         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3908         * tree-vrp.c: Likewise.
3909         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3910         (thread_around_empty_blocks): Change type of path vector argument to
3911         an edge,type pair from just an edge.  Initialize both elements when
3912         appending to a jump threading path.  Tweak references to elements
3913         appropriately.
3914         (thread_across_edge): Similarly.  Release memory for the elements
3915         as needed.
3916         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3917         (dump_jump_thread_path): New function broken out from
3918         register_jump_thread.
3919         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3920         path vector entries.  Search the path for NULL edges and dump
3921         the path if one is found.  Tweak the conversion of path to 3-edge
3922         form to use the block copy type information embedded in the path.
3924 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3926         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3928 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3929             Vladimir Makarov  <vmakarov@redhat.com>
3931         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3932         from the least significant bit.
3933         (strip_address_mutations): Add bitfield operations handling.
3934         (must_be_index_p): Add shifting and rotate operations handling.
3935         (set_address_base): Use must_be_base_p predicate.
3936         (set_address_index): Use must_be_index_p predicate.
3938 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3939             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3940             Sergey Lega  <sergey.s.lega@intel.com>
3941             Anna Tikhonova  <anna.tikhonova@intel.com>
3942             Ilya Tocar  <ilya.tocar@intel.com>
3943             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3944             Ilya Verbin  <ilya.verbin@intel.com>
3945             Kirill Yukhin  <kirill.yukhin@intel.com>
3946             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3948         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
3949         Use new names.
3950         (ix86_expand_vector_move_misalign): Support new unaligned load and
3951         stores and use new names.
3952         (CODE_FOR_sse2_storedqu): Rename to ...
3953         (CODE_FOR_sse2_storedquv16qi): ... this.
3954         (CODE_FOR_sse2_loaddqu): Rename to ...
3955         (CODE_FOR_sse2_loaddquv16qi): ... this.
3956         (CODE_FOR_avx_loaddqu256): Rename to ...
3957         (CODE_FOR_avx_loaddquv32qi): ... this.
3958         (CODE_FOR_avx_storedqu256): Rename to ...
3959         (CODE_FOR_avx_storedquv32qi): ... this.
3960         * config/i386/i386.md (fpint_logic): New.
3961         * config/i386/sse.md (VMOVE): Extend for AVX512.
3962         (VF): Ditto.
3963         (VF_128_256): New.
3964         (VF_512): Ditto.
3965         (VI_UNALIGNED_LOADSTORE): Ditto.
3966         (sse2_avx_avx512f): Ditto.
3967         (sse2_avx2): Extend for AVX512.
3968         (sse4_1_avx2): Ditto.
3969         (avx2_avx512f): New.
3970         (sse): Extend for AVX512.
3971         (sse2): Ditto.
3972         (sse4_1): Ditto.
3973         (avxsizesuffix): Ditto.
3974         (sseintvecmode): Ditto.
3975         (ssePSmode): Ditto.
3976         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
3977         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
3978         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3979         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
3980         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3981         (<sse2_avx_avx512f>_storedqu<mode): ... this.
3982         (<sse>_movnt<mode>): Replace constraint "x" with "v".
3983         (STORENT_MODE): Extend for AVX512.
3984         (*absneg<mode>2): Replace constraint "x" with "v".
3985         (*mul<mode>3): Ditto.
3986         (*ieee_smin<mode>3): Ditto.
3987         (*ieee_smax<mode>3): Ditto.
3988         (avx_cmp<mode>3): Replace VF with VF_128_256.
3989         (*<sse>_maskcmp<mode>3_comm): Ditto.
3990         (<sse>_maskcmp<mode>3): Ditto.
3991         (<sse>_andnot<mode>3): Extend for AVX512.
3992         (<code><mode>3, anylogic): Replace VF with VF_128_256.
3993         (<code><mode>3, fpint_logic): New.
3994         (*<code><mode>3): Extend for AVX512.
3995         (avx512flogicsuff): New.
3996         (avx512f_<logic><mode>): Ditto.
3997         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
3998         VF_128_256.
3999         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4000         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4001         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4002         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4003         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4004         (xop_vpermil2<mode>3): Ditto.
4005         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4006         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4007         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4008         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4010 2013-09-25  Tom Tromey  <tromey@redhat.com>
4012         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4013         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4014         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4015         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4016         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4017         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4018         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4019         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4020         (GRAPHITE_HTAB_H): Remove.
4022 2013-09-25  Tom Tromey  <tromey@redhat.com>
4024         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4026 2013-09-25  Tom Tromey  <tromey@redhat.com>
4028         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4030 2013-09-25  Tom Tromey  <tromey@redhat.com>
4032         * config/i386/t-i386 (i386.o): Remove.
4033         (i386-c.o): Use COMPILE and POSTCOMPILE.
4035 2013-09-25  Tom Tromey  <tromey@redhat.com>
4037         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4039 2013-09-25  Tom Tromey  <tromey@redhat.com>
4041         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4042         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4043         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4044         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4045         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4046         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4047         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4048         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4049         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4050         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4051         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4052         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4053         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4054         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4055         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4056         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4057         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4058         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4059         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4060         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4061         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4062         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4063         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4064         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4065         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4066         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4067         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4068         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4069         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4070         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4071         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4072         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4073         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4074         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4075         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4076         (graphite-clast-to-gimple.o, graphite-dependences.o)
4077         (graphite-interchange.o, graphite-poly.o)
4078         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4079         (graphite-optimize-isl.o, tree-vect-loop.o)
4080         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4081         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4082         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4083         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4084         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4085         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4086         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4087         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4088         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4089         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4090         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4091         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4092         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4093         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4094         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4095         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4096         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4097         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4098         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4099         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4100         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4101         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4102         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4103         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4104         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4105         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4106         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4107         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4108         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4109         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4110         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4111         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4112         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4113         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4114         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4115         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4116         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4117         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4118         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4119         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4120         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4121         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4122         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4123         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4124         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4125         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4126         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4127         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4128         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4129         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4130         (gcov-dump.o): Remove.
4131         (default-c.o): Use COMPILE and POSTCOMPILE.
4132         (CFLAGS-gcc.o): New variable.
4133         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4135 2013-09-25  Tom Tromey  <tromey@redhat.com>
4137         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4138         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4139         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4140         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4141         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4142         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4143         (c-family/array-notation-common.o, c-family/stub-objc.o)
4144         (c-family/c-ubsan.o): Remove.
4146 2013-09-25  Tom Tromey  <tromey@redhat.com>
4148         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4150 2013-09-25  Tom Tromey  <tromey@redhat.com>
4152         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4153         to add -DENABLE_SHARED_LIBGCC.
4154         (gcc.o): Don't use subshell.
4156 2013-09-25  Tom Tromey  <tromey@redhat.com>
4158         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4159         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4160         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4161         * configure, config.in: Rebuild.
4163 2013-09-25  Tom Tromey  <tromey@redhat.com>
4165         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4166         (COMPILE, POSTCOMPILE): New variables.
4167         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4168         (DEPFILES): New variable.
4169         Include ".Po" files.
4170         * configure.ac: Add checks for dependency checking.
4171         * configure, aclocal.m4: Regenerate.
4173 2013-09-25  Tom Tromey  <tromey@redhat.com>
4175         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4176         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4178 2013-09-25  Tom Tromey  <tromey@redhat.com>
4180         * Makefile.in (generated_files): Add options.h,
4181         target-hooks-def.h, insn-opinit.h,
4182         common/common-target-hooks-def.h, pass-instances.def,
4183         c-family/c-target-hooks-def.h.
4185 2013-09-25  Jeff Law  <law@redhat.com>
4187         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4188         than foo[foo.length () - 1] to access last member in a vec.
4189         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4191 2013-09-25  Richard Biener  <rguenther@suse.de>
4193         PR middle-end/58521
4194         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4196 2013-09-25  Jan Hubicka  <jh@suse.cz>
4198         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4199         test.
4201 2013-09-25  Marek Polacek  <polacek@redhat.com>
4203         PR sanitizer/58420
4204         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4205         when determining the type name.
4207 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4209         * config/sh/sh.md: Fix formatting.
4211 2013-09-24  Xinliang David Li  <davidxl@google.com>
4213         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4214         max peel iterations parameter.
4215         * param.def: New parameter.
4216         * doc/invoke.texi: Document New parameter.
4218 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4220         * gimple-pretty-print.c: Various whitespace tweaks.
4221         * tree-core.h: Likewise.
4222         * tree-pretty-print.c: Likewise.
4223         * tree-ssa-alias.c: Likewise.
4224         * tree-ssa-copy.c: Likewise.
4225         * tree-ssanames.c: Likewise.
4226         * tree-ssanames.h: Likewise.
4227         * tree-vrp.c: Likewise.
4229 2013-09-24  Alan Modra  <amodra@gmail.com>
4231         PR middle-end/57134
4232         PR middle-end/57586
4233         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4234         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4235         inout operands.
4237 2013-09-24  Richard Biener  <rguenther@suse.de>
4239         PR middle-end/58513
4240         * tree.c (reference_alias_ptr_type): Move ...
4241         * alias.c (reference_alias_ptr_type): ... here and implement
4242         in terms of the new reference_alias_ptr_type_1.
4243         (ref_all_alias_ptr_type_p): New helper.
4244         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4245         use ref_all_alias_ptr_type_p.
4246         (get_deref_alias_set): Add flag_strict_aliasing check here.
4247         (reference_alias_ptr_type_1): New function, split out from ...
4248         (get_alias_set): ... here.
4249         (alias_ptr_types_compatible_p): New function.
4250         * alias.h (reference_alias_ptr_type): Declare.
4251         (alias_ptr_types_compatible_p): Likewise.
4252         * tree.h (reference_alias_ptr_type): Remove.
4253         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4254         to compare MEM_REF alias types.
4256 2013-09-24  Richard Biener  <rguenther@suse.de>
4258         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4260 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4262         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4263         reload helper function arrays into a single array reg_addr.
4264         (reload_fpr_gpr): Likewise.
4265         (reload_gpr_vsx): Likewise.
4266         (reload_vsx_gpr): Likewise.
4267         (struct rs6000_reg_addr): Likewise.
4268         (reg_addr): Likewise.
4269         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4270         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4271         (rs6000_init_hard_regno_mode_ok): Likewise.
4272         (rs6000_secondary_reload_direct_move): Likewise.
4273         (rs6000_secondary_reload): Likewise.
4275         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4276         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4277         power8 changes.  Put wg constraint in alphabetical order.
4279         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4280         for future work to add ISA 2.07 VSX single precision support.
4281         (-mvsx-scalar-double): Change default from -1 to 1, update
4282         documentation comment.
4283         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4284         (-mupper-regs-df): New debug switch to control whether DF values
4285         can go in the traditional Altivec registers.
4286         (-mupper-regs-sf): New debug switch to control whether SF values
4287         can go in the traditional Altivec registers.
4289         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4290         and wy constraints.
4291         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4292         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4293         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4294         (rs6000_debug_legitimate_address_p): Print if we are running
4295         before, during, or after reload.
4296         (rs6000_secondary_reload): Add a comment.
4297         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4299         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4300         constraints.  Update documentation string.
4301         (wd constraint): Likewise.
4302         (wf constraint): Likewise.
4303         (wg constraint): Likewise.
4304         (wn constraint): Likewise.
4305         (ws constraint): Likewise.
4306         (wt constraint): Likewise.
4307         (wx constraint): Likewise.
4308         (wz constraint): Likewise.
4309         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4310         instructions.
4311         (ww constraint): Likewise.
4312         (wy constraint): Likewise.
4313         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4314         in the previous submissions.
4315         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4317 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4319         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4320         (REG_BR_PROB): Say that the probability is stored in an int_list.
4321         * reg-notes.def: Update commentary to mention INT_LIST.
4322         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4323         (INT_LIST): New rtx.
4324         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4325         * rtlanal.c (int_reg_note_p): New function.
4326         (alloc_reg_note): Assert that the note does not have an int argument.
4327         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4328         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4329         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4330         rather than an INSN_LIST.  Handle INT_LIST.
4331         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4332         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4333         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4334         Manipulate them as ints rather than rtxes.
4335         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4336         * regmove.c (copy_src_to_dest): Likewise.
4337         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4339         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4340         into the cases that need it.
4341         * config/arm/arm.c (arm_unwind_emit): Likewise.
4343         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4344         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4345         * loop-doloop.c (add_test, doloop_modify): Likewise.
4346         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4347         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4348         * predict.c (combine_predictions_for_insn): Likewise.
4349         * print-rtl.c (print_rtx): Handle INT_LIST.
4350         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4351         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4352         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4353         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4354         (ix86_split_fp_branch, predict_jump): Likewise.
4355         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4356         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4357         * config/spu/spu.c (ea_load_store_inline): Likewise.
4359         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4360         value of a REG_BR_PROB note.
4361         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4362         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4363         * emit-rtl.c (try_split): Likewise.
4364         * predict.c (br_prob_note_reliable_p): Likewise.
4365         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4366         * reorg.c (mostly_true_jump): Likewise.
4367         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4368         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4369         * config/i386/i386.c (ix86_print_operand): Likewise.
4370         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4371         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4372         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4373         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4374         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4375         * config/sparc/sparc.c (output_cbranch): Likewise.
4376         * config/spu/spu.c (get_branch_target): Likewise.
4377         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4378         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4380 2013-09-23  Jan Hubicka  <jh@suse.cz>
4382         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4383         for ipa-devirt.
4384         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4385         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4386         of external calls
4387         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4388         (gimple_fold_call): Dump inconsistent devirtualizations; add
4389         sanity check for type based devirtualizations.
4390         * ipa-prop.c: Include ipa-utils.h
4391         (ipa_intraprocedural_devirtualization): Add sanity check.
4392         (try_make_edge_direct_virtual_call): Likewise.
4394 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4396         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4397         assignment statements.
4399 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4401         * gimple-pretty-print.c (dump_ssaname_info): New function.
4402         (dump_gimple_phi): Call it.
4403         (pp_gimple_stmt_1): Likewise.
4404         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4405         (range_info_def): Declare.
4406         * tree-pretty-print.c (pp_double_int): New function.
4407         (dump_generic_node): Call it.
4408         * tree-pretty-print.h (pp_double_int): Declare.
4409         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4410         * tree-ssanames.h (range_info_def): New structure.
4411         (value_range_type): Move definition here.
4412         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4413         Declare.
4414         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4415         initialization.
4416         (set_range_info): New function.
4417         (get_range_info): Likewise.
4418         (duplicate_ssa_name_range_info): Likewise.
4419         (duplicate_ssa_name_fn): Check pointer type and call
4420         duplicate_ssa_name_range_info.
4421         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4422         * tree-vrp.c (value_range_type): Remove definition, now in
4423         tree-ssanames.h.
4424         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4425         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4426         (SSA_NAME_RANGE_INFO): New macro.
4428 2013-09-23  Richard Biener  <rguenther@suse.de>
4430         PR tree-optimization/58464
4431         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4432         (phi_trans_add): Change to add conditionally on being not present.
4433         (phi_translate_1): Remove recursion detection here.
4434         (phi_translate): Pre-seed the cache with NULL to catch
4435         recursion here in a more generic way.
4436         (bitmap_find_leader): Adjust comment.
4437         (get_representative_for): Dump value-numbers.
4438         (create_expression_by_pieces): Likewise.
4439         (insert_into_preds_of_block): Likewise.
4441 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4443         PR target/58475
4444         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4445         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4447 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4449         Revert r202780:
4450         2013-09-20  Renlin Li  <renlin.li@arm.com>
4452         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4453         plus_constant.
4454         (aarch64_expand_epilogue): Likewise.
4455         (aarch64_legitimize_reload_address): Likewise.
4457 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4459         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4460         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4462 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4464         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4465         comparison operators when -fno-trapping-math is in effect.
4466         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4467         comparison operators in TFmode and assert that unsupported operators
4468         cannot reach here.
4469         (ia64_print_operand): Likewise.
4471 2013-09-21  Jan Hubicka  <jh@suse.cz>
4473         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4474         (sse_typeless_stores): Enable for core
4475         (sse_load0_by_pxor): Likewise.
4476         (four_jump_limit): Disable for core.
4477         (pad_returns): Likewise.
4478         (avoid_vector_decode): Likewise.
4479         (fuse_cmp_and_branch): Enable for cores.
4480         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4482 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4484         PR middle-end/56791
4485         * config/pa/pa.c (pa_option_override): Disable auto increment and
4486         decrement instructions until reload is completed.
4488         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4489         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4490         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4492 2013-09-20  DJ Delorie  <dj@redhat.com>
4493             Nick Clifton  <nickc@redhat.com>
4495         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4496         (need_to_save): Save bank 0 on interrupts.
4497         (characterize_address): Strip far address wrappers.
4498         (rl78_as_legitimate_address): Likewise.
4499         (transcode_memory_rtx): Likewise.
4500         (rl78_peep_movhi_p): Disable this peephole after devirt.
4501         (rl78_propogate_register_origins): Forget all origins when a
4502         CLOBBER is seen.
4503         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4504         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4505         peephole2's.
4506         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4507         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4508         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4509         relocating.
4510         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4512 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4514         * config/pa/pa.md: In "scc" insn patterns, change output template to
4515         handle const0_rtx in reg_or_0_operand operands.
4517 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4519         PR target/56875
4520         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4521         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4523 2013-09-20  Richard Biener  <rguenther@suse.de>
4525         PR middle-end/58484
4526         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4527         remembering SSA name version and block index.
4528         (new_scev_info_str): Adjust.
4529         (hash_scev_info): Likewise.  Also hash the block index.
4530         (eq_scev_info): Adjust.
4531         (find_var_scev_info): Likewise.
4532         (struct instantiate_cache_entry): Remove.
4533         (struct instantiate_cache_type): Use a htab to map name, block
4534         to chrec.
4535         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4536         (get_instantiated_value_entry): Likewise.
4537         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4538         (instantiate_scev_name): Adjust.
4540 2013-09-20  Jeff Law  <law@redhat.com>
4542         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4544 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4546         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4547         Call aarch64_simd_expand_args to update op[argc].
4549 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4551         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4552         plugin argument.
4554 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4556         * gengtype.c (file_rules): Added rule for *.cc files.
4557         (get_output_file_with_visibility): Give fatal message when no
4558         rules found.
4560 2013-09-20  Renlin Li  <renlin.li@arm.com>
4562         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4563         (aarch64_expand_epilogue): Likewise.
4564         (aarch64_legitimize_reload_address): Likewise.
4566 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4568         PR middle-end/57748
4569         * expr.c (expand_assignment): Remove misalignp code path.
4571 2013-09-20  Marek Polacek  <polacek@redhat.com>
4573         PR sanitizer/58413
4574         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4575         TYPE_PRECISION.  Add asserts.
4577 2013-09-20  Richard Biener  <rguenther@suse.de>
4579         PR tree-optimization/58453
4580         * tree-loop-distribution.c (distribute_loop): Apply the cost
4581         model for -ftree-loop-distribute-patterns, too.
4583 2013-09-20  Richard Biener  <rguenther@suse.de>
4585         PR middle-end/58473
4586         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4587         make type comparison less strict.
4589 2013-09-20  Alan Modra  <amodra@gmail.com>
4591         * configure: Regenerate.
4592         * aclocal.m4: Regenerate.
4594 2013-09-20  Marek Polacek  <polacek@redhat.com>
4596         PR other/58467
4597         * doc/extend.texi: Document that attribute used is meant to be used
4598         on variables with static storage duration.
4600 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4602         PR tree-optimization/58472
4603         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4604         simd_lane_access set inv_p = false.
4605         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4606         the simduid magic VAR_DECL.
4608 2013-09-19  Jan Hubicka  <jh@suse.cz>
4610         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4612 2013-09-17  Jeff Law  <law@redhat.com>
4614         * tree-ssa-dom.c (record_temporary_equivalences): New function
4615         split out of dom_opt_dom_walker::after_dom_children.
4616         (dom_opt_dom_walker::thread_across_edge): Move common code
4617         in here from dom_opt_dom_walker::after_dom_children.
4618         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4620 2013-09-19  Jan Hubicka  <jh@suse.cz>
4622         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4623         (TARGET_GENERIC): Use PROCESOR_GENERIC
4624         (enum processor_type): Unify generic32 and 64.
4625         * i386.md (cpu): Likewise.
4626         * x86-tune.def (use_leave): Enable for generic32.
4627         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4628         * athlon.md: Change generic64 to generic in all occurences.
4629         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4630         (ix86_target_macros_internal): Likewise.
4631         * driver-i386.c (host_detect_local_cpu): Likewise.
4632         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4633         to ..
4634         (generic_memcpy, generic_memset, generic_cost): This one.
4635         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4636         (m_GENERIC32, m_GENERIC64): Remove.
4637         (m_GENERIC): Turn into one flag.
4638         (processor_target): Unify generic tunnings.
4639         (ix86_option_override_internal): Replace generic32/64 by generic.
4640         (ix86_issue_rate): Likewise.
4641         (ix86_adjust_cost): Likewise.
4643 2013-09-19  Jan Hubicka  <jh@suse.cz>
4645         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4646         of speculative flag.
4648 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4650         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4651         GOMP_sections_start, even if !exit_reachable.
4653 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4655         * lra-constraints.c (need_for_all_save_p): Use macro
4656         HARD_REGNO_CALL_PART_CLOBBERED.
4657         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4658         set up pseudo conflict hard regs.
4660 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4662         PR target/58452
4663         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4664         operands.
4666 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4668         PR rtl-optimization/58438
4669         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4670         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4671         from previous subpasses.
4673 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4675         * arm.c (arm_get_frame_offsets): Validate architecture supports
4676         LDRD/STRD before accepting the tuning preference.
4677         (arm_expand_prologue): Likewise.
4678         (arm_expand_epilogue): Likewise.
4680 2013-09-18  Richard Biener  <rguenther@suse.de>
4682         PR tree-optimization/58417
4683         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4684         have chrecs with symbols defined in the loop as operands.
4685         (chrec_fold_multiply): Likewise.
4686         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4687         parameters before folding binary operations.
4688         (struct instantiate_cache_entry_hasher): Remove.
4689         (struct instantiate_cache_type): Use a pointer-map.
4690         (instantiate_cache_type::instantiate_cache_type): New function.
4691         (instantiate_cache_type::get): Likewise.
4692         (instantiate_cache_type::set): Likewise.
4693         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4694         (get_instantiated_value_entry): Likewise.
4695         (global_cache): New global.
4696         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4697         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4698         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4699         (instantiate_scev_name): Adjust.
4700         (instantiate_scev): Construct global instead of local cache.
4701         (resolve_mixers): Likewise.
4703 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4704             Paolo Carlini  <paolo.carlini@oracle.com>
4706         PR c++/58458
4707         * doc/implement-cxx.texi: Fix references to the C++ standards.
4709 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4711         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4712         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4713         simd_lane_access drs, update also DR_ALIGNED_TO.
4715 2013-09-18  Marek Polacek  <polacek@redhat.com>
4717         PR sanitizer/58411
4718         * doc/extend.texi: Document no_sanitize_undefined attribute.
4719         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4720         no_sanitize_undefined attribute.
4722 2013-09-18  Nick Clifton  <nickc@redhat.com>
4724         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4725         (ASM_DECLARE_FUNCTION_NAME): Define.
4727 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4729         * compare-elim.c (find_comparison_dom_walker): New class
4730         (find_comparisons_in_bb): Rename to
4731         find_comparison_dom_walker::before_dom_children
4732         (find_comparisons): Adjust
4733         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4734         adjust.
4735         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4736         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4737         (init_walk_dominator_tree): Remove declaration.
4738         (fini_walk_dominator_tree): Remove declaration.
4739         * fwprop.c (single_def_use_dom_walker): New class
4740         (single_def_use_enter_block): Convert to
4741         single_def_use_dom_walker::before_dom_children.
4742         (single_def_use_leave_block): Convert to
4743         single_def_use_dom_walker::after_dom_children.
4744         (build_single_def_use_links): Adjust.
4745         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4746         class.
4747         (find_candidates_in_block): Convert to
4748         find_candidates_dom_walker::before_dom_children.
4749         (execute_strength_reduction): Adjust.
4750         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4751         (sese_dom_walker): New class.
4752         (sese_dom_walker::sese_dom_walker): New constructor.
4753         (sese_dom_walker::~sese_dom_walker): New destructor.
4754         (build_sese_conditions_before): Convert to
4755         sese_dom_walker::before_dom_children.
4756         (build_sese_conditions_after): Convert to
4757         sese_dom_walker::after_dom_children.
4758         (build_poly_scop): Adjust
4759         * tree-into-ssa.c (rewrite_dom_walker): New class
4760         (rewrite_enter_block): Convert to
4761         rewrite_dom_walker::before_dom_children.
4762         (rewrite_leave_block): Convert to
4763         rewrite_dom_walker::after_dom_children.
4764         (rewrite_update_dom_walker): New class.
4765         (rewrite_update_enter_block): Convert to
4766         rewrite_update_dom_walker::before_dom_children.
4767         (rewrite_update_leave_block): Convert to
4768         rewrite_update_dom_walker::after_dom_children.
4769         (rewrite_blocks, rewrite_into_ssa): Adjust.
4770         (mark_def_dom_walker): New class.
4771         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4772         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4773         (mark_def_sites_blocks): Convert to
4774         mark_def_dom_walker::before_dom_children.
4775         (mark_def_site_blocks): Remove.
4776         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4777         (tree_ssa_dominator_optimize): Adjust.
4778         (dom_thread_across_edge): Convert to method
4779         dom_opt_dom_walker::thread_across_edge.
4780         (dom_opt_enter_block): Convert to member function
4781         dom_opt_dom_walker::before_dom_children.
4782         (dom_opt_leave_block): Convert to member function
4783         dom_opt_dom_walker::after_dom_children.
4784         * tree-ssa-dse.c (dse_dom_walker): New class.
4785         (dse_enter_block): Convert to member function
4786         dse_dom_walker::before_dom_children.
4787         (tree_ssa_dse): Adjust.
4788         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4789         (determine_invariantness_stmt): Convert to method
4790         invariantness_dom_walker::before_dom_children.
4791         (determine_invariantness): Remove
4792         (move_computations_dom_walker): New class.
4793         (move_computations_stmt): Convert to method
4794         move_computations_dom_walker::before_dom_children.
4795         (move_computations, tree_ssa_lim): Adjust.
4796         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4797         (nt_init_block): Make method
4798         notrappping_dom_walker::before_dom_children.
4799         (nt_fini_block): Make
4800         method nontrapping_dom_walker::after_dom_children.
4801         (get_non_trapping): Adjust.
4802         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4803         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4804         (eliminate_leave_block): Make method.
4805         eliminate_dom_walker::after_dom_children.
4806         (eliminate): Adjust
4807         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4808         (strlen_enter_block): Make method
4809         strlen_dom_walker::before_dom_children.
4810         (strlen_leave_block): Make
4811         method strlen_dom_walker::after_dom_children.
4812         (tree_ssa_strlen): Adjust.
4813         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4814         (tree_ssa_uncprop): Adjust.
4815         (uncprop_leave_block): Make method
4816         uncprop_dom_walker::after_dom_children.
4817         (uncprop_leave_block): Make method
4818         uncprop_dom_walker::before_dom_children.
4820 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4822         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4823         branch in same basic block.  Check both src and dest of the move insn.
4825 2013-09-17  Nick Clifton  <nickc@redhat.com>
4827         * config/rl78/rl78-real.md (bf): New pattern.
4828         (bt): New pattern.
4829         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4830         (rl78_print_operand): Do not put a # before a %B.
4831         * config/rl78/rl78.opt: Tweak doc strings.
4833 2013-09-17  DJ Delorie  <dj@redhat.com>
4835         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4836         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4837         (rl78_option_override): Likewise, if -mallregs.
4838         (is_virtual_register): Likewise.
4839         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4840         (REGNO_OK_FOR_BASE_P): Likewise.
4841         * config/rl78/rl78.opt (-mallregs): New.
4843 2013-09-17  Nick Clifton  <nickc@redhat.com>
4845         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4846         For interrupt functions: save all call clobbered registers if the
4847         interrupt handler is not a leaf function.
4848         (rl78_expand_prologue): Always recompute the frame information.
4849         For interrupt functions: only select bank 0 if one of the bank 0
4850         registers is going to be psuhed.
4852 2013-09-17  DJ Delorie  <dj@redhat.com>
4854         * config/rl78/constraints.md: For each W* constraint, rename to C*
4855         and create a W* constraint that checks for an optional ES: prefix
4856         pattern also.
4857         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4858         (es_addr): New.  Used to wrap far addresses.
4859         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4860         (rl78_es_base): New.
4861         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4862         wrapped far addresses.
4863         (rl78_print_operand_1): Unwrap far addresses before processing.
4864         (rl78_lo16): Wrap far addresses in unspecs.
4865         (rl78_es_addr): New.
4866         (rl78_es_base): New.
4867         (insn_ok_now): Check for not-yet-wrapped far addresses.
4868         (transcode_memory_rtx): Properly re-wrap far addresses.
4870 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4872         * config/sparc/t-rtems: Add leon3 multilibs.
4874 2013-09-17  Cong Hou  <congh@google.com>
4876         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4877         when checking the dot production pattern. The type of rhs operand
4878         of multiply is now checked correctly.
4880 2013-09-17  Jeff Law  <law@redhat.com>
4882         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4883         edge implied equivalences into successor phis.
4884         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4885         here from tree-ssa-threadedge.c.
4886         (mark_threaded_blocks): When threading through a joiner, if both
4887         successors of the joiner's clone reach the same block, verify the
4888         PHI arguments are equal.  If not, cancel the jump threading request.
4889         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4890         tree-ssa-threadupdate.c
4891         (thread_across_edge): Don't check PHI argument equality when
4892         threading through joiner block here.
4894 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4896         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4897         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4898         (warn_uninit, warn_uninitialized_vars,
4899         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4900         Move to tree-ssa-uninit.c.
4901         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4902         (has_undefined_value_p): New.  Pass dependant parts of
4903         ssa_undefined_value_p.
4904         (uninit_undefined_value_p): Use has_undefined_value_p.
4905         (warn_uninit, warn_uninitialized_vars,
4906         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4907         Move from tree-ssa.c.
4908         * tree-ssa.h: Adjust prototypes.
4910 2013-09-17  Jan Hubicka  <jh@suse.cz>
4912         PR middle-end/58332
4913         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4914         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4915         FUNCTION_NOT_OPTIMIZED.
4916         * ipa-inline-analysis.c (compute_inline_parameters): Function
4917         not optimized is not inlinable unless it is alwaysinline.
4918         (inline_analyze_function): Force calls in not optimized
4919         function not inlinable.
4921 2013-09-17  Jan Hubicka  <jh@suse.cz>
4923         PR middle-end/58329
4924         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4925         to return NULL.
4926         * ipa.c (function_and_variable_visibility): Likewise.
4927         * ipa-profile.c (ipa_profile): Likewise.
4929 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4931         PR ipa/58398
4932         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4933         attribute, and don't inline the resolver in this case.
4935 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4937         * coverage.c (get_coverage_counts): Add missing newline.
4939 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4941         PR tree-optimization/58088
4942         * fold-const.c (mask_with_trailing_zeros): New function.
4943         (fold_binary_loc): Make sure we don't recurse infinitely
4944         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
4945         Use mask_with_trailing_zeros where appropriate.
4947 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
4949         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
4950         of 'prev' var to get better distance estimation.
4952 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
4954         * tree-inline.h (struct copy_body_data): Add transform_parameter.
4955         * tree-inline.c (is_parameter_of): New predicate.
4956         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
4957         a parameter has been remapped.
4958         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
4959         (optimize_inline_calls): Initialize transform_parameter.
4960         (copy_gimple_seq_and_replace_locals): Likewise.
4961         (tree_function_versioning): Likewise.
4962         (maybe_inline_call_in_expr): Likewise.
4964 2013-09-17  Nick Clifton  <nickc@redhat.com>
4966         * config/msp430/msp430-protos.h: Add prototypes for new functions.
4967         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
4968         interrupt handlers.
4969         (is_attr_func): New function.
4970         (msp430_is_interrupt_func): New function.
4971         (is_naked_func): New function.
4972         (is_reentrant_func): New function.
4973         (is_critical_func): New function.
4974         (msp430_start_function): Add annotations for function attributes.
4975         (msp430_attr): New function.
4976         (msp430_attribute_table): New.
4977         (msp430_function_section): New function.
4978         (TARGET_ASM_FUNCTION_SECTION): Define.
4979         (msp430_builtin): New enum.
4980         (msp430_init_builtins): New function.
4981         (msp430_builtin_devl): New function.
4982         (msp430_expand_builtin): New function.
4983         (TARGET_INIT_BUILTINS): Define.
4984         (TARGET_EXPAND_BUILTINS): Define.
4985         (TARGET_BUILTIN_DECL): Define.
4986         (msp430_expand_prologue): Add support for naked, interrupt,
4987         critical and reentrant functions.
4988         (msp430_expand_epilogue): Likewise.
4989         (msp430_print_operand): Handle 'O' character.
4990         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
4991         NO_TRAMPOLINES.
4992         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
4993         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
4994         (pushm): Use a 'n' rather than an 'i' constraint.
4995         (msp_return): Add generation of the interrupt return instruction.
4996         (disable_interrupts): New pattern.
4997         (enable_interrupts): New pattern.
4998         (push_intr_state): New pattern.
4999         (pop_intr_state): New pattern.
5000         (bic_SR): New pattern.
5001         (bis_SR): New pattern.
5002         * doc/extend.texi: Document MSP430 function attributes and builtin
5003         functions.
5005 2013-09-17  Richard Biener  <rguenther@suse.de>
5007         PR tree-optimization/58432
5008         * tree-loop-distribution.c (tree_loop_distribution): Also
5009         scan PHIs for outside loop uses and seed a partition from them.
5011 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5013         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5014         (restructure_reference): Call backtrace_base_for_ref.
5016 2013-09-17  Alan Modra  <amodra@gmail.com>
5018         PR target/57589
5019         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5020         patch.
5022 2013-09-16  DJ Delorie  <dj@redhat.com>
5024         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5025         vregs location for RL78/G10.
5026         (rl78_expand_prologue): Avoid SEL on G10.
5027         (rl78_expand_epilogue): Likewise.
5028         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5029         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5030         __RL78_G10__ when appropriate.
5031         (ASM_SPEC): Pass -mg10 along to the assembler.
5032         * config/rl78/rl78.md (sel_rb): Disable for G10.
5033         * config/rl78/rl78.opt: Add -mg10 option.
5034         * config/rl78/t-rl78: Add -mg10 multilib.
5036 2013-09-16  Xinliang David Li  <davidxl@google.com>
5038         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5039         * omp-low.c (omp_max_vf): Ditto.
5040         (expand_omp_simd): Ditto.
5041         * tree-vectorizer.c (vectorize_loops): Ditto.
5042         (gate_vect_slp): Ditto.
5043         (gate_increase_alignment): Ditto.
5044         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5045         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5046         (gate_tree_vectorize): Name change.
5047         (tree_vectorize): Ditto.
5048         (pass_vectorize::gate): Call new function.
5049         (pass_vectorize::execute): Ditto.
5050         * opts.c: O3 default setting change.
5051         (finish_options): Check new flag.
5052         * doc/invoke.texi: Document new flags.
5053         * common.opt: New flags.
5055 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5057         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5058         (Condition Code): Don't reference it.
5059         * doc/tm.texi: Regenerate.
5061 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5063         PR middle-end/58418
5064         * lra-constraints.c (undo_optional_reloads): Consider all optional
5065         reload even if it did not get a hard reg.
5067 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5069         * dumpfile.c (dump_loc): Remove newline emission.
5070         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5071         emission to dump_printf_loc calls where missing.
5072         (vect_mark_for_runtime_alias_test): Ditto.
5073         (vect_analyze_data_ref_dependence): Ditto.
5074         (vect_analyze_data_ref_dependences): Ditto.
5075         (vect_slp_analyze_data_ref_dependence): Ditto.
5076         (vect_slp_analyze_data_ref_dependences): Ditto.
5077         (vect_compute_data_ref_alignment): Ditto.
5078         (vect_update_misalignment_for_peel): Ditto.
5079         (vect_verify_datarefs_alignment): Ditto.
5080         (vector_alignment_reachable_p): Ditto.
5081         (vect_get_data_access_cost): Ditto.
5082         (vect_enhance_data_refs_alignment): Ditto.
5083         (vect_find_same_alignment_drs): Ditto.
5084         (vect_analyze_data_refs_alignment): Ditto.
5085         (vect_analyze_group_access): Ditto.
5086         (vect_analyze_data_ref_access): Ditto.
5087         (vect_analyze_data_ref_accesses): Ditto.
5088         (vect_prune_runtime_alias_test_list): Ditto.
5089         (vect_analyze_data_refs): Ditto.
5090         (vect_create_addr_base_for_vector_ref): Ditto.
5091         (vect_create_data_ref_ptr): Ditto.
5092         (vect_grouped_store_supported): Ditto.
5093         (vect_grouped_load_supported): Ditto.
5094         * value-prof.c (check_counter): Ditto.
5095         (check_ic_target): Ditto.
5096         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5097         (vect_recog_widen_mult_pattern): Ditto.
5098         (vect_recog_widen_sum_pattern): Ditto.
5099         (vect_recog_over_widening_pattern): Ditto.
5100         (vect_recog_widen_shift_pattern): Ditto.
5101         (vect_recog_rotate_pattern): Ditto.
5102         (vect_recog_vector_vector_shift_pattern): Ditto.
5103         (vect_recog_divmod_pattern): Ditto.
5104         (vect_recog_mixed_size_cond_pattern): Ditto.
5105         (vect_recog_bool_pattern): Ditto.
5106         (vect_pattern_recog_1): Ditto.
5107         (vect_pattern_recog): Ditto.
5108         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5109         (vect_is_simple_iv_evolution): Ditto.
5110         (vect_analyze_scalar_cycles_1): Ditto.
5111         (vect_get_loop_niters): Ditto.
5112         (vect_analyze_loop_1): Ditto.
5113         (vect_analyze_loop_form): Ditto.
5114         (vect_analyze_loop_operations): Ditto.
5115         (vect_analyze_loop_2): Ditto.
5116         (vect_analyze_loop): Ditto.
5117         (report_vect_op): Ditto.
5118         (vect_is_slp_reduction): Ditto.
5119         (vect_is_simple_reduction_1): Ditto.
5120         (vect_get_known_peeling_cost): Ditto.
5121         (vect_estimate_min_profitable_iters): Ditto.
5122         (vect_model_reduction_cost): Ditto.
5123         (vect_model_induction_cost): Ditto.
5124         (get_initial_def_for_induction): Ditto.
5125         (vect_create_epilog_for_reduction): Ditto.
5126         (vectorizable_reduction): Ditto.
5127         (vectorizable_induction): Ditto.
5128         (vectorizable_live_operation): Ditto.
5129         (vect_loop_kill_debug_uses): Ditto.
5130         (vect_transform_loop): Ditto.
5131         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5132         (vect_stmt_relevant_p): Ditto.
5133         (process_use): Ditto.
5134         (vect_mark_stmts_to_be_vectorized): Ditto.
5135         (vect_model_simple_cost): Ditto.
5136         (vect_model_promotion_demotion_cost): Ditto.
5137         (vect_model_store_cost): Ditto.
5138         (vect_get_store_cost): Ditto.
5139         (vect_model_load_cost): Ditto.
5140         (vect_get_load_cost): Ditto.
5141         (vect_init_vector_1): Ditto.
5142         (vect_get_vec_def_for_operand): Ditto.
5143         (vect_finish_stmt_generation): Ditto.
5144         (vectorizable_call): Ditto.
5145         (vectorizable_conversion): Ditto.
5146         (vectorizable_assignment): Ditto.
5147         (vectorizable_shift): Ditto.
5148         (vectorizable_operation): Ditto.
5149         (vectorizable_store): Ditto.
5150         (vectorizable_load): Ditto.
5151         (vectorizable_condition): Ditto.
5152         (vect_analyze_stmt): Ditto.
5153         (vect_transform_stmt): Ditto.
5154         (vect_is_simple_use): Ditto.
5155         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5156         (vect_can_advance_ivs_p): Ditto.
5157         (vect_update_ivs_after_vectorizer): Ditto.
5158         (vect_do_peeling_for_loop_bound): Ditto.
5159         (vect_gen_niters_for_prolog_loop): Ditto.
5160         (vect_update_inits_of_drs): Ditto.
5161         (vect_create_cond_for_alias_checks): Ditto.
5162         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5163         (vect_build_slp_tree_1): Ditto.
5164         (vect_supported_load_permutation_p): Ditto.
5165         (vect_analyze_slp_instance): Ditto.
5166         (vect_analyze_slp): Ditto.
5167         (vect_make_slp_decision): Ditto.
5168         (vect_detect_hybrid_slp): Ditto.
5169         (vect_bb_vectorization_profitable_p): Ditto.
5170         (vect_slp_analyze_bb_1): Ditto.
5171         (vect_update_slp_costs_according_to_vf): Ditto.
5172         (vect_get_mask_element): Ditto.
5173         (vect_transform_slp_perm_load): Ditto.
5174         (vect_schedule_slp_instance): Ditto.
5175         (vect_schedule_slp): Ditto.
5176         (vect_slp_transform_bb): Ditto.
5177         * profile.c (read_profile_edge_counts): Ditto.
5178         (compute_branch_probabilities): Ditto.
5179         * coverage.c (get_coverage_counts): Ditto.
5181 2013-09-16  Diego Novillo  <dnovillo@google.com>
5183         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5185 2013-09-16  Jan Hubicka  <jh@suse.cz>
5187         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5188         abstract functions; for static functions check the presence of body.
5190 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5192         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5193         * config/aarch64/aarch64-simd.md
5194         (aarch64_mla_elt<mode>): New.
5195         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5196         (aarch64_mls_elt<mode>): Likewise.
5197         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5198         (aarch64_fma4_elt<mode>): Likewise.
5199         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5200         (aarch64_fma4_elt_to_128v2df): Likewise.
5201         (aarch64_fma4_elt_to_64df): Likewise.
5202         (fnma<mode>4): Likewise.
5203         (aarch64_fnma4_elt<mode>): Likewise.
5204         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5205         (aarch64_fnma4_elt_to_128v2df): Likewise.
5206         (aarch64_fnma4_elt_to_64df): Likewise.
5207         * config/aarch64/iterators.md (VDQSF): New.
5208         * config/aarch64/arm_neon.h
5209         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5210         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5212 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5214         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5215         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5216         (aarch64_mul3_elt_to_128df): Likewise.
5217         (aarch64_mul3_elt_to_64v2df): Likewise.
5218         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5219         (VMUL): New.
5220         (VMUL_CHANGE_NLANES) Likewise.
5221         (h_con): Likewise.
5222         (f): Likewise.
5223         * config/aarch64/arm_neon.h
5224         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5226 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5228         * config/aarch64/arm_neon.h
5229         (vcvtx_high_f32_f64): Fix parameters.
5231 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5232             Uros Bizjak  <ubizjak@gmail.com>
5234         * config/alpha.c: Include tree-ssanames.h.
5236 2013-09-16  Richard Biener  <rguenther@suse.de>
5238         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5239         (dot_rdg_1): Handle control_dd.
5240         (create_edge_for_control_dependence): New function.
5241         (create_rdg_edges): Add control dependences if asked for.
5242         (build_rdg): Likewise.
5243         (generate_loops_for_partition): If there are not necessary
5244         control stmts remove all their dependencies.
5245         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5246         (distribute_loop): Pass on control dependences.
5247         (tree_loop_distribution): Compute control dependences and remove
5248         restriction on number of loop nodes.
5250 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5252         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5253         for internal calls.
5255 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5257         * cse.c (try_const_anchors): Punt on CC modes.
5259 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5261         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5263 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5265         PR target/58382
5266         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5267         calls to word_mode.
5269 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5271         PR target/48094
5272         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5273         seen.
5274         (darwin_objc1_section): Likewise.
5275         (darwin_file_end): Emit Image Info section when required.
5277 2013-09-14  Jan Hubicka  <jh@suse.cz>
5279         * tree-into-ssa.c (gate_into_ssa): New.
5280         (pass_data_build_ssa): Use it.
5281         * cgraph.h (expand_thunk): Update prototype.
5282         * cgraphunit.c (analyze_function): Expand thunks early.
5283         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5284         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5285         set lowered flag; do not add new function.
5286         (assemble_thunks_and_aliases): Update.
5287         * tree-ssa.c (gate_init_datastructures): New gate.
5288         (pass_data_init_datastructures): Use it.
5290 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5292         PR target/58269
5293         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5294         xmm register set described in the psABI.
5296 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5298         * dwarf2out.c (should_emit_struct_debug): Add check
5299         for type_decl variable is not NULL.
5301 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5303         * config.gcc: Use new winnt-c.c target hooks
5304         * config/t-winnt: New file
5305         * config/winnt-c.c: New file
5306         * doc/tm.texi.in: Document new hook
5307         * doc/tm.texi: Regenerated
5309 2013-09-13  Jan Hubicka  <jh@suse.cz>
5311         PR middle-end/58094
5312         * ipa-inline.c (check_callers): New function.
5313         (check_caller_edge): Remove.
5314         (want_inline_function_to_all_callers_p): Also permit alises that are
5315         called dirrectly.
5316         (inline_to_all_callers): Terminate the walk when devirtualization
5317         introduce new calls.
5319 2013-09-13  Jan Hubicka  <jh@suse.cz>
5321         * ipa-inline-analysis.c (struct growth_data): Add node.
5322         (do_estimate_growth_1): Fix detection of recursion.
5324 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5326         PR tree-optimization/58392
5327         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5328         to avoid shadowing of outer loop variable.  If
5329         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5330         replace_by_duplicate_decl simduid of loops that have it set and
5331         set dest_cfun->has_simduid_loops and/or
5332         dest_cfun->has_force_vect_loops.
5333         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5334         instead of maybe_lookup_decl.
5335         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5336         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5337         call.  Copy over force_vect and copy and remap simduid.  Set
5338         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5339         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5340         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5341         caller.  Don't set cfun->has_simduid_loops and/or
5342         cfun->has_force_vect_loops here.
5343         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5344         caller.
5345         (expand_call_inline, tree_function_versioning): Adjust copy_body
5346         callers.
5348 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5350         PR bootstrap/58388
5351         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5352         the assert if the edge was a speculative one.
5354 2013-09-13  Richard Biener  <rguenther@suse.de>
5356         * tree-data-ref.h (known_dependences_p): Move here ...
5357         * tree-loop-distribution.c (known_dependences_p): ... from here.
5358         (dump_rdg_component, debug_rdg_component): Remove.
5359         (dump_rdg): Adjust.
5360         (generate_loops_for_partition): Use gimple_uid instead of
5361         relying on matching stmt visit order.
5362         (rdg_build_partitions): Take starting stmt vector.
5363         (ldist_gen): Merge into ...
5364         (distribute_loop): ... this function.  Do not compute starting
5365         vertices vector.
5366         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5368 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5370         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5371         Set type attribute correctly. Set predicable_short_it attribute.
5372         (cmpsi_shiftsi): Remove %? from output template.
5374 2013-09-13  Richard Biener  <rguenther@suse.de>
5376         * tree-loop-distribution.c (struct rdg_component,
5377         rdg_defs_used_in_other_loops_p, free_rdg_components,
5378         rdg_build_components): Remove.
5379         (stmts_from_loop): Do not record virtual PHIs.
5380         (generate_loops_for_partition): Skip virtual PHIs.
5381         (build_rdg_partition_for_component): Rename to ...
5382         (build_rdg_partition_for_vertex): ... this and adjust.
5383         (rdg_build_partitions): Take a vector of starting vertices
5384         instead of components.  Remove unnecessary leftover handling.
5385         (ldist_gen): Do not build components or record other stores.
5386         (distribute_loop): Do not distribute loops containing stmts
5387         with side-effects.
5389 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5391         PR target/58314
5392         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5394 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5396         * config.gcc: Separate cases for mingw and cygwin targets,
5397         and add 64-bit cygwin target case.
5399         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5400         dll-export inline-functions.
5401         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5404 2013-09-13  Jeff Law  <law@redhat.com>
5406         PR middle-end/58387
5407         Revert:
5408         2013-09-06  Jeff Law  <law@redhat.com>
5410         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5411         edge implied equivalences into successor phis.
5413 2013-09-12  DJ Delorie  <dj@redhat.com>
5415         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5416         separators.
5418 2013-09-12  Brooks Moses  <bmoses@google.com>
5420         PR driver/42955
5421         * Makefile.in: Do not install driver binaries in $(target)/bin.
5423 2013-09-12  DJ Delorie  <dj@redhat.com>
5425         * config/rl78/rl78.opt (mrelax): New.
5426         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5427         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5429         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5430         between SP and FP.
5431         (rl78_expand_epilogue): Likewise.
5433 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5435         PR middle-end/58335
5436         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5437         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5438         pointer elimination with using remove_reg_equal_offset_note.
5440 2013-09-12  DJ Delorie  <dj@redhat.com>
5442         * config/msp430/: New port.
5443         * config.gcc (msp430): Added.
5444         * doc/invoke.texi: Document MSP430 options.
5445         * doc/install.texi: Document msp430-elf
5446         * doc/md.texi: Document msp430-elf
5447         * doc/contrib.texi: Document msp430-elf
5449         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5450         PSImode.
5452 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5454         PR ipa/58389
5455         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5456         reference descriptor is NULL.
5457         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5458         edge to NULL.
5460 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5462         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5463         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5464         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5465         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5466         make_temp_ssa_name): move to tree-ssanames.h
5467         * tree-ssa-alias.h: Move prototype.
5468         * tree-ssa.h: Include tree-ssanames.h.
5469         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5470         * tree-ssanames.h: New.  Move items from tree-flow*.h
5471         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5473 2013-09-12  Richard Biener  <rguenther@suse.de>
5475         PR tree-optimization/58404
5476         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5477         propagate non-invariant addresses into dereferences wrapped
5478         in component references.
5480 2013-09-12  Richard Biener  <rguenther@suse.de>
5482         PR tree-optimization/58402
5483         * passes.def: Move pass_late_warn_uninitialized later.
5485 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5487         * tree-ssa.h: New.  Move content from tree-flow.h and
5488         tree-flow-inline.h.
5489         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5490         Move prototypes belonging to tree-ssa.c.
5491         * tree-flow-inline.h (redirect_edge_var_map_def,
5492         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5493         tree-ssa.h.
5494         * gimple.h: Adjust prototypes.
5495         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5496         to...
5497         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5498         * tree.h: Move prototype to tree-ssa.h.
5499         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5500         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5501         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5502         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5503         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5504         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5505         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5506         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5507         graphite-blocking.c, graphite-clast-to-gimple.c,
5508         graphite-dependences.c, graphite-interchange.c,
5509         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5510         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5511         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5512         ipa-reference.c, ipa-split.c, ipa-utils.c,
5513         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5514         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5515         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5516         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5517         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5518         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5519         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5520         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5521         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5522         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5523         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5524         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5525         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5526         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5527         value-prof.c, var-tracking.c,
5528         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5530 2013-09-12  Richard Biener  <rguenther@suse.de>
5532         PR tree-optimization/58396
5533         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5534         (build_rdg): Take a loop-nest parameter, fix memleaks.
5535         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5537 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5539         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5540         for SLM.
5542 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5544         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5545         The FMA instruction names should have a 'v' prefix.
5547 2013-09-12  Richard Biener  <rguenther@suse.de>
5549         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5550         (dot_rdg): Use popen instead of system in optional code.
5551         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5552         (already_processed_vertex_p): Adjust.
5553         (has_anti_or_output_dependence, predecessor_has_mem_write,
5554         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5555         rdg_flag_uses): Remove.
5556         (rdg_flag_vertex): Simplify.
5557         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5558         remove recursion.
5559         (build_rdg_partition_for_component): Process the first vertex
5560         of a component only.
5561         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5563 2013-09-12  Alan Modra  <amodra@gmail.com>
5565         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5567 2013-09-11  DJ Delorie  <dj@redhat.com>
5568             Nick Clifton  <nickc@redhat.com>
5570         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5571         (rl78_stack_based_mem): New.
5572         * config/rl78/constraints.md (Iv08): New.
5573         (Iv16): New.
5574         (Iv24): New.
5575         (Is09): New.
5576         (Is17): New.
5577         (Is25): New.
5578         (ISsi): New.
5579         (IShi): New.
5580         (ISqi): New.
5581         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5582         (movhi): Likewise.
5583         (movsi): Change from expand to insn-and-split.
5584         (ashrsi3): Clobber AX.
5585         (lshrsi3): New.
5586         (ashlsi3): New.
5587         (cbranchsi4): New.
5588         * config/rl78/rl78.md (CC_REG): Fix.
5589         (addsi3): Allow memory and immediate operands.
5590         (addsi3_internal): Split into...
5591         (addsi3_internal_virt): ...new, and ...
5592         (addsi3_internal_real): ...new.
5593         (subsi): New.
5594         (subsi3_internal_virt): New.
5595         (subsi3_internal_real): New.
5596         (mulsi3): Add memory operand.
5597         (mulsi3_rl78): Likewise.
5598         (mulsi3_g13): Likewise.
5599         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5600         (cbranchqi4_real): Add more constraint options.
5601         (cbranchhi4_real): Expand pattern.
5602         (cbranchhi4_real_signed): New.
5603         (cbranchhi4_real_inverted): New.
5604         (cbranchsi4_real_lt): New.
5605         (cbranchsi4_real_ge): New.
5606         (cbranchsi4_real_signed): New.
5607         (cbranchsi4_real): New.
5608         (peephole2): New.
5609         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5610         constant shifts.
5611         (lshrsi3_virt): Likewise.
5612         (ashlsi3_virt): Likewise.
5613         (cbranchqi4_virt_signed): New.
5614         (cbranchhi4_virt_signed): New.
5615         (cbranchsi4_virt): New.
5616         * config/rl78/rl78.c: Whitespace fixes throughout.
5617         (move_elim_pass): New.
5618         (pass_data_rl78_move_elim): New.
5619         (pass_rl78_move_elim): New.
5620         (make_pass_rl78_move_elim): New.
5621         (rl78_devirt_info): Run devirt earlier.
5622         (rl78_move_elim_info): New.
5623         (rl78_asm_file_start): Register it.
5624         (rl78_split_movsi): New.
5625         (rl78_as_legitimate_address): Allow virtual base registers when
5626         appropriate.
5627         (rl78_addr_space_convert): Remove spurious debug stuff.
5628         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5629         (rl78_print_operand): More cases for not printing '#'.
5630         (rl78_expand_compare): Remove most of the logic.
5631         (content_memory): New.
5632         (clear_content_memory): New.
5633         (get_content_index): New.
5634         (get_content_name): New.
5635         (display_content_memory): New.
5636         (update_content): New.
5637         (record_content): New.
5638         (already_contains): New.
5639         (insn_ok_now): Re-recog insns with virtual registers.
5640         (add_postponed_content_update): New.
5641         (process_postponed_content_update): New.
5642         (gen_and_emit_move): New.
5643         (transcode_memory_rtx): Record new location content.
5644         Use gen_and_emit_move.
5645         (force_into_acc): New.
5646         (move_to_acc): Use gen_and_emit_move.
5647         (move_from_acc): Likewise.
5648         (move_acc_to_reg): Likewise.
5649         (move_to_x): Likewise.
5650         (move_to_hl): Likewise.
5651         (move_to_de): Likewise.
5652         (rl78_alloc_physical_registers_op1): Record location content.
5653         (has_constraint): New.
5654         (rl78_alloc_physical_registers_op2): Record location content.
5655         Optimize use of HL.
5656         (rl78_alloc_physical_registers_ro1): Likewise.
5657         (rl78_alloc_physical_registers_cmp): Likewise.
5658         (rl78_alloc_physical_registers_umul): Likewise.
5659         (rl78_alloc_address_registers_macax): New.
5660         (rl78_alloc_physical_registers): Initialize and set location
5661         content memory as needed.
5662         (rl78_reorg): Make sure split2 is called.
5663         (rl78_rtx_costs): New.
5665 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5667         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5668         for (not (neg ...)) and (neg (not ...)) cases.
5670 2013-09-11  Richard Biener  <rguenther@suse.de>
5672         PR middle-end/58377
5673         * passes.def: Split critical edges before late uninit warning passes.
5674         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5676 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5678         PR tree-optimization/58385
5679         * fold-const.c (build_range_check): If both low and high are NULL,
5680         use omit_one_operand_loc to preserve exp side-effects.
5682 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5684         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5686 2013-09-11  Richard Biener  <rguenther@suse.de>
5688         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5689         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5690         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5691         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5692         stmts_from_loop, known_dependences_p, build_empty_rdg,
5693         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5694         * tree-loop-distribution.c: ... here.
5695         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5696         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5697         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5698         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5699         * tree-loop-distribution.c: ... here.
5700         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5701         (struct partition_s): Add loops member.
5702         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5703         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5704         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5706 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5707             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5708             Sergey Lega  <sergey.s.lega@intel.com>
5709             Anna Tikhonova  <anna.tikhonova@intel.com>
5710             Ilya Tocar  <ilya.tocar@intel.com>
5711             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5712             Ilya Verbin  <ilya.verbin@intel.com>
5713             Kirill Yukhin  <kirill.yukhin@intel.com>
5714             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5716         * config/i386/constraints.md (k): New.
5717         (Yk): Ditto.
5718         * config/i386/i386.c (const regclass_map): Add new mask registers.
5719         (dbx_register_map): Ditto.
5720         (dbx64_register_map): Ditto.
5721         (svr4_dbx_register_map): Ditto.
5722         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5723         disabled.
5724         (ix86_preferred_reload_class): Disable constants for mask registers.
5725         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5726         (ix86_hard_regno_mode_ok): Support new mask registers.
5727         (x86_order_regs_for_local_alloc): Ditto.
5728         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5729         (FIXED_REGISTERS): Add new mask registers.
5730         (CALL_USED_REGISTERS): Ditto.
5731         (REG_ALLOC_ORDER): Ditto.
5732         (VALID_MASK_REG_MODE): New.
5733         (FIRST_MASK_REG): Ditto.
5734         (LAST_MASK_REG): Ditto.
5735         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5736         (MAYBE_MASK_CLASS_P): New.
5737         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5738         (REG_CLASS_CONTENTS): Ditto.
5739         (MASK_REGNO_P): New.
5740         (ANY_MASK_REG_P): Ditto.
5741         (HI_REGISTER_NAMES): Add new mask registers.
5742         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5743         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5744         mask registers.
5745         (attribute "type"): Add mskmov, msklog.
5746         (attribute "length_immediate"): Support them.
5747         (attribute "memory"): Ditto.
5748         (attribute "prefix_0f"): Ditto.
5749         (*movhi_internal): Support new mask registers.
5750         (*movqi_internal): Ditto.
5751         (define_split): Split out clobber pattern is a logic
5752         insn on mask registers.
5753         (*k<logic><mode>): New.
5754         (*andhi_1): Extend to support mask regs.
5755         (*andqi_1): Extend to support mask regs.
5756         (kandn<mode>): New.
5757         (define_split): Split and-not to and and not if operands
5758         are not mask regs.
5759         (*<code><mode>_1): Separate HI mode to new pattern...
5760         (*<code>hi_1): This.
5761         (*<code>qi_1): Extend to support mask regs.
5762         (kxnor<mode>): New.
5763         (kortestzhi): Ditto.
5764         (kortestchi): Ditto.
5765         (kunpckhi): Ditto.
5766         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5767         (*one_cmplhi2_1): ...Here, now with mask registers support.
5768         (*one_cmplqi2_1): Support new mask registers.
5769         (HI/QImode arithmetics splitter): Don't split if mask registers
5770         are used.
5771         (HI/QImode not splitter): Ditto.
5772         * config/i386/predicated.md (mask_reg_operand): New.
5773         (general_reg_operand): Ditto.
5775 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5777         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5778         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5780 2013-09-10  Jeff Law  <law@redhat.com>
5782         PR tree-optimization/58380
5783         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5784         of threading through a buried loop header.
5786         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5787         return value for single successor case.
5789 2013-09-10  Jan Hubicka  <jh@suse.cz>
5791         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5793 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5795         PR target/58361
5796         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5797         support conditional execution.
5798         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5800 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5802         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5803         constraint pass.
5804         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5805         Check destination too to check move insn.
5806         (undo_optional_reloads): Add check that the original peudo did not
5807         changed its allocation and the optional reload was inherited on last
5808         inheritance pass.  Break loop after deciding to keep optional reload.
5809         (lra_undo_inheritance): Add check that inherited pseudo still in
5810         memory.
5812 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5814         * config/aarch64/aarch64.md (generic_sched): New.
5815         * config/aarch64/aarch64-generic.md (load): Make conditional
5816         on generic_sched attribute.
5817         (nonload): Likewise.
5819 2013-09-10  Jan Hubicka  <jh@suse.cz>
5821         * lto-cgraph.c: Include ipa-utils.h.
5822         (compute_ltrans_boundary): Also add possible targets into the boundary.
5824 2013-09-10  Jan Hubicka  <jh@suse.cz>
5826         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5827         VAR_DECL of vtable rather than full expression.
5829 2013-09-10  Jan Hubicka  <jh@suse.cz>
5830             Paolo Carlini  <paolo.carlini@oracle.com>
5832         * cgraphunit.c (analyze_functions): Save input_location, set it
5833         to UNKNOWN_LOCATION and restore it at the end.
5835 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5837         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5838         represented by a thunk.
5840 2013-09-10  Jeff Law  <law@redhat.com>
5842         PR tree-optimization/58343
5843         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5844         jump threading requests through loop headers buried in the middle
5845         of a jump threading path.
5847         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5848         in return value/type.
5850 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5852         PR rtl-optimization/58365
5853         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5854         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5855         it differs.
5857 2013-09-10  Richard Biener  <rguenther@suse.de>
5859         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5860         * tree-data-ref.c (create_rdg_vertices): Collect all data
5861         references, signal failure to the caller, use data-ref API.
5862         (build_rdg): Compute data references only once.  Maintain lifetime
5863         of data references and data dependences from within RDG.
5864         (free_rdg): Free dependence relations.
5865         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5866         inventing extra dependences.
5867         (distribute_loop): Update for RDG API changes.
5869 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5871         * doc/invoke.texi (fms-extensions): Document changed
5872         behavior for ms-abi targets.
5873         * config/i386/i386.c (ix86_option_override_internal):
5874         Set default value of option -fms-extension for ms-abi targets.
5876 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5878         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5880 2013-09-10  Alan Modra  <amodra@gmail.com>
5882         PR target/58330
5883         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5885 2013-09-10  Alan Modra  <amodra@gmail.com>
5887         * config/rs6000/predicates.md (add_cint_operand): New.
5888         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5889         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5890         using add_cint_operand.
5891         (largetoc_high_plus_aix): Likewise.
5893 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5895         PR tree-optimization/58364
5896         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5897         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5898         the current range can't be an unconditional true or false.
5900 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5902         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5904 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5906         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5908 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5910         PR c++/43452
5911         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5913 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5915         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5916         NO_REGS for immediate that can't be moved directly into FP_REGS.
5918 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5920         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5921         comparison with negated operand.
5922         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5923         RTL form.
5925 2013-09-09  Richard Biener  <rguenther@suse.de>
5927         PR middle-end/58326
5928         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5929         of a subloop record all its block as affecting loop-closed SSA form.
5931 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5933         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5934         of an rtx/bitpos pair.
5935         (store_fixed_bit_field): Update accordingly.
5937 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5939         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5940         GEN_INT.
5941         * builtins.c (expand_errno_check): Likewise.
5942         * dwarf2cfi.c (init_return_column_size): Likewise.
5943         * except.c (sjlj_mark_call_sites): Likewise.
5944         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
5945         * lra-constraints.c (emit_inc): Likewise.
5946         * ree.c (combine_set_extension): Likewise.
5947         * regmove.c (fixup_match_2): Likewise.
5948         * reload1.c (inc_for_reload): Likewise.
5950 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5952         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
5953         (force_to_mode, simplify_shift_const_1, simplify_comparison):
5954         Use gen_int_mode with the mode of the associated simplify_* call.
5955         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
5956         * expmed.c (expand_shift_1): Likewise.
5957         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5958         * loop-iv.c (iv_number_of_iterations): Likewise.
5959         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5960         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
5962 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5964         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
5965         of the associated expand_* call.
5966         (asan_emit_stack_protection): Likewise.
5967         * builtins.c (round_trampoline_addr): Likewise.
5968         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
5969         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
5970         (emit_store_flag): Likewise.
5971         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
5972         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
5973         Likewise.
5974         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5975         * ifcvt.c (noce_try_store_flag_constants): Likewise.
5976         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5977         * modulo-sched.c (generate_prolog_epilog): Likewise.
5978         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
5979         (expand_ctz, expand_ffs, expand_unop): Likewise.
5981 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5983         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
5984         of the associated gen_rtx_* call.
5985         * caller-save.c (init_caller_save): Likewise.
5986         * combine.c (find_split_point, make_extraction): Likewise.
5987         (make_compound_operation): Likewise.
5988         * dwarf2out.c (mem_loc_descriptor): Likewise.
5989         * explow.c (plus_constant, probe_stack_range): Likewise.
5990         * expmed.c (expand_mult_const): Likewise.
5991         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
5992         * reload1.c (init_reload): Likewise.
5993         * valtrack.c (cleanup_auto_inc_dec): Likewise.
5994         * var-tracking.c (adjust_mems): Likewise.
5995         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
5996         rather than gen_rtx_fmt_ee.
5998 2013-09-09  Jan Hubicka  <jh@suse.cz>
6000         PR middle-end/58294
6001         * value-prof.c (gimple_ic): Copy also abnormal edges.
6003 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6005         * asan.c (asan_shadow_cst): Use gen_int_mode.
6007 2013-09-08  Jan Hubicka  <jh@suse.cz>
6009         * ipa-profile.c: Add toplevel comment.
6010         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6011         (contains_hot_call_p): New function.
6012         (ipa_propagate_frequency): Set frequencies based on counts when
6013         profile is read.
6014         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6015         profile; do not tamper with profile after inlining if it is read.
6017 2013-09-08  Jan Hubicka  <jh@suse.cz>
6019         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6020         speculative edges.
6022 2013-09-08  Jan Hubicka  <jh@suse.cz>
6024         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6025         summary generation.
6027 2013-09-08  Jeff Law  <law@redhat.com>
6029         PR bootstrap/58340
6030         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6031         of 'found'.
6033 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6035         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6037 2013-09-08  Jan Hubicka  <jh@suse.cz>
6039         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6041 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6043         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6044         for non-debug insns.
6045         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6046         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6047         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6048         accordingly.
6050 2013-09-08  Jan Hubicka  <jh@suse.cz>
6052         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6053         targets and devirtualize to BUILT_IN_UNREACHABLE.
6054         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6055         * ipa.c (walk_polymorphic_call_targets): New function.
6056         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6057         functions; use the new timevar.
6058         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6059         was removed from the program.
6060         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6061         not consider it in the walk when its vtable is dead.
6062         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6063         record_binfo.
6064         (devirt_variable_node_removal_hook): New function.
6065         (possible_polymorphic_call_targets): Also register
6066         devirt_variable_node_removal_hook.
6067         (ipa_devirt): Do not do non-speculative devirtualization.
6068         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6070 2013-09-08  Jan Hubicka  <jh@suse.cz>
6072         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6073         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6074         varpool_remove_variable_insertion_hook): Declare.
6075         * varpool.c (varpool_node_hook_list): New structure.
6076         (first_varpool_node_removal_hook,
6077         first_varpool_variable_insertion_hook): New variables.
6078         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6079         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6080         varpool_remove_variable_insertion_hook,
6081         varpool_call_variable_insertion_hooks): New functions.
6082         (varpool_remove_node): Use it.
6084 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6086         PR c++/54941
6087         * diagnostic.c (diagnostic_build_prefix): When s.file is
6088         "<built-in>" don't output line and column numbers.
6090 2013-09-06  Jan Hubicka  <jh@suse.cz>
6092         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6093         * lto-streamer-out.c: Stream thunks, too.
6094         * lto-streamer-in.c (input_function): Pop cfun here
6095         (lto_read_body): Instead of here.
6097 2013-09-06  Caroline Tice  <cmtice@google.com>
6099         * doc/install.texi: Add documentation for the --enable-vtable-verify
6100         and the --disable-libvtv configure options.
6102 2013-09-06  Jeff Law  <law@redhat.com>
6104         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6105         edge implied equivalences into successor phis.
6107 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6109         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6111 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6113         * resource.c (mark_target_live_regs): Compute resources taking
6114         into account if a call is predicated or not.
6116 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6118         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6119         the compiler in the function names.
6121 2013-09-06  Jan Hubicka  <jh@suse.cz>
6123         PR middle-end/58094
6124         * ipa-inline.c (has_caller_p): New function.
6125         (want_inline_function_to_all_callers_p): Use it.
6126         (sum_callers, inline_to_all_callers): Break out from ...
6127         (ipa_inline): ... here.
6129 2013-09-06  Jan Hubicka  <jh@suse.cz>
6131         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6132         only when AVX is enabled.
6134 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6136         * config/aarch64/aarch64.md
6137         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6138         is fpsimd_<load/store>2.
6139         (load_pair<mode>): Likewise.
6140         (store_pair<mode>): Likewise.
6142 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6144         * config/arm/types.md (type): Add "mrs" type.
6145         * config/aarch64/aarch64.md
6146         (aarch64_load_tp_hard): Make type "mrs".
6147         * config/arm/arm.md
6148         (load_tp_hard): Make type "mrs".
6149         * config/arm/cortex-a15.md: Update with new attributes.
6150         * config/arm/cortex-a5.md: Update with new attributes.
6151         * config/arm/cortex-a53.md: Update with new attributes.
6152         * config/arm/cortex-a7.md: Update with new attributes.
6153         * config/arm/cortex-a8.md: Update with new attributes.
6154         * config/arm/cortex-a9.md: Update with new attributes.
6155         * config/arm/cortex-m4.md: Update with new attributes.
6156         * config/arm/cortex-r4.md: Update with new attributes.
6157         * config/arm/fa526.md: Update with new attributes.
6158         * config/arm/fa606te.md: Update with new attributes.
6159         * config/arm/fa626te.md: Update with new attributes.
6160         * config/arm/fa726te.md: Update with new attributes.
6162 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6164         * config/aarch64/aarch64.md
6165         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6166         (*movtf_aarch64): Likewise.
6167         * config/arm/arm.md
6168         (thumb1_movdi_insn): Use "multiple" for type where more than one
6169         instruction is used for a move.
6170         (*arm32_movhf): Likewise.
6171         (*thumb_movdf_insn): Likewise.
6173 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6175         * config/arm/types.md (type): Rename fcpys to fmov.
6176         * config/arm/vfp.md
6177         (*arm_movsi_vfp): Rename type fcpys as fmov.
6178         (*thumb2_movsi_vfp): Likewise
6179         (*movhf_vfp_neon): Likewise
6180         (*movhf_vfp): Likewise
6181         (*movsf_vfp): Likewise
6182         (*thumb2_movsf_vfp): Likewise
6183         (*movsfcc_vfp): Likewise
6184         (*thumb2_movsfcc_vfp): Likewise
6185         * config/aarch64/aarch64-simd.md
6186         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6187         * config/aarch64/aarch64.md
6188         (*movsi_aarch64): Replace type mov_reg with fmovs.
6189         (*movdi_aarch64): Likewise
6190         (*movsf_aarch64): Likewise
6191         (*movdf_aarch64): Likewise
6192         * config/arm/arm.c
6193         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6194         * config/arm/iwmmxt.md
6195         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6196         * config/arm/arm1020e.md: Update with new attributes.
6197         * config/arm/cortex-a15-neon.md: Update with new attributes.
6198         * config/arm/cortex-a5.md: Update with new attributes.
6199         * config/arm/cortex-a53.md: Update with new attributes.
6200         * config/arm/cortex-a7.md: Update with new attributes.
6201         * config/arm/cortex-a8-neon.md: Update with new attributes.
6202         * config/arm/cortex-a9.md: Update with new attributes.
6203         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6204         * config/arm/cortex-r4f.md: Update with new attributes.
6205         * config/arm/marvell-pj4.md: Update with new attributes.
6206         * config/arm/vfp11.md: Update with new attributes.
6208 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6210         * config/aarch64/aarch64.md
6211         (*madd<mode>): Fix type attribute.
6212         (*maddsi_uxtw): Likewise.
6213         (*msub<mode>): Likewise.
6214         (*msubsi_uxtw): Likewise.
6215         (<su_optab>maddsidi4): Likewise.
6216         (<su_optab>msubsidi4): Likewise.
6218 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6220         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6221         * config/arm/arm.md (core_cycles): Remove fdiv.
6222         * config/arm/vfp.md:
6223         (*sqrtsf2_vfp): Update for attribute changes.
6224         (*sqrtdf2_vfp): Likewise.
6225         * config/aarch64/aarch64.md:
6226         (sqrt<mode>2): Update for attribute changes.
6227         * config/arm/arm1020e.md: Update with new attributes.
6228         * config/arm/cortex-a15-neon.md: Update with new attributes.
6229         * config/arm/cortex-a5.md: Update with new attributes.
6230         * config/arm/cortex-a53.md: Update with new attributes.
6231         * config/arm/cortex-a7.md: Update with new attributes.
6232         * config/arm/cortex-a8-neon.md: Update with new attributes.
6233         * config/arm/cortex-a9.md: Update with new attributes.
6234         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6235         * config/arm/cortex-r4f.md: Update with new attributes.
6236         * config/arm/marvell-pj4.md: Update with new attributes.
6237         * config/arm/vfp11.md: Update with new attributes.
6239 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6241         * config/arm/types.md
6242         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6243         * config/aarch64/aarch64.md
6244         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6245         new attributes.
6246         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6247         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6248         (float<GPI:mode><GPF:mode>2): Likewise.
6249         * config/arm/vfp.md
6250         (*truncsisf2_vfp): Update with new attributes.
6251         (*truncsidf2_vfp): Likewise.
6252         (fixuns_truncsfsi2): Likewise.
6253         (fixuns_truncdfsi2): Likewise.
6254         (*floatsisf2_vfp): Likewise.
6255         (*floatsidf2_vfp): Likewise.
6256         (floatunssisf2): Likewise.
6257         (floatunssidf2): Likewise.
6258         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6259         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6260         * config/arm/arm1020e.md: Update with new attributes.
6261         * config/arm/cortex-a15-neon.md: Update with new attributes.
6262         * config/arm/cortex-a5.md: Update with new attributes.
6263         * config/arm/cortex-a53.md: Update with new attributes.
6264         * config/arm/cortex-a7.md: Update with new attributes.
6265         * config/arm/cortex-a8-neon.md: Update with new attributes.
6266         * config/arm/cortex-a9.md: Update with new attributes.
6267         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6268         * config/arm/cortex-r4f.md: Update with new attributes.
6269         * config/arm/marvell-pj4.md: Update with new attributes.
6270         * config/arm/vfp11.md: Update with new attributes.
6272 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6274         * config/aarch64/arm_neon.h
6275         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6276         (vqtbx<1,2,3,4><q>_s8): Likewise.
6278 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6280         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6281         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6282         (add<mode>3): Add type attribute.
6283         (add<mode>3): Likewise.
6284         (usadd<mode>3): Likewise.
6285         (ssadd<mode>3): Likewise.
6286         (sub<mode>3): Likewise.
6287         (sub<mode>3): Likewise.
6288         (ussub<mode>3): Likewise.
6289         (sssub<mode>3): Likewise.
6290         (ssmulsa3): Likewise.
6291         (usmulusa3): Likewise.
6292         (arm_usatsihi): Likewise.
6293         * config/arm/vfp.md
6294         (*movdi_vfp): Add types for all instructions.
6295         (*movdi_vfp_cortexa8): Likewise.
6296         (*movhf_vfp_neon): Likewise.
6297         (*movhf_vfp): Likewise.
6298         (*movdf_vfp): Likewise.
6299         (*thumb2_movdf_vfp): Likewise.
6300         (*thumb2_movdfcc_vfp): Likewise.
6301         * config/arm/arm.md: Add type attribute to all insn patterns.
6302         (*thumb1_adddi3): Add type attribute.
6303         (*arm_adddi3): Likewise.
6304         (*adddi_sesidi_di): Likewise.
6305         (*adddi_zesidi_di): Likewise.
6306         (*thumb1_addsi3): Likewise.
6307         (addsi3_compare0): Likewise.
6308         (*addsi3_compare0_scratch): Likewise.
6309         (*compare_negsi_si): Likewise.
6310         (cmpsi2_addneg): Likewise.
6311         (*addsi3_carryin_<optab>): Likewise.
6312         (*addsi3_carryin_alt2_<optab>): Likewise.
6313         (*addsi3_carryin_clobercc_<optab>): Likewise.
6314         (*subsi3_carryin): Likewise.
6315         (*subsi3_carryin_const): Likewise.
6316         (*subsi3_carryin_compare): Likewise.
6317         (*subsi3_carryin_compare_const): Likewise.
6318         (*arm_subdi3): Likewise.
6319         (*thumb_subdi3): Likewise.
6320         (*subdi_di_zesidi): Likewise.
6321         (*subdi_di_sesidi): Likewise.
6322         (*subdi_zesidi_di): Likewise.
6323         (*subdi_sesidi_di): Likewise.
6324         (*subdi_zesidi_ze): Likewise.
6325         (thumb1_subsi3_insn): Likewise.
6326         (*arm_subsi3_insn): Likewise.
6327         (*anddi3_insn): Likewise.
6328         (*anddi_zesidi_di): Likewise.
6329         (*anddi_sesdi_di): Likewise.
6330         (*ne_zeroextracts): Likewise.
6331         (*ne_zeroextracts): Likewise.
6332         (*ite_ne_zeroextr): Likewise.
6333         (*ite_ne_zeroextr): Likewise.
6334         (*anddi_notdi_di): Likewise.
6335         (*anddi_notzesidi): Likewise.
6336         (*anddi_notsesidi): Likewise.
6337         (andsi_notsi_si): Likewise.
6338         (thumb1_bicsi3): Likewise.
6339         (*iordi3_insn): Likewise.
6340         (*iordi_zesidi_di): Likewise.
6341         (*iordi_sesidi_di): Likewise.
6342         (*thumb1_iorsi3_insn): Likewise.
6343         (*xordi3_insn): Likewise.
6344         (*xordi_zesidi_di): Likewise.
6345         (*xordi_sesidi_di): Likewise.
6346         (*arm_xorsi3): Likewise.
6347         (*andsi_iorsi3_no): Likewise.
6348         (*smax_0): Likewise.
6349         (*smax_m1): Likewise.
6350         (*arm_smax_insn): Likewise.
6351         (*smin_0): Likewise.
6352         (*arm_smin_insn): Likewise.
6353         (*arm_umaxsi3): Likewise.
6354         (*arm_uminsi3): Likewise.
6355         (*minmax_arithsi): Likewise.
6356         (*minmax_arithsi_): Likewise.
6357         (*satsi_<SAT:code>): Likewise.
6358         (arm_ashldi3_1bit): Likewise.
6359         (arm_ashrdi3_1bit): Likewise.
6360         (arm_lshrdi3_1bit): Likewise.
6361         (*arm_negdi2): Likewise.
6362         (*thumb1_negdi2): Likewise.
6363         (*arm_negsi2): Likewise.
6364         (*thumb1_negsi2): Likewise.
6365         (*negdi_extendsid): Likewise.
6366         (*negdi_zero_extend): Likewise.
6367         (*arm_abssi2): Likewise.
6368         (*thumb1_abssi2): Likewise.
6369         (*arm_neg_abssi2): Likewise.
6370         (*thumb1_neg_abss): Likewise.
6371         (one_cmpldi2): Likewise.
6372         (extend<mode>di2): Likewise.
6373         (*compareqi_eq0): Likewise.
6374         (*arm_extendhisi2addsi): Likewise.
6375         (*arm_movdi): Likewise.
6376         (*thumb1_movdi_insn): Likewise.
6377         (*arm_movt): Likewise.
6378         (*thumb1_movsi_insn): Likewise.
6379         (pic_add_dot_plus_four): Likewise.
6380         (pic_add_dot_plus_eight): Likewise.
6381         (tls_load_dot_plus_eight): Likewise.
6382         (*thumb1_movhi_insn): Likewise.
6383         (*thumb1_movsf_insn): Likewise.
6384         (*movdf_soft_insn): Likewise.
6385         (*thumb_movdf_insn): Likewise.
6386         (cbranchsi4_insn): Likewise.
6387         (cbranchsi4_scratch): Likewise.
6388         (*negated_cbranchsi4): Likewise.
6389         (*tbit_cbranch): Likewise.
6390         (*tlobits_cbranch): Likewise.
6391         (*tstsi3_cbranch): Likewise.
6392         (*cbranchne_decr1): Likewise.
6393         (*addsi3_cbranch): Likewise.
6394         (*addsi3_cbranch_scratch): Likewise.
6395         (*arm_cmpdi_insn): Likewise.
6396         (*arm_cmpdi_unsig): Likewise.
6397         (*arm_cmpdi_zero): Likewise.
6398         (*thumb_cmpdi_zero): Likewise.
6399         (*deleted_compare): Likewise.
6400         (*mov_scc): Likewise.
6401         (*mov_negscc): Likewise.
6402         (*mov_notscc): Likewise.
6403         (*cstoresi_eq0_thumb1_insn): Likewise.
6404         (cstoresi_nltu_thumb1): Likewise.
6405         (cstoresi_ltu_thu): Likewise.
6406         (thumb1_addsi3_addgeu): Likewise.
6407         (*arm_jump): Likewise.
6408         (*thumb_jump): Likewise.
6409         (*check_arch2): Likewise.
6410         (arm_casesi_internal): Likewise.
6411         (thumb1_casesi_dispatch): Likewise.
6412         (*arm_indirect_jump): Likewise.
6413         (*thumb1_indirect_jump): Likewise.
6414         (nop): Likewise.
6415         (*and_scc): Likewise.
6416         (*ior_scc): Likewise.
6417         (*compare_scc): Likewise.
6418         (*cond_move): Likewise.
6419         (*cond_arith): Likewise.
6420         (*cond_sub): Likewise.
6421         (*cmp_ite0): Likewise.
6422         (*cmp_ite1): Likewise.
6423         (*cmp_and): Likewise.
6424         (*cmp_ior): Likewise.
6425         (*ior_scc_scc): Likewise.
6426         (*ior_scc_scc_cmp): Likewise.
6427         (*and_scc_scc): Likewise.
6428         (*and_scc_scc_cmp): Likewise.
6429         (*and_scc_scc_nod): Likewise.
6430         (*negscc): Likewise.
6431         (movcond_addsi): Likewise.
6432         (movcond): Likewise.
6433         (*ifcompare_plus_move): Likewise.
6434         (*if_plus_move): Likewise.
6435         (*ifcompare_move_plus): Likewise.
6436         (*if_move_plus): Likewise.
6437         (*ifcompare_arith_arith): Likewise.
6438         (*if_arith_arith): Likewise.
6439         (*ifcompare_arith_move): Likewise.
6440         (*if_arith_move): Likewise.
6441         (*ifcompare_move_arith): Likewise.
6442         (*if_move_arith): Likewise.
6443         (*ifcompare_move_not): Likewise.
6444         (*if_move_not): Likewise.
6445         (*ifcompare_not_move): Likewise.
6446         (*if_not_move): Likewise.
6447         (*ifcompare_shift_move): Likewise.
6448         (*if_shift_move): Likewise.
6449         (*ifcompare_move_shift): Likewise.
6450         (*if_move_shift): Likewise.
6451         (*ifcompare_shift_shift): Likewise.
6452         (*ifcompare_not_arith): Likewise.
6453         (*ifcompare_arith_not): Likewise.
6454         (*if_arith_not): Likewise.
6455         (*ifcompare_neg_move): Likewise.
6456         (*if_neg_move): Likewise.
6457         (*ifcompare_move_neg): Likewise.
6458         (*if_move_neg): Likewise.
6459         (prologue_thumb1_interwork): Likewise.
6460         (*cond_move_not): Likewise.
6461         (*sign_extract_onebit): Likewise.
6462         (*not_signextract_onebit): Likewise.
6463         (stack_tie): Likewise.
6464         (align_4): Likewise.
6465         (align_8): Likewise.
6466         (consttable_end): Likewise.
6467         (consttable_1): Likewise.
6468         (consttable_2): Likewise.
6469         (consttable_4): Likewise.
6470         (consttable_8): Likewise.
6471         (consttable_16): Likewise.
6472         (*thumb1_tablejump): Likewise.
6473         (prefetch): Likewise.
6474         (force_register_use): Likewise.
6475         (thumb_eh_return): Likewise.
6476         (load_tp_hard): Likewise.
6477         (load_tp_soft): Likewise.
6478         (tlscall): Likewise.
6479         (*arm_movtas_ze): Likewise.
6480         (*arm_rev): Likewise.
6481         (*arm_revsh): Likewise.
6482         (*arm_rev16): Likewise.
6483         * config/arm/thumb2.md
6484         (*thumb2_smaxsi3): Likewise.
6485         (*thumb2_sminsi3): Likewise.
6486         (*thumb32_umaxsi3): Likewise.
6487         (*thumb2_uminsi3): Likewise.
6488         (*thumb2_negdi2): Likewise.
6489         (*thumb2_abssi2): Likewise.
6490         (*thumb2_neg_abss): Likewise.
6491         (*thumb2_movsi_insn): Likewise.
6492         (tls_load_dot_plus_four): Likewise.
6493         (*thumb2_movhi_insn): Likewise.
6494         (*thumb2_mov_scc): Likewise.
6495         (*thumb2_mov_negs): Likewise.
6496         (*thumb2_mov_negs): Likewise.
6497         (*thumb2_mov_nots): Likewise.
6498         (*thumb2_mov_nots): Likewise.
6499         (*thumb2_movsicc_): Likewise.
6500         (*thumb2_movsfcc_soft_insn): Likewise.
6501         (*thumb2_indirect_jump): Likewise.
6502         (*thumb2_and_scc): Likewise.
6503         (*thumb2_ior_scc): Likewise.
6504         (*thumb2_ior_scc_strict_it): Likewise.
6505         (*thumb2_cond_move): Likewise.
6506         (*thumb2_cond_arith): Likewise.
6507         (*thumb2_cond_ari): Likewise.
6508         (*thumb2_cond_sub): Likewise.
6509         (*thumb2_negscc): Likewise.
6510         (*thumb2_movcond): Likewise.
6511         (thumb2_casesi_internal): Likewise.
6512         (thumb2_casesi_internal_pic): Likewise.
6513         (*thumb2_alusi3_short): Likewise.
6514         (*thumb2_mov<mode>_shortim): Likewise.
6515         (*thumb2_addsi_short): Likewise.
6516         (*thumb2_subsi_short): Likewise.
6517         (thumb2_addsi3_compare0): Likewise.
6518         (*thumb2_cbz): Likewise.
6519         (*thumb2_cbnz): Likewise.
6520         (*thumb2_one_cmplsi2_short): Likewise.
6521         (*thumb2_negsi2_short): Likewise.
6522         (*orsi_notsi_si): Likewise.
6523         * config/arm/arm1020e.md: Update with new attributes.
6524         * config/arm/arm1026ejs.md: Update with new attributes.
6525         * config/arm/arm1136jfs.md: Update with new attributes.
6526         * config/arm/arm926ejs.md: Update with new attributes.
6527         * config/arm/cortex-a15.md: Update with new attributes.
6528         * config/arm/cortex-a5.md: Update with new attributes.
6529         * config/arm/cortex-a53.md: Update with new attributes.
6530         * config/arm/cortex-a7.md: Update with new attributes.
6531         * config/arm/cortex-a8.md: Update with new attributes.
6532         * config/arm/cortex-a9.md: Update with new attributes.
6533         * config/arm/cortex-m4.md: Update with new attributes.
6534         * config/arm/cortex-r4.md: Update with new attributes.
6535         * config/arm/fa526.md: Update with new attributes.
6536         * config/arm/fa606te.md: Update with new attributes.
6537         * config/arm/fa626te.md: Update with new attributes.
6538         * config/arm/fa726te.md: Update with new attributes.
6540 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6542         * config/aarch64/aarch64-simd.md
6543         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6544         <vwx> iterator to ensure correct register choice.
6545         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6546         (aarch64_sqdmull_n<mode>): Likewise.
6547         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6548         * config/aarch64/arm_neon.h
6549         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6550         (vml<as><q>_n_<su>16): Likewise.
6551         (vml<as>l_high_lane<q>_<su>16): Likewise.
6552         (vml<as>l_high_n_<su>16): Likewise.
6553         (vml<as>l_lane<q>_<su>16): Likewise.
6554         (vml<as>l_n_<su>16): Likewise.
6555         (vmul<q>_lane<q>_<su>16): Likewise.
6556         (vmul<q>_n_<su>16): Likewise.
6557         (vmull_lane<q>_<su>16): Likewise.
6558         (vmull_n_<su>16): Likewise.
6559         (vmull_high_lane<q>_<su>16): Likewise.
6560         (vmull_high_n_<su>16): Likewise.
6561         (vqrdmulh<q>_n_s16): Likewise.
6563 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6565         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6566         have the correct lane parameter.
6568 2013-09-06  Richard Biener <rguenther@suse.de>
6570         * cfganal.c (control_dependences::~control_dependences):
6571         Properly free all of the vector.
6573 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6575         PR target/58269
6576         * config/i386/i386.c (ix86_conditional_register_usage):
6577         Proper initialize extended SSE registers.
6579 2013-09-06  Jan Hubicka  <jh@suse.cz>
6581         PR tree-optimization/58311
6582         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6584 2013-09-06  Jan Hubicka  <jh@suse.cz>
6586         * Makefile.in (tree-sra.o): Update dependencies.
6587         * tree-sra.c: Include ipa-utils.h
6588         (scan_function): Use recursive_call_p.
6589         (has_caller_p): New function.
6590         (cgraph_for_node_and_aliases): Count also callers of aliases.
6592 2013-09-06  Jan Hubicka  <jh@suse.cz>
6594         PR middle-end/58094
6595         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6596         * tree-tailcall.c: Include ipa-utils.h.
6597         (find_tail_calls): Use it.
6598         * ipa-pure-const.c (check_call): Likewise.
6599         * ipa-utils.c (recursive_call_p): New function.
6600         * ipa-utils.h (recursive_call_p): Dclare.
6601         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6602         (symtab_semantically_equivalent_p): New function.
6603         * Makefile.in (tree-tailcall.o): Update dependencies.
6605 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6607         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6608         non-inlinable part.
6610 2013-09-06  Richard Biener  <rguenther@suse.de>
6612         * lto-streamer.h (lto_global_var_decls): Remove.
6613         * Makefile.in (OBJS): Remove lto-symtab.o.
6614         (lto-symtab.o): Remove.
6615         (GTFILES): Remove lto-symtab.c
6616         * lto-symtab.c: Move to lto/
6618 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6620         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6621         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6622         (UNSPEC_FPINT_RINT): New constant definitions.
6623         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6624         definition with 2 attributes.
6625         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6626         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6627         definitions.
6629 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6631         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6632         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6633         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6634         attribute to "z196_alone".
6635         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6636         "zEC12_simple".
6638 2013-09-06  Richard Biener  <rguenther@suse.de>
6640         * basic-block.h (class control_dependences): New.
6641         * tree-ssa-dce.c (control_dependence_map): Remove.
6642         (cd): New global.
6643         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6644         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6645         find_pdom, find_control_dependence, find_all_control_dependences):
6646         Move to cfganal.c.
6647         (mark_control_dependent_edges_necessary,
6648         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6649         tree_dce_done, perform_tree_ssa_dce): Adjust.
6650         * cfganal.c (set_control_dependence_map_bit,
6651         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6652         find_all_control_dependences): Move from tree-ssa-dce.c and
6653         implement as methods of control_dependences class.
6654         (control_dependences::control_dependences): New.
6655         (control_dependences::~control_dependences): Likewise.
6656         (control_dependences::get_edges_dependent_on): Likewise.
6657         (control_dependences::get_edge): Likewise.
6659 2013-09-04  Jan Hubicka  <jh@suse.cz>
6661         * tree.c (types_same_for_odr): Drop overactive check.
6662         * ipa-devirt.c (hash_type_name): Likewise.
6664 2013-09-04  Jan Hubicka  <jh@suse.cz>
6666         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6667         (analyze_functions): ... here.
6669 2013-09-04  Jan Hubicka  <jh@suse.cz>
6671         PR middle-end/58201
6672         * cgraphunit.c (analyze_functions): Clear AUX fields
6673         after processing; initialize assembler name has.
6675 2013-09-05  Jeff Law  <law@redhat.com>
6677         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6678         from thread_around_empty_block.  Record threading path into PATH.
6679         Recurse if threading through the initial block is successful.
6680         (thread_across_edge): Corresponding changes to slightly simplify.
6682 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6684         * config/aarch64/aarch64.md
6685         (type): Remove frecpe, frecps, frecpx.
6686         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6687         fix to be a TARGET_SIMD instruction.
6688         (aarch64_frecps): Remove.
6689         * config/aarch64/aarch64-simd.md
6690         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6691         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6693 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6694             Sofiane Naci  <sofiane.naci@arm.com>
6696         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6697         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6698         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6699         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6700         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6701         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6702         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6703         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6704         into "clz, "rbit".  Rename "shift" to "shift_imm".
6705         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6706         changes.  Update for attribute changes all occurrences of arlo_* and
6707         shift* types.
6708         * config/arm/arm-fixed.md: Update for attribute changes
6709         all occurrences of arlo_* types.
6710         * config/arm/thumb2.md: Update for attribute changes all occurrences
6711         of arlo_* types.
6712         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6713         (cortexa7_older_only): Likewise.
6714         (cortexa7_younger):  Likewise.
6715         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6716         (1020alu_shift_op): Likewise.
6717         (1020alu_shift_reg_op): Likewise.
6718         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6719         (alu_shift_op): Likewise.
6720         (alu_shift_reg_op): Likewise.
6721         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6722         (11_alu_shift_op): Likewise.
6723         (11_alu_shift_reg_op): Likewise.
6724         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6725         (9_alu_shift_reg_op): Likewise.
6726         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6727         attribute changes.
6728         (cortex_a15_alu_shift): Likewise.
6729         (cortex_a15_alu_shift_reg): Likewise.
6730         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6731         attribute changes.
6732         (cortex_a5_alu_shift): Likewise.
6733         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6734         attribute changes.
6735         (cortex_a53_alu_shift): Likewise.
6736         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6737         attribute changes.
6738         (cortex_a7_alu_reg): Likewise.
6739         (cortex_a7_alu_shift): Likewise.
6740         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6741         attribute changes.
6742         (cortex_a8_alu_shift): Likewise.
6743         (cortex_a8_alu_shift_reg): Likewise.
6744         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6745         (cortex_a9_dp_shift): Likewise.
6746         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6747         attribute changes.
6748         * config/arm/cortex-r4.md
6749         (cortex_r4_alu): Update for attribute changes.
6750         (cortex_r4_mov): Likewise.
6751         (cortex_r4_alu_shift_reg): Likewise.
6752         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6753         (526_alu_shift_op): Likewise.
6754         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6755         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6756         (626te_alu_shift_op): Likewise.
6757         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6758         (726te_alu_shift_op): Likewise.
6759         (726te_alu_shift_reg_op): Likewise.
6760         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6761         (mp626_alu_shift_op): Likewise.
6762         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6763         (pj4_alu_conds): Likewise.
6764         (pj4_shift): Likewise.
6765         (pj4_shift_conds): Likewise.
6766         (pj4_alu_shift): Likewise.
6767         (pj4_alu_shift_conds): Likewise.
6768         * config/aarch64/aarch64.md: Update for attribute change
6769         all occurrences of arlo_* and shift* types.
6771 2013-09-05  Mike Stump  <mikestump@comcast.net>
6773         * tree.h: Move documentation for tree_function_decl to tree-core.h
6774         with the declaration.
6776 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6778         PR target/58139
6779         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6780         looking for widest mode.
6782 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6784         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6786 2013-09-05  Richard Biener  <rguenther@suse.de>
6788         PR tree-optimization/58137
6789         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6790         Do not create vectors of pointers.
6791         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6792         types for the components of the vector initializer.
6793         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6794         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6796 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6798         * ipa-prop.c (remove_described_reference): Accept missing references,
6799         return false if that hppens, otherwise return true.
6800         (cgraph_node_for_jfunc): New function.
6801         (try_decrement_rdesc_refcount): Likewise.
6802         (try_make_edge_direct_simple_call): Use them.
6803         (ipa_edge_removal_hook): Remove references from rdescs.
6804         (ipa_edge_duplication_hook): Clone rdescs and their references
6805         when the new edge has the same caller as the old one.
6806         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6807         reference before removing any edges.
6809 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6811         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6812         initial store.
6813         * thumb2.md (thumb2_storewb_parisi): New pattern.
6815 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6817         * config/aarch64/aarch64-option-extensions.def: Add
6818         AARCH64_OPT_EXTENSION of 'crc'.
6819         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6820         (AARCH64_ISA_CRC): Ditto.
6821         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6822         description of the CRC extension.
6824 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6826         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6827         * config/rs6000/linux.h: Ditto.
6828         * alpha/linux.h: Ditto.
6829         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6830         no_c99_libc_has_function.
6831         * config/c6x/uclinux-elf.h: Ditto.
6832         * config/lm32/uclinux-elf.h: Ditto.
6833         * config/m68k/uclinux.h: Ditto.
6834         * config/moxie/uclinux.h: Ditto.
6835         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6836         (crisv32-*-linux*, cris-*-linux*): Ditto.
6837         * config/bfin/bfin.c: Include "tm_p.h".
6839 2013-09-05  Richard Biener  <rguenther@suse.de>
6841         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6842         check for a definition without a basic-block.
6844 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6845             Sofiane Naci <sofiane.naci@arm.com>
6847         * config/aarch64/aarch64.md
6848         (*movti_aarch64): Rename r_2_f and f_2_r.
6849         (*movsf_aarch64): Likewise.
6850         (*movdf_aarch64): Likewise.
6851         (*movtf_aarch64): Likewise.
6852         (aarch64_movdi_<mode>low): Likewise.
6853         (aarch64_movdi_<mode>high): Likewise.
6854         (aarch64_mov<mode>high_di): Likewise.
6855         (aarch64_mov<mode>low_di): Likewise.
6856         (aarch64_movtilow_tilow): Likewise.
6857         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6858         values to config/arm/types.md
6859         (attribute "conds"): Update for attribute change.
6860         (anddi3_insn): Likewise.
6861         (iordi3_insn): Likewise.
6862         (xordi3_insn): Likewise.
6863         (one_cmpldi2): Likewise.
6864         * config/arm/types.md (type): Add Neon types.
6865         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6866         use "type" attribute.
6867         (movmisalign<mode>_neon_store): Likewise.
6868         (movmisalign<mode>_neon_load): Likewise.
6869         (vec_set<mode>_internal): Likewise.
6870         (vec_setv2di_internal): Likewise.
6871         (vec_extract<mode>): Likewise.
6872         (vec_extractv2di): Likewise.
6873         (add<mode>3_neon): Likewise.
6874         (adddi3_neon): Likewise.
6875         (sub<mode>3_neon): Likewise.
6876         (subdi3_neon): Likewise.
6877         (mul<mode>3_neon): Likewise.
6878         (mul<mode>3add<mode>_neon): Likewise.
6879         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6880         (fma<VCVTF:mode>4)): Likewise.
6881         (fma<VCVTF:mode>4_intrinsic): Likewise.
6882         (fmsub<VCVTF:mode>4)): Likewise.
6883         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6884         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6885         (ior<mode>3): Likewise.
6886         (and<mode>3): Likewise.
6887         (anddi3_neon): Likewise.
6888         (orn<mode>3_neon): Likewise.
6889         (orndi3_neon): Likewise.
6890         (bic<mode>3_neon): Likewise.
6891         (bicdi3_neon): Likewise.
6892         (xor<mode>3): Likewise.
6893         (one_cmpl<mode>2): Likewise.
6894         (abs<mode>2): Likewise.
6895         (neg<mode>2): Likewise.
6896         (umin<mode>3_neon): Likewise.
6897         (umax<mode>3_neon): Likewise.
6898         (smin<mode>3_neon): Likewise.
6899         (smax<mode>3_neon): Likewise.
6900         (vashl<mode>3): Likewise.
6901         (vashr<mode>3_imm): Likewise.
6902         (vlshr<mode>3_imm): Likewise.
6903         (ashl<mode>3_signed): Likewise.
6904         (ashl<mode>3_unsigned): Likewise.
6905         (neon_load_count): Likewise.
6906         (ashldi3_neon_noclobber): Likewise.
6907         (signed_shift_di3_neon): Likewise.
6908         (unsigned_shift_di3_neon): Likewise.
6909         (ashrdi3_neon_imm_noclobber): Likewise.
6910         (lshrdi3_neon_imm_noclobber): Likewise.
6911         (widen_ssum<mode>3): Likewise.
6912         (widen_usum<mode>3): Likewise.
6913         (quad_halves_<code>v4si): Likewise.
6914         (quad_halves_<code>v4sf): Likewise.
6915         (quad_halves_<code>v8hi): Likewise.
6916         (quad_halves_<code>v16qi): Likewise.
6917         (reduc_splus_v2di): Likewise.
6918         (neon_vpadd_internal<mode>): Likewise.
6919         (neon_vpsmin<mode>): Likewise.
6920         (neon_vpsmax<mode>): Likewise.
6921         (neon_vpumin<mode>): Likewise.
6922         (neon_vpumax<mode>): Likewise.
6923         (ss_add<mode>_neon): Likewise.
6924         (us_add<mode>_neon): Likewise.
6925         (ss_sub<mode>_neon): Likewise.
6926         (us_sub<mode>_neon): Likewise.
6927         (neon_vadd<mode>_unspec): Likewise.
6928         (neon_vaddl<mode>): Likewise.
6929         (neon_vaddw<mode>): Likewise.
6930         (neon_vhadd<mode>): Likewise.
6931         (neon_vqadd<mode>): Likewise.
6932         (neon_vaddhn<mode>): Likewise.
6933         (neon_vmul<mode>): Likewise.
6934         (neon_vmla<mode>): Likewise.
6935         (neon_vmlal<mode>): Likewise.
6936         (neon_vmls<mode>): Likewise.
6937         (neon_vmlsl<mode>): Likewise.
6938         (neon_vqdmulh<mode>): Likewise.
6939         (neon_vqdmlal<mode>): Likewise.
6940         (neon_vqdmlsl<mode>): Likewise.
6941         (neon_vmull<mode>): Likewise.
6942         (neon_vqdmull<mode>): Likewise.
6943         (neon_vsub<mode>_unspec): Likewise.
6944         (neon_vsubl<mode>): Likewise.
6945         (neon_vsubw<mode>): Likewise.
6946         (neon_vqsub<mode>): Likewise.
6947         (neon_vhsub<mode>): Likewise.
6948         (neon_vsubhn<mode>): Likewise.
6949         (neon_vceq<mode>): Likewise.
6950         (neon_vcge<mode>): Likewise.
6951         (neon_vcgeu<mode>): Likewise.
6952         (neon_vcgt<mode>): Likewise.
6953         (neon_vcgtu<mode>): Likewise.
6954         (neon_vcle<mode>): Likewise.
6955         (neon_vclt<mode>): Likewise.
6956         (neon_vcage<mode>): Likewise.
6957         (neon_vcagt<mode>): Likewise.
6958         (neon_vtst<mode>): Likewise.
6959         (neon_vabd<mode>): Likewise.
6960         (neon_vabdl<mode>): Likewise.
6961         (neon_vaba<mode>): Likewise.
6962         (neon_vabal<mode>): Likewise.
6963         (neon_vmax<mode>): Likewise.
6964         (neon_vmin<mode>): Likewise.
6965         (neon_vpaddl<mode>): Likewise.
6966         (neon_vpadal<mode>): Likewise.
6967         (neon_vpmax<mode>): Likewise.
6968         (neon_vpmin<mode>): Likewise.
6969         (neon_vrecps<mode>): Likewise.
6970         (neon_vrsqrts<mode>): Likewise.
6971         (neon_vqabs<mode>): Likewise.
6972         (neon_vqneg<mode>): Likewise.
6973         (neon_vcls<mode>): Likewise.
6974         (clz<mode>2): Likewise.
6975         (popcount<mode>2): Likewise.
6976         (neon_vrecpe): Likewise.
6977         (neon_vrsqrte): Likewise.
6978         (neon_vget_lane<mode>_sext_internal): Likewise.
6979         (neon_vget_lane<mode>_zext_internal): Likewise.
6980         (neon_vdup_n<mode>): Likewise.
6981         (neon_vdup_nv2di): Likewise.
6982         (neon_vdpu_lane<mode>_internal): Likewise.
6983         (neon_vswp<mode>): Likewise.
6984         (float<mode><V_cvtto>2): Likewise.
6985         (floatuns<mode><V_cvtto>2): Likewise.
6986         (fix_trunc<mode><V_cvtto>)2): Likewise
6987         (fixuns_trunc<mode><V_cvtto)2): Likewise.
6988         (neon_vcvt<mode>): Likewise.
6989         (neon_vcvtv4sfv4hf): Likewise.
6990         (neon_vcvtv4hfv4sf): Likewise.
6991         (neon_vcvt_n<mode>): Likewise.
6992         (neon_vmovn<mode>): Likewise.
6993         (neon_vqmovn<mode>): Likewise.
6994         (neon_vqmovun<mode>): Likewise.
6995         (neon_vmovl<mode>): Likewise.
6996         (neon_vmul_lane<mode>): Likewise.
6997         (neon_vmull_lane<mode>): Likewise.
6998         (neon_vqdmull_lane<mode>): Likewise.
6999         (neon_vqdmulh_lane<mode>): Likewise.
7000         (neon_vmla_lane<mode>): Likewise.
7001         (neon_vmlal_lane<mode>): Likewise.
7002         (neon_vqdmlal_lane<mode>): Likewise.
7003         (neon_vmls_lane<mode>): Likewise.
7004         (neon_vmlsl_lane<mode>): Likewise.
7005         (neon_vqdmlsl_lane<mode>): Likewise.
7006         (neon_vext<mode>): Likewise.
7007         (neon_vrev64<mode>): Likewise.
7008         (neon_vrev32<mode>): Likewise.
7009         (neon_vrev16<mode>): Likewise.
7010         (neon_vbsl<mode>_internal): Likewise.
7011         (neon_vshl<mode>): Likewise.
7012         (neon_vqshl<mode>): Likewise.
7013         (neon_vshr_n<mode>): Likewise.
7014         (neon_vshrn_n<mode>): Likewise.
7015         (neon_vqshrn_n<mode>): Likewise.
7016         (neon_vqshrun_n<mode>): Likewise.
7017         (neon_vshl_n<mode>): Likewise.
7018         (neon_vqshl_n<mode>): Likewise.
7019         (neon_vqshlu_n<mode>): Likewise.
7020         (neon_vshll_n<mode>): Likewise.
7021         (neon_vsra_n<mode>): Likewise.
7022         (neon_vsri_n<mode>): Likewise.
7023         (neon_vsli_n<mode>): Likewise.
7024         (neon_vtbl1v8qi): Likewise.
7025         (neon_vtbl2v8qi): Likewise.
7026         (neon_vtbl3v8qi): Likewise.
7027         (neon_vtbl4v8qi): Likewise.
7028         (neon_vtbx1v8qi): Likewise.
7029         (neon_vtbx2v8qi): Likewise.
7030         (neon_vtbx3v8qi): Likewise.
7031         (neon_vtbx4v8qi): Likewise.
7032         (neon_vtrn<mode>_internal): Likewise.
7033         (neon_vzip<mode>_internal): Likewise.
7034         (neon_vuzp<mode>_internal): Likewise.
7035         (neon_vld1<mode>): Likewise.
7036         (neon_vld1_lane<mode>): Likewise.
7037         (neon_vld1_dup<mode>): Likewise.
7038         (neon_vld1_dupv2di): Likewise.
7039         (neon_vst1<mode>): Likewise.
7040         (neon_vst1_lane<mode>): Likewise.
7041         (neon_vld2<mode>): Likewise.
7042         (neon_vld2_lane<mode>): Likewise.
7043         (neon_vld2_dup<mode>): Likewise.
7044         (neon_vst2<mode>): Likewise.
7045         (neon_vst2_lane<mode>): Likewise.
7046         (neon_vld3<mode>): Likewise.
7047         (neon_vld3qa<mode>): Likewise.
7048         (neon_vld3qb<mode>): Likewise.
7049         (neon_vld3_lane<mode>): Likewise.
7050         (neon_vld3_dup<mode>): Likewise.
7051         (neon_vst3<mode>): Likewise.
7052         (neon_vst3qa<mode>): Likewise.
7053         (neon_vst3qb<mode>): Likewise.
7054         (neon_vst3_lane<mode>): Likewise.
7055         (neon_vld4<mode>): Likewise.
7056         (neon_vld4qa<mode>): Likewise.
7057         (neon_vld4qb<mode>): Likewise.
7058         (neon_vld4_lane<mode>): Likewise.
7059         (neon_vld4_dup<mode>): Likewise.
7060         (neon_vst4<mode>): Likewise.
7061         (neon_vst4qa<mode>): Likewise.
7062         (neon_vst4qb<mode>): Likewise.
7063         (neon_vst4_lane<mode>): Likewise.
7064         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7065         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7066         (neon_vec_<US>mult_lo_<mode>): Likewise.
7067         (neon_vec_<US>mult_hi_<mode>): Likewise.
7068         (neon_vec_<US>shiftl_<mode>): Likewise.
7069         (neon_unpack<US>_<mode>): Likewise.
7070         (neon_vec_<US>mult_<mode>): Likewise.
7071         (vec_pack_trunc_<mode>): Likewise.
7072         (neon_vec_pack_trunk_<mode>): Likewise.
7073         (neon_vabd<mode>_2): Likewise.
7074         (neon_vabd<mode>_3): Likewise.
7075         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7076         (thumb2_movsi_vfp): Likewise.
7077         (movdi_vfp): Likewise.
7078         (movdi_vfp_cortexa8): Likewise.
7079         (movhf_vfp_neon): Likewise.
7080         (movhf_vfp): Likewiwse.
7081         (movsf_vfp): Likewiwse.
7082         (thumb2_movsf_vfp): Likewiwse.
7083         (movdf_vfp): Likewise.
7084         (thumb2_movdf_vfp): Likewise.
7085         (movsfcc_vfp): Likewise.
7086         (thumb2_movsfcc_vfp): Likewise.
7087         (movdfcc_vfp): Likewise.
7088         (thumb2_movdfcc_vfp): Likewise.
7089         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7090         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7091         (v10_v2c): Likewise.
7092         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7093         attribute change.
7094         (cortex_a15_neon_int_2): Likewise.
7095         (cortex_a15_neon_int_3): Likewise.
7096         (cortex_a15_neon_int_4): Likewise.
7097         (cortex_a15_neon_int_5): Likewise.
7098         (cortex_a15_neon_vqneg_vqabs): Likewise.
7099         (cortex_a15_neon_vmov): Likewise.
7100         (cortex_a15_neon_vaba): Likewise.
7101         (cortex_a15_neon_vaba_qqq): Likewise.
7102         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7103         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7104         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7105         scalar_64_32_long_scalar): Likewise.
7106         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7107         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7108         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7109         lotype_qdd_64_32_long): Likewise.
7110         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7111         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7112         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7113         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7114         (cortex_a15_neon_shift_1): Likewise.
7115         (cortex_a15_neon_shift_2): Likewise.
7116         (cortex_a15_neon_shift_3): Likewise.
7117         (cortex_a15_neon_vshl_ddd): Likewise.
7118         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7119         (cortex_a15_neon_vsra_vrsra): Likewise.
7120         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7121         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7122         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7123         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7124         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7125         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7126         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7127         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7128         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7129         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7130         (cortex_a15_neon_bp_simple): Likewise.
7131         (cortex_a15_neon_bp_2cycle): Likewise.
7132         (cortex_a15_neon_bp_3cycle): Likewise.
7133         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7134         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7135         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7136         (cortex_a15_neon_vld2_4_regs): Likewise.
7137         (cortex_a15_neon_vld3_vld4): Likewise.
7138         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7139         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7140         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7141         (cortex_a15_neon_vst3_vst4): Likewise.
7142         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7143         (cortex_a15_neon_vld3_vld4_lane" 10
7144         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7145         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7146         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7147         (cortex_a15_neon_ldm_2): Likewise.0
7148         (cortex_a15_neon_stm_2): Likewise.
7149         (cortex_a15_neon_mcr): Likewise.
7150         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7151         (cortex_a15_neon_mrc): Likewise.
7152         (cortex_a15_neon_mrrc): Likewise.
7153         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7154         change.
7155         (cortex_a15_alu_shift): Likewise.
7156         (cortex_a15_alu_shift_reg): Likewise.
7157         (cortex_a15_mult32): Likewise.
7158         (cortex_a15_mult64): Likewise.
7159         (cortex_a15_block): Likewise.
7160         (cortex_a15_branch): Likewise.
7161         (cortex_a15_load1): Likewise.
7162         (cortex_a15_load3): Likewise.
7163         (cortex_a15_store1): Likewise.
7164         (cortex_a15_store3): Likewise.
7165         (cortex_a15_call): Likewise.
7166         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7167         (cortex_a5_f2r): Likewise.
7168         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7169         change.
7170         (cortex_a53_f2r): Likewise.
7171         * config/arm/cortex-a7.md
7172         (cortex_a7_branch): Update for attribute change.
7173         (cortex_a7_call): Likewise.
7174         (cortex_a7_alu_imm): Likewise.
7175         (cortex_a7_alu_reg): Likewise.
7176         (cortex_a7_alu_shift): Likewise.
7177         (cortex_a7_mul): Likewise.
7178         (cortex_a7_load1): Likewise.
7179         (cortex_a7_store1): Likewise.
7180         (cortex_a7_load2): Likewise.
7181         (cortex_a7_store2): Likewise.
7182         (cortex_a7_load3): Likewise.
7183         (cortex_a7_store3): Likewise.
7184         (cortex_a7_load4): Likewise.
7185         (cortex_a7_store4): Likewise.
7186         (cortex_a7_fpalu): Likewise.
7187         (cortex_a7_fconst): Likewise.
7188         (cortex_a7_fpmuls): Likewise.
7189         (cortex_a7_neon_mul): Likewise.
7190         (cortex_a7_fpmacs): Likewise.
7191         (cortex_a7_neon_mla: Likewise.
7192         (cortex_a7_fpmuld: Likewise.
7193         (cortex_a7_fpmacd: Likewise.
7194         (cortex_a7_fpfmad: Likewise.
7195         (cortex_a7_fdivs: Likewise.
7196         (cortex_a7_fdivd: Likewise.
7197         (cortex_a7_r2f: Likewise.
7198         (cortex_a7_f2r: Likewise.
7199         (cortex_a7_f_flags: Likewise.
7200         (cortex_a7_f_loads: Likewise.
7201         (cortex_a7_f_loadd: Likewise.
7202         (cortex_a7_f_stores: Likewise.
7203         (cortex_a7_f_stored: Likewise.
7204         (cortex_a7_neon): Likewise.
7205         * config/arm/cortex-a8-neon.md
7206         (cortex_a8_neon_mrc): Update for attribute change.
7207         (cortex_a8_neon_mrrc): Likewise.
7208         (cortex_a8_neon_int_1): Likewise.
7209         (cortex_a8_neon_int_2): Likewise.
7210         (cortex_a8_neon_int_3): Likewise.
7211         (cortex_a8_neon_int_4): Likewise.
7212         (cortex_a8_neon_int_5): Likewise.
7213         (cortex_a8_neon_vqneg_vqabs): Likewise.
7214         (cortex_a8_neon_vmov): Likewise.
7215         (cortex_a8_neon_vaba): Likewise.
7216         (cortex_a8_neon_vaba_qqq): Likewise.
7217         (cortex_a8_neon_vsma): Likewise.
7218         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7219         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7220         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7221         long_scalar): Likewise.
7222         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7223         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7224         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7225         long_scalar_qdd_64_32_long): Likewise.
7226         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7227         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7228         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7229         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7230         (cortex_a8_neon_shift_1): Likewise.
7231         (cortex_a8_neon_shift_2): Likewise.
7232         (cortex_a8_neon_shift_3): Likewise.
7233         (cortex_a8_neon_vshl_ddd): Likewise.
7234         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7235         (cortex_a8_neon_vsra_vrsra): Likewise.
7236         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7237         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7238         (cortex_a8_neon_fp_vsum): Likewise.
7239         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7240         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7241         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7242         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7243         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7244         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7245         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7246         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7247         (cortex_a8_neon_bp_simple): Likewise.
7248         (cortex_a8_neon_bp_2cycle): Likewise.
7249         (cortex_a8_neon_bp_3cycle): Likewise.
7250         (cortex_a8_neon_ldr): Likewise.
7251         (cortex_a8_neon_str): Likewise.
7252         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7253         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7254         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7255         (cortex_a8_neon_vld2_4_regs): Likewise.
7256         (cortex_a8_neon_vld3_vld4): Likewise.
7257         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7258         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7259         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7260         (cortex_a8_neon_vst3_vst4): Likewise.
7261         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7262         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7263         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7264         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7265         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7266         (cortex_a8_neon_mcr): Likewise.
7267         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7268         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7269         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7270         change.
7271         (ca9_neon_mrrc): Likewise.
7272         (cortex_a9_neon_int_1): Likewise.
7273         (cortex_a9_neon_int_2): Likewise.
7274         (cortex_a9_neon_int_3): Likewise.
7275         (cortex_a9_neon_int_4): Likewise.
7276         (cortex_a9_neon_int_5): Likewise.
7277         (cortex_a9_neon_vqneg_vqabs): Likewise.
7278         (cortex_a9_neon_vmov): Likewise.
7279         (cortex_a9_neon_vaba): Likewise.
7280         (cortex_a9_neon_vaba_qqq): Likewise.
7281         (cortex_a9_neon_vsma): Likewise.
7282         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7283         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7284         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7285         long_scalar): Likewise.
7286         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7287         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7288         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7289         long_scalar_qdd_64_32_long): Likewise.
7290         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7291         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7292         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7293         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7294         (cortex_a9_neon_shift_1): Likewise.
7295         (cortex_a9_neon_shift_2): Likewise.
7296         (cortex_a9_neon_shift_3): Likewise.
7297         (cortex_a9_neon_vshl_ddd): Likewise.
7298         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7299         (cortex_a9_neon_vsra_vrsra): Likewise.
7300         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7301         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7302         (cortex_a9_neon_fp_vsum): Likewise.
7303         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7304         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7305         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7306         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7307         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7308         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7309         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7310         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7311         (cortex_a9_neon_bp_simple): Likewise.
7312         (cortex_a9_neon_bp_2cycle): Likewise.
7313         (cortex_a9_neon_bp_3cycle): Likewise.
7314         (cortex_a9_neon_ldr): Likewise.
7315         (cortex_a9_neon_str): Likewise.
7316         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7317         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7318         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7319         (cortex_a9_neon_vld2_4_regs): Likewise.
7320         (cortex_a9_neon_vld3_vld4): Likewise.
7321         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7322         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7323         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7324         (cortex_a9_neon_vst3_vst4): Likewise.
7325         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7326         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7327         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7328         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7329         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7330         (cortex_a9_neon_mcr): Likewise.
7331         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7332         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7333         (cortex_a9_fps): Likewise.
7334         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7335         change.
7336         (cortex_m4_fmuls): Likewise.
7337         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7338         change.
7339         (cortex_r4_mrc): Likewise.
7340         * config/arm/iterators.md: Update comment referring to neon_type.
7341         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7342         (iwmmxt_movsi_insn): Likewise.
7343         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7344         attribute change.
7345         (pj4_core_to_vfp): Likewise.
7346         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7347         attribute change.
7348         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7349         (vfp_fstore): Likewise.
7350         * doc/md.texi: Change references to neon_type to refer to type.
7352 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7354         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7356 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7358         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7359         lglobl if not weak.
7361 2013-09-04  Easwaran Raman  <eraman@google.com>
7363         PR middle-end/57370
7364         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7365         (build_and_add_sum): Use it.
7366         (appears_later_in_bb): Simplify code.
7368 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7370         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7372 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7374         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7376 2013-09-04  Jan Hubicka  <jh@suse.cz>
7378         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7379         * ipa-devirt.c: Include diganostic.h
7380         (odr_type_d): Add types and types_set.
7381         (hash_type_name): Work for types with vtables during LTO.
7382         (odr_hasher::remove): Fix comment; destroy types_set.
7383         (add_type_duplicate): New function,
7384         (get_odr_type): Use it.
7385         (dump_type_inheritance_graph): Dump type duplicates.
7386         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7387         graph.
7388         * tree.c (types_same_for_odr): Give exact answers on types with
7389         virtual tables.
7391 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7393         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7394         explaining their differences.
7396 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7398         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7400 2013-09-03  Jeff Law  <law@redhat.com>
7402         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7403         when not threading through a joiner block.  Pass joiner/no joiner
7404         state to register_jump_thread.
7405         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7406         state from argument rather than implying on path length.
7407         Dump the entire jump thread path into debugging dump.
7408         * tree-flow.h (register_jump_thread): Update prototype.
7410 2013-08-29  Xinliang David Li  <davidxl@google.com>
7412         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7413         Remove a trivial gcc_assert.
7415 2013-08-29  Xinliang David Li  <davidxl@google.com>
7417         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7418         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7419         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7420         Delay base decl alignment adjustment.
7421         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7422         * tree-vectorizer.h: New data structure.
7423         (set_dr_misalignment): New function.
7424         (dr_misalignment): Ditto.
7425         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7426         (vectorizable_load): Ditto.
7427         (ensure_base_align): New function.
7428         (vectorize_loops): Add dbg_cnt support.
7429         (execute_vect_slp): Ditto.
7430         * dbgcnt.def: New debug counter.
7431         * Makefile: New dependency.
7433 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7435         Revert:
7437         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7439         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7441 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7443         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7444         function descriptor.
7446 2013-09-03  Richard Biener  <rguenther@suse.de>
7448         * tree-affine.c (add_elt_to_tree): Fix association issue,
7449         avoid useless converts and make sure to always return a
7450         properly typed result.
7452 2013-09-03  Richard Biener  <rguenther@suse.de>
7454         PR middle-end/57656
7455         * fold-const.c (negate_expr_p): Fix division case.
7456         (negate_expr): Likewise.
7458 2013-09-03  Richard Biener  <rguenther@suse.de>
7460         PR lto/58285
7461         * tree-streamer-out.c: Include tm.h.
7462         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7464 2013-09-03  Jan Hubicka  <jh@suse.cz>
7466         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7468 2013-09-03  Alan Modra  <amodra@gmail.com>
7470         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7471         multilibs to big-endian target and vice versa.
7472         * config/rs6000/t-linux64: Use := assignment on all vars.
7473         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7474         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7475         * config/rs6000/t-linux64le: New file.
7476         * config/rs6000/t-linux64bele: New file.
7477         * config/rs6000/t-linux64lebe: New file.
7479 2013-09-02  Jan Hubicka  <jh@suse.cz>
7481         * ipa-inline-transform.c (inline_transform): Do not
7482         optimize_inline_calls when not optimizing.
7484 2013-09-02  Jan Hubicka  <jh@suse.cz>
7486         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7487         duplicated nodes for assembler names.
7488         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7489         hard registers from assembler name hash.
7491 2013-09-02  Jan Hubicka  <jh@suse.cz>
7493         * ipa-split.c (execute_split_functions): Split externally visible
7494         functions called once.
7496 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7498         PR ipa/58106
7499         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7500         linked list.  When finding the correct duplicate, also consider also
7501         the caller in additon to its inlined_to node.
7503 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7505         * config/aarch64/aarch64-simd-builtins.def
7506         (dup_lane_scalar): Remove.
7507         * config/aarch64/aarch64-simd.md
7508         (aarch64_simd_dup): Add 'w->w' alternative.
7509         (aarch64_dup_lane<mode>): Allow for VALL.
7510         (aarch64_dup_lane_scalar<mode>): Remove.
7511         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7512         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7513         (aarch64_get_lane_unsigned<mode>): Likewise.
7514         (aarch64_get_lane<mode>): Likewise.
7515         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7516         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7517         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7518         (VCON): Change container of V2SF.
7519         (vswap_width_name): Likewise.
7520         * config/aarch64/arm_neon.h
7521         (__aarch64_vdup_lane_any): New.
7522         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7523         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7524         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7526 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7528         PR middle-end/56382
7529         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7530         the source or the destination is a single hard register.
7532 2013-09-02  Richard Biener  <rguenther@suse.de>
7534         PR middle-end/57511
7535         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7536         non-linear SCEVs.
7538 2013-09-02  Richard Biener  <rguenther@suse.de>
7540         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7541         arithmetic to sizetype.
7543 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7545         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7546         Find auto-increment use both before and after candidate.
7548 2013-09-02  Marek Polacek  <polacek@redhat.com>
7550         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7552 2013-09-01  Jan Hubicka  <jh@suse.cz>
7554         * Makefile.in: Add ipa-profile.o
7555         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7556         * cgraph.c (struct cgraph_propagate_frequency_data,
7557         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7558         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7559         * cgraph.h (cgraph_propagate_frequency): Remove.
7560         * ipa-inline-analysis.c: Include ipa-utils.h;
7561         drop duplicated cfgloop.h.
7562         (inline_update_callee_summaries): Update.
7563         * ipa-profile.c: New file.
7564         * ipa-utils.h (ipa_propagate_frequency): Declare.
7565         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7566         data-streamer.h, value-prof.h.
7567         (symtab_remove_unreachable_nodes): Update profile.
7568         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7569         account_time_size, cmp_counts, dump_histogram,
7570         ipa_profile_generate_summary, ipa_profile_write_summary,
7571         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7572         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7573         Move to ipa-profile.c.
7575 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7577         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7579 2013-09-01  Jan Hubicka  <jh@suse.cz>
7581         * common.opt (fdevirtualize-speculatively): New function.
7582         * invoke.texi (fdevirtualize-speculatively): Document.
7583         * ipa-devirt.c: Include ipa-inline.h
7584         (likely_target_p): New function.
7585         (ipa_devirt): New function.
7586         (gate_ipa_devirt): New function.
7587         (pass_data_ipa_devirt): New static var.
7588         (pass_ipa_devirt): Likewise.
7589         (make_pass_ipa_devirt): New function.
7590         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7591         (common_handle_option): Disable devirtualization when
7592         value range profiling is available.
7593         * passes.def (pass_ipa_devirt): Add.
7594         * timever.def (TV_IPA_DEVIRT): New timevar.
7595         * tree-pass.h (make_pass_ipa_devirt):
7597 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7599         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7600         include sanitize(undefined).
7602 2013-08-31  Diego Novillo  <dnovillo@google.com>
7604         * Makefile.in (TREE_CORE_H): Define.
7605         (TREE_H): Use.
7606         (GTFILES): Add tree-core.h.
7607         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7608         size the array.
7609         * tree-core.h: New file.
7610         Move all data structures, enum, typedefs, global
7611         declarations and constants from ...
7612         * tree.h: ... here.
7614 2013-08-31  Jan Hubicka  <jh@suse.cz>
7616         * bulitins.c (expand_builtin): Do not early exit for gcov
7617         instrumented functions.
7619 2013-08-31  Marek Polacek  <polacek@redhat.com>
7621         * ubsan.c: Include tm_p.h.
7623 2013-08-31  Jan Hubicka  <jh@suse.cz>
7625         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7626         warning.
7628         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7629         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7630         * tree-cfg.c (verify_gimple_label): ... here.
7631         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7632         (ipa_merge_profiles): New function.
7633         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7634         (lto_input_function_body): Likewise.
7635         * ipa-utils.h (ipa_merge_profiles): Declare.
7636         * lto-streamer.h (lto_input_function_body): Update prototype.
7637         (emit_label_in_global_context_p): Remove.
7638         * lto-symtab.c: Include ipa-utils.h
7639         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7641 2013-08-31  Jan Hubicka  <jh@suse.cz>
7643         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7645 2013-08-31  Jan Hubicka  <jh@suse.cz>
7647         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7649 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7651         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7652         "cmp" RTX before signed_comparison_operator check to account
7653         for "code" changes.
7655 2013-08-30  Jan Hubicka  <jh@suse.cz>
7657         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7658         (detect_type_change_1): Rename to ...
7659         (detect_type_change): ... this one; early return on non-polymorphic
7660         types.
7661         (detect_type_change_ssa): Add comp_type parameter; update
7662         use of detect_type_change.
7663         (compute_complex_assign_jump_func): Add param_type parameter;
7664         update use of detect_type_change_ssa.
7665         (compute_complex_ancestor_jump_func): Likewise.
7666         (ipa_get_callee_param_type): New function.
7667         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7668         update calls to the jump function computation functions.
7670 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7671             Steven Bosscher  <steven@gcc.gnu.org>
7673         * cfgrtl.c (fixup_new_cold_bb): New routine.
7674         (commit_edge_insertions): Invoke fixup_partitions.
7675         (find_partition_fixes): New routine.
7676         (fixup_partitions): Ditto.
7677         (verify_hot_cold_block_grouping): Update comments.
7678         (rtl_verify_edges): Invoke find_partition_fixes.
7679         (rtl_verify_bb_pointers): Update comments.
7680         (rtl_verify_bb_layout): Ditto.
7681         * basic-block.h (probably_never_executed_edge_p): Declare.
7682         (fixup_partitions): Ditto.
7683         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7684         * bb-reorder.c (sanitize_hot_paths): New function.
7685         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7686         sanitize_hot_paths.
7687         * predict.c (probably_never_executed_edge_p): New routine.
7688         * cfg.c (check_bb_profile): Add partition insanity warnings.
7690 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7692         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7694 2013-08-30  Marek Polacek  <polacek@redhat.com>
7696         * Makefile.in (ubsan.o): Add.
7697         (c-family/c-ubsan.o): Add.
7698         (builtins.o): Add ubsan.h dependency.
7699         * ubsan.h: New file.
7700         * ubsan.c: New file.
7701         * common.opt: Add -fsanitize=undefined option.
7702         (flag_sanitize): Add variable.
7703         (fsanitize=): Add option.  Add Driver.
7704         (fsanitize=thread): Remove option.
7705         (fsanitize=address): Likewise.
7706         (static-libubsan): New option.
7707         * doc/invoke.texi: Document the new flag and -static-libubsan.
7708         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7709         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7710         * builtin-attrs.def (ATTR_COLD): Define.
7711         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7712         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7713         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7714         * flag-types.h (sanitize_code): New enum.
7715         * opts.c (common_handle_option): Parse command line arguments
7716         of -fsanitize=.  Add -fsanitize=unreachable option.
7717         * varasm.c (get_variable_section): Adjust.
7718         (assemble_noswitch_variable): Likewise.
7719         (assemble_variable): Likewise.
7720         (output_constant_def_contents): Likewise.
7721         (categorize_decl_for_section): Likewise.
7722         (place_block_symbol): Likewise.
7723         (output_object_block): Likewise.
7724         * builtins.def: Likewise.
7725         * toplev.c (compile_file): Likewise.
7726         (process_options): Likewise.
7727         * cppbuiltin.c: Likewise.
7728         * tsan.c (tsan_pass): Likewise.
7729         (tsan_gate): Likewise.
7730         (tsan_gate_O0): Likewise.
7731         * cfgexpand.c (partition_stack_vars): Likewise.
7732         (expand_stack_vars): Likewise.
7733         (defer_stack_allocation): Likewise.
7734         (expand_used_vars): Likewise.
7735         * cfgcleanup.c (old_insns_match_p): Likewise.
7736         * asan.c (asan_finish_file): Likewise.
7737         (asan_instrument): Likewise.
7738         (gate_asan): Likewise.
7739         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7740         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7741         (asan_global_struct): Use pointer_sized_int_node instead
7742         calling build_nonstandard_integer_type.
7743         (initialize_sanitizer_builtins): Likewise.
7744         (asan_finish_file): Likewise.
7745         * gcc.c: Document %{%:function(args):X}.
7746         (static_spec_functions): Add sanitize.
7747         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7748         store funcval != NULL there.
7749         (do_spec_1): Adjust handle_spec_function caller.
7750         (handle_braces): Allow %:function(args) as condition.
7751         (sanitize_spec_function): New function.
7752         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7753         (LIBUBSAN_SPEC): Likewise.
7754         (LIBUBSAN_EARLY_SPEC): Likewise.
7755         (SANITIZER_SPEC): Handle libubsan.
7756         (SANITIZER_EARLY_SPEC): Likewise.
7757         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7758         instead of fsanitize=address.
7759         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7760         instead of fsanitize=address*.
7761         * builtins.c: Include ubsan.h.
7762         (fold_builtin_0): Instrument __builtin_unreachable.
7763         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7764         instead of flag_asan.
7765         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7766         (pointer_sized_int_node): Define.
7767         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7769 2013-08-30  Mike Stump  <mikestump@comcast.net>
7771         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7772         with RE patterns.
7774 2013-08-29  Jan Hubicka  <jh@suse.cz>
7776         * cgraph.c (cgraph_function_body_availability): Handle weakref
7777         correctly.
7778         * passes.def: Remove pass_fixup_cfg.
7779         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7780         track when we need to remove functions.
7781         (gate_ipa_inline): Execute inlining always; add comment why.
7782         (pass_data_ipa_inline): Remove TODO_remove_functions.
7783         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7784         do not produce summaries.
7785         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7786         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7787         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7788         constant pool and vtable.
7790 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7792         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7793         New arm_neon.h's internal macros to specify 64-bit constants.
7794         Avoid using stdint.h's macros.
7796 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7798         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7799         before applying REG_P.
7801 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7803         PR tree-optimization/58277
7804         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7805         after seeing too many stmts with vdef in between dombb and current
7806         bb, invalidate everything.
7808 2013-08-30  Richard Biener  <rguenther@suse.de>
7810         * fold-const.c (fold_single_bit_test): Fix overflow test.
7812 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7814         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7815         and which can live in a register, always retrieve the value on entry.
7816         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7817         passed by invisible reference specially.
7818         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7819         (vt_add_function_parameter): Correctly deal with a parameter passed by
7820         invisible reference.
7822 2013-08-30  Jan Hubicka  <jh@suse.cz>
7824         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7826 2013-08-30  Richard Biener  <rguenther@suse.de>
7828         PR tree-optimization/58228
7829         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7830         allow invariant loads in nested loop vectorization.
7832 2013-08-30  Richard Biener  <rguenther@suse.de>
7834         PR tree-optimization/58223
7835         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7836         (has_anti_or_output_dependence): ... this and adjust to also
7837         look for output dependences.
7838         (mark_nodes_having_upstream_mem_writes): Adjust.
7839         (rdg_flag_uses): Likewise.
7841 2013-08-30  Richard Biener  <rguenther@suse.de>
7843         PR tree-optimization/58010
7844         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7845         assert that we have a loop-closed PHI.
7847 2013-08-29  Jan Hubicka  <jh@suse.cz>
7849         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7850         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7851         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7852         (lto_free_function_in_decl_state_for_node): New function.
7854 2013-08-29  Xinliang David Li  <davidxl@google.com>
7856         * loop-unroll.c (report_unroll_peel): Minor message change.
7857         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7858         Emit alignment peeling message with default -fopt-info.
7859         (vect_loop_versioning): Emit loop version info message.
7860         * tree-vectorizer.c (vectorize_loops): Minor message change.
7861         (execute_vect_slp): Ditto.
7863 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7865         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7866         of the clone from the DECL_NAME of the original function.
7868 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7870         * passes.c (register_pass): Add overload.
7871         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7873 2013-08-29  Jan Hubicka  <jh@suse.cz>
7875         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7876         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7877         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7878         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7879         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7880         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7881         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7882         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7884 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7886         * dumpfile.c (dump_loc): Output column number.
7887         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7888         * doc/invoke.texi: Document optall -fopt-info flag.
7889         * profile.c (read_profile_edge_counts): Use new dump framework.
7890         (compute_branch_probabilities): Ditto.
7891         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7892         when pass not in any opt group.
7893         * pass_manager.h (pass_manager::get_pass_profile): New method.
7894         * value-prof.c (check_counter): Use new dump framework.
7895         (check_ic_target): Ditto.
7896         * coverage.c (get_coverage_counts): Ditto.
7897         (coverage_init): Setup new dump framework.
7899 2013-08-29  Richard Biener  <rguenther@suse.de>
7901         PR tree-optimization/58246
7902         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7903         handle the dominance check inside a basic-block.
7905 2013-08-29  Richard Biener  <rguenther@suse.de>
7907         PR middle-end/57287
7908         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7909         of default defs that appear in abnormal PHI nodes.
7911 2013-08-29  Richard Biener  <rguenther@suse.de>
7913         PR tree-optimization/57685
7914         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7915         single-use operands to avoid exponential complexity.
7917 2013-08-28  Dehao Chen  <dehao@google.com>
7919         * ipa-inline.c (edge_badness): Fix integer underflow.
7921 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7923         * gtm-builtins.def (_ITM_free): Declare leaf.
7925 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7927         PR target/58067
7928         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7929         (*tls_local_dynamic_base_64_largepic): Likewise.
7930         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7931         Remove predicate from call operand.
7932         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7933         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7935 2013-08-28  Jeff Law  <law@redhat.com>
7937         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7938         checks for the number of predecessors and successors allowed.
7939         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7940         which require copying a joiner block if there is a request which
7941         is a subpath that requires no joiner block copying.
7943 2013-08-28  Jan Hubicka  <jh@suse.cz>
7945         * lto-streamer-out.c (DFS_write_tree_body): Drop
7946         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
7947         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
7948         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
7949         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
7950         Do not read DECL_ERROR_ISSUED.
7951         (unpack_ts_decl_with_vis_value_fields): Do not read
7952         DECL_DEFER_OUTPUT.
7953         (lto_input_ts_binfo_tree_pointers): Do not read
7954         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
7955         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
7956         write DECL_ERROR_ISSUED..
7957         (pack_ts_decl_with_vis_value_fields): Do not write
7958         DECL_DEFER_OUTPUT.
7959         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
7960         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
7961         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
7962         * tree.h (tree_decl_common): Update comment.
7963         (DECL_ERROR_ISSUED): Remove.
7965 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7967         PR middle-end/58257
7968         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
7970 2013-08-28  Jan Hubicka  <jh@suse.cz>
7972         * builtins.def (free): Declare leaf.
7974 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
7976         * gdbhooks.py: New.
7977         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
7978         * configure: Regenerate.
7980 2013-08-27  Martin Jambor  <mjambor@suse.cz>
7982         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
7983         (ipa_ancestor_jf_data): Likewise.
7984         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
7985         (ipa_get_jf_pass_through_type_preserved): New function.
7986         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
7987         (ipa_get_jf_ancestor_type_preserved): New function.
7988         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
7989         (ipa_get_jf_ancestor_result): Likewise.
7990         (propagate_vals_accross_pass_through): Use
7991         ipa_get_jf_pass_through_result to do all the value mappings.
7992         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
7993         type_preserved flag.
7994         (ipa_set_jf_cst_copy): New function.
7995         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
7996         (ipa_set_jf_arith_pass_through): Likewise.
7997         (ipa_set_ancestor_jf): Likewise.
7998         (compute_complex_assign_jump_func): Set type_preserved instead of
7999         punting.
8000         (ipa_compute_jump_functions_for_edge): Likewise.
8001         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8002         (update_jump_functions_after_inlining): Update type_preserved.
8003         Explicitely create jump functions when combining one with pass_through.
8004         (ipa_write_jump_function): Stream the type_preserved flags.
8005         (ipa_read_jump_function): Likewise.
8007 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8008             Aldy Hernandez  <aldyh@redhat.com>
8010         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8011         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8012         * function.h (struct function): Add has_force_vect_loops and
8013         has_simduid_loops.
8014         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8015         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8016         handle it.
8017         * gimple.def: Update CLAUSES comments.
8018         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8019         (gimple_build_omp_for): Add argument to prototype.
8020         (gimple_omp_for_kind): New.
8021         (gimple_omp_for_set_kind): New.
8022         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8023         GOVD_DATA_SHARE_CLASS.
8024         (enum omp_region_type): Add ORT_SIMD.
8025         (gimple_add_tmp_var): Handle ORT_SIMD.
8026         (gimplify_var_or_parm_decl): Same.
8027         (is_gimple_stmt): Same.
8028         (omp_firstprivatize_variable): Same.
8029         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8030         (omp_notice_variable): Handle ORT_SIMD.
8031         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8032         (omp_check_private): Handle ORT_SIMD.
8033         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8034         OMP_CLAUSE_SAFELEN.
8035         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8036         Handle OMP_CLAUSE_LASTPRIVATE.
8037         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8038         OMP_CLAUSE_SAFELEN.
8039         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8040         (gimplify_expr): Handle OMP_SIMD.
8041         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8042         (expand_GOMP_SIMD_VF): New.
8043         (expand_GOMP_SIMD_LAST_LANE): New.
8044         * internal-fn.def (GOMP_SIMD_LANE): New.
8045         (GOMP_SIMD_VF): New.
8046         (GOMP_SIMD_LAST_LANE): New.
8047         * omp-low.c: Include target.h.
8048         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8049         OMP_CLAUSE_SAFELEN.
8050         (check_omp_nesting_restrictions): Same.
8051         (omp_max_vf): New.
8052         (lower_rec_simd_input_clauses): New.
8053         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8054         OMP_CLAUSE_LINEAR.
8055         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8056         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8057         (expand_omp_build_assign): New.
8058         (expand_omp_for_init_counts): New.
8059         (expand_omp_for_init_vars): New.
8060         (extract_omp_for_update_vars): New.
8061         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8062         and rewrite accordingly.
8063         (expand_omp_simd): New.
8064         (expand_omp_for): Use expand_omp_simd.
8065         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8066         (lower_omp_for): Do not lower the body.
8067         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8068         in their own loops.
8069         * tree-flow.h (find_omp_clause): Remove prototype.
8070         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8071         forcing vectorization of the loop, or if flag_tree_vectorize.
8072         (gate_tree_if_conversion): Similarly.
8073         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8074         gimple_build_omp_for.
8075         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8076         * tree-parloops (create_parallel_loop): Pass kind argument to
8077         gimple_build_omp_for.
8078         * tree-pretty-print.c (dump_omp_clause): Add cases for
8079         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8080         OMP_CLAUSE__SIMDUID_.
8081         (dump_generic_node): Handle OMP_SIMD.
8082         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8083         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8084         unroll OMP_SIMD loops here.
8085         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8086         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8087         loop->safelen.
8088         (vect_analyze_data_refs): Handle simd loops.
8089         * tree-vect-loop.c (vectorizable_live_operation): Handle
8090         IFN_GOMP_SIMD*.
8091         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8092         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8093         (vectorizable_load): Same.
8094         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8095         (struct simduid_to_vf): New.
8096         (simduid_to_vf::hash): New.
8097         (simduid_to-vf::equal): New.
8098         (struct simd_array_to_simduid): New.
8099         (simd_array_to_simduid::hash): New.
8100         (simd_array_to_simduid::equal): New.
8101         (adjust_simduid_builtins): New.
8102         (struct note_simd_array_uses_struct): New.
8103         (note_simd_array_uses_cb): New.
8104         (note_simd_array_uses): New.
8105         (vectorize_loops): Handle simd hints and adjust simd builtins
8106         accordingly.
8107         * tree-vectorizer.h (struct _stmt_vec_info): Add
8108         simd_lane_access_p field.
8109         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8110         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8111         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8112         (omp_clause_code_name): Same.
8113         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8114         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8115         * tree.def (OMP_SIMD): New entry.
8116         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8117         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8118         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8119         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8120         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8121         (OMP_CLAUSE_LINEAR_STEP): New.
8122         (OMP_CLAUSE_SAFELEN_EXPR): New.
8123         (OMP_CLAUSE__SIMDUID__DECL): New.
8124         (find_omp_clause): New prototype.
8126 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8128         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8129         Haswell processor detection.
8131 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8133         PR target/57927
8134         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8135         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8136         AVX2 capable processors.
8138 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8140         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8141         of vget_low_* with implementations in terms of other intrinsics.
8143 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8145         PR middle-end/57219
8146         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8147         values to -1, 0 and 1.
8149 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8151         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8152         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8153         (<optab><mode>3_insn): Remove.
8154         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8155         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8156         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8157         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8158         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8159         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8160         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8161         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8162         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8163         (ror<mode>3_insn): Likewise.
8164         * config/aarch64/predicates.md (aarch64_simd_register): New.
8166 2013-08-27  Richard Biener  <rguenther@suse.de>
8168         PR tree-optimization/57521
8169         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8170         one edge is non-critical.
8171         (find_phi_replacement_condition): Make sure to use a non-critical
8172         edge.  Cleanup and remove old bug workarounds.
8173         (bb_postdominates_preds): Remove.
8174         (if_convertible_loop_p_1): Do not compute post-dominators.
8175         (combine_blocks): Do not free post-dominators.
8176         (main_tree_if_conversion): Likewise.
8177         (pass_data_if_conversion): Add TODO_verify_ssa.
8179 2013-08-27  DJ Delorie  <dj@redhat.com>
8181         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8183 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8185         * function.c (assign_parm_find_data_types): Set passed_mode and
8186         nominal_mode to the TYPE_MODE of nominal_type for the built
8187         pointer type in case of the struct-pass-by-reference.
8189 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8191         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8192         (UINT16_TYPE): Change default to "unsigned int".
8194         * config/avr/avr.opt (mfract-convert-truncate): New option.
8195         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8196         is set, round negative fractional integers according to n1169
8197         when converting to integer types.
8199 2013-08-26  Jan Hubicka  <jh@suse.cz>
8201         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8202         methods can not be called indirectly when their address is not taken.
8204 2013-08-26  Jan Hubicka  <jh@suse.cz>
8206         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8207         ctor_for_folding.
8209 2013-08-26  Jan Hubicka  <jh@suse.cz>
8211         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8212         can be unshared.
8214 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8216         * reload.c (find_valid_class): Allow classes that do not include
8217         FIRST_PSEUDO_REGISTER - 1.
8219 2013-08-26  Jan Hubicka  <jh@suse.cz>
8221         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8222         fix edge count/frequency when speculation failed; fix type check
8223         for the direct call.
8225 2013-08-26  Jan Hubicka  <jh@suse.cz>
8227         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8229 2013-08-26  Jan Hubicka  <jh@suse.cz>
8231         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8232         to be changed by edge redirection.
8234 2013-08-26  Jan Hubicka  <jh@suse.cz>
8236         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8237         formating; add sanity check.
8238         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8239         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8240         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8241         (resolve_noninline_speculation): Update callee keys, too.
8243 2013-08-26  Jan Hubicka  <jh@suse.cz>
8245         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8246         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8248 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8250         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8251         into proper place.
8253 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8255         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8256         (x86_64_elf_select_section): Ditto.
8257         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8258         arguments.
8259         (ix86_pass_by_reference): Ditto.
8260         (output_set_got): Ditto.
8261         (ix86_unary_operator_ok): Ditto.
8262         (ix86_expand_builtin): Ditto.
8264 2013-08-23  Jan Hubicka  <jh@suse.cz>
8266         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8268 2013-08-23  Jan Hubicka  <jh@suse.cz>
8270         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8271         (tree_decl_with_vis): Add FINAL field.
8273 2013-08-23  Jeff Law  <law@redhat.com>
8275         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8276         the debugging dump when the expression is fully redundant.
8278 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8280         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8281         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8282         * pretty-print.c (pp_formatted_text_data): Likewise.
8283         (pp_write_text_to_stream): Likewise.
8284         (pp_write_text_as_dot_label_to_stream): Likewise.
8285         (pp_append_r): Likewise.
8286         (pp_format): Likewise.
8287         (pp_flush): Likewise.
8288         (pp_clear_output_area): Likewise.
8289         (pp_append_text): Likewise.
8290         (pp_formatted_text): Likewise.
8291         (pp_remaining_character_count_for_line): Likewise.
8292         (pp_newline): Likewise.
8293         (pp_character): Likewise.
8294         (output_buffer::~output_buffer): Define.
8295         (pretty_printer::~pretty_printer): Destruct output buffer.
8296         * pretty-print.h (output_buffer::~output_buffer): Declare.
8297         (pretty_printer::~pretty_printer): Declare virtual.
8299 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8301         PR other/57324
8302         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8303         HOST_WIDE_INT_M1U): New macros.
8304         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8305         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8306         unsigned -1 for lshift.
8307         * cse.c (cse_insn): Likewise.
8308         * double-int.c (rshift_double, lshift_double): Likewise.
8309         * builtins.c (fold_builtin_bitop): Likewise.
8310         * combine.c (force_to_mode): Likewise.
8311         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8312         * simplify-rtx.c (simplify_const_unary_operation,
8313         simplify_const_binary_operation): Likewise.
8314         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8315         check_va_list_escapes): Likewise.
8316         * rtlanal.c (nonzero_bits1): Likewise.
8317         * expmed.c (expand_smod_pow2): Likewise.
8318         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8320 2013-08-23  Jan Hubicka  <jh@suse.cz>
8322         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8323         as having address taken.
8325 2013-08-23  Jan Hubicka  <jh@suse.cz>
8327         * ipa-utils.h (method_class_type): Declare.
8328         * ipa-devirt.c (method_class_type): Export.
8330         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8331         do not walk base classes of anonymous types.
8333 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8335         PR rtl-optimization/58220
8336         PR regression/58221
8337         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8338         handle SEQUENCE insns properly.
8340 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8342         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8343         definition.
8344         (pp_newline_and_indent): Likewise.
8345         (pp_separate_with): Likewise.
8346         * pretty-print.c (pp_newline_and_flush): Define.
8347         (pp_newline_and_indent): Likewise.
8348         (pp_separate_with): Likewise.
8350 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8352         PR target/58218
8353         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8354         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8356 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8358         * config/i386/predicates.md (ext_sse_reg_operand): New.
8359         * config/i386/i386.md (*movti_internal): Use
8360         predicate to determine if EVEX is needed.
8361         (*movsi_internal): Ditto.
8362         (*movdf_internal): Ditto.
8363         (*movsf_internal): Ditto.
8364         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8366 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8368         PR tree-optimization/58209
8369         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8370         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8371         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8372         emit POINTER_PLUS_EXPR.
8373         (create_tailcall_accumulator): For pointer result type accumulate in
8374         sizetype type.
8376 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8378         * configure.ac: Add backslashes missing from the last change.
8379         * configure: Regenerate.
8381 2013-08-22  Jan Hubicka  <jh@susue.cz>
8383         * ipa.c (function_and_variable_visibility): First remember function
8384         was global and then make it local.
8386 2013-08-22  Julian Brown  <julian@codesourcery.com>
8388         * configure.ac: Add aarch64 to list of arches which use "nop" in
8389         debug_line test.
8390         * configure: Regenerate.
8392 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8394         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8395         gnu_libc_has_function.
8396         * config/s390/tpf.h: Likewise.
8398 2013-08-22  Jan Hubicka  <jh@susue.cz>
8400         * timevar.c (validate_phases): Add cast.
8402 2013-08-22  Jan Hubicka  <jh@susue.cz>
8404         * timevar.c (validate_phases): Use size_t for memory.
8405         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8407 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8409         * pretty-print.h (output_buffer::output_buffer): Declare.
8410         (pretty_printer::pretty_printer): Likewise.
8411         (pp_construct): Remove.
8412         * pretty-print.c (output_buffer::output_buffer): Define.
8413         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8414         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8415         (print_gimple_expr): Likewise.
8416         (print_gimple_seq): Likewise.
8417         (gimple_dump_bb): Likewise.
8418         * sched-vis.c (dump_value_slim): Likewise.
8419         (dump_insn_slim): Likewise.
8420         (dump_rtl_slim): Likewise.
8421         (str_pattern_slim): Likewise.
8422         * tree-mudflap.c (mf_varname_tree): Likewise.
8423         * graph.c (print_graph_cfg): Likewise.
8424         (start_graph_dump): Likewise.
8425         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8426         placement-new.
8427         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8428         pretty printer initialization.
8429         * coretypes.h (diagnostic_context): Remove superflous type alias
8430         declaration.
8431         (pretty_printer): Likewise.  Declare directly as a class.
8432         (pretty_print_info): Remove declaration as class.
8433         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8434         and pp_clear_output_area.
8435         (asan_add_global): Likewise.
8437 2013-08-22  Jan Hubicka  <jh@suse.cz>
8439         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8440         * ipa-utils.h (update_type_inheritance_graph): Declare.
8441         (possible_polymorphic_call_target_p): Declare.
8442         (possible_polymorphic_call_target_p): New.
8443         * ipa-devirt.c: Update toplevel comments.
8444         (cached_polymorphic_call_targets): Move up.
8445         (odr_type_d): Move ID down.
8446         (polymorphic_type_binfo_p): Update comment.
8447         (odr_hasher::remove): Likewise;
8448         (get_odr_type): Set anonymous_namespace.
8449         (dump_odr_type): Dump it.
8450         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8451         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8452         (record_binfo): Add comment.
8453         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8454         built.
8455         (devirt_node_removal_hook): Do not iCE when cache is freed.
8456         (possible_polymorphic_call_target_p): New predicate.
8457         (update_type_inheritance_graph): New function.
8459 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8460             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8461             Sergey Lega  <sergey.s.lega@intel.com>
8462             Anna Tikhonova  <anna.tikhonova@intel.com>
8463             Ilya Tocar  <ilya.tocar@intel.com>
8464             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8465             Ilya Verbin  <ilya.verbin@intel.com>
8466             Kirill Yukhin  <kirill.yukhin@intel.com>
8467             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8469         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8470         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8471         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8472         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8473         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8474         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8475         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8476         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8477         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8478         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8479         OPT_mavx512pf, OPT_mavx512er cases.
8480         * config/i386/constraints.md (v): New constraint.
8481         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8482         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8483         (bit_AVX512CD): New.
8484         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8485         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8486         * config/i386/i386-c.c (ix86_target_macros_internal):
8487         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8488         __AVX512PF__.
8489         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8490         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8491         * config/i386/i386.c (regclass_map, dbx_register_map)
8492         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8493         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8494         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8495         -mavx512pf options.
8496         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8497         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8498         -mavx512pf options.  Fix formatting.
8499         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8500         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8501         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8502         -mavx512cd, -mavx512pf options.
8503         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8504         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8505         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8506         ALL_SSE_REGS.
8507         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8508         (ix86_set_reg_reg_cost): Ditto.
8509         (x86_order_regs_for_local_alloc): Ditto.
8510         (MAX_VECT_LEN): Extend to 64-byte.
8511         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8512         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8513         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8514         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8515         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8516         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8517         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8518         (SSE_REG_MODE_P): Support new modes.
8519         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8520         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8521         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8522         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8523         (REG_CLASS_CONTENTS): Add new registers.
8524         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8525         (EXT_REX_SSE_REGNO_P): New.
8526         (HI_REGISTER_NAMES): Add new registers.
8527         * config/i386/i386.md: Define constants for new registers.
8528         (mode): Add new 512-bit modes.
8529         (prefix): Support evex prefix.
8530         (isa): Support avx512f, noavx512f, fma_avx512f.
8531         (ssemodesuffix): Add new 512-bit modes.
8532         (movxi): New.
8533         (*movxi_internal_avx512f): Ditto.
8534         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8535         Support MODE_XI.
8536         (*movsi_internal): Likewise.
8537         (*movdf_internal): Likewise.
8538         (*movsf_internal): Likewise.
8539         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8540         (<code><mode>3): Likewise.
8541         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8542         New.
8543         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8544         with the new constraint "v".
8545         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8546         modes.
8547         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8548         with the new constraint "v".
8549         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8550         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8551         (*<plusminus_insn><mode>3): Likewise.
8552         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8553         (*mul<mode>3): Likewise.
8554         (<sse>_vmmul<mode>3): Likewise.
8555         (<sse>_div<mode>3): Likewise.
8556         (<sse>_vmdiv<mode>3): Likewise.
8557         (<sse>_sqrt<mode>2): Likewise.
8558         (<sse>_vmsqrt<mode>2): Likewise.
8559         (*<code><mode>3_finite): Likewise.
8560         (*<code><mode>3) <smaxmin>: Likewise.
8561         (<sse>_vm<code><mode>3): Likewise.
8562         (*<code><mode>3) <any_logic>: Likewise.
8563         (*fma_fmadd_<mode>): Likewise.
8564         (*fma_fmsub_<mode>): Likewise.
8565         (*fma_fnmadd_<mode>): Likewise.
8566         (*fma_fnmsub_<mode>): Likewise.
8567         (*fma_fmaddsub_<mode>): Likewise.
8568         (*fma_fmsubadd_<mode>): Likewise.
8569         (*fmai_fmadd_<mode>): Likewise.
8570         (*fmai_fmsub_<mode>): Likewise.
8571         (*fmai_fnmadd_<mode>): Likewise.
8572         (*fmai_fnmsub_<mode>): Likewise.
8573         (sse_cvtsi2ss): Likewise.
8574         (sse_cvtsi2ssq): Likewise.
8575         (sse_cvtss2si): Likewise.
8576         (sse_cvtss2si_2): Likewise.
8577         (sse_cvtss2siq): Likewise.
8578         (sse_cvtss2siq_2): Likewise.
8579         (sse_cvttss2si): Likewise.
8580         (sse_cvtss2siq_2): Likewise.
8581         (float<sseintvecmodelower><mode>2): Likewise.
8582         (sse2_cvtsd2si_2): Likewise.
8583         (sse2_cvtsd2siq_2): Likewise.
8584         (*<plusminus_insn><mode>3): Likewise.
8585         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8586         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8587         (ashr<mode>3): Likewise.
8588         (<shift_insn><mode>3): Likewise.
8589         (avx2_<code><mode>3): Likewise.
8590         (*avx2_<code><mode>3): Likewise.
8591         (*andnot<mode>3): Likewise.
8592         (*<code><mode>3) <any_logic>: Likewise.
8593         (abs<mode>2): Likewise.
8594         (avx2_permvar<mode>): Likewise.
8595         (avx2_perm<mode>_1): Likewise.
8596         (*avx_vpermilp<mode>): Likewise.
8597         (avx_vpermilvar<mode>3): Likewise.
8598         (avx2_ashrv<mode>): Likewise.
8599         (avx2_<shift_insn>v<mode>): Likewise.
8600         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8601         -mavx512cd.
8602         * doc/rtl.texi: Document XImode.
8604 2013-08-21  Jeff Law  <law@redhat.com>
8606         * tree-flow.h (register_jump_thread): Pass vector of edges
8607         instead of each important edge.
8608         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8609         thread path into a vector and pass that to register_jump_thread.
8610         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8611         passed in edge vector to the current 3-edge form.
8613         Revert:
8614         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8616         * dce.c (fini_dce): Call df_analyze again just in case
8617         delete_unmarked_insns removed anything.
8619 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8621         * reload.h (struct reg_equivs): Rename to ..
8622         (struct reg_equivs_s): .. this.
8624 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8626         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8628 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8630         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8632 2013-08-21  Jeff Law  <law@redhat.com>
8634         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8635         simplify assignments too.  If the RHS collapses to a singleton
8636         range, then return the value for the range.
8638 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8640         * config/i386/sse.md (V16): Rename to...
8641         (VMOVE): this.
8642         (mov<mode>): Update iterator name.
8643         (*mov<mode>_internal): Ditto.
8644         (push<mode>1): Ditto.
8645         (movmisalign<mode>): Ditto.
8647 2013-08-20  Jan Hubicka  <jh@suse.cz>
8649         PR bootstrap/58186
8650         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8651         entry for direct edges.
8652         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8654 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8656         Revert my last two changes, r201865 and r201864:
8658         Revert r201865:
8659         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8661         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8662         instances can own GC refs.
8664         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8665         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8666         (gcc::context::gt_pch_nx): Likewise.
8667         (gcc::context::gt_pch_nx):  Likewise.
8668         * ggc.h (gt_ggc_mx <T>): New.
8669         (gt_pch_nx_with_op <T>): New.
8670         (gt_pch_nx <T>): New.
8671         * passes.c (opt_pass::gt_ggc_mx): New.
8672         (opt_pass::gt_pch_nx): New.
8673         (opt_pass::gt_pch_nx_with_op): New.
8674         (pass_manager::gt_ggc_mx): New.
8675         (pass_manager::gt_pch_nx): New.
8676         (pass_manager::gt_pch_nx_with_op): New.
8677         (pass_manager::operator new): Use
8678         ggc_internal_cleared_alloc_stat rather than xcalloc.
8679         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8680         (pass_manager::gt_ggc_mx): New.
8681         (pass_manager::gt_pch_nx): New.
8682         (pass_manager::gt_pch_nx_with_op): New.
8683         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8684         (opt_pass::operator new): New.
8685         (opt_pass::gt_ggc_mx): New.
8686         (opt_pass::gt_pch_nx): New.
8687         (opt_pass::gt_pch_nx_with_op): New.
8689         Revert r201864:
8690         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8692         * Makefile.in (GTFILES): Add context.h.
8693         * context.c (gcc::context::operator new): New.
8694         (gcc::context::gt_ggc_mx): New.
8695         (gcc::context::gt_pch_nx): New.
8696         (gcc::context::gt_pch_nx): New.
8697         * context.h (gcc::context): Add GTY((user)) marking.
8698         (gcc::context::operator new): New.
8699         (gcc::context::gt_ggc_mx): New.
8700         (gcc::context::gt_pch_nx): New.
8701         (gcc::context::gt_pch_nx): New.
8702         (g): Add GTY marking.
8703         (gt_ggc_mx (gcc::context *)): New.
8704         (gt_pch_nx (gcc::context *)): New.
8705         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8706         void *cookie)): New.
8707         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8709 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8711         * dce.c (fini_dce): Call df_analyze again just in case
8712         delete_unmarked_insns removed anything.
8714 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8716         PR rtl-optimizations/57451
8717         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8718         from crossing split section boundaries.
8720 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8722         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8723         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8724         with MULTLIB_DEFAULTS.
8726 2013-08-20  Nick Clifton  <nickc@redhat.com>
8728         * target.def (narrow_volatile_bitfield): Note that the default
8729         value is false, not !TARGET_STRICT_ALIGN.
8730         * doc/tm.texi: Regenerate.
8732 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8734         Fix LIB_SPEC for systems without libpthread.
8736         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8737         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8738         for Android.
8739         * config/i386/linux-common.h: Likewise.
8740         * config/mips/linux-common.h: Likewise.
8742 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8744         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8745         checks.
8747 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8749         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8750         instances can own GC refs.
8752         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8753         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8754         (gcc::context::gt_pch_nx): Likewise.
8755         (gcc::context::gt_pch_nx):  Likewise.
8756         * ggc.h (gt_ggc_mx <T>): New.
8757         (gt_pch_nx_with_op <T>): New.
8758         (gt_pch_nx <T>): New.
8759         * passes.c (opt_pass::gt_ggc_mx): New.
8760         (opt_pass::gt_pch_nx): New.
8761         (opt_pass::gt_pch_nx_with_op): New.
8762         (pass_manager::gt_ggc_mx): New.
8763         (pass_manager::gt_pch_nx): New.
8764         (pass_manager::gt_pch_nx_with_op): New.
8765         (pass_manager::operator new): Use
8766         ggc_internal_cleared_alloc_stat rather than xcalloc.
8767         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8768         (pass_manager::gt_ggc_mx): New.
8769         (pass_manager::gt_pch_nx): New.
8770         (pass_manager::gt_pch_nx_with_op): New.
8771         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8772         (opt_pass::operator new): New.
8773         (opt_pass::gt_ggc_mx): New.
8774         (opt_pass::gt_pch_nx): New.
8775         (opt_pass::gt_pch_nx_with_op): New.
8777 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8779         * Makefile.in (GTFILES): Add context.h.
8780         * context.c (gcc::context::operator new): New.
8781         (gcc::context::gt_ggc_mx): New.
8782         (gcc::context::gt_pch_nx): New.
8783         (gcc::context::gt_pch_nx): New.
8784         * context.h (gcc::context): Add GTY((user)) marking.
8785         (gcc::context::operator new): New.
8786         (gcc::context::gt_ggc_mx): New.
8787         (gcc::context::gt_pch_nx): New.
8788         (gcc::context::gt_pch_nx): New.
8789         (g): Add GTY marking.
8790         (gt_ggc_mx (gcc::context *)): New.
8791         (gt_pch_nx (gcc::context *)): New.
8792         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8793         void *cookie)): New.
8794         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8796 2013-08-20  Alan Modra  <amodra@gmail.com>
8798         PR target/57865
8799         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8800         (rs6000_emit_epilogue): Likewise.
8802 2013-08-19  Dehao Chen  <dehao@google.com>
8804         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8806 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8807             Jakub Jelinek  <jakub@redhat.com>
8809         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8810         (BUILT_IN_FABSD64): Likewise.
8811         (BUILT_IN_FABSD128): Likewise.
8812         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8813         (fold_builtin_1): Likewise.
8814         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8815         destination and source operands.
8816         (*abstd2_fpr): Likewise.
8817         (*nabstd2_fpr): Likewise.
8819 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8821         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8822         JUMP_P and INSN_P.
8824 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8826         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8827         incomplete.
8829 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8831         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8832         * builtins.c (default_libc_has_function): New.
8833         (gnu_libc_has_function): Ditto.
8834         (no_c99_libc_has_function): Ditto.
8835         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8836         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8837         (fold_builtin_sincos): Likewise.
8838         (fold_builtin_cexp): Likewise.
8839         * builtins.def (DEF_C94_BUILTIN): Likewise.
8840         (DEF_C99_BUILTIN): Likewise.
8841         (DEF_C99_C90RES_BUILTIN): Likewise.
8842         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8843         definitions to using this define.
8844         * config/darwin-protos.h (darwin_libc_has_function): New.
8845         * config/darwin.c (darwin_libc_has_function): Ditto.
8846         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8847         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8848         * config/darwin.h: Ditto.
8849         * config/elfos.h: Ditto.
8850         * config/freebsd.h: Ditto.
8851         * config/i386/cygming.h: Ditto.
8852         * config/i386/djgpp.h: Ditto.
8853         * config/i386/i386-interix.h: Ditto.
8854         * config/microblaze/microblaze.h: Ditto.
8855         * config/mmix/mmix.h: Ditto.
8856         * config/gnu-user.h: Ditto.
8857         * config/ia64/hpux.h: Ditto.
8858         * config/pa/pa-hpux.h: Ditto.
8859         * config/pdp11/pdp11.h: Ditto.
8860         * config/picochip/picochip.h: Ditto.
8861         * config/linux.h: Ditto.
8862         * config/netbsd.h: Ditto.
8863         * config/openbsd.h: Ditto.
8864         * config/rs6000/aix43.h: Ditto.
8865         * config/rs6000/aix51.h: Ditto.
8866         * config/rs6000/aix52.h: Ditto.
8867         * config/rs6000/aix53.h: Ditto.
8868         * config/rs6000/aix61.h: Ditto.
8869         * config/rs6000/darwin.h: Ditto.
8870         * config/rs6000/linux.h: Ditto.
8871         * config/rs6000/linux64.h: Ditto.
8872         * config/s390/tpf.h: Ditto.
8873         * config/sol2-10.h: Ditto.
8874         * config/sol2.h: Ditto.
8875         * config/vms/vms.h: Ditto.
8876         * config/vxworks.h: Ditto.
8877         * config/linux-android.c (linux_android_libc_has_function):
8878         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8879         * config/linux-protos.h (linux_android_libc_has_function):
8880         New declaration.
8881         * config/i386/i386.c (ix86_libc_has_function): New.
8882         * config/i386/i386-protos.h
8883         (ix86_libc_has_function): New declaration.
8884         * config/i386/i386.md
8885         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8886         ("isinf<mode>2): Likewise.
8887         * convert.c (convert_to_integer): Using new target hook
8888         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8889         TARGET_C99_FUNCTIONS.
8890         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8891         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8892         * coretypes.h (function_class): New enum for different
8893         classes of functions.
8894         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8895         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8896         (TARGET_HAS_SINCOS): Likewise.
8897         (TARGET_LIBC_HAS_FUNCTION): New.
8898         * doc/tm.texi: Regenerated.
8899         * targhooks.h (default_libc_has_function): New declaration.
8900         (no_c99_libc_has_function): Ditto.
8901         (gnu_libc_has_function): Ditto.
8902         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8903         and TARGET_HAS_SINCOS.
8905 2013-08-18  Jan Hubicka  <jh@suse.cz>
8907         * Makeifle-in (ipa-devirt.o): New.
8908         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8909         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8910         (analyze_functions): Look into possible targets of polymorphic call.
8911         * dumpfile.c (dump_files): Add type-inheritance dump.
8912         * dumpfile.h (TDI_inheritance): New.
8913         * ipa-devirt.c: New file.
8914         * ipa-utils.h (odr_type_d): Forward declare.
8915         (odr_type): New type.
8916         (build_type_inheritance_graph): Declare.
8917         (possible_polymorphic_call_targets): Declare and introduce inline
8918         variant when only edge is pased.
8919         (dump_possible_polymorphic_call_targets): Likewise.
8920         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8921         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8922         (types_same_for_odr): ... here.
8923         * tree.h (type_in_anonymous_namespace_p): Declare.
8925 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8927         PR tree-optimization/58006
8928         * tree-parloops.c (take_address_of): Don't ICE if get_name
8929         returns NULL.
8930         (eliminate_local_variables_stmt): Remove clobber stmts.
8932 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8934         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8935         error message is issued for an alias to undefined symbol.
8937 2013-08-18  Jan Hubicka  <jh@suse.cz>
8939         * cgraph.c (cgraph_create_indirect_edge): Discover
8940         polymorphic calls and record basic info into indirect_info.
8941         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8942         devirtualization, ignore objc function calls.
8943         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
8944         call with no parm index info.
8945         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
8946         * tree.c (virtual_method_call_p): New function.
8947         * tree.h (virtual_method_call_p): Declare.
8949 2013-08-16  Jan Hubicka  <jh@suse.cz>
8951         PR middle-end/58179
8952         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
8954 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
8956         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
8957         attribute.
8959 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8961         * gengtype.c (type_for_name): Add special-case support for
8962         locating types within the "gcc::" namespace.
8963         (open_base_files): Emit a "using namespace gcc" directive.
8965 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8967         PR target/58160
8968         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
8969         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
8971         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
8972         array instead of each individual operand as a separate argument.
8973         (emit_fusion_gpr_load): Likewise.
8974         (expand_fusion_gpr_load): Add new function declaration.
8976         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
8977         signature to have the operands passed as an array, instead of as
8978         separate arguments.  Allow ZERO_EXTEND to be in the memory
8979         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
8980         depend on the register live/dead flags when peepholes are run.
8981         (expand_fusion_gpr_load): New function to be called from the
8982         peephole2 pass, to change the register that addis sets to be the
8983         target register.
8984         (emit_fusion_gpr_load): Change the calling signature to have the
8985         operands passed as an array, instead of as separate arguments.
8986         Allow ZERO_EXTEND to be in the memory address, and also
8987         SIGN_EXTEND if -mpower8-fusion-sign.
8989         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
8990         unspec enumeration.
8991         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
8992         adjust the register addis loads up in the peephole2 pass.  Do not
8993         depend on the register live/dead state when the peephole pass is done.
8995 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8997         * gengtype.c (create_user_defined_type): Ensure that the kind
8998         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
8999         declaration is seen before the GTY((user)) marking.
9001 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9003         PR target/58105
9004         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9006 2013-08-16  Jan Hubicka  <jh@suse.cz>
9008         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9009         arugment expected_type.
9010         (gimple_fold_call): Use it.
9011         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9012         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9013         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9014         (try_make_edge_direct_virtual_call): Likewise.
9015         * tree.c (obj_type_ref_class): New.
9016         * tree.h (obj_type_ref_class): Use it.
9018 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9020         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9021         (rtl_slim_pp): Likewise.
9022         (init_rtl_slim_pretty_print): Likewise.
9023         (dump_value_slim):  Don't call it.  Use local pretty printer.
9024         (dump_insn_slim): Likewise.
9025         (dump_rtl_slim): Likewise.
9026         (str_pattern_slim): Likewise.
9027         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9028         Simplify.
9030 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9032         PR tree-optimization/58164
9033         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9034         walk gimple_goto_dest of GIMPLE_GOTO.
9036         PR tree-optimization/58165
9037         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9038         bi_call must be the last stmt in a bb, don't split_block, instead
9039         use fallthru edge from it and give up if there is none.
9040         Release conds vector when returning early.
9042 2013-08-14  Xinliang David Li  <davidxl@google.com>
9044         * config/i386/i386.c (ix86_option_override_internal):
9045         Remove unused variable and field.
9047 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9049         PR target/57949
9050         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9051         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9052         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9053         and Linux, correct BLKmode alignment when 128-bit alignment is
9054         required and compatibility flag is not set.
9055         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9056         for zero-size arguments when compatibility flag is not set.
9058 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9060         PR tree-optimization/58145
9061         * tree-sra.c (build_ref_for_offset): If prev_base has
9062         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9064 2013-08-14  Xinliang David Li  <davidxl@google.com>
9066         * config/i386/i386.c (ix86_option_override_internal):
9067         Fix uninitialized variable error.
9069 2013-08-14  Xinliang David Li  <davidxl@google.com>
9071         * config/i386/i386.opt: Define two new options.
9072         * config/i386/x86-tune.def: Add arch selector field in macros.
9073         * config/i386/i386.h: Adjust macro definition.
9074         * config/i386/i386.c (ix86_option_override_internal):
9075         Refactor the code.
9076         (parse_mtune_ctrl_str): New function.
9077         (set_ix86_tune_features): New function.
9078         (ix86_function_specific_restore): Call the new helper function.
9080 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9082         PR rtl-optimization/57662
9083         * sel-sched.c (code_motion_process_successors): When the current insn
9084         is removed after the recursive traversal, break from the loop.
9085         Add comments and debug printouts.
9087 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9088             Alexandre Oliva  <aoliva@redhat.com>
9090         PR target/58067
9091         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9092         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9093         there also UNSPEC_PLTOFF.
9095 2013-08-14  Marek Polacek  <polacek@redhat.com>
9097         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9098         NUM_CONDITIONS bit positions.
9100 2013-08-13  Cary Coutant  <ccoutant@google.com>
9102         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9103         (attr_checksum): Hash vector contents instead of pointer.
9104         (attr_checksum_ordered): Likewise.
9106 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9108         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9109         when Pmode != word_mode.  Add length_address attribute.
9110         (sse3_monitor_<mode>): Merge from sse3_monitor and
9111         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9112         Pmode != word_mode.  Update insn length attribute.
9113         * config/i386/i386.c (ix86_option_override_internal): Update
9114         ix86_gen_monitor selection for merged sse3_monitor insn.
9116 2013-08-13  Julian Brown  <julian@codesourcery.com>
9118         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9119         perform invalid legitimization on greater-than-word-size modes for
9120         TARGET_E500_DOUBLE.
9122 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9124         * ira.c (setup_class_translate_array): Use aclass instead of cl
9125         for classes not fully covered by allocno classes.
9127 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9129         PR tree-optimization/57661
9130         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9131         * tree-inline.c (tree_function_versioning): Initialize it.
9132         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9133         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9134         that is not being copied.
9136         PR sanitizer/56417
9137         * asan.c (instrument_strlen_call): Fix typo in comment.
9138         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9140 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9142         * config/mips/mips.md (prefetch): Use lw instead of ld on
9143         loongson in 32bit mode.
9145 2013-08-13  Nick Clifton  <nickc@redhat.com>
9147         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9149 2013-08-13  Jan Hubicka  <jh@suse.cz>
9151         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9152         introduced edge; fix typo in sanity check.
9153         (cgraph_resolve_speculation): Export; improve diagnostic.
9154         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9155         speculation at type mismatch.
9156         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9157         (cgraph_resolve_speculation): Declare.
9158         (symtab_can_be_discarded): New function.
9159         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9160         * ipa-inline-transform.c (speculation_removed): New global var.
9161         (clone_inlined_nodes): See if speculation can be removed.
9162         (inline_call): If speculations was removed, we growths may not match.
9163         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9164         (speculation_useful_p): New function.
9165         (resolve_noninline_speculation): New function.
9166         (inline_small_functions): Resolve useless speculations.
9167         * ipa-inline.h (speculation_useful_p): Declare
9168         * ipa.c (can_replace_by_local_alias): Simplify.
9169         (ipa_profile): Produce speculative calls in non-lto, too;
9170         add simple cost model; produce local aliases.
9172 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9174         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9175         PASS_MANAGER_H.
9177 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9179         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9181 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9183         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9184         iterator instead of X87MODEF.
9186 2013-08-12  Perez Read  <netfirewall@gmail.com>
9188         PR target/58132
9189         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9190         operand 0 for intel asm alternative.
9191         (*movabs<mode>_2): Ditto for operand 1.
9193 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9195         * config/aarch64/arm_none.h
9196         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9198 2013-08-12  Nick Clifton  <nickc@redhat.com>
9200         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9202 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9204         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9205         expand for QI/HImode operand to produce more effictive code for
9206         unsigned char(short) --> float(double) conversion.
9208 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9210         * doc/invoke.texi: Mention that -ftls-model does not force the final
9211         model.
9213 2013-08-12  Marek Polacek  <polacek@redhat.com>
9214             Marc Glisse  <marc.glisse@inria.fr>
9216         PR tree-optimization/57980
9217         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9218         when creating -1 constant.
9220 2013-08-10  Jan Hubicka  <jh@suse.cz>
9222         Workaround binutils PR14342.
9223         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9224         (gimple_init_edge_profiler): Likewise.
9225         (gimple_gen_ic_func_profiler): Likewise.
9227 2013-08-09  Jan Hubicka  <jh@suse.cz>
9229         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9231 2013-08-09  Xinliang David Li  <davidxl@google.com>
9233         * config/i386/stringop.def: New file.
9234         * config/i386/stringop.opt: New file.
9235         * config/i386/i386-opts.h: Include stringopt.def.
9236         * config/i386/i386.opt: Include stringopt.opt.
9237         * config/i386/i386.c (ix86_option_override_internal):
9238         Override default size based stringop inline strategies with options.
9239         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9240         New function.
9242 2013-08-09  Jan Hubicka  <jh@suse.cz>
9244         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9246 2013-08-09  Jan Hubicka  <jh@suse.cz>
9248         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9249         CGRAPH_FREQ_MAX.
9250         (dump_cgraph_node): Dump profile-id.
9251         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9252         and common_target_probability.
9253         * lto-cgraph.c (lto_output_edge): Stream common targets.
9254         (lto_output_node): Stream profile ids.
9255         (input_node): Stream profile ids.
9256         (input_edge): Stream common targets.
9257         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9258         * ipa.c: Include value-prof.h
9259         (ipa_profile_generate_summary): Turn indirect call statement histograms
9260         into common targets.
9261         (ipa_profile): Turn common targets into speculative edges.
9263 2013-08-09  Jan Hubicka  <jh@suse.cz>
9265         * cgraph.h (cgraph_node): Add profile_id.
9266         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9267         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9268         (del_node_map): Update.
9269         (find_func_by_funcdef_no): Replace by ...
9270         (find_func_by_profile_id): ... this one.
9271         (gimple_ic_transform): Do not remove useful histograms when
9272         speculation is not done; dump info when indirect call removal
9273         can happen at LTO.
9274         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9275         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9276         (__gcov_indirect_call_profiler_v2): .. this one.
9277         * profile.h (init_node_map): Update.
9278         * coverage.c (coverage_compute_profile_id): New function.
9279         * coverage.h (coverage_compute_profile_id): Declare.
9280         * tree-profile.c (init_ic_make_global_vars): Make
9281         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9282         (gimple_init_edge_profiler): Update prototype of
9283         __gcov_indirect_call_profiler.
9284         (gimple_gen_ic_func_profiler): Simplify.
9285         (tree_profiling): Use init_node_map
9287 2013-08-09  Jan Hubicka  <jh@suse.cz>
9289         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9290         non-speculative refs.
9291         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9292         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9293         (cgraph_set_call_stmt): Likewise.
9294         (cgraph_create_edge_1): Fix release checking compilatoin;
9295         clear lto_stmt_uid.
9296         (cgraph_free_edge): Free indirect info.
9297         (cgraph_turn_edge_to_speculative): New function.
9298         (cgraph_speculative_call_info): New function.
9299         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9300         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9301         (dump_cgraph_node): Dump speculation.
9302         (verify_edge_count_and_frequency): Accept speculative edges.
9303         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9304         (verify_cgraph_node): Handle speculation.
9305         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9306         (cgraph_set_call_stmt): Update prototype.
9307         (cgraph_make_edge_direct): Update prototype.
9308         (cgraph_speculative_call_info): Declare.
9309         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9310         to change; update call of ipa_find_references.
9311         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9312         lto_stmt_uid and speculative flags.
9313         (ipa_dump_references): Dump speculation.
9314         (ipa_clone_references): Clone speculative flag.
9315         (ipa_clone_referring): Likewise.
9316         (ipa_clone_ref): New function.
9317         (ipa_find_reference): Look into lto_stmt_uids
9318         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9319         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9320         (ipa_find_reference): Update declaration.
9321         (ipa_clone_ref): Declare.
9322         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9323         stream speculative flag.
9324         (lto_output_ref): Stream statements uids and speculation.
9325         (input_ref): Likewise.
9326         (input_edge): Stream speuclation.
9327         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9328         (cgraph_set_call_stmt_including_clones): Handle speculation.
9329         * ipa-inline.c (heap_edge_removal_hook): New function.
9330         (inline_small_functions): Register it.
9331         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9332         also initialize refs.
9333         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9334         edge to change.
9335         (try_make_edge_direct_simple_call): Likewise.
9336         (try_make_edge_direct_simple_call): Likewise.
9337         (update_indirect_edges_after_inlining): Likewise.
9338         (remove_described_reference): Look proper lto_stmt_uid.
9339         (propagate_controlled_uses): Likewise.
9340         (propagate_controlled_uses): Liekwise.
9341         * tree-inline.c (copy_bb): Copy speculative edges.
9342         (redirect_all_calls): New function.
9343         (copy_cfg_body): Do redirection after loop info is updated.
9344         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9346 2013-08-09  Jan Hubicka  <jh@suse.cz>
9348         * lto-streamer-out.c (output_function): Renumber PHIs.
9349         * lto-streamer-in.c (input_function): Likewise.
9351 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9353         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9354         (get_lane_unsigned): Likewise.
9355         (dup_lane_scalar): Likewise.
9356         (get_lane): enable for VALL.
9357         * config/aarch64/aarch64-simd.md
9358         (aarch64_dup_lane_scalar<mode>): Remove.
9359         (aarch64_get_lane_signed<mode>): Likewise.
9360         (aarch64_get_lane_unsigned<mode>): Likewise.
9361         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9362         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9363         (aarch64_get_lane<mode>): Enable for all vector modes.
9364         (aarch64_get_lanedi): Remove misleading constraints.
9365         * config/aarch64/arm_neon.h
9366         (__aarch64_vget_lane_any): Define.
9367         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9368         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9369         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9370         * config/aarch64/iterators.md (VDQQH): New.
9371         (VDQQHS): Likewise.
9372         (vwcore): Likewise.
9374 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9376         * configure.ac: Add GAS check for LEON instructions on SPARC.
9377         * configure: Regenerate.
9378         * config.in: Likewise.
9379         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9380         sparc*-*-* block.
9381         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9382         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9383         for LEON or LEON3.
9384         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9385         (AS_LEON_FLAG): New macro.
9386         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9387         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9388         Deal with LEON and LEON3 for the memory model.
9389         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9390         (atomic_compare_and_swap<mode>_1): Likewise.
9391         (*atomic_compare_and_swap<mode>_1): Likewise.
9393 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9395         * config/arm/neon.md (vcond): Fix floating-point vector
9396         comparisons against 0.
9398 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9400         * lra-constraints.c (emit_spill_move): Remove assert.
9401         (process_alt_operands): Add more debugging
9402         output.  Increase reject for spilling into memory.  Decrease
9403         reject for reloading scratch.
9404         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9406 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9408         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9409         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9410         micromips incompatible.  Add nan2008.
9411         (MULTILIB_DIRNAMES): Add nan2008.
9412         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9413         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9414         and micromips incompatible.  Add nan2008.
9415         (MULTILIB_DIRNAMES): Add nan2008.
9416         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9418 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9420         PR rtl-optimization/58079
9421         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9422         simplify_comparison has widened a comparison with an integer.
9424 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9426         * config/arm/neon.md (movmisalign<mode>): Disable when we
9427         don't allow unaligned accesses.
9428         (*movmisalign<mode>_neon_store): Likewise.
9429         (*movmisalign<mode>_neon_load): Likewise.
9430         (*movmisalign<mode>_neon_store): Likewise.
9431         (*movmisalign<mode>_neon_load): Likewise.
9433 2013-08-08  Jan Hubicka  <jh@suse.cz>
9435         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9436         (make_pass_rebuild_cgraph_edges): Also clear references.
9437         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9438         * ipa-inline-transform.c (inline_transform): Remove all references
9439         after inlining.
9440         * cgraphunit.c (expand_function): Remove all references after
9441         expansion.
9442         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9443         (ipa_find_reference): Rewrite to iterator.
9444         (remove_stmt_references): Likewise.
9445         (ipa_clear_stmts_in_references): New function.
9446         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9447         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9448         clear references.
9449         * ipa-split.c (split_function): Remove references in split function.
9451 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9453         PR target/57431
9454         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9455         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9457 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9459         PR target/56979
9460         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9461         suggested mode for the assignment isn't compatible with the
9462         registers required.
9464 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9466         PR target/58065
9467         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9469 2013-08-07  Xinliang David Li  <davidxl@google.com>
9471         * config/i386/i386.opt: New option -mtune-ctrl=.
9472         * config/i386/x86-tune.def: New file.
9473         * config/i386/i386.h: include x86-tune.def.
9474         * config/i386/i386.c (ix86_option_override_internal):
9475         Parsing -mtune-ctrl= option and set tune features.
9477 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9479         PR other/12081
9480         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9481         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9482         to gen_2arg_fn_t.
9484 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9486         * rtl.h (update_alignments): Declare.
9487         * final.c (grow_label_align): New function extracted from...
9488         (shorten_branches): ...here.  Call it.
9489         (update_alignments): New function.
9490         * reorg.c (sibling_labels): New variable.
9491         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9492         the new label along with it onto the sibling_labels vector.
9493         (fill_simple_delay_slots): Adjust call to get_label_before.
9494         (fill_slots_from_thread): Likewise.
9495         (relax_delay_slots): Likewise.
9496         (make_return_insns): Likewise.
9497         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9499 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9501         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9502         document its semantics.
9503         (diagnostic_report_diagnostic): Adjust accordingly.
9505 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9507         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9508         (sparc_option_override): ...and port to new C++ pass API.
9509         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9511 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9513         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9515 2013-08-06  Caroline Tice  <cmtice@google.com>
9517         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9518         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9519         * tree-pass.h: Add pass_vtable_verify.
9520         * varasm.c (assemble_variable): Add code to properly set the comdat
9521         section and name for the .vtable_map_vars section.
9522         (assemble_vtyv_preinit_initializer): New function.
9523         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9524         LINK_ONCE flag.
9525         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9526         * vtable-verify.c: New file.
9527         * vtable-verify.h: New file.
9528         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9529         initialiation levels.
9530         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9531         * passes.def: Insert pass_vtable_verify.
9532         * aclocal.m4: Reorder includes.
9533         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9534         -fvtv-counts options.
9535         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9536         as appropriate, if -fvtable-verify=... is used.
9537         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9538         -fvtable-verify=... is used.
9539         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9540         (vtable-verify.o): Add new build rule.
9541         (GTFILES): Add vtable-verify.c to list.
9542         * common.opt (fvtable-verify=): New flag.
9543         (vtv_priority): Values for fvtable-verify= flag.
9544         (fvtv-counts): New flag.
9545         (fvtv-debug): New flag.
9546         * tree.h (save_vtable_map_decl): New extern function decl.
9548 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9550         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9551         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9552         (pass_data_rl78_devirt): ...new pass_data instance and...
9553         (make_pass_rl78_devirt): ...new function.
9554         (rl78_asm_file_start): Port pass registration to new C++ API.
9556 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9558         * coretypes.h (rtl_opt_pass): Add.
9559         (gcc::context): Add.
9560         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9561         (epiphany_init): Port to new C++ pass API.
9562         (epiphany_optimize_mode_switching): Likewise.
9563         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9564         (pass_manager::get_pass_mode_switching): New.
9565         (pass_manager::get_pass_peephole2): New.
9566         * mode-switching.c (pass_mode_switching): Add clone method.
9567         * recog.c (pass_peephole2): Add clone method.
9568         (pass_split_all_insns): Add clone method.
9570 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9572         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9573         (mips_option_override): ...here, porting to new C++ API for passes.
9575 2013-08-06  Jan Hubicka  <jh@suse.cz>
9577         * cgraph.c (cgraph_get_body): New function based on lto.c
9578         implementation.
9579         * cgraph.h (cgraph_get_body): Declare.
9580         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9581         LTO paths.
9582         * cgraphunit.c (expand_function): Get body prior expanding.
9583         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9584         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9585         really need.
9586         * passes.c (do_per_function_toporder): Get body.
9587         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9588         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9590 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9592         PR fortran/57987
9593         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9594         is not re-finalized.  Rename second parameter to no_collect.
9596 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9598         PR middle-end/58041
9599         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9600         MEM_REF has proper alignment information.
9602 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9604         PR other/12081
9605         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9606         class insn_gen_fn.
9607         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9608         rtx (*) (rtx, ...) with insn_gen_fn.
9609         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9610         insn_gen_fn::stored_funcptr.  Add initializer braces.
9612 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9614         Rewrite how instances of passes are cloned to remove assumptions
9615         about their sizes (thus allowing pass subclasses to have
9616         additional data fields, albeit non-GC-managed ones at this point).
9618         * passes.c (make_pass_instance): Now that passes have clone
9619         methods, rewrite this function to eliminate XNEW and memcpy
9620         calls that used hardcoded sizes.  Since this function no longer
9621         creates pass instances, rename it to...
9622         (add_pass_instance): ...this.  Document the old way that passes were
9623         numbered and flagged, and rework this function to continue using it.
9624         (next_pass_1): Add an initial_pass argument for use by
9625         add_pass_instance.
9626         (position_pass): When adding multiple instances of a pass, use
9627         the pass's clone method, rather than relying on the XNEW/memcpy
9628         within the former make_pass_instance (now add_pass_instance).
9629         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9630         the initial instance of the current pass within the pass manager.
9632 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9634         This is the automated part of the conversion of passes from C
9635         structs to C++ classes.
9637         Patch autogenerated by refactor_passes.py from
9638         https://github.com/davidmalcolm/gcc-refactoring-scripts
9639         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9641         * asan.c (pass_asan): Convert from a global struct to a subclass of
9642         gimple_opt_pass along with...
9643         (pass_data_asan): ...new pass_data instance and...
9644         (make_pass_asan): ...new function.
9645         (pass_asan_O0): Convert from a global struct to a subclass of
9646         gimple_opt_pass along with...
9647         (pass_data_asan_O0): ...new pass_data instance and...
9648         (make_pass_asan_O0): ...new function.
9649         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9650         subclass of rtl_opt_pass along with...
9651         (pass_data_inc_dec): ...new pass_data instance and...
9652         (make_pass_inc_dec): ...new function.
9653         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9654         a subclass of rtl_opt_pass along with...
9655         (pass_data_reorder_blocks): ...new pass_data instance and...
9656         (make_pass_reorder_blocks): ...new function.
9657         (pass_duplicate_computed_gotos): Convert from a global struct to a
9658         subclass of rtl_opt_pass along with...
9659         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9660         (make_pass_duplicate_computed_gotos): ...new function.
9661         (pass_partition_blocks): Convert from a global struct to a subclass of
9662         rtl_opt_pass along with...
9663         (pass_data_partition_blocks): ...new pass_data instance and...
9664         (make_pass_partition_blocks): ...new function.
9665         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9666         struct to a subclass of rtl_opt_pass along with...
9667         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9668         and...
9669         (make_pass_branch_target_load_optimize1): ...new function.
9670         (pass_branch_target_load_optimize2): Convert from a global struct to a
9671         subclass of rtl_opt_pass along with...
9672         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9673         and...
9674         (make_pass_branch_target_load_optimize2): ...new function.
9675         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9676         of rtl_opt_pass along with...
9677         (pass_data_jump): ...new pass_data instance and...
9678         (make_pass_jump): ...new function.
9679         (pass_jump2): Convert from a global struct to a subclass of
9680         rtl_opt_pass along with...
9681         (pass_data_jump2): ...new pass_data instance and...
9682         (make_pass_jump2): ...new function.
9683         * cfgexpand.c (pass_expand): Convert from a global struct to a
9684         subclass of rtl_opt_pass along with...
9685         (pass_data_expand): ...new pass_data instance and...
9686         (make_pass_expand): ...new function.
9687         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9688         of rtl_opt_pass along with...
9689         (pass_data_free_cfg): ...new pass_data instance and...
9690         (make_pass_free_cfg): ...new function.
9691         (pass_into_cfg_layout_mode): Convert from a global struct to a
9692         subclass of rtl_opt_pass along with...
9693         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9694         (make_pass_into_cfg_layout_mode): ...new function.
9695         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9696         subclass of rtl_opt_pass along with...
9697         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9698         (make_pass_outof_cfg_layout_mode): ...new function.
9699         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9700         struct to a subclass of gimple_opt_pass along with...
9701         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9702         (make_pass_build_cgraph_edges): ...new function.
9703         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9704         subclass of gimple_opt_pass along with...
9705         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9706         (make_pass_rebuild_cgraph_edges): ...new function.
9707         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9708         subclass of gimple_opt_pass along with...
9709         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9710         and...
9711         (make_pass_remove_cgraph_callee_edges): ...new function.
9712         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9713         struct to a subclass of rtl_opt_pass along with...
9714         (pass_data_stack_adjustments): ...new pass_data instance and...
9715         (make_pass_stack_adjustments): ...new function.
9716         * combine.c (pass_combine): Convert from a global struct to a subclass
9717         of rtl_opt_pass along with...
9718         (pass_data_combine): ...new pass_data instance and...
9719         (make_pass_combine): ...new function.
9720         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9721         global struct to a subclass of rtl_opt_pass along with...
9722         (pass_data_compare_elim_after_reload): ...new pass_data instance
9723         and...
9724         (make_pass_compare_elim_after_reload): ...new function.
9725         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9726         of rtl_opt_pass along with...
9727         (pass_data_rtl_cprop): ...new pass_data instance and...
9728         (make_pass_rtl_cprop): ...new function.
9729         * cse.c (pass_cse): Convert from a global struct to a subclass of
9730         rtl_opt_pass along with...
9731         (pass_data_cse): ...new pass_data instance and...
9732         (make_pass_cse): ...new function.
9733         (pass_cse2): Convert from a global struct to a subclass of
9734         rtl_opt_pass along with...
9735         (pass_data_cse2): ...new pass_data instance and...
9736         (make_pass_cse2): ...new function.
9737         (pass_cse_after_global_opts): Convert from a global struct to a
9738         subclass of rtl_opt_pass along with...
9739         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9740         (make_pass_cse_after_global_opts): ...new function.
9741         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9742         of rtl_opt_pass along with...
9743         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9744         (make_pass_ud_rtl_dce): ...new function.
9745         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9746         rtl_opt_pass along with...
9747         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9748         (make_pass_fast_rtl_dce): ...new function.
9749         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9750         a subclass of rtl_opt_pass along with...
9751         (pass_data_df_initialize_opt): ...new pass_data instance and...
9752         (make_pass_df_initialize_opt): ...new function.
9753         (pass_df_initialize_no_opt): Convert from a global struct to a
9754         subclass of rtl_opt_pass along with...
9755         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9756         (make_pass_df_initialize_no_opt): ...new function.
9757         (pass_df_finish): Convert from a global struct to a subclass of
9758         rtl_opt_pass along with...
9759         (pass_data_df_finish): ...new pass_data instance and...
9760         (make_pass_df_finish): ...new function.
9761         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9762         rtl_opt_pass along with...
9763         (pass_data_rtl_dse1): ...new pass_data instance and...
9764         (make_pass_rtl_dse1): ...new function.
9765         (pass_rtl_dse2): Convert from a global struct to a subclass of
9766         rtl_opt_pass along with...
9767         (pass_data_rtl_dse2): ...new pass_data instance and...
9768         (make_pass_rtl_dse2): ...new function.
9769         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9770         subclass of rtl_opt_pass along with...
9771         (pass_data_dwarf2_frame): ...new pass_data instance and...
9772         (make_pass_dwarf2_frame): ...new function.
9773         * except.c (pass_set_nothrow_function_flags): Convert from a global
9774         struct to a subclass of rtl_opt_pass along with...
9775         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9776         and...
9777         (make_pass_set_nothrow_function_flags): ...new function.
9778         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9779         subclass of rtl_opt_pass along with...
9780         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9781         and...
9782         (make_pass_convert_to_eh_region_ranges): ...new function.
9783         * final.c (pass_compute_alignments): Convert from a global struct to a
9784         subclass of rtl_opt_pass along with...
9785         (pass_data_compute_alignments): ...new pass_data instance and...
9786         (make_pass_compute_alignments): ...new function.
9787         (pass_final): Convert from a global struct to a subclass of
9788         rtl_opt_pass along with...
9789         (pass_data_final): ...new pass_data instance and...
9790         (make_pass_final): ...new function.
9791         (pass_shorten_branches): Convert from a global struct to a subclass of
9792         rtl_opt_pass along with...
9793         (pass_data_shorten_branches): ...new pass_data instance and...
9794         (make_pass_shorten_branches): ...new function.
9795         (pass_clean_state): Convert from a global struct to a subclass of
9796         rtl_opt_pass along with...
9797         (pass_data_clean_state): ...new pass_data instance and...
9798         (make_pass_clean_state): ...new function.
9799         * function.c (pass_instantiate_virtual_regs): Convert from a global
9800         struct to a subclass of rtl_opt_pass along with...
9801         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9802         (make_pass_instantiate_virtual_regs): ...new function.
9803         (pass_leaf_regs): Convert from a global struct to a subclass of
9804         rtl_opt_pass along with...
9805         (pass_data_leaf_regs): ...new pass_data instance and...
9806         (make_pass_leaf_regs): ...new function.
9807         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9808         subclass of rtl_opt_pass along with...
9809         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9810         and...
9811         (make_pass_thread_prologue_and_epilogue): ...new function.
9812         (pass_match_asm_constraints): Convert from a global struct to a
9813         subclass of rtl_opt_pass along with...
9814         (pass_data_match_asm_constraints): ...new pass_data instance and...
9815         (make_pass_match_asm_constraints): ...new function.
9816         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9817         subclass of rtl_opt_pass along with...
9818         (pass_data_rtl_fwprop): ...new pass_data instance and...
9819         (make_pass_rtl_fwprop): ...new function.
9820         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9821         rtl_opt_pass along with...
9822         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9823         (make_pass_rtl_fwprop_addr): ...new function.
9824         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9825         rtl_opt_pass along with...
9826         (pass_data_rtl_pre): ...new pass_data instance and...
9827         (make_pass_rtl_pre): ...new function.
9828         (pass_rtl_hoist): Convert from a global struct to a subclass of
9829         rtl_opt_pass along with...
9830         (pass_data_rtl_hoist): ...new pass_data instance and...
9831         (make_pass_rtl_hoist): ...new function.
9832         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9833         subclass of gimple_opt_pass along with...
9834         (pass_data_lower_cf): ...new pass_data instance and...
9835         (make_pass_lower_cf): ...new function.
9836         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9837         from a global struct to a subclass of gimple_opt_pass along with...
9838         (pass_data_strength_reduction): ...new pass_data instance and...
9839         (make_pass_strength_reduction): ...new function.
9840         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9841         of rtl_opt_pass along with...
9842         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9843         (make_pass_rtl_ifcvt): ...new function.
9844         (pass_if_after_combine): Convert from a global struct to a subclass of
9845         rtl_opt_pass along with...
9846         (pass_data_if_after_combine): ...new pass_data instance and...
9847         (make_pass_if_after_combine): ...new function.
9848         (pass_if_after_reload): Convert from a global struct to a subclass of
9849         rtl_opt_pass along with...
9850         (pass_data_if_after_reload): ...new pass_data instance and...
9851         (make_pass_if_after_reload): ...new function.
9852         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9853         a subclass of rtl_opt_pass along with...
9854         (pass_data_initialize_regs): ...new pass_data instance and...
9855         (make_pass_initialize_regs): ...new function.
9856         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9857         of ipa_opt_pass_d along with...
9858         (pass_data_ipa_cp): ...new pass_data instance and...
9859         (make_pass_ipa_cp): ...new function.
9860         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9861         global struct to a subclass of gimple_opt_pass along with...
9862         (pass_data_inline_parameters): ...new pass_data instance and...
9863         (make_pass_inline_parameters): ...new function.
9864         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9865         subclass of gimple_opt_pass along with...
9866         (pass_data_early_inline): ...new pass_data instance and...
9867         (make_pass_early_inline): ...new function.
9868         (pass_ipa_inline): Convert from a global struct to a subclass of
9869         ipa_opt_pass_d along with...
9870         (pass_data_ipa_inline): ...new pass_data instance and...
9871         (make_pass_ipa_inline): ...new function.
9872         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9873         struct to a subclass of gimple_opt_pass along with...
9874         (pass_data_local_pure_const): ...new pass_data instance and...
9875         (make_pass_local_pure_const): ...new function.
9876         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9877         ipa_opt_pass_d along with...
9878         (pass_data_ipa_pure_const): ...new pass_data instance and...
9879         (make_pass_ipa_pure_const): ...new function.
9880         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9881         to a subclass of ipa_opt_pass_d along with...
9882         (pass_data_ipa_reference): ...new pass_data instance and...
9883         (make_pass_ipa_reference): ...new function.
9884         * ipa-split.c (pass_split_functions): Convert from a global struct to
9885         a subclass of gimple_opt_pass along with...
9886         (pass_data_split_functions): ...new pass_data instance and...
9887         (make_pass_split_functions): ...new function.
9888         (pass_feedback_split_functions): Convert from a global struct to a
9889         subclass of gimple_opt_pass along with...
9890         (pass_data_feedback_split_functions): ...new pass_data instance and...
9891         (make_pass_feedback_split_functions): ...new function.
9892         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9893         global struct to a subclass of simple_ipa_opt_pass along with...
9894         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9895         instance and...
9896         (make_pass_ipa_function_and_variable_visibility): ...new function.
9897         (pass_ipa_free_inline_summary): Convert from a global struct to a
9898         subclass of simple_ipa_opt_pass along with...
9899         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9900         (make_pass_ipa_free_inline_summary): ...new function.
9901         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9902         subclass of ipa_opt_pass_d along with...
9903         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9904         and...
9905         (make_pass_ipa_whole_program_visibility): ...new function.
9906         (pass_ipa_profile): Convert from a global struct to a subclass of
9907         ipa_opt_pass_d along with...
9908         (pass_data_ipa_profile): ...new pass_data instance and...
9909         (make_pass_ipa_profile): ...new function.
9910         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9911         ipa_opt_pass_d along with...
9912         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9913         (make_pass_ipa_cdtor_merge): ...new function.
9914         * ira.c (pass_ira): Convert from a global struct to a subclass of
9915         rtl_opt_pass along with...
9916         (pass_data_ira): ...new pass_data instance and...
9917         (make_pass_ira): ...new function.
9918         (pass_reload): Convert from a global struct to a subclass of
9919         rtl_opt_pass along with...
9920         (pass_data_reload): ...new pass_data instance and...
9921         (make_pass_reload): ...new function.
9922         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9923         subclass of rtl_opt_pass along with...
9924         (pass_data_cleanup_barriers): ...new pass_data instance and...
9925         (make_pass_cleanup_barriers): ...new function.
9926         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9927         of rtl_opt_pass along with...
9928         (pass_data_loop2): ...new pass_data instance and...
9929         (make_pass_loop2): ...new function.
9930         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9931         rtl_opt_pass along with...
9932         (pass_data_rtl_loop_init): ...new pass_data instance and...
9933         (make_pass_rtl_loop_init): ...new function.
9934         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9935         rtl_opt_pass along with...
9936         (pass_data_rtl_loop_done): ...new pass_data instance and...
9937         (make_pass_rtl_loop_done): ...new function.
9938         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9939         subclass of rtl_opt_pass along with...
9940         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9941         (make_pass_rtl_move_loop_invariants): ...new function.
9942         (pass_rtl_unswitch): Convert from a global struct to a subclass of
9943         rtl_opt_pass along with...
9944         (pass_data_rtl_unswitch): ...new pass_data instance and...
9945         (make_pass_rtl_unswitch): ...new function.
9946         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
9947         subclass of rtl_opt_pass along with...
9948         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
9949         and...
9950         (make_pass_rtl_unroll_and_peel_loops): ...new function.
9951         (pass_rtl_doloop): Convert from a global struct to a subclass of
9952         rtl_opt_pass along with...
9953         (pass_data_rtl_doloop): ...new pass_data instance and...
9954         (make_pass_rtl_doloop): ...new function.
9955         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
9956         a subclass of rtl_opt_pass along with...
9957         (pass_data_lower_subreg): ...new pass_data instance and...
9958         (make_pass_lower_subreg): ...new function.
9959         (pass_lower_subreg2): Convert from a global struct to a subclass of
9960         rtl_opt_pass along with...
9961         (pass_data_lower_subreg2): ...new pass_data instance and...
9962         (make_pass_lower_subreg2): ...new function.
9963         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
9964         struct to a subclass of ipa_opt_pass_d along with...
9965         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
9966         (make_pass_ipa_lto_gimple_out): ...new function.
9967         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
9968         of ipa_opt_pass_d along with...
9969         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
9970         (make_pass_ipa_lto_finish_out): ...new function.
9971         * mode-switching.c (pass_mode_switching): Convert from a global struct
9972         to a subclass of rtl_opt_pass along with...
9973         (pass_data_mode_switching): ...new pass_data instance and...
9974         (make_pass_mode_switching): ...new function.
9975         * modulo-sched.c (pass_sms): Convert from a global struct to a
9976         subclass of rtl_opt_pass along with...
9977         (pass_data_sms): ...new pass_data instance and...
9978         (make_pass_sms): ...new function.
9979         * omp-low.c (pass_expand_omp): Convert from a global struct to a
9980         subclass of gimple_opt_pass along with...
9981         (pass_data_expand_omp): ...new pass_data instance and...
9982         (make_pass_expand_omp): ...new function.
9983         (pass_lower_omp): Convert from a global struct to a subclass of
9984         gimple_opt_pass along with...
9985         (pass_data_lower_omp): ...new pass_data instance and...
9986         (make_pass_lower_omp): ...new function.
9987         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
9988         of gimple_opt_pass along with...
9989         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
9990         (make_pass_diagnose_omp_blocks): ...new function.
9991         * passes.c (pass_early_local_passes): Convert from a global struct to
9992         a subclass of simple_ipa_opt_pass along with...
9993         (pass_data_early_local_passes): ...new pass_data instance and...
9994         (make_pass_early_local_passes): ...new function.
9995         (pass_all_early_optimizations): Convert from a global struct to a
9996         subclass of gimple_opt_pass along with...
9997         (pass_data_all_early_optimizations): ...new pass_data instance and...
9998         (make_pass_all_early_optimizations): ...new function.
9999         (pass_all_optimizations): Convert from a global struct to a subclass
10000         of gimple_opt_pass along with...
10001         (pass_data_all_optimizations): ...new pass_data instance and...
10002         (make_pass_all_optimizations): ...new function.
10003         (pass_all_optimizations_g): Convert from a global struct to a subclass
10004         of gimple_opt_pass along with...
10005         (pass_data_all_optimizations_g): ...new pass_data instance and...
10006         (make_pass_all_optimizations_g): ...new function.
10007         (pass_rest_of_compilation): Convert from a global struct to a subclass
10008         of rtl_opt_pass along with...
10009         (pass_data_rest_of_compilation): ...new pass_data instance and...
10010         (make_pass_rest_of_compilation): ...new function.
10011         (pass_postreload): Convert from a global struct to a subclass of
10012         rtl_opt_pass along with...
10013         (pass_data_postreload): ...new pass_data instance and...
10014         (make_pass_postreload): ...new function.
10015         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10016         subclass of rtl_opt_pass along with...
10017         (pass_data_gcse2): ...new pass_data instance and...
10018         (make_pass_gcse2): ...new function.
10019         * postreload.c (pass_postreload_cse): Convert from a global struct to
10020         a subclass of rtl_opt_pass along with...
10021         (pass_data_postreload_cse): ...new pass_data instance and...
10022         (make_pass_postreload_cse): ...new function.
10023         * predict.c (pass_profile): Convert from a global struct to a subclass
10024         of gimple_opt_pass along with...
10025         (pass_data_profile): ...new pass_data instance and...
10026         (make_pass_profile): ...new function.
10027         (pass_strip_predict_hints): Convert from a global struct to a subclass
10028         of gimple_opt_pass along with...
10029         (pass_data_strip_predict_hints): ...new pass_data instance and...
10030         (make_pass_strip_predict_hints): ...new function.
10031         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10032         of rtl_opt_pass along with...
10033         (pass_data_peephole2): ...new pass_data instance and...
10034         (make_pass_peephole2): ...new function.
10035         (pass_split_all_insns): Convert from a global struct to a subclass of
10036         rtl_opt_pass along with...
10037         (pass_data_split_all_insns): ...new pass_data instance and...
10038         (make_pass_split_all_insns): ...new function.
10039         (pass_split_after_reload): Convert from a global struct to a subclass
10040         of rtl_opt_pass along with...
10041         (pass_data_split_after_reload): ...new pass_data instance and...
10042         (make_pass_split_after_reload): ...new function.
10043         (pass_split_before_regstack): Convert from a global struct to a
10044         subclass of rtl_opt_pass along with...
10045         (pass_data_split_before_regstack): ...new pass_data instance and...
10046         (make_pass_split_before_regstack): ...new function.
10047         (pass_split_before_sched2): Convert from a global struct to a subclass
10048         of rtl_opt_pass along with...
10049         (pass_data_split_before_sched2): ...new pass_data instance and...
10050         (make_pass_split_before_sched2): ...new function.
10051         (pass_split_for_shorten_branches): Convert from a global struct to a
10052         subclass of rtl_opt_pass along with...
10053         (pass_data_split_for_shorten_branches): ...new pass_data instance
10054         and...
10055         (make_pass_split_for_shorten_branches): ...new function.
10056         * ree.c (pass_ree): Convert from a global struct to a subclass of
10057         rtl_opt_pass along with...
10058         (pass_data_ree): ...new pass_data instance and...
10059         (make_pass_ree): ...new function.
10060         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10061         subclass of rtl_opt_pass along with...
10062         (pass_data_stack_regs): ...new pass_data instance and...
10063         (make_pass_stack_regs): ...new function.
10064         (pass_stack_regs_run): Convert from a global struct to a subclass of
10065         rtl_opt_pass along with...
10066         (pass_data_stack_regs_run): ...new pass_data instance and...
10067         (make_pass_stack_regs_run): ...new function.
10068         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10069         subclass of rtl_opt_pass along with...
10070         (pass_data_cprop_hardreg): ...new pass_data instance and...
10071         (make_pass_cprop_hardreg): ...new function.
10072         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10073         subclass of rtl_opt_pass along with...
10074         (pass_data_reginfo_init): ...new pass_data instance and...
10075         (make_pass_reginfo_init): ...new function.
10076         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10077         of rtl_opt_pass along with...
10078         (pass_data_regmove): ...new pass_data instance and...
10079         (make_pass_regmove): ...new function.
10080         * regrename.c (pass_regrename): Convert from a global struct to a
10081         subclass of rtl_opt_pass along with...
10082         (pass_data_regrename): ...new pass_data instance and...
10083         (make_pass_regrename): ...new function.
10084         * reorg.c (pass_delay_slots): Convert from a global struct to a
10085         subclass of rtl_opt_pass along with...
10086         (pass_data_delay_slots): ...new pass_data instance and...
10087         (make_pass_delay_slots): ...new function.
10088         (pass_machine_reorg): Convert from a global struct to a subclass of
10089         rtl_opt_pass along with...
10090         (pass_data_machine_reorg): ...new pass_data instance and...
10091         (make_pass_machine_reorg): ...new function.
10092         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10093         of rtl_opt_pass along with...
10094         (pass_data_sched): ...new pass_data instance and...
10095         (make_pass_sched): ...new function.
10096         (pass_sched2): Convert from a global struct to a subclass of
10097         rtl_opt_pass along with...
10098         (pass_data_sched2): ...new pass_data instance and...
10099         (make_pass_sched2): ...new function.
10100         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10101         to a subclass of rtl_opt_pass along with...
10102         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10103         (make_pass_stack_ptr_mod): ...new function.
10104         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10105         to a subclass of rtl_opt_pass along with...
10106         (pass_data_rtl_store_motion): ...new pass_data instance and...
10107         (make_pass_rtl_store_motion): ...new function.
10108         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10109         of gimple_opt_pass along with...
10110         (pass_data_tracer): ...new pass_data instance and...
10111         (make_pass_tracer): ...new function.
10112         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10113         to a subclass of gimple_opt_pass along with...
10114         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10115         (make_pass_diagnose_tm_blocks): ...new function.
10116         (pass_lower_tm): Convert from a global struct to a subclass of
10117         gimple_opt_pass along with...
10118         (pass_data_lower_tm): ...new pass_data instance and...
10119         (make_pass_lower_tm): ...new function.
10120         (pass_tm_init): Convert from a global struct to a subclass of
10121         gimple_opt_pass along with...
10122         (pass_data_tm_init): ...new pass_data instance and...
10123         (make_pass_tm_init): ...new function.
10124         (pass_tm_mark): Convert from a global struct to a subclass of
10125         gimple_opt_pass along with...
10126         (pass_data_tm_mark): ...new pass_data instance and...
10127         (make_pass_tm_mark): ...new function.
10128         (pass_tm_edges): Convert from a global struct to a subclass of
10129         gimple_opt_pass along with...
10130         (pass_data_tm_edges): ...new pass_data instance and...
10131         (make_pass_tm_edges): ...new function.
10132         (pass_tm_memopt): Convert from a global struct to a subclass of
10133         gimple_opt_pass along with...
10134         (pass_data_tm_memopt): ...new pass_data instance and...
10135         (make_pass_tm_memopt): ...new function.
10136         (pass_ipa_tm): Convert from a global struct to a subclass of
10137         simple_ipa_opt_pass along with...
10138         (pass_data_ipa_tm): ...new pass_data instance and...
10139         (make_pass_ipa_tm): ...new function.
10140         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10141         subclass of gimple_opt_pass along with...
10142         (pass_data_call_cdce): ...new pass_data instance and...
10143         (make_pass_call_cdce): ...new function.
10144         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10145         subclass of gimple_opt_pass along with...
10146         (pass_data_build_cfg): ...new pass_data instance and...
10147         (make_pass_build_cfg): ...new function.
10148         (pass_split_crit_edges): Convert from a global struct to a subclass of
10149         gimple_opt_pass along with...
10150         (pass_data_split_crit_edges): ...new pass_data instance and...
10151         (make_pass_split_crit_edges): ...new function.
10152         (pass_warn_function_return): Convert from a global struct to a
10153         subclass of gimple_opt_pass along with...
10154         (pass_data_warn_function_return): ...new pass_data instance and...
10155         (make_pass_warn_function_return): ...new function.
10156         (pass_warn_function_noreturn): Convert from a global struct to a
10157         subclass of gimple_opt_pass along with...
10158         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10159         (make_pass_warn_function_noreturn): ...new function.
10160         (pass_warn_unused_result): Convert from a global struct to a subclass
10161         of gimple_opt_pass along with...
10162         (pass_data_warn_unused_result): ...new pass_data instance and...
10163         (make_pass_warn_unused_result): ...new function.
10164         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10165         a subclass of gimple_opt_pass along with...
10166         (pass_data_merge_phi): ...new pass_data instance and...
10167         (make_pass_merge_phi): ...new function.
10168         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10169         a subclass of gimple_opt_pass along with...
10170         (pass_data_lower_complex): ...new pass_data instance and...
10171         (make_pass_lower_complex): ...new function.
10172         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10173         gimple_opt_pass along with...
10174         (pass_data_lower_complex_O0): ...new pass_data instance and...
10175         (make_pass_lower_complex_O0): ...new function.
10176         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10177         subclass of gimple_opt_pass along with...
10178         (pass_data_lower_eh): ...new pass_data instance and...
10179         (make_pass_lower_eh): ...new function.
10180         (pass_refactor_eh): Convert from a global struct to a subclass of
10181         gimple_opt_pass along with...
10182         (pass_data_refactor_eh): ...new pass_data instance and...
10183         (make_pass_refactor_eh): ...new function.
10184         (pass_lower_resx): Convert from a global struct to a subclass of
10185         gimple_opt_pass along with...
10186         (pass_data_lower_resx): ...new pass_data instance and...
10187         (make_pass_lower_resx): ...new function.
10188         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10189         of gimple_opt_pass along with...
10190         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10191         (make_pass_lower_eh_dispatch): ...new function.
10192         (pass_cleanup_eh): Convert from a global struct to a subclass of
10193         gimple_opt_pass along with...
10194         (pass_data_cleanup_eh): ...new pass_data instance and...
10195         (make_pass_cleanup_eh): ...new function.
10196         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10197         to a subclass of simple_ipa_opt_pass along with...
10198         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10199         (make_pass_ipa_lower_emutls): ...new function.
10200         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10201         a subclass of gimple_opt_pass along with...
10202         (pass_data_if_conversion): ...new pass_data instance and...
10203         (make_pass_if_conversion): ...new function.
10204         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10205         subclass of gimple_opt_pass along with...
10206         (pass_data_build_ssa): ...new pass_data instance and...
10207         (make_pass_build_ssa): ...new function.
10208         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10209         global struct to a subclass of gimple_opt_pass along with...
10210         (pass_data_loop_distribution): ...new pass_data instance and...
10211         (make_pass_loop_distribution): ...new function.
10212         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10213         subclass of gimple_opt_pass along with...
10214         (pass_data_mudflap_1): ...new pass_data instance and...
10215         (make_pass_mudflap_1): ...new function.
10216         (pass_mudflap_2): Convert from a global struct to a subclass of
10217         gimple_opt_pass along with...
10218         (pass_data_mudflap_2): ...new pass_data instance and...
10219         (make_pass_mudflap_2): ...new function.
10220         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10221         subclass of gimple_opt_pass along with...
10222         (pass_data_mudflap_1): ...new pass_data instance and...
10223         (make_pass_mudflap_1): ...new function.
10224         (pass_mudflap_2): Convert from a global struct to a subclass of
10225         gimple_opt_pass along with...
10226         (pass_data_mudflap_2): ...new pass_data instance and...
10227         (make_pass_mudflap_2): ...new function.
10228         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10229         gimple_opt_pass along with...
10230         (pass_data_nrv): ...new pass_data instance and...
10231         (make_pass_nrv): ...new function.
10232         (pass_return_slot): Convert from a global struct to a subclass of
10233         gimple_opt_pass along with...
10234         (pass_data_return_slot): ...new pass_data instance and...
10235         (make_pass_return_slot): ...new function.
10236         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10237         to a subclass of gimple_opt_pass along with...
10238         (pass_data_object_sizes): ...new pass_data instance and...
10239         (make_pass_object_sizes): ...new function.
10240         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10241         global struct to a subclass of gimple_opt_pass along with...
10242         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10243         and...
10244         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10245         (pass_fixup_cfg): Convert from a global struct to a subclass of
10246         gimple_opt_pass along with...
10247         (pass_data_fixup_cfg): ...new pass_data instance and...
10248         (make_pass_fixup_cfg): ...new function.
10249         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10250         (make_pass_mudflap_1): ...new function.
10251         (pass_mudflap_2): Replace declaration with that of...
10252         (make_pass_mudflap_2): ...new function.
10253         (pass_asan): Replace declaration with that of...
10254         (make_pass_asan): ...new function.
10255         (pass_asan_O0): Replace declaration with that of...
10256         (make_pass_asan_O0): ...new function.
10257         (pass_tsan): Replace declaration with that of...
10258         (make_pass_tsan): ...new function.
10259         (pass_tsan_O0): Replace declaration with that of...
10260         (make_pass_tsan_O0): ...new function.
10261         (pass_lower_cf): Replace declaration with that of...
10262         (make_pass_lower_cf): ...new function.
10263         (pass_refactor_eh): Replace declaration with that of...
10264         (make_pass_refactor_eh): ...new function.
10265         (pass_lower_eh): Replace declaration with that of...
10266         (make_pass_lower_eh): ...new function.
10267         (pass_lower_eh_dispatch): Replace declaration with that of...
10268         (make_pass_lower_eh_dispatch): ...new function.
10269         (pass_lower_resx): Replace declaration with that of...
10270         (make_pass_lower_resx): ...new function.
10271         (pass_build_cfg): Replace declaration with that of...
10272         (make_pass_build_cfg): ...new function.
10273         (pass_early_tree_profile): Replace declaration with that of...
10274         (make_pass_early_tree_profile): ...new function.
10275         (pass_cleanup_eh): Replace declaration with that of...
10276         (make_pass_cleanup_eh): ...new function.
10277         (pass_sra): Replace declaration with that of...
10278         (make_pass_sra): ...new function.
10279         (pass_sra_early): Replace declaration with that of...
10280         (make_pass_sra_early): ...new function.
10281         (pass_early_ipa_sra): Replace declaration with that of...
10282         (make_pass_early_ipa_sra): ...new function.
10283         (pass_tail_recursion): Replace declaration with that of...
10284         (make_pass_tail_recursion): ...new function.
10285         (pass_tail_calls): Replace declaration with that of...
10286         (make_pass_tail_calls): ...new function.
10287         (pass_tree_loop): Replace declaration with that of...
10288         (make_pass_tree_loop): ...new function.
10289         (pass_tree_loop_init): Replace declaration with that of...
10290         (make_pass_tree_loop_init): ...new function.
10291         (pass_lim): Replace declaration with that of...
10292         (make_pass_lim): ...new function.
10293         (pass_tree_unswitch): Replace declaration with that of...
10294         (make_pass_tree_unswitch): ...new function.
10295         (pass_predcom): Replace declaration with that of...
10296         (make_pass_predcom): ...new function.
10297         (pass_iv_canon): Replace declaration with that of...
10298         (make_pass_iv_canon): ...new function.
10299         (pass_scev_cprop): Replace declaration with that of...
10300         (make_pass_scev_cprop): ...new function.
10301         (pass_empty_loop): Replace declaration with that of...
10302         (make_pass_empty_loop): ...new function.
10303         (pass_record_bounds): Replace declaration with that of...
10304         (make_pass_record_bounds): ...new function.
10305         (pass_graphite): Replace declaration with that of...
10306         (make_pass_graphite): ...new function.
10307         (pass_graphite_transforms): Replace declaration with that of...
10308         (make_pass_graphite_transforms): ...new function.
10309         (pass_if_conversion): Replace declaration with that of...
10310         (make_pass_if_conversion): ...new function.
10311         (pass_loop_distribution): Replace declaration with that of...
10312         (make_pass_loop_distribution): ...new function.
10313         (pass_vectorize): Replace declaration with that of...
10314         (make_pass_vectorize): ...new function.
10315         (pass_slp_vectorize): Replace declaration with that of...
10316         (make_pass_slp_vectorize): ...new function.
10317         (pass_complete_unroll): Replace declaration with that of...
10318         (make_pass_complete_unroll): ...new function.
10319         (pass_complete_unrolli): Replace declaration with that of...
10320         (make_pass_complete_unrolli): ...new function.
10321         (pass_parallelize_loops): Replace declaration with that of...
10322         (make_pass_parallelize_loops): ...new function.
10323         (pass_loop_prefetch): Replace declaration with that of...
10324         (make_pass_loop_prefetch): ...new function.
10325         (pass_iv_optimize): Replace declaration with that of...
10326         (make_pass_iv_optimize): ...new function.
10327         (pass_tree_loop_done): Replace declaration with that of...
10328         (make_pass_tree_loop_done): ...new function.
10329         (pass_ch): Replace declaration with that of...
10330         (make_pass_ch): ...new function.
10331         (pass_ccp): Replace declaration with that of...
10332         (make_pass_ccp): ...new function.
10333         (pass_phi_only_cprop): Replace declaration with that of...
10334         (make_pass_phi_only_cprop): ...new function.
10335         (pass_build_ssa): Replace declaration with that of...
10336         (make_pass_build_ssa): ...new function.
10337         (pass_build_alias): Replace declaration with that of...
10338         (make_pass_build_alias): ...new function.
10339         (pass_build_ealias): Replace declaration with that of...
10340         (make_pass_build_ealias): ...new function.
10341         (pass_dominator): Replace declaration with that of...
10342         (make_pass_dominator): ...new function.
10343         (pass_dce): Replace declaration with that of...
10344         (make_pass_dce): ...new function.
10345         (pass_dce_loop): Replace declaration with that of...
10346         (make_pass_dce_loop): ...new function.
10347         (pass_cd_dce): Replace declaration with that of...
10348         (make_pass_cd_dce): ...new function.
10349         (pass_call_cdce): Replace declaration with that of...
10350         (make_pass_call_cdce): ...new function.
10351         (pass_merge_phi): Replace declaration with that of...
10352         (make_pass_merge_phi): ...new function.
10353         (pass_split_crit_edges): Replace declaration with that of...
10354         (make_pass_split_crit_edges): ...new function.
10355         (pass_pre): Replace declaration with that of...
10356         (make_pass_pre): ...new function.
10357         (pass_profile): Replace declaration with that of...
10358         (make_pass_profile): ...new function.
10359         (pass_strip_predict_hints): Replace declaration with that of...
10360         (make_pass_strip_predict_hints): ...new function.
10361         (pass_lower_complex_O0): Replace declaration with that of...
10362         (make_pass_lower_complex_O0): ...new function.
10363         (pass_lower_complex): Replace declaration with that of...
10364         (make_pass_lower_complex): ...new function.
10365         (pass_lower_vector): Replace declaration with that of...
10366         (make_pass_lower_vector): ...new function.
10367         (pass_lower_vector_ssa): Replace declaration with that of...
10368         (make_pass_lower_vector_ssa): ...new function.
10369         (pass_lower_omp): Replace declaration with that of...
10370         (make_pass_lower_omp): ...new function.
10371         (pass_diagnose_omp_blocks): Replace declaration with that of...
10372         (make_pass_diagnose_omp_blocks): ...new function.
10373         (pass_expand_omp): Replace declaration with that of...
10374         (make_pass_expand_omp): ...new function.
10375         (pass_expand_omp_ssa): Replace declaration with that of...
10376         (make_pass_expand_omp_ssa): ...new function.
10377         (pass_object_sizes): Replace declaration with that of...
10378         (make_pass_object_sizes): ...new function.
10379         (pass_strlen): Replace declaration with that of...
10380         (make_pass_strlen): ...new function.
10381         (pass_fold_builtins): Replace declaration with that of...
10382         (make_pass_fold_builtins): ...new function.
10383         (pass_stdarg): Replace declaration with that of...
10384         (make_pass_stdarg): ...new function.
10385         (pass_early_warn_uninitialized): Replace declaration with that of...
10386         (make_pass_early_warn_uninitialized): ...new function.
10387         (pass_late_warn_uninitialized): Replace declaration with that of...
10388         (make_pass_late_warn_uninitialized): ...new function.
10389         (pass_cse_reciprocals): Replace declaration with that of...
10390         (make_pass_cse_reciprocals): ...new function.
10391         (pass_cse_sincos): Replace declaration with that of...
10392         (make_pass_cse_sincos): ...new function.
10393         (pass_optimize_bswap): Replace declaration with that of...
10394         (make_pass_optimize_bswap): ...new function.
10395         (pass_optimize_widening_mul): Replace declaration with that of...
10396         (make_pass_optimize_widening_mul): ...new function.
10397         (pass_warn_function_return): Replace declaration with that of...
10398         (make_pass_warn_function_return): ...new function.
10399         (pass_warn_function_noreturn): Replace declaration with that of...
10400         (make_pass_warn_function_noreturn): ...new function.
10401         (pass_cselim): Replace declaration with that of...
10402         (make_pass_cselim): ...new function.
10403         (pass_phiopt): Replace declaration with that of...
10404         (make_pass_phiopt): ...new function.
10405         (pass_forwprop): Replace declaration with that of...
10406         (make_pass_forwprop): ...new function.
10407         (pass_phiprop): Replace declaration with that of...
10408         (make_pass_phiprop): ...new function.
10409         (pass_tree_ifcombine): Replace declaration with that of...
10410         (make_pass_tree_ifcombine): ...new function.
10411         (pass_dse): Replace declaration with that of...
10412         (make_pass_dse): ...new function.
10413         (pass_nrv): Replace declaration with that of...
10414         (make_pass_nrv): ...new function.
10415         (pass_rename_ssa_copies): Replace declaration with that of...
10416         (make_pass_rename_ssa_copies): ...new function.
10417         (pass_sink_code): Replace declaration with that of...
10418         (make_pass_sink_code): ...new function.
10419         (pass_fre): Replace declaration with that of...
10420         (make_pass_fre): ...new function.
10421         (pass_check_data_deps): Replace declaration with that of...
10422         (make_pass_check_data_deps): ...new function.
10423         (pass_copy_prop): Replace declaration with that of...
10424         (make_pass_copy_prop): ...new function.
10425         (pass_vrp): Replace declaration with that of...
10426         (make_pass_vrp): ...new function.
10427         (pass_uncprop): Replace declaration with that of...
10428         (make_pass_uncprop): ...new function.
10429         (pass_return_slot): Replace declaration with that of...
10430         (make_pass_return_slot): ...new function.
10431         (pass_reassoc): Replace declaration with that of...
10432         (make_pass_reassoc): ...new function.
10433         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10434         (make_pass_rebuild_cgraph_edges): ...new function.
10435         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10436         (make_pass_remove_cgraph_callee_edges): ...new function.
10437         (pass_build_cgraph_edges): Replace declaration with that of...
10438         (make_pass_build_cgraph_edges): ...new function.
10439         (pass_local_pure_const): Replace declaration with that of...
10440         (make_pass_local_pure_const): ...new function.
10441         (pass_tracer): Replace declaration with that of...
10442         (make_pass_tracer): ...new function.
10443         (pass_warn_unused_result): Replace declaration with that of...
10444         (make_pass_warn_unused_result): ...new function.
10445         (pass_diagnose_tm_blocks): Replace declaration with that of...
10446         (make_pass_diagnose_tm_blocks): ...new function.
10447         (pass_lower_tm): Replace declaration with that of...
10448         (make_pass_lower_tm): ...new function.
10449         (pass_tm_init): Replace declaration with that of...
10450         (make_pass_tm_init): ...new function.
10451         (pass_tm_mark): Replace declaration with that of...
10452         (make_pass_tm_mark): ...new function.
10453         (pass_tm_memopt): Replace declaration with that of...
10454         (make_pass_tm_memopt): ...new function.
10455         (pass_tm_edges): Replace declaration with that of...
10456         (make_pass_tm_edges): ...new function.
10457         (pass_split_functions): Replace declaration with that of...
10458         (make_pass_split_functions): ...new function.
10459         (pass_feedback_split_functions): Replace declaration with that of...
10460         (make_pass_feedback_split_functions): ...new function.
10461         (pass_strength_reduction): Replace declaration with that of...
10462         (make_pass_strength_reduction): ...new function.
10463         (pass_ipa_lower_emutls): Replace declaration with that of...
10464         (make_pass_ipa_lower_emutls): ...new function.
10465         (pass_ipa_function_and_variable_visibility): Replace declaration with
10466         that of...
10467         (make_pass_ipa_function_and_variable_visibility): ...new function.
10468         (pass_ipa_tree_profile): Replace declaration with that of...
10469         (make_pass_ipa_tree_profile): ...new function.
10470         (pass_early_local_passes): Replace declaration with that of...
10471         (make_pass_early_local_passes): ...new function.
10472         (pass_ipa_whole_program_visibility): Replace declaration with that
10473         of...
10474         (make_pass_ipa_whole_program_visibility): ...new function.
10475         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10476         (make_pass_ipa_lto_gimple_out): ...new function.
10477         (pass_ipa_increase_alignment): Replace declaration with that of...
10478         (make_pass_ipa_increase_alignment): ...new function.
10479         (pass_ipa_inline): Replace declaration with that of...
10480         (make_pass_ipa_inline): ...new function.
10481         (pass_ipa_free_lang_data): Replace declaration with that of...
10482         (make_pass_ipa_free_lang_data): ...new function.
10483         (pass_ipa_free_inline_summary): Replace declaration with that of...
10484         (make_pass_ipa_free_inline_summary): ...new function.
10485         (pass_ipa_cp): Replace declaration with that of...
10486         (make_pass_ipa_cp): ...new function.
10487         (pass_ipa_reference): Replace declaration with that of...
10488         (make_pass_ipa_reference): ...new function.
10489         (pass_ipa_pure_const): Replace declaration with that of...
10490         (make_pass_ipa_pure_const): ...new function.
10491         (pass_ipa_pta): Replace declaration with that of...
10492         (make_pass_ipa_pta): ...new function.
10493         (pass_ipa_lto_finish_out): Replace declaration with that of...
10494         (make_pass_ipa_lto_finish_out): ...new function.
10495         (pass_ipa_tm): Replace declaration with that of...
10496         (make_pass_ipa_tm): ...new function.
10497         (pass_ipa_profile): Replace declaration with that of...
10498         (make_pass_ipa_profile): ...new function.
10499         (pass_ipa_cdtor_merge): Replace declaration with that of...
10500         (make_pass_ipa_cdtor_merge): ...new function.
10501         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10502         of...
10503         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10504         (pass_init_datastructures): Replace declaration with that of...
10505         (make_pass_init_datastructures): ...new function.
10506         (pass_fixup_cfg): Replace declaration with that of...
10507         (make_pass_fixup_cfg): ...new function.
10508         (pass_expand): Replace declaration with that of...
10509         (make_pass_expand): ...new function.
10510         (pass_instantiate_virtual_regs): Replace declaration with that of...
10511         (make_pass_instantiate_virtual_regs): ...new function.
10512         (pass_rtl_fwprop): Replace declaration with that of...
10513         (make_pass_rtl_fwprop): ...new function.
10514         (pass_rtl_fwprop_addr): Replace declaration with that of...
10515         (make_pass_rtl_fwprop_addr): ...new function.
10516         (pass_jump): Replace declaration with that of...
10517         (make_pass_jump): ...new function.
10518         (pass_jump2): Replace declaration with that of...
10519         (make_pass_jump2): ...new function.
10520         (pass_lower_subreg): Replace declaration with that of...
10521         (make_pass_lower_subreg): ...new function.
10522         (pass_cse): Replace declaration with that of...
10523         (make_pass_cse): ...new function.
10524         (pass_fast_rtl_dce): Replace declaration with that of...
10525         (make_pass_fast_rtl_dce): ...new function.
10526         (pass_ud_rtl_dce): Replace declaration with that of...
10527         (make_pass_ud_rtl_dce): ...new function.
10528         (pass_rtl_dce): Replace declaration with that of...
10529         (make_pass_rtl_dce): ...new function.
10530         (pass_rtl_dse1): Replace declaration with that of...
10531         (make_pass_rtl_dse1): ...new function.
10532         (pass_rtl_dse2): Replace declaration with that of...
10533         (make_pass_rtl_dse2): ...new function.
10534         (pass_rtl_dse3): Replace declaration with that of...
10535         (make_pass_rtl_dse3): ...new function.
10536         (pass_rtl_cprop): Replace declaration with that of...
10537         (make_pass_rtl_cprop): ...new function.
10538         (pass_rtl_pre): Replace declaration with that of...
10539         (make_pass_rtl_pre): ...new function.
10540         (pass_rtl_hoist): Replace declaration with that of...
10541         (make_pass_rtl_hoist): ...new function.
10542         (pass_rtl_store_motion): Replace declaration with that of...
10543         (make_pass_rtl_store_motion): ...new function.
10544         (pass_cse_after_global_opts): Replace declaration with that of...
10545         (make_pass_cse_after_global_opts): ...new function.
10546         (pass_rtl_ifcvt): Replace declaration with that of...
10547         (make_pass_rtl_ifcvt): ...new function.
10548         (pass_into_cfg_layout_mode): Replace declaration with that of...
10549         (make_pass_into_cfg_layout_mode): ...new function.
10550         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10551         (make_pass_outof_cfg_layout_mode): ...new function.
10552         (pass_loop2): Replace declaration with that of...
10553         (make_pass_loop2): ...new function.
10554         (pass_rtl_loop_init): Replace declaration with that of...
10555         (make_pass_rtl_loop_init): ...new function.
10556         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10557         (make_pass_rtl_move_loop_invariants): ...new function.
10558         (pass_rtl_unswitch): Replace declaration with that of...
10559         (make_pass_rtl_unswitch): ...new function.
10560         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10561         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10562         (pass_rtl_doloop): Replace declaration with that of...
10563         (make_pass_rtl_doloop): ...new function.
10564         (pass_rtl_loop_done): Replace declaration with that of...
10565         (make_pass_rtl_loop_done): ...new function.
10566         (pass_web): Replace declaration with that of...
10567         (make_pass_web): ...new function.
10568         (pass_cse2): Replace declaration with that of...
10569         (make_pass_cse2): ...new function.
10570         (pass_df_initialize_opt): Replace declaration with that of...
10571         (make_pass_df_initialize_opt): ...new function.
10572         (pass_df_initialize_no_opt): Replace declaration with that of...
10573         (make_pass_df_initialize_no_opt): ...new function.
10574         (pass_reginfo_init): Replace declaration with that of...
10575         (make_pass_reginfo_init): ...new function.
10576         (pass_inc_dec): Replace declaration with that of...
10577         (make_pass_inc_dec): ...new function.
10578         (pass_stack_ptr_mod): Replace declaration with that of...
10579         (make_pass_stack_ptr_mod): ...new function.
10580         (pass_initialize_regs): Replace declaration with that of...
10581         (make_pass_initialize_regs): ...new function.
10582         (pass_combine): Replace declaration with that of...
10583         (make_pass_combine): ...new function.
10584         (pass_if_after_combine): Replace declaration with that of...
10585         (make_pass_if_after_combine): ...new function.
10586         (pass_ree): Replace declaration with that of...
10587         (make_pass_ree): ...new function.
10588         (pass_partition_blocks): Replace declaration with that of...
10589         (make_pass_partition_blocks): ...new function.
10590         (pass_match_asm_constraints): Replace declaration with that of...
10591         (make_pass_match_asm_constraints): ...new function.
10592         (pass_regmove): Replace declaration with that of...
10593         (make_pass_regmove): ...new function.
10594         (pass_split_all_insns): Replace declaration with that of...
10595         (make_pass_split_all_insns): ...new function.
10596         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10597         (make_pass_fast_rtl_byte_dce): ...new function.
10598         (pass_lower_subreg2): Replace declaration with that of...
10599         (make_pass_lower_subreg2): ...new function.
10600         (pass_mode_switching): Replace declaration with that of...
10601         (make_pass_mode_switching): ...new function.
10602         (pass_sms): Replace declaration with that of...
10603         (make_pass_sms): ...new function.
10604         (pass_sched): Replace declaration with that of...
10605         (make_pass_sched): ...new function.
10606         (pass_ira): Replace declaration with that of...
10607         (make_pass_ira): ...new function.
10608         (pass_reload): Replace declaration with that of...
10609         (make_pass_reload): ...new function.
10610         (pass_clean_state): Replace declaration with that of...
10611         (make_pass_clean_state): ...new function.
10612         (pass_branch_prob): Replace declaration with that of...
10613         (make_pass_branch_prob): ...new function.
10614         (pass_value_profile_transformations): Replace declaration with that
10615         of...
10616         (make_pass_value_profile_transformations): ...new function.
10617         (pass_postreload_cse): Replace declaration with that of...
10618         (make_pass_postreload_cse): ...new function.
10619         (pass_gcse2): Replace declaration with that of...
10620         (make_pass_gcse2): ...new function.
10621         (pass_split_after_reload): Replace declaration with that of...
10622         (make_pass_split_after_reload): ...new function.
10623         (pass_branch_target_load_optimize1): Replace declaration with that
10624         of...
10625         (make_pass_branch_target_load_optimize1): ...new function.
10626         (pass_thread_prologue_and_epilogue): Replace declaration with that
10627         of...
10628         (make_pass_thread_prologue_and_epilogue): ...new function.
10629         (pass_stack_adjustments): Replace declaration with that of...
10630         (make_pass_stack_adjustments): ...new function.
10631         (pass_peephole2): Replace declaration with that of...
10632         (make_pass_peephole2): ...new function.
10633         (pass_if_after_reload): Replace declaration with that of...
10634         (make_pass_if_after_reload): ...new function.
10635         (pass_regrename): Replace declaration with that of...
10636         (make_pass_regrename): ...new function.
10637         (pass_cprop_hardreg): Replace declaration with that of...
10638         (make_pass_cprop_hardreg): ...new function.
10639         (pass_reorder_blocks): Replace declaration with that of...
10640         (make_pass_reorder_blocks): ...new function.
10641         (pass_branch_target_load_optimize2): Replace declaration with that
10642         of...
10643         (make_pass_branch_target_load_optimize2): ...new function.
10644         (pass_leaf_regs): Replace declaration with that of...
10645         (make_pass_leaf_regs): ...new function.
10646         (pass_split_before_sched2): Replace declaration with that of...
10647         (make_pass_split_before_sched2): ...new function.
10648         (pass_compare_elim_after_reload): Replace declaration with that of...
10649         (make_pass_compare_elim_after_reload): ...new function.
10650         (pass_sched2): Replace declaration with that of...
10651         (make_pass_sched2): ...new function.
10652         (pass_stack_regs): Replace declaration with that of...
10653         (make_pass_stack_regs): ...new function.
10654         (pass_stack_regs_run): Replace declaration with that of...
10655         (make_pass_stack_regs_run): ...new function.
10656         (pass_df_finish): Replace declaration with that of...
10657         (make_pass_df_finish): ...new function.
10658         (pass_compute_alignments): Replace declaration with that of...
10659         (make_pass_compute_alignments): ...new function.
10660         (pass_duplicate_computed_gotos): Replace declaration with that of...
10661         (make_pass_duplicate_computed_gotos): ...new function.
10662         (pass_variable_tracking): Replace declaration with that of...
10663         (make_pass_variable_tracking): ...new function.
10664         (pass_free_cfg): Replace declaration with that of...
10665         (make_pass_free_cfg): ...new function.
10666         (pass_machine_reorg): Replace declaration with that of...
10667         (make_pass_machine_reorg): ...new function.
10668         (pass_cleanup_barriers): Replace declaration with that of...
10669         (make_pass_cleanup_barriers): ...new function.
10670         (pass_delay_slots): Replace declaration with that of...
10671         (make_pass_delay_slots): ...new function.
10672         (pass_split_for_shorten_branches): Replace declaration with that of...
10673         (make_pass_split_for_shorten_branches): ...new function.
10674         (pass_split_before_regstack): Replace declaration with that of...
10675         (make_pass_split_before_regstack): ...new function.
10676         (pass_convert_to_eh_region_ranges): Replace declaration with that
10677         of...
10678         (make_pass_convert_to_eh_region_ranges): ...new function.
10679         (pass_shorten_branches): Replace declaration with that of...
10680         (make_pass_shorten_branches): ...new function.
10681         (pass_set_nothrow_function_flags): Replace declaration with that of...
10682         (make_pass_set_nothrow_function_flags): ...new function.
10683         (pass_dwarf2_frame): Replace declaration with that of...
10684         (make_pass_dwarf2_frame): ...new function.
10685         (pass_final): Replace declaration with that of...
10686         (make_pass_final): ...new function.
10687         (pass_rtl_seqabstr): Replace declaration with that of...
10688         (make_pass_rtl_seqabstr): ...new function.
10689         (pass_release_ssa_names): Replace declaration with that of...
10690         (make_pass_release_ssa_names): ...new function.
10691         (pass_early_inline): Replace declaration with that of...
10692         (make_pass_early_inline): ...new function.
10693         (pass_inline_parameters): Replace declaration with that of...
10694         (make_pass_inline_parameters): ...new function.
10695         (pass_update_address_taken): Replace declaration with that of...
10696         (make_pass_update_address_taken): ...new function.
10697         (pass_convert_switch): Replace declaration with that of...
10698         (make_pass_convert_switch): ...new function.
10699         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10700         to a subclass of simple_ipa_opt_pass along with...
10701         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10702         (make_pass_ipa_tree_profile): ...new function.
10703         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10704         subclass of gimple_opt_pass along with...
10705         (pass_data_sra_early): ...new pass_data instance and...
10706         (make_pass_sra_early): ...new function.
10707         (pass_sra): Convert from a global struct to a subclass of
10708         gimple_opt_pass along with...
10709         (pass_data_sra): ...new pass_data instance and...
10710         (make_pass_sra): ...new function.
10711         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10712         gimple_opt_pass along with...
10713         (pass_data_early_ipa_sra): ...new pass_data instance and...
10714         (make_pass_early_ipa_sra): ...new function.
10715         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10716         subclass of gimple_opt_pass along with...
10717         (pass_data_ccp): ...new pass_data instance and...
10718         (make_pass_ccp): ...new function.
10719         (pass_fold_builtins): Convert from a global struct to a subclass of
10720         gimple_opt_pass along with...
10721         (pass_data_fold_builtins): ...new pass_data instance and...
10722         (make_pass_fold_builtins): ...new function.
10723         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10724         subclass of gimple_opt_pass along with...
10725         (pass_data_copy_prop): ...new pass_data instance and...
10726         (make_pass_copy_prop): ...new function.
10727         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10728         global struct to a subclass of gimple_opt_pass along with...
10729         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10730         (make_pass_rename_ssa_copies): ...new function.
10731         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10732         subclass of gimple_opt_pass along with...
10733         (pass_data_dce): ...new pass_data instance and...
10734         (make_pass_dce): ...new function.
10735         (pass_dce_loop): Convert from a global struct to a subclass of
10736         gimple_opt_pass along with...
10737         (pass_data_dce_loop): ...new pass_data instance and...
10738         (make_pass_dce_loop): ...new function.
10739         (pass_cd_dce): Convert from a global struct to a subclass of
10740         gimple_opt_pass along with...
10741         (pass_data_cd_dce): ...new pass_data instance and...
10742         (make_pass_cd_dce): ...new function.
10743         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10744         subclass of gimple_opt_pass along with...
10745         (pass_data_dominator): ...new pass_data instance and...
10746         (make_pass_dominator): ...new function.
10747         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10748         gimple_opt_pass along with...
10749         (pass_data_phi_only_cprop): ...new pass_data instance and...
10750         (make_pass_phi_only_cprop): ...new function.
10751         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10752         subclass of gimple_opt_pass along with...
10753         (pass_data_dse): ...new pass_data instance and...
10754         (make_pass_dse): ...new function.
10755         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10756         a subclass of gimple_opt_pass along with...
10757         (pass_data_forwprop): ...new pass_data instance and...
10758         (make_pass_forwprop): ...new function.
10759         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10760         struct to a subclass of gimple_opt_pass along with...
10761         (pass_data_tree_ifcombine): ...new pass_data instance and...
10762         (make_pass_tree_ifcombine): ...new function.
10763         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10764         subclass of gimple_opt_pass along with...
10765         (pass_data_ch): ...new pass_data instance and...
10766         (make_pass_ch): ...new function.
10767         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10768         subclass of gimple_opt_pass along with...
10769         (pass_data_tree_loop): ...new pass_data instance and...
10770         (make_pass_tree_loop): ...new function.
10771         (pass_tree_loop_init): Convert from a global struct to a subclass of
10772         gimple_opt_pass along with...
10773         (pass_data_tree_loop_init): ...new pass_data instance and...
10774         (make_pass_tree_loop_init): ...new function.
10775         (pass_lim): Convert from a global struct to a subclass of
10776         gimple_opt_pass along with...
10777         (pass_data_lim): ...new pass_data instance and...
10778         (make_pass_lim): ...new function.
10779         (pass_tree_unswitch): Convert from a global struct to a subclass of
10780         gimple_opt_pass along with...
10781         (pass_data_tree_unswitch): ...new pass_data instance and...
10782         (make_pass_tree_unswitch): ...new function.
10783         (pass_predcom): Convert from a global struct to a subclass of
10784         gimple_opt_pass along with...
10785         (pass_data_predcom): ...new pass_data instance and...
10786         (make_pass_predcom): ...new function.
10787         (pass_vectorize): Convert from a global struct to a subclass of
10788         gimple_opt_pass along with...
10789         (pass_data_vectorize): ...new pass_data instance and...
10790         (make_pass_vectorize): ...new function.
10791         (pass_graphite): Convert from a global struct to a subclass of
10792         gimple_opt_pass along with...
10793         (pass_data_graphite): ...new pass_data instance and...
10794         (make_pass_graphite): ...new function.
10795         (pass_graphite_transforms): Convert from a global struct to a subclass
10796         of gimple_opt_pass along with...
10797         (pass_data_graphite_transforms): ...new pass_data instance and...
10798         (make_pass_graphite_transforms): ...new function.
10799         (pass_check_data_deps): Convert from a global struct to a subclass of
10800         gimple_opt_pass along with...
10801         (pass_data_check_data_deps): ...new pass_data instance and...
10802         (make_pass_check_data_deps): ...new function.
10803         (pass_iv_canon): Convert from a global struct to a subclass of
10804         gimple_opt_pass along with...
10805         (pass_data_iv_canon): ...new pass_data instance and...
10806         (make_pass_iv_canon): ...new function.
10807         (pass_scev_cprop): Convert from a global struct to a subclass of
10808         gimple_opt_pass along with...
10809         (pass_data_scev_cprop): ...new pass_data instance and...
10810         (make_pass_scev_cprop): ...new function.
10811         (pass_record_bounds): Convert from a global struct to a subclass of
10812         gimple_opt_pass along with...
10813         (pass_data_record_bounds): ...new pass_data instance and...
10814         (make_pass_record_bounds): ...new function.
10815         (pass_complete_unroll): Convert from a global struct to a subclass of
10816         gimple_opt_pass along with...
10817         (pass_data_complete_unroll): ...new pass_data instance and...
10818         (make_pass_complete_unroll): ...new function.
10819         (pass_complete_unrolli): Convert from a global struct to a subclass of
10820         gimple_opt_pass along with...
10821         (pass_data_complete_unrolli): ...new pass_data instance and...
10822         (make_pass_complete_unrolli): ...new function.
10823         (pass_parallelize_loops): Convert from a global struct to a subclass
10824         of gimple_opt_pass along with...
10825         (pass_data_parallelize_loops): ...new pass_data instance and...
10826         (make_pass_parallelize_loops): ...new function.
10827         (pass_loop_prefetch): Convert from a global struct to a subclass of
10828         gimple_opt_pass along with...
10829         (pass_data_loop_prefetch): ...new pass_data instance and...
10830         (make_pass_loop_prefetch): ...new function.
10831         (pass_iv_optimize): Convert from a global struct to a subclass of
10832         gimple_opt_pass along with...
10833         (pass_data_iv_optimize): ...new pass_data instance and...
10834         (make_pass_iv_optimize): ...new function.
10835         (pass_tree_loop_done): Convert from a global struct to a subclass of
10836         gimple_opt_pass along with...
10837         (pass_data_tree_loop_done): ...new pass_data instance and...
10838         (make_pass_tree_loop_done): ...new function.
10839         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10840         struct to a subclass of gimple_opt_pass along with...
10841         (pass_data_cse_reciprocals): ...new pass_data instance and...
10842         (make_pass_cse_reciprocals): ...new function.
10843         (pass_cse_sincos): Convert from a global struct to a subclass of
10844         gimple_opt_pass along with...
10845         (pass_data_cse_sincos): ...new pass_data instance and...
10846         (make_pass_cse_sincos): ...new function.
10847         (pass_optimize_bswap): Convert from a global struct to a subclass of
10848         gimple_opt_pass along with...
10849         (pass_data_optimize_bswap): ...new pass_data instance and...
10850         (make_pass_optimize_bswap): ...new function.
10851         (pass_optimize_widening_mul): Convert from a global struct to a
10852         subclass of gimple_opt_pass along with...
10853         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10854         (make_pass_optimize_widening_mul): ...new function.
10855         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10856         subclass of gimple_opt_pass along with...
10857         (pass_data_phiopt): ...new pass_data instance and...
10858         (make_pass_phiopt): ...new function.
10859         (pass_cselim): Convert from a global struct to a subclass of
10860         gimple_opt_pass along with...
10861         (pass_data_cselim): ...new pass_data instance and...
10862         (make_pass_cselim): ...new function.
10863         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10864         subclass of gimple_opt_pass along with...
10865         (pass_data_phiprop): ...new pass_data instance and...
10866         (make_pass_phiprop): ...new function.
10867         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10868         subclass of gimple_opt_pass along with...
10869         (pass_data_pre): ...new pass_data instance and...
10870         (make_pass_pre): ...new function.
10871         (pass_fre): Convert from a global struct to a subclass of
10872         gimple_opt_pass along with...
10873         (pass_data_fre): ...new pass_data instance and...
10874         (make_pass_fre): ...new function.
10875         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10876         subclass of gimple_opt_pass along with...
10877         (pass_data_reassoc): ...new pass_data instance and...
10878         (make_pass_reassoc): ...new function.
10879         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10880         subclass of gimple_opt_pass along with...
10881         (pass_data_sink_code): ...new pass_data instance and...
10882         (make_pass_sink_code): ...new function.
10883         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10884         subclass of gimple_opt_pass along with...
10885         (pass_data_strlen): ...new pass_data instance and...
10886         (make_pass_strlen): ...new function.
10887         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10888         struct to a subclass of gimple_opt_pass along with...
10889         (pass_data_build_alias): ...new pass_data instance and...
10890         (make_pass_build_alias): ...new function.
10891         (pass_build_ealias): Convert from a global struct to a subclass of
10892         gimple_opt_pass along with...
10893         (pass_data_build_ealias): ...new pass_data instance and...
10894         (make_pass_build_ealias): ...new function.
10895         (pass_ipa_pta): Convert from a global struct to a subclass of
10896         simple_ipa_opt_pass along with...
10897         (pass_data_ipa_pta): ...new pass_data instance and...
10898         (make_pass_ipa_pta): ...new function.
10899         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10900         subclass of gimple_opt_pass along with...
10901         (pass_data_uncprop): ...new pass_data instance and...
10902         (make_pass_uncprop): ...new function.
10903         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10904         global struct to a subclass of gimple_opt_pass along with...
10905         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10906         (make_pass_late_warn_uninitialized): ...new function.
10907         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10908         to a subclass of gimple_opt_pass along with...
10909         (pass_data_init_datastructures): ...new pass_data instance and...
10910         (make_pass_init_datastructures): ...new function.
10911         (pass_early_warn_uninitialized): Convert from a global struct to a
10912         subclass of gimple_opt_pass along with...
10913         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10914         (make_pass_early_warn_uninitialized): ...new function.
10915         (pass_update_address_taken): Convert from a global struct to a
10916         subclass of gimple_opt_pass along with...
10917         (pass_data_update_address_taken): ...new pass_data instance and...
10918         (make_pass_update_address_taken): ...new function.
10919         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10920         struct to a subclass of gimple_opt_pass along with...
10921         (pass_data_release_ssa_names): ...new pass_data instance and...
10922         (make_pass_release_ssa_names): ...new function.
10923         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10924         subclass of gimple_opt_pass along with...
10925         (pass_data_stdarg): ...new pass_data instance and...
10926         (make_pass_stdarg): ...new function.
10927         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10928         global struct to a subclass of gimple_opt_pass along with...
10929         (pass_data_convert_switch): ...new pass_data instance and...
10930         (make_pass_convert_switch): ...new function.
10931         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10932         to a subclass of gimple_opt_pass along with...
10933         (pass_data_tail_recursion): ...new pass_data instance and...
10934         (make_pass_tail_recursion): ...new function.
10935         (pass_tail_calls): Convert from a global struct to a subclass of
10936         gimple_opt_pass along with...
10937         (pass_data_tail_calls): ...new pass_data instance and...
10938         (make_pass_tail_calls): ...new function.
10939         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10940         struct to a subclass of gimple_opt_pass along with...
10941         (pass_data_lower_vector): ...new pass_data instance and...
10942         (make_pass_lower_vector): ...new function.
10943         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
10944         gimple_opt_pass along with...
10945         (pass_data_lower_vector_ssa): ...new pass_data instance and...
10946         (make_pass_lower_vector_ssa): ...new function.
10947         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
10948         to a subclass of gimple_opt_pass along with...
10949         (pass_data_slp_vectorize): ...new pass_data instance and...
10950         (make_pass_slp_vectorize): ...new function.
10951         (pass_ipa_increase_alignment): Convert from a global struct to a
10952         subclass of simple_ipa_opt_pass along with...
10953         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
10954         (make_pass_ipa_increase_alignment): ...new function.
10955         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
10956         gimple_opt_pass along with...
10957         (pass_data_vrp): ...new pass_data instance and...
10958         (make_pass_vrp): ...new function.
10959         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
10960         subclass of simple_ipa_opt_pass along with...
10961         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
10962         (make_pass_ipa_free_lang_data): ...new function.
10963         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
10964         gimple_opt_pass along with...
10965         (pass_data_tsan): ...new pass_data instance and...
10966         (make_pass_tsan): ...new function.
10967         (pass_tsan_O0): Convert from a global struct to a subclass of
10968         gimple_opt_pass along with...
10969         (pass_data_tsan_O0): ...new pass_data instance and...
10970         (make_pass_tsan_O0): ...new function.
10971         * var-tracking.c (pass_variable_tracking): Convert from a global
10972         struct to a subclass of rtl_opt_pass along with...
10973         (pass_data_variable_tracking): ...new pass_data instance and...
10974         (make_pass_variable_tracking): ...new function.
10975         * web.c (pass_web): Convert from a global struct to a subclass of
10976         rtl_opt_pass along with...
10977         (pass_data_web): ...new pass_data instance and...
10978         (make_pass_web): ...new function.
10979         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
10980         declaration with that of...
10981         (make_pass_mode_switch_use): ...new function.
10982         (pass_resolve_sw_modes): Replace declaration with that of...
10983         (make_pass_resolve_sw_modes): ...new function.
10984         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
10985         from a global struct to a subclass of rtl_opt_pass along with...
10986         (pass_data_mode_switch_use): ...new pass_data instance and...
10987         (make_pass_mode_switch_use): ...new function.
10988         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
10989         from a global struct to a subclass of rtl_opt_pass along with...
10990         (pass_data_resolve_sw_modes): ...new pass_data instance and...
10991         (make_pass_resolve_sw_modes): ...new function.
10992         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
10993         struct to a subclass of rtl_opt_pass along with...
10994         (pass_data_insert_vzeroupper): ...new pass_data instance and...
10995         (make_pass_insert_vzeroupper): ...new function.
10996         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
10997         global struct to a subclass of rtl_opt_pass along with...
10998         (pass_data_work_around_errata): ...new pass_data instance and...
10999         (make_pass_work_around_errata): ...new function.
11000         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11001         struct to a subclass of rtl_opt_pass along with...
11002         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11003         (make_pass_mips_machine_reorg2): ...new function.
11005 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11007         * passes.c (pass_manager::operator new): New.
11009 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11011         Handwritten part of conversion of passes to C++ classes.
11013         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11014         (toplev.o): Add dep on PASS_MANAGER_H.
11015         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11016         of early local pases to reflect this moving from a global to a
11017         member of gcc::pass_manager.
11018         (cgraph_add_new_function): Likewise.
11019         * lto-cgraph.c (lto_output_node): Update for conversion of
11020         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11021         * passes.c (opt_pass::clone): New.
11022         (opt_pass::gate): New.
11023         (opt_pass::execute): New.
11024         (opt_pass::opt_pass): New.
11025         (pass_manager::execute_early_local_passes): New.
11026         (pass_manager::execute_pass_mode_switching): new.
11027         (finish_optimization_passes): Convert to...
11028         (pass_manager::finish_optimization_passes): ...this.
11029         (finish_optimization_passes): Update for conversion of passes to
11030         C++ classes.
11031         (register_dump_files_1): Use has_gate since we cannot portably
11032         check a vtable entry against NULL.
11033         (dump_one_pass): Likewise.
11034         (ipa_write_summaries_2): Likewise.
11035         (ipa_write_optimization_summaries_1): Likewise.
11036         (ipa_read_summaries_1): Likewise.
11037         (ipa_read_optimization_summaries_1): Likewise.
11038         (execute_ipa_stmt_fixups): Likewise.
11039         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11040         pass-creation functions rather than wiring up globals, and
11041         storing the results in fields of pass_manager generated using
11042         pass-instances.def.
11043         (pass_manager::dump_profile_report): Update for conversion of
11044         passes to C++ classes.
11045         (pass_manager::execute_ipa_summary_passes): Likewise.
11046         (execute_one_ipa_transform_pass): Likewise.
11047         (execute_one_pass): Use has_gate and has_execute since we cannot
11048         portably check a vtable entry against NULL.
11049         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11050         (pass_manager): Use pass-instances.def to add fields for the
11051         various pass instances.
11052         * toplev.c (finalize): Update for move of
11053         finish_optimization_passes to a method of gcc::pass_manager.
11054         * toplev.h (finish_optimization_passes): Move to method of class
11055         pass_manager.
11056         * tree-pass.h (struct pass_data): New.
11057         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11058         (opt_pass::gate): Convert to virtual function.
11059         (opt_pass::~opt_pass): New.
11060         (opt_pass::clone): New.
11061         (opt_pass::execute): Convert to virtual function.
11062         (opt_pass::opt_pass): New.
11063         (opt_pass::ctxt_): new.
11064         (gimple_opt_pass): Convert to subclass of opt_pass.
11065         (gimple_opt_pass::gimple_opt_pass): New.
11066         (rtl_opt_pass): Convert to subclass of opt_pass.
11067         (rtl_opt_pass::rtl_opt_pass): New.
11068         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11069         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11070         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11071         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11072         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11073         invocation of pass_mode_switching to reflect this moving from a
11074         global to a member of gcc::pass_manager.
11075         (ix86_option_override): Rework how pass_insert_vzeroupper is
11076         added to the pass_manager to reflect autogenerated changes.
11077         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11079 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11081         PR rtl-optimization/57708
11082         * recog.c (peep2_find_free_register): Validate all regs in a
11083         multi-reg mode.
11085 2013-08-05  Jan Hubicka  <jh@suse.cz>
11087         PR lto/57602
11088         * cgraph.c (verify_cgraph_node): Accept local flags from other
11089         partitions.
11090         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11091         (function_and_variable_visibility): Likewise.
11092         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11094 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11096         * graph.c (init_graph_slim_pretty_print): Remove.
11097         (print_graph_cfg): Do not call it.  Use local pretty printer.
11098         (start_graph_dump): Likewise.
11100 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11102         * gimple-pretty-print.c (buffer): Remove.
11103         (initialized): Likewise.
11104         (maybe_init_pretty_print): Likewise.
11105         (print_gimple_stmt): Do not call it.  Use non-static local
11106         pretty_printer variable.
11107         (print_gimple_expr): Likewise.
11108         (print_gimple_seq): Likewise.
11109         (gimple_dump_bb): Likewise.
11111 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11113         * asan.c (asan_pp): Remove.
11114         (asan_pp_initialized): Likewise.
11115         (asan_pp_initialize): Likewise.
11116         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11117         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11118         (asan_add_global): Likewise.
11120 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11122         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11123         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11124         * pretty-print.c (pp_formatted_text_data): Likewise.
11125         (pp_indent): Rename from pp_base_indent.
11126         (pp_format): Rename from pp_base_format.
11127         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11128         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11129         (pp_flush): Rename from pp_base_flush.
11130         (pp_set_line_maximum_length): Rename from
11131         pp_base_set_line_maximum_length.
11132         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11133         (pp_set_prefix): Rename from pp_base_set_prefix.
11134         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11135         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11136         (pp_append_text): Rename from pp_base_append_text.
11137         (pp_formatted_text): Rename from pp_base_formatted_text.
11138         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11139         (pp_remaining_character_count_for_line): Rename from
11140         pp_base_remaining_character_count_for_line.
11141         (pp_newline): Rename from pp_base_newline.
11142         (pp_character): Rename from pp_base_character.
11143         (pp_string): Rename from pp_base_string.
11144         (pp_maybe_space): Rename from pp_base_maybe_space.
11145         * asan.c (asan_pp_string): Adjust.
11146         (asan_emit_stack_protection): Likewise.
11147         (asan_add_global): Likewise.
11148         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11149         * tree-mudflap.c (mf_varname_tree): Likewise.
11150         * tree-pretty-print.c (pp_tree_identifier): Rename from
11151         pp_base_tree_identifier.
11152         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11153         Declare as function.
11155 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11157         * pretty-print.h (pp_bar_bar): New.
11158         (pp_ampersand_ampersand): Likewise.
11159         (pp_less_equal): Likewise.
11160         (pp_greater_equal): Likewise.
11161         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11162         printer functions instead of pp_string or operators and punctuators.
11163         (dump_gimple_call): Likewise.
11164         (dump_gimple_omp_for): Likewise.
11165         (dump_gimple_transaction): Likewise.
11166         (dump_gimple_phi): Likewise.
11167         (pp_gimple_stmt_1): Likewise.
11168         * sched-vis.c (print_insn): Likewise.
11169         * tree-mudflap.c (mf_varname_tree): Likewise.
11170         * tree-pretty-print.c (dump_block_node): Likewise.
11171         (dump_generic_node): Likewise.
11173 2013-08-02  Jan Hubicka  <jh@suse.cz>
11175         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11176         boundaries.
11177         * lto-streamer-out.c (tree_is_indexable): Results decls and
11178         parm decls are not indexable.
11179         (DFS_write_tree_body): Do not follow args and results.
11180         (hash_tree): Likewise.
11181         (output_functions): Rearrange so struct function is needed
11182         only when real body is output; be able to also ouptut abstract
11183         functions; output DECL_ARGUMENTS and DECL_RESULT.
11184         (lto_output): When not in WPA, ale store abstract functions.
11185         (write_symbol): Do not care about RESULT_DECL.
11186         (output_symbol_p): Handle correctly sbtract decls.
11187         * lto-streamer-in.c (input_function): Rearrange so struct
11188         function can be NULL at entry; allow streaming of
11189         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11190         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11191         sanity check during LTO.
11192         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11193         RESULT_DECl and DECL_ARGUMENTS.
11194         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11195         Likewise.
11197 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11199         * pretty-print.h (pp_underscore): New.
11200         (pp_comma): Tidy.
11201         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11202         printer functions instead of pp_character.
11203         (dump_binary_rhs): Likewise.
11204         (dump_ternary_rhs): Likewise.
11205         (dump_gimple_call_args): Likewise.
11206         (pp_points_to_solution): Likewise.
11207         (dump_gimple_call): Likewise.
11208         (dump_gimple_switch): Likewise.
11209         (dump_gimple_cond): Likewise.
11210         (dump_gimple_bind): Likewise.
11211         (dump_gimple_try): Likewise.
11212         (dump_gimple_omp_for): Likewise.
11213         (dump_gimple_omp_continue): Likewise.
11214         (dump_gimple_omp_single): Likewise.
11215         (dump_gimple_omp_sections): Likewise.
11216         (dump_gimple_omp_block): Likewise.
11217         (dump_gimple_omp_critical): Likewise.
11218         (dump_gimple_transaction): Likewise.
11219         (dump_gimple_asm): Likewise.
11220         (dump_gimple_phi): Likewise.
11221         (dump_gimple_omp_parallel): Likewise.
11222         (dump_gimple_omp_task): Likewise.
11223         (dump_gimple_omp_atomic_load): Likewise.
11224         (dump_gimple_omp_atomic_store): Likewise.
11225         (dump_gimple_mem_ops): Likewise.
11226         (pp_gimple_stmt_1): Likewise.
11227         (pp_cfg_jump): Likewise.
11228         (dump_implicit_edges): Likewise.
11229         (gimple_dump_bb_for_graph): Likewise.
11230         * graph.c (draw_cfg_node): Likewise.
11231         * langhooks.c (lhd_print_error_function): Likewise.
11232         * sched-vis.c (print_exp): Likewise.
11233         (print_value): Likewise.
11234         (print_pattern): Likewise.
11235         (print_insn): Likewise.
11236         (rtl_dump_bb_for_graph): Likewise.
11237         * tree-pretty-print.c (dump_function_declaration): Likewise.
11238         (dump_array_domain): Likewise.
11239         (dump_omp_clause): Likewise.
11240         (dump_location): Likewise.
11241         (dump_generic_node): Likewise.
11242         (print_struct_decl): Likewise.
11243         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11245 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11247         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11248         candidate table when replacing a candidate statement.
11249         (replace_rhs_if_not_dup): Likewise.
11250         (replace_one_candidate): Likewise.
11252 2013-08-02  Jan Hubicka  <jh@suse.cz>
11253             Martin Liska  <marxin.liska@gmail.com>
11255         * cgraphunit.c (add_new_function): Fix logic when adding from
11256         late IPA pass.
11257         (assemble_thunk): Rename to ...
11258         (expand_thunk); .. this one; export; get it working with
11259         general functions; make produced gimple valid.
11260         * cgraph.h (expand_thunk): Declare.
11262 2013-08-02  Jan Hubicka  <jh@suse.cz>
11264         * ipa-cp.c (gather_context_independent_values): Use
11265         ipa_get_param_move_cost.
11266         (get_replacement_map): Remove PARAM; move parameter folding
11267         into tree-inline.c
11268         (create_specialized_node): Update.
11269         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11270         assert that we have gimple body; update move_cost.
11271         (count_formal_params): Assert that we have gimple body.
11272         (ipa_dump_param): New function.
11273         (ipa_alloc_node_params): Break out from ...
11274         (ipa_initialize_node_params): ... here.
11275         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11276         (ipa_write_node_info): Stream move costs.
11277         (ipa_read_node_info): Read move costs.
11278         (ipa_update_after_lto_read): Do not recompute node params.
11279         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11280         (ipa_get_param): Check we are not in WPA.
11281         (ipa_get_param_move_cost): New.
11282         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11283         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11284         parm numbers to be present.
11286 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11288         PR rtl-optimization/58048
11289         * lra-constraints.c (process_alt_operands): Don't check asm
11290         operand on register.
11292 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11294         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11295         the implied StoreLoad barrier for atomic operations if before.
11297 2013-08-02  Jan Hubicka  <jh@suse.cz>
11298             Martin Liska  <marxin.liska@gmail.com>
11300         * cgraph.c (cgraph_function_body_availability): Do not check
11301         cgraph flags.
11302         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11303         symtab_node_availability): Declare.
11304         * ipa.c (can_replace_by_local_alias): New.
11305         (function_and_variable_visibility): Use it.
11306         * symtab.c (symtab_for_node_and_aliases,
11307         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11309 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11311         PR rtl-optimization/57963
11312         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11313         (lra_constraints): Use them.
11315 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11317         * config/arm/types.md (define_attr "type"): Add "load_acq"
11318         and "store_rel".
11319         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11320         changes.
11321         (cortex_a53_store1): Likewise.
11323 2013-08-01  Jan Hubicka  <jh@suse.cz>
11325         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11326         partitions are not needed.
11328 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11330         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11331         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11332         MAYBE_NON_Q_CLASS_P where appropriate.
11334 2013-08-01  Jan Hubicka  <jh@suse.cz>
11336         * cgraph.h (release_function_body): Declare.
11337         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11338         of unused delcarations.
11340 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11342         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11343         RTL form when subtracting a constant.
11345 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11347         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11348         Generate canonical plus rtx with negated immediate instead of minus
11349         where appropriate.
11350         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11352 2013-08-01  Jan Hubicka  <jh@suse.cz>
11354         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11355         (cgraph_release_function_body): Likewise.
11356         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11357         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11358         to used_as_abstract_origin.
11359         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11360         symbols used as abstract origins.
11361         * cgraphunit.c (analyze_functions): Update.
11362         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11363         used_as_abstract_origin.
11364         * tree-inline.c (tree_function_versioning): Update
11365         used_as_abstract_origin; be ready for DECL_RESULT and
11366         DECL_ARGUMENTS to be NULL.
11368         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11369         for abstract functions.
11370         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11371         real symbols.
11373 2013-08-01  Jan Hubicka  <jh@suse.cz>
11375         * profile.c (compute_value_histograms): Fix thinko.
11377 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11379         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11380         aarch-common-protos.h to extra_headers.
11381         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11382         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11383         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11385 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11387         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11388         Include "../arm/types.md".  Define "type" attribute for all patterns.
11389         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11390         attribute changes.
11392 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11394         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11395         to support power8 load fusion.
11396         (fusion_gpr_mem_load): Likewise.
11398         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11400         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11401         declarations for power8 load fusion.
11402         (emit_fusion_gpr_load): Likewise.
11404         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11405         tuning for power8, turn on fusion mode by default.  Turn on sign
11406         extending fusion mode if normal fusion mode is on, and we are at
11407         -O2 or -O3.
11408         (fusion_gpr_load_p): New function, return true if we can fuse an
11409         addis instruction with a dependent load to a GPR.
11410         (emit_fusion_gpr_load): Emit the instructions for power8 load
11411         fusion to GPRs.
11413         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11414         (VSX load fusion peepholes): New peepholes to fuse together an
11415         addi instruction with a VSX load instruction.
11417         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11418         peepholes to fuse an addis instruction with a load to a GPR base
11419         register.  If we are supporting sign extending fusions, convert
11420         sign extending loads to zero extending loads and add an explicit
11421         sign extension.
11423 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11425         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11426         aarch-common-protos.h to extra_headers.
11427         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11428         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11429         (arm_early_store_addr_dep): Likewise.
11430         (arm_no_early_alu_shift_dep): Likewise.
11431         (arm_no_early_alu_shift_value_dep): Likewise.
11432         (arm_no_early_mul_dep): Likewise.
11433         (arm_no_early_store_addr_dep): Likewise.
11434         (arm_mac_accumulator_is_mul_result): Likewise.
11435         (arm_mac_accumulator_is_result): Likewise.
11436         * config/arm/aarch-common.c: ... here.  New file.
11437         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11438         here to ...
11439         (arm_early_store_addr_dep): Likewise.
11440         (arm_no_early_alu_shift_dep): Likewise.
11441         (arm_no_early_alu_shift_value_dep): Likewise.
11442         (arm_no_early_mul_dep): Likewise.
11443         (arm_no_early_store_addr_dep): Likewise.
11444         (arm_mac_accumulator_is_mul_result): Likewise.
11445         (arm_mac_accumulator_is_result): Likewise.
11446         * config/arm/aarch-common-protos.h: ... here.  New file.
11447         * config/arm/t-arm (aarch-common.o): Define.
11449 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11451         * config/arm/arm.md: Include new file "types.md".
11452         (define_attr "type"): Move from here to ...
11453         (define_attr "mul32"): Likewise.
11454         (define_attr "mul64"): Likewise.
11455         * config/arm/types.md: ... here.  New file.
11457 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11459         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11460         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11462 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11464         * gen-pass-instances.awk: Fix offset of substr().
11466 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11468         * Makefile.in (pass-instances.def): New.
11469         (passes.o): Replace dependency on passes.def with one on
11470         pass-instances.def
11472         * gen-pass-instances.awk: New.
11474         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11475         rather than passes.def, updating local definition of NEXT_PASS
11476         macro to add an extra NUM parameter (currently unused).
11478 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11480         * Makefile.in (PASS_MANAGER_H): New.
11481         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11482         (passes.o): Likewise.
11483         (statistics.o): Likewise.
11484         (cgraphunit.o): Likewise.
11485         (context.o): Depend on PASS_MANAGER_H.
11487         * pass_manager.h: New.
11489         * cgraphunit.c (cgraph_add_new_function): Update for moves
11490         of globals to fields of pass_manager.
11491         (analyze_function): Likewise.
11492         (expand_function): Likewise.
11493         (ipa_passes): Likewise.
11494         (compile): Likewise.
11496         * context.c (context::context): New.
11497         * context.h  (context::context): New.
11498         (context::get_passes): New.
11499         (context::passes_): New.
11501         * lto-cgraph.c (input_node): Update for moves of globals to
11502         fields of pass_manager.
11504         * passes.c (all_passes): Remove, in favor of a field of the
11505         same name within the new class pass_manager.
11506         (all_small_ipa_passes): Likewise.
11507         (all_lowering_passes): Likewise.
11508         (all_regular_ipa_passes): Likewise.
11509         (all_late_ipa_passes): Likewise.
11510         (all_lto_gen_passes): Likewise.
11511         (passes_by_id): Likewise.
11512         (passes_by_id_size): Likewise.
11513         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11514         the new class pass_manager.
11515         (set_pass_for_id): Convert to...
11516         (pass_manager::set_pass_for_id): ...method.
11517         (get_pass_for_id): Convert to...
11518         (pass_manager::get_pass_for_id): ...method.
11519         (register_one_dump_file): Move body of implementation into...
11520         (pass_manager::register_one_dump_file): ...here.
11521         (register_dump_files_1): Convert to...
11522         (pass_manager::register_dump_files_1): ...method.
11523         (register_dump_files): Convert to...
11524         (pass_manager::register_dump_files): ...method.
11525         (create_pass_tab): Update for moves of globals to fields of
11526         pass_manager.
11527         (dump_passes): Move body of implementation into...
11528         (pass_manager::dump_passes): ...here.
11529         (register_pass): Move body of implementation into...
11530         (pass_manager::register_pass): ...here.
11531         (init_optimization_passes): Convert into...
11532         (pass_manager::pass_manager): ...constructor for new
11533         pass_manager class, and initialize the pass_lists array.
11534         (check_profile_consistency): Update for moves of globals to
11535         fields of pass_manager.
11536         (dump_profile_report): Move body of implementation into...
11537         (pass_manager::dump_profile_report): ...here.
11538         (ipa_write_summaries_1): Update for moves of pass lists from
11539         being globals to fields of pass_manager.
11540         (ipa_write_optimization_summaries): Likewise.
11541         (ipa_read_summaries):  Likewise.
11542         (ipa_read_optimization_summaries): Likewise.
11543         (execute_all_ipa_stmt_fixups): Likewise.
11545         * statistics.c (statistics_fini): Update for moves of globals to
11546         fields of pass_manager.
11548         * toplev.c (general_init): Replace call to
11549         init_optimization_passes with construction of the pass_manager
11550         instance.
11552         * tree-pass.h (all_passes): Remove, in favor of a field of the
11553         same name within the new class pass_manager.
11554         (all_small_ipa_passes): Likewise.
11555         (all_lowering_passes): Likewise.
11556         (all_regular_ipa_passes): Likewise.
11557         (all_lto_gen_passes): Likewise.
11558         (all_late_ipa_passes): Likewise.
11559         (passes_by_id): Likewise.
11560         (passes_by_id_size): Likewise.
11561         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11562         the new class pass_manager.
11563         (get_pass_for_id): Remove.
11565 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11567         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11568         configs.
11570 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11572         * arm.md (mulhi3): New expand pattern.
11574 2013-07-30  Jan Hubicka  <jh@suse.cz>
11575             Martin Liska  <marxin.liska@gmail.com>
11577         * profile.c (compute_value_histograms): Do not ICE when
11578         there is mismatch only on some counters.
11580 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11582         PR rtl-optimization/57637
11583         * function.c (move_insn_for_shrink_wrap): Also check the
11584         GEN set of the LIVE problem for the liveness analysis
11585         if it exists, otherwise give up.
11587 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11589         PR tree-optimization/57993
11590         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11591         replaced statement in the candidate table.
11592         (phi_add_costs): Return infinite cost when the hidden basis does
11593         not dominate all phis on which the candidate is dependent.
11594         (replace_one_candidate): Record replaced statement in the
11595         candidate table.
11597 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11599         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11600         (ashlv2si3): New expander.
11601         (*ashlv2si3_i): New define_insn_and_split.
11602         * predicates.md (float_operation): Allow patterns with three
11603         basic sub-patterns.
11605         PR rtl-optimization/58021
11606         * mode-switching.c (create_pre_exit): Always split off preceding
11607         insns if we are not at the basic block head.
11609 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11611         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11612         (UCLIBC_DYNAMIC_LINKER): New macro.
11613         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11614         `-mnan=2008'.
11615         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11616         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11617         `-mnan=2008'.
11618         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11619         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11620         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11621         for SF and DF modes.  Use ieee_quad_format for TF mode.
11622         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11623         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11624         (mips_option_override): Handle `-mnan=legacy'.
11625         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11626         `-mabs=2008' and `-mnan=2008'.
11627         (OPTION_DEFAULT_SPECS): Add "nan" default.
11628         (ASM_SPEC): Handle `-mnan='.
11629         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11630         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11631         comment accordingly.
11632         (neg<mode>2): Likewise.
11633         * config/mips/mips.opt (mabs, mnan): New options.
11634         * doc/install.texi (Configuration): Document `--with-nan=' option.
11635         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11636         `-mnan=' options.
11637         (MIPS Options): Document them.
11638         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11639         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11640         * configure: Regenerate.
11641         * config.in: Regenerate.
11643 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11645         * config/i386/i386.md (float post-reload splitters): Do not check
11646         for subregs of SSE registers.
11648 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11649             H.J. Lu  <hongjiu.lu@intel.com>
11651         PR target/57954
11652         PR target/57988
11653         * config/i386/i386.md (post-reload splitter
11654         to avoid partial SSE reg dependency stalls): New pattern.
11656 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11658         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11660 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11662         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11663         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11665 2013-07-26  Cary Coutant  <ccoutant@google.com>
11667         * dwarf2out.c (die_checksum_ordered): Don't include template
11668         instantiations in signature.
11669         (is_template_parameter): New function.
11670         (is_template_instantiation): New function.
11671         (generate_skeleton_bottom_up): Don't include template instantiations
11672         in type unit DIE.
11673         (generate_skeleton): Likewise.
11674         (break_out_comdat_types): Move recursive call to break out nested
11675         types earlier.
11676         (prune_unused_types_mark_generic_parms_dies): Call
11677         is_template_parameter.
11679 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11681         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11682         uses vector registers.
11683         * config/aarch64/iterators.md: Add attributes rtn and vas.
11685 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11686             Richard Earnshaw  <richard.earnshaw@arm.com>
11688         * combine.c (simplify_comparison): Re-canonicalize operands
11689         where appropriate.
11690         * config/arm/arm.md (movcond_addsi): New splitter.
11692 2013-07-25  Sterling Augustine  <saugustine@google.com>
11694         * dwarf2out.c (size_of_pubnames): Move code to...
11695         (include_pubname_in_output): ...here.  New.
11696         (want_pubnames): Rearrange.
11697         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11699 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11701         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11703 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11705         PR target/38836
11706         * doc/extend.texi: Remove obsolete builtins. Fix
11707         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11709 2013-07-25  Jan Hubicka  <jh@suse.cz>
11711         * cgraph.c (release_function_body): Break out from ...
11712         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11713         and DECL_ARGUMENTS.
11714         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11715         old_tree in the map.
11716         (create_specialized_node): Update.
11717         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11718         into index.
11719         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11720         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11721         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11722         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11723         DECL_RESULT.
11725 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11727         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11728         addsi3_carryin_alt2_<optab>): Correct output template.
11730 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11732         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11733         Adjust for arm_restrict_it.
11734         Remove trailing whitespace.
11736 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11738         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11739         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11741         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11743 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11745         PR rtl-optimization/57960
11746         * lra-constraints.c (process_alt_operands): Use the right mode
11747         when checking strict_low.
11749 2013-07-25  Jan Hubicka  <jh@suse.cz>
11751         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11752         * cgraph.c (cgraph_remove_node): Do not release function body
11753         when in cgraph streaming.
11754         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11755         in other partitions are not considered reachable; fix handling of
11756         clones.
11758 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11760         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11762 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11764         PR target/19599
11765         PR target/57731
11766         PR target/57837
11767         * config/arm/arm.md ("*sibcall_insn): Replace use of
11768         Ss with US. Adjust output for v5 and v4t.
11769         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11771         * config/arm/constraints.md ("Ss"): Rename to US.
11773 2013-07-25  Terry Guo  <terry.guo@arm.com>
11775         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11776         shift_add/shift_sub0/shift_sub1 RTXs.
11778 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11779             Anton Blanchard  <anton@au1.ibm.com>
11781         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11782         (altivec_vpks<VI_char>ss): Likewise.
11783         (altivec_vpks<VI_char>us): Likewise.
11784         (altivec_vpku<VI_char>us): Likewise.
11785         (altivec_vpku<VI_char>um): Likewise.
11787 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11789         Introduce context class.
11791         * Makefile.in (CONTEXT_H): New.
11792         (OBJS): Add context.o.
11793         (toplev.o): Add CONTEXT_H to dependencies.
11794         (context.o): New.
11796         * toplev.c (general_init): Create the singleton gcc::context instance.
11798         * context.c: New.
11800         * context.h: New.
11802 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11804         PR rtl-optimization/57968
11805         * mode-switching.c (create_pre_exit): Allow instructions that
11806         don't set a return register to need a non-exit mode.
11808 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11809             Anton Blanchard  <anton@au1.ibm.com>
11811         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11812         operands to vperm for little endian.
11813         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11814         of lvsl to create the control mask for a vperm for little endian.
11816 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11817             Anton Blanchard  <anton@au1.ibm.com>
11819         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11820         two operands for little-endian.
11822 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11824         * config/mips/mips.c (mips_case_values_threshold): New.
11825         (TARGET_CASE_VALUES_THRESHOLD): Define.
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): Correct
11831         selection of field for vector splat in little endian mode.
11833 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11835         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11836         expanders to rs6000.md.
11837         (ior<mode>3): Likewise.
11838         (and<mode>3): Likewise.
11839         (one_cmpl<mode>2): Likewise.
11840         (nor<mode>3): Likewise.
11841         (andc<mode>3): Likewise.
11842         (eqv<mode>3): Likewise.
11843         (nand<mode>3): Likewise.
11844         (orc<mode>3): Likewise.
11846         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11847         declaration.
11849         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11850         to split multi-word logical operations.
11851         (rs6000_split_logical_di): Likewise.
11852         (rs6000_split_logical): Likewise.
11854         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11855         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11856         and allow TImode operations in 32-bit.
11857         (vsx_and<mode>3_64bit): Likewise.
11858         (vsx_ior<mode>3_32bit): Likewise.
11859         (vsx_ior<mode>3_64bit): Likewise.
11860         (vsx_xor<mode>3_32bit): Likewise.
11861         (vsx_xor<mode>3_64bit): Likewise.
11862         (vsx_one_cmpl<mode>2_32bit): Likewise.
11863         (vsx_one_cmpl<mode>2_64bit): Likewise.
11864         (vsx_nor<mode>3_32bit): Likewise.
11865         (vsx_nor<mode>3_64bit): Likewise.
11866         (vsx_andc<mode>3_32bit): Likewise.
11867         (vsx_andc<mode>3_64bit): Likewise.
11868         (vsx_eqv<mode>3_32bit): Likewise.
11869         (vsx_eqv<mode>3_64bit): Likewise.
11870         (vsx_nand<mode>3_32bit): Likewise.
11871         (vsx_nand<mode>3_64bit): Likewise.
11872         (vsx_orc<mode>3_32bit): Likewise.
11873         (vsx_orc<mode>3_64bit): Likewise.
11875         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11876         logical types in GPRs.
11878         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11879         logical insns to rs6000.md, and allow TImode operations in
11880         32-bit.
11881         (altivec_ior<mode>3): Likewise.
11882         (altivec_xor<mode>3): Likewise.
11883         (altivec_one_cmpl<mode>2): Likewise.
11884         (altivec_nor<mode>3): Likewise.
11885         (altivec_andc<mode>3): Likewise.
11887         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11888         attributes for moving the 128-bit logical operations into
11889         rs6000.md.
11890         (BOOL_REGS_OUTPUT): Likewise.
11891         (BOOL_REGS_OP1): Likewise.
11892         (BOOL_REGS_OP2): Likewise.
11893         (BOOL_REGS_UNARY): Likewise.
11894         (BOOL_REGS_AND_CR0): Likewise.
11895         (one_cmpl<mode>2): Add support for DI logical operations on
11896         32-bit, splitting the operations to 32-bit.
11897         (anddi3): Likewise.
11898         (iordi3): Likewise.
11899         (xordi3): Likewise.
11900         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11901         changes to combine the 32/64-bit code, allow logical operations on
11902         TI mode in 32-bit, and to use similar match_operator patterns like
11903         scalar mode uses.  Combine the Altivec and VSX code for logical
11904         operations, and move it here.
11905         (ior<mode>3, 128-bit types): Likewise.
11906         (xor<mode>3, 128-bit types): Likewise.
11907         (one_cmpl<mode>3, 128-bit types): Likewise.
11908         (nor<mode>3, 128-bit types): Likewise.
11909         (andc<mode>3, 128-bit types): Likewise.
11910         (eqv<mode>3, 128-bit types): Likewise.
11911         (nand<mode>3, 128-bit types): Likewise.
11912         (orc<mode>3, 128-bit types): Likewise.
11913         (and<mode>3_internal): Likewise.
11914         (bool<mode>3_internal): Likewise.
11915         (boolc<mode>3_internal1): Likewise.
11916         (boolc<mode>3_internal2): Likewise.
11917         (boolcc<mode>3_internal1): Likewise.
11918         (boolcc<mode>3_internal2): Likewise.
11919         (eqv<mode>3_internal1): Likewise.
11920         (eqv<mode>3_internal2): Likewise.
11921         (one_cmpl1<mode>3_internal): Likewise.
11923 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11925         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11926         Rename flag_stack_usage to flag_stack_usage_info.
11928 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11930         * config/microblaze/sync.md: New file.
11931         * config/microblaze/microblaze.md: Include sync.md
11932         * config/microblaze/microblaze.c: Add print_operand 'y'.
11933         * config/microblaze/constraints.md: Add memory_contraint
11934         'Q' which is a single register.
11936 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11938         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11940 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11942         * reload.c (find_reloads): Exit loop once we find this operand
11943         cannot be reloaded somehow for this alternative.
11945         * reload.c (find_reloads): Exit loop once we find a hard register.
11947         * rtlanal.c (computed_jump_p): Exit loop once we find label
11948         reference is used.
11950         * i386.c (ix86_pad_returns): Exit loop after setting replace.
11952         * cfgloopmanip.c (remove_path): Exit loop after setting
11953         irred_invalidated.
11955         * gensupport.c (subst_dup): Avoid loop if code is not
11956         MATCH_DUP nor MATCH_OP_DUP.
11958 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
11960         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
11962 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11964         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
11965         true for SP_REGNUM if mode == ptr_mode.
11966         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
11967         with value R0_REGNUM + 31.
11969 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11971         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
11972         pad pointer-typed argument downward.
11974 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11976         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
11977         and __ILP32__ when the ILP32 model is in use.
11979 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11981         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
11982         (aarch64_load_symref_appropriately): In the case of
11983         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
11984         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
11985         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
11986         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
11987         if 'mode' doesn't equal to 'ptr_mode'.
11988         (aarch64_output_mi_thunk): Add an assertion on the alignment of
11989         'vcall_offset'; change to call aarch64_emit_move differently depending
11990         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
11991         to calculate the upper bound of 'vcall_offset'.
11992         (aarch64_cannot_force_const_mem): Change to also return true if
11993         mode != ptr_mode.
11994         (aarch64_legitimize_reload_address): In the case of large
11995         displacements, add new local variable 'xmode' and an assertion
11996         based on it; change to use 'xmode' to generate the new rtx and
11997         reload.
11998         (aarch64_asm_trampoline_template): Change to generate the template
11999         differently depending on TARGET_ILP32 or not; change to use
12000         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12001         (aarch64_trampoline_size): Removed.
12002         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12003         and replace immediate literals with it.  Change to use 'ptr_mode'
12004         instead of 'DImode' and call convert_memory_address if the mode
12005         of 'fnaddr' doesn't equal to 'ptr_mode'.
12006         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12007         to output symbol.
12008         (aarch64_elf_asm_destructor): Likewise.
12009         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12010         on TARGET_ILP32 instead of aarch64_trampoline_size.
12011         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12012         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12013         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12014         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12015         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12016         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12017         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12018         depending on the value of 'mode'.
12019         (add_losym_<mode>): New.
12020         (ldr_got_small_<mode>): New, based on ldr_got_small.
12021         (ldr_got_small): Remove.
12022         (ldr_got_small_sidi): New.
12023         * config/aarch64/iterators.md (P): New.
12024         (PTR): Change to 'ptr_mode' in the condition.
12026 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12028         * config.gcc (aarch64*-*-*): Support --with-abi.
12029         (aarch64*-*-elf): Support --with-multilib-list.
12030         (aarch64*-*-linux*): Likewise.
12031         (supported_defaults): Add abi to aarch64*-*-*.
12032         * configure.ac: Mention AArch64 for --with-multilib-list.
12033         * configure: Re-generated.
12034         * config/aarch64/biarchilp32.h: New file.
12035         * config/aarch64/biarchlp64.h: New file.
12036         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12037         (ABI_SPEC): Ditto.
12038         (MULTILIB_DEFAULTS): Ditto.
12039         (DRIVER_SELF_SPECS): Ditto.
12040         (ASM_SPEC): Update to also substitute -mabi.
12041         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12042         file whose name depends on -mabi= and -mbig-endian.
12043         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12044         TARGET_ILP32.
12045         (POINTER_SIZE): New define.
12046         (POINTERS_EXTEND_UNSIGNED): Ditto.
12047         (enum aarch64_abi_type): New enumeration tag.
12048         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12049         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12050         (TARGET_ILP32): New define.
12051         * config/aarch64/aarch64.opt (mabi): New.
12052         (aarch64_abi): New.
12053         (ilp32, lp64): New values for -mabi.
12054         * config/aarch64/t-aarch64 (comma): New define.
12055         (MULTILIB_OPTIONS): Ditto.
12056         (MULTILIB_DIRNAMES): Ditto.
12057         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12058         * doc/invoke.texi: Document -mabi for AArch64.
12060 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12062         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12064 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12065             Anton Blanchard  <anton@au1.ibm.com>
12067         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12068         endianness when selecting field to splat.
12070 2013-07-22  Eric Christopher  <echristo@gmail.com>
12072         * dwarf2out.c (die_odr_checksum): New function to use
12073         CHECKSUM_ macros and ULEB128 for DIE tag.
12074         (generate_type_signature): Use.
12076 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12078         * config.gcc (sparc*-*-*): Accept leon3 processor.
12079         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12080         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12081         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12082         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12083         (mfix-ut699): Adjust comment.
12084         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12085         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12086         (CPP_CPU_SPEC): Likewise.
12087         (ASM_CPU_SPEC): Likewise.
12088         * config/sparc/sparc.c (leon3_cost): New constant.
12089         (sparc_option_override): Add leon3 support.
12090         (mem_ref): New function.
12091         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12092         (sparc_do_work_around_errata): Look into the instruction in the delay
12093         slot and adjust accordingly.  Add fix for the data cache nullify issues
12094         of the UT699.  Change insertion position for the NOP.
12095         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12096         (leon3_load): New reservation.
12097         (leon_store): Bump latency to 2.
12098         (grfpu): New automaton.
12099         (grfpu_alu): New unit.
12100         (grfpu_ds): Likewise.
12101         (leon_fp_alu): Adjust.
12102         (leon_fp_mult): Delete.
12103         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12104         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12105         * config/sparc/sparc.md (cpu): Add leon3.
12106         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12107         (swapsi): Likewise.
12108         (atomic_test_and_set): Likewise.
12109         (ldstub): Likewise.
12111 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12113         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12114         default for R5900 targets.
12115         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12116         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12117         * config/mips/mips.c (mips_option_override): Report an error for
12118         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12119         for -march=r5900 -mhard-float.
12121 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12123         * df-problems.c (can_move_insns_across): Exit loop once we
12124         find a non-fixed, non-global register.
12126         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12127         setting can_throw.
12129         * omega.c (omega_eliminate_red): Break after setting red_found.
12130         (omega_problem_has_red_equations): Similarly after setting found.
12131         (omega_query_variable): Similarly after setting coupled.
12133 2013-07-22  Marek Polacek  <polacek@redhat.com>
12135         * gimplify.c: Don't include gimple.h twice.
12137 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12139         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12140         instead of TARGET_THUMB1.
12141         (Pz): New constraint.
12142         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12143         encodings.
12144         (compare_negsi_si): Likewise.
12145         (compare_addsi2_op0): Likewise.
12146         (compare_addsi2_op1): Likewise.
12147         (addsi3_carryin_<optab>): Likewise.
12148         (addsi3_carryin_alt2_<optab>): Likewise.
12149         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12150         for arm_restrict_it.
12151         (subsi3_carryin): Likewise.
12152         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12153         (minmax_arithsi): Disable for arm_restrict_it.
12154         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12155         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12156         (satsi_<SAT:code>_shift): Likewise.
12157         (arm_shiftsi3): Add alternative for 16-bit encoding.
12158         (arm32_movhf): Disable for arm_restrict_it.
12159         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12160         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12162 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12164         * config/arm/arm.md (attribute "insn"): Delete.
12165         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12166         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12167         (not_shiftsi): Update for attribute change.
12168         (not_shiftsi_compare0): Likewise.
12169         (not_shiftsi_compare0_scratch): Likewise.
12170         (arm_one_cmplsi2): Likewise.
12171         (thumb1_one_cmplsi2): Likewise.
12172         (notsi_compare0): Likewise.
12173         (notsi_compare0_scratch): Likewise.
12174         (thumb1_movdi_insn): Likewise.
12175         (arm_movsi_insn): Likewise.
12176         (movhi_insn_arch4): Likewise.
12177         (movhi_bytes): Likewise.
12178         (arm_movqi_insn): Likewise.
12179         (thumb1_movqi_insn): Likewise.
12180         (arm32_movhf): Likewise.
12181         (thumb1_movhf): Likewise.
12182         (arm_movsf_soft_insn): Likewise.
12183         (thumb1_movsf_insn): Likewise.
12184         (thumb_movdf_insn): Likewise.
12185         (movsicc_insn): Likewise.
12186         (movsfcc_soft_insn): Likewise.
12187         (and_scc): Likewise.
12188         (cond_move): Likewise.
12189         (if_move_not): Likewise.
12190         (if_not_move): Likewise.
12191         (if_shift_move): Likewise.
12192         (if_move_shift): Likewise.
12193         (if_shift_shift): Likewise.
12194         (if_not_arith): Likewise.
12195         (if_arith_not): Likewise.
12196         (cond_move_not): Likewise.
12197         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12198         (neon_mov<mode>): Likewise.
12199         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12200         (thumb2_movsi_vfp): Likewise.
12201         (movsf_vfp): Likewise.
12202         (thumb2_movsf_vfp): Likewise.
12203         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12204         change.
12205         (cortexa7_older_only): Likewise.
12206         (cortexa7_younger): Likewise.
12207         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12208         (1020alu_shift_op): Likewise.
12209         (1020alu_shift_reg_op): Likewise.
12210         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12211         (alu_shift_op): Likewise.
12212         (alu_shift_reg_op): Likewise.
12213         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12214         (11_alu_shift_op): Likewise.
12215         (11_alu_shift_reg_op): Likewise.
12216         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12217         (9_alu_shift_reg_op): Likewise.
12218         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12219         change.
12220         (cortex_a15_alu_shift): Likewise.
12221         (cortex_a15_alu_shift_reg): Likewise.
12222         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12223         (cortex_a5_alu_shift): Likewise.
12224         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12225         change.
12226         (cortex_a53_alu_shift): Likewise.
12227         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12228         change.
12229         (cortex_a7_alu_reg): Likewise.
12230         (cortex_a7_alu_shift): Likewise.
12231         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12232         (cortex_a8_alu_shift): Likewise.
12233         (cortex_a8_alu_shift_reg): Likewise.
12234         (cortex_a8_mov): Likewise.
12235         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12236         (cortex_a9_dp_shift): Likewise.
12237         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12238         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12239         (cortex_r4_mov): Likewise.
12240         (cortex_r4_alu_shift): Likewise.
12241         (cortex_r4_alu_shift_reg): Likewise.
12242         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12243         (526_alu_shift_op): Likewise.
12244         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12245         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12246         (626te_alu_shift_op): Likewise.
12247         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12248         (726te_alu_op): Likewise.
12249         (726te_alu_shift_op): Likewise.
12250         (726te_alu_shift_reg_op): Likewise.
12251         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12252         (mp626_alu_shift_op): Likewise.
12253         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12254         (pj4_alu_e1_conds): Likewise.
12255         (pj4_alu): Likewise.
12256         (pj4_alu_conds): Likewise.
12257         (pj4_shift): Likewise.
12258         (pj4_shift_conds): Likewise.
12259         (pj4_alu_shift): Likewise.
12260         (pj4_alu_shift_conds): Likewise.
12262 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12264         * config/arm/predicates.md (shiftable_operator_strict_it):
12265         New predicate.
12266         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12267         Disable cond_exec version for arm_restrict_it.
12268         (thumb2_smaxsi3): Convert to generate cond_exec.
12269         (thumb2_sminsi3): Likewise.
12270         (thumb32_umaxsi3): Likewise.
12271         (thumb2_uminsi3): Likewise.
12272         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12273         (thumb2_neg_abssi2): Likewise.
12274         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12275         (thumb2_movsicc_insn): Adjust alternatives.
12276         (thumb2_mov_negscc): Disable for arm_restrict_it.
12277         (thumb2_mov_negscc_strict_it): New pattern.
12278         (thumb2_mov_notscc_strict_it): New pattern.
12279         (thumb2_mov_notscc): Disable for arm_restrict_it.
12280         (thumb2_ior_scc): Likewise.
12281         (thumb2_ior_scc_strict_it): New pattern.
12282         (thumb2_cond_move): Adjust for arm_restrict_it.
12283         (thumb2_cond_arith): Disable for arm_restrict_it.
12284         (thumb2_cond_arith_strict_it): New pattern.
12285         (thumb2_cond_sub): Adjust for arm_restrict_it.
12286         (thumb2_movcond): Likewise.
12287         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12288         (thumb2_zero_extendhisi2_v6): Likewise.
12289         (thumb2_zero_extendqisi2_v6): Likewise.
12290         (orsi_notsi_si): Likewise.
12291         (orsi_not_shiftsi_si): Likewise.
12293 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12295         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12296         instruction sequence is 1 byte shorter.
12298 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12300         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12301         it is not needed after split.
12303 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12305         PR target/51784
12306         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12307         second label for nonlocal goto receivers. Don't output pic base labels
12308         unless we're producing PIC; mark that action unreachable().
12309         (ix86_save_reg): If the function contains a nonlocal label, save the
12310         PIC base reg.
12311         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12312         * config/darwin.c (emitted_pic_label_num): New GTY.
12313         (update_pic_label_number_if_needed): New.
12314         (machopic_output_function_base_name): Adjust for nonlocal receiver
12315         case.
12316         (machopic_should_output_picbase_label): New.
12317         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12318         (nonlocal_goto_receiver): New insn and split.
12320 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12322         * config/aarch64/aarch64-builtins.c
12323         (aarch64_fold_builtin): Fold abs in all modes.
12324         * config/aarch64/aarch64-simd-builtins.def
12325         (abs): Enable for all modes.
12326         * config/aarch64/arm_neon.h
12327         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12328         (vabs_f64): Add missing intrinsic.
12330 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12332         * config/aarch64/arm_neon.h (vabs_s64): New function
12334 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12336         PR target/57516
12337         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12338         * config/avr/avr.md (adjust_len): Add `round'.
12339         * config/avr/avr-protos.h (avr_out_round): New prototype.
12340         (avr_out_plus): Add `out_label' argument.
12341         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12342         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12343         Handle the case where `insn' is just a pattern.
12344         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12345         (avr_out_round): New function.
12346         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12348 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12350         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12351         Add check for flag_stack_usage to handle -fstack-usage support
12353 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12355         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12356         interaction for new Power8 flags and VSX.
12358 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12360         PR middle-end/57698
12361         * tree-inline.c (expand_call_inline): Emit errors during
12362         early_inlining only if optimization is not turned on.
12364 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12366         * passes.def: New.
12368         * passes.c (init_optimization_passes): Move the construction of
12369         the pass hierarchy into a new passes.def file.
12371         * Makefile.in (passes.o): Add dependency on passes.def.
12373 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12375         * passes.c (init_optimization_passes): Introduce macros for
12376         constructing the tree of passes (INSERT_PASSES_AFTER,
12377         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12378         TERMINATE_PASS_LIST).
12380 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12381             Wei Mi  <wmi@google.com>
12383         PR rtl-optimization/57878
12384         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12385         top.
12386         (reload_pseudo_compare_func): Check nregs first for reload
12387         pseudos.
12389 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12391         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12393 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12395         * read-rtl.c (validate_const_int): Once an invalid character is
12396         seen, quit the loop.
12398         * gengtype.c (write_roots): Similarly once we find the "deletable"
12399         or "if_marked" option.
12401 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12403         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12404         "xtab" and "sat".  Move value "clz" from here to ...
12405         (attriubte "type"): ... here.
12406         (satsi_<SAT:code>): Delete "insn" attribute.
12407         (satsi_<SAT:code>_shift): Likewise.
12408         (arm_zero_extendqisi2addsi): Likewise.
12409         (arm_extendqisi2addsi): Likewise.
12410         (clzsi2): Update for attribute changes.
12411         (rbitsi2): Likewise.
12412         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12413         attribute.
12414         (arm_usatsihi): Likewise.
12415         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12417 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12419         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12420         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12421         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12422         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12423         in alphabetical order.
12424         (attribute "core_cycles"): Update for attribute changes.
12425         (arm_addsi3): Likewise.
12426         (addsi3_compare0): Likewise.
12427         (addsi3_compare0_scratch): Likewise.
12428         (addsi3_compare_op1): Likewise.
12429         (addsi3_compare_op2): Likewise.
12430         (compare_addsi2_op0): Likewise.
12431         (compare_addsi2_op1): Likewise.
12432         (addsi3_carryin_shift_<optab>): Likewise.
12433         (subsi3_carryin_shift): Likewise.
12434         (rsbsi3_carryin_shift): Likewise.
12435         (arm_subsi3_insn): Likewise.
12436         (subsi3_compare0): Likewise.
12437         (subsi3_compare): Likewise.
12438         (arm_andsi3_insn): Likewise.
12439         (thumb1_andsi3_insn): Likewise.
12440         (andsi3_compare0): Likewise.
12441         (andsi3_compare0_scratch): Likewise.
12442         (zeroextractsi_compare0_scratch
12443         (andsi_not_shiftsi_si): Likewise.
12444         (iorsi3_insn): Likewise.
12445         (iorsi3_compare0): Likewise.
12446         (iorsi3_compare0_scratch): Likewise.
12447         (arm_xorsi3): Likewise.
12448         (thumb1_xorsi3_insn): Likewise.
12449         (xorsi3_compare0): Likewise.
12450         (xorsi3_compare0_scratch): Likewise.
12451         (satsi_<SAT:code>_shift): Likewise.
12452         (rrx): Likewise.
12453         (arm_shiftsi3): Likewise.
12454         (shiftsi3_compare0): Likewise.
12455         (not_shiftsi): Likewise.
12456         (not_shiftsi_compare0): Likewise.
12457         (not_shiftsi_compare0_scratch): Likewise.
12458         (arm_one_cmplsi2): Likewise.
12459         (thumb_one_complsi2): Likewise.
12460         (notsi_compare0): Likewise.
12461         (notsi_compare0_scratch): Likewise.
12462         (thumb1_zero_extendhisi2): Likewise.
12463         (arm_zero_extendhisi2): Likewise.
12464         (arm_zero_extendhisi2_v6): Likewise.
12465         (arm_zero_extendhisi2addsi): Likewise.
12466         (thumb1_zero_extendqisi2): Likewise.
12467         (thumb1_zero_extendqisi2_v6): Likewise.
12468         (arm_zero_extendqisi2): Likewise.
12469         (arm_zero_extendqisi2_v6): Likewise.
12470         (arm_zero_extendqisi2addsi): Likewise.
12471         (thumb1_extendhisi2): Likewise.
12472         (arm_extendhisi2): Likewise.
12473         (arm_extendhisi2_v6): Likewise.
12474         (arm_extendqisi): Likewise.
12475         (arm_extendqisi_v6): Likewise.
12476         (arm_extendqisi2addsi): Likewise.
12477         (thumb1_extendqisi2): Likewise.
12478         (thumb1_movdi_insn): Likewise.
12479         (arm_movsi_insn): Likewise.
12480         (movsi_compare0): Likewise.
12481         (movhi_insn_arch4): Likewise.
12482         (movhi_bytes): Likewise.
12483         (arm_movqi_insn): Likewise.
12484         (thumb1_movqi_insn): Likewise.
12485         (arm32_movhf): Likewise.
12486         (thumb1_movhf): Likewise.
12487         (arm_movsf_soft_insn): Likewise.
12488         (thumb1_movsf_insn): Likewise.
12489         (movdf_soft_insn): Likewise.
12490         (thumb_movdf_insn): Likewise.
12491         (arm_cmpsi_insn): Likewise.
12492         (cmpsi_shiftsi): Likewise.
12493         (cmpsi_shiftsi_swp): Likewise.
12494         (arm_cmpsi_negshiftsi_si): Likewise.
12495         (movsicc_insn): Likewise.
12496         (movsfcc_soft_insn): Likewise.
12497         (arith_shiftsi): Likewise.
12498         (arith_shiftsi_compare0
12499         (arith_shiftsi_compare0_scratch
12500         (sub_shiftsi): Likewise.
12501         (sub_shiftsi_compare0
12502         (sub_shiftsi_compare0_scratch
12503         (and_scc): Likewise.
12504         (cond_move): Likewise.
12505         (if_plus_move): Likewise.
12506         (if_move_plus): Likewise.
12507         (if_move_not): Likewise.
12508         (if_not_move): Likewise.
12509         (if_shift_move): Likewise.
12510         (if_move_shift): Likewise.
12511         (if_shift_shift): Likewise.
12512         (if_not_arith): Likewise.
12513         (if_arith_not): Likewise.
12514         (cond_move_not): Likewise.
12515         (thumb1_ashlsi3): Set type attribute.
12516         (thumb1_ashrsi3): Likewise.
12517         (thumb1_lshrsi3): Likewise.
12518         (thumb1_rotrsi3): Likewise.
12519         (shiftsi3_compare0_scratch): Likewise.
12520         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12521         (neon_mov<mode>): Likewise.
12522         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12523         attribute changes.
12524         (thumb2_movsi_insn): Likewise.
12525         (thumb2_cmpsi_neg_shiftsi): Likewise.
12526         (thumb2_extendqisi_v6): Likewise.
12527         (thumb2_zero_extendhisi2_v6): Likewise.
12528         (thumb2_zero_extendqisi2_v6): Likewise.
12529         (thumb2_shiftsi3_short): Likewise.
12530         (thumb2_addsi3_compare0_scratch): Likewise.
12531         (orsi_not_shiftsi_si): Likewise.
12532         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12533         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12534         changes.
12535         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12536         (1020alu_shift_op): Likewise.
12537         (1020alu_shift_reg_op): Likewise.
12538         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12539         (alu_shift_op): Likewise.
12540         (alu_shift_reg_op): Likewise.
12541         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12542         (11_alu_shift_op): Likewise.
12543         (11_alu_shift_reg_op): Likewise.
12544         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12545         (9_alu_shift_reg_op): Likewise.
12546         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12547         changes.
12548         (cortex_a15_alu_shift): Likewise.
12549         (cortex_a15_alu_shift_reg): Likewise.
12550         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12551         changes.
12552         (cortex_a5_alu_shift): Likewise.
12553         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12554         changes.
12555         (cortex_a53_alu_shift): Likewise.
12556         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12557         changes.
12558         (cortex_a7_alu_reg): Likewise.
12559         (cortex_a7_alu_shift): Likewise.
12560         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12561         changes.
12562         (cortex_a8_alu_shift): Likewise.
12563         (cortex_a8_alu_shift_reg): Likewise.
12564         (cortex_a8_mov): Likewise.
12565         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12566         (cortex_a9_dp_shift): Likewise.
12567         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12568         changes.
12569         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12570         changes.
12571         (cortex_r4_mov): Likewise.
12572         (cortex_r4_alu_shift): Likewise.
12573         (cortex_r4_alu_shift_reg): Likewise.
12574         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12575         (526_alu_shift_op): Likewise.
12576         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12577         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12578         (626te_alu_shift_op): Likewise.
12579         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12580         (726te_alu_op): Likewise.
12581         (726te_alu_shift_op): Likewise.
12582         (726te_alu_shift_reg_op): Likewise.
12583         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12584         (mp626_alu_shift_op): Likewise.
12585         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12586         (pj4_alu_e1_conds): Likewise.
12587         (pj4_alu): Likewise.
12588         (pj4_alu_conds): Likewise.
12589         (pj4_shift): Likewise.
12590         (pj4_shift_conds): Likewise.
12591         (pj4_alu_shift): Likewise.
12592         (pj4_alu_shift_conds): Likewise.
12593         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12594         changes.
12595         (cortexa7_older_only): Likewise.
12596         (cortexa7_younger): Likewise.
12598 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12600         * ipa-pure-const.c (generate_summary): Rename to...
12601         (pure_const_generate_summary): ... this.
12603 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12605         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12607 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12609         PR target/57909
12610         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12611         usage in HI mode.
12613 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12615         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12616         enabled without -march=zEC12.
12617         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12618         flags to be set.
12620 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12622         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12623         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12624         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12625         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12626         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12627         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12628         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12629         ISA_HAS_FP4.
12630         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12631         and ISA_HAS_NMADD3_NMSUB3.
12632         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12633         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12634         (nmsub4<mode>, nmsub3<mode>): Likewise.
12635         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12637 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12639         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12640         TARGET_MIPS5400 checking.
12642 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12643             Peter Bergner  <bergner@vnet.ibm.com>
12645         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12646         registers in the comment.
12647         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12648         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12649         rather than FIRST_PSEUDO_REGISTERS.
12651 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12653         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12654         enable extra ISA flags with TARGET_HTM.
12656 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12658         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12659         Fix comment typos.
12661 2013-07-15  Cong Hou  <congh@google.com>
12663         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12664         in compare function for sorting.
12666 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12668         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12669         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12670         * config/rs6000/rs6000.opt: Add -mhtm option.
12671         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12672         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12673         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12674         __HTM__ if the HTM instructions are available.
12675         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12676         htm_spr_reg_operand): New define_predicates.
12677         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12678         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12679         Include htm.md.
12680         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12681         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12682         HTM builtin functions.
12683         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12684         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12685         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12686         (rs6000_builtin_mask_calculate): Likewise.
12687         (rs6000_option_override_internal): Likewise.
12688         (bdesc_htm): Add new HTM builtin support.
12689         (htm_spr_num): New function.
12690         (htm_spr_regno): Likewise.
12691         (rs6000_htm_spr_icode): Likewise.
12692         (htm_expand_builtin): Likewise.
12693         (htm_init_builtins): Likewise.
12694         (rs6000_expand_builtin): Add support for HTM builtin functions.
12695         (rs6000_init_builtins): Likewise.
12696         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12697         option.
12698         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12699         (TARGET_HTM, MASK_HTM): Define macros.
12700         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12701         (FIXED_REGISTERS): Likewise.
12702         (CALL_USED_REGISTERS): Likewise.
12703         (CALL_REALLY_USED_REGISTERS): Likewise.
12704         (REG_ALLOC_ORDER): Likewise.
12705         (enum reg_class): Likewise.
12706         (REG_CLASS_NAMES): Likewise.
12707         (REG_CLASS_CONTENTS): Likewise.
12708         (REGISTER_NAMES): Likewise.
12709         (ADDITIONAL_REGISTER_NAMES): Likewise.
12710         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12711         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12712         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12713         * config/rs6000/htm.md: New file.
12714         * config/rs6000/htmintrin.h: New file.
12715         * config/rs6000/htmxlintrin.h: New file.
12717 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12719         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12720         Define SYMBOL_TINY_GOT, update comment.
12721         * config/aarch64/aarch64.c
12722         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12723         (aarch64_expand_mov_immediate): Likewise.
12724         (aarch64_print_operand): Likewise.
12725         (aarch64_classify_symbol): Likewise.
12726         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12727         (ldr_got_tiny): Define.
12729 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12731         PR tree-optimization/54094
12732         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12733           scheduling dimension for the parallelism check from the polyhedral
12734           information in the AST.
12735         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12736           in 2D + 1 form.
12738 2013-07-13  Jason Merrill  <jason@redhat.com>
12740         * print-tree.c (debug_vec_tree): Use debug_raw.
12741         (debug_raw (vec<tree, va_gc> &)): New.
12742         (debug_raw (vec<tree, va_gc> *)): New.
12743         * tree.h: Declare them.
12745 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12747         * ifcvt.c (ifcvt_after_combine): New static variable.
12748         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12749         for size.
12750         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12751         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12752         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12754 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12756         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12758 2013-07-12  Michael Matz  <matz@suse.de>
12760         PR middle-end/55771
12761         * convert.c (convert_to_real): Reject non-float inner types.
12763 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12765         * config/aarch64/aarch64-protos.h
12766         (aarch64_simd_immediate_valid_for_move): Remove.
12767         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12768         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12769         cases.
12770         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12772 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12774         * config/mips/mips.c (mips_conditional_register_usage): Do not
12775         use t[0-7] registers in MIPS16 mode when optimizing for size.
12777 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12779         * config/i386/i386.c (dispatch_function_versions): Fix array
12780         indexing of function_version_info to match actual_versions.
12782 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12784         * vec.h (struct va_gc): Move release out-of-line.
12785         (va_gc::release): Call ggc_free on released vec.
12787 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12789         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12790         Require GOT register as additional operand in UNSPEC.
12791         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12792         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12793         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12794         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12795         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12796         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12797         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12799 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12801         PR target/57631
12802         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12803         name seen by assembler/linker rather if available.
12805 2013-07-11  Andreas Schwab  <schwab@suse.de>
12807         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12809 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12811         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12813 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12815         * doc/tm.texi.in: Move hook documentation to ....
12816         * target.def: ... here.
12818         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12819         text on @hook line.
12820         * doc/tm.texi: Regenerate.
12822 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12824         PR c++/57869
12825         * doc/invoke.texi: Document -Wconditionally-supported.
12827 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12829         PR target/57844
12830         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12831         of my_fp.
12833 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12835         PR target/57506
12836         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12837         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12838         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12839         Remove duplicate devices.
12840         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12841         * config/avr/t-multilib: Regenerate.
12842         * config/avr/avr-tables.opt: Regenerate.
12843         * doc/avr-mmcu.texi: Regenerate.
12845 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12847         PR target/56987
12848         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12850 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12852         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12853         the cost of MULT when optimizing for size.
12855 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12857         * config/cr16/cr16-protos.h: Don't include target.h.
12859 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12861         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12862         adjust register size for TDmode and TFmode for VSX registers.
12864 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12866         PR target/56892
12867         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12868         hook_bool_const_tree_true.
12870 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12872         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12873         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12874         * config/s390/s390.md: Define FPR*_REGNUM constants.
12875         Fix FPR2_REGNUM constant (18 -> 17).
12876         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12877         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12878         ("trunc<BFP:mode><DFP_ALL:mode>2")
12879         ("trunc<DFP_ALL:mode><BFP:mode>2")
12880         ("*extend<BFP:mode><DFP_ALL:mode>2")
12881         ("*extend<DFP_ALL:mode><BFP:mode>2")
12882         ("extend<BFP:mode><DFP_ALL:mode>2")
12883         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12884         FPR4_REGNUM.
12886 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12888         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12890 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12892         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12893         and cfun_fpr_bit_p to cfun_fpr_save_p.
12894         (s390_frame_area, s390_register_info, s390_frame_info)
12895         (s390_emit_prologue, s390_emit_epilogue)
12896         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12897         register numbers.
12898         * config/s390/s390.h: Define *_REGNUM macros for floating point
12899         register numbers.
12901 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12903         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12905 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12907         PR rtl-optimization/57786
12908         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12909         and break out of the loop when it is set to false.
12911 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12913         PR target/57819
12914         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12915         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12916         (const_int 63)) 0)).
12917         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12918         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12919         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12921         PR rtl-optimization/57829
12922         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12923         mask bits outside of mode are just sign-extension from mode to HWI.
12925 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12927         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12928         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12929         adjust_address instead of change_address to keep info about alignment.
12930         (emit_strmov): Remove.
12931         (emit_memmov): New function.
12932         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12933         (expand_movmem_epilogue): Likewise and return updated rtx for
12934         destination.
12935         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12936         destination and source.
12937         (decide_alignment): Refactor, handle vector_loop.
12938         (ix86_expand_movmem): Likewise.
12939         (ix86_expand_setmem): Likewise.
12940         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12942 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
12944         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
12945         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
12947 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12949         * config/i386/sse.md (sse_movlhps): Change alternative 3
12950         of operand 2 to "m".
12952 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12954         PR target/57807
12955         * config/i386/sse.md (iptr): New mode attribute.
12956         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
12957         (<sse>_vm<plusminus_insn><mode>3): Ditto.
12958         (<sse>_vmmul<mode>3): Ditto.
12959         (<sse>_vmdiv<mode>3): Ditto.
12960         (sse_vmrcpv4sf2): Ditto.
12961         (<sse>_vmsqrt<mode>2): Ditto.
12962         (sse_vmrsqrtv4sf2): Ditto.
12963         (<sse>_vm<code><mode>3): Ditto.
12964         (avx_vmcmp<mode>3): Ditto.
12965         (<sse>_vmmaskcmp<mode>3): Ditto.
12966         (<sse>_comi): Ditto.
12967         (<sse>_ucomi): Ditto.
12968         (*xop_vmfrcz_<mode>): Ditto.
12969         (*fmai_fmadd_<mode>): Ditto.
12970         (*fmai_fmsub_<mode>): Ditto.
12971         (*fmai_fnmadd_<mode>): Ditto.
12972         (*fmai_fnmsub_<mode>): Ditto.
12973         (*fma4i_vmfmadd_<mode>): Ditto.
12974         (*fma4i_vmfmsub_<mode>): Ditto.
12975         (*fma4i_vmfnmadd_<mode>): Ditto.
12976         (*fma4i_vmfnmsub_<mode>): Ditto.
12977         (*xop_vmfrcz_<mode>): Ditto.
12978         (sse_cvtps2pi): Ditto.
12979         (sse_cvttps2pi): Ditto.
12980         (sse_cvtss2si): Ditto.
12981         (sse_cvtss2si_2): Ditto.
12982         (sse_cvtss2siq_2): Ditto.
12983         (sse_cvttss2si): Ditto.
12984         (sse_cvttss2siq): Ditto.
12985         (sse_cvtsd2si): Ditto.
12986         (sse_cvtsd2si_2): Ditto.
12987         (sse_cvtsd2siq_2): Ditto.
12988         (sse_cvttsd2si): Ditto.
12989         (sse_cvttsd2siq): Ditto.
12990         (sse_cvtsd2ss): Ditto.
12991         (sse_cvtss2sd): Ditto.
12992         (avx2_pbroadcast<mode>): Ditto.
12993         (avx2_pbroadcast<mode>_1): Ditto.
12994         (*avx_vperm_broadcast_v4sf): Ditto.
12996         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
12997         (sse_movlhps): Ditto.
12998         (sse_storehps): Ditto.
12999         (sse_loadhps): Ditto.
13000         (sse_storelps): Ditto.
13001         (sse_loadlps): Ditto.
13002         (*vec_concatv4sf): Ditto.
13003         (*vec_interleave_highv2df): Ditto.
13004         (*vec_interleave_lowv2df): Ditto.
13005         (*vec_extractv2df_1_sse): Ditto.
13006         (*vec_extractv2df_0_sse): Ditto.
13007         (sse2_storelpd): Ditto.
13008         (sse2_loadlpd): Ditto.
13009         (sse2_movsd): Ditto.
13010         (*vec_concatv4si): Ditto.
13011         (vec_concatv2di): Ditto.
13013         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13014         for Intel asm dialect.
13015         (mmx_punpcklwd): Ditto.
13016         (mmx_punpckldq): Ditto.
13018         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13019         for intel assembler dialect.
13021 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13023         PR target/29776
13024         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13025         for BUILT_IN_C{LZ,LRSB}*.
13026         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13027         * tree-vrp.c (extract_range_basic): Handle
13028         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13029         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13030         fall thru to code calling set_value*.
13031         * builtins.c (expand_builtin): Remove *IMAX cases.
13032         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13033         if width is bigger than 2*HWI.
13035 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13037         PR rtl-optimization/55342
13038         * lra-int.h (lra_subreg_reload_pseudos): New.
13039         * lra.c: Add undoing optional reloads to the block diagram.
13040         (lra_subreg_reload_pseudos): New.
13041         (lra_optional_reload_pseudos): Change comments.
13042         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13043         lra_optional_reload_pseudos after undo transformations.
13044         * lra-assigns.c (pseudo_prefix_title): New.
13045         (lra_setup_reg_renumber): Use it.
13046         (spill_for): Ditto.  Check subreg reload pseudos too.
13047         (assign_by_spills): Consider subreg reload pseudos too.
13048         * lra-constraints.c (simplify_operand_subreg): Use
13049         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13050         (curr_insn_transform): Recognize and do optional reloads.
13051         (undo_optional_reloads): New.
13052         (lra_undo_inheritance): Call undo_optional_reloads.
13054 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13056         * tree-complex.c (expand_complex_operations_1): Fix typo.
13058 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13060         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13061         (tune_params): New member 'const vec_costs'.
13062         * config/aarch64/aarch64.c (generic_vector_cost): New.
13063         (generic_tunings): New member 'generic_vector_cost'.
13064         (aarch64_builtin_vectorization_cost): New.
13065         (aarch64_add_stmt_cost): New.
13066         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13067         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13069 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13071         PR target/57777
13072         * config/i386/predicates.md (vsib_address_operand): Disallow
13073         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13075 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13077         PR middle-end/55030
13078         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13079         expand_builtin_setjmp_receiver.
13080         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13081         with NULL for the label parameter.
13082         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13083         the frame-pointer.  Adjust comments.
13084         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13085         only if LABEL is non-NULL.
13087 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13089         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13090         (ARM_ABI_AAPCS64): Ditto.
13091         (arm_abi): Ditto.
13092         (ARM_DEFAULT_ABI): Ditto.
13094 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13096         * config/aarch64/aarch64-builtins.c
13097         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13098         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13099         (st1): Likewise.
13100         * config/aarch64/aarch64-simd.md
13101         (aarch64_ld1<VALL:mode>): New.
13102         (aarch64_st1<VALL:mode>): Likewise.
13103         * config/aarch64/arm_neon.h
13104         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13106 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13108         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13109         target ISA is AVX.
13110         (ix86_option_override_internal):Turn on all -mavx target flags by
13111         default as they are dependent on AVX anyway.
13113 2013-07-02  Cary Coutant  <ccoutant@google.com>
13115         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13116         deterministic hash.
13117         (loc_checksum_ordered): Likewise.
13118         (hash_loc_operands): Remove inline keyword.
13120 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13122         PR tree-optimization/57741
13123         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13124         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13125         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13126         Allow REAL_CST step_exprs if flag_associative_math.
13127         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13129 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13131         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13133 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13135         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13137 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13139         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13140         encoding.
13141         (iorsi3_insn): Likewise.
13142         (arm_xorsi3): Likewise.
13144 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13146         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13147         to ...
13148         (attribute "type"): ... here, and prefix with "wmmx_".
13149         (attribute "core_cycles"): Update for attribute changes.
13150         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13151         (tbcstv4hi): Likewise.
13152         (tbcstv2si): Likewise.
13153         (iwmmxt_iordi3): Likewise.
13154         (iwmmxt_xordi3): Likewise.
13155         (iwmmxt_anddi3): Likewise.
13156         (iwmmxt_nanddi3): Likewise.
13157         (iwmmxt_arm_movdi): Likewise.
13158         (iwmmxt_movsi_insn): Likewise.
13159         (mov<mode>_internal): Likewise.
13160         (and<mode>3_iwmmxt): Likewise.
13161         (ior<mode>3_iwmmxt): Likewise.
13162         (xor<mode>3_iwmmxt): Likewise.
13163         (add<mode>3_iwmmxt): Likewise.
13164         (ssaddv8qi3): Likewise.
13165         (ssaddv4hi3): Likewise.
13166         (ssaddv2si3): Likewise.
13167         (usaddv8qi3): Likewise.
13168         (usaddv4hi3): Likewise.
13169         (usaddv2si3): Likewise.
13170         (sub<mode>3_iwmmxt): Likewise.
13171         (sssubv8qi3): Likewise.
13172         (sssubv4hi3): Likewise.
13173         (sssubv2si3): Likewise.
13174         (ussubv8qi3): Likewise.
13175         (ussubv4hi3): Likewise.
13176         (ussubv2si3): Likewise.
13177         (mulv4hi3_iwmmxt): Likewise.
13178         (smulv4hi3_highpart): Likewise.
13179         (umulv4hi3_highpart): Likewise.
13180         (iwmmxt_wmacs): Likewise.
13181         (iwmmxt_wmacsz): Likewise.
13182         (iwmmxt_wmacu): Likewise.
13183         (iwmmxt_wmacuz): Likewise.
13184         (iwmmxt_clrdi): Likewise.
13185         (iwmmxt_clrv8qi): Likewise.
13186         (iwmmxt_clr4hi): Likewise.
13187         (iwmmxt_clr2si): Likewise.
13188         (iwmmxt_uavgrndv8qi3): Likewise.
13189         (iwmmxt_uavgrndv4hi3): Likewise.
13190         (iwmmxt_uavgv8qi3): Likewise.
13191         (iwmmxt_uavgv4hi3): Likewise.
13192         (iwmmxt_tinsrb): Likewise.
13193         (iwmmxt_tinsrh): Likewise.
13194         (iwmmxt_tinsrw): Likewise.
13195         (iwmmxt_textrmub): Likewise.
13196         (iwmmxt_textrmsb): Likewise.
13197         (iwmmxt_textrmuh): Likewise.
13198         (iwmmxt_textrmsh): Likewise.
13199         (iwmmxt_textrmw): Likewise.
13200         (iwmxxt_wshufh): Likewise.
13201         (eqv8qi3): Likewise.
13202         (eqv4hi3): Likewise.
13203         (eqv2si3): Likewise.
13204         (gtuv8qi3): Likewise.
13205         (gtuv4hi3): Likewise.
13206         (gtuv2si3): Likewise.
13207         (gtv8qi3): Likewise.
13208         (gtv4hi3): Likewise.
13209         (gtv2si3): Likewise.
13210         (smax<mode>3_iwmmxt): Likewise.
13211         (umax<mode>3_iwmmxt): Likewise.
13212         (smin<mode>3_iwmmxt): Likewise.
13213         (umin<mode>3_iwmmxt): Likewise.
13214         (iwmmxt_wpackhss): Likewise.
13215         (iwmmxt_wpackwss): Likewise.
13216         (iwmmxt_wpackdss): Likewise.
13217         (iwmmxt_wpackhus): Likewise.
13218         (iwmmxt_wpackwus): Likewise.
13219         (iwmmxt_wpackdus): Likewise.
13220         (iwmmxt_wunpckihb): Likewise.
13221         (iwmmxt_wunpckihh): Likewise.
13222         (iwmmxt_wunpckihw): Likewise.
13223         (iwmmxt_wunpckilb): Likewise.
13224         (iwmmxt_wunpckilh): Likewise.
13225         (iwmmxt_wunpckilw): Likewise.
13226         (iwmmxt_wunpckehub): Likewise.
13227         (iwmmxt_wunpckehuh): Likewise.
13228         (iwmmxt_wunpckehuw): Likewise.
13229         (iwmmxt_wunpckehsb): Likewise.
13230         (iwmmxt_wunpckehsh): Likewise.
13231         (iwmmxt_wunpckehsw): Likewise.
13232         (iwmmxt_wunpckelub): Likewise.
13233         (iwmmxt_wunpckeluh): Likewise.
13234         (iwmmxt_wunpckeluw): Likewise.
13235         (iwmmxt_wunpckelsb): Likewise.
13236         (iwmmxt_wunpckelsh): Likewise.
13237         (iwmmxt_wunpckelsw): Likewise.
13238         (ror<mode>3): Likewise.
13239         (ashr<mode>3_iwmmxt): Likewise.
13240         (lshr<mode>3_iwmmxt): Likewise.
13241         (ashl<mode>3_iwmmxt): Likewise.
13242         (ror<mode>3_di): Likewise.
13243         (ashr<mode>3_di): Likewise.
13244         (lshr<mode>3_di): Likewise.
13245         (ashl<mode>3_di): Likewise.
13246         (iwmmxt_wmadds): Likewise.
13247         (iwmmxt_wmaddu): Likewise.
13248         (iwmmxt_tmia): Likewise.
13249         (iwmmxt_tmiaph): Likewise.
13250         (iwmmxt_tmiabb): Likewise.
13251         (iwmmxt_tmiatb): Likewise.
13252         (iwmmxt_tmiabt): Likewise.
13253         (iwmmxt_tmiatt): Likewise.
13254         (iwmmxt_tmovmskb): Likewise.
13255         (iwmmxt_tmovmskh): Likewise.
13256         (iwmmxt_tmovmskw): Likewise.
13257         (iwmmxt_waccb): Likewise.
13258         (iwmmxt_wacch): Likewise.
13259         (iwmmxt_waccw): Likewise.
13260         (iwmmxt_waligni): Likewise.
13261         (iwmmxt_walignr): Likewise.
13262         (iwmmxt_walignr0): Likewise.
13263         (iwmmxt_walignr1): Likewise.
13264         (iwmmxt_walignr2): Likewise.
13265         (iwmmxt_walignr3): Likewise.
13266         (iwmmxt_wsadb): Likewise.
13267         (iwmmxt_wsadh): Likewise.
13268         (iwmmxt_wsadbz): Likewise.
13269         (iwmmxt_wsadhz): Likewise.
13270         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13271         (iwmmxt_wabsdiffb): Likewise.
13272         (iwmmxt_wabsdiffh): Likewise.
13273         (iwmmxt_wabsdiffw): Likewise.
13274         (iwmmxt_waddsubhx): Likewise
13275         (iwmmxt_wsubaddhx): Likewise.
13276         (addc<mode>3): Likewise.
13277         (iwmmxt_avg4): Likewise.
13278         (iwmmxt_avg4r): Likewise.
13279         (iwmmxt_wmaddsx): Likewise.
13280         (iwmmxt_wmaddux): Likewise.
13281         (iwmmxt_wmaddsn): Likewise.
13282         (iwmmxt_wmaddun): Likewise.
13283         (iwmmxt_wmulwsm): Likewise.
13284         (iwmmxt_wmulwum): Likewise.
13285         (iwmmxt_wmulsmr): Likewise.
13286         (iwmmxt_wmulumr): Likewise.
13287         (iwmmxt_wmulwsmr): Likewise.
13288         (iwmmxt_wmulwumr): Likewise.
13289         (iwmmxt_wmulwl): Likewise.
13290         (iwmmxt_wqmulm): Likewise.
13291         (iwmmxt_wqmulwm): Likewise.
13292         (iwmmxt_wqmulmr): Likewise.
13293         (iwmmxt_wqmulwmr): Likewise.
13294         (iwmmxt_waddbhusm): Likewise.
13295         (iwmmxt_waddbhusl): Likewise.
13296         (iwmmxt_wqmiabb): Likewise.
13297         (iwmmxt_wqmiabt): Likewise.
13298         (iwmmxt_wqmiatb): Likewise.
13299         (iwmmxt_wqmiatt): Likewise.
13300         (iwmmxt_wqmiabbn): Likewise.
13301         (iwmmxt_wqmiabtn): Likewise.
13302         (iwmmxt_wqmiatbn): Likewise.
13303         (iwmmxt_wqmiattn): Likewise.
13304         (iwmmxt_wmiabb): Likewise.
13305         (iwmmxt_wmiabt): Likewise.
13306         (iwmmxt_wmiatb): Likewise.
13307         (iwmmxt_wmiatt): Likewise.
13308         (iwmmxt_wmiabbn): Likewise.
13309         (iwmmxt_wmiabtn): Likewise.
13310         (iwmmxt_wmiatbn): Likewise.
13311         (iwmmxt_wmiattn): Likewise.
13312         (iwmmxt_wmiawbb): Likewise.
13313         (iwmmxt_wmiawbt): Likewise.
13314         (iwmmxt_wmiawtb): Likewise.
13315         (iwmmxt_wmiawtt): Likewise.
13316         (iwmmxt_wmiawbbn): Likewise.
13317         (iwmmxt_wmiawbtn): Likewise.
13318         (iwmmxt_wmiawtbn): Likewise.
13319         (iwmmxt_wmiawttn): Likewise.
13320         (iwmmxt_wmerge): Likewise.
13321         (iwmmxt_tandc<mode>3): Likewise.
13322         (iwmmxt_torc<mode>3): Likewise.
13323         (iwmmxt_torvsc<mode>3): Likewise.
13324         (iwmmxt_textrc<mode>3): Likewise.
13325         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13326         (wmmxt_pack): Likewise.
13327         (wmmxt_mult_c1): Likewise.
13328         (wmmxt_mult_c2): Likewise.
13329         (wmmxt_alu_c1): Likewise.
13330         (wmmxt_alu_c2): Likewise.
13331         (wmmxt_alu_c3): Likewise.
13332         (wmmxt_transfer_c1): Likewise.
13333         (wmmxt_transfer_c2): Likewise.
13334         (wmmxt_transfer_c3): Likewise.
13335         (marvell_f_iwmmxt_wstr): Likewise.
13336         (marvell_f_iwmmxt_wldr): Likewise.
13338 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13340         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13342 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13344         Revert:
13345         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13346         * lra-constraints.c (need_for_split_p): Check call used hard regs
13347         living through calls.
13349         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13350         call used regs for call insn.
13352 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13354         PR target/57736
13355         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13356         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13358 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13360         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13362 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13364         * lra-constraints.c (need_for_split_p): Check call used hard regs
13365         living through calls.
13367 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13369         PR target/57744
13370         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13371         to tie with any other modes.  Eliminate Altivec vector mode tests,
13372         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13373         code, to return 0 if testing MODE2 for a condition, if we've
13374         already tested MODE1 for the same condition.
13376 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13378         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13379         layout.
13381 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13383         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13384         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13386 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13388         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13389         Define.
13390         (aarch64_symbolic_constant_p): Remove.
13391         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13392         static.  Fix line length and white space.
13393         (aarch64_symbolic_constant_p): Remove.
13394         * config/aarch64/predicates.md (aarch64_valid_symref):
13395         Use aarch64_classify_symbol_expression.
13397 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13399         * config/arm/constraints.md (Ts): New constraint.
13400         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13401         16-bit encodings.
13402         (compare_scc): Use "Ts" constraint for operand 0.
13403         (ior_scc_scc): Likewise.
13404         (and_scc_scc): Likewise.
13405         (and_scc_scc_nodom): Likewise.
13406         (ior_scc_scc_cmp): Likewise for operand 7.
13407         (and_scc_scc_cmp): Likewise.
13408         * config/arm/thumb2.md (thumb2_movsi_insn):
13409         Add alternatives for 16-bit encodings.
13410         (thumb2_movhi_insn): Likewise.
13411         (thumb2_movsicc_insn): Likewise.
13412         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13413         (thumb2_negscc): Use "Ts" constraint.
13414         Move mvn instruction outside cond_exec block.
13415         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13416         for 16-bit encodings.
13418 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13420         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13421         encoding.
13422         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13423         (mulsi3subsi): Likewise.
13424         (mulsidi3adddi): Likewise.
13425         (mulsidi3_v6): Likewise.
13426         (umulsidi3_v6): Likewise.
13427         (umulsidi3adddi_v6): Likewise.
13428         (smulsi3_highpart_v6): Likewise.
13429         (umulsi3_highpart_v6): Likewise.
13430         (mulhisi3tb): Likewise.
13431         (mulhisi3bt): Likewise.
13432         (mulhisi3tt): Likewise.
13433         (maddhisi4): Likewise.
13434         (maddhisi4tb): Likewise.
13435         (maddhisi4tt): Likewise.
13436         (maddhidi4): Likewise.
13437         (maddhidi4tb): Likewise.
13438         (maddhidi4tt): Likewise.
13439         (zeroextractsi_compare0_scratch): Likewise.
13440         (insv_zero): Likewise.
13441         (insv_t2): Likewise.
13442         (anddi_notzesidi_di): Likewise.
13443         (anddi_notsesidi_di): Likewise.
13444         (andsi_notsi_si): Likewise.
13445         (iordi_zesidi_di): Likewise.
13446         (xordi_zesidi_di): Likewise.
13447         (andsi_iorsi3_notsi): Likewise.
13448         (smax_0): Likewise.
13449         (smax_m1): Likewise.
13450         (smin_0): Likewise.
13451         (not_shiftsi): Likewise.
13452         (unaligned_loadsi): Likewise.
13453         (unaligned_loadhis): Likewise.
13454         (unaligned_loadhiu): Likewise.
13455         (unaligned_storesi): Likewise.
13456         (unaligned_storehi): Likewise.
13457         (extv_reg): Likewise.
13458         (extzv_t2): Likewise.
13459         (divsi3): Likewise.
13460         (udivsi3): Likewise.
13461         (arm_zero_extendhisi2addsi): Likewise.
13462         (arm_zero_extendqisi2addsi): Likewise.
13463         (compareqi_eq0): Likewise.
13464         (arm_extendhisi2_v6): Likewise.
13465         (arm_extendqisi2addsi): Likewise.
13466         (arm_movt): Likewise.
13467         (thumb2_ldrd): Likewise.
13468         (thumb2_ldrd_base): Likewise.
13469         (thumb2_ldrd_base_neg): Likewise.
13470         (thumb2_strd): Likewise.
13471         (thumb2_strd_base): Likewise.
13472         (thumb2_strd_base_neg): Likewise.
13473         (arm_negsi2): Add alternative for 16-bit encoding.
13474         (arm_one_cmplsi2): Likewise.
13476 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13478         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13479         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13480         (movdfcc): Likewise.
13481         * config/arm/vfp.md (*thumb2_movsf_vfp):
13482         Disable predication for arm_restrict_it.
13483         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13484         (*thumb2_movdfcc_vfp): Likewise.
13485         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13486         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13487         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13488         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13489         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13490         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13491         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13492         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13493         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13494         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13495         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13496         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13497         Disable predication for arm_restrict_it.
13499 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13501         * config/i386/bmiintrin.h (_bextr_u32): New.
13502         (_bextr_u64): Ditto.
13504 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13506         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13507         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13508         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13509         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13510         * config/mips/n32-elf.h: ...this new file.
13512 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13514         PR target/57224
13515         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13516         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13518 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13520         * config/mips/mips-tables.opt: Regenerate.
13521         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13522         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13523         * doc/invoke.texi: Add -m14kc.
13525 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13527         PR target/57623
13528         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13529         constraints of operand 1 and 2.
13531         PR target/57623
13532         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13533         to match RTL canonicalization.  Swap predicates and
13534         constraints of operand 1 and 2.
13536 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13538         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13539         Process OP_INOUT regs for splitting too.
13541 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13543         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13544         decl before the loop, initialize to NULL.
13545         (vectorizable_load): Initialize ptr_incr to NULL.
13547 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13549         PR lto/57208
13550         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13551         * ipa-ref.c (ipa_maybe_record_reference): New function.
13552         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13553         * ipa-cp.c (create_specialized_node): Record potential references from
13554         aggvals.
13555         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13557 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13559         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13560         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13561         'mode' to force_reg.
13562         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13563         (aarch64_expand_mov_immediate): Likewise.
13565 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13567         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13568         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13570 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13572         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13573         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13574         (struct machine_function): Add tbegin_p.
13575         (s390_canonicalize_comparison): Fold CC mode compares to
13576         conditional jump if possible.
13577         (s390_emit_jump): Return the emitted jump.
13578         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13579         Handle CCRAWmode compares.
13580         (s390_option_override): Default to -mhtm if available.
13581         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13582         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13583         FPRs instead of df_regs_ever_live_p.
13584         (s390_optimize_nonescaping_tx): New function.
13585         (s390_init_frame_layout): Extend clobbered_regs array to cover
13586         FPRs as well.
13587         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13588         (s390_expand_tbegin): New function.
13589         (enum s390_builtin): New enum definition.
13590         (code_for_builtin): New array definition.
13591         (s390_init_builtins): New function.
13592         (s390_expand_builtin): New function.
13593         (TARGET_INIT_BUILTINS): Define.
13594         (TARGET_EXPAND_BUILTIN): Define.
13595         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13596         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13597         (s390_alc_comparison): Likewise.
13598         * config/s390/s390-modes.def: Add CCRAWmode.
13599         * config/s390/s390.h (processor_flags): Add PF_TX.
13600         (TARGET_CPU_HTM): Define macro.
13601         (TARGET_HTM): Define macro.
13602         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13603         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13604         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13605         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13606         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13607         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13608         ("*cc_to_int"): Move up.
13609         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13610         constants other than 0.
13611         ("*ccraw_to_int"): New insn and splitter definition.
13612         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13613         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13614         ("tx_assist"): New expander.
13615         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13616         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13617         * config/s390/s390.opt: Add -mhtm option.
13618         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13619         * config/s390/htmxlintrin.h: New file.
13620         * config/s390/htmintrin.h: New file.
13621         * config/s390/s390intrin.h: New file.
13622         * doc/extend.texi: Document htm builtins.
13623         * config.gcc: Add the new header files to extra_headers.
13625 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13627         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13628         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13630 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13631             Pat Haugen  <pthaugen@us.ibm.com>
13632             Peter Bergner  <bergner@vnet.ibm.com>
13634         * config/rs6000/power8.md: New.
13635         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13636         setting for power8 entry.
13637         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13638         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13639         test for Power4/Power5 only.
13640         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13641         support.
13642         (force_new_group): Adjust comment.
13643         * config/rs6000/rs6000.md: Include power8.md.
13645 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13647         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13648         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13649         declaration.
13650         (tune_params): Update comment.
13651         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13652         (arm_max_conditional_execute): New function.
13653         (thumb2_final_prescan_insn): Use max_insn_skipped and
13654         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13656 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13658         PR tree-optimization/57705
13659         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13660         SSA_NAME step, provided that it is not defined inside the loop.
13661         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13662         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13664 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13666         PR middle-end/57670
13667         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13668         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13669         calls in the dump.
13670         (ipa_note_param_call): Initialize member_ptr flag.
13671         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13672         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13673         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13674         (ipa_read_indirect_edge_info): Likewise.
13676 2013-06-25  Richard Biener  <rguenther@suse.de>
13678         PR middle-end/56977
13679         * passes.c (init_optimization_passes): Move pass_fold_builtins
13680         and pass_dce earlier with -Og.
13682 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13684         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13685         <BIT_FIELD_REF>: Remove trailing TAB.
13686         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13687         remove blank line.
13689 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13691         PR tree-optimization/57358
13692         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13693         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13694         (ipa_analyze_params_uses): Generate pessimistic info when true.
13696 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13698         PR tree-optimization/57539
13699         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13700         global.inlined_to of the new node to it.  All callers changed.
13701         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13702         inlining_into, pass it to cgraph_clone_node.
13703         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13704         ipa_free_edge_args_substructures.
13705         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13706         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13707         Assert we have found an rdesc in the rdesc list.
13709 2013-06-24  Richard Biener  <rguenther@suse.de>
13711         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13712         (pointer_set_lookup): Declare.
13713         (class pointer_map): New template class implementing a
13714         generic pointer to T map.
13715         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13716         pointer_map<T>::contains, pointer_map<T>::insert,
13717         pointer_map<T>::traverse): New functions.
13718         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13719         (pointer_set_lookup): New function.
13720         (pointer_set_contains): Use pointer_set_lookup.
13721         (pointer_set_insert): Likewise.
13722         (insert_aux): Remove.
13723         (struct pointer_map_t): Embed a pointer_set_t.
13724         (pointer_map_create): Adjust.
13725         (pointer_map_destroy): Likewise.
13726         (pointer_map_contains): Likewise.
13727         (pointer_map_insert): Likewise.
13728         (pointer_map_traverse): Likewise.
13729         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13730         pointer_map<unsigned> instead of a pointer_map_t.
13731         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13732         (streamer_tree_cache_lookup): Likewise.
13733         (streamer_tree_cache_create): Likewise.
13734         (streamer_tree_cache_delete): Likewise.
13735         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13736         pointer_map<unsigned> instead of a pointer_map_t.
13737         (lto_init_tree_ref_encoder): Adjust.
13738         (lto_destroy_tree_ref_encoder): Likewise.
13739         * lto-section-out.c (lto_output_decl_index): Likewise.
13740         (lto_record_function_out_decl_state): Likewise.
13741         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13743 2013-06-24  Richard Biener  <rguenther@suse.de>
13745         PR tree-optimization/57488
13746         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13748 2013-06-24  Alan Modra  <amodra@gmail.com>
13750         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13751         (gen_easy_altivec_constant): Likewise.
13752         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13753         easy_vector_constant_msb): Likewise.
13755 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13757         PR target/57688
13758         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13759         add missing return true.
13761 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13763         PR target/52483
13764         * config/sh/predicates.md (general_extend_operand): Invoke
13765         general_movsrc_operand for memory operands.
13766         (general_movsrc_operand): Allow reg+reg addressing, do not use
13767         general_operand for memory operands.
13769 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13771         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13772         when current target options does not apply.
13773         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13774         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13775         * config/i386/bmiintrin.h: Pass appropriate target
13776         attributes to header.
13777         * config/i386/mmintrin.h: Ditto.
13778         * config/i386/nmmintrin.h: Ditto.
13779         * config/i386/avx2intrin.h: Ditto.
13780         * config/i386/fxsrintrin.h: Ditto.
13781         * config/i386/tbmintrin.h: Ditto.
13782         * config/i386/xsaveintrin.h: Ditto.
13783         * config/i386/f16cintrin.h: Ditto.
13784         * config/i386/xtestintrin.h: Ditto.
13785         * config/i386/xsaveoptintrin.h: Ditto.
13786         * config/i386/bmi2intrin.h: Ditto.
13787         * config/i386/lzcntintrin.h: Ditto.
13788         * config/i386/smmintrin.h: Ditto.
13789         * config/i386/wmmintrin.h: Ditto.
13790         * config/i386/x86intrin.h: Remove all header include guards.
13791         * config/i386/prfchwintrin.h: Ditto.
13792         * config/i386/pmmintrin.h: Ditto.
13793         * config/i386/tmmintrin.h: Ditto.
13794         * config/i386/xmmintrin.h: Ditto.
13795         * config/i386/popcntintrin.h: Ditto.
13796         * config/i386/rdseedintrin.h: Ditto.
13797         * config/i386/ammintrin.h: Ditto.
13798         * config/i386/emmintrin.h: Ditto.
13799         * config/i386/immintrin.h: Remove all header include guards.
13800         * config/i386/fma4intrin.h: Ditto.
13801         * config/i386/lwpintrin.h: Ditto.
13802         * config/i386/xopintrin.h: Ditto.
13803         * config/i386/ia32intrin.h: Ditto.
13804         * config/i386/avxintrin.h: Ditto.
13805         * config/i386/rtmintrin.h: Ditto.
13806         * config/i386/fmaintrin.h: Ditto.
13807         * config/i386/mm3dnow.h: Ditto.
13809 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13811         * common/config/i386/i386-common.c: Handle LZCNT.
13813 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13815         * doc/extend.texi: Use __atomic_store_n instead of
13816         __atomic_store in HLE example.
13818 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13820         * config/sh/sh.c: Remove <cstdlib> workaround.
13822 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13824         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13826 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13828         * doc/extend.texi: Document that __atomic_clear and
13829           __atomic_test_and_set should only be used with bool.
13831 2013-06-20  Jan Hubicka  <jh@suse.cz>
13833         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13834         types_same_for_odr.
13835         * tree.c (decls_same_for_odr): New function.
13836         (same_for_edr): New function.
13837         (types_same_for_odr): New function.
13838         (get_binfo_at_offset): Use it.
13839         * tree.h (types_same_for_odr): Declare.
13841 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13842             Jason Merrill  <jason@redhat.com>
13844         * system.h: Include <cstdlib> as well as <stdlib.h>.
13846 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13848         PR target/57655
13849         * config/i386/i386.c (construct_container): Report error if
13850         long double is used with disabled x87 float returns.
13852 2013-06-20  Jan Hubicka  <jh@suse.cz>
13854         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13856 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13858         PR rtl-optimization/57425
13859         PR rtl-optimization/57569
13860         * alias.c (write_dependence_p): Remove parameters mem_mode and
13861         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13862         Changed all callers.
13863         (canon_anti_dependence): Get comments and semantics in sync.
13864         Add parameter mem_canonicalized.  Changed all callers.
13865         * rtl.h (canon_anti_dependence): Update prototype.
13867 2013-06-20  Richard Biener  <rguenther@suse.de>
13869         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13870         byte case, inline streamer_read_uchar and defer section
13871         overrun check.
13873 2013-06-20  Richard Biener  <rguenther@suse.de>
13875         PR tree-optimization/57584
13876         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13877         SSA names into the expanded expression that take part in
13878         abnormal coalescing.
13880 2013-06-19  Sharad Singhai  <singhai@google.com>
13882         * gcov.c (print_usage): Handle new option.
13883         (process_args): Ditto.
13884         (get_gcov_intermediate_filename): New function.
13885         (output_intermediate_file): New function.
13886         (output_gcov_file): New function
13887         (generate_results): Handle new option.
13888         (release_function): Relase demangled name.
13889         (read_graph_file): Handle demangled name.
13890         (output_lines): Ditto.
13891         * doc/gcov.texi: Document gcov intermediate format.
13893 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13895         PR bootstrap/57604
13896         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13897         (lra_emit_add): Use the functions.  Add comment about Y as an
13898         address segment.
13900 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13902         PR driver/57652
13903         * collect2.c (collect_atexit): New.
13904         (collect_exit): Delete.
13905         (main): Register collect_atexit with atexit.
13906         (collect_wait): Change collect_exit to exit.
13907         (do_wait): Same.
13908         * collect2.h (collect_exit): Delete.
13909         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13911 2013-06-19  Wei Mi  <wmi@google.com>
13913         PR rtl-optimization/57518
13914         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13915         if regno is used in paradoxical subreg.
13916         (update_equiv_regs): Check pdx_subregs[regno] before
13917         set a reg to be equivalent with a mem.
13919 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13921         PR driver/57651
13922         * file-find.h (find_a_file): Add a mode parameter.
13923         * file-find.c (find_a_file): Likewise.
13924         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13925         with X_OK for the executables.
13926         * collect2.c (main): Call find_a_file with X_OK.
13928 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13930         PR target/56942
13931         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13932         Use NEXT_INSN instead of next_real_insn.
13934 2013-06-19  Jan Hubicka  <jh@suse.cz>
13936         * cgraph.h (const_value_known_p): Replace by ...
13937         (ctor_for_folding): .. this one.
13938         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13939         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13940         * expr.c (expand_expr_real_1): Likewise.
13941         (string_constant): Likewise.
13942         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
13943         * ipa.c (process_references): Likewise.
13944         (symtab_remove_unreachable_nodes): Likewise.
13945         * ipa-inline-analysis.c (param_change_prob): Likewise.
13946         * gimple-fold.c (canonicalize_constructor_val): Likewise.
13947         (get_base_constructor): Likwise.
13948         * varpool.c (varpool_remove_node): Likewise.
13949         (varpool_remove_initializer): LIkewise.
13950         (dump_varpool_node): LIkwise.
13951         (const_value_known_p): Rewrite to ...
13952         (ctor_for_folding): ... this one.
13954 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
13956         PR driver/57651
13957         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
13958         PERSONALITY in $PATH derived prefixes.
13960 2013-06-19  Jeff Law  <law@redhat.com>
13962         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
13963         in comment.
13965         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
13966         (simplify_bitwise_binary): Use it to simpify certain binary ops on
13967         booleans.
13969 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
13971         * config/arm/vfp.md: Move VFP instruction classification documentation
13972         to ...
13973         * config/arm/arm.md: ... here.  Update instruction classification
13974         documentation.
13976 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
13978         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
13979         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
13980         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
13982 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
13984         PR target/57609
13985         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13986         with NEXT_INSN.  Use tablejump_p to check for jump table data
13987         insns.
13989 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
13991         PR c++/56544
13992         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
13993         that now in C++ the value is correct per the C++ standards.
13995 2013-06-19  Richard Biener  <rguenther@suse.de>
13997         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
13998         for global context.
14000 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14002         Revert:
14003         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14005         PR target/57609
14006         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14007         with next_active_insn.
14009 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14011         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14012         functions are inlined during failures to flag an error.
14013         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14014         in early inline pass.
14016 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14018         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14019         in comments.
14021 2013-06-18  Julian Brown  <julian@codesourcery.com>
14023         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14024         Permit virtual register pre-reload if !strict.
14025         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14026         change.
14027         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14028         prototype.
14029         * config/arm/neon.md (movmisalign<mode>): Use
14030         neon_perm_struct_or_reg_operand instead of
14031         neon_struct_or_register_operand.
14032         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14033         neon_permissive_struct_operand instead of neon_struct_operand.
14034         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14035         neon_vector_mem_operand.
14036         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14037         neon_vector_mem_operand.
14038         (neon_permissive_struct_operand): New.
14039         (neon_struct_or_register_operand): Rename to...
14040         (neon_perm_struct_or_reg_operand): This. Adjust call to
14041         neon_vector_mem_operand.
14043 2013-06-18  Richard Biener  <rguenther@suse.de>
14045         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14046         * lto-streamer.h: Include pointer-set.h.
14047         (struct lto_decl_slot): Remove.
14048         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14049         Remove next_index entry.
14050         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14051         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14052         (lto_init_tree_ref_encoder): Adjust.
14053         (lto_destroy_tree_ref_encoder): Likewise.
14054         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14055         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14056         (lto_output_decl_index): Adjust.
14057         (lto_new_out_decl_state): Likewise.
14058         (lto_record_function_out_decl_state): Likewise.
14059         * lto-streamer-out.c (copy_function): Likewise.
14061 2013-06-18  Richard Biener  <rguenther@suse.de>
14063         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14064         * cgraphunit.c: Include cfgloop.h.
14065         (init_lowered_empty_function): Initialize the loop tree.
14066         (assemble_thunk): Insert new BBs into loops.
14068 2013-06-18  Richard Biener  <rguenther@suse.de>
14070         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14071         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14072         the map from cache entry to cache index optional.
14073         (streamer_tree_cache_replace_tree): Adjust accordingly.
14074         (streamer_tree_cache_append): Likewise.
14075         (streamer_tree_cache_delete): Likewise.
14076         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14077         streamer cache map from cache entry to cache index.
14078         * lto-streamer-out.c (create_output_block): Adjust.
14080 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14082         * config/arm/arm.md (attribute "insn"): Move multiplication and
14083         division attributes to...
14084         (attribute "type"): ... here.  Remove mult.
14085         (attribute "mul32"): New attribute.
14086         (attribute "mul64"): Add umaal.
14087         (*arm_mulsi3): Update attributes.
14088         (*arm_mulsi3_v6): Likewise.
14089         (*thumb_mulsi3): Likewise.
14090         (*thumb_mulsi3_v6): Likewise.
14091         (*mulsi3_compare0): Likewise.
14092         (*mulsi3_compare0_v6): Likewise.
14093         (*mulsi_compare0_scratch): Likewise.
14094         (*mulsi_compare0_scratch_v6): Likewise.
14095         (*mulsi3addsi): Likewise.
14096         (*mulsi3addsi_v6): Likewise.
14097         (*mulsi3addsi_compare0): Likewise.
14098         (*mulsi3addsi_compare0_v6): Likewise.
14099         (*mulsi3addsi_compare0_scratch): Likewise.
14100         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14101         (*mulsi3subsi): Likewise.
14102         (*mulsidi3adddi): Likewise.
14103         (*mulsi3addsi_v6): Likewise.
14104         (*mulsidi3adddi_v6): Likewise.
14105         (*mulsidi3_nov6): Likewise.
14106         (*mulsidi3_v6): Likewise.
14107         (*umulsidi3_nov6): Likewise.
14108         (*umulsidi3_v6): Likewise.
14109         (*umulsidi3adddi): Likewise.
14110         (*umulsidi3adddi_v6): Likewise.
14111         (*smulsi3_highpart_nov6): Likewise.
14112         (*smulsi3_highpart_v6): Likewise.
14113         (*umulsi3_highpart_nov6): Likewise.
14114         (*umulsi3_highpart_v6): Likewise.
14115         (mulhisi3): Likewise.
14116         (*mulhisi3tb): Likewise.
14117         (*mulhisi3bt): Likewise.
14118         (*mulhisi3tt): Likewise.
14119         (maddhisi4): Likewise.
14120         (*maddhisi4tb): Likewise.
14121         (*maddhisi4tt): Likewise.
14122         (maddhidi4): Likewise.
14123         (*maddhidi4tb): Likewise.
14124         (*maddhidi4tt): Likewise.
14125         (divsi3): Likewise.
14126         (udivsi3): Likewise.
14127         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14128         (thumb2_mulsi_short_compare0): Likewise.
14129         (thumb2_mulsi_short_compare0_scratch): Likewise.
14130         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14131         (1020mult2): Likewise.
14132         (1020mult3): Likewise.
14133         (1020mult4): Likewise.
14134         (1020mult5): Likewise.
14135         (1020mult6): Likewise.
14136         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14137         change.
14138         (cortex_a15_mult64): Likewise.
14139         (cortex_a15_sdiv): Likewise.
14140         (cortex_a15_udiv): Likewise.
14141         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14142         (mult2): Likewise.
14143         (mult3): Likewise.
14144         (mult4): Likewise.
14145         (mult5): Likewise.
14146         (mult6): Likewise.
14147         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14148         (pj4_ir_div): Likewise.
14149         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14150         (11_mult2): Likewise.
14151         (11_mult3): Likewise.
14152         (11_mult4): Likewise.
14153         (11_mult5): Likewise.
14154         (11_mult6): Likewise.
14155         (11_mult7): Likewise.
14156         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14157         (cortex_a8_mla): Likewise.
14158         (cortex_a8_mull): Likewise.
14159         (cortex_a8_smulwy): Likewise.
14160         (cortex_a8_smlald): Likewise.
14161         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14162         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14163         (cortex_r4_mul_3): Likewise.
14164         (cortex_r4_mla_4): Likewise.
14165         (cortex_r4_mla_3): Likewise.
14166         (cortex_r4_smlald): Likewise.
14167         (cortex_r4_mull): Likewise.
14168         (cortex_r4_sdiv): Likewise.
14169         (cortex_r4_udiv): Likewise.
14170         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14171         (cortex_a7_idiv): Likewise.
14172         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14173         (9_mult2): Likewise.
14174         (9_mult3): Likewise.
14175         (9_mult4): Likewise.
14176         (9_mult5): Likewise.
14177         (9_mult6): Likewise.
14178         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14179         (cortex_a53_sdiv): Likewise.
14180         (cortex_a53_udiv): Likewise.
14181         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14182         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14183         (mp626_mult2): Likewise.
14184         (mp626_mult3): Likewise.
14185         (mp626_mult4): Likewise.
14186         * config/arm/fa526.md (526_mult1): Update attribute change.
14187         (526_mult2): Likewise.
14188         * config/arm/arm-generic.md (mult): Update attribute change.
14189         (mult_ldsched_strongarm): Likewise.
14190         (mult_ldsched): Likewise.
14191         (multi_cycle): Likewise.
14192         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14193         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14194         (606te_mult2): Likewise.
14195         (606te_mult3): Likewise.
14196         (606te_mult4): Likewise.
14197         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14198         (cortex_a9_mac16): Likewise.
14199         (cortex_a9_multiply): Likewise.
14200         (cortex_a9_mac): Likewise.
14201         (cortex_a9_multiply_long): Likewise.
14202         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14203         (626te_mult2): Likewise.
14204         (626te_mult3): Likewise.
14205         (626te_mult4): Likewise.
14207 2013-06-18  Richard Biener  <rguenther@suse.de>
14209         PR lto/57334
14210         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14212 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14214         PR target/57609
14215         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14216         with next_active_insn.
14218 2013-06-18  Alan Modra  <amodra@gmail.com>
14220         * config/rs6000/rs6000.h (enum data_align): New.
14221         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14222         (DATA_ABI_ALIGNMENT): Define.
14223         (CONSTANT_ALIGNMENT): Correct comment.
14224         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14225         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14227 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14229         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14230         ATTRIBUTE_UNUSED marking.
14232 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14234         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14235         alternative and update.
14236         (aarch64_dup_lanedi): Delete.
14237         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14238         * config/aarch64/aarch64-simd-builtins.def: Update.
14240 2013-06-17  Richard Biener  <rguenther@suse.de>
14242         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14243         (lto_input_scc): Declare.
14244         (lto_input_tree_1): Likewise.
14245         (struct lto_stats_d): Add num_tree_bodies_output and
14246         num_pickle_refs_output.
14247         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14248         (lto_read_tree_1): Split out from ...
14249         (lto_read_tree): ... this.
14250         (lto_input_scc): New function.
14251         (lto_input_tree_1): Split out from ...
14252         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14253         (lto_data_in_create): Create the streamer cache without hashes.
14254         * lto-streamer-out.c (create_output_block): Create the streamer
14255         cache with hashes when not doing WPA.
14256         (lto_write_tree_1): Split out from ...
14257         (lto_write_tree): ... this.
14258         (get_symbol_initial_value): New function.
14259         (lto_output_tree_1): Split out from ...
14260         (lto_output_tree): ... this.  Write trees as series of SCCs
14261         using a DFS walk via DFS_write_tree.
14262         (struct sccs, struct scc_entry): New types.
14263         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14264         (DFS_write_tree_body): New function.
14265         (DFS_write_tree): Likewise.
14266         (hash_tree): Likewise.
14267         (scc_entry_compare): Likewise.
14268         (hash_scc): Likewise.
14269         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14270         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14271         TREE_CHAIN as regular reference.
14272         (streamer_read_integer_cst): Remove.
14273         (streamer_get_pickled_tree): Adjust.
14274         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14275         of DECL_EXTERNALs in BLOCK_VARS for now.
14276         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14277         reference.
14278         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14279         Add hash value argument and record that if hashes are recorded
14280         in the cache.
14281         (streamer_tree_cache_insert_1): Adjust.
14282         (streamer_tree_cache_insert): Likewise.
14283         (streamer_tree_cache_insert_at): Rename to ...
14284         (streamer_tree_cache_replace_tree): ... this and adjust.
14285         (streamer_tree_cache_append): Adjust.
14286         (record_common_node): Likewise.
14287         (streamer_tree_cache_create): Add argument whether to
14288         record hash values together with trees.
14289         (streamer_tree_cache_delete): Adjust.
14290         * tree-streamer.h (struct streamer_tree_cache_d): Add
14291         vector of hashes.
14292         (streamer_read_integer_cst): Remove.
14293         (streamer_tree_cache_insert): Adjust.
14294         (streamer_tree_cache_append): Likewise.
14295         (streamer_tree_cache_insert_at): Rename to ...
14296         (streamer_tree_cache_replace_tree): ... this and adjust.
14297         (streamer_tree_cache_create): Add argument whether to record hashes.
14298         (streamer_tree_cache_get): Rename to ...
14299         (streamer_tree_cache_get_tree): ... this.
14300         (streamer_tree_cache_get_hash): New function.
14301         * tree.c (cache_integer_cst): New function.
14302         * tree.h (cache_integer_cst): Declare.
14303         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14304         * lto-symtab.c (lto_varpool_replace_node): Only release
14305         DECL_INITIAL of non-prevailing decls.
14306         * varpool.c (varpool_remove_initializer): Do not release
14307         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14309 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14311         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14312         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14313         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14314         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14315         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14316         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14317         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14318         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14319         instead of TARGET_64BIT.
14320         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14321         Require ISA_HAS_<D>DIV.
14323 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14325         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14326         (mips*-*-linux*): Move default with_llsc setting to where other
14327         defaults are set.
14328         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14329         with_arch block.
14330         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14331         Likewise.  Remove default with_tune setting.  Move default float
14332         setting to its own block.  Handle with_llsc in the same block as above.
14334 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14336         PR rtl-optimization/57425
14337         PR rtl-optimization/57569
14338         * alias.c (write_dependence_p): Add new parameters mem_mode,
14339         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14340         Changed all callers.
14341         (canon_anti_dependence): New function.
14342         * cse.c (check_dependence): Use canon_anti_dependence.
14343         * cselib.c (cselib_invalidate_mem): Likewise.
14344         * rtl.h (canon_anti_dependence): Declare.
14346 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14348         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14349         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14350         ".set mips3" for 64-bit targets.
14352 2013-06-15  Dehao Chen  <dehao@google.com>
14354         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14355         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14356         (gimple_check_call_args): Likewise.
14357         * value-prof.c (check_ic_target): Likewise.
14358         * ipa-inline.c (early_inliner): Likewise.
14359         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14360         * cgraph.c (cgraph_create_edge_1): Likewise.
14361         (cgraph_make_edge_direct): Likewise.
14363 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14365         PR target/57615
14366         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14367         rs6000_output_move_128bit to handle emitting quad memory
14368         operations.  Set attribute length to 8 bytes.
14370 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14372         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14373         New pattern.
14374         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14375         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14376         (aarch64_<su>mlsl<mode>): Likewise.
14378 2013-06-14  Mike Stump  <mikestump@comcast.net>
14380         * Makefile.in (TARGET_H): Add insn-codes.h.
14382 2013-06-14  Alan Modra  <amodra@gmail.com>
14384         PR middle-end/57134
14385         PR middle-end/57586
14386         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14387         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14388         bitfield expansion when EXPAND_MEMORY.
14389         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14391 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14393         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14394         test for clearing quad memory on 32-bit later.
14396 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14398         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14399         (fold_negate_expr): Likewise.
14400         (fold_real_zero_addition_p): Handle vectors.
14401         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14403 2013-06-14  Alan Modra  <amodra@gmail.com>
14405         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14407 2013-06-13  Jan Hubicka  <jh@suse.cz>
14409         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14410         Local comdats are not externally visible.
14411         * symtab.c (dump_symtab_base): Dump externally visible.
14412         (verify_symtab_base): Verify back links in the symtab hash.
14414 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14416         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14417         CONVERT_EXPR as equal nodes.
14419 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14421         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14423 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14425         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14426         Generalize to complex and vector.
14427         * tree.c (build_all_ones_cst): New function.
14428         * tree.h (build_all_ones_cst): Declare it.
14430 2013-06-13  Alan Modra  <amodra@gmail.com>
14432         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14433         * config/rs6000/rs6000.md (signbittf2): New insn.
14434         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14435         (abstf2_internal, cmptf_internal2): Likewise.
14436         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14438 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14439             Pat Haugen  <pthaugen@us.ibm.com>
14440             Peter Bergner  <bergner@vnet.ibm.com>
14442         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14443         power8 byte, half-word, and quad-word atomic instructions.
14444         (emit_store_conditional): Likewise.
14445         (rs6000_expand_atomic_compare_and_swap): Likewise.
14446         (rs6000_expand_atomic_op): Likewise.
14448         * config/rs6000/sync.md (larx): Add new modes for power8.
14449         (stcx): Likewise.
14450         (AINT): New mode iterator to include TImode as well as normal
14451         integer modes on power8.
14452         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14453         that VSX registers are not considered.  Use AINT mode iterator
14454         instead of INT1 to allow inclusion of quad word atomic operations
14455         on power8.
14456         (load_locked<mode>): Likewise.
14457         (store_conditional<mode>): Likewise.
14458         (atomic_compare_and_swap<mode>): Likewise.
14459         (atomic_exchange<mode>): Likewise.
14460         (atomic_nand<mode>): Likewise.
14461         (atomic_fetch_<fetchop_name><mode>): Likewise.
14462         (atomic_nand_fetch<mode>): Likewise.
14463         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14464         each type.
14465         (ATOMIC): On power8, add QImode, HImode modes.
14466         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14467         modes that promote to SImode.
14468         (load_lockedti): Convert TImode arguments to PTImode, so that we
14469         get a guaranteed even/odd register pair.
14470         (load_lockedpti): Likewise.
14471         (store_conditionalti): Likewise.
14472         (store_conditionalpti): Likewise.
14474         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14475         atomic load/store instructions.
14476         (HSI): Likewise.
14478 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14480         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14481         loads.
14482         (insn_count): New attribute, with most cases extracted from...
14483         (length): ...here.  Redefine most cases in terms of insn_count.
14484         (single_insn): Delete.
14485         (can_delay): Use insn_count to check for single instructions.
14486         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14487         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14488         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14489         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14490         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14491         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14492         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14493         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14494         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14495         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14496         rather than "length".
14497         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14498         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14499         Use "insn_count" rather than "length".
14500         * config/mips/mips-dsp.md
14501         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14502         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14503         length attributes.
14505 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14507         PR tree-optimization/57361
14508         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14510 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14512         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14513         to split.
14514         (aarch64_simd_combine<mode>): New instruction expansion.
14515         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14516         function prototype.
14517         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14518         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14520 2013-06-12  Jan Hubicka  <jh@suse.cz>
14522         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14523         decl has when in streaming stage.
14524         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14525         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14527 2013-06-12  Roland Stigge  <stigge@antcom.de>
14529         PR target/57578
14530         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14532 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14534         PR tree-optimization/57537
14535         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14536         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14538 2013-06-12  Richard Biener  <rguenther@suse.de>
14540         * data-streamer.h (streamer_write_char_stream): CSE
14541         obs->current_pointer.
14542         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14543         streamer_write_char_stream manually and optimize the resulting loop.
14544         (streamer_write_hwi_stream): Likewise.
14546 2013-06-12  Jan Hubicka  <jh@suse.cz>
14548         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14549         * cgraph.h (varpool_create_empty_node): Declare.
14550         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14551         duplicated nodes.
14552         * symtab.c (symtab_unregister_node): Be lax about missin entries
14553         in node hash.
14554         (symtab_get_node): Update comment.
14555         * varpool.c (varpool_create_empty_node): Break out from ...
14556         (varpool_node_for_decl): ... here.
14557         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14559 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14561         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14562         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14563         part.  Use straight-line flow at the end.
14564         <COMPONENT_REF>: Remove superfluous else.
14565         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14567 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14569         PR target/56564
14570         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14571         target hook even for !TREE_PUBLIC decls.  If no resolution info
14572         is available, return false for common and external decls.
14574 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14576         * config/rl78/constraints.md (U): New constraint.
14577         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14578         valloc attribute.
14580 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14582         PR target/57589
14583         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14584         to allow returning address to AT_PLATFORM name.
14586 2013-06-11  Jan Hubicka  <jh@suse.cz>
14588         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14589         * cgraph.h (symtab_node_base): Add weakref flag.
14590         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14591         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14592         (output_weakrefs): Use weakref flag.
14593         * fold-const.c (simple_operand_p): Handle WEAK.
14594         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14595         * ipa.c (varpool_externally_visible_p): Drop weakref.
14596         (function_and_variable_visibility): Update comment; fix weakref
14597         sanity checks; do not clear DECL_WEAK on them.
14598         * lto-cgraph.c (lto_output_node): update.
14599         (lto_output_varpool_node): Update.
14600         (input_overwrite_node): Update.
14601         (input_node): Update.
14602         (input_varpool_node): Update.
14603         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14604         (lto_symtab_merge_symbols): Add sanity check.
14605         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14606         * passes.c (rest_of_decl_compilation): Set static flag, too.
14607         * symtab.c (dump_symtab_base): Dump weakref.
14608         (verify_symtab_base): Sanity check weakrefs.
14609         (symtab_make_decl_local): Remove duplicated code.
14610         (symtab_alias_ultimate_target): Simplify.
14611         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14613 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14615         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14616         sequence_vect == NULL.
14618 2013-06-11  DJ Delorie  <dj@redhat.com>
14620         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14621         (rl78_unwind_word_mode): New.
14623 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14625         * final.c (debug_prefix_maps): Make static.
14627 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14629         * function.c (initial_trampoline): Remove stray copy.
14631 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14633         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14635 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14637         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14638         within bounds at the beginning of the function.
14640 2013-06-11  Alan Modra  <amodra@gmail.com>
14642         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14643         reporting.
14644         (get_named_section): Don't NULL !DECL_P decl.
14646 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14648         * doc/invoke.texi (core-avx2): Document.
14649         (slm): Likewise.
14650         (atom): Updated with MOVBE.
14652 2013-06-11  Richard Biener  <rguenther@suse.de>
14654         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14656 2013-06-11  Anton Blanchard  <anton@samba.org>
14658         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14659         correct shift value in little-endian mode.
14661 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14663         PR target/56564
14664         * varasm.c (get_variable_align): Move #endif to the right place.
14666 2013-06-10  Cary Coutant  <ccoutant@google.com>
14668         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14669         for hash so that hash table traversal order is deterministic.
14671 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14672             Pat Haugen  <pthaugen@us.ibm.com>
14673             Peter Bergner  <bergner@vnet.ibm.com>
14675         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14676         of vectors in GPRS if they are direct moves or quad word load or
14677         store moves.
14679         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14680         declaration.
14681         (direct_move_p): Likewise.
14682         (quad_load_store_p): Likewise.
14684         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14685         classes into bins based on the physical register type.
14686         (reg_class_to_reg_type): Likewise.
14687         (IS_STD_REG_TYPE): Likewise.
14688         (IS_FP_VECT_REG_TYPE): Likewise.
14689         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14690         use direct move instructions.
14691         (reload_gpr_vsx): Likewise.
14692         (reload_vsx_gpr): Likewise.
14693         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14694         information that is a simplification of register classes.  Also
14695         precalculate direct move reload helpers.
14696         (direct_move_p): New function to return true if the operation can
14697         be done as a direct move instruciton.
14698         (quad_load_store_p): New function to return true if the operation
14699         is a quad memory operation.
14700         (rs6000_legitimize_address): If quad memory, only allow register
14701         indirect for TImode addresses.
14702         (rs6000_legitimate_address_p): Likewise.
14703         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14704         (rs6000_reload_register_type): Likewise.
14705         (register_to_reg_type): Return register type.
14706         (rs6000_secondary_reload_simple_move): New helper function for
14707         secondary reload and secondary memory needed to identify anything
14708         that is a simple move, and does not need reloading.
14709         (rs6000_secondary_reload_direct_move): New helper function for
14710         secondary reload to identify cases that can be done with several
14711         instructions via the direct move instructions.
14712         (rs6000_secondary_reload_move): New helper function for secondary
14713         reload to identify moves between register types that can be done.
14714         (rs6000_secondary_reload): Add support for quad memory operations
14715         and for direct move.
14716         (rs6000_secondary_memory_needed): Likewise.
14717         (rs6000_debug_secondary_memory_needed): Change argument names.
14718         (rs6000_output_move_128bit): New function to return the move to
14719         use for 128-bit moves, including knowing about the various
14720         limitations of quad memory operations.
14722         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14723         memory operations.  call rs6000_output_move_128bit for the actual
14724         instruciton(s) to generate.
14725         (vsx_movti_64bit): Likewise.
14727         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14728         (UNSPEC_P8V_MTVSRWZ): Likewise.
14729         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14730         (UNSPEC_P8V_MTVSRD): Likewise.
14731         (UNSPEC_P8V_XXPERMDI): Likewise.
14732         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14733         (UNSPEC_FUSION_GPR): Likewise.
14734         (FMOVE128_GPR): New iterator for direct move.
14735         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14736         (f32_sv): Likewise.
14737         (f32_dm): Likewise.
14738         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14739         loads and direct move instructions.
14740         (zero_extendsidi2_lfiwzx): Likewise.
14741         (extendsidi2_lfiwax): Likewise.
14742         (extendsidi2_nocell): Likewise.
14743         (floatsi<mode>2_lfiwax): Likewise.
14744         (lfiwax): Likewise.
14745         (floatunssi<mode>2_lfiwzx): Likewise.
14746         (lfiwzx): Likewise.
14747         (fix_trunc<mode>_stfiwx): Likewise.
14748         (fixuns_trunc<mode>_stfiwx): Likewise.
14749         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14750         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14751         (parity<mode>2_cmpb): Set length/type attr.
14752         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14753         for 'mr.' to fast_compare.
14754         (bpermd_<mode>): Change type attr to popcnt.
14755         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14756         (p8_mtvsrwz_1): Likewise.
14757         (p8_mtvsrwz_2): Likewise.
14758         (reload_fpr_from_gpr<mode>): Likewise.
14759         (p8_mtvsrd_1): Likewise.
14760         (p8_mtvsrd_2): Likewise.
14761         (p8_xxpermdi_<mode>): Likewise.
14762         (reload_vsx_from_gpr<mode>): Likewise.
14763         (reload_vsx_from_gprsf): Likewise.
14764         (p8_mfvsrd_3_<mode>): LIkewise.
14765         (reload_gpr_from_vsx<mode>): Likewise.
14766         (reload_gpr_from_vsxsf): Likewise.
14767         (p8_mfvsrd_4_disf): Likewise.
14768         (multi-word GPR splits): Do not split direct moves or quad memory
14769         operations.
14771 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14773         * tree-into-ssa.c (interesting_blocks): Make static.
14775 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14777         PR target/56564
14778         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14779         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14780         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14781         (get_variable_align): New function.
14782         (get_variable_section, emit_bss, emit_common,
14783         assemble_variable_contents, place_block_symbol): Use
14784         get_variable_align instead of DECL_ALIGN.
14785         (assemble_noswitch_variable): Add align argument, use it
14786         instead of DECL_ALIGN.
14787         (assemble_variable): Adjust caller.  Use get_variable_align
14788         instead of DECL_ALIGN.
14789         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14790         caller.
14791         (DATA_ABI_ALIGNMENT): Define.
14792         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14793         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14794         opt is false, only return the psABI mandated alignment increase.
14795         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14796         (DATA_ABI_ALIGNMENT): ... this.
14797         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14798         (DATA_ABI_ALIGNMENT): ... this.
14799         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14800         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14801         (DATA_ABI_ALIGNMENT): ... this.
14802         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14803         * doc/tm.texi: Regenerated.
14805 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14807         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14808         cmp_code to construct REG_EQUAL note.
14810 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14812         PR target/57568
14813         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14814         that operands[2] doesn't overlap with operands[0].
14816 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14817             Jan Hubicka  <jh@suse.cz>
14819         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14820         hack to mark symbols as used.
14822 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14824         PR rtl-optimization/57559
14825         * lra-constraints.c (process_alt_operands): Don't discourage
14826         memory with known offset for offsetable memory constraint.
14827         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14829 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14831         * varasm.c (struct oc_local_state): Reorder fields.
14832         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14833         and adjust accordingly.
14834         (output_constructor): Reorder initialization code and adjust call to
14835         output_constructor_bitfield.
14837 2013-06-07  Jan Hubicka  <jh@suse.cz>
14839         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14841 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14843         * tree-object-size.c (unknown): Make const.
14845 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14847         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14848         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14849         for last alternative in the cpu_facility attribute.
14851 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14853         PR target/56315
14854         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14855         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14856         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14857         * config/arm/constraints.md (Dg): New constraint.
14858         * config/arm/neon.md (xordi3_neon): Remove.
14859         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14860         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14862 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14864         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14865         Clean up alternatives.
14867 2013-06-07  Alan Modra  <amodra@gmail.com>
14869         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14870         va_list_gpr_size.
14872 2013-06-07  Alan Modra  <amodra@gmail.com>
14874         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14876 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14878         * config/arm/constraints.md (Df): New constraint.
14879         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14880         Correct length attribute for last two alternatives.
14882 2013-06-07  Alan Modra  <amodra@gmail.com>
14884         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14885         override user -mfp-in-toc.
14886         (offsettable_ok_by_alignment): Consider just the current access
14887         rather than the whole object, unless BLKmode.  Handle
14888         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14889         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14890         for -mcmodel=medium.
14891         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14892         override user -mfp-in-toc or -msum-in-toc.  Default to
14893         -mno-fp-in-toc for -mcmodel=medium.
14895 2013-06-06  DJ Delorie  <dj@redhat.com>
14897         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14898         TARGET_VALID_POINTER_MODE.
14900 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14901             Pat Haugen  <pthaugen@us.ibm.com>
14902             Peter Bergner  <bergner@vnet.ibm.com>
14904         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14905         Document new power8 builtins.
14907         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14908         condition code register, to allow 128-bit logical operations to be
14909         done in the VSX or GPR registers.
14910         (nor<mode>3): Use the canonical form for nor.
14911         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14912         vclz*, and vpopcnt* vector instructions.
14913         (nand<mode>3): Likewise.
14914         (orc<mode>3): Likewise.
14915         (clz<mode>2): LIkewise.
14916         (popcount<mode>2): Likewise.
14918         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14919         that only the GPRs are recognized.
14921         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14922         support for new power8 builtins.
14924         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14925         builtin functions.
14926         (xscvdpspn): Likewise.
14927         (vclz): Likewise.
14928         (vclzb): Likewise.
14929         (vclzh): Likewise.
14930         (vclzw): Likewise.
14931         (vclzd): Likewise.
14932         (vpopcnt): Likewise.
14933         (vpopcntb): Likewise.
14934         (vpopcnth): Likewise.
14935         (vpopcntw): Likewise.
14936         (vpopcntd): Likewise.
14937         (vgbbd): Likewise.
14938         (vmrgew): Likewise.
14939         (vmrgow): Likewise.
14940         (eqv): Likewise.
14941         (eqv_v16qi3): Likewise.
14942         (eqv_v8hi3): Likewise.
14943         (eqv_v4si3): Likewise.
14944         (eqv_v2di3): Likewise.
14945         (eqv_v4sf3): Likewise.
14946         (eqv_v2df3): Likewise.
14947         (nand): Likewise.
14948         (nand_v16qi3): Likewise.
14949         (nand_v8hi3): Likewise.
14950         (nand_v4si3): Likewise.
14951         (nand_v2di3): Likewise.
14952         (nand_v4sf3): Likewise.
14953         (nand_v2df3): Likewise.
14954         (orc): Likewise.
14955         (orc_v16qi3): Likewise.
14956         (orc_v8hi3): Likewise.
14957         (orc_v4si3): Likewise.
14958         (orc_v2di3): Likewise.
14959         (orc_v4sf3): Likewise.
14960         (orc_v2df3): Likewise.
14962         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
14963         allow power8 quad mode in 64-bit.
14964         (rs6000_builtin_vectorized_function): Add support to vectorize
14965         ISA 2.07 count leading zeros, population count builtins.
14966         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
14967         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
14968         (builtin_function_type): Add vgbbd builtin function which takes an
14969         unsigned argument.
14970         (altivec_expand_vec_perm_const): Add support for new power8 merge
14971         instructions.
14973         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
14974         that does not include TImdoe for use with 32-bit.
14975         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
14976         instructions.
14977         (UNSPEC_VSX_CVDPSPN): Likewise.
14978         (vsx_xscvdpspn): Likewise.
14979         (vsx_xscvspdpn): Likewise.
14980         (vsx_xscvdpspn_scalar): Likewise.
14981         (vsx_xscvspdpn_directmove): Likewise.
14982         (vsx_and<mode>3): Split logical operations into 32-bit and
14983         64-bit. Add support to do logical operations on TImode as well as
14984         VSX vector types.  Allow logical operations to be done in either
14985         VSX registers or in general purpose registers in 64-bit mode.  Add
14986         splitters if GPRs were used. For AND, add clobber of CCmode to
14987         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
14988         encoding.
14989         (vsx_and<mode>3_32bit): Likewise.
14990         (vsx_and<mode>3_64bit): Likewise.
14991         (vsx_ior<mode>3): Likewise.
14992         (vsx_ior<mode>3_32bit): Likewise.
14993         (vsx_ior<mode>3_64bit): Likewise.
14994         (vsx_xor<mode>3): Likewise.
14995         (vsx_xor<mode>3_32bit): Likewise.
14996         (vsx_xor<mode>3_64bit): Likewise.
14997         (vsx_one_cmpl<mode>2): Likewise.
14998         (vsx_one_cmpl<mode>2_32bit): Likewise.
14999         (vsx_one_cmpl<mode>2_64bit): Likewise.
15000         (vsx_nor<mode>3): Likewise.
15001         (vsx_nor<mode>3_32bit): Likewise.
15002         (vsx_nor<mode>3_64bit): Likewise.
15003         (vsx_andc<mode>3): Likewise.
15004         (vsx_andc<mode>3_32bit): Likewise.
15005         (vsx_andc<mode>3_64bit): Likewise.
15006         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15007         and xxlorc instructions.
15008         (vsx_eqv<mode>3_64bit): Likewise.
15009         (vsx_nand<mode>3_32bit): Likewise.
15010         (vsx_nand<mode>3_64bit): Likewise.
15011         (vsx_orc<mode>3_32bit): Likewise.
15012         (vsx_orc<mode>3_64bit): Likewise.
15014         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15016         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15017         instruction.
15018         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15019         (p8_vmrgow): Likewise.
15020         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15021         GPRs to be split under VSX.
15022         (p8v_clz<mode>2): Add power8 count leading zero support.
15023         (p8v_popcount<mode>2): Add power8 population count support.
15024         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15025         support.
15027         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15028         instruction.
15030         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15031         builtin functions.
15032         (vec_nand): Likewise.
15033         (vec_vclz): Likewise.
15034         (vec_vclzb): Likewise.
15035         (vec_vclzd): Likewise.
15036         (vec_vclzh): Likewise.
15037         (vec_vclzw): Likewise.
15038         (vec_vgbbd): Likewise.
15039         (vec_vmrgew): Likewise.
15040         (vec_vmrgow): Likewise.
15041         (vec_vpopcnt): Likewise.
15042         (vec_vpopcntb): Likewise.
15043         (vec_vpopcntd): Likewise.
15044         (vec_vpopcnth): Likewise.
15045         (vec_vpopcntw): Likewise.
15047 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15049         PR rtl-optimization/57468
15050         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15051         spilled pseudos.
15053 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15055         PR rtl-optimization/57459
15056         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15057         type when setting live regs.
15059 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15061         * config/s390/s390.opt (mlra): New option.
15062         * config/s390/s390.c (s390_decompose_address): Check displacement
15063         for all registers for LRA.
15064         (s390_secondary_reload): Don't used secondary reloads for LRA.
15065         (s390_lra_p): New function.
15066         (TARGET_LRA_P): Define.
15067         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15068         of attribute cpu_facility to zarch for the last alternative.
15069         (*cmpmem_short): Ditto.
15071 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15073         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15074         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15075         (arm_expand_prologue): Likewise.
15077 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15079         PR c++/53743
15080         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15081         as this is now done by redirect_edge_and_branch_force.
15082         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15083         barriers, and fix interaction with splitting.
15084         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15085         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15086         reflect changes made in the routine.
15087         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15088         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15089         since this is called in cfglayout mode, and replace partition fixup
15090         with assert as that is now done by force_nonfallthru_and_redirect.
15091         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15092         already be marked with region crossing note.
15093         (insert_section_boundary_note): Make non-static, gate on flag
15094         has_bb_partition, rewrite to also check for multiple partitions.
15095         (rest_of_handle_reorder_blocks): Remove call to
15096         insert_section_boundary_note, now done later during free_cfg.
15097         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15098         * bb-reorder.h (insert_section_boundary_note): Declare.
15099         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15100         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15101         invoke insert_section_boundary_note.
15102         (try_redirect_by_replacing_jump): Remove unnecessary
15103         check for region crossing note.
15104         (fixup_partition_crossing): New function.
15105         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15106         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15107         in non-cfglayout mode.
15108         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15109         remove old code that tried to do this. Emit barrier correctly
15110         when we are in cfglayout mode.
15111         (last_bb_in_partition): New function.
15112         (rtl_split_edge): Correctly fixup partition boundaries.
15113         (commit_one_edge_insertion): Remove old code that tried to
15114         fixup region crossing edge since this is now handled in
15115         split_block, and set up insertion point correctly since
15116         block may now end in a jump.
15117         (verify_hot_cold_block_grouping): Guard against checking when not in
15118         linearized RTL mode.
15119         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15120         notes.
15121         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15122         rtl_verify_flow_info, so not called in cfglayout mode.
15123         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15124         (fixup_reorder_chain): Remove old code that attempted to fixup region
15125         crossing note as this is now handled in force_nonfallthru_and_redirect.
15126         (duplicate_insn_chain): Don't duplicate switch section notes.
15127         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15128         note.
15129         * basic-block.h (emit_barrier_after_bb): Declare.
15131 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15133         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15134         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15135         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15137 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15139         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15140         where appropriate.
15141         * config/arm/ldmstm.md: Regenerate.
15143 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15145         * config/arm/sync.md (atomic_loaddi_1):
15146         Disable predication for arm_restrict_it.
15147         (arm_load_exclusive<mode>): Likewise.
15148         (arm_load_exclusivesi): Likewise.
15149         (arm_load_exclusivedi): Likewise.
15150         (arm_load_acquire_exclusive<mode>): Likewise.
15151         (arm_load_acquire_exclusivesi): Likewise.
15152         (arm_load_acquire_exclusivedi): Likewise.
15153         (arm_store_exclusive<mode>): Likewise.
15154         (arm_store_exclusive<mode>): Likewise.
15155         (arm_store_release_exclusivedi): Likewise.
15156         (arm_store_release_exclusive<mode>): Likewise.
15158 2013-06-06  Richard Biener  <rguenther@suse.de>
15160         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15161         after LTO_null.
15162         (lto_tag_is_tree_code_p): Adjust.
15163         (lto_tag_is_gimple_code_p): Likewise.
15164         (lto_gimple_code_to_tag): Likewise.
15165         (lto_tag_to_gimple_code): Likewise.
15166         (lto_tree_code_to_tag): Likewise.
15167         (lto_tag_to_tree_code): Likewise.
15168         * data-streamer.h (streamer_write_hwi_in_range): Use
15169         uhwi streaming to stream the normalized range.
15170         (streamer_read_hwi_in_range): Likewise.
15172 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15174         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15175         (predicable_short_it): Likewise.
15176         (predicated): Likewise.
15177         (enabled): Handle above.
15178         (define_cond_exec): Set predicated attribute to yes.
15180 2013-06-05  Mike Stump  <mikestump@comcast.net>
15182         * gdbinit.in (__FUNCTION__): Add.
15184 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15186         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15187         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15189 2013-06-05  Jan Hubicka  <jh@suse.cz>
15191         * varasm.c (mark_decl_referenced): Revert the removal until targets
15192         are fixed.
15194 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15196         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15197         instead of mark_decl_referenced.
15199 2013-06-05  Jan Hubicka  <jh@suse.cz>
15201         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15202         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15203         and symtab_used_from_object_file_p.
15204         (cgraph_make_node_local_1): Clear forced_by_abi.
15205         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15206         * cgraph.h (symtab_node_base): Add forced_by_abi.
15207         (decide_is_variable_needed): Remove.
15208         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15209         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15210         (decide_is_symbol_needed): ... this one; handle symbols in general;
15211         always analyze virtuals; honnor forced_by_abi.
15212         (cgraph_finalize_function): Update.
15213         (varpool_finalize_decl): Update.
15214         (symbol_defined_and_needed): Remove.
15215         (analyze_functions): Update.
15216         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15217         output_refs, input_overwrite_node): Handle forced_by_abi.
15218         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15219         (address_taken_from_non_vtable_p): ... this one.
15220         (comdat_can_be_unshared_p_1): New function.
15221         (cgraph_comdat_can_be_unshared_p): Rename to ...
15222         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15223         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15224         (function_and_variable_visibility): Clear forced_by_abi as needed.
15225         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15226         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15227         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15228         * varpool.c (decide_is_variable_needed): Remove.
15230 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15232         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15233         (arm_option_override): Override arm_restrict_it where appropriate.
15234         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15235         * config/arm/arm.opt (mrestrict-it): New command-line option.
15236         * doc/invoke.texi: Document -mrestrict-it.
15238 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15240         * tsan.c (tsan_atomic_table): Make const.
15242 2013-06-05  Richard Biener  <rguenther@suse.de>
15244         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15245         index associated with the tree we are supposed to replace.
15246         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15247         TREE_ASM_WRITTEN as zero for everything but SSA names.
15249 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15251         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15253 2013-06-04  Jan Hubicka  <jh@suse.cz>
15255         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15256         (input_node, input_varpool_node): Handle correctly external same
15257         body aliases.
15258         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15259         nodes at ltrans stage.
15261 2013-06-04  Jan Hubicka  <jh@suse.cz>
15263         * ipa-inline.c (update_caller_keys): Fix availability test.
15264         (update_callee_keys): Likewise.
15265         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15266         to follow ELF standard.
15268 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15270         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15271         (mips64r5900el-*-elf*): New configurations.
15272         * config/mips/mips-cpus.def (r5900): New processor.
15273         * config/mips/mips-tables.opt: Regenerate.
15274         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15275         (mips_issue_rate): Handle PROCESSOR_R5900.
15276         (mips_reorg_process_insns): Force reorder mode for the R5900.
15277         * config/mips/mips.h (TARGET_MIPS5900): Define.
15278         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15279         TARGET_MIPS5900.
15280         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15281         TARGET_MIPS5900.
15282         * config/mips/mips.md (processor): Add r5900.
15283         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15285 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15287         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15288         into function to generate MOVI instruction.
15289         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15290         (aarch64_preferred_simd_mode): Turn into wrapper.
15291         (aarch64_output_scalar_simd_mov_immediate): New function.
15292         * config/aarch64/aarch64-protos.h: Add prototype for above.
15294 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15296         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15297         element_char member.
15298         (sizetochar): Return signed char.
15299         (aarch64_simd_valid_immediate): Remove elchar and other
15300         unnecessary variables.
15301         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15302         Calculate element_char as required.
15303         * config/aarch64/aarch64-protos.h: Update and move prototype
15304         for aarch64_output_simd_mov_immediate.
15305         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15306         Update arguments.
15308 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15310         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15311         information completed by aarch64_simd_valid_immediate.
15312         (aarch64_legitimate_constant_p): Update arguments.
15313         (aarch64_simd_valid_immediate): Work with struct rather than many
15314         pointers.
15315         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15316         (aarch64_simd_make_constant): Update arguments.
15317         (aarch64_output_simd_mov_immediate): Work with struct rather than
15318         many pointers.  Output immediate directly rather than as operand.
15319         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15320         Update prototype.
15321         * config/aarch64/constraints.md (Dn): Update arguments.
15323 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15325         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15326         longer static.
15327         (aarch64_simd_immediate_valid_for_move): Remove.
15328         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15329         (aarch64_simd_make_constant): Update call.
15330         (aarch64_output_simd_mov_immediate): Update call.
15331         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15332         Add prototype.
15333         * config/aarch64/constraints.md (Dn): Update call.
15335 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15337         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15338         return type to bool for prototype.
15339         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15340         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15341         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15343 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15345         * config/mips/mips.opt (meva): New.
15346         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15347         (ASM_SPEC): Handle -meva.
15348         * doc/invoke.texi (meva):  Document.
15350 2013-06-04  Alan Modra  <amodra@gmail.com>
15352         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15353         constant output.
15355 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15357         * rtl.def: Add extra fourth optional field to define_cond_exec.
15358         * gensupport.c (process_one_cond_exec): Process attributes from
15359         define_cond_exec.
15360         * doc/md.texi: Document fourth field in define_cond_exec.
15362 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15364         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15365         out the processing order as in store_bit_field_1.
15367 2013-06-04  Jan Hubicka  <jh@suse.cz>
15369         PR middle-end/57500
15370         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15371         non-VAR_DECL node if it does not exist yet.
15373 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15375         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15376         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15377         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15378         target_cpu_default setting.
15380 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15382         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15383         default to -fopt-info=optimized instead of all.
15384         * doc/invoke.texi: Ditto.
15385         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15386         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15387         (execute_vect_slp): Emit BB vectorization success under
15388         MSG_OPTIMIZED_LOCATIONS.
15389         * tree-vect-slp.c (vect_slp_transform_bb): Change
15390         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15391         * tree-vect-loop.c (vect_transform_loop): Ditto.
15393 2013-06-03  Jason Merrill  <jason@redhat.com>
15395         PR c++/57415
15396         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15397         Use TARGET_EXPR for C++.
15399 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15401         PR rtl-optimization/57268
15402         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15403         if DEBUG_INSN_P (insn).
15405         Reapply
15406         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15408         PR rtl-optimization/57268
15409         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15410         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15412 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15414         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15415         (ix86_avoid_lea_for_addr): Likewise.
15416         (exact_dependency_1): Likewise.
15417         (ix86_adjust_cost): Likewise.
15418         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15419         removed.
15420         (do_reorder_for_imul): Fix typo, formatting and
15421         !reload_completed check removed.
15422         (ix86_sched_reorder): Fix typo and formatting.
15423         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15424         list.
15426 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15428         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15430 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15432         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15433         <STRING_CST>: Likewise.
15434         <VECTOR_CST>: Likewise.
15436 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15437             Mikael Morin  <mikael@gcc.gnu.org>
15439         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15440         * config.in: Regenerated.
15441         * configure: Regenerated.
15443 2013-06-01  Jan Hubicka  <jh@suse.cz>
15445         PR middle-end/57366
15446         * cgraphunit.c (compile): When weakref is not supported,
15447         set up transparent aliases before final output pass.
15448         * varasm.c (assemble_alias): Do not try to do it here.
15450 2013-06-01  Jan Hubicka  <jh@suse.cz>
15452         PR middle-end/57467
15453         * passes.c (for_per_function): Skip unanalyzed functions.
15455 2013-06-01  Jan Hubicka  <jh@suse.cz>
15457         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15458         (lto_symtab_merge_symbols_1): ... this one.
15459         (lto_symtab_merge_cgraph_nodes): Rename to ...
15460         (lto_symtab_merge_symbols): ... this one; simplify.
15461         * cgraph.c (same_body_aliases_done): Rename to ...
15462         (cpp_implicit_aliases_done): ... this one.
15463         (cgraph_create_function_alias): Update.
15464         (cgraph_same_body_alias): Update.
15465         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15466         (verify_edge_corresponds_to_fndecl): Simplify.
15467         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15468         (cgraph_node): Remove same_body_alias.
15469         (varpool_node): Remove alias_of and extra_name_alias.
15470         (same_body_aliases_done): Rename to ..
15471         (cpp_implicit_aliases_done): ... this one.
15472         (symtab_alias_ultimate_target): Add default parameter.
15473         (symtab_resolve_alias): New function.
15474         (fixup_same_cpp_alias_visibility): Declare.
15475         (cgraph_function_node): Add default parameter.
15476         (cgraph_node_asm_name): Likewise.
15477         (cgraph_function_or_thunk_node): Add default parameter; do
15478         not ICE when it is NULL.
15479         (varpool_variable_node): Likewise.
15480         * tree-emutls.c (create_emultls_var): Update.
15481         (ipa_lower_emutls): Update.
15482         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15483         (cgraph_reset_node): Reset alias info.
15484         (cgraph_finalize_function): Update.
15485         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15486         (analyze_function): Simplify.
15487         (cgraph_process_same_body_aliases): Simplify.
15488         (analyze_functions): Fixup same body aliases.
15489         (handle_alias_pairs): Simplify.
15490         (assemble_thunk): Update.
15491         (assemble_thunks_and_aliases): Update.
15492         (output_weakrefs): Rewrite.
15493         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15494         (lto_output_varpool_node): Likewise.
15495         (compute_ltrans_boundary): Remve assert.
15496         (get_alias_symbol): New functoin.
15497         (input_node): Rewrite alias handling.
15498         (input_varpool_node): Likewise.
15499         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15500         * ipa.c (process_references): Handle weakrefs correctly.
15501         (symtab_remove_unreachable_nodes): Likewise.
15502         * trans-mem.c (get_cg_data): Update.
15503         (ipa_tm_create_version_alias): Update.
15504         (ipa_tm_execute): Update.
15505         * symtab.c (dump_symtab_base): Dump aliases.
15506         (verify_symtab_base): Verify aliases.
15507         (symtab_node_availability): New function.
15508         (symtab_alias_ultimate_target): Simplify.
15509         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15510         handle all the fixup cases.
15511         (symtab_resolve_alias): New function.
15512         * passes.c (ipa_write_summaries): Handle weakrefs.
15513         * varpool.c (varpool_analyze_node): Simplify.
15514         (assemble_aliases): Update.
15515         (varpool_create_variable_alias): Simplify.
15516         (varpool_extra_name_alias): Simplify.
15517         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15518         (lto_symtab_merge_symbols): ... this one.
15520 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15522         Revert
15523         PR rtl-optimization/57268
15524         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15525         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15527 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15529         Partially reverted:
15530         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15532         PR middle-end/57073
15533         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15534         further up.
15536 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15538         PR rtl-optimization/57268
15539         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15540         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15542 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15544         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15545         unordered comparison operators when -fno-trapping-math is in effect
15546         on the e500.
15547         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15548         and implement unordered comparison operators properly on the e500.
15550 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15552         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15553         for constant scalar integers.
15554         (simplify_relational_operation_1): Likewise.
15556 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15558         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15559         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15560         Fix comment.
15562 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15563             Igor Zamyatin  <igor.zamyatin@intel.com>
15565         Silvermont (SLM) architecture performance tuning.
15566         * config/i386/i386.h (enum ix86_tune_indices): Add
15567         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15568         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15570         * config/i386/i386.c (initial_ix86_tune_features)
15571         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15572         (ix86_lea_outperforms): Handle Silvermont tuning.
15573         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15574         call.
15575         (ix86_use_lea_for_mov): Likewise.
15576         (ix86_avoid_lea_for_addr): Likewise.
15577         (ix86_lea_for_add_ok): Likewise.
15578         (exact_dependency_1): New function.
15579         (exact_store_load_dependency): Likewise.
15580         (ix86_adjust_cost): Handle Silvermont tuning.
15581         (do_reoder_for_imul): Likewise.
15582         (swap_top_of_ready_list): New function.
15583         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15585         * config/i386/i386.md (peepholes that split memory operand in fp
15586         converts): New.
15588 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15590         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15591         Remove un-necessary braces.
15593 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15595         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15596         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15598 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15600         PR middle-end/57073
15601         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15603 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15605         PR target/56315
15606         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15607         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15608         * config/arm/neon.md (iordi3_neon): Remove.
15609         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15610         * config/arm/predicates.md (imm_for_neon_logic_operand):
15611         Move to earlier in the file.
15612         (neon_logic_op2): Likewise.
15613         (arm_iordi_operand_neon): New predicate.
15615 2013-05-31  Richard Biener  <rguenther@suse.de>
15617         PR tree-optimization/57478
15618         PR tree-optimization/57453
15619         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15620         are life as well.
15622 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15624         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15625         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15627 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15628             Thomas Koenig  <tkoenig@gcc.gnu.org>
15630         PR middle-end/57073
15631         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15632         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15634 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15636         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15638 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15640         * target.def (register_usage_leveling_p): New hook.
15641         * targhooks.c (default_register_usage_leveling_p): New.
15642         * targhooks.h (default_register_usage_leveling_p): New prototype.
15643         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15644         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15645         * doc/tm.texi: Update.
15646         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15648 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15650         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15651         (*insv_reg<mode>): New define_insn.
15653 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15655         PR rtl-optimization/57439
15656         * postreload.c (move2add_valid_value_p): Check that we have
15657         a zero subreg_regno_offset when accessing the register in
15658         the requested mode.
15660 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15661             Igor Zamyatin  <igor.zamyatin@intel.com>
15663         Silvermont (SLM) architecture pipeline model, tuning and
15664         insn selection.
15665         * config.gcc: Add slm config options and target.
15667         * config/i386/slm.md: New.
15669         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15671         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15672         PROCESSOR_SLM.
15673         (ix86_target_macros_internal): Likewise.
15675         * config/i386/i386.c (slm_cost): New cost.
15676         (m_SLM): New macro flag.
15677         (initial_ix86_tune_features): Set m_SLM.
15678         (x86_accumulate_outgoing_args): Likewise.
15679         (x86_arch_always_fancy_math_387): Likewise.
15680         (processor_target_table): Add slm cost.
15681         (cpu_names): Add slm cpu name.
15682         (x86_option_override_internal): Set SLM ISA.
15683         (ix86_issue_rate): New case PROCESSOR_SLM.
15684         (ia32_multipass_dfa_lookahead): Likewise.
15685         (fold_builtin_cpu): Add slm.
15687         * config/i386/i386.h (TARGET_SLM): New target macro.
15688         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15689         (processor_type): Add PROCESSOR_SLM.
15691         * config/i386/i386.md (cpu): Add new value "slm".
15692         (slm.md): Include slm.md.
15694 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15695             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15697         * config/arm/arm-protos.h: Add and update function protos.
15698         * config/arm/arm.c (use_simple_return_p): New added.
15699         (thumb2_expand_return): Check simple_return flag.
15700         * config/arm/arm.md: Add simple_return and conditional simple_return.
15701         * config/arm/iterators.md: Add iterator for return and simple_return.
15703 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15705         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15706         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15707         (arm_emit_vfp_multi_reg_pop): Likewise.
15708         (thumb2_emit_ldrd_pop): Likewise.
15709         (arm_expand_epilogue): Add misc REG_CFA notes.
15710         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15712 2013-05-29  Lawrence Crowl  <crowl@google.com>
15714         * config/arm/t-arm: Update for below.
15716         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15717         Change type to hash_table.  Update dependent calls and types.
15719         * config/i386/t-cygming: Update for below.
15721         * config/i386/t-interix: Update for below.
15723         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15724         Change type to hash_table.  Update dependent calls and types.
15725         (i386_find_on_wrapper_list::wrappers): Likewise.
15727         * config/ia64/t-ia64: Update for below.
15729         * config/ia64/ia64.c (bundle_state_table):
15730         Change type to hash_table.  Update dependent calls and types.
15732         * config/mips/mips.c (mips_reorg_process_insns::htab):
15733         Change type to hash_table.  Update dependent calls and types.
15735         * config/sol2.c (solaris_comdat_htab):
15736         Change type to hash_table.  Update dependent calls and types.
15738         * config/t-sol2: Update for above.
15740 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15742         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15743         functions are not yet marked as defined.
15745 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15746             Pat Haugen  <pthaugen@us.ibm.com>
15747             Peter Bergner  <bergner@vnet.ibm.com>
15749         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15750         instructions.
15751         (VEC_A): Likewise.
15752         (VEC_C): Likewise.
15753         (vrotl<mode>3): Likewise.
15754         (vashl<mode>3): Likewise.
15755         (vlshr<mode>3): Likewise.
15756         (vashr<mode>3): Likewise.
15758         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15759         support for power8 V2DI builtins.
15761         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15762         power8 V2DI builtins.
15763         (vupkhsw): Likewise.
15764         (vupklsw): Likewise.
15765         (vaddudm): Likewise.
15766         (vminsd): Likewise.
15767         (vmaxsd): Likewise.
15768         (vminud): Likewise.
15769         (vmaxud): Likewise.
15770         (vpkudum): Likewise.
15771         (vpksdss): Likewise.
15772         (vpkudus): Likewise.
15773         (vpksdus): Likewise.
15774         (vrld): Likewise.
15775         (vsld): Likewise.
15776         (vsrd): Likewise.
15777         (vsrad): Likewise.
15778         (vsubudm): Likewise.
15779         (vcmpequd): Likewise.
15780         (vcmpgtsd): Likewise.
15781         (vcmpgtud): Likewise.
15782         (vcmpequd_p): Likewise.
15783         (vcmpgtsd_p): Likewise.
15784         (vcmpgtud_p): Likewise.
15785         (vupkhsw): Likewise.
15786         (vupklsw): Likewise.
15787         (vaddudm): Likewise.
15788         (vmaxsd): Likewise.
15789         (vmaxud): Likewise.
15790         (vminsd): Likewise.
15791         (vminud): Likewise.
15792         (vpksdss): Likewise.
15793         (vpksdus): Likewise.
15794         (vpkudum): Likewise.
15795         (vpkudus): Likewise.
15796         (vrld): Likewise.
15797         (vsld): Likewise.
15798         (vsrad): Likewise.
15799         (vsrd): Likewise.
15800         (vsubudm): Likewise.
15802         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15803         support for power8 V2DI instructions.
15805         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15806         power8 V2DI instructions.  Combine pack and unpack insns to use an
15807         iterator for each mode.  Check whether a particular mode supports
15808         Altivec instructions instead of just checking TARGET_ALTIVEC.
15809         (UNSPEC_VPKUWUM): Likewise.
15810         (UNSPEC_VPKSHSS): Likewise.
15811         (UNSPEC_VPKSWSS): Likewise.
15812         (UNSPEC_VPKUHUS): Likewise.
15813         (UNSPEC_VPKSHUS): Likewise.
15814         (UNSPEC_VPKUWUS): Likewise.
15815         (UNSPEC_VPKSWUS): Likewise.
15816         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15817         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15818         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15819         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15820         (UNSPEC_VUPKHSB): Likewise.
15821         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15822         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15823         (UNSPEC_VUPKHSH): Likewise.
15824         (UNSPEC_VUPKLSB): Likewise.
15825         (UNSPEC_VUPKLSH): Likewise.
15826         (VI2): Likewise.
15827         (VI_char): Likewise.
15828         (VI_scalar): Likewise.
15829         (VI_unit): Likewise.
15830         (VP): Likewise.
15831         (VP_small): Likewise.
15832         (VP_small_lc): Likewise.
15833         (VU_char): Likewise.
15834         (add<mode>3): Likewise.
15835         (altivec_vaddcuw): Likewise.
15836         (altivec_vaddu<VI_char>s): Likewise.
15837         (altivec_vadds<VI_char>s): Likewise.
15838         (sub<mode>3): Likewise.
15839         (altivec_vsubcuw): Likewise.
15840         (altivec_vsubu<VI_char>s): Likewise.
15841         (altivec_vsubs<VI_char>s): Likewise.
15842         (altivec_vavgs<VI_char>): Likewise.
15843         (altivec_vcmpbfp): Likewise.
15844         (altivec_eq<mode>): Likewise.
15845         (altivec_gt<mode>): Likewise.
15846         (altivec_gtu<mode>): Likewise.
15847         (umax<mode>3): Likewise.
15848         (smax<mode>3): Likewise.
15849         (umin<mode>3): Likewise.
15850         (smin<mode>3): Likewise.
15851         (altivec_vpkuhum): Likewise.
15852         (altivec_vpkuwum): Likewise.
15853         (altivec_vpkshss): Likewise.
15854         (altivec_vpkswss): Likewise.
15855         (altivec_vpkuhus): Likewise.
15856         (altivec_vpkshus): Likewise.
15857         (altivec_vpkuwus): Likewise.
15858         (altivec_vpkswus): Likewise.
15859         (altivec_vpks<VI_char>ss): Likewise.
15860         (altivec_vpks<VI_char>us): Likewise.
15861         (altivec_vpku<VI_char>us): Likewise.
15862         (altivec_vpku<VI_char>um): Likewise.
15863         (altivec_vrl<VI_char>): Likewise.
15864         (altivec_vsl<VI_char>): Likewise.
15865         (altivec_vsr<VI_char>): Likewise.
15866         (altivec_vsra<VI_char>): Likewise.
15867         (altivec_vsldoi_<mode>): Likewise.
15868         (altivec_vupkhsb): Likewise.
15869         (altivec_vupkhs<VU_char>): Likewise.
15870         (altivec_vupkls<VU_char>): Likewise.
15871         (altivec_vupkhsh): Likewise.
15872         (altivec_vupklsb): Likewise.
15873         (altivec_vupklsh): Likewise.
15874         (altivec_vcmpequ<VI_char>_p): Likewise.
15875         (altivec_vcmpgts<VI_char>_p): Likewise.
15876         (altivec_vcmpgtu<VI_char>_p): Likewise.
15877         (abs<mode>2): Likewise.
15878         (vec_unpacks_hi_v16qi): Likewise.
15879         (vec_unpacks_hi_v8hi): Likewise.
15880         (vec_unpacks_lo_v16qi): Likewise.
15881         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15882         (vec_unpacks_lo_v8hi): Likewise.
15883         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15884         (vec_pack_trunc_v8h): Likewise.
15885         (vec_pack_trunc_v4si): Likewise.
15886         (vec_pack_trunc_<mode>): Likewise.
15888         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15889         V2DI builtins.
15890         (vec_vmaxsd): Likewise.
15891         (vec_vmaxud): Likewise.
15892         (vec_vminsd): Likewise.
15893         (vec_vminud): Likewise.
15894         (vec_vpksdss): Likewise.
15895         (vec_vpksdus): Likewise.
15896         (vec_vpkudum): Likewise.
15897         (vec_vpkudus): Likewise.
15898         (vec_vrld): Likewise.
15899         (vec_vsld): Likewise.
15900         (vec_vsrad): Likewise.
15901         (vec_vsrd): Likewise.
15902         (vec_vsubudm): Likewise.
15903         (vec_vupkhsw): Likewise.
15904         (vec_vupklsw): Likewise.
15906 2013-05-29  Jan Hubicka  <jh@suse.cz>
15908         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15909         flags; reorder rest of fields in more consistent way.
15910         (varpool_node): Remove analyzed, finalized and alias.
15911         (cgraph_ndoe): Likewise.
15912         (symtab_alias_ultimate_target): New function.
15913         (cgraph_function_node): Move offline.
15914         (cgraph_reset_node): Declare.
15915         (cgraph_comdat_can_be_unshared_p): Remove.
15916         (varpool_remove_initializer): Declare.
15917         (varpool_first_defined_variable, varpool_next_defined_variable
15918         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15919         (cgraph_function_with_gimple_body_p): Update.
15920         (varpool_all_refs_explicit_p): Update.
15921         (symtab_alias_target): New function.
15922         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15923         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15924         (cgraph_function_or_thunk_node): Simplify using
15925         symtab_alias_ultimate_target.
15926         (varpool_variable_node): Likewise.
15927         * cgraph.c (cgraph_create_function_alias): Update.
15928         (cgraph_add_thunk): Update.
15929         (cgraph_remove_node): Update.
15930         (dump_cgraph_node): Do not dump removed flags.
15931         (cgraph_function_body_availability): Update.
15932         (cgraph_propagate_frequency): Update.
15933         (verify_cgraph_node): Check sanity of local flag.
15934         (cgraph_function_node): Move here from cgraph.h; revamp for
15935         cgraph_function_or_thunk_node.
15936         * lto-symtab.c (lto_varpool_replace_node): Update.
15937         (lto_symtab_resolve_can_prevail_p): Update.
15938         (lto_symtab_merge_cgraph_nodes): Update.
15939         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15940         propagate_constants_accross_call, devirtualization_time_bonus,
15941         ipcp_propagate_stage): Update.
15942         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
15943         * ipa-inline-transform.c (clone_inlined_nodes,
15944         preserve_function_body_p): Update.
15945         * ipa-reference.c (propagate): Update.
15946         (write_node_summary_p): Update.
15947         * toplev.c (wrapup_global_declaration_2): Update.
15948         * cgraphunit.c (cgraph_analyze_function): Rename to ...
15949         (analyze_function) ... this one.
15950         (cgraph_process_new_functions): Update.
15951         (cgraph_reset_node): Export.
15952         (cgraph_finalize_function): Update.
15953         (cgraph_add_new_function): Update.
15954         (process_function_and_variable_attributes): Update.
15955         (varpool_finalize_decl): Update.
15956         (symbol_finalized): Remove.
15957         (symbol_finalized_and_needed): Rename to ...
15958         (symbol_defined_and_needed): ... update.
15959         (cgraph_analyze_functions): Update.
15960         (handle_alias_pairs): Update.
15961         (mark_functions_to_output): Update.
15962         (assemble_thunk): Update.
15963         (output_in_order): Update.
15964         (output_weakrefs): Update.
15965         (finalize_compilation_unit): Update.
15966         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
15967         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
15968         input_node, input_varpool_node): Update.
15969         * dbxout.c (dbxout_expand_expr): Update.
15970         * cgraphclones.c (cgraph_clone_node): Update.
15971         (cgraph_copy_node_for_versioning): Update.
15972         (cgraph_materialize_clone): Update.
15973         (cgraph_materialize_all_clones): Update.
15974         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
15975         propagate_pure_const, propagate_nothrow): Update.
15976         * lto-streamer-out.c (lto_output, write_symbol): Update.
15977         * ipa-utils.c (ipa_reverse_postorder): Update.
15978         * ipa-inline.c (can_inline_edge_p): Update.
15979         (update_caller_keys, ipa_inline): Update.
15980         * dwarf2out.c (reference_to_unused,
15981         premark_types_used_by_global_vars_helper): Update.
15982         * tree-eh.c (tree_could_trap_p): Update.
15983         * ipa-split.c (consider_split, execute_split_functions): Update.
15984         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
15985         has_addr_references_p): Update; move ahead in file for better
15986         readability.
15987         (process_references): Simplify.
15988         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
15989         bodies are removed.
15990         (cgraph_comdat_can_be_unshared_p): Make static.
15991         (cgraph_externally_visible_p): Update.
15992         (varpool_externally_visible_p): Update.
15993         (function_and_variable_visibility): Update.
15994         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
15995         ipa_tm_mark_force_output_node): Update.
15996         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
15997         estimate_edge_devirt_benefit, inline_generate_summary,
15998         inline_write_summary): Update.
15999         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16000         * ipa-prop.c (ipa_compute_jump_functions): Update.
16001         (ipa_print_node_params, ipa_prop_read_section,
16002         ipa_update_after_lto_read, read_replacements_section): Update.
16003         * varasm.c (mark_decl_referenced): Update.
16004         (assemble_alias, dump_tm_clone_pairs): Update.
16005         * tree-inline.c (copy_bb): Update.
16006         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16007         Update.
16008         * symtab.c (dump_symtab_base): Print new flags.
16009         (verify_symtab_base): Verify new flags.
16010         (symtab_alias_ultimate_target): New function.
16011         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16012         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16013         Update.
16014         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16015         Update.
16016         * i386.c (ix86_get_function_versions_dispatcher,
16017         ix86_generate_version_dispatcher_body): Update.
16018         (fold_builtin_cpu): Use varpool_add_new_variable.
16019         * varpool.c (varpool_remove_initializer): Break out from ...
16020         (varpool_remove_node): ... this one.
16021         (dump_varpool_node, varpool_node_for_asm,
16022         cgraph_variable_initializer_availability, varpool_analyze_node,
16023         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16024         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16025         Update.
16027 2013-05-29  Jan Hubicka  <jh@suse.cz>
16029         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16031 2013-05-29  Easwaran Raman  <eraman@google.com>
16033         PR tree-optimization/57442
16034         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16035         when control exits the main loop.
16037 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16039         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16040         and RX600.
16041         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16042         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16043         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16044         option.
16046 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16048         PR tree-optimization/57441
16049         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16050         Don't limit size of incr_vec to number of candidates.
16052 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16054         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16055         and mips16 directories.
16056         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16057         (MULTILIB_DIRNAMES): Ditto.
16058         (MULTILIB_EXCEPTIONS): Add new exceptions.
16059         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16060         (MULTILIB_DIRNAMES): Ditto.
16061         (MULTILIB_EXCEPTIONS): Add new exceptions.
16063 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16064             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16066         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16067         SYMBOL_TINY_ABSOLUTE.
16068         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16069         SYMBOL_TINY_ABSOLUTE.
16070         (aarch64_expand_mov_immediate): Likewise.
16071         (aarch64_classify_symbol): Likewise.
16072         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16073         Permit SYMBOL_TINY_ABSOLUTE.
16074         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16076 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16077             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16079         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16080         Refactor if/switch.  Replace gcc_assert with if.
16082 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16084         * config/i386/i386.c (initial_ix86_tune_features): Enable
16085         FP Reassociation for AMD bdver1 and bdver2.
16087 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16089         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16090         and IMAGPART_EXPR do not occur within other handled_components.
16092 2013-05-29  Richard Biener  <rguenther@suse.de>
16094         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16095         access on whether the use is in the BB we currently try to
16096         vectorize.
16097         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16098         vectorize to vect_bb_slp_scalar_cost.
16100 2013-05-29  Richard Biener  <rguenther@suse.de>
16102         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16103         computing scalar cost offsetted by stmts that are kept live
16104         by scalar uses.
16105         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16106         for computation of scalar cost.
16108 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16110         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16112 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16114         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16115         array notation built-in reduction functions.
16116         * doc/passes.texi (Passes): Added documentation about changes done
16117         for Cilk Plus.
16118         * doc/invoke.texi (C Dialect Options): Added documentation about
16119         the -fcilkplus flag.
16120         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16121         (BUILTINS_DEF): Depend on cilkplus.def.
16122         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16123         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16124         * cilkplus.def: New file.
16126 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16128         PR rtl-optimization/57439
16129         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16131 2013-05-28  Easwaran Raman  <eraman@google.com>
16133         PR tree-optimization/57337
16134         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16135         (find_insert_point): Correctly identify the insertion point
16136         when two statements with the same UID is compared.
16138 2013-05-28  Richard Biener  <rguenther@suse.de>
16140         PR tree-optimization/56787
16141         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16142         from the list of data references.
16143         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16144         clobbers.
16145         (vect_analyze_loop_operations): Likewise.
16146         (vect_transform_loop): Remove clobbers.
16148 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16150         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16151         and REALPART_EXPRs have scalar type.
16153 2013-05-28  Richard Biener  <rguenther@suse.de>
16155         PR tree-optimization/57411
16156         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16157         virtual operands.
16158         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16159         virtual operand propagation.
16161 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16163         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16164         destination register for bmasksi_vis.
16165         (vector_init_bshuffle): Likewise.
16166         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16168 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16170         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16171         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16172         mode if the instruction isn't available in the original mode.
16173         * config/sparc/sparc.opt (mfix-ut699): New option.
16174         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16175         (divdf3): Turn into expander.
16176         (divdf3_nofix): New insn.
16177         (divdf3_fix): Likewise.
16178         (divsf3): Disable if -mfix-ut699.
16179         (sqrtdf2): Turn into expander.
16180         (sqrtdf2_nofix): New insn.
16181         (sqrtdf2_fix): Likewise.
16182         (sqrtsf2): Disable if -mfix-ut699.
16184 2013-05-27  Richard Biener  <rguenther@suse.de>
16186         PR middle-end/57412
16187         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16188         block for the new loop.
16190 2013-05-27  Richard Biener  <rguenther@suse.de>
16192         PR tree-optimization/57343
16193         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16194         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16195         (number_of_iterations_cond): Do not build the folded tree.
16197 2013-05-27  Richard Biener  <rguenther@suse.de>
16199         Revert
16200         PR middle-end/57381
16201         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16202         OEP_CONSTANT_ADDRESS_OF retained.
16204         PR tree-optimization/57417
16205         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16206         for unchanged base.
16207         (set_ssa_val_to): Compare addresses using
16208         get_addr_base_and_unit_offset.
16210 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16212         PR rtl-optimization/56833
16213         * postreload.c (move2add_record_mode): New function.
16214         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16215         (move2add_use_add2_insn): Use move2add_record_sym_value.
16216         (move2add_use_add3_insn): Likewise.
16217         (reload_cse_move2add): Use move2add_valid_value_p and
16218         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16219         affected regs by setting reg_mode to VOIDmode.
16220         (move2add_note_store): Don't pretend the inside of a SUBREG is
16221         the actual destination.  Invalidate single/leading registers by
16222         setting reg_mode to VOIDmode.
16223         Use move2add_record_sym_value, move2add_valid_value_p and
16224         move2add_record_mode.
16226 2013-05-27  Richard Biener  <rguenther@suse.de>
16228         PR tree-optimization/57396
16229         * tree-affine.c (double_int_constant_multiple_p): Properly
16230         return false for val == 0 and div != 0.
16232 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16234         * config/mips/mips.h: Use #elif in preprocessor conditions.
16236 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16238         PR target/53916
16239         * config/mips/constraints.md (kl): New constraint.
16240         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16241         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16242         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16243         from LO for MIPS16.
16244         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16246 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16248         PR target/55777
16249         * config/mips/mips.c (mips_can_inline_p): New function.
16250         (TARGET_CAN_INLINE_P): Define.
16252 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16254         * sched-int.h (ds_t, dw_t): Make unsigned int.
16255         Fix documentation that describes how all the ds_t bits are used.
16256         Reserve the last bit for delayed-branch scheduling.
16257         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16258         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16259         (gen_dep_weak_1): Remove prototype.
16260         * sched-deps.c (get_dep_weak_1): Make static.
16261         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16262         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16263         * doc/tm.texi: Regenerate.
16264         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16266 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16268         PR debug/56950
16269         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16271 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16272             Sandra Loosemore  <sandra@codesourcery.com>
16274         * config.gcc (powerpc-*): Allow native for with-cpu.
16276 2013-05-24  Jeff Law  <law@redhat.com>
16278         PR tree-optimization/57124
16279         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16280         conversion feeding a condition if the range has an overflow
16281         if -fstrict-overflow.  Add warnings for when we do make the
16282         transformation.
16284 2013-05-24  Dehao Chen  <dehao@google.com>
16286         * tree-cfg.c (locus_discrim_map): Fix the typo.
16287         (locus_discrim_hasher): Likewise.
16288         (locus_discrim_hasher::hash): Likewise.
16289         (locus_discrim_hasher::equal): Likewise.
16291 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16293         PR tree-optimization/57294
16294         * cgraph.h (ipa_record_stmt_references): Declare.
16295         * cgraphbuild.c (ipa_record_stmt_references): New function.
16296         (build_cgraph_edges): Use ipa_record_stmt_references.
16297         (rebuild_cgraph_edges): Likewise.
16298         (cgraph_rebuild_references): Likewise.
16299         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16300         associated with the old statement and build references from the
16301         newly built statements.
16302         * ipa-ref.c (ipa_remove_stmt_references): New function.
16303         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16305 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16307         * lra-constraints.c (emit_spill_move): Use smaller mode for
16308         mem-mem moves.
16309         (check_and_process_move): Consider mem-reg moves for secondary
16310         too.
16311         (curr_insn_transform): Don't lose insns emitted before for
16312         secondary memory moves.
16313         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16314         reg set up in the current insn.
16316 2013-05-24  Dehao Chen  <dehao@google.com>
16318         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16319         hash function.
16320         (locus_descrim_hasher::equal): Likewise.
16321         (build_gimple_cfg): New discriminator assignment algorithm.
16322         (make_edges): Likewise.
16323         (next_discriminator_for_locus): Likewise.
16324         (same_line_p): Likewise.
16325         (assign_discriminators): Likewise.
16326         (make_cond_expr_edges): Likewise.
16327         (make_gimple_switch_edges): Likewise.
16328         (make_goto_expr_edges): Likewise.
16329         (make_gimple_asm_edges): Likewise.
16331 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16333         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16334         X format specifier to only display bottom 16 bits.
16335         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16336         immediate to match for operand 2, since it will be masked.
16338 2013-05-24  Richard Biener  <rguenther@suse.de>
16340         PR tree-optimization/57287
16341         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16342         all SSA names that occur in abnormal PHIs.
16344 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16346         PR tree-ssa/57385
16347         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16348         that index is not negative.
16350 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16352         PR rtl-optimization/55177
16353         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16354         (simplify_byte_swapping_operation): New.
16355         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16356         (simplify_relational_operation_1): Deal with BSWAP.
16358 2013-05-23  Richard Henderson  <rth@redhat.com>
16360         PR target/56742
16361         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16362         (ix86_reorg): Call it.
16364 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16366         PR target/57379
16367         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16368         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16369         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16371 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16373         PR debug/57351
16374         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16376 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16377             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16379         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16380         * config/aarch64/constraints.md (Usa): Remove.
16381         * doc/md.texi (AArch64 Usa): Remove.
16383 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16384             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16386         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16387         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16388         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16389         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16391 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16393         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16394         instruction (AdvSIMD).
16395         * config/aarch64/aarch64-builtins.c
16396         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16397         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16399 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16401         PR middle-end/57347
16402         * tree.h (contains_bitfld_component_ref_p): Declare.
16403         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16404         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16405         caller.
16406         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16407         not access a bit-field.  Assert all final offsets are byte-aligned.
16409 2013-05-23  Richard Biener  <rguenther@suse.de>
16411         PR tree-optimization/57380
16412         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16413         least one invariant or re-used load.
16414         * passes.c (init_optimization_passes): Move pass_phiprop before
16415         pass_forwprop.
16417 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16419         * config/aarch64/aarch64-simd.md
16420         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16422 2013-05-23  Richard Biener  <rguenther@suse.de>
16424         PR middle-end/57381
16425         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16426         OEP_CONSTANT_ADDRESS_OF retained.
16428 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16430         PR middle-end/57344
16431         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16432         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16434 2013-05-23  Richard Biener  <rguenther@suse.de>
16436         PR rtl-optimization/57341
16437         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16438         instead of true_dependence.
16440 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16442         * bb-reorder.c (branch_threshold): Make const.
16443         (exec_threshold): Ditto.
16445 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16446             Pat Haugen  <pthaugen@us.ibm.com>
16447             Peter Bergner  <bergner@vnet.ibm.com>
16449         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16450         documentation for the power8 crypto builtins.
16452         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16454         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16455         macros for defining power8 builtin functions.
16456         (BU_P8V_AV_2): Likewise.
16457         (BU_P8V_AV_P): Likewise.
16458         (BU_P8V_VSX_1): Likewise.
16459         (BU_P8V_OVERLOAD_1): Likewise.
16460         (BU_P8V_OVERLOAD_2): Likewise.
16461         (BU_CRYPTO_1): Likewise.
16462         (BU_CRYPTO_2): Likewise.
16463         (BU_CRYPTO_3): Likewise.
16464         (BU_CRYPTO_OVERLOAD_1): Likewise.
16465         (BU_CRYPTO_OVERLOAD_2): Likewise.
16466         (XSCVSPDP): Fix typo, point to the correct instruction.
16467         (VCIPHER): Add power8 crypto builtins.
16468         (VCIPHERLAST): Likewise.
16469         (VNCIPHER): Likewise.
16470         (VNCIPHERLAST): Likewise.
16471         (VPMSUMB): Likewise.
16472         (VPMSUMH): Likewise.
16473         (VPMSUMW): Likewise.
16474         (VPERMXOR_V2DI): Likewise.
16475         (VPERMXOR_V4SI: Likewise.
16476         (VPERMXOR_V8HI: Likewise.
16477         (VPERMXOR_V16QI: Likewise.
16478         (VSHASIGMAW): Likewise.
16479         (VSHASIGMAD): Likewise.
16480         (VPMSUM): Likewise.
16481         (VPERMXOR): Likewise.
16482         (VSHASIGMA): Likewise.
16484         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16485         __CRYPTO__ if the crypto instructions are available.
16486         (altivec_overloaded_builtins): Add support for overloaded power8
16487         builtins.
16489         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16490         support for power8 crypto builtins.
16491         (builtin_function_type): Likewise.
16492         (altivec_init_builtins): Add support for builtins that take vector
16493         long long (V2DI) arguments.
16495         * config/rs6000/crypto.md: New file, define power8 crypto
16496         instructions.
16498 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16499             Pat Haugen  <pthaugen@us.ibm.com>
16500             Peter Bergner  <bergner@vnet.ibm.com>
16502         * doc/invoke.texi (Option Summary): Add power8 options.
16503         (RS/6000 and PowerPC Options): Likewise.
16505         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16506         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16507         wm, wn, wr documentation.
16509         * config/rs6000/constraints.md (wm): New constraint for VSX
16510         registers if direct move instructions are enabled.
16511         (wn): New constraint for no registers.
16512         (wq): New constraint for quad word even GPR registers.
16513         (wr): New constraint if 64-bit instructions are enabled.
16514         (wv): New constraint if power8 vector instructions are enabled.
16515         (wQ): New constraint for quad word memory locations.
16517         * config/rs6000/predicates.md (const_0_to_15_operand): New
16518         constraint for 0..15 for crypto instructions.
16519         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16520         as GPR and floating point registers.
16521         (int_reg_operand): New predicate to match only GPR registers.
16522         (base_reg_operand): New predicate to match base registers.
16523         (quad_int_reg_operand): New predicate to match even GPR registers
16524         for quad memory operations.
16525         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16526         operations in both GPR and VSX registers.
16527         (quad_memory_operand): New predicate for quad memory operations.
16528         (reg_or_indexed_operand): New predicate for direct move support.
16530         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16531         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16532         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16533         (POWERPC_MASKS): Add power8 options.
16534         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16535         various options.
16537         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16538         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16540         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16541         (-mpower8-fusion): New power8 options.
16542         (-mpower8-fusion-sign): Likewise.
16543         (-mpower8-vector): Likewise.
16544         (-mcrypto): Likewise.
16545         (-mdirect-move): Likewise.
16546         (-mquad-memory): Likewise.
16548         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16549         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16550         registers.
16551         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16552         (rs6000_debug_vector_unit): Add p8_vector.
16553         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16554         definitions.  Also print fusion state.
16555         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16556         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16557         (rs6000_option_override_internal): Add support for power8.
16558         (rs6000_common_init_builtins): Add debugging for skipped builtins
16559         if -mdebug=builtin.
16560         (rs6000_adjust_cost): Add power8 support.
16561         (rs6000_issue_rate): Likewise.
16562         (insn_must_be_first_in_group): Likewise.
16563         (insn_must_be_last_in_group): Likewise.
16564         (force_new_group): Likewise.
16565         (rs6000_register_move_cost): Likewise.
16566         (rs6000_opt_masks): Likewise.
16568         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16569         power8 capable assembler, default to power7 options.
16570         (TARGET_DIRECT_MOVE): Likewise.
16571         (TARGET_CRYPTO): Likewise.
16572         (TARGET_P8_VECTOR): Likewise.
16573         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16574         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16575         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16576         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16577         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16578         (TARGET_XSCVDPSPN): Likewise.
16579         (TARGET_XSCVSPDPN): Likewsie.
16580         (TARGET_SYNC_HI_QI): Likewise.
16581         (TARGET_SYNC_TI): Likewise.
16582         (MASK_CRYPTO): Likewise.
16583         (MASK_DIRECT_MOVE): Likewise.
16584         (MASK_P8_FUSION): Likewise.
16585         (MASK_P8_VECTOR): Likewise.
16586         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16587         temporary used by some of the direct move instructions to get two FP
16588         temporary registers does not force creation of a stack frame.
16589         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16590         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16591         that any VSX registers are tieable, even if they are also an
16592         Altivec vector mode.
16593         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16594         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16595         (RS6000_BTM_CRYPTO): Likewise.
16596         (RS6000_BTM_COMMON): Likewise.
16598         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16599         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16600         (enum rs6000_vector): Add power8 vector support.
16602 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16604         PR target/19599
16605         PR target/57340
16606         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16607         (any_sibcall_could_use_r3): this and handle indirect calls.
16608         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16610 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16612         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16614 2013-05-22  Richard Biener  <rguenther@suse.de>
16616         PR middle-end/57349
16617         * profile.c (branch_prob): Do not split blocks that are
16618         abnormally receiving from ECF_RETURNS_TWICE functions.
16620 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16622         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16623         address mode.  Move pointer mode initialization to the same place.
16625 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16627         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16628         while it has any effect.
16630 2013-05-21  Easwaran Raman  <eraman@google.com>
16632         PR tree-optimization/57322
16633         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16634         UID of the statement added to the BB to be 1.
16636 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16638         PR tree-optimization/57331
16639         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16640         of conversion from pointer type to integral type with integer.
16642 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16644         PR lto/57289
16645         * ipa-prop.c (ipa_read_node_info): Process param_used and
16646         controlled_uses in the same order as when writing.
16648 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16650         PR plugins/56754
16651         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16653 2013-05-21  Richard Biener  <rguenther@suse.de>
16655         PR tree-optimization/57318
16656         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16657         estimate stmts with side-effects as likely eliminated.
16659 2013-05-21  Richard Biener  <rguenther@suse.de>
16661         PR tree-optimization/57330
16662         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16663         preserve the call stmts fntype.
16665 2013-05-21  Richard Biener  <rguenther@suse.de>
16667         PR tree-optimization/57303
16668         * tree-ssa-sink.c (statement_sink_location): Improve killing
16669         stmt detection and properly handle self-assignments.
16671 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16673         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16674         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16675         registers. Set register size out of the PARALLEL loop.
16677 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16679         PR target/56547
16680         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16681         (*fmasf4, *fmasf4_media): New insns.
16683 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16685         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16686         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16687         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16688         (mips_idiv_insns): Update the comments to say that the returned
16689         instruction counts are in units of BASE_INSN_LENGTH.
16690         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16691         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16692         using 2 rather than 4 as the length of indirect MIPS16 and
16693         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16694         length of a NOP.  Don't divide MIPS16 lengths by 2.
16695         (mips16_split_long_branches): Assume a branch is long if the
16696         length is greater than 4 rather than 8.
16697         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16698         rather than multiplying them by 2.  Multiply instruction counts
16699         by BASE_INSN_LENGTH rather than 4.
16700         (*jump_mips16, tls_get_tp_mips16_<mode>)
16701         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16703 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16705         * config/mips/mips.md (extended_mips16): Remove branch case.
16706         (length): Remove duplicated extended_mips16 test.
16708 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16710         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16712 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16714         * recog.h (Recog_data): Rename to...
16715         (recog_data_d): ...this.
16716         (recog_data): Update accordingly.
16717         * recog.c (recog_data): Likewise.
16718         * reload.c (save_recog_data): Likewise.
16719         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16720         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16722 2013-05-17  Julian Brown  <julian@codesourcery.com>
16724         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16725         found in a REG_EQUAL note, invalidate it.
16727 2013-05-17  Easwaran Raman  <eraman@google.com>
16729         * tree-ssa-reassoc.c (find_insert_point): New function.
16730         (insert_stmt_after): Likewise.
16731         (get_def_stmt): Likewise.
16732         (ensure_ops_are_available): Likewise.
16733         (not_dominated_by): Likewise.
16734         (rewrite_expr_tree): Do not move statements beyond what is
16735         necessary. Remove call to swap_ops_for_binary_stmt...
16736         (reassociate_bb): ... and move it here.
16737         (build_and_add_sum): Assign UIDs for new statements.
16738         (linearize_expr): Likewise.
16739         (do_reassoc): Renumber gimple statement UIDs.
16741 2013-05-17  Jan Hubicka  <jh@suse.cz>
16743         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16744         weakrefs.
16745         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16746         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16747         weakrefs.
16748         (output_weakrefs): Update.
16750 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16751             Martin Jambor  <mjambor@suse.cz>
16753         PR middle-end/57276
16754         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16755         value that corresponds to the given aggval is found in values vector.
16757 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16759         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16760         sse, sse2, sse3, ssse3 and sse4a flags to options.
16762 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16764         * gengtype-state.c: (s_expr_writer): New class, to handle
16765         prettifying of output layout of s-expressions.
16766         (state_writer): New class, to write out gtype.state.
16767         (state_written_type_count): Move this variable into member data of
16768         state_writer.
16769         (s_expr_writer::s_expr_writer): New code: constructor for new class
16770         (state_writer::state_writer(): ditto
16771         (s_expr_writer::write_new_line): New function
16772         (s_expr_writer::write_any_indent): ditto
16773         (s_expr_writer::begin_s_expr): ditto
16774         (s_expr_writer::end_s_expr): ditto
16775         (write_state_fileloc): convert to method of state_writer...
16776         (state_writer:: write_state_fileloc): ...and use methods of
16777         s_expr_writer to write indentation into the gtype.state output file
16778         to visually represent the hierarchical structure of the list
16779         structures
16780         (write_state_fields): ditto, renaming to...
16781         (state_writer::write_state_fields)
16782         (write_state_a_string): ditto, renaming to...
16783         (state_writer::write_state_a_string)
16784         (write_state_string_option): ditto, renaming to...
16785         (state_writer::write_state_string_option)
16786         (write_state_type_option): ditto, renaming to...
16787         (state_writer::write_state_type_option)
16788         (write_state_nested_option): ditto, renaming to...
16789         (state_writer::write_state_nested_option)
16790         (write_state_option): ditto, renaming to...
16791         (state_writer::write_state_option)
16792         (write_state_options): ditto, renaming to...
16793         (state_writer::write_state_options)
16794         (write_state_lang_bitmap): ditto, renaming to...
16795         (state_writer::write_state_lang_bitmap)
16796         (write_state_version): ditto, renaming to...
16797         (state_writer::write_state_version)
16798         (write_state_scalar_type): ditto, renaming to...
16799         (state_writer::write_state_scalar_type)
16800         (write_state_string_type): ditto, renaming to...
16801         (state_writer::write_state_string_type)
16802         (write_state_undefined_type): ditto, renaming to...
16803         (state_writer::write_state_undefined_type)
16804         (write_state_struct_union_type): ditto, renaming to...
16805         (state_writer::write_state_struct_union_type)
16806         (write_state_struct_type): ditto, renaming to...
16807         (state_writer::write_state_struct_type)
16808         (write_state_user_struct_type): ditto, renaming to...
16809         (state_writer::write_state_user_struct_type)
16810         (write_state_lang_struct_type): ditto, renaming to...
16811         (state_writer::write_state_lang_struct_type)
16812         (write_state_param_struct_type): ditto, renaming to...
16813         (state_writer::write_state_param_struct_type)
16814         (write_state_pointer_type): ditto, renaming to...
16815         (state_writer::write_state_pointer_type)
16816         (write_state_array_type): ditto, renaming to...
16817         (state_writer::write_state_array_type)
16818         (write_state_gc_used): ditto, renaming to...
16819         (state_writer::write_state_gc_used)
16820         (write_state_common_type_content): ditto, renaming to...
16821         (state_writer::write_state_common_type_content)
16822         (write_state_type): ditto, renaming to...
16823         (state_writer::write_state_type)
16824         (write_state_pair_list): ditto, renaming to...
16825         (state_writer::write_state_pair_list)
16826         (write_state_pair): ditto, renaming to...
16827         (state_writer::write_state_pair)
16828         (write_state_typedefs): ditto, renaming to...
16829         (state_writer::write_state_typedefs)
16830         (write_state_structures): ditto, renaming to...
16831         (state_writer::write_state_structures)
16832         (write_state_param_structs): ditto, renaming to...
16833         (state_writer::write_state_param_structs)
16834         (write_state_variables): ditto, renaming to...
16835         (state_writer::write_state_variables)
16836         (write_state_srcdir): ditto, renaming to...
16837         (state_writer::write_state_srcdir)
16838         (write_state_files_list): ditto, renaming to...
16839         (state_writer::write_state_files_list)
16840         (write_state_languages): ditto, renaming to...
16841         (state_writer::write_state_languages)
16842         (write_state): create a state_writer instance and use it when
16843         writing out the state file
16845 2013-05-17  Mike Stump  <mikestump@comcast.net>
16847         PR rtl-optimization/57304
16848         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16849         accessing DF_REF_REAL_LOC.
16851 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16853         PR rtl-optimization/57281
16854         PR rtl-optimization/57300
16855         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16856         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16857         what the other splitter did if the registers are dead.
16859 2013-05-17  Richard Biener  <rguenther@suse.de>
16861         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16862         MEM_REF offsets.
16864 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16866         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16867         linking.
16869 2013-05-17  Marek Polacek  <polacek@redhat.com>
16871         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16872         length when doing non-zero store of storing '\0' to '\0'.
16874 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16876         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16877         vect_external_def oprnd1 with loop_vinfo, try to emit
16878         optional cast, negation and and stmts on the loop preheader
16879         edge instead of into the pattern def seq.
16881         PR tree-optimization/57051
16882         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16883         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16885 2013-05-16  Nick Clifton  <nickc@redhat.com>
16887         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16888         (rl78_is_naked_func): New function.
16889         (rl78_expand_prologue): Skip prologue generation for naked functions.
16890         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16891         * doc/extend.texi (naked): Add RL78 to the list of processors
16892         that supports this attribute.
16894 2013-05-16  Jeff Law  <law@redhat.com>
16896         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16898 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16900         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16901         cache parameters using detect_caches_amd also for CYRIX,
16902         NSC and TM2 signatures.
16904 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16905             Dzianis Kahanovich  <mahatma@eu.by>
16907         PR target/45359
16908         PR target/46396
16909         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16910         VIA/Centaur processors and determine their cache parameters
16911         using detect_caches_amd.
16913 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16915         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16916         (rtl_verify_edges): New function.
16917         (rtl_verify_bb_insns): Ditto.
16918         (rtl_verify_bb_pointers): Ditto.
16919         (rtl_verify_bb_insn_chain): Ditto.
16920         (rtl_verify_fallthru): Ditto.
16921         (rtl_verify_bb_layout): Ditto.
16922         (rtl_verify_flow_info_1): Outline checks into new functions.
16923         (rtl_verify_flow_info): Ditto.
16925 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16927         * cfghooks.c (copy_bbs): Add update_dominance argument.
16928         * cfghooks.h (copy_bbs): Update prototype.
16929         * tree-cfg.c (gimple_duplicate_sese_region):
16930         Add update_dominance argument.
16931         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16932         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16933         gimple_duplicate_sese_region call.
16934         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16935         Update copy_bbs call.
16936         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16937         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16939 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16941         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16942         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
16943         vect_recog_rotate_pattern.
16944         (vect_recog_rotate_pattern): New function.
16946 2013-05-16  Jason Merrill  <jason@redhat.com>
16948         * Makefile.in (LLINKER): New variable.
16949         (mostlyclean): Remove link mutex.
16950         * configure.ac: Handle --enable-link-mutex.
16951         * lock-and-run.sh: New script.
16953 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16955         PR target/19599
16956         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
16957         for NULL decl.
16959 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16961         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
16963 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16965         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
16966         * config/arm/arm.c (next_consecutive_mem): New function.
16967         (gen_movmem_ldrd_strd): Likewise.
16968         * config/arm/arm.md (movmemqi): Update condition and code.
16969         (unaligned_loaddi, unaligned_storedi): New patterns.
16971 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16973         * config.gcc: Obsolete *-*-solaris2.9*.
16974         * doc/install.texi (Specific, *-*-solaris2*): Document it.
16976 2013-05-16  Richard Biener  <rguenther@suse.de>
16978         * passes.c (init_optimization_passes): Move pass_parallelize_loops
16979         earlier, after GRAPHITE transforms and IV canonicalization.
16981 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16983         * omp-low.c (extract_omp_for_data): For collapsed loops,
16984         if at least one of the loops is known at compile time to
16985         iterate zero times, set count to 0.
16986         (expand_omp_regimplify_p): New function.
16987         (expand_omp_for_generic): For collapsed loops, if at least
16988         one of the loops isn't known to iterate at least once,
16989         add runtime check with setting count to 0.
16990         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
16991         For unsigned types if it isn't known at compile time that
16992         the loop will iterate at least once, add runtime check to bypass
16993         the whole loop if initial condition isn't true.
16995 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
16997         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
16999 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17001         PR middle-end/57286
17002         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17003         transformations to avoid an infinite loop.
17005 2013-05-16  Marek Polacek  <polacek@redhat.com>
17007         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17009 2013-05-15  Leif Ekblad  <leif@rdos.net>
17011         * config/i386/i386.c (ix86_decompose_address): Use
17012         DEFAULT_TLS_SEG_REG to access TLS segment register.
17013         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17014         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17015         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17017 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17019         PR target/57260
17020         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17021         sibling calls to functions that would normally be lazily bound,
17022         unless $gp is call-clobbered.
17024 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17026         * config/i386/i386.c (ix86_option_override_internal): Update
17027         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17028         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17029         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17030         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17031         of TARGET_3DNOW.
17032         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17034 2013-05-15  Andreas Schwab  <schwab@suse.de>
17036         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17037         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17038         third operand.
17040 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17042         * loop-unroll.c (report_unroll_peel): Check decision before
17043         emitting unroll/peel message.
17045 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17047         * function.h (has_bb_partition): New rtl_data flag.
17048         (bb_reorder_complete): Ditto.
17049         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17050         instead of flag_reorder_blocks_and_partition.
17051         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17052         with some enhancements.
17053         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17054         * bb-reorder.c (connect_traces): Check for has_bb_partition
17055         instead of flag_reorder_blocks_and_partition.
17056         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17057         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17058         verify_hot_cold_block_grouping.
17059         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17061 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17063         PR target/19599
17064         * config/arm/predicates.md (call_insn_operand): New predicate.
17065         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17066         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17067         if insn is not a tail call.
17068         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17069         registers.
17070         * config/arm/arm.h (enum reg_class): New caller save register class.
17071         (REG_CLASS_NAMES): Likewise.
17072         (REG_CLASS_CONTENTS): Likewise.
17073         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17074         without decls.
17076 2013-05-15  Richard Biener  <rguenther@suse.de>
17078         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17079         of MSG_OPTIMIZED_LOCATIONS.
17080         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17081         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17082         message.
17083         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17084         of MSG_OPTIMIZED_LOCATIONS.
17085         (execute_vect_slp): Likewise.
17086         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17087         (vect_create_cond_for_alias_checks): Likewise.
17088         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17089         (vect_recog_widen_mult_pattern): Likewise.
17090         (vect_recog_widen_sum_pattern): Likewise.
17091         (vect_recog_over_widening_pattern): Likewise.
17092         (vect_recog_widen_shift_pattern): Likewise.
17093         (vect_recog_vector_vector_shift_pattern): Likewise.
17094         (vect_recog_divmod_pattern): Likewise.
17095         (vect_recog_mixed_size_cond_pattern): Likewise.
17096         (vect_recog_bool_pattern): Likewise.
17097         (vect_pattern_recog_1): Likewise.
17099 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17101         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17102         non-functions to builtin_unreachable.
17103         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17104         correct when new direct edges were discovered.
17106 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17108         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17109         header, print symbol order instead of node uid, print more information
17110         about indirect edge targets.
17111         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17112         uids.
17113         (ipa_make_edge_direct_to_target): Likewise.
17114         (remove_described_reference): Likewise.
17115         (propagate_controlled_uses): Likewise.
17116         (ipa_print_node_params): Also print symbol order.
17117         (ipcp_transform_function): Print symbol order instead of node uids.
17118         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17119         (cgraph_get_create_real_symbol_node): Likewise.
17120         * ipa-cp.c (print_lattice): Likewise.
17121         (print_all_lattices): Likewise.
17122         (determine_versionability): Likewise.
17123         (initialize_node_lattices): Likewise.
17124         (estimate_local_effects): Likewise.
17125         (update_profiling_info): Likewise.
17126         (create_specialized_node): Likewise.
17127         (perhaps_add_new_callers): Likewise.
17128         (decide_about_value): Likewise.
17129         (decide_whether_version_node): Likewise.
17130         (identify_dead_nodes): Likewise.
17131         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17132         (dump_inline_summary): Likewise.
17133         (estimate_node_size_and_time): Likewise.
17134         (inline_analyze_function): Likewise.
17135         * ipa-inline.c (report_inline_failed_reason): Likewise.
17136         (want_early_inline_function_p): Likewise.
17137         (edge_badness): Likewise.
17138         (update_edge_key): Likewise.
17139         (inline_small_functions): Likewise.  Add dumping of order to two other
17140         dumps.
17141         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17142         instead of node uids.
17143         (propagate_pure_const): Likewise.
17144         (propagate_pure_const): Likewise.
17145         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17146         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17147         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17148         of node uids.
17149         * tree-pretty-print.c (dump_function_header): Likewise.
17150         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17151         Print symbol order instead of node uids.
17153 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17155         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17156         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17158 2013-05-15  Richard Biener  <rguenther@suse.de>
17160         PR tree-optimization/57275
17161         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17162         return value for fail to do runtime alias checks for gather loads.
17164 2013-05-15  Jan Hubicka  <jh@suse.cz>
17166         PR lto/57038
17167         PR lto/47375
17168         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17169         weakrefs are not external.
17170         (lto_symtab_merge_decls): Fix thinko when dealing with
17171         non-lto_symtab decls.
17172         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17173         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17174         * varpool.c (dump_varpool_node): Dump more flags.
17176 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17178         * config/i386/i386.c (processor_alias_table): Add instruction
17179         FSGSBASE for AMD bdver3 architecture.
17181 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17183         * tree.c (warn_deprecated_use): Print file:line using locus color.
17184         * diagnostic.c (diagnostic_report_current_module): Print file:line
17185         and file:line:column using locus color.
17187 2013-05-14  Mike Stump  <mikestump@comcast.net>
17189         * gdbinit.in: Add __null.
17191 2013-05-14  Mike Stump  <mikestump@comcast.net>
17193         * recog.h: Rename struct recog_data to Recog_data.
17194         * recog.c: Likewise.
17195         * reload.c (can_reload_into): Likewise.
17196         * config/picochip/picochip.c: Likewise.
17198 2013-05-14  Mike Stump  <mikestump@comcast.net>
17200         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17202 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17204         * resource.h (struct resources): Remove unch_memory member.
17205         (CLEAR_RESOURCE): Don't clear unch_memory.
17206         * resource.c (mark_referenced_resources): Don't set it.
17207         (mark_set_resources): Likewise.
17208         (mark_target_live_regs): Don't clear it.
17209         (init_resource_info): Likewise.
17210         * reorg.c (resource_conflicts_p): Don't compare it.
17211         (redundant_insn): Don't set it.
17213         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17214         Remove prototypes.
17215         * emit-rtl.c (next_label): Remove unused function.
17216         (skip_consecutive_labels, link_cc0_insns): Move to ...
17217         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17218         only place where these functions are used, and make them static.
17220 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17222         * fold-const.c (fold_negate_expr): Handle vectors.
17223         (fold_truth_not_expr): Make it static.
17224         (fold_invert_truthvalue): New static function.
17225         (invert_truthvalue_loc): Handle vectors. Do not call
17226         fold_truth_not_expr directly.
17227         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17228         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17229         (fold_comparison): Handle vector constants.
17230         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17231         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17232         * tree.h (fold_truth_not_expr): Remove declaration.
17234 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17236         * config/aarch64/aarch64-simd.md
17237         (aarch64_vcond_internal<mode>): Rename to...
17238         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17239         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17240         float modes. Clarify all iterator modes.
17241         (vcond<mode><mode>): Use new name for vcond expanders.
17242         (vcond<v_cmp_result><mode>): Likewise.
17243         (vcondu<mode><mode>: Likewise.
17244         * config/aarch64/iterators.md (VDQF_COND): New.
17246 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17248         PR bootstrap/57266
17249         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17250         variable for the shift amount. Check that we shift by non-negative
17251         amounts.
17253 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17255         PR target/42017
17256         * config/arm/arm.h (EPILOGUE_USES): Only return true
17257         for LR_REGNUM after epilogue_completed.
17259 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17261         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17262         is error_mark_node.
17264 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17266         PR target/57261
17267         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17268         and Solaris 11+/x86 with gld.
17269         * configure: Regenerate.
17271 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17273         * expmed.c (expand_shift_1): Canonicalize rotates by
17274         constant bitsize / 2 to bitsize - 1.
17275         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17276         case ROTATERT>: Likewise.
17278         Revert:
17279         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17281         * config/i386/i386.md (rotateinv): New code attr.
17282         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17283         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17284         roll $31, %eax, etc.
17286 2013-05-14  Richard Biener  <rguenther@suse.de>
17288         PR middle-end/57235
17289         * tree-eh.c (sink_clobbers): Give up for successors with
17290         multiple predecessors and no virtual uses.
17292 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17294         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17295         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17297 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17299         PR middle-end/57251
17300         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17301         the case when both op0 and op1 have VOIDmode.
17303 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17305         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17306         in multiply-accumulate mode.
17308 2013-05-13  Guozhi Wei  <carrot@google.com>
17310         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17312 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17314         PR target/56975
17315         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17316         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17317         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17318         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17319         (ix86_expand_prologue): Likewise.
17320         (ix86_expand_split_stack_prologue): Likewise.
17321         (legitimate_pic_address_disp_p): Likewise.
17322         (legitimize_pic_address): Likewise.
17323         (legitimize_tls_address): Likewise.
17324         (legitimize_pe_coff_symbol): Likewise.
17325         (output_pic_addr_const): Likewise.
17326         (construct_plt_address): Likewise.
17327         (ix86_expand_call): Likewise.
17328         (x86_output_mi_thunk): Likewise.
17329         (x86_function_profiler): Likewise.
17331 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17333         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17334         similar switch cases.
17335         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17336         (aarch64_simd_mov_to_<mode>low): Delete.
17337         (aarch64_simd_mov_to_<mode>high): Delete.
17338         (move_lo_quad_<mode>): Add w<-r alternative.
17339         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17340         (aarch64_simd_mov_from_*): Update type attribute.
17341         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17342         statement.
17344 2013-05-13  Jan Hubicka  <jh@suse.cz>
17346         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17347         * config/i386/i386.c (ix86_compute_frame_layout,
17348         ix86_expand_epilogue, emit_i387_cw_initialization,
17349         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17350         ix86_local_alignment): Fix use of size/speed predicates.
17352 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17354         PR tree-optimization/45216
17355         PR tree-optimization/57157
17356         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17357         the (-Y) & (B - 1) variant if OP is |.
17358         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17359         return shifted.  Use (-op1) & (prec - 1) as other_amount
17360         instead of prec - op1.
17362 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17364         PR middle-end/42371
17365         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17366         (ipa_constant_data): New type.
17367         (ipa_jump_func): Use ipa_constant_data to hold information about
17368         constant jump functions.
17369         (ipa_get_jf_constant): Adjust to jump function type changes.
17370         (ipa_get_jf_constant_rdesc): New function.
17371         (ipa_param_descriptor): New field controlled_uses.
17372         (ipa_get_controlled_uses): New function.
17373         (ipa_set_controlled_uses): Likewise.
17374         * ipa-ref.h (ipa_find_reference): Declare.
17375         * ipa-prop.c (ipa_cst_ref_desc): New type.
17376         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17377         changes.
17378         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17379         New parameter cs.  Adjust all callers.
17380         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17381         (remove_described_reference): New function.
17382         (jfunc_rdesc_usable): Likewise.
17383         (try_make_edge_direct_simple_call): Decrement controlled use count,
17384         attempt to remove reference if it hits zero.
17385         (combine_controlled_uses_counters): New function.
17386         (propagate_controlled_uses): Likewise.
17387         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17388         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17389         (ipa_print_node_params): Print described use counter.
17390         (ipa_write_jump_function): Adjust to jump function type changes.
17391         (ipa_read_jump_function): New parameter CS, pass it to
17392         ipa_set_jf_constant.  Adjust caller.
17393         (ipa_write_node_info): Stream controlled use count
17394         (ipa_read_node_info): Likewise.
17395         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17396         asserting.
17397         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17398         count.  Remove cloning-added reference if it reaches zero.
17399         * ipa-ref.c (ipa_find_reference): New function.
17401 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17403         * config/i386/i386.c (processor_target_table): Modified default
17404         alignment values for AMD BD and BT architectures.
17406 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17408         * tree-vect-generic.c (uniform_vector_p): Move ...
17409         * tree.c (uniform_vector_p): ... here.
17410         * tree.h (uniform_vector_p): Declare it.
17411         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17412         into a scalar.
17414 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17416         PR tree-optimization/57230
17417         * tree-ssa-strlen.c (handle_char_store): Record length for
17418         array store from STRING_CST.
17420         PR tree-optimization/57230
17421         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17422         check.
17424 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17426         * config/epiphany/epiphany.c (epiphany_init): Check size of
17427         NUM_MODES_FOR_MODE_SWITCHING.
17428         (epiphany_expand_prologue):
17429         Remove CONFIG_REGNUM initial value handling code.
17430         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17431         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17432         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17433         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17434         Don't return 1 for FP_MODE_NONE.
17435         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17436         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17437         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17438         * config/epiphany/epiphany.md (save_config): New pattern.
17440 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17442         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17444 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17446         * config/i386/i386.md (memory): Handle sseishft1.
17447         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17448         (*vec_extractv2di_1): Ditto.
17450 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17452         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17453         saved registers.
17455 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17457         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17458         Add mthumb/march=armv7-a multilib.
17459         Add mthumb/march=armv7-r multilib.
17460         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17462 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17464         * config/v850/t-rtems: Add more multilibs.
17466 2013-05-10  Richard Biener  <rguenther@suse.de>
17468         PR tree-optimization/57214
17469         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17470         not propagate from SSA names that occur in abnormal PHI nodes.
17472 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17474         * stor-layout.c (element_precision): New function.
17475         * machmode.h (element_precision): Declare it.
17476         * tree.c (build_minus_one_cst): New function.
17477         (element_precision): Likewise.
17478         * tree.h (build_minus_one_cst): Declare new function.
17479         (element_precision): Likewise.
17480         * fold-const.c (operand_equal_p): Use element_precision.
17481         (fold_binary_loc): Handle vector types.
17482         * convert.c (convert_to_integer): Use element_precision.
17483         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17484         separately.
17486 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17488         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17489         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17490         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17491         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17492         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17493         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17494         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17495         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17496         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17497         (Uuw8): New constraints.
17498         (Usb4): Move into alphabetical order.
17499         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17500         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17501         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17502         previously unnamed patterns.
17503         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17504         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17505         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17506         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17507         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17508         of specific lengths.
17510 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17512         * config/i386/i386.md (rotateinv): New code attr.
17513         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17514         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17515         roll $31, %eax, etc.
17517         PR tree-optimization/45216
17518         PR tree-optimization/57157
17519         * tree-ssa-forwprop.c (simplify_rotate): New function.
17520         (ssa_forward_propagate_and_combine): Call it.
17522 2013-05-10  Richard Biener  <rguenther@suse.de>
17524         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17525         disable peeling when we version for aliasing.
17526         (vector_alignment_reachable_p): Honor explicit user alignment.
17527         (vect_supportable_dr_alignment): Likewise.
17528         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17529         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17530         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17531         then peeling to arrange for the cost-model check to come first.
17533 2013-05-10  Alan Modra  <amodra@gmail.com>
17535         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17536         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17537         * configure: Regenerate.
17539 2013-05-10  Alan Modra  <amodra@gmail.com>
17541         PR target/55033
17542         * varasm.c (default_elf_select_section): Move !DECL_P check..
17543         (get_named_section): ..to here before calling get_section_name.
17544         Adjust assertion.
17545         (default_section_type_flags): Add DECL_P check.
17546         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17547         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17549 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17551         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17552         When using gen_stack_adjust_str with a register offset, add a
17553         REG_FRAME_RELATED_EXPR note.
17555 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17557         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17558         (*vec_extractv4si_zext_mem): Ditto.
17559         (*vec_extractv2di): Add 0->x and x->x alternatives.
17560         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17561         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17563 2013-05-09  Jason Merrill  <jason@redhat.com>
17565         N3639 C++1y VLA support
17566         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17567         DECL_VALUE_EXPR.
17569         * tree.c (build_constructor_va): New.
17570         * tree.h: Declare it.
17572 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17574         PR lto/57084
17575         * gimple-fold.c (canonicalize_constructor_val): Call
17576         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17578 2013-05-09  Jan Hubicka  <jh@suse.cz>
17579             Richard Biener  <rguenther@suse.de>
17581         PR lto/54095
17582         * symtab.c (symtab_make_decl_local): Do not add private names.
17584 2013-05-09  Jan Hubicka  <jh@suse.cz>
17586         PR lto/54095
17587         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17588         (unlink_from_assembler_name_hash): Likewise.
17589         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17590         symtab_unregister_node, symtab_initialize_asm_name_hash,
17591         change_decl_assembler_name): Update.
17593 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17595         * config/aarch64/aarch64.md: New movtf split.
17596         (*movtf_aarch64): Update.
17597         (aarch64_movdi_tilow): Handle TF modes and rename to
17598         aarch64_movdi_<mode>low.
17599         (aarch64_movdi_tihigh): Handle TF modes and rename to
17600         aarch64_movdi_<mode>high
17601         (aarch64_movtihigh_di): Handle TF modes and rename to
17602         aarch64_mov<mode>high_di
17603         (aarch64_movtilow_di): Handle TF modes and rename to
17604         aarch64_mov<mode>low_di
17605         (aarch64_movtilow_tilow): Remove spurious whitespace.
17606         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17607         splits.
17608         (aarch64_print_operand): Update.
17610 2013-05-09  Alan Modra  <amodra@gmail.com>
17612         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17613         powerpc64le.
17614         * configure: Regenerate.
17616 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17618         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17619         splitter preparation statements.
17620         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17621         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17622         adjust_address_nv.
17624 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17626         * gimple-ssa-strength-reduction.c (count_candidates): Change
17627         return value to int.
17628         (analyze_candidates_and_replace): Change type of length to int.
17630 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17632         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17633         (*vec_extract<mode>): Use VI12_128 mode iterator.
17634         (*vec_extract<mode>_mem): Ditto.
17635         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17636         attribute.
17638 2013-05-08  Diego Novillo  <dnovillo@google.com>
17640         PR bootstrap/54659
17642         Revert:
17643         2012-08-17  Diego Novillo  <dnovillo@google.com>
17645         PR bootstrap/54281
17646         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17647         * config.in: Regenerate.
17648         * configure: Regenerate.
17649         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17651 2013-05-08  Jan Hubicka  <jh@suse.cz>
17653         PR lto/54095
17654         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17655         * cgraph.h (symtab_node_base): Add unique_name.
17656         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17657         input_overwrite_node, input_varpool_node): Stream unique_name.
17658         * cgraphclones.c (cgraph_create_virtual_clone,
17659         cgraph_function_versioning): Set unique_name.
17660         * ipa.c (function_and_variable_visibility): Set unique_name.
17662 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17664         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17665         (alloc_cand_and_find_basis): Restrict conditional candidate
17666         processing to CAND_MULTs.
17668 2013-05-08  Jan Hubicka  <jh@suse.cz>
17670         PR lto/54095
17671         lto-symtab.c (lto_symtab_symbol_p): New function.
17672         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17673         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17674         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17675         Skip static symbols.
17677 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17679         PR tree-optimization/57200
17680         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17681         Only call inform if the preceding warning_at returns true.
17683 2013-05-07  Han Shen  <shenhan@google.com>
17685         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17686         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17687         * common.opt (fstack-protector-strong): New option.
17688         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17689         * doc/invoke.texi (Optimization Options): Document
17690         "-fstack-protector-strong".
17691         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17693 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17695         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17697 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17699         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17700         it is not lhs of the insn.
17701         (setup_reg_equiv): Remove insn having equiv memory which it is not
17702         lhs of the insn.
17703         * lra-constraints.c (process_address): Try to improve generation
17704         code for address base + disp.
17705         (lra_constraints): Make correct the code for checking insn setting
17706         up backward equivalence.  Remove insn only if it is in the init
17707         insn list.
17708         * lra-eliminations.c (update_reg_eliminate): Change return value.
17709         (lra_eliminate): Use the result.
17711 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17713         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17714         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17715         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17716         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17717         register target operands.
17718         (*vec_extractv8hi_sse2): New pattern.
17719         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17720         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17721         (*vec_extract<mode>_mem): New insn and split pattern.
17723 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17725         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17726         (TARGET_ASAN_SHADOW_OFFSET): Define.
17727         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17728         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17730 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17732         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17733         (incr_vec_index): Return -1 if increment not found.
17734         (create_add_on_incoming_edge): Assert if increment not found.
17735         (record_increment): Limit number of increments recorded.
17736         (all_phi_incrs_profitable): Return false if an increment not found.
17737         (replace_profitable_candidates): Don't process increments that were
17738         not recorded.
17739         (analyze_candidates_and_replace): Limit size of incr_vec.
17741 2013-05-07  Richard Biener  <rguenther@suse.de>
17743         * calls.c (special_function_p): setjmp-like functions are leaf.
17744         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17745         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17747 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17749         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17750         (aarch64_simd_mov<mode>): New expander.
17751         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17752         (aarch64_simd_mov_to_<mode>high): Likewise.
17753         (aarch64_simd_mov_from_<mode>low): Likewise.
17754         (aarch64_simd_mov_from_<mode>high): Likewise.
17755         (aarch64_dup_lane<mode>): Update.
17756         (aarch64_dup_lanedi): New instruction pattern.
17757         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17758         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17760 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17762         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17763         (replace_mult_candidate): Remove unnecessary argument; remove
17764         unnecessary parameter from call to introduce_cast_before_cand.
17765         (replace_unconditional_candidate): Remove unnecessary parameter
17766         from call to replace_mult_candidate.
17767         (replace_conditional_candidate): Likewise.
17768         (insert_initializers): Use make_temp_ssa_name.
17769         (introduce_cast_before_cand): Remove unnecessary argument; use
17770         make_temp_ssa_name.
17771         (replace_one_candidate): Remove unnecessary argument; remove
17772         unnecessary parameter from calls to introduce_cast_before_cand.
17773         (replace_profitable_candidates): Remove unnecessary parameters
17774         from calls to replace_one_candidate.
17776 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17778         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17779         phi def as possibly hiding a basis for a CAND_ADD whose operands
17780         have been commuted in the analysis.
17781         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17783 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17785         * config/aarch64/aarch64.md
17786         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17787         shift value between 0-4.
17789 2013-05-07  Richard Biener  <rguenther@suse.de>
17791         * double-int.h (rshift): New overload.
17792         * double-int.c (rshift): New function.
17793         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17794         (create_reference_ops_from_ref): Remove.
17795         (vn_reference_insert): Use shared ops for constructing the
17796         reference and copy it.
17798 2013-05-07  Richard Biener  <rguenther@suse.de>
17800         PR middle-end/57190
17801         * tree-eh.c (sink_clobbers): Properly propagate
17802         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17804 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17806         PR tree-optimization/57149
17807         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17808         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17809         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17810         uninit_undefined_value_p instead of ssa_undefined_value_p.
17812         PR debug/57184
17813         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17814         for modifier == EXPAND_INITIALIZER.
17816 2013-05-07  Anton Blanchard  <anton@samba.org>
17818         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17819         for powerpc64 little endian.
17820         * configure: Regenerate.
17822 2013-05-06  Graham Stott  <grahams@btinternet.com>
17824         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17825         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17826         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17827         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17829 2013-05-06  Graham Stott  <grahams@btinternet.com>
17831         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17832         codes which allow non-lvalues.
17834 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17836         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17837         components are all 1s.
17838         (integer_minus_onep): New function.
17839         * tree.h (integer_minus_onep): Declare it.
17840         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17841         integer_minus_onep instead of integer_all_onesp.
17843 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17845         PR target/52933
17846         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17847         variations of these patterns.
17849 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17851         * config/i386/i386.md (isa): Add x64_sse4 member.
17852         (enabled): Handle x64_sse4.
17853         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17854         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17855         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17856         instruction for SSE4_1 targets.  Update insn attributes.
17857         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17858         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17859         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17860         const_1 selector.
17861         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17862         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17863         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17865 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17867         PR target/57108
17868         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17870 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17872         * final.c (do_assembler_dialects): Don't handle curly braces and
17873         vertical bar escaped by % as dialect delimiters.
17874         (output_asm_insn): Print curly braces and vertical bar if escaped
17875         by % and ASSEMBLER_DIALECT defined.
17876         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17877         * doc/tm.texi: Regenerated.
17879 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17881         * config/mips/mips.c: Include tree-pass.h.
17882         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17883         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17884         (pass_mips_machine_reorg2): New machine specific pass.
17885         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17886         (mips_option_override): Register the new pass.
17887         * rtl.h (cleanup_barriers): Remove prototype.
17888         (dbr_schedule): Likewise.
17889         * jump.c (cleanup_barriers): Make static.
17890         * reorg.c (dbr_schedule): Likewise.
17892 2013-05-06  Richard Biener  <rguenther@suse.de>
17894         PR tree-optimization/57185
17895         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17896         SSA names properly.
17898 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17900         PR target/57106
17901         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17902         between "first_arg" and "insn", not "last" and "insn".
17904 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17906         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17907         (find_candidates_in_block): Re-enable slsr_process_phi.
17908         (create_phi_basis): Fix double counting of candidate adjustment.
17910 2013-05-06  Richard Biener  <rguenther@suse.de>
17912         PR middle-end/57147
17913         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17914         the edge is also fallthru, preserve it and just clear the
17915         abnormal flag.
17916         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17917         also complex, preserve that and just clear the fallthru flag.
17918         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17919         update virtual operands.
17921 2013-05-06  Alan Modra  <amodra@gmail.com>
17923         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17924         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17925         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17926         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17927         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17928         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17930 2013-05-06  Alan Modra  <amodra@gmail.com>
17932         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17933         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17934         (DEFAULT_ASM_ENDIAN): Define.
17935         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17936         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17937         Update -K PIC clause from sysv4.h.
17938         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17939         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17941 2013-05-06  Alan Modra  <amodra@gmail.com>
17943         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
17944         twice for little-endian.
17945         (ashrdi3_no_power, ashrdi3): Support little-endian.
17947 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17949         PR target/55303
17950         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
17951         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
17952         related expanders.
17953         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
17954         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
17955         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
17956         New predicates.
17958 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
17959             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17961         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
17962         * config/pa/pa.opt: Make mbig-switch a no-op.
17963         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
17964         (CASE_VECTOR_MODE): Always return SImode.
17965         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
17966         for the !TARGET_BIG_SWITCH case.
17967         * config/pa/pa-linux.h: Likewise.
17968         * config/pa/pa-openbsd.h: Likewise.
17969         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
17970         * config/pa/pa.md (short_jump): Remove define_insn.
17971         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
17972         (casesi0): Remove define_insn.
17973         (type): Remove btable_branch.
17974         (pa_combine_type): Likewise.
17975         (in_nullified_branch_delay): Likewise.
17976         (in_call_delay): Likewise.
17977         (define_delay): Likewise.
17978         (define_insn_reservation "Z3"): Likewise.
17979         (define_insn_reservation "Z4"): Likewise.
17980         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
17981         (pa_adjust_insn_length): Remove adjustment for btable branches.
17982         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
17983         and mno-big-switch
17985 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
17987         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17988         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
17989         Add m->r,x alternatives.
17990         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
17991         splitters using SWI48x mode iterator.
17992         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
17993         TARGET_64BIT.  Add m->x alternative.
17994         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
17995         Add o->x alternative.  Enable for TARGET_SSE.
17996         (sse_storeq): Remove expander.
17997         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
17998         with memory input operand.
17999         (*vec_extractv2di_1 splitter): New.
18000         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18001         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18003 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18005         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18006         (INT_LOWPART): Delete.
18007         (extract_MB): Adjust.
18008         (extract_ME): Adjust.
18009         (print_operand): Adjust.
18011 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18013         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18014         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18015         (reg_or_logical_cint_operand, easy_fp_constant,
18016         logical_const_operand): Delete "CONST_DOUBLE" case.
18017         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18018         "HOST_BITS_PER_WIDE_INT == 64" test.
18019         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18020         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18021         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18022         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18023         test.
18024         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18025         CONST_DOUBLE DImode/VOIDmode case.
18026         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18027         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18028         CONST_DOUBLE VOIDmode case.
18029         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18030         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18031         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18032         Delete CONST_DOUBLE case.
18033         (splitters for mov FMOVE64 const_double): Delete
18034         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18035         "HOST_BITS_PER_WIDE_INT >= 64" test.
18036         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18037         case.
18038         (mov DI const_double): Delete.
18040 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18042         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18043         on op shows all bits zero in mode of a lowpart subreg, return zero.
18045 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18047         PR target/57150
18048         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18049         to save TFmode registers and DImode to save TImode registers for
18050         caller save operations.
18051         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18052         mark being partially clobbered since they only use the first
18053         double word.
18055         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18056         and TDmode only use the upper 64-bits of each VSX register.
18058 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18060         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18061         (find_candidates_in_block): Disable slsr_process_phi.
18063 2013-05-03  Guozhi Wei  <carrot@google.com>
18065         * coverage.c (coverage_obj_init): Move the construction of gcov
18066         constructor to ...
18067         (build_init_ctor): ... here.
18069 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18071         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18072         (slsr_cand_d): Redefine def_phi.
18073         (stride_status, phi_adjust_status, count_phis_status): New enums.
18074         (find_phi_def): New.
18075         (find_basis_for_base_expr): New.
18076         (find_basis_for_candidate): Handle hidden bases.
18077         (alloc_cand_and_find_basis): Handle phi candidates.
18078         (slsr_process_phi): New.
18079         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18080         (create_mul_imm_cand): Likewise.
18081         (create_add_ssa_cand): Exclude phi base candidates.
18082         (create_add_imm_cand): Likewise.
18083         (slsr_process_cast): Likewise.
18084         (slsr_process_copy): Likewise.
18085         (find_candidates_in_block): Handle phi candidates.
18086         (dump_candidate): Likewise.
18087         (unconditional_cands): Delete.
18088         (unconditional_cands_with_known_stride_p): Delete.
18089         (phi_dependent_cand_p): New.
18090         (cand_increment): Handle phi-dependent candidates.
18091         (replace_dependent): Delete.
18092         (replace_mult_candidate): New.
18093         (replace_unconditional_candidate): New.
18094         (incr_vec_index): Move to avoid forward reference.
18095         (create_add_on_incoming_edge): New.
18096         (create_phi_basis): New.
18097         (replace_dependents): Delete.
18098         (replace_conditional_candidate): New.
18099         (phi_add_costs): New.
18100         (replace_uncond_cands_and_profitable_phis): New.
18101         (record_increment): Handle phi adjustments.
18102         (record_phi_increments): New.
18103         (record_increments): Handle phi adjustments.
18104         (phi_incr_cost): New.
18105         (lowest_cost_path): Handle phis.
18106         (total_savings): Likewise.
18107         (analyze_increments): Likewise.
18108         (ncd_with_phi): New.
18109         (ncd_of_cand_and_phis): New.
18110         (nearest_common_dominator_for_cands): Handle phi increments.
18111         (all_phi_incrs_profitable): New.
18112         (replace_profitable_candidates): Handle phi-dependent candidates.
18113         (analyze_candidates_and_replace): Likewise.
18115 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18117         PR bootstrap/57154
18118         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18119         do not exceed REG_BR_PROB_BASE.
18121 2013-05-03  Jeff Law  <law@redhat.com>
18123         PR tree-optimization/57144
18124         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18125         operand of the condition will bit into the new type when eliminating
18126         a cast feeding a condition.
18128 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18130         PR rtl-optimization/57130
18131         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18132         of COMPARE as in_code to the recursive call if needed.
18134 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18136         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18137         (enabled): Handle new members.
18138         * config/i386/sse.md (*vec_concatv2si): Merge from
18139         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18140         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18142 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18144         PR tree-optimization/57027
18145         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18146         for fnms opportunity, check we got the prerequisite kind
18147         of tree / gimple before using accessor functions.
18149 2013-05-03  Richard Biener  <rguenther@suse.de>
18151         * double-int.h (lshift): New overload without precision
18152         and arith argument.
18153         (operator *=, operator +=, operator -=): Move ...
18154         * double-int.c (operator *=, operator +=, operator -=): ... here
18155         and implement more efficiently.
18156         (mul_double_with_sign): Remove.
18157         (lshift_double): Adjust to take unsinged shift argument, push
18158         dispatching code to callers.
18159         (mul_double_wide_with_sign): Add early out for callers that
18160         are not interested in high parts or overflow.
18161         (lshift): New function.
18162         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18163         dispatch code here.
18164         (lrotate, rrotate): Use logical shifts.
18165         * expr.c (get_inner_reference): Use lshift.
18166         * fixed-value.c (do_fixed_divide): Likewise.
18167         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18168         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18169         (indirect_refs_may_alias_p): Likewise.
18170         (stmt_kills_ref_p_1): Likewise.
18172 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18174         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18176 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18178         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18179         scalar form of FABD instruction.
18181 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18183         * lra-constraints.c (process_alt_operands): Add checking alt
18184         number to choose the best alternative.
18186 2013-05-02  Richard Biener  <rguenther@suse.de>
18188         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18189         bitmap and its handling.
18190         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18192 2013-05-02  Richard Biener  <rguenther@suse.de>
18194         PR middle-end/57140
18195         * tree-inline.c (copy_loops): Properly handle removed loops.
18196         (copy_cfg_body): Mark destination loops for fixup if source
18197         loops needed fixup.
18199 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18201         PR target/56732
18202         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18203         generating simple_return for naked functions.
18205 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18207         PR middle-end/56988
18208         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18209         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18210         flags match.
18211         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18212         ipa_agg_replacement_value structures.
18213         (known_aggs_to_agg_replacement_list): Likewise.
18214         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18215         (read_agg_replacement_chain): Likewise.
18216         (ipcp_transform_function): Also check that by_ref flags match.
18218 2013-05-02  Richard Biener  <rguenther@suse.de>
18220         * graphds.h (struct graph): Add obstack member.
18221         * graphds.c (new_graph): Initialize obstack and allocate
18222         vertices from it.
18223         (add_edge): Allocate edge from the obstack.
18224         (free_graph): Free the obstack instead of all edges and vertices.
18226 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18228         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18229         divides.
18230         * cfg.c (update_bb_profile_for_threading): Ditto.
18231         * tree-inline.c (copy_bb): Ditto.
18232         (copy_edges_for_bb): Ditto.
18233         (initialize_cfun): Ditto.
18234         (copy_cfg_body): Ditto.
18235         (expand_call_inline): Ditto.
18236         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18237         (estimate_node_size_and_time): Ditto.
18238         (inline_merge_summary): Ditto.
18239         * cgraphclones.c (cgraph_clone_edge): Ditto.
18240         (cgraph_clone_node): Ditto.
18241         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18242         (compute_trg_info): Ditto.
18244 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18246         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18247         S reg when fp attribute set.
18248         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18250 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18252         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18253         New pattern.
18254         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18255         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18256         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18258 2013-05-02  Richard Biener  <rguenther@suse.de>
18260         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18261         (struct instantiate_cache_entry): New type.
18262         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18263         (struct instantiate_cache_type): New type.
18264         (set_instantiated_value, get_instantiated_value): Remove.
18265         (get_instantiated_value_entry): New function.
18266         (instantiate_scev_name): Use the new cache and adjust.
18267         (instantiate_scev_poly): Adjust.
18268         (instantiate_scev_binary): Likewise.
18269         (instantiate_array_ref): Likewise.
18270         (instantiate_scev_convert): Likewise.
18271         (instantiate_scev_not): Likewise.
18272         (instantiate_scev_3): Likewise.
18273         (instantiate_scev_2): Likewise.
18274         (instantiate_scev_r): Likewise.
18275         (instantiate_scev): Likewise.
18276         (resolve_mixers): Likewise.
18278 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18280         PR target/57091
18281         * lra-constraints.c (best_small_class_operands_num): Remove.
18282         (process_alt_operands): Remove small_class_operands_num.  Take
18283         small classes operands into losers and only if the operand is not
18284         matched.  Modify debugging output.
18285         (curr_insn_transform): Remove best_small_class_operands_num.
18286         Print insn name.
18288 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18290         * config/aarch64/aarch64-builtins.c
18291         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18292         * config/aarch64/aarch64-simd-builtins.def
18293         (reduc_splus_): Add new modes.
18294         (reduc_uplus_): New.
18295         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18296         (reduc_uplus_v4sf): Likewise.
18297         (reduc_splus_v4sf): Likewise.
18298         (aarch64_addv<mode>): Likewise.
18299         (reduc_uplus_<mode>): Likewise.
18300         (reduc_splus_<mode>): Likewise.
18301         (aarch64_addvv2di): Likewise.
18302         (reduc_uplus_v2di): Likewise.
18303         (reduc_splus_v2di): Likewise.
18304         (aarch64_addvv2si): Likewise.
18305         (reduc_uplus_v2si): Likewise.
18306         (reduc_splus_v2si): Likewise.
18307         (reduc_<sur>plus_<mode>): New.
18308         (reduc_<sur>plus_v2di): Likewise.
18309         (reduc_<sur>plus_v2si): Likewise.
18310         (reduc_<sur>plus_v4sf): Likewise.
18311         (aarch64_addpv4sf): Likewise.
18312         * config/aarch64/arm_neon.h
18313         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18314         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18315         add UNSPEC_SADDV, UNSPEC_UADDV.
18316         (SUADDV): New.
18317         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18319 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18321         * config/aarch64/arm_neon.h
18322         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18324 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18326         * config/aarch64/aarch64-builtins
18327         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18329 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18331         * config/aarch64/aarch64-simd-builtins.def
18332         (reduc_smax_): New.
18333         (reduc_smin_): Likewise.
18334         (reduc_umax_): Likewise.
18335         (reduc_umin_): Likewise.
18336         (reduc_smax_nan_): Likewise.
18337         (reduc_smin_nan_): Likewise.
18338         (fmax): Remove.
18339         (fmin): Likewise.
18340         (smax): Update for V2SF, V4SF and V2DF modes.
18341         (smin): Likewise.
18342         (smax_nan): New.
18343         (smin_nan): Likewise.
18344         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18345         (<su><maxmin><mode>3): ...This, refactor.
18346         (s<maxmin><mode>3): New.
18347         (<maxmin_uns><mode>3): Likewise.
18348         (reduc_<maxmin_uns>_<mode>): Refactor.
18349         (reduc_<maxmin_uns>_v4sf): Likewise.
18350         (reduc_<maxmin_uns>_v2si): Likewise.
18351         (aarch64_<fmaxmin><mode>: Remove.
18352         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18353         new builtin names.
18354         (vmin<q>_f<32,64>): Likewise.
18355         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18356         (FMAXMIN): New.
18357         (su): Add mappings for smax, smin, umax, umin.
18358         (maxmin): New.
18359         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18360         (FMAXMIN): Rename as...
18361         (FMAXMIN_UNS): ...This.
18362         (maxminv): Remove.
18363         (fmaxminv): Likewise.
18364         (fmaxmin): Likewise.
18365         (maxmin_uns): New.
18366         (maxmin_uns_op): Likewise.
18368 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18370         * config/aarch64/arm_neon.h
18371         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18372         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18373         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18375 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18377         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18378         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18380 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18382         * config/aarch64/aarch64-simd.md
18383         (vcond<mode>_internal): Handle special cases for constant masks.
18384         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18385         (vcondu<mode><mode>): Likewise.
18386         (vcond<v_cmp_result><mode>): New.
18388 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18390         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18391         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18392         * config/aarch64/aarch64-simd-builtins.def
18393         (cmeq): Update to BUILTIN_VALLDI.
18394         (cmgt): Likewise.
18395         (cmge): Likewise.
18396         (cmle): Likewise.
18397         (cmlt): Likewise.
18398         * config/aarch64/arm_neon.h
18399         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18400         to builtins or C as appropriate.
18402 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18404         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18405         (cmgeu): ...This.
18406         (cmhi): Rename to...
18407         (cmgtu): ...This.
18408         * config/aarch64/aarch64-simd.md
18409         (simd_mode): Add SF.
18410         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18411         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18412         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18413         (cstore<mode>_neg): ...This.
18414         * config/aarch64/iterators.md
18415         (VALLF): new.
18416         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18417         (COMPARISONS): New.
18418         (UCOMPARISONS): Likewise.
18419         (optab): Add missing comparisons.
18420         (n_optab): New.
18421         (cmp_1): Likewise.
18422         (cmp_2): Likewise.
18423         (CMP): Likewise.
18424         (cmp): Remove.
18425         (VCMP_S): Likewise.
18426         (VCMP_U): Likewise.
18427         (V_cmp_result): Add DF, SF modes.
18428         (v_cmp_result): Likewise.
18429         (v): Likewise.
18430         (vmtype): Likewise.
18431         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18433 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18435         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18436         define_insn to define_insn_and_split.
18437         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18438         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18439         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18440         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18441         (thumb2_negscc): Likewise.
18443 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18445         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18447 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18449         * config/arm/thumb2.md: Remove trailing whitespaces.
18451 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18453         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18454         Use gen_int_mode rather than GEN_INT.
18456 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18458         * value-prof.c (stream_in_histogram_value): Remove the strayed
18459         debug_gimple_stmt.
18461 2013-04-30  Richard Biener  <rguenther@suse.de>
18463         PR middle-end/57122
18464         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18466 2013-04-30  Richard Biener  <rguenther@suse.de>
18468         PR middle-end/57107
18469         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18471 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18473         PR rtl-optimization/56957
18474         PR rtl-optimization/57105
18475         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18476         variable.  Use just INSN_UID for determining whether an insn
18477         should be only disconnected from the insn stream.
18478         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18480 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18482         PR tree-optimization/57104
18483         * tsan.c (instrument_expr): Don't instrument accesses to
18484         DECL_HARD_REGISTER VAR_DECLs.
18486 2013-04-30  Richard Biener  <rguenther@suse.de>
18488         * function.h (loops_for_fn): New inline function.
18489         (set_loops_for_fn): Likewise.
18490         * cfgloop.h (place_new_loop): Add struct function parameter.
18491         (get_loop): Likewise.
18492         (get_loops): Likewise.
18493         (number_of_loops): Likewise.
18494         (fel_next): Adjust.
18495         (fel_init): Likewise.
18496         * cfg.c (get_loop_copy): Adjust.
18497         * cfgloop.c (flow_loops_dump): Likewise.
18498         (record_loop_exits): Likewise.
18499         (verify_loop_structure): Likewise.
18500         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18501         (estimate_reg_pressure_cost): Likewise.
18502         (mark_loop_exit_edges): Likewise.
18503         * cfgloopmanip.c (place_new_loop): Likewise.
18504         (add_loop): Likewise.
18505         (duplicate_loop): Likewise.
18506         * graph.c (draw_cfg_nodes): Likewise.
18507         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18508         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18509         (extract_affine_chrec): Likewise.
18510         (build_scop_iteration_domain): Likewise.
18511         * graphite.c (graphite_initialize): Likewise.
18512         * ira-build.c (create_loop_tree_nodes): Likewise.
18513         (more_one_region_p): Likewise.
18514         (rebuild_regno_allocno_maps): Likewise.
18515         (mark_loops_for_removal): Likewise.
18516         (mark_all_loops_for_removal): Likewise.
18517         (remove_unnecessary_regions): Likewise.
18518         (ira_build): Likewise.
18519         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18520         * loop-init.c (fix_loop_structure): Likewise.
18521         (gate_rtl_move_loop_invariants): Likewise.
18522         (gate_rtl_unswitch): Likewise.
18523         (gate_rtl_unroll_and_peel_loops): Likewise.
18524         (rtl_doloop): Likewise.
18525         * lto-streamer-in.c (input_cfg): Likewise.
18526         * lto-streamer-out.c (output_cfg): Likewise.
18527         * modulo-sched.c (sms_schedule): Likewise.
18528         * predict.c (tree_estimate_probability): Likewise.
18529         (tree_estimate_probability_driver): Likewise.
18530         (estimate_loops): Likewise.
18531         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18532         (move_sese_region_to_fn): Likewise.
18533         (debug_loop_num): Likewise.
18534         * tree-chrec.c (chrec_evaluate): Likewise.
18535         (hide_evolution_in_other_loops_than_loop): Likewise.
18536         (chrec_component_in_loop_num): Likewise.
18537         (reset_evolution_in_loop): Likewise.
18538         (evolution_function_is_invariant_rec_p): Likewise.
18539         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18540         * tree-inline.c (copy_loops): Likewise.
18541         (copy_cfg_body): Likewise.
18542         (tree_function_versioning): Likewise.
18543         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18544         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18545         Likewise.
18546         (add_to_evolution_1): Likewise.
18547         (scev_const_prop): Likewise.
18548         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18549         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18550         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18551         (tree_ssa_lim_initialize): Likewise.
18552         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18553         (verify_loop_closed_ssa): Likewise.
18554         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18555         (tree_ssa_loop_im): Likewise.
18556         (tree_ssa_loop_unswitch): Likewise.
18557         (tree_vectorize): Likewise.
18558         (check_data_deps): Likewise.
18559         (tree_ssa_loop_ivcanon): Likewise.
18560         (tree_ssa_loop_bounds): Likewise.
18561         (tree_complete_unroll): Likewise.
18562         (tree_complete_unroll_inner): Likewise.
18563         (tree_parallelize_loops): Likewise.
18564         (tree_ssa_loop_prefetch): Likewise.
18565         (tree_ssa_loop_ivopts): Likewise.
18566         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18567         * tree-vectorizer.c (vectorize_loops): Likewise.
18569 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18571         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18572         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18573         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18574         with EABI_LINK_SPEC.
18576 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18578         PR target/44578
18579         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18580         alternative.
18582 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18584         PR target/57097
18585         * lra-constraints.c (process_alt_operands): Discourage a bit more
18586         using memory for pseudos.  Print cost dump for alternatives.
18587         Modify cost values for conflicts with early clobbers.
18588         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18590 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18592         PR target/57098
18593         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18595 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18597         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18598         from/to S register.
18599         (movdi_aarch64): Support LDR/STR from/to D register.
18601 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18603         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18604         or higher by default.
18606 2013-04-29  Richard Biener  <rguenther@suse.de>
18608         PR middle-end/57075
18609         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18610         even if not adding abnormal edges for calls that can make
18611         abnormal gotos.
18613 2013-04-29  Richard Biener  <rguenther@suse.de>
18615         PR middle-end/57103
18616         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18617         TREE_BLOCK.
18618         (move_stmt_r): Remove redundant checking.
18620 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18622         PR bootstrap/57077
18623         * basic-block.h (apply_scale): New function.
18624         (apply_probability): Use apply_scale.
18625         * gimple-streamer-in.c (input_bb): Ditto.
18626         * lto-streamer-in.c (input_cfg): Ditto.
18627         * lto-cgraph.c (merge_profile_summaries): Ditto.
18628         * tree-optimize.c (execute_fixup_cfg): Ditto.
18629         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18630         (copy_edges_for_bb): Ditto.
18631         (copy_cfg_body): Ditto.
18633 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18635         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18636         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18637         (tail_merge_optimize): Handle current_loops == NULL.
18639 2013-04-26  Jeff Law  <law@redhat.com>
18641         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18642         (simplify_cond_using_ranges): Generalize code to simplify
18643         COND_EXPRs where one argument is a constant and the other
18644         is an SSA_NAME created by an integral type conversion.
18646 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18648         * config/arm/arm.md (store_minmaxsi): Use only when
18649         optimize_insn_for_size_p.
18651 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18653         PR target/57108
18654         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18656 2013-04-29  Richard Biener  <rguenther@suse.de>
18658         PR middle-end/57089
18659         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18660         loop tree make sure to schedule a fixup for the child as well.
18661         (expand_omp_for_generic): Properly add loops.
18662         (expand_omp_for_static_nochunk): Likewise.
18663         (expand_omp_for_static_chunk): Likewise.
18664         (expand_omp_for): For the degenerate case fixup loops.
18665         (expand_omp_sections): Fix default bb placement in loops.
18666         (expand_omp_atomic_pipeline): Properly add loops.
18668 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18670         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18672 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18674         * tree-ssa-tail-merge.c: Update header comment.
18676 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18678         * config/aarch64/arm_neon.h
18679         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18680         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18681         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18682         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18683         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18684         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18685         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18686         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18688 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18690         * config/aarch64/aarch64-simd.md
18691         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18692         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18693         fix_trunc, fixuns_trunc.
18694         (ftrunc<VDQF:mode>2): New.
18695         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18696         (fix_trunc_optab): New.
18698 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18700         * config/aarch64/aarch64-builtins.c
18701         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18702         iceilf, lround, iroundf.
18704 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18706         PR target/54349
18707         * config/i386/i386.h (enum ix86_tune_indices)
18708         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18709         New, split from X86_TUNE_INTER_UNIT_MOVES.
18710         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18711         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18712         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18713         (TARGET_INTER_UNIT_MOVES): Remove.
18714         * config/i386/i386.c (initial_ix86_tune_features): Update.
18715         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18716         (ix86_expand_convert_uns_didf_sse): Use
18717         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18718         (ix86_expand_vector_init_one_nonzero): Ditto.
18719         (ix86_expand_vector_init_interleave): Ditto.
18720         (inline_secondary_memory_needed): Return true for moves from SSE class
18721         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18722         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18723         * config/i386/constraints.md (Yi, Ym): Depend on
18724         TARGET_INTER_UNIT_MOVES_TO_VEC.
18725         (Yj, Yn): New constraints.
18726         * config/i386/i386.md (*movdi_internal): Change constraints of
18727         operand 1 from Yi to Yj and from Ym to Yn.
18728         (*movsi_internal): Ditto.
18729         (*movdf_internal): Ditto.
18730         (*movsf_internal): Ditto.
18731         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18732         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18733         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18734         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18735         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18736         * config/i386/sse.md (movdi_to_sse): Ditto.
18737         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18738         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18739         TARGET_INTER_UNIT_MOVES.
18740         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18741         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18742         instead of TARGET_INTER_UNIT_MOVES.
18743         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18744         operand 1 from Yi to Yj and from Ym to Yn.
18746 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18748         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18749         (float_truncate_hi_): Likewise.
18750         (float_extend_lo_): Likewise.
18751         (float_truncate_lo_): Likewise.
18752         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18753         (aarch64_float_extend_lo_v2df): Likewise.
18754         (vec_unpacks_hi_v4sf): Likewise.
18755         (aarch64_float_truncate_lo_v2sf): Likewise.
18756         (aarch64_float_truncate_hi_v4sf): Likewise.
18757         (vec_pack_trunc_v2df): Likewise.
18758         (vec_pack_trunc_df): Likewise.
18760 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18762         * config/aarch64/aarch64-builtins.c
18763         (aarch64_fold_builtin): Fold float conversions.
18764         * config/aarch64/aarch64-simd-builtins.def
18765         (floatv2si, floatv4si, floatv2di): New.
18766         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18767         * config/aarch64/aarch64-simd.md
18768         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18769         * config/aarch64/iterators.md (FLOATUORS): New.
18770         (optab): Add float, floatuns.
18771         (su_optab): Likewise.
18773 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18775         * config/aarch64/aarch64-builtins.c
18776         (aarch64_builtin_vectorized_function): Use new names for
18777         fcvt builtins.
18778         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18779         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18780         (fcvtzu): Split as...
18781         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18782         (fcvtas): Split as...
18783         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18784         (fcvtau): Split as...
18785         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18786         (fcvtps): Split as...
18787         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18788         (fcvtpu): Split as...
18789         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18790         (fcvtms): Split as...
18791         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18792         (fcvtmu): Split as...
18793         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18794         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18795         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18796         (lfrintnusf, lfrintnudf): Likewise.
18797         * config/aarch64/aarch64-simd.md
18798         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18799         define_insn.
18800         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18801         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18802         (fcvt_pattern): Likewise.
18804 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18806         * config/aarch64/aarch64-simd.md
18807         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18808         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18810 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18812         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18813         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18814         (vrnd<a,m,n,p>_f32): Implement using builtins.
18815         (vrnd<i,x><q>_f<32, 64>): New.
18817 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18819         * config/aarch64/aarch64-builtins.c
18820         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18821         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18822         (frintz): Rename to...
18823         (btrunc): ...this.
18824         (frintp): Rename to...
18825         (ceil): ...this.
18826         (frintm): Rename to...
18827         (floor): ...this.
18828         (frinti): Rename to...
18829         (nearbyint): ...this.
18830         (frintx): Rename to...
18831         (rint): ...this.
18832         (frinta): Rename to...
18833         (round): ...this.
18834         * config/aarch64/aarch64-simd.md
18835         (aarch64_frint<frint_suffix><mode>): Delete.
18836         (<frint_pattern><mode>2): Convert to insn.
18837         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18838         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18839         (frint_pattern): Likewise.
18840         (frint_suffix): Likewise.
18842 2013-04-29  Richard Biener  <rguenther@suse.de>
18844         PR tree-optimization/57081
18845         * loop-init.c: Include tree-flow.h.
18846         (loop_optimizer_finalize): Free number of iteration estimates.
18847         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18849 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18851         PR tree-optimization/57083
18852         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18853         non-singleton shift count range, zero extend low_bound for uns case.
18855         * config/i386/predicates.md (general_vector_operand): New predicate.
18856         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18857         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18858         if they aren't nonimmediate operands.  If their original values
18859         satisfy const_vector_equal_evenodd_p, don't shift them.
18860         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18861         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18862         if not nonimmediate_operand.
18863         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18864         instead of register_operand.
18865         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18867 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18869         * stor-layout.c (finalize_size_functions): Allocate a structure and
18870         reset cfun before dumping the functions.
18872 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18874         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18876         PR target/56866
18877         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18878         use xop_pmacsdqh if uns_p.
18879         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18880         the immediate rotate count.
18882 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18884         * rtl.h (struct rtx_def): Add comment for field jump.
18885         (LRA_SUBREG_P): New macro.
18886         * recog.c (register_operand): Check LRA_SUBREG_P.
18887         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18888         stack frame.
18889         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18890         (lra_final_code_change): Skip subreg change for operators.
18891         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18892         if there are no operand changes.
18893         * lra-constraints.c (curr_insn_set): New.
18894         (match_reload): Set LRA_SUBREG_P.
18895         (emit_spill_move): Ditto.
18896         (check_and_process_move): Use curr_insn_set. Process only single
18897         set insns.  Don't initialize sec_mem_p and change_p.
18898         (simplify_operand_subreg): Use LRA_SUBREG_P.
18899         (reg_in_class_p): New function.
18900         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18901         of #ifdef.  Add code to remove cycling.
18902         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18903         non-null disp.  Reload inner instead of disp when base and index
18904         are null.  Try to put lo_sum into register.
18905         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18906         (check_and_process_move): Move code for move cost check to
18907         simple_move_p.  Remove equiv_substitution.
18908         (simple_move_p): New function.
18909         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18910         curr_insn_set.  Call check_and_process_move only for single set
18911         insns.  Use the new function.  Move call of check_and_process_move
18912         after operand equiv substitution and address process.
18914 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18916         PR go/57045
18917         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18918         with nonlocal goto receivers or returns twice calls, ignore
18919         unininitialized values from abnormal edges to nl goto receiver
18920         or returns twice call.
18922 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18924         PR tree-optimization/57051
18925         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18926         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18927         bitsize.
18929 2013-04-26  Richard Biener  <rguenther@suse.de>
18931         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18932         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18933         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18934         (fixup_loop_arrays_after_move): New function.
18935         (move_sese_region_to_fn): Properly outline the loop tree parts
18936         of the SESE region.
18938 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18940         * config/i386/i386.md (type, unit): Fix long lines.
18942 2013-04-26  Richard Biener  <rguenther@suse.de>
18944         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
18945         (lto-streamer-out.o): Likewise.
18946         * cfgloop.c (init_loops_structure): Export, add struct function
18947         argument and adjust.
18948         (flow_loops_find): Adjust.
18949         * cfgloop.h (enum loop_estimation): Add EST_LAST.
18950         (init_loops_structure): Declare.
18951         * lto-streamer-in.c: Include cfgloop.h.
18952         (input_cfg): Input the loop tree.
18953         * lto-streamer-out.c: Include cfgloop.h.
18954         (output_cfg): Output the loop tree.
18955         (output_struct_function_base): Do not drop PROP_loops.
18957 2013-03-26  Richard Biener  <rguenther@suse.de>
18959         * tree-cfg.c (execute_build_cfg): Build the loop tree.
18960         (pass_build_cfg): Provide PROP_loops.
18961         (move_sese_region_to_fn): Remove loops that are outlined into fn
18962         for now.
18963         * tree-inline.c: Include cfgloop.h.
18964         (initialize_cfun): Do not drop PROP_loops.
18965         (copy_loops): New function.
18966         (copy_cfg_body): Copy loop structure.
18967         (tree_function_versioning): Initialize destination loop tree.
18968         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
18969         (pass_parallelize_loops): Do IL verification.
18970         * loop-init.c (loop_optimizer_init): Fixup loops if required.
18971         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
18972         the CFG make sure we fixup loops as well.
18973         * tree-ssa-tail-merge.c: Include cfgloop.h.
18974         (replace_block_by): When merging loop latches mark loops for fixup.
18975         * lto-streamer-out.c (output_struct_function_base): Drop
18976         PROP_loops for now.
18977         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
18978         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
18979         * ipa-split.c: Include cfgloop.h.
18980         (split_function): Add the new return block to the loop tree root.
18981         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
18982         whether we have removed the forwarder block.
18983         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
18984         * cfgloop.h (place_new_loop): Declare.
18985         * cfgloopmanip.c (place_new_loop): Export.
18986         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
18987         (tree-switch-conversion.o): Likewise.
18988         (tree-complex.o): Likewise.
18989         (tree-inline.o): Likewise.
18990         (tree-ssa-tailmerge.o): Likewise.
18991         (ipa-split.o): Likewise.
18992         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
18993         (tree-ssa-copy.o): Likewise.
18994         * tree-switch-conversion.c: Include cfgloop.h
18995         (process_switch): If we emit a bit-test cascade, schedule loops
18996         for fixup.
18997         * tree-complex.c: Include cfgloop.h.
18998         (expand_complex_div_wide): Properly add new basic-blocks to loops.
18999         * asan.c: Include cfgloop.h.
19000         (create_cond_insert_point): Properly add new basic-blocks to
19001         loops, schedule loop fixup.
19002         * cfgloop.c (verify_loop_structure): Check that looks are not
19003         marked for fixup.
19004         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19005         to loops.
19006         (expand_omp_for_generic): Likewise.
19007         (expand_omp_sections): Likewise.
19008         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19009         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19010         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19011         is initialized, not when loops are present.
19012         * tree-parloops.c (parallelize_loops): Remove checking here.
19013         * passes.c (init_optimization_passes): Schedule a copy-propagation
19014         pass before complete unrolling of inner loops.
19016 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19018         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19019         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19020         (colorize_init): Add argument to _WIN32 version.
19021         * toplev.c: Include diagnostic-color.h.
19022         (process_options): Default to -fdiagnostics-color=auto if
19023         GCC_COLORS env var is in the environment.
19024         * common.opt (fdiagnostics-color=): Add Var and Init.
19025         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19026         env var is in the environment, the default is auto rather than never.
19028         * diagnostic.h (file_name_as_prefix): Add context argument.
19029         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19030         the string as locus.
19031         * langhooks.c (lhd_print_error_function): Adjust caller.
19033 2013-04-25  Lawrence Crowl  <crowl@google.com>
19035         * var-tracking.c (shared_hash_def::htab):
19036         Change type to hash_table.  Update dependent calls and types.
19038 2013-04-25  Lawrence Crowl  <crowl@google.com>
19040         * Makefile.in: Update as needed below.
19042         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19043         Move declaration to after the type's method definitons.
19045         * attribs.c (htab_t scoped_attributes::attribute_hash):
19046         Change type to hash_table.  Update dependent calls and types.
19048         * bitmap.c (htab_t bitmap_desc_hash):
19049         Change type to hash_table.  Update dependent calls and types.
19051         * cselib.c (htab_t cselib_hash_table):
19052         Change type to hash_table.  Update dependent calls and types.
19054         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19055         (hash_string_slot_node): Move implementation into lto-streamer.h
19056         struct string_slot_hasher.
19057         (eq_string_slot_node): Likewise.
19059         * data-streamer-out.c: Update output_block::string_hash_table
19060         dependent calls and types.
19062         * dwarf2cfi.c (htab_t trace_index):
19063         Change type to hash_table.  Update dependent calls and types.
19065         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19066         Change type to hash_table.  Update dependent calls and types.
19067         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19068         (htab_t optimize_external_refs::map): Likewise.
19069         (htab_t output_comp_unit::extern_map): Likewise.
19070         (htab_t output_comdat_type_unit::extern_map): Likewise.
19071         (htab_t output_macinfo::macinfo_htab): Likewise.
19072         (htab_t optimize_location_lists::htab): Likewise.
19073         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19075         * except.c (htab_t ehspec_hash_type):
19076         Change type to hash_table.  Update dependent calls and types.
19077         (assign_filter_values::ttypes): Likewise.
19078         (assign_filter_values::ehspec): Likewise.
19079         (sjlj_assign_call_site_values::ar_hash): Likewise.
19080         (convert_to_eh_region_ranges::ar_hash): Likewise.
19082         * gcse.c (htab_t pre_ldst_table):
19083         Change type to hash_table.  Update dependent calls and types.
19085         * ggc-common.c (htab_t saving_htab):
19086         Change type to hash_table.  Update dependent calls and types.
19087         (htab_t loc_hash): Likewise.
19088         (htab_t ptr_hash): Likewise.
19089         (call_count): Rename ggc_call_count.
19090         (call_alloc): Rename ggc_call_alloc.
19091         (loc_descriptor): Rename make_loc_descriptor.
19092         (add_statistics): Rename ggc_add_statistics.
19094         * ggc-common.c (saving_htab):
19095         Change type to hash_table.  Update dependent calls and types.
19097         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19098         (push_gimplify_context): Likewise.
19099         (pop_gimplify_context): Likewise.
19100         (struct gimple_temp_hash_elt): Added.
19101         (struct gimplify_hasher): Likewise.
19102         (struct gimplify_ctx.temp_htab):
19103         Change type to hash_table.  Update dependent calls and types.
19105         * gimple-fold.c: Include gimplify-ctx.h.
19107         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19108         Change type to hash_table.  Update dependent calls and types.
19109         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19110         avoid potential global name collision.
19112         * gimplify.c: Include gimplify-ctx.h.
19113         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19114         (htab_t gimplify_ctx::temp_htab):
19115         Update dependent calls and types for new type hash_table.
19116         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19117         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19119         * gimplify-ctx.h: New.
19120         (struct gimple_temp_hash_elt): Move from gimplify.c.
19121         (class gimplify_hasher): New.
19122         (struct gimplify_ctx): Move from gimple.h.
19123         (htab_t gimplify_ctx::temp_htab):
19124         Change type to hash_table.  Update dependent calls and types.
19126         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19127         (htab_t ivs_params::newivs_index):
19128         Change type to hash_table.  Update dependent calls and types.
19129         (htab_t ivs_params::params_index): Likewise.
19130         (htab_t print_generated_program::params_index): Likewise.
19131         (htab_t gloog::newivs_index): Likewise.
19132         (htab_t gloog::params_index): Likewise.
19134         * graphite.c: Include graphite-htab.h.
19135         4htab_t graphite_transform_loops::bb_pbb_mapping):
19136         Change type to hash_table.  Update dependent calls and types.
19138         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19139         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19140         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19142         * graphite-dependences.c: Include graphite-htab.h.
19143         (loop_is_parallel_p): Change hash table type of parameter.
19145         * graphite-htab.h: New.
19146         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19147         (extern find_pbb_via_hash): Move from graphite-poly.h.
19148         (extern loop_is_parallel_p): Move from graphite-poly.h.
19149         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19151         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19152         (extern loop_is_parallel_p): Move to graphite-htab.h.
19153         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19155         * haifa-sched.c (htab_t delay_htab):
19156         Change type to hash_table.  Update dependent calls and types.
19157         (htab_t delay_htab_i2): Likewise.
19159         * ira-color.c (htab_t allocno_hard_regs_htab):
19160         Change type to hash_table.  Update dependent calls and types.
19162         * ira-costs.c (htab_t cost_classes_htab):
19163         Change type to hash_table.  Update dependent calls and types.
19165         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19166         Change type to hash_table.  Update dependent calls and types.
19168         * loop-iv.c (htab_t bivs):
19169         Change type to hash_table.  Update dependent calls and types.
19171         * loop-unroll.c (htab_t opt_info::insns_to_split):
19172         Change type to hash_table.  Update dependent calls and types.
19173         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19175         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19176         (struct string_slot_hasher): New.
19177         (htab_t output_block::string_hash_table):
19178         Change type to hash_table.  Update dependent calls and types.
19180         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19181         (htab_t file_name_hash_table):
19182         Change type to hash_table.  Update dependent calls and types.
19184         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19185         calls and types.
19187         * lto-streamer.c (htab_t tree_htab):
19188         Change type to hash_table.  Update dependent calls and types.
19190         * omp-low.c: Include gimplify-ctx.h.
19192         * passes.c (htab_t name_to_pass_map):
19193         Change type to hash_table.  Update dependent calls and types.
19194         (pass_traverse): Rename to passes_pass_traverse.
19196         * plugin.c (htab_t event_tab):
19197         Change type to hash_table.  Update dependent calls and types.
19199         * postreload-gcse.c (htab_t expr_table):
19200         Change type to hash_table.  Update dependent calls and types.
19201         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19203         * sese.c (debug_rename_map_1): Make extern.
19204         (htab_t copy_bb_and_scalar_dependences::rename_map):
19205         Change type to hash_table.  Update dependent calls and types.
19207         * sese.h (extern debug_rename_map): Move to .c file.
19209         * store-motion.c (htab_t store_motion_mems_table):
19210         Change type to hash_table.  Update dependent calls and types.
19212         * trans-mem.c (htab_t tm_new_mem_hash):
19213         Change type to hash_table.  Update dependent calls and types.
19215         * tree-browser.c (htab_t TB_up_ht):
19216         Change type to hash_table.  Update dependent calls and types.
19218         * tree-cfg.c (htab_t discriminator_per_locus):
19219         Change type to hash_table.  Update dependent calls and types.
19221         * tree-complex.c: Include tree-hasher.h
19222         (htab_t complex_variable_components):
19223         Change type to hash_table.  Update dependent calls and types.
19225         * tree-eh.c (htab_t finally_tree):
19226         Change type to hash_table.  Update dependent calls and types.
19228         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19229         struct int_tree_hasher.
19230         (extern int_tree_map_eq): Likewise.
19231         (uid_decl_map_hash): Removed.
19232         (extern decl_tree_map_eq): Likewise.
19234         * tree-hasher.h: New.
19235         (struct int_tree_hasher): New.
19236         (typedef int_tree_htab_type): New.
19238         * tree-inline.c: Include gimplify-ctx.h.
19240         * tree-mudflap.c: Include gimplify-ctx.h.
19242         * tree-parloops.c: Include tree-hasher.h.
19243         (htab_t eliminate_local_variables_stmt::decl_address):
19244         Change type to hash_table.  Update dependent calls and types.
19245         (htab_t separate_decls_in_region::decl_copies): Likewise.
19247         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19248         Change type to hash_table.  Update dependent calls and types.
19250         * tree-sra.c (candidates):
19251         Change type to hash_table.  Update dependent calls and types.
19253         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19254         in tree-flow.h.
19255         (int_tree_map_hash): Likewise.
19257         * tree-ssa-dom.c (htab_t avail_exprs):
19258         Change type to hash_table.  Update dependent calls and types.
19260         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19261         Change type to hash_table.  Update dependent calls and types.
19263         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19264         Change type to hash_table.  Update dependent calls and types.
19266         * tree-ssa-phiopt.c (seen_ssa_names):
19267         Change type to hash_table.  Update dependent calls and types.
19269         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19270         Change type to hash_table.  Update dependent calls and types.
19272         * tree-ssa-uncprop.c (equiv):
19273         Change type to hash_table.  Update dependent calls and types.
19275 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19277         PR rtl-optimization/57003
19278         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19279         call note_stores with kill_clobbered_value callback again after
19280         killing regs_invalidated_by_call.
19282 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19284         * config/aarch64/aarch64-simd.md
19285         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19286         (aarch64_simd_bsl<mode>): Likewise.
19287         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19289 2013-04-25  Marek Polacek  <polacek@redhat.com>
19291         PR tree-optimization/57066
19292         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19294 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19296         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19298 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19300         * config/aarch64/aarch64-builtins.c
19301         (aarch64_fold_builtin): New.
19302         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19303         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19304         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19305         * config/aarch64/arm_neon.h
19306         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19308 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19309             Tejas Belagod  <tejas.belagod@arm.com>
19311         * config/aarch64/aarch64-builtins.c
19312         (aarch64_gimple_fold_builtin): New.
19313         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19314         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19315         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19316         (addvv4sf): Update.
19317         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19319 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19321         * config/aarch64/aarch64.md
19322         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19324 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19326         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19327         (*ngcsi_uxtw): New pattern.
19329 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19330             Julian Brown  <julian@codesourcery.com>
19332         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19333         (TB_DREG): Add T_V4HF.
19334         (v4hf_UP): New macro.
19335         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19336         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19337         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19338         built-ins.
19339         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19340         (arm_vector_mode_supported_p): Handle V4HF.
19341         (arm_mangle_map): Handle V4HFmode.
19342         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19343         * config/arm/arm_neon_builtins.def: Add entries for
19344         vcvtv4hfv4sf, vcvtv4sfv4hf.
19345         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19346         (neon_vcvtv4hfv4sf): Likewise.
19347         * config/arm/neon-gen.ml: Handle half-precision floating point
19348         features.
19349         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19350         * config/arm/arm_neon.h: Regenerate.
19351         * config/arm/neon.ml (type elts): Add F16.
19352         (type vectype): Add T_float16x4, T_floatHF.
19353         (type vecmode): Add V4HF.
19354         (type features): Add Requires_FP_bit feature.
19355         (elt_width): Handle F16.
19356         (elt_class): Likewise.
19357         (elt_of_class_width): Likewise.
19358         (mode_of_elt): Refactor.
19359         (type_for_elt): Handle F16, fix error messages.
19360         (vectype_size): Handle T_float16x4.
19361         (vcvt_sh): New function.
19362         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19363         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19364         (string_of_mode): Handle V4HF.
19365         * doc/arm-neon-intrinsics.texi: Regenerate.
19367 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19369         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19370         format specifier in 'X' case.
19372 2013-04-25  Alan Modra  <amodra@gmail.com>
19374         PR target/57052
19375         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19376         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19377         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19378         Repeat for many other rotate/shift and mask patterns using subregs.
19379         Name lshiftrt insns.
19380         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19381         on WORDS_BIG_ENDIAN.
19383 2013-04-25  Alan Modra  <amodra@gmail.com>
19385         * config.gcc: Support little-endian powerpc-linux targets.
19386         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19387         (LINK_OS_LINUX_SPEC): Define.
19388         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19389         Preserve MASK_LITTLE_ENDIAN.
19390         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19391         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19392         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19393         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19394         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19395         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19396         Correct fp word order for little-endian.  Don't shift toc entries
19397         smaller than a word for little-endian.
19398         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19399         (bswapdi2 splits): Correct low-part subreg for little-endian.
19400         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19401         low/high where such is correct only for be.
19402         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19403         little-endian for -mcall-aixdesc.
19405 2013-04-25  Alan Modra  <amodra@gmail.com>
19407         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19408         replace_equiv_address_nv.
19410 2013-04-25  Alan Modra  <amodra@gmail.com>
19412         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19414 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19416         Revert:
19417         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19418         * rtl.h (struct rtx_def): ...
19420 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19422         PR rtl-optimizations/57046
19423         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19424         for multi-reg splits.
19426 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19428         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19430 2013-04-24  Sterling Augustine  <saugustine@google.com>
19432         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19433         (comp_dir_string, debug_str_dwo_section): New.
19434         (DEBUG_STR_DWO_SECTION): Rename to ...
19435         (DEBUG_DWO_STR_SECTION): ... this.
19436         (DEBUG_NORM_STR_SECTION): Delete.
19437         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19438         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19439         (find_AT_string): Move most logic to ...
19440         (find_AT_string_in_table): ... here.  New.
19441         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19442         add_skeleton_AT_string.  Delete logic.
19443         (output_skeleton_debug_sections): Remove call to
19444         add_top_level_skeleton_die_attrs.
19445         (add_comp_dir_attribute): Move logic to comp_dir_string.
19446         (dwarf2out_init): Initialize debug_str_dwo_section.
19447         (output_indirect_string): Call find_string_form.
19448         (output_indirect_strings): Rewrite.
19449         (prune_unused_types): Empty skeleton_debug_str_hash.
19450         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19451         (dwarf2out_finish):  Call output_indirect_strings.
19453 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19455         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19457 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19459         * rtl.h (struct rtx_def): Add comment for field jump.
19460         (LRA_SUBREG_P): New macro.
19461         * recog.c (register_operand): Check LRA_SUBREG_P.
19462         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19463         stack frame.
19464         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19465         (lra_final_code_change): Skip subreg change for operators.
19466         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19467         if there are no operand changes.
19468         * lra-constraints.c (curr_insn_set): New.
19469         (match_reload): Set LRA_SUBREG_P.
19470         (emit_spill_move): Ditto.
19471         (check_and_process_move): Use curr_insn_set. Process only single
19472         set insns.  Don't initialize sec_mem_p and change_p.
19473         (simplify_operand_subreg): Use LRA_SUBREG_P.
19474         (reg_in_class_p): New function.
19475         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19476         of #ifdef.  Add code to remove cycling.
19477         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19478         non-null disp.  Reload inner instead of disp when base and index
19479         are null.  Try to put lo_sum into register.
19480         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19481         (check_and_process_move): Move code for move cost check to
19482         simple_move_p.  Remove equiv_substitution.
19483         (simple_move_p): New function.
19484         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19485         curr_insn_set.  Call check_and_process_move only for single set
19486         insns.  Use the new function.  Move call of check_and_process_move
19487         after operand equiv substitution and address process.
19489 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19491         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19492         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19493         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19495 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19497         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19499 2013-04-24  Marek Polacek  <polacek@redhat.com>
19501         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19502         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19503         (select_loops_exit_conditions): Likewise.
19504         (number_of_iterations_for_all_loops): Likewise.
19505         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19506         (scev_analysis): Likewise.
19508 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19509             Chao-ying Fu  <fu@mips.com>
19511         * config/mips/micromips.md (jraddiusp): New pattern.
19512         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19513         instruction if possible.
19515 2013-04-24  Alan Modra  <amodra@gmail.com>
19517         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19519 2013-04-24  Julian Brown  <julian@codesourcery.com>
19520             Chung-Lin Tang  <cltang@codesourcery.com>
19522         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19523         dependency behavior in enumeration type DIE generation. Add TODO note
19524         to comments about future DW_FORM_sdata/udata re-work of related code.
19526 2013-04-23  Lawrence Crowl  <crowl@google.com>
19528         * Makefile.in: Update as needed below.
19530         * hash-table.h (class hash_table):
19531         Correct many methods with parameter types compare_type to the correct
19532         value_type.  (Correct code was unlikely to notice the change.)
19533         (hash_table::elements_with_deleted) New.
19534         (class hashtable::iterator): New.
19535         (hashtable::begin()): New.
19536         (hashtable::end()): New.
19537         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19539         * statistics.c (statistics_hashes):
19540         Change type to hash_table.  Update dependent calls and types.
19542         * tree-into-ssa.c (var_infos):
19543         Change type to hash_table.  Update dependent calls and types.
19545         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19546         Change type to hash_table.  Update dependent calls and types.
19548         * tree-ssa-loop-im.c (struct mem_ref.refs):
19549         Change type to hash_table.  Update dependent calls and types.
19551         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19552         Change type to hash_table.  Update dependent calls and types.
19554         * tree-ssa-sccvn.c (vn_tables_s::nary):
19555         Change type to hash_table.  Update dependent calls and types.
19556         (vn_tables_s::phis): Likewise.
19557         (vn_tables_s::references): Likewise.
19559         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19560         (vn_reference_eq): Update parameter and return types.
19562         * tree-ssa-structalias.c (pointer_equiv_class_table):
19563         Change type to hash_table.  Update dependent calls and types.
19564         (location_equiv_class_table): Likewise.
19566         * tree-vect-data-refs.c: Consequential changes for making
19567         peeling a hash_table.
19569         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19570         (destroy_loop_vec_info): Dependent hash_table update.
19572         * tree-vectorizer.h (peeling_htab):
19573         Change type to hash_table.  Update dependent calls and types.
19575 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19577         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19578         to check the register content is equal or not.
19579         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19580         to assign register content record.
19581         * lra-eliminations.c (update_reg_eliminate): Use
19582         lra_update_reg_val_offset to update register content offset.
19583         * lra-int.h (struct lra_reg): Add offset member.
19584         (lra_reg_val_equal_p): New static inline function.
19585         (lra_update_reg_val_offset): New static inline function.
19586         (lra_assign_reg_val): New static inline function.
19587         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19588         to assign register content record.
19589         (initialize_lra_reg_info_element): Initial offset to zero.
19591 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19593         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19594         operands.  Record compression.
19596 2013-04-23  Xinliang David Li  <davidxl@google.com>
19598         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19600 2013-04-23  Richard Biener  <rguenther@suse.de>
19602         PR middle-end/57036
19603         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19604         parameter, only add abnormal goto edges from the copied body
19605         if the call could perform abnormal gotos.
19606         (copy_cfg_body): Adjust.
19608 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19610         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19612 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19614         * coretypes.h (gimple_stmt_iterator): Add struct to make
19615         compatible with C.
19617 2013-04-23  Richard Biener  <rguenther@suse.de>
19619         PR tree-optimization/57026
19620         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19621         from SSA names occuring in abnormal PHI nodes.
19623 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19625         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19627 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19629         * lto/lto.c (print_lto_report_1): Declare early.
19630         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19632 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19634         * common.opt (-flto-report-wpa): Add.
19635         * doc/invoke.texi (-flto-report-wpa): Add.
19636         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19637         (lto_main): dito.
19639 2013-04-22  Xinliang David Li  <davidxl@google.com>
19641         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19642         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19643         * Makefile.in: New dependency
19645             David Daney <ddaney.cavm@gmail.com>
19647         * configure.ac (gcc_cv_as_micromips_support):  Use the
19648         --fatal-warnings option.
19649         * configure: Regenerate.
19651 2013-04-22  Marek Polacek  <polacek@redhat.com>
19653         PR sanitizer/56990
19654         * tsan.c (instrument_expr): Don't instrument expression
19655         in case its size is zero.
19657 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19659         PR target/57032
19660         Revert:
19661         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19663         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19665 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19667         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19668         (gimple_stmt_iterator): New typedef.
19669         * gimple.h (gimple_stmt_iterator): Rename to...
19670         (gimple_stmt_iterator_d): ... This.
19671         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19672         trees be valid for GIMPLE and GENERIC.
19673         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19674         * gimple-fold.c (gimple_fold_call): Call target hook
19675         gimple_fold_builtin.
19676         * hooks.c (hook_bool_gsiptr_false): New.
19677         * hooks.h (hook_bool_gsiptr_false): New.
19678         * target.def (fold_stmt): New.
19679         * doc/tm.texi: Regenerate.
19681 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19683         PR target/57018
19684         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19685         a set sp if no stack realignment.
19687 2013-04-22  Nick Clifton  <nickc@redhat.com>
19689         * config.gcc (tilegx-linux): Extend extra_objs rather than
19690         overwriting it.
19691         (tilepro-linux): Likewise.
19693 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19695         * config/aarch64/aarch64-builtins.c
19696         (CF): Remove.
19697         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19698         (VAR<1-12>): Add MAP parameter.
19699         (BUILTIN_*): Likewise.
19700         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19701         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19702         (aarch64_ushl_n<mode>): Likewise.
19703         (aarch64_sshr_n<mode>): Likewise.
19704         (aarch64_ushr_n<mode>): Likewise.
19705         (aarch64_<maxmin><mode>): Likewise.
19706         (aarch64_sqrt<mode>): Likewise.
19707         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19708         (vshr<q>_n_*): Likewise.
19710 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19712         * config/aarch64/aarch64-builtins.c
19713         (aarch64_simd_builtin_type_mode): Handle SF types.
19714         (sf_UP): Define.
19715         (BUILTIN_GPF): Define.
19716         (aarch64_init_simd_builtins): Handle SF types.
19717         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19718         (frecps): Likewise.
19719         (frecpx): Likewise.
19720         * config/aarch64/aarch64-simd.md
19721         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19722         (aarch64_frecpe<mode>): New.
19723         (aarch64_frecps<mode>): Likewise.
19724         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19725         (v8type): Add frecp<esx>.
19726         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19727         (aarch64_frecps<mode>): Likewise.
19728         * config/aarch64/iterators.md (FRECP): New.
19729         (frecp_suffix): Likewise.
19730         * config/aarch64/arm_neon.h
19731         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19733 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19735         PR target/56995
19736         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19737         (REG_CLASS_NAMES): Idem.
19738         (REG_CLASS_CONTENTS): Idem.
19739         (REGCLASS_HAS_FP_REG): Idem.
19740         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19741         (sh_conditional_register_usage): Idem.
19743 2013-04-21  Jeff Law  <law@redhat.com>
19745         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19746         (ssa_forward_propagate_and_combine): Use it.
19748 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19750         * lra.c: Update the flow chart diagram.
19752 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19754         PR rtl-optimization/56847
19755         * lra-constraints.c (process_alt_operands): Discourage alternative
19756         with non-matche doffsettable memory constraint fro memory with
19757         known offset.
19759 2013-04-19  Richard Biener  <rguenther@suse.de>
19761         PR tree-optimization/56982
19762         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19763         function.
19764         * gimplify.c (gimplify_call_expr): Notice special calls.
19765         (gimplify_modify_expr): Likewise.
19766         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19767         abnormal control flow receivers.
19768         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19769         in the same way as cfun->has_nonlocal_labels.
19770         (gimple_purge_dead_abnormal_call_edges): Likewise.
19771         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19772         receivers start a basic-block.
19774 2013-04-19  Richard Biener  <rguenther@suse.de>
19776         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19777         member ...
19778         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19779         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19780         (SLP_TREE_LOAD_PERMUTATION): Add.
19781         (vect_transform_slp_perm_load): Adjust prototype.
19782         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19783         (vect_free_slp_instance): Likewise.
19784         (vect_create_new_slp_node): Likewise.
19785         (vect_supported_slp_permutation_p): Remove.
19786         (vect_slp_rearrange_stmts): Adjust.
19787         (vect_supported_load_permutation_p): Likewise.  Inline
19788         vect_supported_slp_permutation_p here.
19789         (vect_analyze_slp_instance): Compute load permutations per
19790         slp node instead of per instance.
19791         (vect_get_slp_defs): Adjust.
19792         (vect_transform_slp_perm_load): Likewise.
19793         (vect_schedule_slp_instance): Remove redundant code.
19794         (vect_schedule_slp): Remove hack for PR56270, add it ...
19795         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19796         CSE loads for SLP.  Adjust.
19798 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19800         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19801         spelling in two comments.
19803 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19805         PR target/56797
19806         * config/arm/arm.c (load_multiple_sequence): Require SP
19807         as base register for loads if SP is in the register list.
19809 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19811         PR tree-optimization/56718
19812         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19813         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19814         and made public.  Adjusted all callers.
19815         (ipa_intraprocedural_devirtualization): New function.
19816         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19817         (ipa_intraprocedural_devirtualization): Likewise.
19818         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19820 2013-04-19  Richard Biener  <rguenther@suse.de>
19822         PR tree-optimization/57000
19823         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19825 2013-04-19  Terry Guo  <terry.guo@arm.com>
19827         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19828         Replace with ...
19829         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19830         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19831         (cortex_m4_fmacs): Use new reservations.
19832         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19834 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19836         PR rtl-optimization/56999
19837         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19838         related code.
19839         (lra_coalesce): Remove split_origin_bitmap and related code.
19840         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19841         ranges if necessary.
19843 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19845         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19846         New array.
19847         (ix86_expand_call): Remove clobbered_registers array and use
19848         x86_64_ms_sysv_extra_clobbered_registers instead.
19849         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19850         Declare here.
19851         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19852         predicate.
19853         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19854         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19855         (*call_value_rex64_ms_sysv): Ditto.
19857 2013-04-18  Cary Coutant  <ccoutant@google.com>
19859         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19860         parent instead of die_mark.
19862 2013-04-18  Diego Novillo  <dnovillo@google.com>
19864         * gimple.c (create_gimple_tmp): New.
19865         (get_expr_type): New.
19866         (build_assign): New.
19867         (build_type_cast): New.
19868         * gimple.h (enum ssa_mode): Define.
19869         (gimple_seq_set_location): New.
19870         * asan.c (build_check_stmt): Change some gimple_build_* calls
19871         to use build_assign and build_type_cast.
19873 2013-04-18  Richard Biener  <rguenther@suse.de>
19875         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19876         handle negative step.  Remove redundant checks.
19877         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19878         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19879         for negative step and grouped loads fail to vectorize.
19881 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19883         * emit-rtl.c (reset_insn_used_flags): New function.
19884         (reset_all_used_flags): Use it.
19885         (verify_insn_sharing): New function.
19886         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19888 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19890         PR tree-optimization/56984
19891         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19892         and (x >> M) >= N don't register any assertion if N << M is the
19893         minimum value.
19895 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19897         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19898         do not delete_insn insns that have not yet been emitted, only
19899         unlink them with remove_insn.
19900         * df-scan.c (df_insn_delete): Revert r197492.
19902 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19904         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19905         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19907 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19909         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19910         define_insn_and_split.
19911         (and_scc,ior_scc,negscc): Likewise.
19912         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19914 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19916         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19917         can benefit from using a sequence of LDRD instructions in epilogue
19918         instead of a single LDM instruction.
19920 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19922         PR 45688
19923         * doc/extend.texi: Fix typo.
19925 2013-04-17  Richard Biener  <rguenther@suse.de>
19927         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19928         (vect_build_slp_tree): ... here.
19929         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19930         match.  Remove special-casing of mismatched complex loads.
19931         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19932         re-try the match with swapped commutative operands.
19933         (vect_supported_load_permutation_p): Remove special-casing of
19934         mismatched complex loads.
19935         (vect_analyze_slp_instance): Adjust.
19937 2013-04-17  Richard Biener  <rguenther@suse.de>
19939         PR rtl-optimization/56921
19940         * cfgloop.h (struct loop): Add simple_loop_desc member.
19941         (struct niter_desc): Mark with GTY(()).
19942         (simple_loop_desc): Do not use aux field but simple_loop_desc.
19943         * loop-iv.c (get_simple_loop_desc): Likewise.
19944         (free_simple_loop_desc): Likewise.
19946         Revert
19947         2013-04-16  Richard Biener  <rguenther@suse.de>
19949         PR rtl-optimization/56921
19950         * loop-init.c (pass_rtl_move_loop_invariants): Add
19951         TODO_do_not_ggc_collect to todo_flags_finish.
19952         (pass_rtl_unswitch): Same.
19953         (pass_rtl_unroll_and_peel_loops): Same.
19954         (pass_rtl_doloop): Same.
19956 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
19958         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
19959         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
19960         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
19961         references.
19962         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
19963         * tree-streamer.c (record_common_node): Adjust reference in comment.
19965 2013-04-17  Terry Guo  <terry.guo@arm.com>
19967         * config/arm/cortex-m4.md: Add a new bypass.
19969 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19971         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
19972         New pattern.
19973         (*subs_<optab><mode>_multp2): New pattern.
19974         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19975         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19977 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19979         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
19980         (*subs_mul_imm_<mode>): New pattern.
19982 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
19984         PR target/56948
19985         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
19986         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
19987         (vsx_movti_32bit): Change j->wa to O->wa.
19989 2013-04-16  Richard Biener  <rguenther@suse.de>
19991         PR rtl-optimization/56921
19992         * loop-init.c (pass_rtl_move_loop_invariants): Add
19993         TODO_do_not_ggc_collect to todo_flags_finish.
19994         (pass_rtl_unswitch): Same.
19995         (pass_rtl_unroll_and_peel_loops): Same.
19996         (pass_rtl_doloop): Same.
19998 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20000         * config/arm/arm.c (emit_multi_reg_push): New declaration
20001         for an existing function.
20002         (arm_emit_strd_push): New function.
20003         (arm_expand_prologue): Used here.
20004         (arm_emit_ldrd_pop): New function.
20005         (arm_expand_epilogue): Used here.
20006         (arm_get_frame_offsets): Update condition.
20007         (arm_emit_multi_reg_pop): Add a special case for load of a single
20008         register with writeback.
20010 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20012         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20013         description.
20015 2013-04-16  Richard Biener  <rguenther@suse.de>
20017         PR tree-optimization/56756
20018         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20019         (first_mem_ref_loc): New.
20020         (execute_sm): Place the load temporarily before a previous
20021         access instead of in the latch edge to ensure its SSA dependencies
20022         are defined at points dominating the load.
20024 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20026         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20027         correct fix by moving header and footer insn to the footer of
20028         the merged basic block.  Clear BB_END of the merged-away block.
20030         PR middle-end/43631
20031         * emit-rtl.c (make_note_raw): New function.
20032         (link_insn_into_chain): New static inline function.
20033         (add_insn): Use it.
20034         (add_insn_before, add_insn_after): Factor insn chain linking code...
20035         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20036         using link_insn_into_chain.
20037         (note_outside_basic_block_p): New helper function for emit_note_after
20038         and emit_note_before.
20039         (emit_note_after): Use nobb variant of add_insn_after if the note
20040         should not be contained in a basic block.
20041         (emit_note_before): Use nobb variant of add_insn_before if the note
20042         should not be contained in a basic block.
20043         (emit_note_copy): Use make_note_raw.
20044         (emit_note): Likewise.
20045         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20046         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20047         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20048         the moved barrier the tail of the basic block it follows.
20049         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20051 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20053         PR tree-optimization/56962
20054         * gimple-ssa-strength-reduction.c (record_increment): Only set
20055         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20056         either rhs1 or rhs2 is equal to c->base_expr.
20058 2013-04-15  Richard Biener  <rguenther@suse.de>
20060         PR tree-optimization/56933
20061         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20062         member.
20063         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20064         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20065         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20066         dependence check ...
20067         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20068         ... here.
20069         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20070         GROUP_READ_WRITE_DEPENDENCE.
20072 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20074         * emit-rtl.c (reset_all_used_flags): New function.
20075         (verify_rtl_sharing): Call reset_all_used_flags before and after
20076         performing the checks.
20078 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20080         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20081         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20082         * config/arm/constraints.md (De): New constraint.
20083         * config/arm/neon.md (anddi3_neon): Delete.
20084         (neon_vand<mode>): Expand to standard anddi3 pattern.
20085         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20086         Move earlier in the file.
20087         (neon_inv_logic_op2): Likewise.
20088         (arm_anddi_operand_neon): New predicate.
20090 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20092         * configure.ac (gcc_cv_ld_as_needed): Set
20093         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20094         Use -z ignore, -z record on *-*-solaris2*.
20095         (HAVE_LD_AS_NEEDED): Update comment.
20096         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20097         * configure: Regenerate.
20098         * config.in: Regenerate.
20099         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20100         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20101         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20102         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20103         equivalents.  Fix markup.
20104         * doc/tm.texi: Regenerate.
20106 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20108         * config/i386/i386.opt: New option mstack-protector-guard=.
20109         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20110         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20111         TARGET_SSP_TLS_GUARD.
20112         * config/i386/i386.c (ix86_option_override_internal): Set
20113         ix86_stack_protector_guard.
20114         * config/i386/i386.md (stack_protect_set): Enable for
20115         TARGET_SSP_TLS_GUARD only.
20116         (stack_protect_set_<mode>): Ditto.
20117         (stack_protect_test): Ditto.
20118         (stack_protect_test_<mode>): Ditto.
20119         * doc/invoke.texi (i386 Option): Document.
20121 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20123         PR target/56890
20124         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20125         (S_MODES): Set H_MODE bit.
20126         (SF_MODES): Set only S_MODE and SF_MODE bits.
20127         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20128         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20129         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20130         <MODE_FLOAT>: Likewise.
20132 2013-04-15  Joey Ye  <joey.ye@arm.com>
20134         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20136 2013-04-15  Joey Ye  <joey.ye@arm.com>
20138         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20139         for real far jump.
20140         (thumb_far_jump_used_p): Count instruction size and set
20141         far_jump_used.
20143 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20145         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20146         * resource.c (mark_target_live_regs): Reformat conditional block.
20148 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20150         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20151         notes, they are emitted only just before final.
20152         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20154 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20156         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20157         * cfgrtl.c (delete_insn): Call it here instead.
20158         * lra-spills.c (lra_final_code_change): Use delete_insn.
20159         * haifa-sched.c (sched_remove_insn): Likewise.
20160         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20161         returning to the nop pool.
20162         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20163         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20165 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20167         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20168         * doc/tm.texi: Regenerated.
20170 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20172         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20173         QImode checks.
20175 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20177         * df-core.c (df_find_def): Compare register numbers.
20178         (df_find_use): Likewise.
20180 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20182         PR target/56903
20183         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20184         lra_in_progress for return.
20186 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20188         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20189         define_insn into define_insn_and_split and emit movsicc patterns.
20191 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20193         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20195 2013-04-12  Richard Biener  <rguenther@suse.de>
20197         * tree-pass.h (TODO_do_not_ggc_collect): New.
20198         * passes.c (execute_one_ipa_transform_pass): Honor
20199         TODO_do_not_ggc_collect.
20200         (execute_one_pass): Likewise.
20202         Revert
20203         2013-04-10  Richard Biener  <rguenther@suse.de>
20205         * passes.c (init_optimization_passes): Remove reload pass.
20206         * ira.c (do_reload): Merge into ...
20207         (ira): ... this.
20208         (rest_of_handle_reload): Remove.
20209         (pass_reload): Likewise.
20210         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20211         of reload for vzeroupper pass placement.
20213 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20215         PR tree-optimization/56918
20216         PR tree-optimization/56920
20217         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20218         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20219         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20220         use wide_mul_with_sign method.
20222 2013-04-12  Richard Biener  <rguenther@suse.de>
20224         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20225         not be considered a gimple constant.
20227 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20229         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20230         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20231         operations.
20233 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20234             Jakub Jelinek  <jakub@redhat.com>
20236         * opts.c: Include diagnostic-color.h.
20237         (common_handle_option): Handle OPT_fdiagnostics_color_.
20238         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20239         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20240         (diagnostic-color.o): New.
20241         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20242         (diagnostic_color_rule): New enum.
20243         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20244         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20245         the location string.
20246         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20247         either NULL, or color kind.
20248         * diagnostic-color.c: New file.
20249         * diagnostic-color.h: New file.
20250         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20251         arguments.
20252         * doc/invoke.texi (-fdiagnostics-color): Document.
20253         * pretty-print.h (pp_show_color): Define.
20254         (struct pretty_print_info): Add show_color field.
20255         * diagnostic.c: Include diagnostic-color.h.
20256         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20257         macros.  Colorize error:, warning: etc. strings and also the location
20258         string.
20259         (diagnostic_show_locus): Colorize the caret line.
20260         * pretty-print.c: Include diagnostic-color.h.
20261         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20262         inside of %< %> quotes or quoted through q format modifier.
20264 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20266         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20268 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20270         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20271         code in CC_NZ mode.
20272         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20273         pattern.
20275 2013-04-11  Marek Polacek  <polacek@redhat.com>
20277         PR tree-optimization/48184
20278         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20280 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20282         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20283         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20284         (skip_simple_arithmetic): Tidy up.
20285         * tree.h (skip_simple_constant_arithmetic): Declare.
20287 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20289         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20291 2013-04-11  Richard Biener  <rguenther@suse.de>
20293         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20294         generate vector constants.
20296 2013-04-11  Richard Biener  <rguenther@suse.de>
20298         PR tree-optimization/56878
20299         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20300         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20301         New function.
20302         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20303         Prefer to align the DR with the most invariant base address.
20305 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20307         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20308         comment.
20310 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20312         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20313         floating-point vector comparisons against 0.
20315 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20317         PR tree-optimization/56899
20318         * fold-const.c (extract_muldiv_1): Apply distributive law
20319         only if TYPE_OVERFLOW_WRAPS (ctype).
20321 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20323         PR target/56124
20324         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20325         loading has side effect.
20327 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20329         * config/sparc/sparc.c: Include tree-pass.h.
20330         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20331         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20332         head of file.  Change return type.  Split off gate function.
20333         (sparc_gate_work_around_errata): New function.
20334         (pass_work_around_errata): New pass definition.
20335         (insert_pass_work_around_errata) New pass insert definition to
20336         insert pass_work_around_errata just after delayed-branch scheduling.
20337         (sparc_option_override): Insert the pass.
20338         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20340 2013-04-10  David S. Miller  <davem@davemloft.net>
20342         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20343         or -mcpu=hypersparc.
20345         * target.def (cstore_mode): New hook.
20346         * target.h: Include insn-codes.h
20347         * targhooks.c: Likewise.
20348         (default_cstore_mode): New function.
20349         * targhooks.h: Declare it.
20350         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20351         * doc/tm.texi: Rebuild.
20352         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20353         target hook, rather than inspecting the insn_data.
20354         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20355         (TARGET_CSTORE_MODE): Redefine.
20356         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20357         result patterns.
20358         * config/sparc/predicates.md (cstore_result_operand): New special
20359         predicate.
20360         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20361         Use it for operand 0.
20362         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20363         (*snesi_special): Likewise.
20364         (*snesi_zero): Likewise.
20365         (*seqsi_zero): Likewise.
20366         (*sltu_insn): Likewise.
20367         (*sgeu_insn): Likewise.
20368         (*seqdi_special): Make operand 0 and comparison operation be of
20369         DImode.
20370         (*snedi_special): Likewise.
20371         (*snedi_special_vis3): Likewise.
20372         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20373         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20374         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20375         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20376         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20377         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20378         (*sltu_extend_sp64): Likewise.
20379         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20380         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20381         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20382         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20383         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20385 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20387         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20388         (aarch64_start_file): Use the new function.
20390 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20391             Jason Merrill  <jason@redhat.com>
20393         * common.opt: Add -gdwarf.
20394         * opts.c (common_handle_option): Handle it.
20395         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20397 2013-04-10  Richard Biener  <rguenther@suse.de>
20399         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20400         (execute_one_ipa_transform_pass): But unconditionally here.
20401         (execute_one_pass): And here.
20402         (init_optimization_passes): Remove reload pass.
20403         * tree-pass.h (TODO_ggc_collect): Remove.
20404         (pass_reload): Likewise.
20405         * ira.c (do_reload): Merge into ...
20406         (ira): ... this.
20407         (rest_of_handle_reload): Remove.
20408         (pass_reload): Likewise.
20409         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20410         of reload for vzeroupper pass placement.
20411         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20412         and todo_flags_finish of all passes.
20414 2013-04-10  Richard Biener  <rguenther@suse.de>
20416         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20417         first_const_oprnd field, rename first_def_type to first_op_type.
20418         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20419         (vect_get_and_check_slp_defs): Always use the type of the
20420         operand.  Allow mixed vect_external_def, vect_constant_def types.
20421         (vect_get_constant_vectors): Handle mixed vect_external_def,
20422         vect_constant_def types.
20424 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20426         PR tree-optimization/55524
20427         * tree-ssa-math-opts.c
20428         (convert_mult_to_fma): Don't use an fms construct
20429         when we don't have an fms operation, but fnma, and it looks
20430         likely that we'll be able to use the latter.
20432 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20434         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20435         function.
20436         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20437         inline fail caused by overwritable functions.
20439 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20441         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20442         unnecessary bits in the constant power of two case.
20444 2013-04-10  Richard Biener  <rguenther@suse.de>
20446         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20447         broken code swapping operands.
20448         (vect_build_slp_tree): Do not compute load permutations here.
20449         (vect_analyze_slp_instance): Compute load permutations here,
20450         after building the SLP tree.
20452 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20454         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20455         of next/prev_real_insn.
20457 2013-04-09  Jan Hubicka  <jh@suse.cz>
20459         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20460         Drop aliased parameter.
20461         (function_and_variable_visibility): Do not handle alias pairs.
20462         * cgraph.c (varpool_externally_visible_p): Update prototype.
20463         * varpool.c (varpool_add_new_variable): Update.
20465 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20467         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20469 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20471         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20473         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20475 2013-04-09  Marek Polacek  <polacek@redhat.com>
20477         PR tree-optimization/48762
20478         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20480 2013-04-09  Richard Biener  <rguenther@suse.de>
20482         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20483         dealing with cost.
20484         (vect_build_slp_tree): Likewise.
20485         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20486         calculating the cost of a SLP instance.
20487         (vect_analyze_slp_instance): Use it from here, after building
20488         the SLP tree.
20490 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20492         PR middle-end/56883
20493         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20494         expand_omp_for_static_chunk): Use simple_p = true in
20495         force_gimple_operand_gsi calls when assigning to addressable decls.
20497 2013-04-09  Jeff Law  <law@redhat.com>
20499         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20500         when the boolean was created by converting a wider object which
20501         had a boolean range.
20503 2013-04-09  Richard Biener  <rguenther@suse.de>
20505         * tree-vectorizer.h (slp_void_p): Remove.
20506         (slp_tree): Typedef before _slp_tree declaration.
20507         (struct _slp_tree): Use a vector of slp_tree as children.
20508         (vect_get_place_in_interleaving_chain): Remove.
20509         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20510         Move ...
20511         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20512         and make static.
20513         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20514         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20515         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20516         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20517         Use slp_node instead of slp_void_p and adjust.
20519 2013-04-09  Richard Biener  <rguenther@suse.de>
20521         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20522         work that is not necessary.
20524 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20526         PR tree-optimization/56854
20527         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20528         forward into clobber stmts if it would change MEM_REF lhs into
20529         non-MEM_REF.
20531 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20533         * tree.c (type_hash_lookup, type_hash_add): Make static.
20534         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20536 2013-04-09  Richard Biener  <rguenther@suse.de>
20538         * tree.h (unsave_expr_now): Remove.
20539         * tree-inline.c (mark_local_for_remap_r): Remove.
20540         (unsave_expr_1): Likewise.
20541         (unsave_r): Likewise.
20542         (unsave_expr_now): Likewise.
20543         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20544         (propagate_tree_value): Likewise.
20546 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20548         * doc/rtl.texi (sequence): Rewrite documentation to match the
20549         current use of SEQUENCE rtl objects.
20550         * rtl.def (SEQUENCE): Likewise.
20552         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20553         Update documentation.
20554         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20555         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20557         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20559 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20561         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20562         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20563         methods.
20564         (estimate_edge_size_and_time): Add comment to suggest using rounding
20565         methods.
20566         (estimate_node_size_and_time): Ditto.
20567         (remap_edge_change_prob): Use helper rounding divide methods.
20568         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20569         (gimple_mod_pow2_value_transform): Ditto.
20570         (gimple_mod_subtract_transform): Ditto.
20571         (gimple_ic_transform): Ditto.
20572         (gimple_stringops_transform): Ditto.
20573         * stmt.c (conditional_probability): Ditto.
20574         (emit_case_dispatch_table): Ditto.
20575         * lto-cgraph.c (merge_profile_summaries): Ditto.
20576         * tree-optimize.c (execute_fixup_cfg): Ditto.
20577         * cfgcleanup.c (try_forward_edges): Ditto.
20578         * cfgloopmanip.c (scale_loop_profile): Ditto.
20579         (loopify): Ditto.
20580         (duplicate_loop_to_header_edge): Ditto.
20581         (lv_adjust_loop_entry_edge): Ditto.
20582         * tree-vect-loop.c (vect_transform_loop): Ditto.
20583         * profile.c (compute_branch_probabilities): Ditto.
20584         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20585         * lto-streamer-in.c (input_cfg): Ditto.
20586         * gimple-streamer-in.c (input_bb): Ditto.
20587         * ipa-cp.c (update_profiling_info): Ditto.
20588         (update_specialized_profile): Ditto.
20589         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20590         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20591         rounding methods.
20592         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20593         (compute_trg_info): Ditto.
20594         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20595         (purge_dead_edges): Ditto.
20596         * loop-unswitch.c (unswitch_loop): Ditto.
20597         * cgraphclones.c (cgraph_clone_edge): Ditto.
20598         (cgraph_clone_node): Ditto.
20599         * tree-inline.c (copy_bb): Ditto.
20600         (copy_edges_for_bb): Ditto.
20601         (initialize_cfun): Ditto.
20602         (copy_cfg_body): Ditto.
20603         (expand_call_inline): Ditto.
20605 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20607         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20608         TARGET_CYGWIN64 by TARGET_64BIT.
20610 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20612         * config/epiphany/epiphany.md (GPR_1): New constant.
20613         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20614         * config/epiphany/epiphany.c (gen_compare_reg):
20615         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20616         is already in place.
20617         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20618         Don't require being called during rtl expansion; If y operlaps r0,
20619         return 0.
20620         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20621         (epiphany_expand_epilogue): Likewise.
20623         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20624         Don't use CC_FPmode for ORDERED / UNORDERED.
20625         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20627         * config/epiphany/constraints.md (CnL): New constraint.
20628         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20629         * config/epiphany/predicates.md (add_operand): Allow 1024.
20631         * config/epiphany/epiphany.md (logical_op): New code iterator.
20632         (op_mnc): New code attribute.
20633         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20634         (mov_f+1, mov_f+2): New peephole2 patterns.
20636         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20637         (cstoresi4): Also allow re-use of zero result when doing a NE
20638         comparison to a non-zero operand.
20639         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20641         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20642         Use gen_addsi3_i / gen_subsi3_i.
20644 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20646         PR c++/34949
20647         PR c++/50243
20648         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20649         contain anything but clobbers, at most one __builtin_stack_restore,
20650         optionally debug stmts and final resx, and if it has at least one
20651         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20652         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20653         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20654         which isn't defaut definition, remove them.
20655         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20656         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20657         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20658         with MEM_REF LHS with SSA_NAME address.
20660 2013-04-08  Jeff Law  <law@redhat.com>
20662         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20664 2013-04-08  Richard Biener  <rguenther@suse.de>
20666         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20667         extra newline.
20668         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20669         determined vector type.
20670         (vect_analyze_data_refs): Likewise.
20671         (vect_get_new_vect_var): Adjust.
20672         (vect_create_destination_var): Preserve SSA name versions.
20673         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20674         not dump anything here.
20676 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20678         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20679         Add member lr_slot_known.
20680         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20681         if necessary.
20682         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20683         Remove code that sets lr_slot_offset according to what a previous
20684         version of epiphany_emit_save_restore used to do.
20685         (epiphany_emit_save_restore): When doing an lr save or restore,
20686         set/verify lr_slot_known and lr_slot_offset.
20688 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20690         PR target/54338
20691         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20692         in ALL_REGS.
20694 2013-04-08  Richard Biener  <rguenther@suse.de>
20696         * alias.c (find_base_term): Fix thinko in previous change.
20698 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20700         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20701         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20702         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20703         if possible to compute val.
20704         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20705         For QImode integers don't require anything about precision.  Use
20706         const_with_all_bytes_same to find out if the constant doesn't have
20707         repeated bytes in it.
20709 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20711         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20712         within mode size.
20714 2013-04-08  Marek Polacek  <polacek@redhat.com>
20716         PR rtl-optimization/48182
20717         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20718         value to 1.
20720 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20722         PR target/55487
20723         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20724         nuses, make sure we have a label.
20726 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20728         PR target/56843
20729         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20730         (rs6000_emit_swdiv_low_precision): Remove.
20731         (rs6000_emit_swdiv): Rewrite to handle between one and four
20732         iterations of Newton-Raphson generally; modify required number of
20733         iterations for some cases.
20734         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20736 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20738         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20739         set-but-unused variable.
20741         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20742         basic blocks of released function bodies garbage-collectable.
20744         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20745         (struct rtl_opt_pass): Add TODO_df_finish.
20747         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20749 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20751         * config/arm/constraints.md (q): New constraint.
20752         * config/arm/ldrdstrd.md: New file.
20753         * config/arm/arm.md (ldrdstrd.md) New include.
20754         (arm_movdi): Use "q" instead of "r" constraint
20755         for double-word memory access.
20756         (movdf_soft_insn): Likewise.
20757         * config/arm/vfp.md (movdi_vfp): Likewise.
20758         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20759         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20760         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20761         (mem_ok_for_ldrd_strd): Likewise.
20762         (output_move_double): Update assertion.
20764 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20766         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20768 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20770         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20771         define_insn_and_split.
20772         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20774 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20776         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20777         define_insn_and_split.
20778         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20779         (shiftsi3_compare): New pattern.
20780         (rrx): New pattern.
20781         * config/arm/unspecs.md (UNSPEC_RRX): New.
20783 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20785         * config/arm/arm.md (negdi_extendsidi): New pattern.
20786         (negdi_zero_extendsidi): Likewise.
20788 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20790         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20791         define_insn_and_split.
20792         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20793         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20795 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20797         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20798         define_insn_and_split.
20799         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20800         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20802 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20804         * config/arm/arm.md (subsi3_carryin): New pattern.
20805         (subsi3_carryin_const): Likewise.
20806         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20807         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20809 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20811         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20813 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20815         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20816         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20818 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20820         * config/arm/arm.c (arm_expand_builtin): Change fcode
20821         type to unsigned int.
20823 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20825         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20827 2013-04-04  Ian Lance Taylor  <iant@google.com>
20829         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20830         language standard.
20832 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20834         PR middle-end/56729
20835         * df-scan.c (df_insn_delete): Disable failing assert.
20837 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20839         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20840         New function prototype.
20841         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20842         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20843         (arm_builtin_vectorized_function): New function.
20845 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20847         * config/arm/arm_neon_builtins.def: New file.
20848         * config/arm/arm.c (neon_builtin_data): Move contents to
20849         arm_neon_builtins.def.
20850         (enum arm_builtins): Include neon builtin definitions.
20851         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20852         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20854 2013-04-04  Marek Polacek  <polacek@redhat.com>
20856         PR tree-optimization/48186
20857         * predict.c (maybe_hot_frequency_p): Return false if
20858         HOT_BB_FREQUENCY_FRACTION is 0.
20859         (cgraph_maybe_hot_edge_p): Likewise.
20861 2013-04-04  Richard Biener  <rguenther@suse.de>
20863         PR tree-optimization/56826
20864         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20865         more accurately.
20867 2013-04-04  Richard Biener  <rguenther@suse.de>
20869         PR tree-optimization/56213
20870         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20871         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20873 2013-04-04  Richard Biener  <rguenther@suse.de>
20875         PR tree-optimization/56837
20876         * tree-loop-distribution.c (classify_partition): For non-zero
20877         values require that the value has the same precision as its
20878         mode to be useful as memset value.
20880 2013-04-03  Nick Clifton  <nickc@redhat.com>
20882         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20883         (fmssf4): Use fmsf.s on E3V5 architectures.
20884         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20885         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20887 2013-04-03  Jeff Law  <law@redhat.com>
20889         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20890         (lra-eliminations.o): Likewise.
20892 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20894         * gcov-io.c (compute_working_sets): Moved most of body of old
20895         compute_working_sets here from profile.c.
20896         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20897         (gcov_working_set_t): Moved typedef here from basic-block.h
20898         (compute_working_set): Declare.
20899         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20900         (get_working_sets): Renamed from compute_working_set,
20901         replace most of body with call to new compute_working_sets.
20902         (get_exec_counts): Replace call to compute_working_sets
20903         to get_working_sets.
20904         * profile.h (get_working_sets): Renamed from compute_working_set.
20905         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20906         to get_working_sets.
20907         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20908         * gcov-dump.c (dump_working_sets): New function.
20910 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20912         * hwint.c (sext_hwi, zext_hwi): New functions.
20913         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20914         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20915         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20916         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20917         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20918         (sext_hwi, zext_hwi): New functions.
20920 2013-04-03  Jeff Law  <law@redhat.com>
20922         PR tree-optimization/56799
20923         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20924         back test for widening conversion erroneously dropped in prior change.
20926 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20928         PR target/56809
20929         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20930         instead of next_real_insn.
20932 2013-04-03  Marek Polacek  <polacek@redhat.com>
20934         PR sanitizer/55702
20935         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20937 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20939         PR target/56809
20940         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20941         next_real_insn.
20942         (thumb1_output_casesi): Likewise.
20943         (thumb2_output_casesi): Likewise.
20945 2013-04-03  Richard Biener  <rguenther@suse.de>
20947         PR tree-optimization/56817
20948         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
20949         Split out ...
20950         (tree_unroll_loops_completely_1): ... new function to manually
20951         walk the loop tree, properly defering outer loops of unrolled
20952         loops to later iterations.
20954 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20956         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
20957         (vectorizable_load): Likewise.
20958         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
20959         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
20961 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20963         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
20964         BIT_FIELD_REF.
20966 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
20968         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
20970 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
20972         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
20974 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20976         PR tree-optimization/56790
20977         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
20978         folding.
20980 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20982         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
20983         Handle VEC_MERGE.
20984         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
20985         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
20986         equal arguments.
20988 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
20990         PR c/19449
20991         * tree.h (force_folding_builtin_constant_p): New decl.
20992         * builtins.c (force_folding_builtin_constant_p): New variable.
20993         (fold_builtin_constant_p): Fold immediately also if
20994         force_folding_builtin_constant_p.
20996 2013-04-03  Richard Biener  <rguenther@suse.de>
20998         PR tree-optimization/56812
20999         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21000         DRs of the same interleaving chain are independent.
21002 2013-04-02  Jason Merrill  <jason@redhat.com>
21004         * gdbinit.in (pbb): Use debug fn.
21006 2013-04-02  Lawrence Crowl  <crowl@google.com>
21008         * sese.h (struct ivtype_map_elt_s): Remove unused.
21009         (extern debug_ivtype_map): Remove unused.
21010         (extern eq_ivtype_map_elts): Remove unused.
21011         * sese.c (debug_ivtype_map): Removed unused.
21012         (debug_ivtype_map_1): Removed unused.
21013         (debug_ivtype_elt): Remove unused.
21014         (eq_ivtype_map_elts): Remove unused.
21017 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21019         PR target/52790
21020         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21021         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21022         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21023         function.
21024         (legitimize_pe_coff_symbol): Likewise.
21025         (is_imported_p): New helper-function.
21026         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21027         for Windows x64 targets.
21028         (ix86_expand_prologue): Optimize for pe-coff targets.
21029         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21030         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21031         medium/large code-model.
21032         (legitimize_pic_address): Likewise.
21033         (legitimize_tls_address): Likewise.
21034         (ix86_expand_call): Likewise.
21035         (x86_output_mi_thunk): Likewise.
21036         (get_dllimport_decl): Add new beimport argument.
21037         (construct_plt_address): Don't assert for x64 pe-coff targets.
21038         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21039         targets.
21040         (SYMBOL_FLAG_STUBVAR): New macro.
21041         (SYMBOL_REF_STUBVAR_P): Likewise.
21042         * config/i386/winnt.c (stub_list): New structure.
21043         (stub_head): New local variable.
21044         (i386_pe_record_stub): New function.
21045         (i386_pe_file_end): Emit refptr-stubs.
21047 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21049         PR rtl-optimization/56745
21050         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21051         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21053         PR c++/34949
21054         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21055         and both of them are MEM_REFs, just compare first argument for
21056         equality and attempt to deal even with differing offsets.
21058         PR c++/34949
21059         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21060         of gimple_clobber_p to be MEM_REF.
21061         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21062         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21063         after gimplification.
21064         * asan.c (get_mem_ref_of_assignment): Don't instrument
21065         gimple_clobber_p stmts.
21066         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21067         gimple_clobber_p stmt if they have MEM_REF lhs and
21068         are dead because of another gimple_clobber_p stmt.
21069         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21070         gimple_clobber_p stmts like debug stmts.
21071         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21072         that refer to unused VAR_DECLs or uninitialized values.
21073         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21074         gimple_clobber_p stmts if they refer to removed parameters.
21075         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21076         formatting.
21078 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21080         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21081         using SWI48 mode attribute.
21083 2013-04-02  Wei Mi  <wmi@google.com>
21085         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21086         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21087         *<rotate_insn><mode>3_mask in i386.md.
21089 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21091         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21093 2013-04-02  Richard Biener  <rguenther@suse.de>
21095         PR tree-optimization/56778
21096         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21097         Runtime alias tests are not supported for gather loads.
21098         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21099         stmts referenced from SSA operands before updating SSA form.
21101 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21102             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21104         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21105         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21106         * config/arm/cortex-a53.md: New file.
21107         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21108         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21109         * config/arm/arm.c (arm_issue_rate): Likewise.
21110         * config/arm/arm-tune.md: Regenerate
21111         * config/arm/arm-tables.opt: Regenerate.
21112         * config/arm/arm-cores.def: Add cortex-a53.
21114 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21116         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21117         non-static link.
21119 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21121         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21122         scalar load/store operations using B/H registers.
21123         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21125 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21127         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21128         scalar move.
21129         * config/aarch64/aarch64.c
21130         (aarch64_simd_scalar_immediate_valid_for_move): New.
21131         * config/aarch64/aarch64-protos.h
21132         (aarch64_simd_scalar_immediate_valid_for_move): New.
21133         * config/aarch64/constraints.md (Dh, Dq): New.
21134         * config/aarch64/iterators.md (hq): New.
21136 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21138         * reorg.c (get_branch_condition): Deal with conditional returns.
21139         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21141 2013-04-01  Wei Mi  <wmi@google.com>
21143         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21144         Truncate operand 2 using %b asm operand modifier.
21145         (*<shift_insn><mode>3_mask): Ditto.
21146         (*<rotate_insn><mode>3_mask): Ditto.
21148 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21150         PR middle-end/56798
21151         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21153 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21155         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21156         of next_real_insn.
21157         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21159 2013-03-30  Lawrence Crowl  <crowl@google.com>
21161         * dse.c (clear_alias_sets): Remove never set.
21162         (disqualified_clear_alias_sets): Remove never set.
21163         (clear_alias_mode_pool): Remove never set.
21164         (dse_step0): Remove condition that is never true.
21165         (canon_address): Remove condition that is never true.
21166         (dse_step7): Remove condition that is never true.
21167         (rest_of_handle_dse): Remove condition that is never true.
21168         (rest_of_handle_dse::did_global): Remove never read from above.
21169         (dse_step2_spill): Remove never called from above.
21170         (dse_step5_spill): Remove never called from above.
21172 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21174         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21175         JUMP_TABLE_DATA changes.
21176         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21177         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21178         (Insns) <jump_table_data>: New entry.
21179         * doc/tm.texi: Regenerate.
21181         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21183         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21184         for table jump at the end of a basic block using tablejump_p.
21185         * targhooks.c (default_invalid_within_doloop): Likewise.
21186         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21187         target hook implementation that is identical to the default hook.
21188         (rs6000_invalid_within_doloop): Remove.
21190         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21191         unused variable from tablejump_p call.
21193         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21194         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21195         (INSN_DELETED_P): Likewise.
21196         (emit_jump_table_data): New prototype.
21197         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21198         after 4th as unused.
21199         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21200         * sched-vis.c (print_insn): Likewise.
21201         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21202         insn for compatibility with back ends that use next_active_insn to
21203         identify jump table data.
21204         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21205         (remove_insn): Likewise.
21206         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21207         to be emitted.
21208         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21209         (emit_jump_table_data): New function.
21211         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21212         basic block, a JUMP_TABLE_DATA never is.
21213         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21214         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21215         off from code handling real insns.
21216         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21217         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21218         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21219         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21220         is not a NONDEBUG_INSN_P.
21221         * ira-costs.c (scan_one_insn): Likewise.
21222         * jump.c (mark_all_labels): Likewise.
21223         (mark_jump_label_1): Likewise.
21224         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21225         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21226         a basic block.
21227         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21228         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21229         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21230         JUMP_TABLE_DATA_P insns.
21231         (calculate_elim_costs_all_insns): Likewise.
21232         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21233         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21234         (delete_output_reload): Code style fixups.
21235         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21236         insn flags on this non-insn.
21237         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21238         as scheduling barriers, for pre-change compatibility.
21239         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21240         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21242         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21243         redundant JUMP_TABLE_DATA_P test.
21244         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21245         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21246         (frv_for_each_packet): Likewise.
21247         * config/i386/i386.c (min_insn_size): Likewise.
21248         (ix86_avoid_jump_mispredicts): Likewise.
21249         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21250         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21251         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21252         (mips16_insn_length): Robustify.
21253         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21254         (mips16_split_long_branches): Likewise.
21255         * config/pa/pa.c (pa_combine_instructions): Likewise.
21256         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21257         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21258         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21259         as contributing to pool range lengths.
21260         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21261         Remove redundant JUMP_TABLE_DATA_P test.
21262         (sh_loop_align): Likewise.
21263         (split_branches): Likewise.
21264         (sh_insn_length_adjustment): Likewise.
21265         * config/spu/spu.c (get_branch_target): Likewise.
21267 2013-03-29  Jan Hubicka  <jh@suse.cz>
21269         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21270         gcov streaming; stream hot bb threshold to ltrans.
21271         * predict.c (get_hot_bb_threshold): Break out from ....
21272         (maybe_hot_count_p): ... here.
21273         (set_hot_bb_threshold): New function.
21274         * lto-section-in.c (lto_section_name): Add profile.
21275         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21276         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21277         and data-streamer.h
21278         (histogram_entry): New structure.
21279         (histogram, histogram_pool): New global vars.
21280         (histogram_hash): New structure.
21281         (histogram_hash::hash): New method.
21282         (histogram_hash::equal): Likewise.
21283         (account_time_size): New function.
21284         (cmp_counts): New function.
21285         (dump_histogram): New function.
21286         (ipa_profile_generate_summary): New function.
21287         (ipa_profile_write_summary): New function.
21288         (ipa_profile_read_summary): New function.
21289         (ipa_profile): Decide on threshold.
21290         (pass_ipa_profile): Add ipa_profile_write_summary and
21291         ipa_profile_read_summary.
21292         * Makefile.in (ipa.o): Update dependencies.
21293         * lto-streamer.h (LTO_section_ipa_profile): New section.
21295 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21297         * tree.h (VAR_P): New.
21299 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21301         PR lto/56777
21302         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21304 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21306         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21307         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21308         (control_flow_insn_p): Likewise.
21309         * cfgrtl.c (duplicate_insn_chain): Likewise.
21310         * final.c (get_attr_length_1): Likewise.
21311         (shorten_branches): Likewise.
21312         (final_scan_insn): Likewise.
21313         * function.c (instantiate_virtual_regs): Likewise.
21314         * gcse.c (insert_insn_end_basic_block): Likewise.
21315         * ira-costs.c (scan_one_insn): Likewise.
21316         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21317         * lra.c (check_rtl): Likewise.
21318         * reload1.c (elimination_costs_in_insn): Likewise.
21319         * reorg.c (follow_jumps): Likewise.
21321         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21322         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21323         (thumb_far_jump_used_p): Likewise.
21324         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21325         (workaround_speculation): Likewise.
21326         (add_sched_insns_for_speculation): Likewise.
21327         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21328         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21329         (frv_for_each_packet): Likewise.
21330         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21331         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21332         (final_emit_insn_group_barriers): Likewise.
21333         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21334         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21335         (mips16_insn_length): Likewise.
21336         * config/pa/pa.c (pa_reorg): Likewise.
21337         (pa_combine_instructions): Likewise.
21338         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21339         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21340         (sh_reorg): Likewise.
21341         (split_branches): Likewise.
21342         * config/spu/spu.c (get_branch_target): Likewise.
21344         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21345         JUMP_TABLE_DATA_P.
21347 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21349         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21350         Fix declaration name.
21352 2013-03-28  Lawrence Crowl  <crowl@google.com>
21354         * graphds.h (struct graph.indicies): Remove unused.
21355         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21356         (SCOP_ORIGINAL_PDDRS): Remove unused.
21357         * sese.h (extern insert_loop_close_phis): Removed unused.
21358         (extern insert_guard_phis): Removed unused.
21359         (extern ivtype_map_elt_info): Removed unused.
21360         (new_ivtype_map_elt): Removed unused.
21361         * sese.c (ivtype_map_elt_info): Removed unused.
21363 2013-03-28  Lawrence Crowl  <crowl@google.com>
21365         * Makefile.in: Add several missing include dependences.
21366         (DUMPFILE_H): New.
21367         (test-dump.o): New.  This object is not added to any executable,
21368         but is present for ad-hoc testing.
21369         * bitmap.c
21370         (debug (const bitmap_head_def &)): New.
21371         (debug (const bitmap_head_def *)): New.
21372         * bitmap.h
21373         (extern debug (const bitmap_head_def &)): New.
21374         (extern debug (const bitmap_head_def *)): New.
21375         * cfg.c
21376         (debug (edge_def &)): New.
21377         (debug (edge_def *)): New.
21378         * cfghooks.c
21379         (debug (basic_block_def &)): New.
21380         (debug (basic_block_def *)): New.
21381         * dumpfile.h
21382         (dump_node (const_tree, int, FILE *)): Correct source file.
21383         * dwarf2out.c
21384         (debug (die_struct &)): New.
21385         (debug (die_struct *)): New.
21386         * dwarf2out.h
21387         (extern debug (die_struct &)): New.
21388         (extern debug (die_struct *)): New.
21389         * gimple-pretty-print.c
21390         (debug (gimple_statement_d &)): New.
21391         (debug (gimple_statement_d *)): New.
21392         * gimple-pretty-print.h
21393         (extern debug (gimple_statement_d &)): New.
21394         (extern debug (gimple_statement_d *)): New.
21395         * ira-build.c
21396         (debug (ira_allocno_copy &)): New.
21397         (debug (ira_allocno_copy *)): New.
21398         (debug (ira_allocno &)): New.
21399         (debug (ira_allocno *)): New.
21400         * ira-int.h
21401         (extern debug (ira_allocno_copy &)): New.
21402         (extern debug (ira_allocno_copy *)): New.
21403         (extern debug (ira_allocno &)): New.
21404         (extern debug (ira_allocno *)): New.
21405         * ira-lives.c
21406         (debug (live_range &)): New.
21407         (debug (live_range *)): New.
21408         * lra-int.h
21409         (debug (lra_live_range &)): New.
21410         (debug (lra_live_range *)): New.
21411         * lra-lives.c
21412         (debug (lra_live_range &)): New.
21413         (debug (lra_live_range *)): New.
21414         * omega.c
21415         (debug (omega_pb_d &)): New.
21416         (debug (omega_pb_d *)): New.
21417         * omega.h
21418         (extern debug (omega_pb_d &)): New.
21419         (extern debug (omega_pb_d *)): New.
21420         * print-rtl.c
21421         (debug (const rtx_def &)): New.
21422         (debug (const rtx_def *)): New.
21423         * print-tree.c
21424         (debug_tree (tree): Move within file.
21425         (debug_raw (const tree_node &)): New.
21426         (debug_raw (const tree_node *)): New.
21427         (dump_tree_via_hooks (const tree_node *, int)): New.
21428         (debug (const tree_node &)): New.
21429         (debug (const tree_node *)): New.
21430         (debug_verbose (const tree_node &)): New.
21431         (debug_verbose (const tree_node *)): New.
21432         (debug_head (const tree_node &)): New.
21433         (debug_head (const tree_node *)): New.
21434         (debug_body (const tree_node &)): New.
21435         (debug_body (const tree_node *)): New.
21436         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21437         (debug (vec<tree, va_gc> &)): New.
21438         (debug (vec<tree, va_gc> *)): New.
21439         * rtl.h
21440         (extern debug (const rtx_def &)): New.
21441         (extern debug (const rtx_def *)): New.
21442         * sbitmap.c
21443         (debug_raw (simple_bitmap_def &)): New.
21444         (debug_raw (simple_bitmap_def *)): New.
21445         (debug (simple_bitmap_def &)): New.
21446         (debug (simple_bitmap_def *)): New.
21447         * sbitmap.h
21448         (extern debug (simple_bitmap_def &)): New.
21449         (extern debug (simple_bitmap_def *)): New.
21450         (extern debug_raw (simple_bitmap_def &)): New.
21451         (extern debug_raw (simple_bitmap_def *)): New.
21452         * sel-sched-dump.c
21453         (debug (vinsn_def &)): New.
21454         (debug (vinsn_def *)): New.
21455         (debug_verbose (vinsn_def &)): New.
21456         (debug_verbose (vinsn_def *)): New.
21457         (debug (expr_def &)): New.
21458         (debug (expr_def *)): New.
21459         (debug_verbose (expr_def &)): New.
21460         (debug_verbose (expr_def *)): New.
21461         (debug (vec<rtx> &)): New.
21462         (debug (vec<rtx> *)): New.
21463         * sel-sched-dump.h
21464         (extern debug (vinsn_def &)): New.
21465         (extern debug (vinsn_def *)): New.
21466         (extern debug_verbose (vinsn_def &)): New.
21467         (extern debug_verbose (vinsn_def *)): New.
21468         (extern debug (expr_def &)): New.
21469         (extern debug (expr_def *)): New.
21470         (extern debug_verbose (expr_def &)): New.
21471         (extern debug_verbose (expr_def *)): New.
21472         (extern debug (vec<rtx> &)): New.
21473         (extern debug (vec<rtx> *)): New.
21474         * sel-sched-ir.h
21475         (_list_iter_cond_expr): Make inline instead of static.
21476         * sreal.c
21477         (debug (sreal &)): New.
21478         (debug (sreal *)): New.
21479         * sreal.h
21480         (extern debug (sreal &)): New.
21481         (extern debug (sreal *)): New.
21482         * tree.h
21483         (extern debug_raw (const tree_node &)): New.
21484         (extern debug_raw (const tree_node *)): New.
21485         (extern debug (const tree_node &)): New.
21486         (extern debug (const tree_node *)): New.
21487         (extern debug_verbose (const tree_node &)): New.
21488         (extern debug_verbose (const tree_node *)): New.
21489         (extern debug_head (const tree_node &)): New.
21490         (extern debug_head (const tree_node *)): New.
21491         (extern debug_body (const tree_node &)): New.
21492         (extern debug_body (const tree_node *)): New.
21493         (extern debug (vec<tree, va_gc> &)): New.
21494         (extern debug (vec<tree, va_gc> *)): New.
21495         * tree-cfg.c
21496         (debug (struct loop &)): New.
21497         (debug (struct loop *)): New.
21498         (debug_verbose (struct loop &)): New.
21499         (debug_verbose (struct loop *)): New.
21500         * tree-dump.c: Add header dependence.
21501         * tree-flow.h
21502         (extern debug (struct loop &)): New.
21503         (extern debug (struct loop *)): New.
21504         (extern debug_verbose (struct loop &)): New.
21505         (extern debug_verbose (struct loop *)): New.
21506         * tree-data-ref.c
21507         (debug (data_reference &)): New.
21508         (debug (data_reference *)): New.
21509         (debug (vec<data_reference_p> &)): New.
21510         (debug (vec<data_reference_p> *)): New.
21511         (debug (vec<ddr_p> &)): New.
21512         (debug (vec<ddr_p> *)): New.
21513         * tree-data-ref.h
21514         (extern debug (data_reference &)): New.
21515         (extern debug (data_reference *)): New.
21516         (extern debug (vec<data_reference_p> &)): New.
21517         (extern debug (vec<data_reference_p> *)): New.
21518         (extern debug (vec<ddr_p> &)): New.
21519         (extern debug (vec<ddr_p> *)): New.
21520         * tree-ssa-alias.c
21521         (debug (pt_solution &)): New.
21522         (debug (pt_solution *)): New.
21523         * tree-ssa-alias.h
21524         (extern debug (pt_solution &)): New.
21525         (extern debug (pt_solution *)): New.
21526         * tree-ssa-alias.c
21527         (debug (_var_map &)): New.
21528         (debug (_var_map *)): New.
21529         (debug (tree_live_info_d &)): New.
21530         (debug (tree_live_info_d *)): New.
21531         * tree-ssa-alias.h
21532         (extern debug (_var_map &)): New.
21533         (extern debug (_var_map *)): New.
21534         (extern debug (tree_live_info_d &)): New.
21535         (extern debug (tree_live_info_d *)): New.
21537 2013-03-28  Jan Hubicka  <jh@suse.cz>
21539         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21541 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21543         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21544         record only when desired or required.
21546 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21548         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21549         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21551 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21553         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21554         (*andsi3_compare0_uxtw): New pattern.
21555         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21556         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21558 2013-03-28  Jan Hubicka  <jh@suse.cz>
21560         * data-streamer-in.c (streamer_read_gcov_count): New function.
21561         * gimple-streamer-out.c: Include value-prof.h.
21562         (output_gimple_stmt): Output histogram.
21563         (output_bb): Use streamer_write_gcov_count.
21564         * value-prof.c: Include data-streamer.h
21565         (dump_histogram_value): Add HIST_TYPE_MAX.
21566         (stream_out_histogram_value): New function.
21567         (stream_in_histogram_value): New function.
21568         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21569         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21570         * data-streamer-out.c (streamer_write_gcov_count): New function.
21571         (streamer_write_gcov_count_stream): New function.
21572         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21573         (lto_output_node): Likewise.
21574         (input_node, input_edge): Likewise.
21575         * lto-streamer-out.c (output_cfg): Update streaming.
21576         * lto-streamer-in.c (input_cfg): Likewise.
21577         * data-streamer.h (streamer_write_gcov_count,
21578         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21579         * gimple-streamer-in.c: Include value-prof.h
21580         (input_gimple_stmt): Input histograms.
21581         (input_bb): Update profile streaming.
21583 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21585         * genmodes.c (emit_max_int): New function.
21586         (emit_insn_modes_h): Added call to emit_max_function.
21587         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21588         Added doc.
21589         * machmode.def: Fixed comment.
21591 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21593         * combine.c (try_combine): Removed useless assert.
21594         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21596 2013-03-28  Marek Polacek  <polacek@redhat.com>
21597             Richard Biener  <rguenther@suse.de>
21599         PR tree-optimization/56695
21600         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21601         build signed result of a vector comparison.
21602         * tree-cfg.c (verify_gimple_comparison): Check that a result
21603         of a vector comparison has signed type.
21605 2013-03-28  Richard Biener  <rguenther@suse.de>
21607         PR tree-optimization/37021
21608         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21609         do not restrict gaps between groups.
21610         * tree-vect-stmts.c (vectorizable_load): Properly account for
21611         a gap between groups.
21613 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21615         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21616         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21617         is not enabled.
21619 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21621         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21622         * doc/extend.texi (Named Address Spaces): Ditto.
21623         (Variable Attributes): Ditto.
21625 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21627         * config.build: Add support for cygwin x64 target.
21628         * config.gcc: Likewise.
21629         * config.host: Likewise.
21630         * configure.ac: Likewise
21631         * configure: Regenerated.
21633 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21635         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21636         * config/i386/t-cygwin-w64: New file.
21637         * config/i386/cygwin-w64.h: New file.
21638         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21639         and add support for x64-cygwin target.
21640         (CPP_SPEC): Likewise.
21641         (CXX_WRAP_SPEC_LIST): Undefine before define.
21642         (LIBGCJ_SONAME): Use 15 as version.
21644 2013-03-27  Richard Biener  <rguenther@suse.de>
21646         PR tree-optimization/56716
21647         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21648         dumping for ref nodes.
21650 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21652         PR tree-optimization/55334
21653         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21654         restricted pointers to arrays.
21656 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21658         * Makefile.in (.SUFFIXES): Add .cc.
21659         (.c.o): Apply same recipe for implicit rule .cc.o.
21661 2013-03-27  Richard Biener  <rguenther@suse.de>
21663         PR tree-optimization/37021
21664         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21665         REALPART/IMAGPART_EXPRs around the supported refs.
21666         * tree-ssa-structalias.c (find_func_aliases): Assume that
21667         floating-point values are not used to transfer pointers.
21669 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21671         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21672         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21673         * doc/tm.texi: Regenerate.
21674         * targhooks.h (default_has_ifunc_p): New.
21675         * targhooks.c (default_has_ifunc_p): Ditto.
21676         * config/linux-protos.h: New file.
21677         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21678         hook for linux which disables support of indirect functions in android.
21679         * config/linux-android.c: New file.
21680         * config/t-linux-android.c: Ditto.
21681         * config.gcc: Added new object file linux-android.o.
21682         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21683         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21684         * varasm.c (do_assemble_alias): Likewise.
21685         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21686         doesn't support indirect functions.
21687         * configure: Regenerate.
21689 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21691         PR target/56102
21692         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21693         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21694         mult-word mode.
21696 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21698         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21700 2013-03-27  Terry Guo  <terry.guo@arm.com>
21702         * config/arm/arm-cores.def: Added core cortex-r7.
21703         * config/arm/arm-tune.md: Regenerated.
21704         * config/arm/arm-tables.opt: Regenerated.
21705         * doc/invoke.texi: Added entry for core cortex-r7.
21707 2013-03-27  Walter Lee  <walt@tilera.com>
21709         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21710         double-decrement of next_scratch_regno.
21712 2013-03-27  Walter Lee  <walt@tilera.com>
21714         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21715         input operands.
21716         (insn_v1mulus): Ditto.
21717         (insn_v2muls): Ditto.
21719 2013-03-27  Walter Lee  <walt@tilera.com>
21721         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21722         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21724 2013-03-27  Walter Lee  <walt@tilera.com>
21726         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21727         (*sibcall_value): Ditto.
21729 2013-03-27  Walter Lee  <walt@tilera.com>
21731         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21732         (insn_mnz_v8qi): ... this ...
21733         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21734         vector equivalent.
21735         (insn_v<n>mnz): Replaced by ...
21736         (insn_v1mnz): ... this ...
21737         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21738         equivalent.
21739         (insn_mz_<mode>): Replaced by ...
21740         (insn_mz_v8qi): ... this ...
21741         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21742         vector equivalent.
21743         (insn_v<n>mz): Replaced by ...
21744         (insn_v1mz): ... this ...
21745         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21746         equivalent.
21748 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21750         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21752 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21754         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21755         than fprintf with a non-constant, non-format string.
21757 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21759         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21760         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21761         operand 0 predicate.
21762         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21763         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21764         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21765         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21766         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21767         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21768         (mov<mode>_insv_1): Remove expander.  Merge insn with
21769         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21770         Use general_x64nomem_operand as operand 1 predicate.
21771         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21772         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21773         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21774         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21775         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21776         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21777         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21778         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21779         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21780         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21781         (general_x64nomem_operand): Ditto.
21783 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21785         * config/rtems.opt: Add -pthread option.
21787 2013-03-26  Richard Biener  <rguenther@suse.de>
21789         * alias.c (find_base_term): Avoid redundant and not used recursion.
21790         (base_alias_check): Get the initial base term from the caller.
21791         (true_dependence_1): Compute and pass base terms to base_alias_check.
21792         (write_dependence_p): Likewise.
21793         (may_alias_p): Likewise.
21795 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21797         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21798         PC-relative load in SI modes and above only.
21800 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21802         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21803         * config/arm/iwmmxt.md (WCGR0): Update.
21804         (WCGR1, WCGR2, WCGR3): Likewise.
21806 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21808         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21809         Use x64 and nox64 isa attributes.
21811 2013-03-26  Richard Biener  <rguenther@suse.de>
21813         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21814         alignment computations and rely on get_object_alignment_1
21815         for the !TYPE_P case.
21816         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21818 2013-03-26  Walter Lee  <walt@tilera.com>
21820         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21821         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21823 2013-03-25  Jeff Law  <law@redhat.com>
21825         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21826         check for INTEGRAL_TYPE_P that was missing due to checking in
21827         wrong version of prior patch.
21829 2013-03-25  Walter Lee  <walt@tilera.com>
21831         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21832         TILEGX_INSN_SHUFFLEBYTES1.
21833         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21834         shufflebytes1.
21835         (tilegx_builtins): Ditto.
21836         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21838 2013-03-25  Walter Lee  <walt@tilera.com>
21840         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21841         (floatunssisf2): New pattern.
21842         (floatsidf2): New pattern.
21843         (floatunssidf2): New pattern.
21845 2013-03-25  Walter Lee  <walt@tilera.com>
21847         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21848         tests for constraint J, K, N, P.
21850 2013-03-25  Walter Lee  <walt@tilera.com>
21852         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21853         Use indirect/pcrel encoding.
21854         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21855         Ditto.
21857 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21859         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21860         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21861         * config/mips/mips.c (mips_option_override): Set IMADD default.
21862         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21863         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21864         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21865         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21866         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21868 2013-03-25  Jeff Law  <law@redhat.com>
21870         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21871         slightly to avoid creating and folding useless trees.  Simplify
21872         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21874 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21876         * config/i386/i386.md (*zero_extendsidi2): Merge with
21877         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21878         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21879         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21880         targets.  Clarify comment.
21882 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21884         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21885         pass-through jump functions differently.
21886         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21887         up jump functions.
21889 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21891         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21892         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21893         process it.
21894         (ipa_get_indirect_edge_target): New function.
21895         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21896         ipa_get_indirect_edge_target.  Update all callers.
21897         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21898         ipa_get_indirect_edge_target_1 instead of calling
21899         ipa_get_indirect_edge_target.
21900         (create_specialized_node): Pass aggvlas to
21901         ipcp_discover_new_direct_edges.
21903 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21905         * config/arm/arm.md (f_sels, f_seld): New types.
21906         (*cmov<mode>): New pattern.
21907         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21908         predicate.
21910 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21912         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21913         POSIX-printf for mingw-hosted  builds.
21915 2013-03-25  Richard Biener  <rguenther@suse.de>
21917         PR middle-end/56694
21918         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21919         must-not-throw stmt location.
21921 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21923         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21924         Emit load-acquire versions when acq is true.
21925         (arm_emit_store_exclusive): Add rel parameter.
21926         Emit store-release versions when rel is true.
21927         (arm_split_compare_and_swap): Use acquire-release instructions
21928         instead.
21929         of barriers when appropriate.
21930         (arm_split_atomic_op): Likewise.
21931         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21932         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21933         (VUNSPEC_SLX): Likewise.
21934         (VUNSPEC_LDA): Likewise.
21935         (VUNSPEC_STL): Likewise.
21936         * config/arm/sync.md (atomic_load<mode>): New pattern.
21937         (atomic_store<mode>): Likewise.
21938         (arm_load_acquire_exclusive<mode>): Likewise.
21939         (arm_load_acquire_exclusivesi): Likewise.
21940         (arm_load_acquire_exclusivedi): Likewise.
21941         (arm_store_release_exclusive<mode>): Likewise.
21943 2013-03-25  Catherine Moore  <clm@codesourcery.com>
21945         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
21946         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
21947         * config/mip/predicates.md (lwsp_swsp_operand,
21948         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
21949         sb16_operand, db4_operand, db7_operand, ib3_operand,
21950         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
21951         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
21952         andi16_operand): New predicates.
21953         * config/mips/mips.md (compression): New attribute.
21954         (enabled): New attribute.
21955         (length): Consider compression in computing length.
21956         (shift_compression): New code attribute.
21957         (*add<mode>3): New operands. Record compression.
21958         (sub<mode>3): Likewise.
21959         (one_cmpl<mode>2): Likewise.
21960         (*and<mode>3): Likewise.
21961         (*ior<mode>3): Likewise.
21962         (unnamed pattern for xor): Likewise.
21963         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
21964         (*<optab><mode>3): Likewise.
21965         (*mov<mode>_internal: Likewise.
21966         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
21967         (mips_unsigned_immediate_p): New.
21968         (umips_lwsp_swsp_address_p): New.
21969         (m16_based_address_p): New.
21970         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
21971         (mips_unsigned_immediate_p): New prototype.
21972         (lwsp_swsp_address_p): New prototype.
21973         (m16_based_address_p): New prototype.
21974         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
21975         (mips_signed_immediate_p): New function.
21976         (m16_based_address_p): New function.
21977         (lwsp_swsp_address_p): New function.
21978         (mips_print_operand_punctuation): Recognize short delay slot insns
21979         for microMIPS.add<mode>3"
21981 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21983         PR target/56720
21984         * config/arm/iterators.md (v_cmp_result): New mode attribute.
21985         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
21987 2013-03-25  Richard Biener  <rguenther@suse.de>
21989         PR tree-optimization/56689
21990         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
21991         any edge.
21993 2013-03-25  Richard Biener  <rguenther@suse.de>
21995         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
21996         of bitmap.
21997         (memory_references): Likewise.
21998         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
21999         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22000         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22001         (gather_mem_refs_in_loops): Fold into ...
22002         (analyze_memory_references): ... this.  Move initialization
22003         to tree_ssa_lim_initialize.
22004         (fill_always_executed_in): Rename to ...
22005         (fill_always_executed_in_1): ... this.
22006         (fill_always_executed_in): Move contains_call computation to
22007         this new function from ...
22008         (tree_ssa_lim_initialize): ... here.
22009         (tree_ssa_lim): Call fill_always_executed_in.
22011 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22013         * postreload.c (reload_combine): Fix code detecting returns.
22015 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22017         * function.c (emit_use_return_register_into_block): On cc0 targets,
22018         do not emit the sequence between cc0 setter and user.
22020 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22022         * config/i386/predicates.md (local_symbolic_operand): Interpret
22023         dll-imported symbols as none-local.
22025 2013-03-25  Richard Biener  <rguenther@suse.de>
22027         * tree-ssa-loop-im.c (struct depend): Remove.
22028         (struct lim_aux_data): Make depends a vec of gimples.
22029         (free_lim_aux_data): Adjust.
22030         (add_dependency): Likewise.
22031         (set_level): Likewise.
22033 2013-03-25  Richard Biener  <rguenther@suse.de>
22035         PR middle-end/56434
22036         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22037         the pointer returned by calls with ECF_MALLOC set.
22039 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22041         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22043 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22045         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22046         using MMXMODE mode iterator.
22047         (*move<mode>_internal): Merge with *movv2sf_internal and
22048         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22050 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22052         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22053         (record_last_mem_set_info): Likewise.
22055         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22056         of XNEWVEC followed by memset.
22057         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22059 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22061         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22062         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22063         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22064         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22065         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22066         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22067         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22068         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22069         BARRIER_P instead of GET_CODE.
22071 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22073         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22074         inaccuracy in the probing code.
22076         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22077         (ctrapdi4): Likewise.
22079 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22081         * calls.c (expand_call): Add missing guard to code handling return
22082         of non-BLKmode structures in MSB.
22083         * function.c (expand_function_end): Likewise.
22085 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22087         * combine.c (try_combine): Adjust comment.  Do not add the set of
22088         insn #0 if the destination indirectly is set or dies in insn #2.
22089         Tidy up code to distribute a new note.
22091 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22093         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22094         also for alternatives 16 and 17.
22096 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22098         * config/i386/sse.md (*mov<mode>_internal): Merge with
22099         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22100         Emit insn template depending on type attribute.  Use
22101         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22102         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22103         attribute calculation.  Remove unit attribute calculation.
22104         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22105         Set prefix_data16 attribute for DImode ssemov types.
22106         Use Ym instead of y for SSE-MMX conversion alternatives.
22107         Reorder operand constraints.
22109 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22111         * df.h (df_insn_delete): Adjust prototype.
22112         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22113         and let it decide whether mark the basic block dirty.
22114         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22115         * df-scan.c (df_insn_info_delete): New helper function, split
22116         off from df_insn_delete.
22117         (df_scan_free_bb_info): Use it.
22118         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22119         Likewise.
22120         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22121         that the insn is actually an insn and it has a non-NULL basic block.
22122         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22124 2013-03-22  Richard Biener  <rguenther@suse.de>
22126         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22127         dep_ref members.
22128         (mem_ref_alloc): Do not allocate them.
22129         (refs_independent_p): Do not query or maintain a cache.
22131 2013-03-22  Richard Biener  <rguenther@suse.de>
22133         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22134         (gather_mem_refs_in_loops): Do not compute it.
22135         (analyze_memory_references): Do not allocate it.
22136         (tree_ssa_lim_finalize): Do not free it.
22137         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22139 2013-03-22  Richard Biener  <rguenther@suse.de>
22141         * is-a.h (as_a): Use gcc_checking_assert.
22143 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22145         * config/aarch64/aarch64.c (aarch64_print_operand): New
22146         format specifier for printing a constant in hex.
22147         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22148         format specifier for printing second operand.
22150 2013-03-22  Richard Biener  <rguenther@suse.de>
22152         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22153         bitmaps.
22154         (gather_mem_refs_in_loops): Perform store accumulation here.
22155         (create_vop_ref_mapping_loop): Remove.
22156         (create_vop_ref_mapping): Likewise.
22157         (analyze_memory_references): Initialize refs_stored_in_loop.
22158         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22159         (record_indep_loop): Remove.
22160         (record_dep_loop): New function.
22161         (ref_indep_loop_p_1): Adjust to only walk over references
22162         in the loop, not its subloops.
22163         (ref_indep_loop_p): Rename to ...
22164         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22165         maintaining a more fine-grained cache.
22166         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22167         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22169 2013-03-22  Richard Biener  <rguenther@suse.de>
22171         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22172         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22173         aggregate mem_ref_loc.
22174         (free_mem_ref_locs): Inline into ...
22175         (memref_free): ... this and adjust.
22176         (mem_ref_alloc): Adjust.
22177         (mem_ref_locs_alloc): Remove.
22178         (record_mem_ref_loc): Adjust.
22179         (get_all_locs_in_loop): Rewrite into ...
22180         (for_all_locs_in_loop): ... this iterator.
22181         (rewrite_mem_ref_loc): New functor.
22182         (rewrite_mem_refs): Use for_all_locs_in_loop.
22183         (sm_set_flag_if_changed): New functor.
22184         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22185         (ref_always_accessed): New functor.
22186         (ref_always_accessed_p): Use for_all_locs_in_loop.
22188 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22190         * tree-pass.h (PROP_gimple_lvec): New.
22191         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22192         (init_optimization_passes): Move pass_lower_vector.
22193         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22194         PROP_gimple_lvec.
22195         (pass_lower_vector): Provide PROP_gimple_lvec.
22196         (pass_lower_vector_ssa): Likewise.
22197         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22199 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22201         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22203 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22205         * config/i386/i386.md (*movdi_internal): Disparage slightly
22206         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22207         conversion alternatives.
22209 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22211         PR middle-end/48087
22212         * diagnostic.def (DK_WERROR): New kind.
22213         * diagnostic.h (werrorcount): Define.
22214         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22215         promoted to DK_ERROR, increment DK_WERROR counter instead of
22216         DK_ERROR counter.
22217         * toplev.c (toplev_main): Call print_ignored_options even if
22218         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22219         even if just werrorcount is non-zero.
22221         PR debug/55608
22222         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22223         on failure.
22224         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22225         (string_cst_pool_decl): New function.
22226         (optimize_one_addr_into_implicit_ptr): New function.
22227         (resolve_addr_in_expr): Optimize DWARF location expression
22228         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22229         which doesn't live in memory, but has DW_AT_location or
22230         DW_AT_const_value, or refers to a string literal, into
22231         DW_OP_GNU_implicit_pointer.
22232         (optimize_location_into_implicit_ptr): New function.
22233         (resolve_addr): If removing DW_AT_location of a variable because
22234         it was DW_OP_addr of address of the variable, but the variable doesn't
22235         live in memory, try to emit const value attribute for the initializer.
22237 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22239         * tree.h (VECTOR_TYPE_P): New macro.
22240         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22241         TYPE_MODE): Use it.
22242         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22243         VEC_COND_EXPR cannot be lvalues.
22244         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22246 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22248         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22249         Restrict the transformation to equal modes.
22251 2013-03-21  Richard Biener  <rguenther@suse.de>
22253         PR tree-optimization/39326
22254         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22255         (MEM_ANALYZABLE): Adjust.
22256         (record_mem_ref_loc): Move bitmap ops ...
22257         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22258         unanalyzable refs, do not record locations for it.
22259         (analyze_memory_references): Allocate ref zero as shared
22260         unanalyzable ref.
22261         (refs_independent_p): Do not test for unanalyzed mems here.
22262         (ref_indep_loop_p_1): Special-case disambiguation against
22263         the unanalyzed ref.
22264         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22266 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22268         * config/arm/arm-protos.h (tune_params): Add
22269         prefer_neon_for_64bits field.
22270         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22271         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22272         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22273         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22274         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22275         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22276         (arm_option_override): Handle -mneon-for-64bits new option.
22277         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22278         (prefer_neon_for_64bits): Declare new variable.
22279         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22280         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22281         (arch_enabled): Handle new arch types. Remove support for onlya8
22282         and nota8.
22283         (one_cmpldi2): Use new arch names.
22284         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22285         * config/arm/arm.opt (mneon-for-64bits): Add option.
22286         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22287         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22288         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22289         of onlya8.
22290         * doc/invoke.texi (-mneon-for-64bits): Document.
22292 2013-03-21  Richard Biener  <rguenther@suse.de>
22294         PR tree-optimization/39326
22295         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22296         (sort_bbs_in_loop_postorder_cmp): New function.
22297         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22299 2013-03-21  Richard Biener  <rguenther@suse.de>
22301         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22302         (vect_insert_into_interleaving_chain): Likewise.
22303         (vect_drs_dependent_in_basic_block): Inline ...
22304         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22305         split out from ...
22306         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22307         (vect_check_interleaving): Simplify.
22308         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22309         (vect_slp_analyze_data_ref_dependences): ... this new function.
22310         (dr_group_sort_cmp): New function.
22311         (vect_analyze_data_ref_accesses): Compute data-reference groups
22312         here instead of in vect_analyze_data_ref_dependence.  Use
22313         a more efficient algorithm.
22314         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22315         vect_slp_analyze_data_ref_dependences.  Call
22316         vect_analyze_data_ref_accesses earlier.
22317         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22318         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22319         (vect_slp_analyze_data_ref_dependences): New prototype.
22321 2013-03-21  Richard Biener  <rguenther@suse.de>
22323         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22324         ref is stored in the loop.
22325         (find_refs_for_sm): Walk only over all stores.
22326         (store_motion_loop): Allocate from lim_bitmap_obstack.
22327         (store_motion): Likewise.
22329 2013-03-21  Richard Biener  <rguenther@suse.de>
22331         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22332         Update virtual SSA form.
22334 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22336         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22337         * configure: Regenerate.
22338         * config.in: Regenerate.
22339         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22340         if !HAVE_LD_EH_FRAME_CIEV3.
22342 2013-03-21  Richard Biener  <rguenther@suse.de>
22344         * tree-cfg.c (verify_expr_no_block): New function.
22345         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22346         nor DECL_VALUE_EXPR have locations with associated blocks.
22347         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22348         (clear_unused_block_pointer): Remove code dealing with
22349         blocks in DECL_DEBUG_EXPR locations.
22351 2013-03-21  Richard Biener  <rguenther@suse.de>
22353         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22354         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22355         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22356         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22357         instead of DECL_DEBUG_EXPR_IS_FROM.
22358         * gimplify.c (gimplify_modify_expr): Likewise.
22359         * tree-cfg.c (verify_expr_location_1): Likewise.
22360         * tree-complex.c (create_one_component_var): Likewise.
22361         * tree-sra.c (create_access_replacement): Likewise.
22362         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22363         (clear_unused_block_pointer): Likewise.
22364         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22365         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22366         * var-tracking.c (var_debug_decl): Likewise.
22367         (track_expr_p): Likewise.
22368         * tree-inline.c (add_local_variables): Likewise.  Set
22369         DECL_HAS_DEBUG_EXPR_P after copying it.
22370         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22371         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22373 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22375         PR bootstrap/56656
22376         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22377         * configure: Regenerate.
22378         * config.in: Regenerate.
22379         * config/i386/i386.md (*movdf_internal): Use
22380         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22381         movd instead of movq mnemonic for interunit moves.
22382         (*movdi_internal): Ditto.
22384 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22386         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22387         (abd<mode>_3): New pattern.
22388         (aba<mode>_3): New pattern.
22389         (fabd<mode>_3): New pattern.
22391 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22393         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22394         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22395         occurrence of REGISTER_PREFIX as its empty string.
22397 2013-03-20  Jeff Law  <law@redhat.com>
22399         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22400         addititional equivalences for equality comparisons between an SSA_NAME
22401         and a constant where the SSA_NAME was set from a widening conversion.
22403 2013-03-20  Walter Lee  <walt@tilera.com>
22405         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22407 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22409         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22410         depending on type attribute.
22411         (*movti_internal): Ditto.
22412         (*movtf_internal): Ditto.
22413         (*movxf_internal): Ditto.
22414         (*movdf_internal): Ditto.
22415         (*movsf_internal): Ditto.
22417 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22419         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22420         maybe_vex for sselog1 and ssemov types.
22421         (*movdi_internal): Reorder operand constraints.
22422         (*movsi_internal): Ditto.  Set prefix attribute to
22423         maybe_vex for sselog1 and ssemov types.
22424         (*movtf_internal): Set prefix attribute to maybe_vex
22425         for sselog1 and ssemov types.
22426         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22427         DImode ssemov types.  Reorder operand constraints.
22428         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22429         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22430         attribute for SImode ssemov types.  Reorder operand constraints.
22432 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22434         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22435         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22437 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22439         * config/rs6000/predicates.md (indexed_address, update_address_mem
22440         update_indexed_address_mem): New predicates.
22441         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22442         attribute for load/store instructions.
22443         * config/rs6000/dfp.md (movsd_store): Likewise.
22444         (movsd_load): Likewise.
22445         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22446         (unnamed HI->DI extend define_insn): Likewise.
22447         (unnamed SI->DI extend define_insn): Likewise.
22448         (unnamed QI->SI extend define_insn): Likewise.
22449         (unnamed QI->HI extend define_insn): Likewise.
22450         (unnamed HI->SI extend define_insn): Likewise.
22451         (unnamed HI->SI extend define_insn): Likewise.
22452         (extendsfdf2_fpr): Likewise.
22453         (movsi_internal1): Likewise.
22454         (movsi_internal1_single): Likewise.
22455         (movhi_internal): Likewise.
22456         (movqi_internal): Likewise.
22457         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22458         attribute for load/store instructions.
22459         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22460         instructions.
22461         (mov<mode>_softfloat): Likewise.
22462         (mov<mode>_hardfloat32): Likewise.
22463         (mov<mode>_hardfloat64): Likewise.
22464         (mov<mode>_softfloat64): Likewise.
22465         (movdi_internal32): Likewise.
22466         (movdi_internal64): Likewise.
22467         (probe_stack_<mode>): Likewise.
22469 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22471         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22472         floating point, and decimal floating point to reload iterator.
22474         * config/rs6000/constraints.md (wl constraint): New constraints to
22475         return FLOAT_REGS if certain options are used to reduce the number
22476         of separate patterns that exist in the file.
22477         (wx constraint): Likewise.
22478         (wz constraint): Likewise.
22480         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22481         -mdebug=reg, print wg, wl, wx, and wz constraints.
22482         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22483         Initialize the reload functions for 64-bit binary/decimal floating
22484         point types.
22485         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22486         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22487         create the buffer on the stack to overcome not having a 32-bit
22488         load and store.
22489         (rs6000_emit_move): Likewise.
22490         (rs6000_secondary_memory_needed_rtx): Likewise.
22491         (rs6000_alloc_sdmode_stack_slot): Likewise.
22492         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22493         via xxlxor, just like DFmode 0.0.
22495         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22496         define as 1 if we are running on a power7 or newer.
22497         (enum r6000_reg_class_enum): Add new constraints.
22499         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22500         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22501         with other moves by using conditional constraits (wg).  Use LFIWZX
22502         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22503         (movsd splitter): Likewise.
22504         (movsd_hardfloat): Likewise.
22505         (movsd_softfloat): Likewise.
22507         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22508         binary and decimal floating point moves.
22509         (fmove_ok): New attributes to combine binary and decimal floating
22510         point moves, and to combine power6x (mfpgpr) moves along normal
22511         floating moves.
22512         (real_value_to_target): Likewise.
22513         (f32_lr): Likewise.
22514         (f32_lm): Likewise.
22515         (f32_li): Likewise.
22516         (f32_sr): Likewise.
22517         (f32_sm): Likewise.
22518         (f32_si): Likewise.
22519         (movsf): Combine binary and decimal floating point moves.  Combine
22520         power6x (mfpgpr) moves with other moves by using conditional
22521         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22522         (mov<mode> for SFmode/SDmode); Likewise.
22523         (SFmode/SDmode splitters): Likewise.
22524         (movsf_hardfloat): Likewise.
22525         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22526         (movsf_softfloat): Likewise.
22527         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22529         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22530         wx and wz constraints.
22532         * config/rs6000/constraints.md (wg constraint): New constraint to
22533         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22535         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22536         constraint.
22538         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22539         -mdebug=reg, print wg, wl, wx, and wz constraints.
22540         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22541         Initialize the reload functions for 64-bit binary/decimal floating
22542         point types.
22543         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22544         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22545         create the buffer on the stack to overcome not having a 32-bit
22546         load and store.
22547         (rs6000_emit_move): Likewise.
22548         (rs6000_secondary_memory_needed_rtx): Likewise.
22549         (rs6000_alloc_sdmode_stack_slot): Likewise.
22550         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22551         via xxlxor, just like DFmode 0.0.
22553         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22554         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22555         with other moves by using conditional constraits (wg).  Use LFIWZX
22556         and STFIWX for loading SDmode on power7.
22557         (movdd splitters): Likewise.
22558         (movdd_hardfloat32): Likewise.
22559         (movdd_softfloat32): Likewise.
22560         (movdd_hardfloat64_mfpgpr): Likewise.
22561         (movdd_hardfloat64): Likewise.
22562         (movdd_softfloat64): Likewise.
22564         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22565         64-bit binary and decimal floating point moves.
22566         (FMOVE64X): Likewise.
22567         (movdf): Combine 64-bit binary and decimal floating point moves.
22568         Combine power6x (mfpgpr) moves with other moves by using
22569         conditional constraits (wg).
22570         (mov<mode> for DFmode/DDmode): Likewise.
22571         (DFmode/DDmode splitters): Likewise.
22572         (movdf_hardfloat32): Likewise.
22573         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22574         (movdf_softfloat32): Likewise.
22575         (movdf_hardfloat64_mfpgpr): Likewise.
22576         (movdf_hardfloat64): Likewise.
22577         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22578         (movdf_softfloat64): Likewise.
22579         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22580         (reload_<mode>_load): Move to later in the file so they aren't in
22581         the middle of the floating point move insns.
22582         (reload_<mode>_store): Likewise.
22584         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22585         constraint.
22587         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22588         constraint if -mdebug=reg.
22589         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22590         Enable using dd reload support if needed.
22592         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22593         binary and decimal floating point moves in rs6000.md.
22594         (movtd_internal): Likewise.
22596         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22597         decimal floating point moves.
22598         (movtf): Likewise.
22599         (movtf_internal): Likewise.
22600         (mov<mode>_internal, TDmode/TFmode): Likewise.
22601         (movtf_softfloat): Likewise.
22602         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22604         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22605         movdi_internal64, using wg constraint for move direct operations.
22606         (movdi_internal64): Likewise.
22608         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22609         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22610         the various virtual registers. Use GPR/FPR first/last values,
22611         instead of hard coding the register numbers.  Print which modes
22612         have reload functions registered.
22613         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22614         settings before/after setting cpu, target and subtarget settings.
22615         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22616         and for secondary reload failures in rs6000_secondary_reload_inner.
22617         (rs6000_secondary_reload_fail): Likewise.
22618         (rs6000_secondary_reload_inner): Likewise.
22620         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22621         macros for first/last GPR and FPR registers.
22622         (LAST_GPR_REGNO): Likewise.
22623         (FIRST_FPR_REGNO): Likewise.
22624         (LAST_FPR_REGNO): Likewise.
22626         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22627         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22628         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22629         (vcond<mode><mode>): Likewise.
22630         (vcondu<mode><mode>): Likewise.
22631         (vector_gtu<mode>): Likewise.
22632         (vector_gte<mode>): Likewise.
22633         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22634         to prevent the compiler from converting DImode operations to TImode.
22635         (ior<mode>3): Likewise.
22636         (and<mode>3): Likewise.
22637         (one_cmpl<mode>2): Likewise.
22638         (nor<mode>3): Likewise.
22639         (andc<mode>3): Likewise.
22641         * config/rs6000/constraints.md (wt constraint): New constraint
22642         that returns VSX_REGS if TImode is allowed in VSX registers.
22644         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22645         constant under VSX.
22647         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22648         similar to TImode, but it is restricted to being in the GPRs.
22650         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22651         TImode to occupy a single VSX register.
22653         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22654         -mvsx-timode for power7/power8.
22655         (power7 cpu): Likewise.
22656         (power8 cpu): Likewise.
22658         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22659         sure that TFmode/TDmode take up two registers if they are ever
22660         allowed in the upper VSX registers.
22661         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22662         registers.
22663         (rs6000_init_hard_regno_mode_ok): Likewise.
22664         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22665         constraint.  Print if LRA is turned on.
22666         (rs6000_option_override_internal): Give an error if -mvsx-timode
22667         and VSX is not enabled.
22668         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22669         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22670         to reg+offset addressing.  Use PTImode when checking offset
22671         addresses for validity.
22672         (reg_offset_addressing_ok_p): Likewise.
22673         (rs6000_legitimate_offset_address_p): Likewise.
22674         (rs6000_legitimize_address): Likewise.
22675         (rs6000_legitimize_reload_address): Likewise.
22676         (rs6000_legitimate_address_p): Likewise.
22677         (rs6000_eliminate_indexed_memrefs): Likewise.
22678         (rs6000_emit_move): Likewise.
22679         (rs6000_secondary_reload): Likewise.
22680         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22681         reloads to fpr registers to continue to use reg+offset addressing,
22682         but 64-bit reloads to altivec registers need reg+reg addressing.
22683         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22684         it.  Treat LO_SUM like a PLUS operation.
22685         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22686         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22687         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22688         registers to share a register with a smaller sized type, since VSX
22689         puts scalars in the upper 64-bits.
22690         (print_operand): Add support for PTImode.
22691         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22692         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22693         registers, but don't have arithmetic support.
22694         (rs6000_memory_move_cost): Add test for VSX.
22695         (rs6000_opt_masks): Add -mvsx-timode.
22697         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22698         for TImode.
22699         (VSs): Likewise.
22700         (VSr): Use wt constraint for TImode.
22701         (VSv): Drop TImode support.
22702         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22703         (vsx_movti_64bit): Likewise.
22704         (vsx_movti_32bit): Likewise.
22705         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22706         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22707         one '?' on the appropriate output constraint.  Do not allow TImode
22708         logical operations on 32-bit systems.
22709         (vsx_ior<mode>3): Likewise.
22710         (vsx_xor<mode>3): Likewise.
22711         (vsx_one_cmpl<mode>2): Likewise.
22712         (vsx_nor<mode>3): Likewise.
22713         (vsx_andc<mode>3): Likewise.
22714         (vsx_concat_<mode>): Likewise.
22715         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22717         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22718         OPTION_MASK_VSX_TIMODE.
22719         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22720         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22722         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22723         (TI2 iterator): New iterator for TImode, PTImode.
22724         (wd mode attribute): Add values for vector types.
22725         (movti_string): Replace TI move operations with operations for TImode
22726         and PTImode.  Add support for TImode being allowed in VSX registers.
22727         (mov<mode>_string, TImode/PTImode): Likewise.
22728         (movti_ppc64): Likewise.
22729         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22730         (TI mode splitters): Likewise.
22732         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22733         constraint.
22735 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22737         PR tree-optimization/56355
22738         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22739         Also handle integers with undefined overflow.
22741 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22742             Maciej W. Rozycki  <macro@codesourcery.com>
22743             Tom de Vries  <tom@codesourcery.com>
22744             Nathan Sidwell  <nathan@codesourcery.com>
22745             Iain Sandoe  <iain@codesourcery.com>
22746             Nathan Froyd  <froydnj@codesourcery.com>
22747             Chao-ying Fu  <fu@mips.com>
22749         * doc/extend.texi: (micromips, nomicromips, nocompression):
22750         Document new function attributes.
22751         * doc/invoke.texi (minterlink-compressed, mmicromips,
22752         m14k, m14ke, m14kec): Document new options.
22753         (minterlink-mips16): Update documentation.
22754         * doc/md.texi (ZC, ZD): Document new constraints.
22755         * configure.ac (gcc_cv_as_micromips): Check if linker
22756         supports the .set micromips directive.
22757         * configure: Regenerate.
22758         * config.in: Regenerate.
22759         * config/mips/mips-tables.opt: Regenerate.
22760         * config/mips/micromips.md: New file.
22761         * constraints.md (ZC, ZD): New constraints.
22762         * config/mips/predicates.md (movep_src_register): New predicate.
22763         (movep_src_operand): New predicate.
22764         (non_volatile_mem_operand): New predicate.
22765         * config/mips/mips.md (multimem): New type.
22766         (length): Differentiate between 17-bit and 18-bit branch offsets.
22767         (MOVEP1, MOVEP2): New mode iterator.
22768         (mov_<load>l): Use ZC constraint.
22769         (mov_<load>r): Likewise.
22770         (mov_<store>l): Likewise.
22771         (mov_<store>r): Likewise.
22772         (*branch_equality<mode>_inverted): Add microMIPS support.
22773         (*branch_equality<mode>): Likewise.
22774         (*jump_absolute): Likewise.
22775         (indirect_jump_<mode>): Likewise.
22776         (tablejump_<mode>): Likewise.
22777         (<optab>_internal): Likewise.
22778         (sibcall_internal): Likewise.
22779         (sibcall_value_internal): Likewise.
22780         (prefetch): Use constraint ZD.
22781         * config/mips/mips.opt (minterlink-compressed): New option.
22782         (minterlink-mips16): Now an alias for minterlink-compressed.
22783         (mmicromips): New option.
22784         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22785         (compare_and_swap_12): Likewise.
22786         (sync_add<mode>): Likewise.
22787         (sync_<optab>_12): Likewise.
22788         (sync_old_<optab>_12): Likewise.
22789         (sync_new_<optab>_12): Likewise.
22790         (sync_nand_12): Likewise.
22791         (sync_old_nand_12): Likewise.
22792         (sync_new_nand_12): Likewise.
22793         (sync_sub<mode>): Likewise.
22794         (sync_old_add<mode>): Likewise.
22795         (sync_old_sub<mode>): Likewise.
22796         (sync_new_add<mode>): Likewise.
22797         (sync_new_sub<mode>): Likewise.
22798         (sync_<optab><mode>): Likewise.
22799         (sync_old_<optab><mode>): Likewise.
22800         (sync_new_<optab><mode>): Likewise.
22801         (sync_nand<mode>): Likewise.
22802         (sync_old_nand<mode>): Likewise.
22803         (sync_new_nand<mode>): Likewise.
22804         (sync_lock_test_and_set<mode>): Likewise.
22805         (test_and_set_12): Likewise.
22806         (atomic_compare_and_swap<mode>): Likewise.
22807         (atomic_exchange<mode>_llsc): Likewise.
22808         (atomic_fetch_add<mode>_llsc): Likewise.
22809         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22810         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22811         (umips_save_restore_pattern_p): Likewise.
22812         (umips_load_store_pair_p): Likewise.
22813         (umips_output_load_store_pair): Likewise.
22814         (umips_movep_target_p): Likewise.
22815         (umips_12bit_offset_address_p): Likewise.
22816         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22817         (mips_base_mips16): Rename this...
22818         (mips_base_compression_flags): ...to this. Update all uses.
22819         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22820         (mips_mips16_decl_p): Delete.
22821         (mips_nomips16_decl_p): Delete.
22822         (mips_get_compress_on_flags): New function.
22823         (mips_get_compress_off_flags): New function.
22824         (mips_get_compress_mode): New function.
22825         (mips_get_compress_on_name): New function.
22826         (mips_get_compress_off_name): New function.
22827         (mips_insert_attributes): Support multiple compression types.
22828         (mips_merge_decl_attributes): Likewise.
22829         (umips_12bit_offset_address_p): New function.
22830         (mips_start_function_definition): Emit .set micromips directive.
22831         (mips_call_may_need_jalx_p): New function.
22832         (mips_function_ok_for_sibcall): Add microMIPS support.
22833         (mips_print_operand_punctuation): Support short delay slots and
22834         compact jumps.
22835         (umips_swm_mask, umips_swm_encoding): New.
22836         (umips_build_save_restore): New function.
22837         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22838         (was_mips16_p): Remove.
22839         (old_compression_mode): New.
22840         (mips_set_compression_mode): New function.
22841         (mips_set_current_function): Add microMIPS support.
22842         (mips_option_override): Likewise.
22843         (umips_save_restore_pattern_p): New function.
22844         (umips_output_save_restore): New function.
22845         (umips_load_store_pair_p_1): New function.
22846         (umips_load_store_pair_p): New function.
22847         (umips_output_load_store_pair_1): New function.
22848         (umips_output_load_store_pair): New function.
22849         (umips_movep_target_p) New function.
22850         (mips_prepare_pch_save): Add microMIPS support.
22851         * config/mips/mips.h (TARGET_COMPRESSION): New.
22852         (TARGET_CPU_CPP_BUILTINS): Update macro
22853         to use new compression flags and to support microMIPS.
22854         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22855         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22856         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22857         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22858         (ASM_SPEC): Support mmicromips and mno-micromips.
22859         (M16STORE_REG_P): New macro.
22860         (MIPS_CALL): Support TARGET_MICROMIPS.
22861         (MICROMIPS_J): New macro.
22862         (mips_base_mips16): Rename this...
22863         (mips_base_compression_flags): ...to this.
22864         (UMIPS_12BIT_OFFSET_P): New macro.
22865         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22866         (MULTILIB_DIRNAMES): Likewise.
22867 2013-03-20  Richard Biener  <rguenther@suse.de>
22869         PR tree-optimization/56661
22870         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22871         the result does not have to be distinct.
22873 2013-03-20  Richard Biener  <rguenther@suse.de>
22875         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22876         remap_gimple_op_r.
22878 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22879             Steven Bosscher  <steven@gcc.gnu.org>
22881         PR rtl-optimization/56605
22882         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22884 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22886         PR bootstrap/56656
22887         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22888         that require movd instead of movq.
22890 2013-03-20  Richard Biener  <rguenther@suse.de>
22892         * tree-ssa-structalias.c (struct variable_info): Add pointer
22893         to the first field of an aggregate with sub-vars.  Make
22894         this and the pointer to the next subfield its ID.
22895         (vi_next): New function.
22896         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22897         storedanything_id, integer_id): Increment by one.
22898         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22899         get_call_clobber_vi): Adjust.
22900         (solution_set_expand): Simplify and speedup.
22901         (solution_set_add): Inline into ...
22902         (set_union_with_increment): ... this.  Adjust accordingly.
22903         (do_sd_constraint): Likewise.
22904         (do_ds_constraint): Likewise.
22905         (do_complex_constraint): Simplify.
22906         (build_pred_graph): Adjust.
22907         (solve_graph): Likewise.  Simplify and speedup.
22908         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22909         get_constraint_for_component_ref, get_constraint_for_1,
22910         first_vi_for_offset, first_or_preceding_vi_for_offset,
22911         create_function_info_for, create_variable_info_for_1,
22912         create_variable_info_for, intra_create_variable_infos): Adjust.
22913         (init_base_vars): Push NULL for ID zero.
22914         (compute_points_to_sets): Adjust.
22916 2013-03-20  Richard Biener  <rguenther@suse.de>
22918         * cfgloop.c (verify_loop_structure): Streamline and avoid
22919         ICEing on corrupt loop tree.
22920         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22921         loop tree.
22923 2013-03-20  Richard Biener  <rguenther@suse.de>
22925         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22926         check whether an SSA update is needed.
22928 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22930         * config/mips/constraints.md (T): Rename to...
22931         (Yf): ...this.
22932         (U): Rename to...
22933         (Yd): ...this.
22934         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22935         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22937 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22939         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22940         (*subsi3_carryin_uxtw): Likewise.
22942 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22944         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
22945         (*rorsi3_insn_uxtw): Likewise.
22947 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22949         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
22950         (*extrsi5_insn_uxtw): Likewise.
22952 2013-03-19  Richard Biener  <rguenther@suse.de>
22954         PR tree-optimization/56273
22955         * passes.c (init_optimization_passes): Move second VRP after DOM.
22957 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
22959         * config/i386/i386.md (*movti_internal): Merge from
22960         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
22961         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
22962         nox64 isa attributes.
22964 2013-03-18  Richard Biener  <rguenther@suse.de>
22966         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
22967         (unite): Likewise.
22968         (merge_node_constraints): Likewise.
22969         (build_succ_graph): Likewise.
22970         (valid_graph_edge): Inline into single caller.
22971         (unify_nodes): Likewise.  Use bitmap_set_bit return value
22972         and cache varinfo.
22973         (scc_visit): Fix formatting and variable use.
22974         (do_sd_constraint): Use gcc_checking_assert.
22975         (do_ds_constraint): Likewise.
22976         (do_complex_constraint): Likewise.
22977         (condense_visit): Likewise.  Cleanup.
22978         (dump_pred_graph): New function.
22979         (perform_var_substitution): Dump the pred-graph before
22980         variable substitution.
22981         (find_equivalent_node): Use gcc_checking_assert.
22982         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
22984 2013-03-18  Richard Biener  <rguenther@suse.de>
22986         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
22987         Remove cond_expr_stmt_list argument and do not gimplify the
22988         built expression.
22989         (vect_loop_versioning): Adjust.
22990         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
22991         Cleanup to use less temporaries.
22992         (vect_create_data_ref_ptr): Cleanup.
22994 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
22996         PR tree-optimization/56635
22997         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
22998         require types_compatible_p types.
23000 2013-03-18  Nick Clifton  <nickc@redhat.com>
23002         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23003         spurious backslash.
23005         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23006         Add missing line to comment describing function.
23008 2013-03-18  Richard Biener  <rguenther@suse.de>
23010         PR tree-optimization/56210
23011         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23012         Handle string / character search functions.
23013         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23015 2013-03-18  Richard Biener  <rguenther@suse.de>
23017         PR middle-end/56483
23018         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23019         and implement properly.
23020         * gimple.h (gimple_cond_single_var_p): Remove.
23022 2013-03-18  Richard Biener  <rguenther@suse.de>
23024         * tree-data-ref.h (find_data_references_in_loop): Declare.
23025         * tree-data-ref.c (get_references_in_stmt): Use a stack
23026         vector pre-allocated in the callers.
23027         (find_data_references_in_stmt): Adjust.
23028         (graphite_find_data_references_in_stmt): Likewise.
23029         (create_rdg_vertices): Likewise.
23030         (find_data_references_in_loop): Export.
23031         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23032         Compute dependences here...
23033         (vect_analyze_data_refs): ...not here.  When we encounter
23034         a non-vectorizable data reference in basic-block vectorization
23035         truncate the data reference vector.  Do not bother to
23036         fixup data-dependence information for gather loads.
23037         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23038         of data references, as reported.
23040 2013-03-18  Richard Biener  <rguenther@suse.de>
23042         PR tree-optimization/3713
23043         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23044         has_constants and expr.
23045         (stmt_has_constants): Properly valueize SSA names when deciding
23046         whether the stmt has constants.
23048 2013-03-18  Richard Biener  <rguenther@suse.de>
23050         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23051         whole function when there is nothing to do.
23052         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23053         * tree-vectorizer.c (vectorize_loops): Update virtual and
23054         loop-closed SSA once.
23055         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23057 2013-03-18  Richard Biener  <rguenther@suse.de>
23059         PR middle-end/56113
23060         * domwalk.c (bb_postorder): New global static.
23061         (cmp_bb_postorder): New function.
23062         (walk_dominator_tree): Replace scheme imposing an order for
23063         visiting dominator sons by one sorting them at the time they
23064         are pushed on the stack.
23066 2013-03-18  Richard Biener  <rguenther@suse.de>
23068         PR tree-optimization/39326
23069         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23070         (struct mem_ref): Replace mem member with ao_ref typed member.
23071         (MEM_ANALYZABLE): Adjust.
23072         (memref_eq): Likewise.
23073         (mem_ref_alloc): Likewise.
23074         (gather_mem_refs_stmt): Likewise.
23075         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23076         (execute_sm_if_changed_flag_set): Adjust.
23077         (execute_sm): Likewise.
23078         (ref_always_accessed_p): Likewise.
23079         (refs_independent_p): Likewise.
23080         (can_sm_ref_p): Likewise.
23082 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23084         PR c/56566
23085         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23086         return 1 even for !unsignedp.
23088 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23090         * config/i386/i386.md (isa): Add x64 and nox64.
23091         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23092         (*pushtf): Enable *roF alternative for x64 isa only.
23093         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23094         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23095         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23096         (*movtf_internal): Merge from *movtf_internal_rex64 and
23097         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23098         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23099         nox64 isa attributes.
23100         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23101         nox64 isa attributes.
23102         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23104 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23106         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23108 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23110         PR target/56640
23111         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23112         class names.  Remove trailing comma after "ALL_REGS".
23114 2013-03-16  Jan Hubicka  <jh@suse.cz>
23116         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23117         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23118         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23119         of cgraph_get_create_node.
23120         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23122 2013-03-16  Jason Merrill  <jason@redhat.com>
23124         PR debug/49090
23125         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23126         with DW_AT_default_value.
23128 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23130         * BASE-VER: Set to 4.9.0.
23132 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23134         PR target/56619
23135         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23136         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23137         Document _x* TSX intrinsics.
23139 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23140             David Holsgrove  <david.holsgrove@xilinx.com>
23142         * configure.ac: Add MicroBlaze TLS support detection.
23143         * configure: Regenerate.
23144         * config/microblaze/microblaze-protos.h
23145         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23146         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23147         * config/microblaze/microblaze.c (microblaze_address_type): Add
23148         ADDRESS_TLS and tls_reloc address types.
23149         (microblaze_address_info): Add tls_reloc.
23150         (TARGET_HAVE_TLS): Define.
23151         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23152          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23153          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23154         load_tls_operand,  microblaze_call_tls_get_addr,
23155         microblaze_legitimize_tls_address): New functions.
23156         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23157         (get_base_reg): Use microblaze_tls_symbol_p.
23158         (microblaze_classify_address): Handle TLS.
23159         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23160         label_mentioned_p and microblaze_tls_referenced_p.
23161         (microblaze_legitimize_address): Handle TLS.
23162         (microblaze_address_insns): Handle ADDRESS_TLS.
23163         (pic_address_needs_scratch): Handle TLS.
23164         (print_operand_address): Handle TLS.
23165         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23166         (microblaze_expand_move): Handle TLS.
23167         (microblaze_legitimate_constant_p): Check
23168         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23169         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23170         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23171         (PIC_OFFSET_TABLE_REGNUM): Set.
23172         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23173         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23174         (addsi3, movsi_internal2, movdf_internal): Update constraints
23175         * config/microblaze/predicates.md (arith_plus_operand): Define
23176         (move_operand): Redefine as move_src_operand,
23177         check microblaze_tls_referenced_p.
23179 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23181         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23182         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23184 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23186         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23187         CC mode for AND.
23189 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23191         PR tree-optimization/53265
23192         * common.opt (Waggressive-loop-optimizations): New option.
23193         * tree-ssa-loop-niter.c: Include tree-pass.h.
23194         (do_warn_aggressive_loop_optimizations): New function.
23195         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23196         if number_of_latch_executions returned constant.
23197         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23198         early.  If number_of_latch_executions returned constant, set
23199         nb_iterations_upper_bound back to it.
23200         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23201         field.
23202         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23203         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23205         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23206         (MULTILIB_OSDIRNAMES): Set.
23207         * genmultilib: If defaultosdirname doesn't start with :: , set
23208         defaultosdirname2 instead, clear it and emit two . multilib_raw
23209         entries instead of just one.
23211 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23213         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23214         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23215         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23216         (SUBTARGET_OVERRIDE_OPTIONS): New.
23218 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23220         PR target/49880
23221         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23222         (musermode): Convert to Var(TARGET_USERMODE).
23223         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23224         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23225         * config/sh/sh.c (sh_option_override): Use
23226         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23227         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23228         condition.
23229         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23230         TARGET_SH4.
23231         (udivsi3_i4_single, divsi3_i4_single): Use
23232         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23234 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23236         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23237         default setting.
23239 2013-03-13  Richard Biener  <rguenther@suse.de>
23241         PR tree-optimization/56608
23242         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23243         calls when vectorizing basic-blocks.
23245 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23247         PR plugins/45078
23248         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23249         tm_file.
23251 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23253         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23255 2013-03-11  Jan Hubicka  <jh@suse.cz>
23257         PR lto/56557
23258         * lto-streamer-out.c (output_symbol_p): Skip references from
23259         constructors of external variables.
23261 2013-03-11  Jan Hubicka  <jh@suse.cz>
23263         PR middle-end/56571
23264         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23265         from pseudos.
23266         * emit-rtl.c (verify_rtx_sharing): Likewise.
23267         (copy_insn_1): Likewise.
23268         * rtl.c (copy_rtx): Likewise.
23270 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23272         PR target/56591
23273         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23274         output_operand_lossage message.
23276 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23278         PR target/56470
23279         * arm.c (shift_op): Validate RTL pattern on the fly.
23280         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23281         the RTL.
23283 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23285         PR target/56347
23286         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23287         new call patterns that clobber %fr12.
23288         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23289         split and postreload patterns.
23290         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23291         registers %fr12 and %fr12R as call used.
23293 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23295         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23296         (canon_address, record_store, replace_read, check_mem_read_rtx,
23297         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23298         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23299         rest_of_handle_dse): Likewise.
23301 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23303         PR middle-end/56524
23304         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23305         Add base_optabs.
23306         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23307         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23308         (save_optabs_if_changed): Replace with...
23309         (init_tree_optimization_optabs): ...this.
23310         * optabs.c (save_optabs_if_changed): Rename to...
23311         (init_tree_optimization_optabs): ...this.  Take the optimization node
23312         as argument.  Do nothing if the base optabs are already correct.
23313         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23314         to recompute optabs.
23315         * function.h (function): Remove optabs field.
23316         * function.c (invoke_set_current_function_hook): Call
23317         init_tree_optimization_optabs.  Use the result to initialize
23318         this_fn_optabs.
23320 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23322         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23323         if GTMA_HAS_NO_INSTRUMENTATION.
23324         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23325         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23326         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23327         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23328         GTMA_HAS_NO_INSTRUMENTATION.
23330 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23332         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23333         libasan_preinit.o.
23335 2013-03-08  Marek Polacek  <polacek@redhat.com>
23336             Jakub Jelinek  <jakub@redhat.com>
23338         PR tree-optimization/56478
23339         * predict.c (is_comparison_with_loop_invariant_p): Change the
23340         type of loop_step to tree.
23341         (predict_loops): Adjust.
23342         (predict_iv_comparison): Perform the computations on double_ints.
23344 2013-03-08  Richard Biener  <rguenther@suse.de>
23346         PR tree-optimization/56570
23347         * tree-cfg.c (verify_expr_location_1): Verify locations for
23348         DECL_DEBUG_EXPR.
23349         * tree-sra.c (create_access_replacement): Strip locations
23350         from DECL_DEBUG_EXPRs.
23352 2013-03-08  Richard Biener  <rguenther@suse.de>
23354         * tree-inline.c (expand_call_inline): Do not associate
23355         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23356         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23358 2013-03-08  Richard Biener  <rguenther@suse.de>
23360         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23361         or block changes with -Og.  Fix for location / block encoding
23362         changes and PHI arguments with locations.
23364 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23366         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23367         for all counters.
23368         (struct output_info): Likewise.
23369         (register_overhead): Remove bad gcc_assert.
23370         (bitmap_find_bit): If there is only a single bitmap element, do not
23371         count a miss as a search.
23372         (print_statistics): Update for counter type changes.
23373         (dump_bitmap_statistics): Likewise.  Print headers such that they
23374         are properly lined up with the printed counters.
23376 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23378         PR tree-optimization/56559
23379         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23380         check that it has only a single use.
23382 2013-03-07  Richard Biener  <rguenther@suse.de>
23384         * doc/invoke.texi (fwhole-program): Discourage use in combination
23385         with -flto.
23387 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23389         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23391         PR tree-optimization/56539
23392         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23393         instead of GSI_CONTINUE_LINKING as last argument to
23394         force_gimple_operand_gsi.  Adjust function comment.
23396         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23397         aarch64-cores.def.
23399         PR middle-end/56548
23400         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23401         promoted mode, convert the result back to the original mode.
23403 2013-03-06  Richard Biener  <rguenther@suse.de>
23405         PR middle-end/56294
23406         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23407         (insert_updated_phi_nodes_compare_uids): New function.
23408         (update_ssa): Sort symbols_to_rename after UID before
23409         traversing it to insert PHI nodes.
23411 2013-03-06  Richard Biener  <rguenther@suse.de>
23413         PR middle-end/50494
23414         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23415         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23417         Revert
23418         2013-02-13  Richard Biener  <rguenther@suse.de>
23420         PR lto/50494
23421         * varasm.c (output_constant_def_1): Get the decl representing
23422         the constant as argument.
23423         (output_constant_def): Wrap output_constant_def_1.
23424         (make_decl_rtl): Use output_constant_def_1 with the decl
23425         representing the constant.
23426         (build_constant_desc): Optionally re-use a decl already
23427         representing the constant.
23428         (tree_output_constant_def): Adjust.
23430 2013-03-06  Joey Ye  <joey.ye@arm.com>
23432         PR lto/50293
23433         * gcc.c (convert_white_space): New function.
23434         (main): Handles white space in function name.
23436 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23438         PR target/56529
23439         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23440         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23441         to SH_DIV_CALL_TABLE for TARGET_SH2.
23442         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23443         list.
23444         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23445         call-table options.
23447 2013-03-05  Sterling Augustine  <saugustine@google.com>
23448             Cary Coutant  <ccoutant@google.com>
23450         PR debug/55364
23451         * dwarf2out.c (resolve_addr): Don't call
23452         remove_loc_list_addr_table_entries a second time for the same
23453         expression.
23455 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23457         PR debug/56510
23458         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23459         (avoid_complex_debug_insns): New function.
23460         (expand_debug_locations): Call it.
23462         PR rtl-optimization/56484
23463         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23464         lifetimes of hard registers on small register class machines.
23466 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23468         * config/microblaze/microblaze-protos.h: Rename
23469         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23470         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23471         fast_interrupt.
23472         (microblaze_fast_interrupt_function_p): New function.
23473         (microblaze_is_interrupt_handler): Rename to
23474         microblaze_is_interrupt_variant and add fast_interrupt check.
23475         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23476         (save_restore_insns): Likewise.
23477         (compute_frame_size): Likewise.
23478         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23479         (microblaze_globalize_label): Likewise.
23480         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23481         * config/microblaze/microblaze.md: Use wrapper
23482         microblaze_is_interrupt_variant.
23484 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23486         * sdbout.c (sdbout_one_type): Switch to current function's section
23487         supporting cold/hot.
23489 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23491         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23492         -mxl-reorder.
23494 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23496         PR middle-end/56461
23497         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23498         if VALGRIND_GET_VBITS is defined, temporarily make object
23499         memory all defined, and restore previous valgrind addressability
23500         and definability afterwards.  Free this_object at the end.
23502         PR middle-end/56461
23503         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23504         right before calling lra_create_live_ranges, also call it
23505         when clearing live_p.  Only call lra_clear_live_ranges
23506         at the end if live_p.
23508         PR middle-end/56461
23509         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23511 2013-03-05  Richard Biener  <rguenther@suse.de>
23513         PR tree-optimization/56521
23514         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23515         value-id.
23517 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23519         PR c++/55135
23520         * except.h (remove_unreachable_eh_regions): New prototype.
23521         * except.c (remove_eh_handler_splicer): New function, split out
23522         of remove_eh_handler.
23523         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23524         warning about running it on many EH regions one at a time.
23525         (remove_unreachable_eh_regions_worker): New function, walk the
23526         EH tree in depth-first order and remove non-marked regions.
23527         (remove_unreachable_eh_regions): New function.
23528         * tree-eh.c (mark_reachable_handlers): New function, split out
23529         from remove_unreachable_handlers.
23530         (remove_unreachable_handlers): Use mark_reachable_handlers and
23531         remove_unreachable_eh_regions.
23532         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23533         and remove_unreachable_eh_regions.
23535 2013-03-05  Richard Biener  <rguenther@suse.de>
23537         PR middle-end/56525
23538         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23539         not freeing them until the end.
23541 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23543         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23545 2013-03-05  Richard Biener  <rguenther@suse.de>
23547         PR tree-optimization/56270
23548         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23549         of loads after scheduling an SLP instance.
23551 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23553         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23554         tic6x.exp.
23555         (check_gcc_parallelize): Run guality.exp as a separate job from
23556         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23557         struct-layout-1.exp with stackalign.exp.
23559         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23561         PR middle-end/56461
23562         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23563         load_index sbitmap even if some bit in it isn't set.
23565         PR middle-end/56461
23566         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23567         (discover_iteration_bound_by_body_walk): Change queues to
23568         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23569         spelling in comment.  Call safe_push on queues[bound_index] directly.
23570         Release queues[queue_index] in every iteration unconditionally.
23571         Release bounds vector.
23573         PR middle-end/56461
23574         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23575         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23576         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23577         inner_phis vector.
23579 2013-03-05  Richard Biener  <rguenther@suse.de>
23581         PR lto/56515
23582         * tree-inline.c (remap_blocks_to_null): New function.
23583         (expand_call_inline): When expanding a call stmt without
23584         an associated block inline remap all callee blocks to NULL.
23586 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23588         PR rtl-optimization/56494
23589         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23590         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23591         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23593         PR middle-end/56461
23594         * sel-sched-ir.c (free_sched_pools): Release
23595         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23596         if succs_info_pool.max_top isn't -1.
23598         PR bootstrap/56509
23599         * opts.c (opts_obstack, opts_concat): Moved to...
23600         * opts-common.c (opts_obstack, opts_concat): ... here.
23602 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23604         PR middle-end/56461
23605         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23607 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23609         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23610         all appropriate places.
23612 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23614         PR tree-optimization/56424
23615         * ipa-split.c (split_function): Do not set the RSO flag if result is
23616         not by reference and its type is a register type.
23618 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23620         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23621         (microblaze_legitimate_pic_operand): Likewise
23622         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23623         new function microblaze_legitimate_pic_operand
23624         * config/microblaze/microblaze-protos.h
23625         (microblaze_legitimate_pic_operand): Declare.
23627 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23629         * config/microblaze/predicates.md (call_insn_simple_operand):
23630         New predicate for supported rtx code types.
23631         * config/microblaze/microblaze.md (call_internal1): Use
23632         call_insn_simple_operand predicate.
23634 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23636         PR middle-end/56461
23637         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23638         partitions.ordered_remove.
23640         PR middle-end/56461
23641         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23642         vec_oprnds0.create (1) for modifier == NONE.
23644         PR middle-end/56461
23645         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23646         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23647         vec_oprnds1 right before pushing anything to it for
23648         scalar_shift_arg.
23650         PR middle-end/56461
23651         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23652         set nbbs to 0 instead of having separate code path.
23653         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23654         instead of false as last argument if returning NULL.
23656 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23658         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23659         the attribute is now called "target" instead of "option".
23660         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23661         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23662         attribute/pragma name for TARGET_OPTION_VALID_P and
23663         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23664         * doc/tm.texi: Regenerated.
23666 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23668         * config/microblaze/microblaze.c:
23669         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23670         * config/microblaze/microblaze.h: Add -mxl-reorder to
23671         DRIVER_SELF_SPECS.
23672         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23673         instructions emitted if TARGET_REORDER.
23674         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23675         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23676         separately.
23678 2013-03-01  Xinliang David Li  <davidxl@google.com>
23680         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23681         walk length.
23683 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23685         PR middle-end/56461
23686         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23687         vector even when returning true.  Fix up function comment formatting.
23689         PR middle-end/56461
23690         * ira-build.c (ira_loop_nodes_count): New variable.
23691         (create_loop_tree_nodes): Initialize it.
23692         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23694         PR middle-end/56461
23695         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23696         method on dr_chain and result_chain.
23697         * tree-vect-stmts.c (vectorizable_store): Only call
23698         result_chain.create if j == 0.
23700         PR middle-end/56461
23701         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23702         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23703         before overwriting it.
23705 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23707         * doc/extended.texi (C Extensions): Change order in @menu
23708         to match @node.
23709         (Other MIPS Built-in Functions): Move last MIPS entry before
23710         "picoChip Built-in Functions".
23711         (SH Built-in Functions): Move after RX Built-in Functions.
23712         * doc/gcc.texi (Introduction): Change order in @menu
23713         to match @node.
23714         * doc/md.texi (Constraints): Ditto.
23715         * gty.texi (Type Information): Ditto.
23716         (User-provided marking routines for template types): Make
23717         subsection.
23718         * doc/invoke.texi (AArch64 Options): Move before
23719         "Adapteva Epiphany Options".
23721 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23722             Jakub Jelinek  <jakub@redhat.com>
23724         PR sanitizer/56454
23725         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23726         no_address_safety_analysis attribute.
23727         * doc/extend.texi (no_address_safety_attribute): Rename to
23728         no_sanitize_address attribute, mention no_address_safety_analysis
23729         attribute as deprecated alias.
23731 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23733         PR middle-end/56461
23734         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23735         type to vec<vec<tree> > *.
23736         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23737         to be vec<tree> instead of vec<tree> *, set vec_defs
23738         to vNULL and call vec_defs.create (number_of_vects), adjust other
23739         uses of vec_defs.
23740         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23741         vectorizable_condition): Adjust vect_get_slp_defs callers.
23743 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23745         * config/aarch64/aarch64.c
23746         (aarch64_float_const_representable): Remove unused variable.
23748 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23750         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23752 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23754         * config/aarch64/aarch64-builtins.c
23755         (aarch64_init_simd_builtins): Make static.
23757 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23759         * config/aarch64/aarch64.c
23760         (aarch64_simd_make_constant): Make static.
23762 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23764         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23765         with no initialization to the RHS of debug statements.
23767 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23769         PR tree-optimization/56294
23770         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23771         Adjust dumping.
23772         (get_access_replacement): Do not call create_access_replacement.
23773         Assert a replacement exists.
23774         (get_repl_default_def_ssa_name): Create the replacement declaration
23775         itself.
23777 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23779         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23780         final_end_function.
23782 2013-02-28  Marek Polacek  <polacek@redhat.com>
23784         PR rtl-optimization/56466
23785         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23786         if we're changing a loop.
23787         (peel_loops_completely): Likewise.
23789 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23791         PR c++/55813
23792         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23794 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23796         PR target/56445
23797         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23798         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23799         INTX_FTYPE_FX, FX_FTYPE_INTX.
23800         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23802 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23804         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23805         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23806         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23807         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23808         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23809         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23810         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23811         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23812         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23813         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23814         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23815         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23816         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23817         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23818         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23819         (avrxmega6): Increase max flash segments from 5 to 6.
23820         * config/avr/t-multilib: Regenerate.
23821         * config/avr/avr-tables.opt: Regenerate.
23822         * doc/avr-mmcu.texi: Regenerate.
23824 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23826         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23827         (avr_device_to_arch): Rename to avr_device_to_ld.
23828         (avr_device_to_as): New prototype.
23829         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23830         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23831         * config/avr/driver-avr.c (avr_device_to_as): New.
23832         (avr_device_to_arch): Rename to avr_device_to_ld.
23834 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23836         PR middle-end/56461
23837         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23838         method on dr_chain and result_chain.
23840         PR middle-end/56461
23841         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23842         pointer_set_destroy on not_executed_last_iteration.
23844         PR middle-end/56461
23845         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23847         PR middle-end/56461
23848         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23849         FOR_EACH_DEFINED_FUNCTION when freeing state.
23851         PR middle-end/56461
23852         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23853         pool_free.
23854         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23855         overwriting it.
23857         PR middle-end/56461
23858         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23859         known_aggs[i].items and release known_aggs vector.
23861         PR middle-end/56461
23862         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23864 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23866         * config/microblaze/microblaze.c (microblaze_emit_compare):
23867         Use xor for EQ/NE comparisions.
23868         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23869         (cbranchsf4): Adjust operator to comparison_operator.
23871 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23873         PR middle-end/56461
23874         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23875         vector.
23876         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23877         vec_safe_push, always update *slot.
23878         (redirect_edge_var_map_clear): Use vec_free.
23879         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23880         (free_var_map_entry): Use vec_free.
23881         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23882         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23884 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23886         PR middle-end/45472
23887         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23888         when the may_trap_p bit of the exprs being merged differs.
23889         Reorder tests for speculativeness in the logical and operator.
23891 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23893         * incpath.c (add_standard_paths): Use reconcat instead of concat
23894         where appropriate and avoid leaking memory.
23896         * opts.h: Include obstack.h.
23897         (opts_concat): New prototype.
23898         (opts_obstack): New declaration.
23899         * opts.c (opts_concat): New function.
23900         (opts_obstack): New variable.
23901         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23902         (finish_options): Use opts_concat instead of concat
23903         and XOBNEWVEC instead of XNEWVEC.
23904         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23905         generate_option): Likewise.
23906         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23907         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23909         PR target/56455
23910         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23911         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23913 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23915         PR middle-end/56461
23916         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23918 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23920         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23921         (arm_block_move_unaligned_straight): Likewise.
23922         (arm_adjust_block_mem): Likewise.
23924 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23926         PR target/48901
23927         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23928         temp, cond and label.
23929         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23931         PR target/52500
23932         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23933         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23935         PR target/52501
23936         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23937         of prologue/epilogue functions.
23939         PR target/52550
23940         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23941         Remove unused variable cfa_offset.
23942         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
23944         PR target/54639
23945         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
23946         type promotion to unsigned.
23948         PR target/54640
23949         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
23950         for HOST_WIDE_INT of 32 bit / same size as int.
23951         (arm_block_move_unaligned_straight): Likewise.
23952         (arm_adjust_block_mem): Likewise.
23954         PR target/54662
23955         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
23956         ALL_CFLAGS.
23958 2013-02-26  Marek Polacek  <polacek@redhat.com>
23960         PR tree-optimization/56426
23961         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
23963 2013-02-26  Richard Biener  <rguenther@suse.de>
23965         PR target/56444
23966         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
23967         unused variable loops.
23969 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23971         PR tree-optimization/56448
23972         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
23973         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
23974         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
23975         later operands of the references, or even first operand for
23976         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
23978         PR tree-optimization/56443
23979         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
23980         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
23981         to type_for_mode langhook.
23983 2013-02-25  Matt Turner  <mattst88@gmail.com>
23985         * doc/invoke.texi: Document r4700.
23987 2013-02-25  Richard Biener  <rguenther@suse.de>
23989         PR tree-optimization/56175
23990         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
23991         split out from ...
23992         (simplify_bitwise_binary): ... here.  Also guard the conversion
23993         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
23995 2013-02-25  Catherine Moore  <clm@codesourcery.com>
23997         Revert:
23998         2013-02-24  Catherine Moore  <clm@codesourcery.com>
23999             Maciej W. Rozycki  <macro@codesourcery.com>
24000             Tom de Vries  <tom@codesourcery.com>
24001             Nathan Sidwell  <nathan@codesourcery.com>
24002             Iain Sandoe  <iain@codesourcery.com>
24003             Nathan Froyd  <froydnj@codesourcery.com>
24004             Chao-ying Fu  <fu@mips.com>
24006         * doc/extend.texi: (micromips, nomicromips, nocompression):
24007         Document new function attributes.
24008         * doc/invoke.texi (minterlink-compressed, mmicromips,
24009         m14k, m14ke, m14kec): Document new options.
24010         (minterlink-mips16): Update documentation.
24011         * doc/md.texi (ZC, ZD): Document new constraints.
24012         * configure.ac (gcc_cv_as_micromips): Check if linker
24013         supports the .set micromips directive.
24014         * configure: Regenerate.
24015         * config.in: Regenerate.
24016         * config/mips/mips-tables.opt: Regenerate.
24017         * config/mips/micromips.md: New file.
24018         * constraints.md (ZC, AD): New constraints.
24019         * config/mips/predicates.md (movep_src_register): New predicate.
24020         (movep_src_operand): New predicate.
24021         (non_volatile_mem_operand): New predicate.
24022         * config/mips/mips.md (multimem): New type.
24023         (length): Differentiate between 17-bit and 18-bit branch offsets.
24024         (MOVEP1, MOVEP2): New mode iterator.
24025         (mov_<load>l): Use ZC constraint.
24026         (mov_<load>r): Likewise.
24027         (mov_<store>l): Likewise.
24028         (mov_<store>r): Likewise.
24029         (*branch_equality<mode>_inverted): Add microMIPS support.
24030         (*branch_equality<mode>): Likewise.
24031         (*jump_absolute): Likewise.
24032         (indirect_jump_<mode>): Likewise.
24033         (tablejump_<mode>): Likewise.
24034         (<optab>_internal): Likewise.
24035         (sibcall_internal): Likewise.
24036         (sibcall_value_internal): Likewise.
24037         (prefetch): Use constraint ZD.
24038         * config/mips/mips.opt (minterlink-compressed): New option.
24039         (minterlink-mips16): Now an alias for minterlink-compressed.
24040         (mmicromips): New option.
24041         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24042         (compare_and_swap_12): Likewise.
24043         (sync_add<mode>): Likewise.
24044         (sync_<optab>_12): Likewise.
24045         (sync_old_<optab>_12): Likewise.
24046         (sync_new_<optab>_12): Likewise.
24047         (sync_nand_12): Likewise.
24048         (sync_old_nand_12): Likewise.
24049         (sync_new_nand_12): Likewise.
24050         (sync_sub<mode>): Likewise.
24051         (sync_old_add<mode>): Likewise.
24052         (sync_old_sub<mode>): Likewise.
24053         (sync_new_add<mode>): Likewise.
24054         (sync_new_sub<mode>): Likewise.
24055         (sync_<optab><mode>): Likewise.
24056         (sync_old_<optab><mode>): Likewise.
24057         (sync_new_<optab><mode>): Likewise.
24058         (sync_nand<mode>): Likewise.
24059         (sync_old_nand<mode>): Likewise.
24060         (sync_new_nand<mode>): Likewise.
24061         (sync_lock_test_and_set<mode>): Likewise.
24062         (test_and_set_12): Likewise.
24063         (atomic_compare_and_swap<mode>): Likewise.
24064         (atomic_exchange<mode>_llsc): Likewise.
24065         (atomic_fetch_add<mode>_llsc): Likewise.
24066         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24067         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24068         (umips_save_restore_pattern_p): Likewise.
24069         (umips_load_store_pair_p): Likewise.
24070         (umips_output_load_store_pair): Likewise.
24071         (umips_movep_target_p): Likewise.
24072         (umips_12bit_offset_address_p): Likewise.
24073         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24074         (mips_base_mips16): Rename this...
24075         (mips_base_compression_flags): ...to this. Update all uses.
24076         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24077         (mips_mips16_decl_p): Delete.
24078         (mips_nomips16_decl_p): Delete.
24079         (mips_get_compress_on_flags): New function.
24080         (mips_get_compress_off_flags): New function.
24081         (mips_get_compress_mode): New function.
24082         (mips_get_compress_on_name): New function.
24083         (mips_get_compress_off_name): New function.
24084         (mips_insert_attributes): Support multiple compression types.
24085         (mips_merge_decl_attributes): Likewise.
24086         (umips_12bit_offset_address_p): New function.
24087         (mips_start_function_definition): Emit .set micromips directive.
24088         (mips_call_may_need_jalx_p): New function.
24089         (mips_function_ok_for_sibcall): Add microMIPS support.
24090         (mips_print_operand_punctuation): Support short delay slots and
24091         compact jumps.
24092         (umips_swm_mask, umips_swm_encoding): New.
24093         (umips_build_save_restore): New function.
24094         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24095         (was_mips16_p): Remove.
24096         (old_compression_mode): New.
24097         (mips_set_compression_mode): New function.
24098         (mips_set_current_function): Add microMIPS support.
24099         (mips_option_override): Likewise.
24100         (umips_save_restore_pattern_p): New function.
24101         (umips_output_save_restore): New function.
24102         (umips_load_store_pair_p_1): New function.
24103         (umips_load_store_pair_p): New function.
24104         (umips_output_load_store_pair_1): New function.
24105         (umips_output_load_store_pair): New function.
24106         (umips_movep_target_p) New function.
24107         (mips_prepare_pch_save): Add microMIPS support.
24108         * config/mips/mips.h (TARGET_COMPRESSION): New.
24109         (TARGET_CPU_CPP_BUILTINS): Update macro
24110         to use new compression flags and to support microMIPS.
24111         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24112         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24113         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24114         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24115         (ASM_SPEC): Support mmicromips and mno-micromips.
24116         (M16STORE_REG_P): New macro.
24117         (MIPS_CALL): Support TARGET_MICROMIPS.
24118         (MICROMIPS_J): New macro.
24119         (mips_base_mips16): Rename this...
24120         (mips_base_compression_flags): ...to this.
24121         (UMIPS_12BIT_OFFSET_P): New macro.
24122         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24123         (MULTILIB_DIRNAMES): Likewise.
24125 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24127         PR rtl-optimization/56131
24128         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24129         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24130         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24132 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24134         * doc/invoke.texi (-fsanitize=): Move from optimization
24135         to debugging options.
24137 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24139         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24141 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24142             Alexander Monakov  <amonakov@ispras.ru>
24144         PR middle-end/56077
24145         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24146         flush pending lists also on non-jumps.  Adjust comment.
24148 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24149             Maciej W. Rozycki  <macro@codesourcery.com>
24150             Tom de Vries  <tom@codesourcery.com>
24151             Nathan Sidwell  <nathan@codesourcery.com>
24152             Iain Sandoe  <iain@codesourcery.com>
24153             Nathan Froyd  <froydnj@codesourcery.com>
24154             Chao-ying Fu  <fu@mips.com>
24156         * doc/extend.texi: (micromips, nomicromips, nocompression):
24157         Document new function attributes.
24158         * doc/invoke.texi (minterlink-compressed, mmicromips,
24159         m14k, m14ke, m14kec): Document new options.
24160         (minterlink-mips16): Update documentation.
24161         * doc/md.texi (ZC, ZD): Document new constraints.
24162         * configure.ac (gcc_cv_as_micromips): Check if linker
24163         supports the .set micromips directive.
24164         * configure: Regenerate.
24165         * config.in: Regenerate.
24166         * config/mips/mips-tables.opt: Regenerate.
24167         * config/mips/micromips.md: New file.
24168         * constraints.md (ZC, AD): New constraints.
24169         * config/mips/predicates.md (movep_src_register): New predicate.
24170         (movep_src_operand): New predicate.
24171         (non_volatile_mem_operand): New predicate.
24172         * config/mips/mips.md (multimem): New type.
24173         (length): Differentiate between 17-bit and 18-bit branch offsets.
24174         (MOVEP1, MOVEP2): New mode iterator.
24175         (mov_<load>l): Use ZC constraint.
24176         (mov_<load>r): Likewise.
24177         (mov_<store>l): Likewise.
24178         (mov_<store>r): Likewise.
24179         (*branch_equality<mode>_inverted): Add microMIPS support.
24180         (*branch_equality<mode>): Likewise.
24181         (*jump_absolute): Likewise.
24182         (indirect_jump_<mode>): Likewise.
24183         (tablejump_<mode>): Likewise.
24184         (<optab>_internal): Likewise.
24185         (sibcall_internal): Likewise.
24186         (sibcall_value_internal): Likewise.
24187         (prefetch): Use constraint ZD.
24188         * config/mips/mips.opt (minterlink-compressed): New option.
24189         (minterlink-mips16): Now an alias for minterlink-compressed.
24190         (mmicromips): New option.
24191         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24192         (compare_and_swap_12): Likewise.
24193         (sync_add<mode>): Likewise.
24194         (sync_<optab>_12): Likewise.
24195         (sync_old_<optab>_12): Likewise.
24196         (sync_new_<optab>_12): Likewise.
24197         (sync_nand_12): Likewise.
24198         (sync_old_nand_12): Likewise.
24199         (sync_new_nand_12): Likewise.
24200         (sync_sub<mode>): Likewise.
24201         (sync_old_add<mode>): Likewise.
24202         (sync_old_sub<mode>): Likewise.
24203         (sync_new_add<mode>): Likewise.
24204         (sync_new_sub<mode>): Likewise.
24205         (sync_<optab><mode>): Likewise.
24206         (sync_old_<optab><mode>): Likewise.
24207         (sync_new_<optab><mode>): Likewise.
24208         (sync_nand<mode>): Likewise.
24209         (sync_old_nand<mode>): Likewise.
24210         (sync_new_nand<mode>): Likewise.
24211         (sync_lock_test_and_set<mode>): Likewise.
24212         (test_and_set_12): Likewise.
24213         (atomic_compare_and_swap<mode>): Likewise.
24214         (atomic_exchange<mode>_llsc): Likewise.
24215         (atomic_fetch_add<mode>_llsc): Likewise.
24216         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24217         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24218         (umips_save_restore_pattern_p): Likewise.
24219         (umips_load_store_pair_p): Likewise.
24220         (umips_output_load_store_pair): Likewise.
24221         (umips_movep_target_p): Likewise.
24222         (umips_12bit_offset_address_p): Likewise.
24223         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24224         (mips_base_mips16): Rename this...
24225         (mips_base_compression_flags): ...to this. Update all uses.
24226         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24227         (mips_mips16_decl_p): Delete.
24228         (mips_nomips16_decl_p): Delete.
24229         (mips_get_compress_on_flags): New function.
24230         (mips_get_compress_off_flags): New function.
24231         (mips_get_compress_mode): New function.
24232         (mips_get_compress_on_name): New function.
24233         (mips_get_compress_off_name): New function.
24234         (mips_insert_attributes): Support multiple compression types.
24235         (mips_merge_decl_attributes): Likewise.
24236         (umips_12bit_offset_address_p): New function.
24237         (mips_start_function_definition): Emit .set micromips directive.
24238         (mips_call_may_need_jalx_p): New function.
24239         (mips_function_ok_for_sibcall): Add microMIPS support.
24240         (mips_print_operand_punctuation): Support short delay slots and
24241         compact jumps.
24242         (umips_swm_mask, umips_swm_encoding): New.
24243         (umips_build_save_restore): New function.
24244         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24245         (was_mips16_p): Remove.
24246         (old_compression_mode): New.
24247         (mips_set_compression_mode): New function.
24248         (mips_set_current_function): Add microMIPS support.
24249         (mips_option_override): Likewise.
24250         (umips_save_restore_pattern_p): New function.
24251         (umips_output_save_restore): New function.
24252         (umips_load_store_pair_p_1): New function.
24253         (umips_load_store_pair_p): New function.
24254         (umips_output_load_store_pair_1): New function.
24255         (umips_output_load_store_pair): New function.
24256         (umips_movep_target_p) New function.
24257         (mips_prepare_pch_save): Add microMIPS support.
24258         * config/mips/mips.h (TARGET_COMPRESSION): New.
24259         (TARGET_CPU_CPP_BUILTINS): Update macro
24260         to use new compression flags and to support microMIPS.
24261         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24262         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24263         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24264         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24265         (ASM_SPEC): Support mmicromips and mno-micromips.
24266         (M16STORE_REG_P): New macro.
24267         (MIPS_CALL): Support TARGET_MICROMIPS.
24268         (MICROMIPS_J): New macro.
24269         (mips_base_mips16): Rename this...
24270         (mips_base_compression_flags): ...to this.
24271         (UMIPS_12BIT_OFFSET_P): New macro.
24272         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24273         (MULTILIB_DIRNAMES): Likewise.
24275 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24277         PR target/52555
24278         * target-globals.c (save_target_globals): For init_reg_sets and
24279         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24281 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24283         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24284         * config/aarch64/t-aarch64
24285         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24287 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24289         PR inline-asm/56148
24290         * lra-constraints.c (process_alt_operands): Reload operand
24291         conflicting with earlier clobber only if no more other conflicting
24292         operands.
24294 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24296         PR sanitizer/56393
24297         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24298         if not linking a shared library.
24300 2013-02-22  Seth LaForge  <sethml@google.com>
24302         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24304 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24306         * config/arm/arm.md (split for extendsidi): Update condition.
24307         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24308         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24309         (qhs_zextenddi_cstr): Likewise.
24311 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24313         PR middle-end/56420
24314         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24315         avoid signed wrapping.
24316         (expand_mult): Handle properly multiplication by
24317         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24318         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24319         in the compiler if coeff is HOST_WIDE_INT_MIN.
24320         (expand_divmod): Don't make ext_op1 static, change it's type to
24321         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24323         PR rtl-optimization/50339
24324         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24325         field.
24326         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24327         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24328         into splitting_ashiftrt field.
24329         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24330         ASHIFTRT.
24331         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24332         choices.
24334 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24336         PR middle-end/56108
24337         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24338         are sure to go irrevocable.
24340 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24342         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24343         scalars are valid operands.
24345 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24347         PR tree-optimization/56310
24348         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24349         only matching indices and non-negative final offsets.
24350         (intersect_aggregates_with_edge): Pass src_idx to
24351         agg_replacements_to_vector.  Pass src_idx insstead of index to
24352         intersect_with_agg_replacements.
24354 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24356         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24357         instead of hard-wired defaults.
24359 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24361         * doc/invoke.texi (MIPS Options): Update documentation of the
24362         floating-point multiply-accumulate instruction restrictions.
24364 2013-02-21  Kostya Serebryany  <kcc@google.com>
24366         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24367         asan_shadow_offset on x86_64 linux.
24369 2013-02-21  Richard Biener  <rguenther@suse.de>
24371         PR tree-optimization/56415
24372         Revert
24373         2013-02-11  Richard Biener  <rguenther@suse.de>
24375         PR tree-optimization/56273
24376         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24377         first VRP run.
24379 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24381         PR bootstrap/56258
24382         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24383         instead of @itemx.
24385         PR inline-asm/56405
24386         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24387         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24389 2013-02-20  Jan Hubicka  <jh@suse.cz>
24391         PR tree-optimization/56265
24392         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24393         when target is referenced for first time.
24395 2013-02-20  Richard Biener  <rguenther@suse.de>
24397         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24398         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24399         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24400         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24401         not return anything.
24402         (rename_ssa_copies): Do not remove unused locals.
24403         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24404         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24405         * passes.c (execute_function_todo): Do not schedule unused locals
24406         removal if cleanup_tree_cfg did something.
24407         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24408         about the number of removed locals.
24410 2013-02-20  Richard Biener  <rguenther@suse.de>
24412         PR tree-optimization/56398
24413         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24415 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24417         PR tree-optimization/55334
24418         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24419         restricted pointers to arrays.
24421 2013-02-20  Richard Biener  <rguenther@suse.de>
24422             Jakub Jelinek  <jakub@redhat.com>
24424         PR tree-optimization/56396
24425         * tree-ssa-ccp.c (n_const_val): New static variable.
24426         (get_value): Return NULL for SSA names we don't have a lattice
24427         entry for.
24428         (ccp_initialize): Initialize n_const_val.
24429         * tree-ssa-copy.c (n_copy_of): New static variable.
24430         (init_copy_prop): Initialize n_copy_of.
24431         (get_value): Return NULL_TREE for SSA names we don't have a
24432         lattice entry for.
24434 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24436         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24438 2013-02-20  Richard Biener  <rguenther@suse.de>
24440         * genpreds.c (write_lookup_constraint): Do not compare first
24441         letter of the constraint again.
24443 2013-02-20  Richard Biener  <rguenther@suse.de>
24445         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24446         and ceil_log2.
24447         (get_use_iv_cost): Terminate hashtable walk when coming across
24448         an empty entry.
24450 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24452         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24453         reassociation for avx2 targets.
24455 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24457         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24458         Add version check for v8.10.a to enable microblaze_has_clz
24459         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24460         version and TARGET_PATTERN_COMPARE check
24461         * config/microblaze/microblaze.md: New clzsi2 instruction
24463 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24465         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24466         function before branching.
24468 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24470         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24471         DUMP_INSN_RTX_UID.
24472         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24474 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24476         PR middle-end/55889
24477         * sel-sched.c: Include ira.h.
24478         (implicit_clobber_conflict_p): New function.
24479         (moveup_expr): Use it.
24480         * Makefile.in (sel-sched.o): Depend on ira.h.
24482 2013-02-19  Richard Biener  <rguenther@suse.de>
24484         PR tree-optimization/56384
24485         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24486         (vn_hash_type): Split out from ...
24487         (vn_hash_constant_with_type): ... here.
24488         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24489         (vn_phi_eq): Compare types from vn_phi_s structure.
24490         (vn_phi_lookup): Populate vn_phi_s type.
24491         (vn_phi_insert): Likewise.
24493 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24495         PR tree-optimization/56350
24496         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24497         if haven't found reduction or nested cycle operand, rather than
24498         asserting we must find it.
24500         PR tree-optimization/56381
24501         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24502         to fold_build3.
24504 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24505             Jakub Jelinek  <jakub@redhat.com>
24507         PR target/52555
24508         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24509         (swap_optab_enable): Same.
24510         (init_all_optabs): Use argument instead of global.
24511         * tree.h (struct tree_optimization_option): New field target_optabs.
24512         * expr.h (init_all_optabs): Add argument to prototype.
24513         (TREE_OPTIMIZATION_OPTABS): New.
24514         (save_optabs_if_changed): Protoize.
24515         * optabs.h: Declare this_fn_optabs.
24516         * optabs.c (save_optabs_if_changed): New.
24517         Declare this_fn_optabs.
24518         (init_optabs): Add argument to init_all_optabs() call.
24519         * function.c (invoke_set_current_function_hook): Handle per
24520         function optabs.
24521         * function.h (struct function): New field optabs.
24522         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24523         optimization_current_node has changed.
24524         * target-globals.h (save_target_globals_default_opts): Protoize.
24525         * target-globals.c (save_target_globals_default_opts): New.
24527 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24529         PR target/56347
24530         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24531         registers %fr12 and %fr12R as call used.
24533         PR target/56214
24534         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24535         and HImode, require all displacements to be an integer multiple of
24536         their mode size.
24537         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24538         only allow QImode and HImode when reload is in progress and strict is
24539         true.  Likewise for symbolic addresses.  Use base14_operand to check
24540         displacements in REG+BASE addresses.
24542 2013-02-18  Richard Biener  <rguenther@suse.de>
24544         PR tree-optimization/56366
24545         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24546         handle sign-conversion of outer-loop initial induction value.
24548 2013-02-18  Richard Biener  <rguenther@suse.de>
24550         PR middle-end/56349
24551         * cfghooks.c (merge_blocks): If we merge a latch into another
24552         block adjust references to it.
24553         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24554         (verify_loop_structure): Verify that a recorded latch is in fact
24555         a latch.
24557 2013-02-18  Richard Biener  <rguenther@suse.de>
24559         PR tree-optimization/56321
24560         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24561         order SSA name release and virtual operand unlinking.
24563 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24565         * config/microblaze/microblaze.md (save_stack_block): Define.
24566         (restore_stack_block): Likewise.
24568 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24570         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24571         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24572         * config/microblaze/microblaze.c (microblaze_option_override):
24573         Bail out early for PIC modes when target does not support PIC.
24575 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24577         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24578         Replace with a microblaze version.
24579         (microblaze_trampoline_init): Adapt for microblaze.
24580         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24581         microblaze.
24583 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24584             Dodji Seketeli  <dodji@redhat.com>
24586         PR asan/56330
24587         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24588         (instrument_mem_region_access): Do not forget to always put
24589         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24590         0) statement, even for cases where either 'base' or 'base + len'
24591         are not instrumented -- because they have been previously
24592         instrumented.  Simplify the logic by putting all the statements
24593         instrument 'base + len' inside a sequence, and then insert that
24594         sequence right before the current insertion point.  Then, to
24595         instrument 'base + len', just get an iterator on that statement.
24596         And do not forget to update the pointer to iterator the function
24597         received as argument.
24599 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24601         PR rtl-optimization/56348
24602         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24604 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24606         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24607         (clean_graph_dump_file): Pass base to start_graph_dump.
24609 2013-02-14  Richard Henderson  <rth@redhat.com>
24611         PR target/55941
24612         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24614 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24616         * collect2-aix.h: Define F_LOADONLY.
24618 2013-02-14  Richard Biener  <rguenther@suse.de>
24620         PR lto/50494
24621         * varasm.c (output_constant_def_1): Get the decl representing
24622         the constant as argument.
24623         (output_constant_def): Wrap output_constant_def_1.
24624         (make_decl_rtl): Use output_constant_def_1 with the decl
24625         representing the constant.
24626         (build_constant_desc): Optionally re-use a decl already
24627         representing the constant.
24628         (tree_output_constant_def): Adjust.
24630 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24632         Fix an asan crash
24633         * asan.c (instrument_builtin_call):  Really put the length of the
24634         second source argument into src1_len.
24636 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24638         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24639         argument.  If it is false, don't create edge from then_bb to
24640         fallthru_bb.
24641         (insert_if_then_before_iter): Pass true to it.
24642         (build_check_stmt): Pass false to it.
24643         (transform_statements): Flush hash table only on extended basic
24644         block boundaries, rather than at the beginning of every bb.
24645         Don't flush hash table on nonfreeing_call_p calls.
24646         * tree-flow.h (nonfreeing_call_p): New prototype.
24647         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24649 2013-02-13  David S. Miller  <davem@davemloft.net>
24651         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24653 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24655         PR target/56184
24656         * ira.c (max_regno_before_ira): Move from ...
24657         (ira): ... here.
24658         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24659         vec_safe_length.
24661 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24663         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24665 2013-02-13  Richard Biener  <rguenther@suse.de>
24667         PR lto/56295
24668         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24669         globals in MEM_REFs.
24671 2013-02-13  Richard Biener  <rguenther@suse.de>
24673         * loop-init.c (loop_optimizer_init): Clear loop state when
24674         re-initializing preserved loops.
24675         * loop-unswitch.c (unswitch_single_loop): Return whether
24676         we unswitched the loop.  Do not verify loop state here.
24677         (unswitch_loops): When we unswitched a loop discover new loops.
24679 2013-02-13  Kostya Serebryany  <kcc@google.com>
24681         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24682         on x86_64 linux.
24683         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24685 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24687         Avoid instrumenting duplicated memory access in the same basic block
24688         * Makefile.in (asan.o): Add new dependency on hash-table.h
24689         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24690         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24691         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24692         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24693         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24694         (get_mem_ref_of_assignment): New functions.
24695         (get_mem_refs_of_builtin_call): Extract from
24696         instrument_builtin_call and tweak a little bit to make it fit with
24697         the new signature.
24698         (instrument_builtin_call): Use the new
24699         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24700         of is_gimple_builtin_call.
24701         (instrument_derefs, instrument_mem_region_access): Insert the
24702         instrumented memory reference into the hash table.
24703         (maybe_instrument_assignment): Renamed instrument_assignment into
24704         this, and change it to advance the iterator when instrumentation
24705         actually happened and return true in that case.  This makes it
24706         homogeneous with maybe_instrument_assignment, and thus give a
24707         chance to callers to be more 'regular'.
24708         (transform_statements): Clear the memory reference hash table
24709         whenever we enter a new BB, when we cross a function call, or when
24710         we are done transforming statements.  Use
24711         maybe_instrument_assignment instead of instrumentation.  No more
24712         need to special case maybe_instrument_assignment and advance the
24713         iterator after calling it; it's now handled just like
24714         maybe_instrument_call.  Update comment.
24716 2013-02-13  Richard Biener  <rguenther@suse.de>
24718         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24719         Fix loop discovery code.
24721 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24723         PR inline-asm/56148
24724         * lra-constraints.c (process_alt_operands): Match early clobber
24725         operand with itself.  Check conflicts with earlyclobber only if
24726         the operand is not reloaded.  Prefer to reload conflicting operand
24727         if earlyclobber and matching operands are the same.
24729 2013-02-12  Richard Biener  <rguenther@suse.de>
24731         PR lto/56297
24732         * lto-streamer-out.c (write_symbol): Do not output symbols
24733         for hard register variables.
24735 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24737         PR target/54222
24738         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24739         (umulsidi3_insn, mulsidi3_insn): New insns.
24741 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24743         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24744         (struct tune_params): Add vec_costs field.
24745         * config/arm/arm.c (arm_builtin_vectorization_cost)
24746         (arm_add_stmt_cost): New functions.
24747         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24748         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24749         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24750         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24751         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24752         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24753         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24755 2013-02-12  Richard Biener  <rguenther@suse.de>
24757         PR lto/56295
24758         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24759         decls again if possible.
24761 2013-02-12  Richard Biener  <rguenther@suse.de>
24763         PR middle-end/56288
24764         * tree-ssa.c (verify_ssa_name): Fix check, move
24765         SSA_NAME_IN_FREE_LIST check up.
24767 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24768             Steven Bosscher   <steven@gcc.gnu.org>
24770         PR rtl-optimization/56151
24771         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24772         equal to op0 or op1, and last_insn pattern is CODE operation
24773         with MEM dest and one of the operands matches that MEM.
24775 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24777         * doc/extend.texi: Document Function Multiversioning and "default"
24778         parameter string to target attribute.
24779         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24780         target attribute parameter is "default".
24781         (ix86_compare_version_priority): Remove checks for target attribute.
24782         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24783         Remove check for target attribute equal to NULL. Add assert.
24784         (ix86_generate_version_dispatcher_body): Change error to sorry.
24786 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24787             Jack Howarth  <howarth@bromo.med.uc.edu>
24788             Patrick Marlier  <patrick.marlier@gmail.com>
24790         PR libitm/55693
24791         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24792         define ENDFILE_SPEC as TM_DESTRUCTOR.
24793         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24795 2013-02-11  Alexander Potapenko  <glider@google.com>
24796             Jack Howarth  <howarth@bromo.med.uc.edu>
24797             Jakub Jelinek  <jakub@redhat.com>
24799         PR sanitizer/55617
24800         * config/darwin.c (cdtor_record): Rename ctor_record.
24801         (sort_cdtor_records): Rename sort_ctor_records.
24802         (finalize_dtors): New routine to sort destructors by
24803         priority before use in assemble_integer.
24804         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24806 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24808         PR rtl-optimization/56275
24809         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24810         offset is non-negative and less than cmode size before
24811         calling simplify_subreg.
24813 2013-02-11  Richard Biener  <rguenther@suse.de>
24815         PR tree-optimization/56264
24816         * cfgloop.h (fix_loop_structure): Adjust prototype.
24817         * loop-init.c (fix_loop_structure): Return the number of
24818         newly discovered loops.
24819         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24820         are discovered, do a full loop-closed SSA rewrite.
24822 2013-02-11  Richard Biener  <rguenther@suse.de>
24824         PR tree-optimization/56273
24825         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24826         first VRP run.
24827         (check_array_ref): Fix missing newline in dumps.
24828         (search_for_addr_array): Likewise.
24830 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24832         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24834 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24836         PR target/56256
24837         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24839 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24841         PR rtl-optimization/56246
24842         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24843         reload pseudo.
24844         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24845         constraints are satisfied.
24847 2013-02-08  Jeff Law  <law@redhat.com>
24849         PR debug/53948
24850         * emit-rtl.c (reg_is_parm_p): New function.
24851         * regs.h (reg_is_parm_p): New prototype.
24852         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24853         callee-clobbered registers.
24855 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24857         PR target/56043
24858         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24859         If there is no implicit builtin declaration, just return NULL.
24861 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24863         * config/i386/sse.md (FMAMODEM): New mode iterator.
24864         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24865         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24867 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24869         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24870         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24871         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24873 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24875         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24876         (microblaze*-*-elf): Likewise.
24877         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24878         LINK_SPEC.
24879         * config/microblaze/microblaze-c.c: Add builtin defines for
24880         _LITTLE_ENDIAN and _BIG_ENDIAN.
24881         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24882         add to TARGET_DEFAULT flags.
24883         Expand ASM_SPEC and LINK_SPEC.
24884         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24885         * config/microblaze/microblaze.md: Update extendsidi2 and
24886         movdi_internal instructions to use low-order / high-order reg
24887         print_operands.
24888         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24889         options and inversemask / mask of LITTLE_ENDIAN.
24890         * config/microblaze/t-microblaze: Expand multilib options to
24891         include mlittle-endian (le) and update exceptions patterns.
24893 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24895         PR rtl-optimization/56195
24896         * lra-constraints.c (get_reload_reg): Don't reuse regs
24897         if they have smaller mode than requested, if they have
24898         wider mode than requested, try to return a SUBREG.
24900         PR tree-optimization/56250
24901         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24902         if type is unsigned and code isn't MULT_EXPR.
24904 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24906         PR tree-optimization/56064
24907         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24908         bits according to mode.
24909         * fixed-value.h (fixed_from_double_int)
24910         (const_fixed_from_double_int): Adjust comments.
24912 2013-02-08  Richard Biener  <rguenther@suse.de>
24914         PR lto/56231
24915         * lto-streamer.h (struct data_in): Remove current_file, current_line
24916         and current_col members.
24917         * lto-streamer-out.c (lto_output_location): Stream changed bits
24918         en-block for efficiency.
24919         * lto-streamer-in.c (clear_line_info): Remove.
24920         (lto_input_location): Cache current file, line and column
24921         globally via local statics.  Read changed bits en-block.
24922         (input_function): Do not call clear_line_info.
24923         (lto_read_body): Likewise.
24924         (lto_input_toplevel_asms): Likewise.
24926 2013-02-08  Michael Matz  <matz@suse.de>
24928         PR tree-optimization/52448
24929         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24930         (nt_call_phase): New static.
24931         (add_or_mark_expr): Only mark accesses with newer phase than any
24932         call seen.
24933         (nonfreeing_call_p): New.
24934         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24935         (nt_fini_block): Keep blocks marked as visited.
24936         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24938 2013-02-08  Richard Biener  <rguenther@suse.de>
24940         * ira.c (ira): Free broken dominator information.
24942 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24944         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
24946 2013-02-08  Marek Polacek  <polacek@redhat.com>
24948         * cfgloop.c (verify_loop_structure): Add more checking of headers.
24950 2013-02-08  Richard Biener  <rguenther@suse.de>
24952         PR middle-end/56181
24953         * cfgloop.h (flow_loops_find): Adjust.
24954         (bb_loop_header_p): Declare.
24955         * cfgloop.c (bb_loop_header_p): New function split out from ...
24956         (flow_loops_find): ... here.  Adjust function signature,
24957         support incremental loop structure update.
24958         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
24959         * cfgloopmanip.c (fix_loop_structure): Move ...
24960         * loop-init.c (fix_loop_structure): ... here.
24961         (apply_loop_flags): Split out from ...
24962         (loop_optimizer_init): ... here.
24963         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
24964         in incremental mode, only remove dead loops here.
24966 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24968         PR target/54222
24969         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
24970         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
24971         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
24972         (*round<mode>3.libgcc): New insns for fixed-modes.
24973         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
24974         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
24975         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
24976         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
24977         implementations.  Define to __builtin_avr_absFX,
24978         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
24979         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
24980         __builtin_avr_countlsFX, respectively.
24981         * config/avr/avr-c.c (target.h): Include it.
24982         (enum avr_builtin_id): New enum.
24983         (avr_resolve_overloaded_builtin): New static function.
24984         (avr_register_target_pragmas): Use it to set
24985         targetm.resolve_overloaded_builtin.
24986         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
24987         tree nodes used by DEF_BUILTIN.
24988         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
24989         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
24990         <AVR_BUILTIN_xxBITS>: Same.
24992 2013-02-08  Richard Biener  <rguenther@suse.de>
24994         * cfgloop.c (verify_loop_structure): Properly handle
24995         a loop exiting to another loop header.
24996         * ira-int.h (ira_loops): Remove.
24997         * ira.c (ira_loops): Remove.
24998         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
24999         (do_reload): Use loop_optimizer_finalize.
25000         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25001         number_of_loops to access the loop tree.
25002         (more_one_region_p): Likewise.
25003         (finish_loop_tree_nodes): Likewise.
25004         (rebuild_regno_allocno_maps): Likewise.
25005         (mark_loops_for_removal): Likewise.
25006         (mark_all_loops_for_removal): Likewise.
25007         (remove_unnecessary_regions): Likewise.
25008         (ira_build): Likewise.
25009         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25011 2013-02-08  Richard Biener  <rguenther@suse.de>
25013         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25014         * ipa-pure-const.c (analyze_function): Avoid calling
25015         mark_irreducible_loops twice.
25016         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25018 2013-02-07  David S. Miller  <davem@davemloft.net>
25020         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25021         on 'reg'.
25022         * var-tracking.c (vt_add_function_parameter): Test the presence of
25023         HAVE_window_save properly and do not remap argument registers when
25024         we have a leaf function.
25026 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25028         PR bootstrap/56227
25029         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25030         instead of "ll".
25031         * config/i386/i386.c (ix86_print_operand): Ditto.
25033 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25035         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25037 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25039         PR rtl-optimization/56225
25040         * lra-constraints.c (process_alt_operands): Check that reload hard
25041         reg can hold value for strict_low_part.
25043 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25045         PR debug/56154
25046         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25047         dwarf2out_end_function.
25048         (in_first_function_p, maybe_at_text_label_p,
25049         first_loclabel_num_not_at_text_label): New variables.
25050         (dwarf2out_var_location): In the first function find out
25051         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25052         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25053         functions.
25055 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25057         PR rtl-optimization/56178
25058         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25059         SUBREG of a register.  Tidy up related block of code.
25060         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25061         note if the source is a register or a SUBREG of a register.
25063 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25065         PR target/56228
25066         * config/rs6000/rs6000.md (ptrm): New mode attr.
25067         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25068         call_value_indirect_aix<pttrsize>,
25069         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25070         m in constraints.
25072 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25074         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25075         if -bnortl. Convert to strcmp and strncmp.
25077 2013-02-07  Alan Modra  <amodra@gmail.com>
25079         PR target/54009
25080         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25081         addresses won't wrap when offsetting.
25082         (rs6000_secondary_reload): Provide secondary reloads needed for
25083         wrapping LO_SUM addresses.
25085 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25087         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25088         MACH, just __MACH__.
25090 2013-02-06  Richard Biener  <rguenther@suse.de>
25092         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25093         instead of calling fix_loop_structure.
25095 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25097         PR middle-end/56217
25098         * omp-low.c (use_pointer_for_field): Return false if
25099         lower_send_shared_vars doesn't generate any copy-out code.
25101 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25103         PR rtl-optimization/56131
25104         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25105         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25106         of the label is NULL.  Add comment.
25108 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25110         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25112         PR sanitizer/55374
25113         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25114         (STATIC_LIBTSAN_LIBS): Likewise.
25115         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25116         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25117         is defined, don't add anything else beyond that.
25118         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25119         (LINK_COMMAND_SPEC): Use them.
25121         PR tree-optimization/56205
25122         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25123         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25124         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25126 2013-02-05  Richard Biener  <rguenther@suse.de>
25128         PR tree-optimization/53342
25129         PR tree-optimization/53185
25130         * tree-vectorizer.h (vect_check_strided_load): Remove.
25131         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25132         not disallow peeling for vectorized strided loads.
25133         (vect_check_strided_load): Make static and simplify.
25134         (vect_analyze_data_refs): Adjust.
25135         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25136         correctly when vectorizing strided loads.
25138 2013-02-05  Richard Biener  <rguenther@suse.de>
25140         * doc/install.texi: Refer to ISL, not PPL.
25142 2013-02-05  Jan Hubicka  <jh@suse.cz>
25144         PR tree-optimization/55789
25145         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25147 2013-02-05  Jan Hubicka  <jh@suse.cz>
25149         PR tree-optimization/55789
25150         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25151         the dead call anyway.
25153 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25155         PR sanitizer/55374
25156         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25158 2013-02-04  Alexander Potapenko  <glider@google.com>
25159             Jack Howarth  <howarth@bromo.med.uc.edu>
25160             Jakub Jelinek  <jakub@redhat.com>
25162         PR sanitizer/55617
25163         * config/darwin.c (sort_ctor_records): Stabilized qsort
25164         on constructor priority by using original position.
25165         (finalize_ctors): New routine to sort constructors by
25166         priority before use in assemble_integer.
25167         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25169 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25171         PR libstdc++/54314
25172         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25173         about visibility on artificial decls.
25174         * config/sol2.c (solaris_assemble_visibility): Likewise.
25176 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25178         PR target/56186
25179         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25180         argument and improve checking of return-argument types for 16-byte
25181         modes.
25182         (ix86_function_value_1): Add additional valtype argument on call
25183         of function_value_64.
25184         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25185         handling infunction_value_64 function.
25187 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25189         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25191 2013-02-04  Richard Biener  <rguenther@suse.de>
25193         PR tree-optimization/56188
25194         * tree-ssa-structalias.c (label_visit): Consider case with
25195         initially non-empty points-to set.
25196         (perform_var_substitution): Dump node mapping and clean up.
25198 2013-02-04  Richard Guenther  <rguenther@suse.de>
25200         PR lto/56168
25201         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25202         node prevail as last resort.
25203         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25204         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25206 2013-02-04  Richard Biener  <rguenther@suse.de>
25208         PR tree-optimization/56113
25209         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25210         Merge into ...
25211         (equiv_class_lookup_or_add): ... this.
25212         (label_visit): Adjust and fix error in previous patch.
25213         (perform_var_substitution): Adjust.
25215 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25217         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25218         * config/sh/sh4-300.md: Likewise.
25219         * config/sh/sh4a.md: Likewise.
25220         * config/sh/constraints.md: Likewise.
25221         * config/sh/sh.md: Likewise.
25222         * config/sh/netbsd-elf.h: Likewise.
25223         * config/sh/predicates.md: Likewise.
25224         * config/sh/sh-protos.h: Likewise.
25225         * config/sh/ushmedia.h: Likewise.
25226         * config/sh/linux.h: Likewise.
25227         * config/sh/sh.c: Likewise.
25228         * config/sh/superh.h: Likewise.
25229         * config/sh/elf.h: Likewise.
25230         * config/sh/sh4.md: Likewise.
25231         * config/sh/sh.h: Likewise.
25233 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25235         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25236         constraint to match_test floating_point_store_memory_operand().
25237         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25238         (base14_operand): New.
25239         (floating_point_store_memory_operand): New.
25240         (integer_store_memory_operand): Revise to use base14_operand and
25241         reg_plus_base_memory_operand.
25242         (move_dest_operand): Allow symbolic_memory_operands.
25243         (symbolic_memory_operand): Check for LO_SOM.
25244         (symbolic_operand): Change default case to break.
25245         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25246         CONST_DOUBLE values to be reloaded by putting them into memory when
25247         the destination is a floating point register.
25248         (movdf): Remove code to handle CONST_DOUBLE.
25249         (movsf): Likewise.
25250         (reload_indf_r1): New.
25251         (reload_insf_r1): New.
25252         Consistently use "Q" and "T" constraints with integer and floating
25253         point move instructions, respectively.
25254         (movdi): Remove FAIL.
25255         Change predicate for source operand unamed DImode move from
25256         general_operand to move_src_operand.
25257         (umulsidi3): Change predicate for destination operand to
25258         register_operand.
25259         Likewise for similar unamed patterns.
25260         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25261         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25262         (hppa_legitimize_address): Simplify mask calculation.
25263         (pa_emit_move_sequence): Revised handling of secondary reloads from
25264         REG+D addresses for floating point loads and stores.  Directly handle
25265         loading CONST0_RTX (mode) to a floating point register.
25266         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25267         to floating point registers.  Don't restrict secondary reloads to
25268         floating point registers to integer modes.  Revise some comments and
25269         cleanup some code.
25270         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25271         (pa_legitimate_address_p): New.
25272         (pa_legitimize_reload_address): New.
25273         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25274         (STRICT_REG_OK_FOR_BASE_P): New.
25275         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25276         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25278 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25279             Andrew Dixie  <andrewd@gentrack.com>
25281         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25282         flag set.
25284 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25286         * expmed.c (extract_bit_field_1): Pass the full width of the
25287         structure to get_best_reg_extraction_insn.
25289 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25291         PR target/54601
25292         * configure.ac (use_cxa_atexit): Add AIX.
25293         * configure: Regenerate.
25295         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25297 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25299         PR debug/54793
25300         * final.c (need_profile_function): New variable.
25301         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25302         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25303         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25304         notes, targetm.asm_out.function_prologue doesn't emit anything,
25305         HAVE_prologue and profiler should be emitted before prologue,
25306         set need_profile_function instead of emitting it.
25307         (final_scan_insn): If need_profile_function, emit
25308         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25309         NOTE_INSN_FUNCTION_BEG note.
25311 2013-02-01  Richard Henderson  <rth@redhat.com>
25313         * config/rs6000/rs6000.md (smulditi3): New.
25314         (umulditi3): New.
25316         * config/alpha/alpha.md (umulditi3): New.
25318 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25320         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25321         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25323 2013-02-01  Richard Biener  <rguenther@suse.de>
25325         PR tree-optimization/56113
25326         * tree-ssa-structalias.c (label_visit): Reduce work for
25327         single-predecessor nodes.
25329 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25331         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25332         range isn't testing for zero.
25334 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25336         PR middle-end/56113
25337         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25339 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25340             Nick Clifton  <nickc@redhat.com>
25342         * config/v850/constraints.md (Q): Define as a memory constraint.
25343         * config/v850/predicates.md (label_ref_operand): New predicate.
25344         (e3v5_shift_operand): New predicate.
25345         (ior_operator): New predicate.
25346         * config/v850/t-v850: Add e3v5 multilib.
25347         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25348         (v850_gen_movdi): Prototype.
25349         * config/v850/v850.c: Add support for e3v5 architecture.
25350         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25351         TARGET_V850E_UP.
25352         (construct_save_jarl): Add e3v5 long JARL support.
25353         (v850_adjust_insn_length): New function.  Adjust length of call
25354         insns when using e3v5 instructions.
25355         (v850_gen_movdi): New function: Generate instructions to move a
25356         DImode value.
25357         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25358         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25359         (TARGET_USE_FPU): Enable for e3v5.
25360         (CONST_OK_FOR_W): New macro.
25361         (ADJUST_INSN_LENGTH): Define.
25362         * config/v850/v850.md (UNSPEC_LOOP): Define.
25363         (attr cpu): Add v850e3v5.
25364         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25365         (movdi): New pattern.
25366         (movdi_internal): New pattern.
25367         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25368         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25369         (cstoresf4): Likewise.
25370         (cstoredf4): Likewise.
25371         (insv): New pattern.
25372         (rotlso3_a): New pattern.
25373         (rotlsi3_b): New pattern
25374         (rotlsi3_v850e3v5): New pattern.
25375         (doloop_begin): New pattern.
25376         (fix_loop_counter): New pattern.
25377         (doloop_end): New pattern.
25378         (branch_normal): Add e3v5 long branch support.
25379         (branch_invert): Likewise.
25380         (branch_z_normal): Likewise.
25381         (branch_z_invert): Likewise.
25382         (branch_nz_normal): Likewise.
25383         (branch_nz_invert): Likewise.
25384         (call_internal_short): Add e3v5 register-indirect JARL support.
25385         (call_internal_long): Likewise.
25386         (call_value_internal_short): Likewise.
25387         (call_value_internal_long): Likewise.
25388         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25389         (mloop): New option.
25390         * config.gcc: Add support for configuring v840e3v5 target.
25391         * doc/invoke.texi: Document new v850 specific command line options.
25393 2013-01-31  Paul Koning  <ni1d@arrl.net>
25395         PR debug/55059
25396         PR debug/54508
25397         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25398         children if parent is a class.
25399         (prune_unused_types_prune): Don't add DW_AT_declaration.
25401 2013-01-31  Richard Biener  <rguenther@suse.de>
25403         PR tree-optimization/56157
25404         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25405         match up operand with SLP child.
25407 2013-01-31  Jason Merrill  <jason@redhat.com>
25409         PR debug/54410
25410         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25411         parameters the first time.
25412         (gen_scheduled_generic_parms_dies): Check completeness here.
25414 2013-01-31  Richard Biener  <rguenther@suse.de>
25416         PR middle-end/53073
25417         * common.opt (faggressive-loop-optimizations): New flag,
25418         enabled by default.
25419         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25420         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25421         infer_loop_bounds_from_undefined by it.
25423 2013-01-31  Richard Biener  <rguenther@suse.de>
25425         PR tree-optimization/56150
25426         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25427         visit virtual operands.
25428         (find_uses_to_rename_bb): Likewise.
25430 2013-01-31  Richard Biener  <rguenther@suse.de>
25432         PR tree-optimization/56150
25433         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25434         mixed store non-store stmts.
25436 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25438         PR sanitizer/55374
25439         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25440         LIBASAN_EARLY_SPEC is defined.
25441         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25442         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25443         before %o.
25444         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25446         PR c++/55742
25447         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25448         invalid args instead of ICEing on it.
25449         (ix86_valid_target_attribute_tree): Return error_mark_node if
25450         ix86_valid_target_attribute_inner_p failed.
25451         (ix86_valid_target_attribute_p): Return false only if
25452         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25453         target("default") attribute.
25454         (sorted_attr_string): Change argument from const char * to tree,
25455         merge in all target attribute arguments rather than just one.
25456         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25457         instead of free.  Avoid using strcat.
25458         (ix86_mangle_function_version_assembler_name): Mangle
25459         target("default") as if no target attribute is present.  Adjust
25460         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25461         instead of xmalloc and XDELETEVEC instead of free.
25462         (ix86_function_versions): Don't return true if one of the decls
25463         doesn't have target attribute.  If they don't and one of the decls
25464         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25465         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25466         (ix86_supports_function_versions): Remove.
25467         (make_name): Fix up formatting.
25468         (make_dispatcher_decl): Remove resolver_name and its initialization.
25469         Avoid leaking memory.
25470         (is_function_default_version): Return true if there is
25471         target("default") attribute rather than no target attribute at all.
25472         (make_resolver_func): Avoid leaking memory.
25473         (ix86_generate_version_dispatcher_body): Likewise.
25474         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25475         * target.def (supports_function_versions): Remove.
25476         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25477         * doc/tm.texi: Regenerated.
25479 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25481         PR rtl-optimization/56144
25482         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25483         for values with side effects.
25485 2013-01-30  Richard Biener  <rguenther@suse.de>
25487         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25488         (sparseset_pop): Likewise.
25489         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25490         to be able to use quick_push in the worker loop.
25492 2013-01-30  Marek Polacek  <polacek@redhat.com>
25494         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25496 2013-01-30  Richard Biener  <rguenther@suse.de>
25498         PR lto/56147
25499         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25501 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25503         PR tree-optimization/56064
25504         * fixed-value.c (fixed_from_double_int): New function.
25505         * fixed-value.h (fixed_from_double_int): New prototype.
25506         (const_fixed_from_double_int): New static inline function.
25507         * fold-const.c (native_interpret_fixed): New static function.
25508         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25509         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25510         (native_encode_fixed): New static function.
25511         (native_encode_expr) <FIXED_CST>: Use it.
25512         (native_interpret_int): Move double_int worker code to...
25513         * double-int.c (double_int::from_buffer): ...this new static method.
25514         * double-int.h (double_int::from_buffer): Prototype it.
25516 2013-01-30  Richard Biener  <rguenther@suse.de>
25518         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25519         New pointer-map and obstack.
25520         (init_alias_vars): Allocate pointer-map and obstack.
25521         (delete_points_to_sets): Free them.
25522         (find_what_var_points_to): Cache result.
25523         (find_what_p_points_to): Adjust for changed interface of
25524         find_what_var_points_to.
25525         (compute_points_to_sets): Likewise.
25526         (ipa_pta_execute): Likewise.
25528 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25530         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25531         * configure: Regenerate.
25532         * config.in: Regenerate.
25533         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25534         #nobits/#progbits if supported.
25536 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25538         PR target/56121
25539         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25540         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25541         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25543 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25545         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25546         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25548 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25550         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25551         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25553 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25555         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25556         declaration.
25557         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25558         * config/arm/cortex-a7.md: New bypasses using
25559         arm_mac_accumulator_is_result.
25561 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25563         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25564         (cortex_a7_neon_mla): Likewise.
25565         (cortex_a7_fpfmad): New reservation.
25566         (cortex_a7_fpmacs): Use ffmas and update required units.
25567         (cortex_a7_fpmuld): Update required units and latency.
25568         (cortex_a7_fpmacd): Likewise.
25569         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25570         (cortex_a7_neon). Likewise.
25571         (bypass) Update participating units.
25573 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25575         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25576         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25577         from fmac to ffma.
25578         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25579         (vfp_fmul): Use ffmad.
25580         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25581         (cortex_r4_fmacd): Use ffmad.
25582         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25583         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25584         (cortex_a9_fmacd): Use ffmad.
25585         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25586         (cortex_a8_vfp_macd): Use ffmad.
25587         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25588         (cortex_a5_fpmacd): Use ffmad.
25589         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25590         (cortex_a15_vfp_macd): Use ffmad.
25591         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25593 2013-01-29  Jason Merrill  <jason@redhat.com>
25595         PR libstdc++/54314
25596         * varasm.c (default_assemble_visibility): Don't warn about
25597         visibility on artificial decls.
25599 2013-01-29  Richard Biener  <rguenther@suse.de>
25601         PR tree-optimization/56113
25602         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25603         the bitmap leader.
25604         (label_visit): Free duplicate bitmaps and record the leader instead.
25605         (perform_var_substitution): Adjust.
25607 2013-01-29  Richard Biener  <rguenther@suse.de>
25609         PR tree-optimization/55270
25610         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25611         the CFG, schedule loops for fixup.
25613 2013-01-29  Nick Clifton  <nickc@redhat.com>
25615         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25616         SP_REG.
25618 2013-01-28  Leif Ekblad  <leif@rdos.net>
25620         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25621         * config/i386/i386.h (TARGET_RDOS): New macro.
25622         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25623         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25624         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25625         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25626         DEFAULT_LARGE_SECTION_THRESHOLD.
25627         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25628         * config/i386/rdos.h: New file.
25629         * config/i386/rdos64.h: New file.
25631 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25633         PR other/54814
25634         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25635         TEST_HARD_REG_BIT.
25637 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25639         PR rtl-optimization/56117
25640         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25641         call cselib_lookup_from_insn on the MEM before calling
25642         add_insn_mem_dependence.
25644 2013-01-28  Richard Biener  <rguenther@suse.de>
25646         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25647         to a stmt that didn't have one.
25648         (copy_phis_for_bb): Likewise for PHI arguments.
25649         (copy_debug_stmt): Likewise for debug stmts.
25651 2013-01-28  Richard Biener  <rguenther@suse.de>
25653         PR tree-optimization/56034
25654         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25655         (partition_builtin_p): Adjust.
25656         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25657         it is the last partition.
25658         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25659         up the vertex for the definition.
25660         (classify_partition): Classify whether a partition is a
25661         PKIND_REDUCTION, thus has uses outside of the loop.
25662         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25663         Merge all PKIND_REDUCTION partitions into the last partition.
25664         (tree_loop_distribution): Seed partitions from reductions as well.
25666 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25668         PR tree-optimization/56125
25669         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25670         pow(x,c) into sqrt(x) * powi(x, n/2) or
25671         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25672         optimizing for size.
25673         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25674         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25675         integer.
25677         PR tree-optimization/56094
25678         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25679         to UNKNOWN_LOCATION while gimplifying expr.
25681 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25683         PR target/56114
25684         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25685         operand 0 in movabs insn template for -masm=intel asm alternative.
25686         (*movabs<mode>_2): Ditto for operand 1.
25688 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25690         PR target/54663
25691         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25692         of microblaze-c.o
25694 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25696         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25697         tm_file.
25699 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25701         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25702         Undef to avoid warning.
25704 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25706         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25707         * configure: Regenerate.
25709 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25711         PR tree-optimization/56098
25712         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25713         for stmts with volatile ops.
25714         (cond_store_replacement): Don't optimize if assign has volatile ops.
25715         (cond_if_else_store_replacement_1): Don't optimize if either
25716         then_assign or else_assign have volatile ops.
25717         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25718         volatile ops.
25720 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25722         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25724 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25726         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25727         missing ':' in asm example.
25729 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25731         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25732         entries into lane and laneq entries.
25733         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25734         Remove AdvSIMD scalar modes.
25735         (aarch64_sq<r>dmulh_laneq<mode>): New.
25736         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25737         modes.
25738         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25739         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25740         * config/aarch64/iterators.md (VCOND): New.
25741         (VCONQ): New.
25743 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25745         PR target/54222
25746         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25747         Add NULL LIBNAME argument to existing definitions.
25748         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25749         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25750         * config/avr/avr.c (DEF_BUILTIN): Same.
25751         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25752         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25753         implementation is available (DECL_ASSEMBLER_NAME is set).
25754         (avr_fold_absfx): New static function.
25755         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25756         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25757         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25758         AVR_BUILTIN_ABSLLK.
25759         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25760         (abshk, absk, abslk, absllk): Provide as static inline functions.
25762 2013-01-25  Marek Polacek  <polacek@redhat.com>
25764         PR tree-optimization/56035
25765         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25767 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25769         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25770         (*movtf_internal_rex64): Add (!o,C) alternative
25771         (*movxf_internal_rex64): Ditto.
25772         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25774 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25776         * doc/invoke.texi: fix typo.
25777         * doc/objc.texi: fix typo.
25779 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25781         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25782         for the first two alternatives.
25784 2013-01-24  Diego Novillo  <dnovillo@google.com>
25786         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25787         (ggc-zone.o): Remove.
25788         * configure.ac: Remove option --with-gc.
25789         * configure: Re-generate.
25790         * doc/install.texi: Remove documentation for --with-gc.
25791         * gengtype.c (write_enum_defn): Remove.  Update all users.
25792         (write_Types_process_field): Remove generation of gt_e_* argument.
25793         (output_type_enum): Remove.  Update all users.
25794         (write_enum_defn): Remove.  Update all users.
25795         (enum alloc_zone): Remove.  Update all users.
25796         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25797         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25798         Update all callers.
25799         (struct ptr_data): Remove field TYPE.  Update all users.
25800         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25801         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25802         Update all users.
25803         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25804         (struct alloc_zone): Remove.
25805         (ggc_internal_alloc_zone_stat): Remove.
25806         (ggc_internal_cleared_alloc_zone_stat): Remove.
25807         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25808         (ggc_pch_count_object): Remove last argument.  Update all users.
25809         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25810         (struct alloc_zone): Remove.
25811         * ggc-zone.c: Remove.
25812         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25813         (struct alloc_zone): Remove.
25814         (ggc_alloc_typed_stat): Remove.
25815         (ggc_alloc_typed): Remove.
25816         (ggc_splay_alloc): Remove first argument.
25817         (rtl_zone): Remove.  Update all users.
25818         (tree_zone): Remove.  Update all users.
25819         (tree_id_zone): Remove.  Update all users.
25820         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25821         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25822         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25823         * tree-ssanames.c: Remove references to zone allocator in comments.
25825 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25827         * config/avr/avr.c (avr_out_fract): Make register numbers that
25828         might be outside of source operand signed.
25830 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25832         * config/i386/constraints.md (Yf): New constraint.
25833         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25834         of f constraint to conditionaly disable x87 register preferences.
25835         (*movdf_internal): Ditto.
25836         (*movsf_internal): Ditto.
25838 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25840         PR inline-asm/55934
25841         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25842         that have operands with impossible constraints.
25843         Add a FIXME for a speed-up opportunity.
25844         * lra-constraints.c (process_alt_operands): Verify that a class
25845         selected from constraints on asms is valid for the operand mode.
25846         (curr_insn_transform): Remove incorrect comment.
25848 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25850         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25851         TOC operand is a valid symbol ref in the constant pool.
25853 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25855         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25857 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25859         PR target/54222
25860         * config/avr/stdfix.h: New file.
25861         * t-avr (stdfix-gcc.h): New rule to build it.
25862         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25864 2013-01-23  Kostya Serebryany  <kcc@google.com>
25866         * config/darwin.h: remove dependency on
25867         CoreFoundation (asan on Mac OS).
25869 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25871         PR target/49069
25872         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25873         instead of cmpdi_operand for first comparison operand.
25874         Don't assert that comparison operands aren't both constants.
25876 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25878         * doc/install.texi (Downloading the Source): Update references to
25879         downloading separate components.
25881 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25883         * doc/extend.texi (__int128): Improve grammar.
25885 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25887         PR target/56028
25888         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25889         alternative to (o,r).
25890         (*movdi_internal_rex64): Remove (!o,n) alternative.
25891         (DImode immediate->memory splitter): Remove.
25892         (DImode immediate->memory peephole2): Remove.
25893         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25894         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25895         alternative to (!o,*r).
25896         (*movtf_internal_sse): New pattern.
25897         (*movxf_internal_rex64): New pattern.
25898         (*movxf_internal): Disable for TARGET_64BIT.
25899         (*movdf_internal_rex64): Remove (!o,F) alternative.
25901 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25903         PR middle-end/56074
25904         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25905         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25906         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25907         stmt locations where LOCATION_LOCUS of the stmt location is
25908         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25910         PR target/55686
25911         * config/i386/i386.md (UNSPEC_STOS): New.
25912         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25913         *strsetqi_1): Add UNSPEC_STOS.
25915 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25917         PR c++/56067
25918         * doc/invoke.texi: Remove left over -Wsynth example.
25920 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25922         PR tree-optimization/56051
25923         * fold-const.c (fold_binary_loc): Don't fold
25924         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25925         a narrowing conversion, or widening conversion from signed
25926         to unsigned.
25928 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25930         PR rtl-optimization/56023
25931         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25932         dependent on debug instruction.
25934 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25936         PR middle-end/56022
25937         * function.c (allocate_struct_function): Call
25938         invoke_set_current_function_hook earlier.
25940 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25942         * reload1.c (init_reload): Only initialize reload_obstack
25943         during the first call.
25945 2013-01-21  Marek Polacek  <polacek@redhat.com>
25947         * cfgloop.c (verify_loop_structure): Fix up grammar.
25949 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
25951         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
25952         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
25954 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25956         PR target/56058
25957         * config/arm/marvell-pj4.md: Update copyright year.
25958         Fix up use of alu to alu_reg and simple_alu_imm.
25960 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25962         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
25964 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
25966         PR target/55433
25967         * lra-constraints.c (curr_insn_transform): Don't reuse original
25968         insn for secondary memory move when memory mode should be different.
25970 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25972         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
25973         atomic_storedi_1): New patterns.
25975 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
25977         btver2 pipeline descriptions.
25978         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
25979         descriptions.
25980         * config/i386/i386.md (btver2_decode): New type attributes.
25981         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
25982         type attributes.
25983         * config/i386/btver2.md: New file describing btver2 pipelines.
25985 2013-01-19  Andrew Pinski  <apinski@cavium.com>
25987         PR tree-optimization/52631
25988         * tree-ssa-sccvn (visit_use): Before looking up the original
25989         statement, try looking up the simplified expression.
25991 2013-01-19  Anthony Green  <green@moxielogic.com>
25993         * config/moxie/moxie.c (moxie_expand_prologue): Set
25994         current_function_static_stack_size.
25996 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
25998         PR tree-optimization/56029
25999         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26000         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26002 2013-01-18  Sharad Singhai  <singhai@google.com>
26004         PR tree-optimization/55995
26005         * dumpfile.c (dump_loc): Print location only if available.
26006         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26008 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26010         PR target/55433
26011         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26012         secondary memory move.
26013         (inherit_reload_reg): Use rclass instead of cl for
26014         check_secondary_memory_needed_p.
26016 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26018         PR middle-end/56015
26019         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26020         the case where writing real complex part of target modifies op1.
26022 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26024         * config/aarch64/aarch64-simd.md
26025         (aarch64_vcond_internal<mode>): Handle unordered cases.
26026         * config/aarch64/iterators.md (v_cmp_result): New.
26028 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26029             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26031         * config/arm/marvell-pj4.md: New file.
26032         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26033         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26034         (generic_vfp): Likewise.
26035         * config/arm/arm-cores.def: Add marvell-pj4.
26036         * config/arm/arm-tune.md: Regenerate.
26037         * config/arm/arm-tables.opt: Regenerate.
26038         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26039         * doc/invoke.texi: Document marvell-pj4.
26041 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26043         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26045 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26047         PR debug/54114
26048         PR debug/54402
26049         PR debug/49888
26050         * var-tracking.c (negative_power_of_two_p): New.
26051         (global_get_addr_cache, local_get_addr_cache): New.
26052         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26053         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26054         heading comment.
26055         (vt_stack_offset_p): Remove.
26056         (vt_canon_true_dep): Always canonicalize loc's address.
26057         (clobber_overlapping_mems): Make sure we have a MEM.
26058         (local_get_addr_clear_given_value): New.
26059         (val_reset): Clear local cached entries.
26060         (compute_bb_dataflow): Create and release the local cache.
26061         Disable duplicate MEMs clobbering.
26062         (emit_notes_in_bb): Clobber MEMs likewise.
26063         (vt_emit_notes): Create and release the local cache.
26064         (vt_initialize, vt_finalize): Create and release the global
26065         cache, respectively.
26066         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26068 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26070         PR libmudflap/53359
26071         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26072         not found in the symtab.
26074 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26076         PR debug/56006
26077         PR rtl-optimization/55547
26078         PR rtl-optimization/53827
26079         PR debug/53671
26080         PR debug/49888
26081         * alias.c (offset_overlap_p): New, factored out of...
26082         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26083         the conservative special case for symbolic constants.  Don't
26084         adjust zero sizes on alignment.
26086 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26088         PR rtl-optimization/52573
26089         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26090         REG_UNUSED for the same register.
26092 2013-01-17  Richard Biener  <rguenther@suse.de>
26093             Marek Polacek  <polacek@redhat.com>
26095         PR rtl-optimization/55833
26096         * loop-unswitch.c (unswitch_loops): Move loop verification...
26097         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26098         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26099         Set it to true when we're removing a loop from hierarchy tree in
26100         an irreducible region.
26101         (fix_bb_placements): Adjust caller.
26102         (fix_loop_placements): Likewise.
26104 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26106         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26107         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26108         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26109         Remove ID.  Adjust comments.
26110         * config/avr/avr-c.c (avr_builtin_name): Remove.
26111         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26112         * config/avr/avr.c (avr_tolower): New static function.
26113         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26114         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26115         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26116         default expansion.
26118 2013-01-17  Jan Hubicka  <jh@suse.cz>
26120         PR tree-optimization/55273
26121         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26123 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26125         PR target/55981
26126         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26127         store through atomic_store<mode>_1.
26128         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26130 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26132         PR tree-optimizations/55264
26133         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26134         for virtual methods.
26135         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26136         virtual methods before inlining is over.
26137         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26138         virtual functions.
26139         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26140         non-virtual.
26142 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26144         PR rtl-optimization/56005
26145         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26146         pending reads for prefetch.
26148 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26150         * config/aarch64/aarch64.md
26151         (*cstoresi_neg_uxtw): New pattern.
26152         (*cmovsi_insn_uxtw): New pattern.
26153         (*<optab>si3_uxtw): New pattern.
26154         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26155         (*<optab>si3_insn_uxtw): New pattern.
26156         (*bswapsi2_uxtw): New pattern.
26158 2013-01-16  Richard Biener  <rguenther@suse.de>
26160         * tree-inline.c (tree_function_versioning): Remove set but
26161         never used variable.
26163 2013-01-16  Richard Biener  <rguenther@suse.de>
26165         PR tree-optimization/55964
26166         * tree-flow.h (rename_variables_in_loop): Remove.
26167         (rename_variables_in_bb): Likewise.
26168         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26169         (copy_loop_before): Adjust and delete update-ssa status.
26170         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26171         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26172         (rename_variables_in_loop): Remove.
26173         (slpeel_update_phis_for_duplicate_loop): Likewise.
26174         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26175         use available cfg machinery instead of duplicating it.
26176         Update PHI nodes and perform poor-mans SSA update here.
26177         (slpeel_tree_peel_loop_to_edge): Adjust.
26179 2013-01-16  Richard Biener  <rguenther@suse.de>
26181         PR tree-optimization/54767
26182         PR tree-optimization/53465
26183         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26184         (vrp_visit_phi_node): For PHI arguments coming via backedges
26185         drop all symbolical range information.
26186         (execute_vrp): Compute backedges.
26188 2013-01-16  Richard Biener  <rguenther@suse.de>
26190         * doc/install.texi: Update CLooG and ISL requirements to
26191         0.18.0 and 0.11.1.
26193 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26195         PR target/55301
26196         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26197         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26198         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26200 2013-01-16  DJ Delorie  <dj@redhat.com>
26202         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26203         (UNSPECV_SP_SWITCH_E): New.
26204         (sp_switch_1): Change to an unspec.
26205         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26206         replace $r15.
26208 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26210         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26211         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26212         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26213         (expand_mem_thread_fence): Ditto.
26214         (expand_mem_signal_fence): Ditto.
26215         (expand_atomic_load): Ditto.
26216         (expand_atomic_store): Ditto.
26218 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26220         PR rtl-optimization/55547
26221         PR rtl-optimization/53827
26222         PR debug/53671
26223         PR debug/49888
26224         * alias.c (memrefs_conflict_p): Set sizes to negative after
26225         AND adjustments.
26227 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26229         PR target/55940
26230         * function.c (thread_prologue_and_epilogue_insns): Always
26231         add crtl->drap_reg to set_up_by_prologue.set, even if
26232         stack_realign_drap is false.
26234 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26236         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26237         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26238         *call): Fix indention.
26240 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26242         PR target/55876
26243         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26244         Update comment.
26246 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26248         PR rtl-optimization/55153
26249         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26251 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26253         PR tree-optimization/55920
26254         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26255         accesses as grp_to_be_debug_replaced.
26257 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26259         PR tree-optimization/55920
26260         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26261         there is non-useless type conversion needed from debug rhs to lhs,
26262         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26264 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26265             Mikael Pettersson  <mikpe@it.uu.se>
26267         PR target/43961
26268         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26269         Thumb.
26270         (ASM_OUTPUT_CASE_LABEL): Remove.
26271         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26272         * final.c (shorten_branches): Update alignment of labels before
26273         jump tables if CASE_VECTOR_SHORTEN_MODE.
26275 2013-01-15  Richard Biener  <rguenther@suse.de>
26277         PR bootstrap/55961
26278         * system.h: Do not include gmp.h for building host tools.
26280 2013-01-15  Richard Biener  <rguenther@suse.de>
26282         PR middle-end/55882
26283         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26284         account for bitpos when computing alignment.
26286 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26288         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26289         (ix86_target_macros_internal): Likewise.
26291         * config/i386/i386.c (m_CORE2I7): Removed.
26292         (m_CORE_HASWELL): New macro.
26293         (m_CORE_ALL): Likewise.
26294         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26295         (initial_ix86_arch_features): Likewise.
26296         (processor_target_table): Initializations for Core avx2.
26297         (cpu_names): New names "core-avx2".
26298         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26299         PROCESSOR_CORE_HASWELL.
26300         (ix86_issue_rate): New case.
26301         (ia32_multipass_dfa_lookahead): Likewise.
26302         (ix86_sched_init_global): Likewise.
26304         * config/i386/i386.h (TARGET_HASWELL): New macro.
26305         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26306         (processor_type): New PROCESSOR_HASWELL.
26308 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26310         PR tree-optimization/55955
26311         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26312         *SHIFT_EXPR and *ROTATE_EXPR codes.
26314         PR tree-optimization/48766
26315         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26316         -ftrapv disable -fwrapv.
26318 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26320         PR target/55974
26321         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26322         etc. to 1 and not to __flash.
26323         Use LL suffix for __INT24_MAX__ with -mint8.
26324         Use ULL suffix for __UINT24_MAX__ with -mint8.
26326 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26328         * config/avr/avr-arch.h
26329         (struct base_arch_s): Use typedef avr_arch_t instead.
26330         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26331         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26332         * config/avr/avr.c: Same.
26333         * config/avr/avr-devices.c: Same.
26334         * config/avr/driver-avr.c: Same.
26335         * config/avr/gen-avr-mmcu-texi.c: Same.
26336         * config/avr/avr-mcus.def: Adjust comment.
26338 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26340         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26341         * config/aarch64/iterators.md (VALLDI): New.
26343 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26344             Andi Kleen  <ak@linux.intel.com>
26346         PR target/55948
26347         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26348         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26349         memmodel flag.
26351 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26353         * config/avr/avr-stdint.h: Remove trailing blanks.
26354         * config/avr/avr-log.h: Same.
26355         * config/avr/avr-arch.h: Same.
26356         * config/avr/avr-devices.c: Same.
26357         * config/avr/avr-dimode.md: Same.
26358         * config/avr/predicates.md: Same.
26359         * config/avr/avr-c.c: Same.  And fix typo.
26361         * config/avr/avr-protos.h: Same.  And:
26362         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26363         (init_cumulative_args): Rename to avr_init_cumulative_args.
26364         (expand_prologue): Rename to avr_expand_prologue.
26365         (expand_epilogue): Rename to avr_expand_epilogue.
26366         (adjust_insn_length): Rename to avr_adjust_insn_length.
26367         (notice_update_cc): Rename to avr_notice_update_cc.
26368         (final_prescan_insn): Rename to avr_final_prescan_insn.
26369         * config/avr/avr.c: Same.
26370         * config/avr/avr.h: Same.
26371         * config/avr/avr.md: Remove trailing blanks.
26372         (prologue): Use avr_expand_prologue.
26373         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26375 2013-01-14  Richard Biener  <rguenther@suse.de>
26377         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26378         verify_location, collect_subblocks): New functions.
26379         (verify_gimple_in_cfg): Verify that locations only reference
26380         BLOCKs in the functions BLOCK tree.
26382 2013-01-14  Richard Biener  <rguenther@suse.de>
26384         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26385         PHI argument.
26386         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26387         unshare reference.
26388         (insert_out_of_ssa_copy_on_edge): Likewise.
26389         (rewrite_close_phi_out_of_ssa): Likewise.
26390         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26391         debug expressions.
26392         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26393         propagated constants.
26394         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26395         can not be shared.
26397 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26399         * config/avr/avr-modes.def: Add GPL copyright notice.
26401 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26403         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26404         MEMMODEL_MASK to determine memory model.
26405         (atomic_store<mode>): Ditto from operands[2].
26406         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26408 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26410         PR fortran/55935
26411         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26412         (fold_gimple_assign): Don't call unshare_expr here.
26413         (fold_ctor_reference): Call unshare_expr.
26415 2013-01-13  Terry Guo  <terry.guo@arm.com>
26417         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26418         * doc/fragments.texi: Document MULTILIB_REUSE.
26419         * gcc.c (multilib_reuse): New internal spec.
26420         (set_multilib_dir): Also search multilib from multilib_reuse.
26421         * genmultilib (tmpmultilib3): Refactor code.
26422         (tmpmultilib4): Ditto.
26423         (multilib_reuse): New multilib argument.
26425 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26427         * Makefile.in: Update copyright.
26429 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26431         PR middle-end/55890
26432         * calls.c (expand_call): Check if arg_nr is valid.
26434 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26436         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26437         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26438         documentation.  Add missing '__' in front of
26439         __builtin_ia32_packssdw256.
26441 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26443         PR target/55719
26444         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26445         NO_REGS for larl operands.
26446         (s390_reload_larl_operand): Use s390_load_address instead of
26447         emit_move_insn.
26449 2013-01-11  Richard Biener  <rguenther@suse.de>
26451         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26452         (verify_node_sharing): ... here.
26453         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26455 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26457         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26458         Substitute TREECHECKING.
26459         * configure: Regenerate.
26460         * Makefile.in (TREECHECKING): New.
26462 2013-01-11  Richard Guenther  <rguenther@suse.de>
26464         PR tree-optimization/44061
26465         * tree-vrp.c (extract_range_basic): Compute zero as
26466         value-range for __builtin_constant_p of function parameters.
26468 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26470         Update copyright years.
26472 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26474         PR rtl-optimization/55672
26475         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26476         const to be eliminable.
26478 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26480         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26481         * configure: Regenerate.
26483 2013-01-10  Richard Biener  <rguenther@suse.de>
26485         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26487 2013-01-10  Richard Biener  <rguenther@suse.de>
26489         PR bootstrap/55792
26490         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26491         locations for virtual PHI arguments.
26492         (rewrite_update_phi_arguments): Likewise.
26494 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26496         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26497         on to assembler.
26499 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26501         PR tree-optimization/55921
26502         * tree-complex.c (expand_complex_asm): New function.
26503         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26505 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26507         PR target/55718
26508         * config/s390/s390.c (s390_symref_operand_p)
26509         (s390_loadrelative_operand_p): Merge the two functions.
26510         (s390_check_qrst_address, print_operand_address): Add parameters
26511         to s390_loadrelative_operand_p invokation.
26512         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26513         (s390_reload_larl_operand, s390_secondary_reload): Use
26514         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26515         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26517 2013-01-09  Mike Stump  <mikestump@comcast.net>
26519         * dse.c (record_store): Remove unnecessary assert.
26521 2013-01-09  Jan Hubicka  <jh@suse.cz>
26523         PR tree-optimization/55569
26524         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26525         * cfgloop.h (scale_loop_profile): Likewise.
26527 2013-01-09  Jan Hubicka  <jh@suse.cz>
26529         PR lto/45375
26530         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26531         functions.
26532         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26534 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26536         PR middle-end/55114
26537         * expr.h (maybe_emit_group_store): Declare.
26538         * expr.c (maybe_emit_group_store): New function.
26539         * builtins.c (expand_builtin_int_roundingfn): Call it.
26540         (expand_builtin_int_roundingfn_2): Likewise.
26542 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26544         PR rtl-optimization/55829
26545         * lra-constraints.c (match_reload): Add code for absent output.
26546         (curr_insn_transform): Add code for reloads of matched inputs
26547         without output.
26549 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26551         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26552         attribute of movddup insn to DF.
26553         (*vec_interleave_lowv2df): Ditto.
26554         (vec_dupv2df): Ditto.
26556 2013-01-09  Jan Hubicka  <jh@suse.cz>
26558         PR tree-optimiation/55875
26559         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26560         EVERY_ITERATION parameter.
26561         (number_of_iterations_exit): Check if exit is executed every iteration.
26562         (idx_infer_loop_bounds): Similarly here.
26563         (n_of_executions_at_most): Simplify
26564         to only test for cases where statement is dominated by the
26565         particular bound; handle correctly the "postdominance" test.
26566         (scev_probably_wraps_p): Use max loop iterations info
26567         as a global bound first.
26569 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26570             Nick Clifton  <nickc@redhat.com>
26572         * config/v850/v850.md (cbranchsf4): New pattern.
26573         (cstoresf4): New pattern.
26574         (cbranchdf4): New pattern.
26575         (cstoredf4): New pattern.
26576         (movsicc): Disallow floating point comparisons.
26577         (cmpsf_le_insn): Fix order of operators.
26578         (cmpsf_lt_insn): Likewise.
26579         (cmpsf_eq_insn): Likewise.
26580         (cmpdf_le_insn): Likewise.
26581         (cmpdf_lt_insn): Likewise.
26582         (cmpdf_eq_insn): Likewise.
26583         (cmpsf_ge_insn): Use LE comparison.
26584         (cmpdf_ge_insn): Likewise.
26585         (cmpsf_gt_insn): Use LT comparison.
26586         (cmpdf_gt_insn): Likewise.
26587         (cmpsf_ne_insn): Delete pattern.
26588         (cmpdf_ne_insn): Delete pattern.
26589         * config/v850/v850.c (v850_gen_float_compare): Use
26590         gen_cmpdf_eq_insn for NE comparison.
26591         (v850_float_z_comparison_operator)
26592         (v850_float_nz_comparison_operator): Move from here ...
26593         * config/v850/predicates.md: ... to here.  Move GT and GE
26594         comparisons into v850_float_z_comparison_operator.
26595         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26596         Delete prototype.
26597         (v850_float_nz_comparison_operator): Likewise.
26599 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26601         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26602         with calls to gen_insvsi/gen_insvdi.
26604 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26606         * config/i386/i386.c (initial_ix86_tune_features): Set up
26607         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26609 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26610             Jakub Jelinek  <jakub@redhat.com>
26612         PR tree-optimization/48189
26613         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26614         If nitercst is 0, don't predict the exit edge.
26616 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26618         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26619         in asm_fprintf with reg_names.
26620         (aarch64_print_operand_address): Likewise.
26621         (aarch64_return_addr): Likewise.
26622         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26624 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26626         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26627         (INT_U6_BITS): Likewise.
26628         * config/pa/predicates.md (uint6_operand): New predicate.
26629         (shift5_operand, shift6_operand): Likewise.
26630         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26631         arith32_operand.
26632         (lshrdi3): Use shift6_operand.
26633         (shrpsi4, shrpdi4): New insn patterns.
26634         (extzv): Delete expander.
26635         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26636         predicates in unamed zero extract patterns.  Tighten common constraint.
26637         (extv): Delete expander.
26638         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26639         predicates in unamed sign extract patterns.  Tighten common constraint.
26640         (insv): Delete expander.
26641         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26642         predicates in unamed insert patterns.  Tighten common constraint.
26643         Change uint32_operand predicate to uint6_operand predicate in unamed
26644         DImode pattern to insert constant values of type 1...1xxxx.
26646 2013-01-04  Jan Hubicka  <jh@suse.cz>
26648         PR tree-optimization/55823
26649         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26650         issue.
26652 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26653             Uros Bizjak  <ubizjak@gmail.com>
26655         PR rtl-optimization/55845
26656         * df-problems.c (can_move_insns_across): Stop scanning at
26657         volatile_insn_p source instruction or give up if
26658         across_from .. across_to range contains any volatile_insn_p
26659         instructions.
26661 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26663         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26664         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26665         Declare.
26666         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26667         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26669 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26671         PR fortran/55341
26672         * asan.c (asan_clear_shadow): New function.
26673         (asan_emit_stack_protection): Use it.
26675 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26677         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26678         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26679         with tab instead of space.
26681 2013-01-08  Nick Clifton  <nickc@redhat.com>
26683         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26684         register bank 0 at the start of an interrupt handler.
26685         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26686         MDBH registers.
26688 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26690         * config/aarch64/aarch64-simd.md
26691         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26692         (aarch64_simd_bsl): Likewise.
26693         (aarch64_vcond_internal<mode>): Likewise.
26694         (vcond<mode><mode>): Likewise.
26695         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26696         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26698 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26700         * config/aarch64/aarch64-builtins.c
26701         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26703 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26705         PR debug/55579
26706         * tree-sra.c (analyze_access_subtree): Return true also after
26707         potentially creating a debug-only replacement.
26709 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26711         PR middle-end/55890
26712         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26714         PR tree-optimization/54120
26715         * tree-vrp.c (range_fits_type_p): Don't allow
26716         src_precision < precision from signed vr to unsigned_p
26717         if vr->min or vr->max is negative.
26718         (simplify_float_conversion_using_ranges): Test can_float_p
26719         against CODE_FOR_nothing.
26721 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26722             Richard Biener  <rguenther@suse.de>
26724         PR middle-end/55851
26725         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26726         types instead of just INTEGER_TYPE types.
26728 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26730         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26731         TF_SIZE): Define.
26733 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26735         PR target/42661
26736         * config/mips/mips.opt: Change mad to mmad to match documentation.
26738 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26740         PR target/55897
26741         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26742         .progmemx.data now.
26744 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26746         PR target/55897
26747         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26748         (avr_addrspace_t): Add .section_name field.
26749         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26750         array size.
26751         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26752         NULL entry.  Put __memx into .progmemx.data.
26753         (progmem_section_prefix): Remove.
26754         (avr_asm_init_sections): No need to initialize progmem_section.
26755         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26756         section name prefix.
26757         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26758         retrieve the progmem section.
26759         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26760         boundary to run over avr_addrspace[].
26761         (avr_register_target_pragmas): Ditto.
26763 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26765         * varasm.c (output_constant_def_contents): For asan_protect_global
26766         protected strings, adjust DECL_ALIGN if needed, before testing for
26767         anchored symbols.
26768         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26769         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26770         normal decls.
26771         (output_object_block): For asan protected decls, emit asan padding
26772         after their contents.
26773         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26774         (asan_finish_file): Test it here instead.
26776 2013-01-07  Nick Clifton  <nickc@redhat.com>
26777             Matthias Klose  <doko@debian.org>
26778             Doug Kwan  <dougkwan@google.com>
26779             H.J. Lu  <hongjiu.lu@intel.com>
26781         PR driver/55470
26782         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26784         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26786         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26788         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26790         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26792 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26794         PR target/54461
26795         * doc/install.texi (Cross-Compiler-Specific Options): Document
26796         --with-avrlibc.
26798 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26800         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26801         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26802         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26803         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26804         vqmovun_high_s64): Fix source operand number and update copyright.
26806 2013-01-07  Richard Biener  <rguenther@suse.de>
26808         PR middle-end/55890
26809         * gimple.h (gimple_call_builtin_p): New overload.
26810         * gimple.c (validate_call): New function.
26811         (gimple_call_builtin_p): Likewise.
26812         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26813         Use gimple_call_builtin_p.
26814         (find_func_clobbers): Likewise.
26815         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26816         (strlen_optimize_stmt): Likewise.
26818 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26820         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26821         (vld1q_dup_*): Likewise.
26822         (vld1_*): Likewise.
26823         (vld1q_*): Likewise.
26824         (vld1_lane_*): Likewise.
26825         (vld1q_lane_*): Likewise.
26827 2013-01-07  Richard Biener  <rguenther@suse.de>
26829         * lto-streamer.h (LTO_minor_version): Bump to 2.
26831 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26833         * config/aarch64/aarch64-protos.h
26834         (aarch64_const_double_zero_rtx_p): Rename to...
26835         (aarch64_float_const_zero_rtx_p): ...this.
26836         (aarch64_float_const_representable_p): New.
26837         (aarch64_output_simd_mov_immediate): Likewise.
26838         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26839         move immediate case.
26840         * config/aarch64/aarch64.c
26841         (aarch64_const_double_zero_rtx_p): Rename to...
26842         (aarch64_float_const_zero_rtx_p): ...this.
26843         (aarch64_print_operand): Allow printing of new constants.
26844         (aarch64_valid_floating_const): New.
26845         (aarch64_legitimate_constant_p): Check for valid floating-point
26846         constants.
26847         (aarch64_simd_valid_immediate): Likewise.
26848         (aarch64_vect_float_const_representable_p): New.
26849         (aarch64_float_const_representable_p): Likewise.
26850         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26851         (aarch64_output_simd_mov_immediate): New.
26852         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26853         (*movdf_aarch64): Likewise.
26854         * config/aarch64/constraints.md (Ufc): New.
26855         (Y): call aarch64_float_const_zero_rtx.
26856         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26858 2013-01-07  Richard Biener  <rguenther@suse.de>
26860         PR tree-optimization/55888
26861         PR tree-optimization/55862
26862         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26863         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26864         not if it is contained therein.
26866 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26868         * config/avr/t-avr: Typo.
26870 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26872         PR55243
26873         * config/avr/t-avr: Don't automatically rebuild
26874         $(srcdir)/config/avr/t-multilib
26875         $(srcdir)/config/avr/avr-tables.opt
26876         $(srcdir)/doc/avr-mmcu.texi
26877         (avr-mcus): New phony target to build them on request.
26878         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26879         * avr/avr-mcus.def: Adjust comments.
26881 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26883         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26885 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26887         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26889 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26891         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26893 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26895         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26896         to generate profiling.
26897         * config/rs6000/aix64.h (LIB_SPEC): Same.
26899 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26901         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26902         New function.
26903         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26905 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26907         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26908         unconditionally.
26909         (ix86_expand_move): Ditto.
26910         (ix86_zero_extend_to_Pmode): Ditto.
26911         (ix86_expand_call): Ditto.
26912         (ix86_expand_special_args_builtin): Ditto.
26913         (ix86_expand_builtin): Ditto.
26915 2013-01-04  Richard Biener  <rguenther@suse.de>
26917         PR tree-optimization/55862
26918         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26919         translating them through PHI nodes.
26921 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26923         PR tree-optimization/55755
26924         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26925         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26927 2013-01-04  Marek Polacek  <polacek@redhat.com>
26929         PR middle-end/55859
26930         * opts.c (default_options_optimization): Clarify error message.
26932 2013-01-04  Richard Biener  <rguenther@suse.de>
26934         PR middle-end/55863
26935         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26936         reassociation.
26938 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26940         PR target/53789
26941         * config/pa/pa.md (movsi): Revert previous change.
26942         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
26943         references.
26945 2013-01-03  Richard Henderson  <rth@redhat.com>
26947         * config/i386/i386.c (ix86_expand_move): Always assign to op1
26948         after eliminating TLS symbols.
26950 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
26952         PR bootstrap/50167
26953         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
26954         * graphite-poly.c (debug_gmp_value): Likewise.
26956 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
26958         PR target/55712
26959         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
26960         selected code model, define __code_mode_small__, __code_model_medium__,
26961         __code_model_large__, __code_model_32__ or __code_model_kernel__.
26962         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
26963         xchg temporary register with %k.  Declare temporary register as
26964         early clobbered.
26965         [__x86_64__]: For medium and large code models, preserve %rbx register.
26967 2013-01-03  Richard Biener  <rguenther@suse.de>
26969         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
26970         (dump_subscript): Adjust.
26971         (finalize_ddr_dependent): Do not dump redundant info.
26972         (analyze_siv_subscript): Adjust.
26973         (subscript_dependence_tester): Likewise.
26974         (compute_affine_dependence): Likewise.
26976 2013-01-03  Richard Biener  <rguenther@suse.de>
26978         Revert
26979         2013-01-03  Richard Biener  <rguenther@suse.de>
26981         PR tree-optimization/55857
26982         * tree-vect-stmts.c (vectorizable_load): Do not setup
26983         re-alignment for invariant loads.
26985         2013-01-02  Richard Biener  <rguenther@suse.de>
26987         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26988         invariant load do not generate a vector load from the scalar location.
26990 2013-01-03  Richard Biener  <rguenther@suse.de>
26992         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
26993         for not vectorizing.
26994         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
26995         not build INDIRECT_REFs, call get_name once only.
26996         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
26997         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
26999 2013-01-03  Richard Biener  <rguenther@suse.de>
27001         PR tree-optimization/55857
27002         * tree-vect-stmts.c (vectorizable_load): Do not setup
27003         re-alignment for invariant loads.
27005 2013-01-03  Richard Biener  <rguenther@suse.de>
27007         PR lto/55848
27008         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27009         prefer a built-in decl.
27011 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27013         * gcc.c (process_command): Update copyright notice dates.
27014         * gcov.c (print_version): Likewise.
27015         * gcov-dump.c (print_version): Likewise.
27017         PR rtl-optimization/55838
27018         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27019         iv0.step, iv1.step and step.
27021 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27022             Marc Glisse  <marc.glisse@inria.fr>
27024         PR tree-optimization/55832
27025         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27026         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27027         integer_{one,zero}_node.
27029 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27031         PR debug/54402
27032         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27033         * var-tracking.c (reverse_op): Don't add reverse ops to
27034         VALUEs that have already
27035         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27037 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27039         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27041 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27043         * dumpfile.c (dump_loc): Print filename with location.
27044         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27045         new location_t parameter to emit complete unroll message with
27046         new dump framework.
27047         (canonicalize_loop_induction_variables): Compute loops location
27048         and pass to try_unroll_loop_completely.
27049         * loop-unroll.c (report_unroll_peel): New function.
27050         (peel_loops_completely): Use new dump format with location
27051         for main dumpfile message, and invoke report_unroll_peel on success.
27052         (decide_unrolling_and_peeling): Ditto.
27053         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27054         by report_unroll_peel.
27055         (decide_peel_completely): Ditto.
27056         (decide_unroll_constant_iterations): Ditto.
27057         (decide_unroll_runtime_iterations): Ditto.
27058         (decide_peel_simple): Ditto.
27059         (decide_unroll_stupid): Ditto.
27060         * cfgloop.c (get_loop_location): New function.
27061         * cfgloop.h (get_loop_location): Declare.
27063 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27065         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27066         NULL.
27068 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27070         PR middle-end/55198
27071         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27072         BLKmode objects when EXPAND_MEMORY is specified.
27074 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27076         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27077         in loop predicate.
27078         (fold_builtin_cpu): Do not share cpu model decls across statements.
27080 2013-01-02  Jason Merrill  <jason@redhat.com>
27082         PR c++/55804
27083         * tree.c (build_array_type_1): Revert earlier change.
27085 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27087         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27088         "cortex-a57".
27089         * config/aarch64/aarch64-tune.md: Re-generate.
27091 2013-01-02  Richard Biener  <rguenther@suse.de>
27093         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27094         invariant load do not generate a vector load from the scalar location.
27096 2013-01-02  Richard Biener  <rguenther@suse.de>
27098         PR bootstrap/55784
27099         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27100         * configure: Regenerate.
27102 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27104         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27105         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27106         (expand_builtin_int_roundingfn_2): Keep the original target around
27107         for the fallback case.
27109 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27111         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27112         to be clear for sign changes.
27114 2013-01-01  Jan Hubicka  <jh@suse.cz>
27116         * ipa-inline-analysis.c: Fix formatting.
27118 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27120         PR tree-optimization/55831
27121         * tree-vect-loop.c (get_initial_def_for_induction): Use
27122         gsi_after_labels instead of gsi_start_bb.
27124 Copyright (C) 2013 Free Software Foundation, Inc.
27126 Copying and distribution of this file, with or without modification,
27127 are permitted in any medium without royalty provided the copyright
27128 notice and this notice are preserved.