[gcc/]
[official-gcc.git] / gcc / ChangeLog
blob62105643d4e60e56bc5fc8531d68f46febe79855
1 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
4         extend and extend_add.
6 2013-10-21  Richard Biener  <rguenther@suse.de>
8         PR tree-optimization/58794
9         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
10         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
12 2013-10-21  Richard Biener  <rguenther@suse.de>
14         PR middle-end/58742
15         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
16         to (T) X for sign-changing conversions (or no conversion).
18 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
20         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
22 2013-10-20  Jan Hubicka  <jh@suse.cz>
24         * config/i386/i386-tune.def: Add comment; organize into categories
26 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
28         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
29         argument.  Update function comment.
30         (expand_set_or_movmem_via_rep): New function combining
31         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
32         (expand_movmem_via_rep_mov): Remove.
33         expand_setmem_via_rep_stos): Remove.
34         (expand_movmem_epilogue): Update calls correspondingly.
35         (expand_setmem_epilogue_via_loop): Likewise.
36         (emit_memset): New.
37         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
38         (expand_set_or_movmem_prologue): New function combining
39         expand_movmem_prologue and expand_setmem_prologue.
40         (expand_movmem_prologue): Remove.
41         (expand_setmem_prologue): Remove.
42         (expand_set_or_movmem_constant_prologue): New function combining
43         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
44         (expand_constant_movmem_prologue): Remove.
45         (expand_constant_setmem_prologue): Remove.
46         (promote_duplicated_reg): Allow vector-const0 value.
47         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
48         and ix86_expand_setmem.
49         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
50         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
52 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
54         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
55         little endian.
56         (vec_unpacku_hi_v8hi): Likewise.
57         (vec_unpacku_lo_v16qi): Likewise.
58         (vec_unpacku_lo_v8hi): Likewise.
60 2013-10-20  Jan Hubicka  <jh@suse.cz>
62         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
63         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
64         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
66 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
68         * config/mips/mips.h (ISA_HAS_WSBH): Define.
69         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
70         constants.
71         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
73 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
75         PR target/58603
76         * system.h: Undef m_slot.
78 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
80         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
81         all elements for both endian flavors.
83 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
85         PR target/58792
86         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
87         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
88         and SI_REG for 64bit SYSV ABI targets.
90 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
92         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
93         to multi_reg_return.  Clarify that we are skipping USEs of multiple
94         return registers.  Use bool type where appropriate.
96 2013-10-18  Jan Hubicka  <jh@suse.cz>
98         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
99         for cold functions.
100         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
101         (X86_TUNE_PUSH_MEMORY): Likewise.
102         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
103         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
104         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
105         New.
106         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
107         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
108         Remove.
109         (ix86_option_override_internal): Update to use tune features instead
110         of variables.
112 2013-10-18  Cong Hou  <congh@google.com>
114         PR tree-optimization/58508
115         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
116         statement that contains data refs with zero-step.
118 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
120         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
121         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
122         * gimple-low.c (gimple_check_call_arg,
123         gimple_check_call_matching_types): Move to cgraph.c.
124         * gimple-low.h: Remove prototype.
125         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
126         Relocate from gimple-low.c.
127         * cgraph.h: Add prototype.  Don't include basic-block.h.
128         * gimplify.c: Add gimple-low to include list.
129         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
130         * tree-eh.c: Add gimple-low to include list.
131         * tree-nested.c: Likewise.
132         * cfgexpand.c: Add tree-ssa-address.h to include list.
133         * expr.c: Likewise.
134         * gimple-fold.c: Likewise.
135         * gimple-ssa-strength-reduction.c: Likewise.
136         * trans-mem.c: Likewise.
137         * tree-mudflap.c: Likewise.
138         * tree-ssa-loop-ivopts.c: Likewise.
139         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
140         (degenerate_phi_result): Move to tree-phinodes.c.
141         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
142         * tree-ssa-threadedge.c: Likewise.
143         * tree-vrp.c: Likewise.
144         * tree-phinodes.c (degenerate_phi_result): Relocate here.
145         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
146         * tree-phinodes.h (degenerate_phi_result): Add prototype.
147         * tree-ssa-copy.c: Include tree-ssa-dom.h.
148         * tree-ssa-forwprop.c: Likewise.
149         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
150         pass_data_cleanup_cfg_post_optimizing,
151         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
152         * tree-optimize.c: Delete File.
153         * graphite.c: Include tree-cfgcleanup.h.
154         * passes.c: Likewise.
155         * tree-cfg.c: Likewise.
156         * tree-profile.c: Likewise.
157         * tree-ssa-dse.c: Likewise.
158         * tree-ssa-loop-ivcanon.c: Likewise.
159         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
160         * tree-outof-ssa.c: Include sbitmap.h.
161         * tree-ssa-live.c: Likewise.
162         * tree-ssa-propagate.c: Likewise.
163         * tree-ssa-structalias.c: Likewise.
164         * tree-stdarg.c: Likewise.
165         * Makefile.in (OBJS): Delete tree-optimize.o.
166         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
167         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
168         * varasm.c: Include basic-block.h.
169         * cfgloop.h: Include function.h instead of basic-block.h
170         (bb_loop_depth): Move to cfgloop.c.
171         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
173 2013-10-18  Teresa Johnson  <tejohnson@google.com>
175         * predict.c (probably_never_executed): Compare frequency-based
176         count to number of training runs.
177         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
179 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
181         * config/arm/arm.c (cortexa9_extra_costs): New table.
182         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
184 2013-10-18  Jeff Law  <law@redhat.com>
186         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
188         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
189         (register_jump_thread): Add "registered_jump_thread" debug
190         counter support.
191         * dbgcnt.def (registered_jump_thread): New debug counter.
193 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
195         * config/rs6000/rs6000.c: Include cgraph.h.
197 2013-10-18  Teresa Johnson  <tejohnson@google.com>
199         * tree-ssa-tail-merge.c (replace_block_by): Update edge
200         weights during merging.
202 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
204         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
205         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
206         * tree-ssa-operands.h: Remove prototype.
207         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
208         * gimple.c (virtual_operand_p): Relocate from gimple.c.
209         * gimple.h: Add prototype.
210         * gimple-ssa.h: Include tree-ssa-operands.h.
211         * tree-dump.c: Add tree-cfg.h to include list.
212         * tree-ssa-alias.c: Add ipa-reference.h to include list.
213         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
214         * config/i386/i386.c: Don't include tree-flow.h.
215         * config/rs6000/rs6000.c: Likewise.
217 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
219         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
221 2013-10-18  Richard Biener  <rguenther@suse.de>
223         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
224         or TYPE_UNSIGNED of integral types.
225         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
226         NULL_TREE for zero-precision integral types.
228 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
230         * config/aarch64/arm_neon.h
231         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
233 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
235         * opth-gen.awk: Define target_flags_explicit.
237 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
239         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
240         fields to the reg_addr array that describes the valid addressing
241         mode for any register, general purpose registers, floating point
242         registers, and Altivec registers.
243         (FIRST_RELOAD_REG_CLASS): Likewise.
244         (LAST_RELOAD_REG_CLASS): Likewise.
245         (struct reload_reg_map_type): Likewise.
246         (reload_reg_map_type): Likewise.
247         (RELOAD_REG_VALID): Likewise.
248         (RELOAD_REG_MULTIPLE): Likewise.
249         (RELOAD_REG_INDEXED): Likewise.
250         (RELOAD_REG_OFFSET): Likewise.
251         (RELOAD_REG_PRE_INCDEC): Likewise.
252         (RELOAD_REG_PRE_MODIFY): Likewise.
253         (reg_addr): Likewise.
254         (mode_supports_pre_incdec_p): New helper functions to say whether
255         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
256         (mode_supports_pre_modify_p): Likewise.
257         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
258         print the valid address mode bits for each mode.
259         (rs6000_debug_print_mode): Likewise.
260         (rs6000_debug_reg_global): Likewise.
261         (rs6000_setup_reg_addr_masks): New function to set up the address
262         mask bits for each type.
263         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
264         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
265         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
266         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
267         PRE_MODIFY are supported.
268         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
269         registers, instead of {src,dest}_av_p.
270         (rs6000_print_options_internal): Tweak the debug output slightly.
272 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
274         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
275         isa attribute.
277 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
279         * tree-flow.h (struct omp_region): Move to omp-low.c.
280         Remove omp_ prototypes and variables.
281         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
282         (copy_var_decl): Relocate prototype from tree-flow.h.
283         * gimple.c (copy_var_decl): Relocate from omp-low.c.
284         * tree.h: Move prototype to omp-low.h.
285         * omp-low.h: New File.  Relocate prototypes here.
286         * omp-low.c (struct omp_region): Make local here.
287         (root_omp_region): Make static.
288         (copy_var_decl) Move to gimple.c.
289         (new_omp_region): Make static.
290         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
291         * tree-cfg.c: Include omp-low.h.
292         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
293         * gimplify.c: Include omp-low.h.
294         * tree-parloops.c: Likewise.
296 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
298         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
299         operands are in memory, prefer to force non-matched operand 1 to
300         the register.
302 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
304         PR target/58673
305         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
306         restrict TImode addresses to single indirect registers if both
307         -mquad-memory and -mvsx-timode are used.
308         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
309         we should emit load/store quad.  Remove using %y for quad memory
310         addresses.
312         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
313         constraints to allow load/store quad on machines where TImode is
314         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
315         for TImode to load integer constants.
317 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
319         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
321 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
323         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
324         handling of STACK_REG.
326 2013-10-17  Richard Biener  <rguenther@suse.de>
328         PR tree-optimization/58143
329         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
330         New function.
331         (rewrite_to_defined_overflow): Likewise.
332         (move_computations_dom_walker::before_dom): Rewrite stmts
333         with undefined signed overflow that are not always executed
334         into unsigned arithmetic.
336 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
338         PR target/57756
339         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
340         explicit isa flag to be an options variable, instead of using
341         global_options_set.  Remove define from rs6000.h.
342         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
344         * config/rs6000/rs6000.c (rs6000_option_override_internal):
345         Initialize rs6000_isa_flags_explicit.
346         (rs6000_function_specific_save): Add gcc_options* parameter, so
347         that the powerpc builds after the 2013-10-15 changes.
348         (rs6000_function_specific_restore): Likewise.
350 2013-10-16  DJ Delorie  <dj@redhat.com>
352         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
353         op is a REG before checking REGNO.
354         (rl78_alloc_physical_registers): Verify pattern is a SET before
355         checking SET_SRC.
357 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
359         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
360         endianness.
361         (vec_unpacks_lo_v4sf): Likewise.
362         (vec_unpacks_float_hi_v4si): Likewise.
363         (vec_unpacks_float_lo_v4si): Likewise.
364         (vec_unpacku_float_hi_v4si): Likewise.
365         (vec_unpacku_float_lo_v4si): Likewise.
367 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
369         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
370         (vsx_concat_v2sf): Likewise.
372 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
374         * config/aarch64/aarch64.md
375         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
377 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
379         PR tree-optimization/58697
380         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
381         estimated_loop_iterations_int.
382         (max_stmt_executions_int): Call get_max_loop_iterations_int.
383         (get_max_loop_iterations_int): New.  HWINT version of
384         get_max_loop_iterations.
385         * cfgloop.h: Add prototypes.
386         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
387         * loop-unroll.c (decide_peel_once_rolling): Call
388         get_estimated_loop_iterations_int.
389         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
390         * tree-ssa-loop-niter.h: Tweak prototypes.
392 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
394         * gengtype-parse.c (struct_field_seq): Ignore access-control
395         keywords ("public:" etc).
397 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
399         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
400         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
402 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
404         * config/arm/arm.opt (mlra): New option.
405         * config/arm/arm.c (arm_lra_p): New function.
406         (TARGET_LRA_P): Define.
408 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
410         * tree-core.h (tree_code_name): Remove.
411         * tree.h (get_tree_code_name): New prototype.
412         * tree.c (tree_code_name): Make static.
413         (get_tree_code_name): New function.
414         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
415         tree_class_check_failed, tree_range_check_failed,
416         tree_not_class_check_failed, omp_clause_check_failed,
417         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
418         wrapper get_tree_code_name instead of calling tree_code_name directly.
419         * tree-vrp.c (dump_asserts_for): Likewise.
420         * tree-dump.c (dequeue_and_dump): Likewise.
421         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
422         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
423         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
424         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
425         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
426         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
427         dump_gimple_omp_for): Likewise.
428         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
429         * tree-ssa-pre.c (print_pre_expr): Likewise.
430         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
431         * print-tree.c (print_node_brief, print_node): Likewise.
432         * gimple.c (gimple_check_failed): Likewise.
433         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
434         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
435         * config/mep/mep.c (mep_validate_vliw): Likewise.
436         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
437         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
439 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
441         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
442         for AMD bdver3.
444 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
446         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
447         (MULTILIB_MATCHES): Add multilib for -march=v8.
449 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
451         PR target/57756
452         * optc-save-gen.awk: Add extra parameter to the save and restore
453         target calls.
454         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
455         * tree.c (build_optimization_node): New parameter.  Add extra parameter
456         to call to cl_optimization_save.
457         (build_target_option_node): New parameter. Add extra parameter
458         to call to cl_target_option_save.
459         * tree.h (build_optimization_node): New parameter.
460         (build_target_option_node): New parameter.
461         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
462         build_optimization_node and build_target_option_node.
463         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
464         (handle_pragma_push_options): Ditto.
465         * toplev.c (process_options): Ditto.
466         * opts.c (init_options_struct): Check for opts_set non-null.
467         * target.def (target_option.save): New parameter.
468         (target_option.restore): New parameter.
469         * tm.texi: Generate.
470         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
471         (ix86_pragma_target_parse): Ditto.
472         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
473         parameters.
474         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
475         to  build_optimization_node and build_target_option_node.
476         (rs6000_valid_attribute_p): Ditto.
477         (rs6000_pragma_target_parse): Ditto.
478         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
479         data.
480         * config/i386/i386.h:
481         TARGET_64BIT_P: New Macro
482         TARGET_MMX_P: New Macro.
483         TARGET_3DNOW_P: New Macro.
484         TARGET_3DNOW_A_P: New Macro.
485         TARGET_SSE_P: New Macro.
486         TARGET_SSE2_P: New Macro.
487         TARGET_SSE3_P: New Macro.
488         TARGET_SSSE3_P: New Macro.
489         TARGET_SSE4_1_P: New Macro.
490         TARGET_SSE4_2_P: New Macro.
491         TARGET_AVX_P: New Macro.
492         TARGET_AVX2_P: New Macro.
493         TARGET_AVX512F_P: New Macro.
494         TARGET_AVX512PF_P: New Macro.
495         TARGET_AVX512ER_P: New Macro.
496         TARGET_AVX512CD_P: New Macro.
497         TARGET_FMA_P: New Macro.
498         TARGET_SSE4A_P: New Macro.
499         TARGET_FMA4_P: New Macro.
500         TARGET_XOP_P: New Macro.
501         TARGET_LWP_P: New Macro.
502         TARGET_ABM_P: New Macro.
503         TARGET_BMI_P: New Macro.
504         TARGET_BMI2_P: New Macro.
505         TARGET_LZCNT_P: New Macro.
506         TARGET_TBM_P: New Macro.
507         TARGET_POPCNT_P: New Macro.
508         TARGET_SAHF_P: New Macro.
509         TARGET_MOVBE_P: New Macro.
510         TARGET_CRC32_P: New Macro.
511         TARGET_AES_P: New Macro.
512         TARGET_PCLMUL_P: New Macro.
513         TARGET_CMPXCHG16B_P: New Macro.
514         TARGET_FSGSBASE_P: New Macro.
515         TARGET_RDRND_P: New Macro.
516         TARGET_F16C_P: New Macro.
517         TARGET_RTM_P: New Macro.
518         TARGET_HLE_P: New Macro.
519         TARGET_RDSEED_P: New Macro.
520         TARGET_PRFCHW_P: New Macro.
521         TARGET_ADX_P: New Macro.
522         TARGET_FXSR_P: New Macro.
523         TARGET_XSAVE_P: New Macro.
524         TARGET_XSAVEOPT_P: New Macro.
525         TARGET_LP64_P: New Macro.
526         TARGET_X32_P: New Macro.
527         TARGET_FPMATH_DEFAULT_P: New Macro.
528         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
529         * config/i386/i386.c (ix86_option_override_internal): New parameters.
530         opts and opts_set.
531         Change ix86_tune_string to access opts->x_ix86_tune_string.
532         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
533         Change ix86_arch_string to access opts->x_ix86_arch_string.
534         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
535         Change ix86_pmode to access opts->x_ix86_pmode.
536         Change ix86_abi to access opts->x_ix86_abi.
537         Change ix86_cmodel to access opts->x_ix86_cmodel.
538         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
539         Change ix86_isa_flags_explicit to access
540         opts->x_ix86_isa_flags_explicit.
541         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
542         Change ix86_regparm to access opts->x_ix86_regparm.
543         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
544         Change ix86_preferred_stack_boundary_arg to access
545         opts->x_ix86_preferred_stack_boundary_arg.
546         Change ix86_force_align_arg_pointer to access
547         opts->x_ix86_force_align_arg_pointer.
548         Change ix86_incoming_stack_boundar_arg to access
549         opts->x_ix86_incoming_stack_boundar_arg.
550         Change ix86_fpmath to access opts->x_ix86_fpmath.
551         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
552         Change ix86_recip_name to access opts->x_ix86_recip_name.
553         Change ix86_stack_protector_guard to access
554         opts->x_ix86_stack_protector_guard.
555         Change ix86_tune_memcpy_strategy to access
556         opts->x_ix86_tune_memcpy_strategy.
557         Change ix86_tune_memset_strategy to access
558         opts->x_ix86_tune_memset_strategy.
559         Change global_options to access opts.
560         Change global_options_set to access opts_set.
561         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
562         Change TARGET_MMX to TARGET_MMX_P (opts->...).
563         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
564         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
565         Change TARGET_SSE to TARGET_SSE_P (opts->...).
566         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
567         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
568         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
569         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
570         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
571         Change TARGET_AVX to TARGET_AVX_P (opts->...).
572         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
573         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
574         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
575         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
576         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
577         Change TARGET_FMA to TARGET_FMA_P (opts->...).
578         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
579         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
580         Change TARGET_XOP to TARGET_XOP_P (opts->...).
581         Change TARGET_LWP to TARGET_LWP_P (opts->...).
582         Change TARGET_ABM to TARGET_ABM_P (opts->...).
583         Change TARGET_BMI to TARGET_BMI_P (opts->...).
584         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
585         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
586         Change TARGET_TBM to TARGET_TBM_P (opts->...).
587         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
588         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
589         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
590         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
591         Change TARGET_AES to TARGET_AES_P (opts->...).
592         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
593         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
594         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
595         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
596         Change TARGET_F16C to TARGET_F16C_P (opts->...).
597         Change TARGET_RTM to TARGET_RTM_P (opts->...).
598         Change TARGET_HLE to TARGET_HLE_P (opts->...).
599         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
600         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
601         Change TARGET_ADX to TARGET_ADX_P (opts->...).
602         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
603         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
604         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
605         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
606         Change TARGET_X32 to TARGET_X32_P (opts->...).
607         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
608         Change TARGET_FLOAT_RETURNS_IN_80387 to
609         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
610         (ix86_function_specific_save): New parameter. Use opts-> fields
611         to replace global fields.
612         (ix86_function_specific_restore): Ditto.
613         (ix86_valid_target_attribute_inner_p): New parameters.
614         Fix recursive call.
615         Fix call to ix86_handle_option and set_option.
616         (ix86_valid_target_attribute_tree): New parameters.
617         Change global_options to access opts.
618         Change global_options_set to access opts_set.
619         Fix call to ix86_valid_target_attribute_inner_p.
620         Change ix86_tune_string to access opts->x_ix86_tune_string.
621         Change ix86_arch_string to access opts->x_ix86_arch_string.
622         Change ix86_fpmath to access opts->x_ix86_fpmath
623         Fix call to ix86_option_override_internal.
624         Fix call to ix86_add_new_builtins.
625         Fix calls to build_optimization_node and build_target_option_node.
626         (ix86_valid_target_attribute_p): Remove access to global_options.
627         Use new gcc_options structure func_options.
628         Fix call to ix86_valid_target_attribute_tree.
629         Fix call to  build_optimization_node.
630         (get_builtin_code_for_version): Fix call to
631         ix86_valid_target_attribute_tree.
633 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
635         * Makefile.in (PICFLAG): New.
636         (enable_host_shared): New.
637         (INTERNAL_CFLAGS): Use PICFLAG.
638         (LIBIBERTY): Use pic build of libiberty.a if configured with
639         --enable-host-shared.
640         * configure.ac: Add --enable-host-shared, setting up new
641         PICFLAG variable.
642         * configure: Regenerate.
643         * doc/install.texi (--enable-shared): Add note contrasting it with ...
644         (--enable-host-shared): New option.
646 2013-10-15  Richard Biener  <rguenther@suse.de>
648         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
649         for built-in functions.
651 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
653         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
654         (optimize_range_tests_1): New function,
655         extracted from optimize_range_tests.
656         (optimize_range_tests_xor): Similarly.
657         (optimize_range_tests_diff): New function.
658         (optimize_range_tests): Use optimize_range_tests_1.
660 2013-10-15  Cong Hou  <congh@google.com>
662         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
663         requirement of the reduction pattern so that one operand of the
664         reduction operation can come from outside of the loop.
666 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
668         * config/arm/neon-schedgen.ml: Remove.
669         * config/arm/cortex-a9-neon.md: Remove comment regarding
670         neon-schedgen.ml.
672 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
674         * config/arm/types: Remove old neon types.
676 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
678         * config/arm/cortex-a7.md
679         (cortex_a7_neon_type): New.
680         (cortex_a7_neon_mul): Update for new types.
681         (cortex_a7_neon_mla): Likewise.
682         (cortex_a7_neon): Likewise.
684 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
686         * config/arm/cortex-a15-neon.md
687         (cortex_a15_neon_type): New,
689         (cortex_a15_neon_int_1): Remove.
690         (cortex_a15_neon_int_2): Likewise.
691         (cortex_a15_neon_int_3): Likewise.
692         (cortex_a15_neon_int_4): Likewise.
693         (cortex_a15_neon_int_5): Likewise.
694         (cortex_a15_neon_vqneg_vqabs): Likewise.
695         (cortex_a15_neon_vmov): Likewise.
696         (cortex_a15_neon_vaba): Likewise.
697         (cortex_a15_neon_vaba_qqq): Likewise.
698         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
699         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
700         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
701         Likewise.
702         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
703         (cortex_a15_neon_mla_qqq_8_16): Likewise.
704         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
705         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
706         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
707         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
708         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
709         (cortex_a15_neon_shift_1): Likewise.
710         (cortex_a15_neon_shift_2): Likewise.
711         (cortex_a15_neon_shift_3): Likewise.
712         (cortex_a15_neon_vshl_ddd): Likewise.
713         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
714         (cortex_a15_neon_vsra_vrsra): Likewise.
715         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
716         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
717         (cortex_a15_neon_bp_3cycle): Likewise.
718         (cortex_a15_neon_ldm_2): Likewise.
719         (cortex_a15_neon_stm_2): Likewise.
720         (cortex_a15_neon_mcr): Likewise.
721         (cortex_a15_neon_mrc): Likewise.
722         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
723         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
724         (cortex_a15_neon_fp_vmul_ddd): Likewise.
725         (cortex_a15_neon_fp_vmul_qqd): Likewise.
726         (cortex_a15_neon_fp_vmla_ddd): Likewise.
727         (cortex_a15_neon_fp_vmla_qqq): Likewise.
728         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
729         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
730         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
731         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
732         (cortex_a15_neon_bp_simple): Likewise.
733         (cortex_a15_neon_bp_2cycle): Likewise.
734         (cortex_a15_neon_bp_3cycle): Likewise.
735         (cortex_a15_neon_vld1_1_2_regs): Likewise.
736         (cortex_a15_neon_vld1_3_4_regs): Likewise.
737         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
738         (cortex_a15_neon_vld2_4_regs): Likewise.
739         (cortex_a15_neon_vld3_vld4): Likewise.
740         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
741         (cortex_a15_neon_vst1_3_4_regs): Likewise.
742         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
743         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
744         (cortex_a15_neon_vst3_vst4): Rename to...
745         (cortex_a15_neon_vst4): This, update for new attributes.
746         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
747         (cortex_a15_neon_vld3_vld4_lane): Likewise.
748         (cortex_a15_neon_vst1_vst2_lane): Likewise.
749         (cortex_a15_neon_vst3_vst4_lane): Likewise.
750         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
751         (cortex_a15_neon_ldm_2): Likewise.
752         (cortex_a15_neon_stm_2): Likewise.
753         (cortex_a15_neon_mcr): Likewise.
754         (cortex_a15_neon_mcr_2_mcrr): Likewise.
755         (cortex_a15_neon_mrc): Likewise.
756         (cortex_a15_neon_mrrc): Likewise.
758         (cortex_a15_neon_abd): New.
759         (cortex_a15_neon_abd_q): Likewise.
760         (cortex_a15_neon_aba): Likewise.
761         (cortex_a15_neon_aba_q): Likewise.
762         (cortex_a15_neon_acc): Likewise.
763         (cortex_a15_neon_acc_q): Likewise.
764         (cortex_a15_neon_arith_basic): Likewise.
765         (cortex_a15_neon_arith_complex): Likewise.
766         (cortex_a15_neon_multiply): Likewise.
767         (cortex_a15_neon_multiply_q): Likewise.
768         (cortex_a15_neon_mla): Likewise.
769         (cortex_a15_neon_mla_q): Likewise.
770         (cortex_a15_neon_sat_mla_long): Likewise.
771         (cortex_a15_neon_shift_acc): Likewise.
772         (cortex_a15_neon_shift_imm_basic): Likewise.
773         (cortex_a15_neon_shift_imm_complex): Likewise.
774         (cortex_a15_neon_shift_reg_basic): Likewise.
775         (cortex_a15_neon_shift_reg_basic_q): Likewise.
776         (cortex_a15_neon_shift_reg_complex): Likewise.
777         (cortex_a15_neon_shift_reg_complex_q): Likewise.
778         (cortex_a15_neon_fp_negabs): Likewise
779         (cortex_a15_neon_fp_arith): Likewise
780         (cortex_a15_neon_fp_arith_q): Likewise
781         (cortex_a15_neon_fp_cvt_int): Likewise
782         (cortex_a15_neon_fp_cvt_int_q): Likewise
783         (cortex_a15_neon_fp_cvt_16): Likewise
784         (cortex_a15_neon_fp_mul): Likewise
785         (cortex_a15_neon_fp_mul_q): Likewise
786         (cortex_a15_neon_fp_mla): Likewise
787         (cortex_a15_neon_fp_mla_q): Likewise
788         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
789         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
790         (cortex_a15_neon_bitops): Likewise.
791         (cortex_a15_neon_bitops_q): Likewise.
792         (cortex_a15_neon_from_gp): Likewise.
793         (cortex_a15_neon_from_gp_q): Likewise.
794         (cortex_a15_neon_tbl3_tbl4): Likewise.
795         (cortex_a15_neon_zip_q): Likewise.
796         (cortex_a15_neon_to_gp): Likewise.
797         (cortex_a15_neon_load_a): Likewise.
798         (cortex_a15_neon_load_b): Likewise.
799         (cortex_a15_neon_load_c): Likewise.
800         (cortex_a15_neon_load_d): Likewise.
801         (cortex_a15_neon_load_e): Likewise.
802         (cortex_a15_neon_load_f): Likewise.
803         (cortex_a15_neon_store_a): Likewise.
804         (cortex_a15_neon_store_b): Likewise.
805         (cortex_a15_neon_store_c): Likewise.
806         (cortex_a15_neon_store_d): Likewise.
807         (cortex_a15_neon_store_e): Likewise.
808         (cortex_a15_neon_store_f): Likewise.
809         (cortex_a15_neon_store_g): Likewise.
810         (cortex_a15_neon_store_h): Likewise.
811         (cortex_a15_vfp_to_from_gp): Likewise.
813 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
815         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
817         (cortex_a9_neon_vshl_ddd): Remove.
818         (cortex_a9_neon_vst3_vst4): Likewise.
819         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
821         (cortex_a9_neon_bit_ops_q): New.
823         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
824         (cortex_a9_neon_int_2): Likewise.
825         (cortex_a9_neon_int_3): Likewise.
826         (cortex_a9_neon_int_4): Likewise.
827         (cortex_a9_neon_int_5): Likewise.
828         (cortex_a9_neon_vqneg_vqabs): Likewise.
829         (cortex_a9_neon_vmov): Likewise.
830         (cortex_a9_neon_vaba): Likewise.
831         (cortex_a9_neon_vaba_qqq): Likewise.
832         (cortex_a9_neon_shift_1): Likewise.
833         (cortex_a9_neon_shift_2): Likewise.
834         (cortex_a9_neon_shift_3): Likewise.
835         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
836         (cortex_a9_neon_vsra_vrsra): Likewise.
837         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
838         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
839         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
840         Likewise.
841         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
842         (cortex_a9_neon_mla_qqq_8_16): Likewise.
843         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
844         Likewise.
845         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
846         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
847         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
848         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
849         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
850         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
851         (cortex_a9_neon_fp_vsum): Likewise.
852         (cortex_a9_neon_fp_vmul_ddd): Likewise.
853         (cortex_a9_neon_fp_vmul_qqd): Likewise.
854         (cortex_a9_neon_fp_vmla_ddd): Likewise.
855         (cortex_a9_neon_fp_vmla_qqq): Likewise.
856         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
857         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
858         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
859         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
860         (cortex_a9_neon_bp_simple): Likewise.
861         (cortex_a9_neon_bp_2cycle): Likewise.
862         (cortex_a9_neon_bp_3cycle): Likewise.
863         (cortex_a9_neon_ldr): Likewise.
864         (cortex_a9_neon_str): Likewise.
865         (cortex_a9_neon_vld1_1_2_regs): Likewise.
866         (cortex_a9_neon_vld1_3_4_regs): Likewise.
867         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
868         (cortex_a9_neon_vld2_4_regs): Likewise.
869         (cortex_a9_neon_vld3_vld4): Likewise.
870         (cortex_a9_neon_vld1_vld2_lane): Likewise.
871         (cortex_a9_neon_vld3_vld4_lane): Likewise.
872         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
873         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
874         (cortex_a9_neon_vst1_3_4_regs): Likewise.
875         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
876         (cortex_a9_neon_vst1_vst2_lane): Likewise.
877         (cortex_a9_neon_vst3_vst4_lane): Likewise.
878         (cortex_a9_neon_mcr): Likewise.
879         (cortex_a9_neon_mcr_2_mcrr): Likewise.
880         (cortex_a9_neon_mrc): Likewise.
881         (cortex_a9_neon_mrrc): Likewise.
883 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
885         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
887         (cortex_a8_neon_vshl_ddd): Remove.
888         (cortex_a8_neon_vst3_vst4): Likewise.
889         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
891         (cortex_a8_neon_bit_ops_q): New.
893         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
894         (cortex_a8_neon_int_2): Likewise..
895         (cortex_a8_neon_int_3): Likewise.
896         (cortex_a8_neon_int_5): Likewise.
897         (cortex_a8_neon_vqneg_vqabs): Likewise.
898         (cortex_a8_neon_int_4): Likewise.
899         (cortex_a8_neon_vaba): Likewise.
900         (cortex_a8_neon_vaba_qqq): Likewise.
901         (cortex_a8_neon_shift_1): Likewise.
902         (cortex_a8_neon_shift_2): Likewise.
903         (cortex_a8_neon_shift_3): Likewise.
904         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
905         (cortex_a8_neon_vsra_vrsra): Likewise.
906         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
907         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
908         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
909         Likewise.
910         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
911         (cortex_a8_neon_mla_qqq_8_16): Likewise.
912         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
913         Likewise.
914         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
915         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
916         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
917         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
918         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
919         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
920         (cortex_a8_neon_fp_vsum): Likewise.
921         (cortex_a8_neon_fp_vmul_ddd): Likewise.
922         (cortex_a8_neon_fp_vmul_qqd): Likewise.
923         (cortex_a8_neon_fp_vmla_ddd): Likewise.
924         (cortex_a8_neon_fp_vmla_qqq): Likewise.
925         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
926         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
927         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
928         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
929         (cortex_a8_neon_bp_simple): Likewise.
930         (cortex_a8_neon_bp_2cycle): Likewise.
931         (cortex_a8_neon_bp_3cycle): Likewise.
932         (cortex_a8_neon_ldr): Likewise.
933         (cortex_a8_neon_str): Likewise.
934         (cortex_a8_neon_vld1_1_2_regs): Likewise.
935         (cortex_a8_neon_vld1_3_4_regs): Likewise.
936         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
937         (cortex_a8_neon_vld2_4_regs): Likewise.
938         (cortex_a8_neon_vld3_vld4): Likewise.
939         (cortex_a8_neon_vld1_vld2_lane): Likewise.
940         (cortex_a8_neon_vld3_vld4_lane): Likewise.
941         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
942         (cortex_a8_neon_vst1_3_4_regs): Likewise.
943         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
944         (cortex_a8_neon_vst1_vst2_lane): Likewise.
945         (cortex_a8_neon_vst3_vst4_lane): Likewise.
946         (cortex_a8_neon_mcr): Likewise.
947         (cortex_a8_neon_mcr_2_mcrr): Likewise.
948         (cortex_a8_neon_mrc): Likewise.
949         (cortex_a8_neon_mrrc): Likewise.
951 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
953         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
954         (fp): New.
955         * config/aarch64/aarch64-simd.md (neon_type): Remove.
956         (aarch64_simd_dup<mode>): Add "type" attribute.
957         (aarch64_dup_lane<mode>): Likewise.
958         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
959         (*aarch64_simd_mov<mode>): Likewise.
960         (aarch64_simd_mov_from_<mode>low): Likewise.
961         (aarch64_simd_mov_from_<mode>high): Likewise.
962         (orn<mode>3): Likewise.
963         (bic<mode>3): Likewise.
964         (add<mode>3): Likewise.
965         (sub<mode>3): Likewise.
966         (mul<mode>3): Likewise.
967         (*aarch64_mul3_elt<mode>): Likewise.
968         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
969         (*aarch64_mul3_elt_to_128df): Likewise.
970         (*aarch64_mul3_elt_to_64v2df): Likewise.
971         (neg<mode>2): Likewise.
972         (abs<mode>2): Likewise.
973         (abd<mode>_3): Likewise.
974         (aba<mode>_3): Likewise.
975         (fabd<mode>_3): Likewise.
976         (*fabd_scalar<mode>3): Likewise.
977         (and<mode>3): Likewise.
978         (ior<mode>3): Likewise.
979         (xor<mode>3): Likewise.
980         (one_cmpl<mode>2): Likewise.
981         (aarch64_simd_vec_set<mode>): Likewise.
982         (aarch64_simd_lshr<mode>): Likewise.
983         (aarch64_simd_ashr<mode>): Likewise.
984         (aarch64_simd_imm_shl<mode>): Likewise.
985         (aarch64_simd_reg_sshl<mode): Likewise.
986         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
987         (aarch64_simd_reg_shl<mode>_signed): Likewise.
988         (aarch64_simd_vec_setv2di): Likewise.
989         (aarch64_simd_vec_set<mode>): Likewise.
990         (aarch64_mla<mode>): Likewise.
991         (*aarch64_mla_elt<mode>): Likewise.
992         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
993         (aarch64_mls<mode>): Likewise.
994         (*aarch64_mls_elt<mode>): Likewise.
995         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
996         (<su><maxmin><mode>3): Likewise.
997         (move_lo_quad_<mode>): Likewise.
998         (aarch64_simd_move_hi_quad_<mode>): Likewise.
999         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1000         (vec_pack_trunc_<mode>): Likewise.
1001         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1002         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1003         (*aarch64_<su>mlal_lo<mode>): Likewise.
1004         (*aarch64_<su>mlal_hi<mode>): Likewise.
1005         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1006         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1007         (*aarch64_<su>mlal<mode>): Likewise.
1008         (*aarch64_<su>mlsl<mode>): Likewise.
1009         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1010         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1011         (add<mode>3): Likewise.
1012         (sub<mode>3): Likewise.
1013         (mul<mode>3): Likewise.
1014         (div<mode>3): Likewise.
1015         (neg<mode>2): Likewise.
1016         (abs<mode>2): Likewise.
1017         (fma<mode>4): Likewise.
1018         (*aarch64_fma4_elt<mode>): Likewise.
1019         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1020         (*aarch64_fma4_elt_to_128df): Likewise.
1021         (*aarch64_fma4_elt_to_64v2df): Likewise.
1022         (fnma<mode>4): Likewise.
1023         (*aarch64_fnma4_elt<mode>): Likewise.
1024         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1025         (*aarch64_fnma4_elt_to_128df): Likewise.
1026         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1027         (<frint_pattern><mode>2): Likewise.
1028         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1029         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1030         (vec_unpacks_lo_v4sf): Likewise.
1031         (aarch64_float_extend_lo_v2df): Likewise.
1032         (vec_unpacks_hi_v4sf): Likewise.
1033         (aarch64_float_truncate_lo_v2sf): Likewise.
1034         (aarch64_float_truncate_hi_v4sf): Likewise.
1035         (aarch64_vmls<mode>): Likewise.
1036         (<su><maxmin><mode>3): Likewise.
1037         (<maxmin_uns><mode>3): Likewise.
1038         (reduc_<sur>plus_<mode>): Likewise.
1039         (reduc_<sur>plus_v2di): Likewise.
1040         (reduc_<sur>plus_v2si): Likewise.
1041         (reduc_<sur>plus_<mode>): Likewise.
1042         (aarch64_addpv4sf): Likewise.
1043         (clz<mode>2): Likewise.
1044         (reduc_<maxmin_uns>_<mode>): Likewise.
1045         (reduc_<maxmin_uns>_v2di): Likewise.
1046         (reduc_<maxmin_uns>_v2si): Likewise.
1047         (reduc_<maxmin_uns>_<mode>): Likewise.
1048         (reduc_<maxmin_uns>_v4sf): Likewise.
1049         (aarch64_simd_bsl<mode>_internal): Likewise.
1050         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1051         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1052         (aarch64_get_lane<mode>): Likewise.
1053         (*aarch64_combinez<mode>): Likewise.
1054         (aarch64_combine<mode>): Likewise.
1055         (aarch64_simd_combine<mode>): Likewise.
1056         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1057         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1058         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1059         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1060         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1061         (aarch64_<sur>h<addsub><mode>): Likewise.
1062         (aarch64_<sur><addsub>hn<mode>): Likewise.
1063         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1064         (aarch64_pmul<mode>): Likewise.
1065         (aarch64_<su_optab><optab><mode>): Likewise.
1066         (aarch64_<sur>qadd<mode>): Likewise.
1067         (aarch64_sqmovun<mode>): Likewise.
1068         (aarch64_<sur>qmovn<mode>): Likewise.
1069         (aarch64_s<optab><mode>): Likewise.
1070         (aarch64_sq<r>dmulh<mode>): Likewise.
1071         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1072         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1073         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1074         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1075         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1076         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1077         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1078         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1079         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1080         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1081         (aarch64_sqdmull<mode>): Likewise.
1082         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1083         (aarch64_sqdmull_n<mode>): Likewise.
1084         (aarch64_sqdmull2<mode>_internal): Likewise.
1085         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1086         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1087         (aarch64_<sur>shl<mode>): Likewise.
1088         (aarch64_<sur>q<r>shl<mode>
1089         (aarch64_<sur>shll_n<mode>): Likewise.
1090         (aarch64_<sur>shll2_n<mode>): Likewise.
1091         (aarch64_<sur>shr_n<mode>): Likewise.
1092         (aarch64_<sur>sra_n<mode>): Likewise.
1093         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1094         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1095         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1096         (aarch64_cm<optab><mode>): Likewise.
1097         (aarch64_cm<optab>di): Likewise.
1098         (aarch64_cm<optab><mode>): Likewise.
1099         (aarch64_cm<optab>di): Likewise.
1100         (aarch64_cmtst<mode>): Likewise.
1101         (aarch64_cmtstdi): Likewise.
1102         (aarch64_cm<optab><mode>): Likewise.
1103         (*aarch64_fac<optab><mode>): Likewise.
1104         (aarch64_addp<mode>): Likewise.
1105         (aarch64_addpdi): Likewise.
1106         (sqrt<mode>2): Likewise.
1107         (vec_load_lanesoi<mode>): Likewise.
1108         (vec_store_lanesoi<mode>): Likewise.
1109         (vec_load_lanesci<mode>): Likewise.
1110         (vec_store_lanesci<mode>): Likewise.
1111         (vec_load_lanesxi<mode>): Likewise.
1112         (vec_store_lanesxi<mode>): Likewise.
1113         (*aarch64_mov<mode>): Likewise.
1114         (aarch64_ld2<mode>_dreg): Likewise.
1115         (aarch64_ld2<mode>_dreg): Likewise.
1116         (aarch64_ld3<mode>_dreg): Likewise.
1117         (aarch64_ld3<mode>_dreg): Likewise.
1118         (aarch64_ld4<mode>_dreg): Likewise.
1119         (aarch64_ld4<mode>_dreg): Likewise.
1120         (aarch64_tbl1<mode>): Likewise.
1121         (aarch64_tbl2v16qi): Likewise.
1122         (aarch64_combinev16qi): Likewise.
1123         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1124         (aarch64_st2<mode>_dreg): Likewise.
1125         (aarch64_st2<mode>_dreg): Likewise.
1126         (aarch64_st3<mode>_dreg): Likewise.
1127         (aarch64_st3<mode>_dreg): Likewise.
1128         (aarch64_st4<mode>_dreg): Likewise.
1129         (aarch64_st4<mode>_dreg): Likewise.
1130         (*aarch64_simd_ld1r<mode>): Likewise.
1131         (aarch64_frecpe<mode>): Likewise.
1132         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1133         (aarch64_frecps<mode>): Likewise.
1135 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1137         * config/arm/iterators.md (V_elem_ch): New.
1138         (q): Likewise.
1139         (VQH_type): Likewise.
1140         * config/arm/arm.md (is_neon_type): New.
1141         (conds): Use is_neon_type.
1142         (anddi3_insn): Update type attribute.
1143         (xordi3_insn): Likewise.
1144         (one_cmpldi2): Likewise.
1145         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1146         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1147         (*movmisalign<mode>_neon_store): Likewise.
1148         (*movmisalign<mode>_neon_load): Likewise.
1149         (vec_set<mode>_internal): Likewise.
1150         (vec_set<mode>_internal): Likewise.
1151         (vec_setv2di_internal): Likewise.
1152         (vec_extract<mode>): Likewise.
1153         (vec_extract<mode>): Likewise.
1154         (vec_extractv2di): Likewise.
1155         (*add<mode>3_neon): Likewise.
1156         (adddi3_neon): Likewise.
1157         (*sub<mode>3_neon): Likewise.
1158         (subdi3_neon): Likewise.
1159         (fma<VCVTF:mode>4): Likewise.
1160         (fma<VCVTF:mode>4_intrinsic): Likewise.
1161         (*fmsub<VCVTF:mode>4): Likewise.
1162         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1163         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1164         (ior<mode>3): Likewise.
1165         (and<mode>3): Likewise.
1166         (orn<mode>3_neon): Likewise.
1167         (orndi3_neon): Likewise.
1168         (bic<mode>3_neon): Likewise.
1169         (bicdi3_neon): Likewise.
1170         (xor<mode>3): Likewise.
1171         (one_cmpl<mode>2): Likewise.
1172         (abs<mode>2): Likewise.
1173         (neg<mode>2): Likewise.
1174         (negdi2_neon): Likewise.
1175         (*umin<mode>3_neon): Likewise.
1176         (*umax<mode>3_neon): Likewise.
1177         (*smin<mode>3_neon): Likewise.
1178         (*smax<mode>3_neon): Likewise.
1179         (vashl<mode>3): Likewise.
1180         (vashr<mode>3_imm): Likewise.
1181         (vlshr<mode>3_imm): Likewise.
1182         (ashl<mode>3_signed): Likewise.
1183         (ashl<mode>3_unsigned): Likewise.
1184         (neon_load_count): Likewise.
1185         (ashldi3_neon_noclobber): Likewise.
1186         (ashldi3_neon): Likewise.
1187         (signed_shift_di3_neon): Likewise.
1188         (unsigned_shift_di3_neon): Likewise.
1189         (ashrdi3_neon_imm_noclobber): Likewise.
1190         (lshrdi3_neon_imm_noclobber): Likewise.
1191         (<shift>di3_neon): Likewise.
1192         (widen_ssum<mode>3): Likewise.
1193         (widen_usum<mode>3): Likewise.
1194         (quad_halves_<code>v4si): Likewise.
1195         (quad_halves_<code>v4sf): Likewise.
1196         (quad_halves_<code>v8hi): Likewise.
1197         (quad_halves_<code>v16qi): Likewise.
1198         (reduc_splus_v2di): Likewise.
1199         (neon_vpadd_internal<mode>): Likewise.
1200         (neon_vpsmin<mode>): Likewise.
1201         (neon_vpsmax<mode>): Likewise.
1202         (neon_vpumin<mode>): Likewise.
1203         (neon_vpumax<mode>): Likewise.
1204         (*ss_add<mode>_neon): Likewise.
1205         (*us_add<mode>_neon): Likewise.
1206         (*ss_sub<mode>_neon): Likewise.
1207         (*us_sub<mode>_neon): Likewise.
1208         (neon_vadd<mode>_unspec): Likewise.
1209         (neon_vaddl<mode>): Likewise.
1210         (neon_vaddw<mode>): Likewise.
1211         (neon_vhadd<mode>): Likewise.
1212         (neon_vqadd<mode>): Likewise.
1213         (neon_vaddhn<mode>): Likewise.
1214         (neon_vmul<mode>): Likewise.
1215         (neon_vfms<VCVTF:mode>): Likewise.
1216         (neon_vmlal<mode>): Likewise.
1217         (neon_vmls<mode>): Likewise.
1218         (neon_vmlsl<mode>): Likewise.
1219         (neon_vqdmulh<mode>): Likewise.
1220         (neon_vqdmlal<mode>): Likewise.
1221         (neon_vqdmlsl<mode>): Likewise.
1222         (neon_vmull<mode>): Likewise.
1223         (neon_vqdmull<mode>): Likewise.
1224         (neon_vsub<mode>_unspec): Likewise.
1225         (neon_vsubl<mode>): Likewise.
1226         (neon_vsubw<mode>): Likewise.
1227         (neon_vqsub<mode>): Likewise.
1228         (neon_vhsub<mode>): Likewise.
1229         (neon_vsubhn<mode>): Likewise.
1230         (neon_vceq<mode>): Likewise.
1231         (neon_vcge<mode>): Likewise.
1232         (neon_vcgeu<mode>): Likewise.
1233         (neon_vcgt<mode>): Likewise.
1234         (neon_vcgtu<mode>): Likewise.
1235         (neon_vcle<mode>): Likewise.
1236         (neon_vclt<mode>): Likewise.
1237         (neon_vcage<mode>): Likewise.
1238         (neon_vcagt<mode>): Likewise.
1239         (neon_vtst<mode>): Likewise.
1240         (neon_vabd<mode>): Likewise.
1241         (neon_vabdl<mode>): Likewise.
1242         (neon_vaba<mode>): Likewise.
1243         (neon_vabal<mode>): Likewise.
1244         (neon_vmax<mode>): Likewise.
1245         (neon_vmin<mode>): Likewise.
1246         (neon_vpaddl<mode>): Likewise.
1247         (neon_vpadal<mode>): Likewise.
1248         (neon_vpmax<mode>): Likewise.
1249         (neon_vpmin<mode>): Likewise.
1250         (neon_vrecps<mode>): Likewise.
1251         (neon_vrsqrts<mode>): Likewise.
1252         (neon_vqabs<mode>): Likewise.
1253         (neon_vqneg<mode>): Likewise.
1254         (neon_vcls<mode>): Likewise.
1255         (clz<mode>2): Likewise.
1256         (popcount<mode>2): Likewise.
1257         (neon_vrecpe<mode>): Likewise.
1258         (neon_vrsqrte<mode>): Likewise.
1259         (neon_vget_lane<mode>_sext_internal): Likewise.
1260         (neon_vget_lane<mode>_zext_internal): Likewise.
1261         (neon_vdup_n<mode>): Likewise.
1262         (neon_vdup_n<mode>): Likewise.
1263         (neon_vdup_nv2di): Likewise.
1264         (neon_vdup_lane<mode>_interal): Likewise.
1265         (*neon_vswp<mode>): Likewise.
1266         (neon_vcombine<mode>): Likewise.
1267         (float<mode><V_cvtto>2): Likewise.
1268         (floatuns<mode><V_cvtto>2): Likewise.
1269         (fix_trunc<mode><V_cvtto>2): Likewise.
1270         (fixuns_trunc<mode><V_cvtto>2
1271         (neon_vcvt<mode>): Likewise.
1272         (neon_vcvt<mode>): Likewise.
1273         (neon_vcvtv4sfv4hf): Likewise.
1274         (neon_vcvtv4hfv4sf): Likewise.
1275         (neon_vcvt_n<mode>): Likewise.
1276         (neon_vcvt_n<mode>): Likewise.
1277         (neon_vmovn<mode>): Likewise.
1278         (neon_vqmovn<mode>): Likewise.
1279         (neon_vqmovun<mode>): Likewise.
1280         (neon_vmovl<mode>): Likewise.
1281         (neon_vmul_lane<mode>): Likewise.
1282         (neon_vmul_lane<mode>): Likewise.
1283         (neon_vmull_lane<mode>): Likewise.
1284         (neon_vqdmull_lane<mode>): Likewise.
1285         (neon_vqdmulh_lane<mode>): Likewise.
1286         (neon_vqdmulh_lane<mode>): Likewise.
1287         (neon_vmla_lane<mode>): Likewise.
1288         (neon_vmla_lane<mode>): Likewise.
1289         (neon_vmlal_lane<mode>): Likewise.
1290         (neon_vqdmlal_lane<mode>): Likewise.
1291         (neon_vmls_lane<mode>): Likewise.
1292         (neon_vmls_lane<mode>): Likewise.
1293         (neon_vmlsl_lane<mode>): Likewise.
1294         (neon_vqdmlsl_lane<mode>): Likewise.
1295         (neon_vext<mode>): Likewise.
1296         (neon_vrev64<mode>): Likewise.
1297         (neon_vrev32<mode>): Likewise.
1298         (neon_vrev16<mode>): Likewise.
1299         (neon_vbsl<mode>_internal): Likewise.
1300         (neon_vshl<mode>): Likewise.
1301         (neon_vqshl<mode>): Likewise.
1302         (neon_vshr_n<mode>): Likewise.
1303         (neon_vshrn_n<mode>): Likewise.
1304         (neon_vqshrn_n<mode>): Likewise.
1305         (neon_vqshrun_n<mode>): Likewise.
1306         (neon_vshl_n<mode>): Likewise.
1307         (neon_vqshl_n<mode>): Likewise.
1308         (neon_vqshlu_n<mode>): Likewise.
1309         (neon_vshll_n<mode>): Likewise.
1310         (neon_vsra_n<mode>): Likewise.
1311         (neon_vsri_n<mode>): Likewise.
1312         (neon_vsli_n<mode>): Likewise.
1313         (neon_vtbl1v8qi): Likewise.
1314         (neon_vtbl2v8qi): Likewise.
1315         (neon_vtbl3v8qi): Likewise.
1316         (neon_vtbl4v8qi): Likewise.
1317         (neon_vtbl1v16qi): Likewise.
1318         (neon_vtbl2v16qi): Likewise.
1319         (neon_vcombinev16qi): Likewise.
1320         (neon_vtbx1v8qi): Likewise.
1321         (neon_vtbx2v8qi): Likewise.
1322         (neon_vtbx3v8qi): Likewise.
1323         (neon_vtbx4v8qi): Likewise.
1324         (*neon_vtrn<mode>_insn): Likewise.
1325         (*neon_vzip<mode>_insn): Likewise.
1326         (*neon_vuzp<mode>_insn): Likewise.
1327         (neon_vld1<mode>): Likewise.
1328         (neon_vld1_lane<mode>): Likewise.
1329         (neon_vld1_lane<mode>): Likewise.
1330         (neon_vld1_dup<mode>): Likewise.
1331         (neon_vld1_dup<mode>): Likewise.
1332         (neon_vld1_dupv2di): Likewise.
1333         (neon_vst1<mode>): Likewise.
1334         (neon_vst1_lane<mode>): Likewise.
1335         (neon_vst1_lane<mode>): Likewise.
1336         (neon_vld2<mode>): Likewise.
1337         (neon_vld2<mode>): Likewise.
1338         (neon_vld2_lane<mode>): Likewise.
1339         (neon_vld2_lane<mode>): Likewise.
1340         (neon_vld2_dup<mode>): Likewise.
1341         (neon_vst2<mode>): Likewise.
1342         (neon_vst2<mode>): Likewise.
1343         (neon_vst2_lane<mode>): Likewise.
1344         (neon_vst2_lane<mode>): Likewise.
1345         (neon_vld3<mode>): Likewise.
1346         (neon_vld3qa<mode>): Likewise.
1347         (neon_vld3qb<mode>): Likewise.
1348         (neon_vld3_lane<mode>): Likewise.
1349         (neon_vld3_lane<mode>): Likewise.
1350         (neon_vld3_dup<mode>): Likewise.
1351         (neon_vst3<mode>): Likewise.
1352         (neon_vst3qa<mode>): Likewise.
1353         (neon_vst3qb<mode>): Likewise.
1354         (neon_vst3_lane<mode>): Likewise.
1355         (neon_vst3_lane<mode>): Likewise.
1356         (neon_vld4<mode>): Likewise.
1357         (neon_vld4qa<mode>): Likewise.
1358         (neon_vld4qb<mode>): Likewise.
1359         (neon_vld4_lane<mode>): Likewise.
1360         (neon_vld4_lane<mode>): Likewise.
1361         (neon_vld4_dup<mode>): Likewise.
1362         (neon_vst4<mode>): Likewise.
1363         (neon_vst4qa<mode>): Likewise.
1364         (neon_vst4qb<mode>): Likewise.
1365         (neon_vst4_lane<mode>): Likewise.
1366         (neon_vst4_lane<mode>): Likewise.
1367         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1368         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1369         (neon_vec_<US>mult_lo_<mode>): Likewise.
1370         (neon_vec_<US>mult_hi_<mode>): Likewise.
1371         (neon_vec_<US>shiftl_<mode>): Likewise.
1372         (neon_unpack<US>_<mode>): Likewise.
1373         (neon_vec_<US>mult_<mode>): Likewise.
1374         (vec_pack_trunc_<mode>): Likewise.
1375         (neon_vec_pack_trunc_<mode>): Likewise.
1376         (neon_vabd<mode>_2): Likewise.
1377         (neon_vabd<mode>_3): Likewise.
1379 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1381         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1382         (load_pair): Update type attribute.
1383         (store_pair): Update type attribute.
1384         * config/aarch64/iterators.md (q): New.
1386 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1388         * config/arm/types.md: Add new types for Neon insns.
1390 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1391             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1392             Sergey Lega  <sergey.s.lega@intel.com>
1393             Anna Tikhonova  <anna.tikhonova@intel.com>
1394             Ilya Tocar  <ilya.tocar@intel.com>
1395             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1396             Ilya Verbin  <ilya.verbin@intel.com>
1397             Kirill Yukhin  <kirill.yukhin@intel.com>
1398             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1400         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1401         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1402         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1403         (rcp14<mode>): New.
1404         (srcp14<mode>): Ditto.
1405         (rsqrt14<mode>): Ditto.
1406         (rsqrt14<mode>): Ditto.
1407         (avx512f_vmscalef<mode>): Ditto.
1408         (avx512f_scalef<mode>): Ditto.
1409         (avx512f_getexp<mode>): Ditto.
1410         (avx512f_sgetexp<mode>): Ditto.
1411         (avx512f_fixupimm<mode>): Ditto.
1412         (avx512f_sfixupimm<mode>): Ditto.
1413         (avx512f_rndscale<mode>): Ditto.
1414         (*avx512er_exp2<mode>): Ditto.
1415         (*avx512er_rcp28<mode>): Ditto.
1416         (avx512er_rsqrt28<mode>): Ditto.
1417         (avx512f_getmant<mode>): Ditto.
1418         (avx512f_getmant<mode>): Ditto.
1419         (avx512f_rndscale<mode>): Fix formatting.
1421 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1423         * ipa-utils.h (ipa_edge_within_scc): Declare.
1424         * ipa-cp.c (edge_within_scc): Moved...
1425         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1427 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1428             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1429             Sergey Lega  <sergey.s.lega@intel.com>
1430             Anna Tikhonova  <anna.tikhonova@intel.com>
1431             Ilya Tocar  <ilya.tocar@intel.com>
1432             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1433             Ilya Verbin  <ilya.verbin@intel.com>
1434             Kirill Yukhin  <kirill.yukhin@intel.com>
1435             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1437         * config/i386/predicates.md (const_8_to_15_operand): New.
1438         (const_16_to_31_operand): Ditto.
1439         * config/i386/sse.md (V8FI): New.
1440         (V16FI): Ditto.
1441         (reduc_splus_v8df): Ditto.
1442         (reduc_splus_v16sf): Ditto.
1443         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1444         (vec_extract_hi_<mode>): Ditto.
1445         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1446         (vec_set_lo_<mode>): Ditto.
1447         (vec_set_hi_<mode>): Ditto.
1448         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1449         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1450         (avx512f_pshufd_1): Ditto.
1451         (avx512f_broadcast<mode>): Ditto.
1452         (avx512f_broadcast<mode>): Ditto.
1453         (define_split): Split vec_extract_lo into move.
1454         (ssequartermode): Ditto.
1455         (ssedoublemode): Extened with wider modes.
1456         (vec_extract_lo_<mode>): Ditto.
1458 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1459             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1460             Sergey Lega  <sergey.s.lega@intel.com>
1461             Anna Tikhonova  <anna.tikhonova@intel.com>
1462             Ilya Tocar  <ilya.tocar@intel.com>
1463             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1464             Ilya Verbin  <ilya.verbin@intel.com>
1465             Kirill Yukhin  <kirill.yukhin@intel.com>
1466             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1468         * config/i386/predicates.md (register_or_constm1_operand): New.
1469         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1470         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1471         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1472         UNSPEC_SCATTER_PREFETCH
1473         (VI48_512): New.
1474         (avx512f_ucmp<mode>3): Ditto.
1475         (avx512f_vternlog<mode>): Ditto.
1476         (avx512f_align<mode>): Ditto.
1477         (<shift_insn><mode>3): Ditto.
1478         (avx512f_<rotate>v<mode>): Ditto.
1479         (avx512f_<rotate><mode>): Ditto.
1480         (avx512f_eq<mode>3): Ditto.
1481         (avx512f_eq<mode>3_1): Ditto.
1482         (avx512f_gt<mode>3): Ditto.
1483         (avx512f_testm<mode>3): Ditto.
1484         (avx512f_testnm<mode>3): Ditto.
1485         (avx512pf_gatherpf<mode>): Ditto.
1486         (*avx512pf_gatherpf<mode>_mask): Ditto.
1487         (*avx512pf_gatherpf<mode>): Ditto.
1488         (avx512pf_scatterpf<mode>): Ditto.
1489         (*avx512pf_scatterpf<mode>_mask): Ditto.
1490         (*avx512pf_scatterpf<mode>): Ditto.
1491         (avx512f_vec_dup_gpr<mode>): Ditto.
1492         (clz<mode>2): Ditto.
1493         (conflict<mode>): Ditto.
1494         (REDUC_SMINMAX_MODE): Extened with wider modes.
1495         (reduc_<code>_<mode>): Ditto.
1496         (vlshr<mode>3): Ditto.
1497         (vashl<mode>3): Ditto.
1499 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1500             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1501             Sergey Lega  <sergey.s.lega@intel.com>
1502             Anna Tikhonova  <anna.tikhonova@intel.com>
1503             Ilya Tocar  <ilya.tocar@intel.com>
1504             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1505             Ilya Verbin  <ilya.verbin@intel.com>
1506             Kirill Yukhin  <kirill.yukhin@intel.com>
1507             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1509         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1510         UNSPEC_SCATTER.
1511         (VI48F_512): New.
1512         (avx512fmaskmode): Ditto.
1513         (bcstscalarsuff): Ditto.
1514         (avx512f_blendm<mode>): Ditto.
1515         (cmp_imm_predicate): Ditto.
1516         (avx512f_cmp<mode>3): Ditto.
1517         (avx512f_vec_dup<mode>): Ditto.
1518         (avx512f_vec_dup_mem<mode>): Ditto.
1519         (avx512f_vpermi2var<mode>3): Ditto.
1520         (avx512f_vpermt2var<mode>3): Ditto.
1521         (vec_init<mode>): Ditto.
1522         (avx512f_gathersi<mode>): Ditto.
1523         (*avx512f_gathersi<mode>): Ditto.
1524         (*avx512f_gathersi<mode>_2): Ditto.
1525         (avx512f_gatherdi<mode>): Ditto.
1526         (*avx512f_gatherdi<mode>): Ditto.
1527         (*avx512f_gatherdi<mode>_2): Ditto.
1528         (avx512f_scattersi<mode>): Ditto.
1529         (*avx512f_scattersi<mode>): Ditto.
1530         (avx512f_scatterdi<mode>): Ditto.
1531         (*avx512f_scatterdi<mode>): Ditto.
1532         (sseintprefix): Extened with wider modes.
1533         (VEC_GATHER_IDXSI): Ditto.
1534         (VEC_GATHER_IDXDI): Ditto.
1535         (VEC_GATHER_SRCDI): Ditto.
1537 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1538             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1540         * config/arm/t-aprofile: New file.
1541         * config.gcc: Handle --with-multilib-list option.
1543 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1545         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1546         the input for r2.
1548 2013-10-15  Richard Biener  <rguenther@suse.de>
1550         * tree-loop-distribution.c (build_empty_rdg): Inline into
1551         single user.
1552         (rdg_flag_vertex): Inline into single user.
1553         (rdg_flag_vertex_and_dependent): Likewise.
1554         (build_rdg_partition_for_vertex): Remove processed bitmap.
1555         (rdg_build_partitions): Simplify.
1557 2013-10-15  Richard Biener  <rguenther@suse.de>
1559         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1560         Restructure forwarding through conversions and copies to
1561         avoid performing copy-propagation the wrong way.  Adjust
1562         recursion invocations.
1563         (forward_propagate_addr_expr): Add argument stating if we
1564         are recursing from a single-use.
1565         (ssa_forward_propagate_and_combine): Adjust.
1567 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1569         * dumpfile.h (gcc::dump_manager): New class, to hold state
1570         relating to dumpfile management.
1571         (get_dump_file_name): Remove in favor of method of dump_manager.
1572         (dump_initialized_p): Likewise.
1573         (dump_start): Likewise.
1574         (dump_finish): Likewise.
1575         (dump_switch_p): Likewise.
1576         (dump_register): Likewise.
1577         (get_dump_file_info): Likewise.
1578         * context.c (gcc::context::context): Construct the dump_manager
1579         instance.
1580         * context.h (gcc::context::get_dumps): New.
1581         (gcc::context::m_dumps): New.
1582         * coverage.c (coverage_init): Port to dump_manager API.
1583         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1584         (extra_dump_files_in_use): Likewise.
1585         (extra_dump_files_alloced): Likewise.
1586         (gcc::dump_manager::dump_manager): New.
1587         (dump_register): Convert to...
1588         (gcc::dump_manager::dump_register): ...method, replacing
1589         function-static next_dump with m_next_dump field.
1590         (get_dump_file_info): Convert to...
1591         (gcc::dump_manager::get_dump_file_info): ...method.
1592         (get_dump_file_name): Convert to...
1593         (gcc::dump_manager::get_dump_file_name): ...method.
1594         (dump_start): Convert to...
1595         (gcc::dump_manager::dump_start): ...method.
1596         (dump_finish): Convert to...
1597         (gcc::dump_manager::dump_finish): ...method.
1598         (dump_begin): Replace body with...
1599         (gcc::dump_manager::dump_begin): ...new method.
1600         (dump_phase_enabled_p): Convert to...
1601         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1602         (dump_phase_enabled_p): Convert to...
1603         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1604         (dump_initialized_p):  Convert to...
1605         (gcc::dump_manager::dump_initialized_p): ...method.
1606         (dump_flag_name): Replace body with...
1607         (gcc::dump_manager::dump_flag_name): ...new method.
1608         (dump_enable_all): Convert to...
1609         (gcc::dump_manager::dump_enable_all): ...new method.
1610         (opt_info_enable_passes): Convert to...
1611         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1612         (dump_switch_p_1): Convert to...
1613         (gcc::dump_manager::dump_switch_p_1): ...new method.
1614         (dump_switch_p):  Convert to...
1615         (gcc::dump_manager::dump_switch_p): ...new method.
1616         (opt_info_switch_p): Port to dump_manager API.
1617         (enable_rtl_dump_file): Likewise.
1618         * opts-global.c (handle_common_deferred_options): Port to new
1619         dump_manager API.
1620         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1621         (pass_manager::register_one_dump_file): Likewise.
1622         (pass_manager::register_pass): Likewise.
1623         (pass_init_dump_file): Likewise.
1624         (pass_fini_dump_file): Likewise.
1625         * statistics.c (statistics_early_init): Likewise.
1627 2013-10-14  Richard Biener  <rguenther@suse.de>
1629         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1630         iterative_hash_canonical_type, gimple_canonical_type_hash,
1631         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1632         gimple_register_canonical_type, print_gimple_types_stats,
1633         free_gimple_type_tables): Move to lto/lto.c
1634         (gt-gimple.h): Do not include.
1635         * gimple.h (gimple_register_canonical_type,
1636         print_gimple_types_stats, free_gimple_type_tables): Remove.
1637         * Makefile.in (GTFILES): Remove gimple.c.
1639 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1641         PR target/58716
1642         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1643         scanning for msp430x targets.
1645 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1647         PR bootstrap/58509
1648         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1649         (registers_ok_for_ldd_peep): Move around.
1650         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1651         * config/sparc/sparc.md (widening peepholes): Use it.
1653 2013-10-14  Richard Biener  <rguenther@suse.de>
1655         PR middle-end/58712
1656         PR middle-end/55358
1657         * gimple.c (iterative_hash_canonical_type): Make sure to
1658         record the hash into the correct hashtable slot.
1660 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1662         PR rtl-optimization/58662
1663         * combine.c (try_combine): Take into account death nodes on I2 when
1664         splitting a PARALLEL of two independent SETs.  Fix dump message.
1666 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1668         PR target/51244
1669         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1670         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1671         * config/sh/t-sh (sh_treg_combine.o): New entry.
1672         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1673         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1674         (register_sh_passes): New function.  Register sh_treg_combine pass.
1675         (sh_option_override): Invoke it.
1676         (sh_canonicalize_comparison): Handle op0_preserve_value.
1677         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1678         opportunities.  Canonicalize branch condition.
1679         (nott): Allow only if pseudos can be created for non-SH2A.
1681 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1683         PR target/58690
1684         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1685         (ix86_expand_movmem): Replace copy_addr_to_reg with
1686         ix86_copy_addr_to_reg.
1687         (ix86_expand_setmem): Likewise.
1689 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1691         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1692         provided by ix86_fp_compare_mode instead of CCFPUmode.
1694 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1696         * config/aarch64/arm_neon.h
1697         (vtbx<1,3>_<psu>8): Fix register constriants.
1699 2013-10-11  Jeff Law  <law@redhat.com>
1701         PR tree-optimization/58640
1702         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1703         threading paths that cross over two loop entry points.
1705 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1707         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1708         handle vector float as well.
1709         (*vsx_le_perm_load_v4si): Likewise.
1710         (*vsx_le_perm_store_v2di): Likewise.
1711         (*vsx_le_perm_store_v4si): Likewise.
1713 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1715         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1716         directly to circumvent subtract from splat{31} workaround.
1717         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1718         prototype.
1719         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1720         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1721         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1722         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1723         separate big and little endian logic.
1724         (*altivec_vperm_<mode>_internal): New define_insn.
1725         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1726         separate big and little endian logic.
1727         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1728         (vec_permv16qi): Add little endian logic.
1730 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1732         * doc/extend.texi (returns_nonnull): Remove arguments.
1734 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1735             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1736             Sergey Lega  <sergey.s.lega@intel.com>
1737             Anna Tikhonova  <anna.tikhonova@intel.com>
1738             Ilya Tocar  <ilya.tocar@intel.com>
1739             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1740             Ilya Verbin  <ilya.verbin@intel.com>
1741             Kirill Yukhin  <kirill.yukhin@intel.com>
1742             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1744         * config/i386/sse.md (VI48F_256_512): New.
1745         (avx2_permvar<mode>): Change to ...
1746         (<avx2_avx512f>_permvar<mode>): This.
1748 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1749             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1750             Sergey Lega  <sergey.s.lega@intel.com>
1751             Anna Tikhonova  <anna.tikhonova@intel.com>
1752             Ilya Tocar  <ilya.tocar@intel.com>
1753             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1754             Ilya Verbin  <ilya.verbin@intel.com>
1755             Kirill Yukhin  <kirill.yukhin@intel.com>
1756             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1758         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1759         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1760         * config/i386/sse.md (VI4_AVX): New.
1761         (sf2simodelower): Ditto.
1762         (sse2_cvtps2dq): Change to ...
1763         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1765 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1766             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1767             Sergey Lega  <sergey.s.lega@intel.com>
1768             Anna Tikhonova  <anna.tikhonova@intel.com>
1769             Ilya Tocar  <ilya.tocar@intel.com>
1770             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1771             Ilya Verbin  <ilya.verbin@intel.com>
1772             Kirill Yukhin  <kirill.yukhin@intel.com>
1773             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1775         * config/i386/sse.md (V_512): New.
1776         (VI_512): Ditto.
1777         (vcond<V_512:mode><VF_512:mode>): Ditto.
1778         (vcond<V_512:mode><VI_512:mode>): Ditto.
1779         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1781 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1782             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1783             Sergey Lega  <sergey.s.lega@intel.com>
1784             Anna Tikhonova  <anna.tikhonova@intel.com>
1785             Ilya Tocar  <ilya.tocar@intel.com>
1786             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1787             Ilya Verbin  <ilya.verbin@intel.com>
1788             Kirill Yukhin  <kirill.yukhin@intel.com>
1789             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1791         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1792         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1793         (FMAMODE): Ditto.
1794         (fma<mode>4): Removed condition.
1795         (fms<mode>4): Ditto.
1796         (fnma<mode>4): Ditto.
1797         (fnms<mode>4): Ditto.
1798         (fma4i_fmadd_<mode>): Ditto.
1799         (*fma_fmadd_<mode>): Ditto.
1800         (*fma_fmsub_<mode>): Ditto.
1801         (*fma_fnmadd_<mode>): Ditto.
1802         (*fma_fnmsub_<mode>): Ditto.
1803         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1804         (*fma_fmaddsub_<mode>): Ditto.
1805         (*fma_fmsubadd_<mode>): Ditto.
1806         (*fmai_fmadd_<mode>): Ditto.
1807         (*fmai_fmsub_<mode>): Ditto.
1808         (*fmai_fnmadd_<mode>): Ditto.
1809         (*fmai_fnmsub_<mode>): Ditto.
1811 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1812             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1813             Sergey Lega  <sergey.s.lega@intel.com>
1814             Anna Tikhonova  <anna.tikhonova@intel.com>
1815             Ilya Tocar  <ilya.tocar@intel.com>
1816             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1817             Ilya Verbin  <ilya.verbin@intel.com>
1818             Kirill Yukhin  <kirill.yukhin@intel.com>
1819             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1821         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1822         (VI124_256): Changed to ...
1823         (VI124_256_48_512): This.
1824         (ssepackmode): Extended with wider modes.
1825         (<code><mode>3): Changed iterator.
1826         (*avx2_<code><mode>3): Ditto.
1827         (vec_pack_trunc_<mode>): Ditto.
1829 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1830             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1831             Sergey Lega  <sergey.s.lega@intel.com>
1832             Anna Tikhonova  <anna.tikhonova@intel.com>
1833             Ilya Tocar  <ilya.tocar@intel.com>
1834             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1835             Ilya Verbin  <ilya.verbin@intel.com>
1836             Kirill Yukhin  <kirill.yukhin@intel.com>
1837             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1839         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1840         (VI8F_256_512): Ditto.
1841         (abs<mode>2): Changed iterator.
1842         (avx2_perm<mode>): Changed to ...
1843         (<avx2_avx512f>_perm<mode>): This.
1844         (avx2_perm<mode>_1): Changed to ...
1845         (<avx2_avx512f>_perm<mode>_1): This.
1847 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1848             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1849             Sergey Lega  <sergey.s.lega@intel.com>
1850             Anna Tikhonova  <anna.tikhonova@intel.com>
1851             Ilya Tocar  <ilya.tocar@intel.com>
1852             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1853             Ilya Verbin  <ilya.verbin@intel.com>
1854             Kirill Yukhin  <kirill.yukhin@intel.com>
1855             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1857         * config/i386/sse.md (VI48_AVX512F): New.
1858         (VI48_AVX2): Changed to ...
1859         (VI48_AVX2_48_AVX512F): This.
1860         (avx2_ashrv<mode>): Changed to ...
1861         (<avx2_avx512f>_ashrv<mode>): This.
1862         (avx2_<shift_insn>v<mode>): Changed to ...
1863         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1865 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1866             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1867             Sergey Lega  <sergey.s.lega@intel.com>
1868             Anna Tikhonova  <anna.tikhonova@intel.com>
1869             Ilya Tocar  <ilya.tocar@intel.com>
1870             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1871             Ilya Verbin  <ilya.verbin@intel.com>
1872             Kirill Yukhin  <kirill.yukhin@intel.com>
1873             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1875         * config/i386/sse.md (VI4_AVX512F): New.
1876         (VI8_AVX2_AVX512F): Ditto.
1877         (mul<mode>3): Extended with wider modes.
1878         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1879         (mul<mode>3): Ditto.
1880         (vec_widen_<s>mult_odd_<mode>): Ditto.
1882 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1883             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1884             Sergey Lega  <sergey.s.lega@intel.com>
1885             Anna Tikhonova  <anna.tikhonova@intel.com>
1886             Ilya Tocar  <ilya.tocar@intel.com>
1887             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1888             Ilya Verbin  <ilya.verbin@intel.com>
1889             Kirill Yukhin  <kirill.yukhin@intel.com>
1890             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1892         * config/i386/sse.md (VI2_AVX512F): New.
1893         (VI124_AVX512F): Ditto.
1894         (sseunpackmode): Extended with wider modes.
1895         (sseunpackfltmode): Ditto.
1896         (vec_unpacks_float_hi_<mode>): Ditto.
1897         (vec_unpacks_float_lo_<mode>): Ditto.
1898         (vec_unpacku_float_hi_<mode>): Ditto.
1899         (vec_unpacku_float_lo_<mode>): Ditto.
1900         (vec_unpacks_lo_<mode>): Ditto.
1901         (vec_unpacks_hi_<mode>): Ditto.
1902         (vec_unpacku_lo_<mode>): Ditto.
1903         (vec_unpacku_hi_<mode>): Ditto.
1905 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1906             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1907             Sergey Lega  <sergey.s.lega@intel.com>
1908             Anna Tikhonova  <anna.tikhonova@intel.com>
1909             Ilya Tocar  <ilya.tocar@intel.com>
1910             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1911             Ilya Verbin  <ilya.verbin@intel.com>
1912             Kirill Yukhin  <kirill.yukhin@intel.com>
1913             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1915         * config/i386/i386.md (multdiv): New.
1916         (multdiv_mnemonic): Ditto.
1917         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1918         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1919         (<sse>_vmdiv<mode>3): Removed.
1921 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1922             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1923             Sergey Lega  <sergey.s.lega@intel.com>
1924             Anna Tikhonova  <anna.tikhonova@intel.com>
1925             Ilya Tocar  <ilya.tocar@intel.com>
1926             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1927             Ilya Verbin  <ilya.verbin@intel.com>
1928             Kirill Yukhin  <kirill.yukhin@intel.com>
1929             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1931         * config/i386/sse.md (V): Extended with wider modes.
1932         (VF2): Ditto.
1933         (ssehalfvecmode): Ditto.
1934         (i128): Ditto.
1935         (ssepackfltmode): Ditto.
1936         (avx_vec_concat<mode>): Ditto.
1937         (V_256_512): New iterator.
1938         (VF2_512_256): Ditto.
1939         (si2dfmode): New attribute.
1940         (si2dfmodelower): Ditto.
1941         (sf2dfmode): Ditto.
1942         (concat_tg_mode): Ditto.
1943         (floatv4siv4df2): Changed to ...
1944         (float<si2dfmodelower><mode>2): This.
1945         (avx_cvtps2pd256): Changed to ...
1946         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1947         (vec_pack_trunc_v4df): Changed to ...
1948         (vec_pack_trunc_<mode>): This.
1949         (avx_vpermil<mode>): Changed to ...
1950         (<sse2_avx_avx512f>_vpermil<mode>): This.
1951         (<fixsuffix>fix_truncv8dfv8si2): New.
1952         (vec_pack_sfix_trunc_v8df): Ditto.
1953         (avx512f_rndscale<mode>): Ditto.
1954         (avx512f_roundpd512): Ditto.
1955         (vec_pack_ufix_trunc_<mode>): Updated iterator.
1957 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1958             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1959             Sergey Lega  <sergey.s.lega@intel.com>
1960             Anna Tikhonova  <anna.tikhonova@intel.com>
1961             Ilya Tocar  <ilya.tocar@intel.com>
1962             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1963             Ilya Verbin  <ilya.verbin@intel.com>
1964             Kirill Yukhin  <kirill.yukhin@intel.com>
1965             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1967         * config/i386/i386.md (any_fix): New iterator.
1968         (fixsuffix): New attribute.
1969         * config/i386/sse.md (VF1): Extened with wider modes.
1970         (VI): Ditto.
1971         (VI_AVX2): Ditto.
1972         (VI8): Ditto.
1973         (sseintvecmodelower): Ditto.
1974         (ssescalarmode): Ditto.
1975         (ssescalarnum): Ditto.
1976         (VF1_128_256): New.
1977         (ssexmmmode): Ditto.
1978         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
1979         (<sse>_rcp<mode>2): Change iterator.
1980         (rsqrt<mode>2): Ditto.
1981         (<sse>_rsqrt<mode>2): Ditto.
1982         (avx2_vec_dup<mode>): Ditto.
1983         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
1984         (round<mode>2_sfix): Ditto.
1985         (avx2_pbroadcast<mode>): Ditto.
1986         (*andnot<mode>3): Handle XI mode.
1987         (*<code><mode>3): Ditto.
1988         (AVXTOSSEMODE): Removed.
1989         (avx_vpermil<mode>): Changed to ...
1990         (<sse2_avx_avx512f>_vpermil<mode>): This.
1992 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1993             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1994             Sergey Lega  <sergey.s.lega@intel.com>
1995             Anna Tikhonova  <anna.tikhonova@intel.com>
1996             Ilya Tocar  <ilya.tocar@intel.com>
1997             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1998             Ilya Verbin  <ilya.verbin@intel.com>
1999             Kirill Yukhin  <kirill.yukhin@intel.com>
2000             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2002         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2003         (<sse>_comi): Ditto.
2004         (<sse>_ucomi): Ditto.
2005         (sse_cvtss2siq_2): Ditto.
2006         (sse2_cvtsd2si): Ditto.
2007         (sse2_cvtsd2siq): Ditto.
2008         (sse2_cvttsd2si): Ditto.
2009         (sse2_cvttsd2siq): Ditto.
2010         (<shift_insn><mode>3): Ditto.
2011         (sse2_cvtsi2sdq): Update constraint and prefix.
2012         (sse_cvtsi2ss): Update prefix.
2013         (sse_cvtsi2ssq): Ditto.
2015 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2017         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2018         ignore internal calls.
2020 2013-10-11  Richard Biener  <rguenther@suse.de>
2022         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2023         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2024         TREE_OVERFLOW is set.
2026 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2028         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2030         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2031         * gimple.h: Likewise.
2033         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2035         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2036         macros.
2038         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2040         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2041         * configure: Regenerate.
2043 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2045         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2046         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2047         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2048         similarly pp_right_paren.
2049         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2050         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2051         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2052         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2053         clauses.
2054         (omp_declare_simd_clauses_equal,
2055         omp_remove_redundant_declare_simd_attrs): New functions.
2056         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2057         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2058         * tree.h (OMP_LOOP_CHECK): Define.
2059         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2060         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2061         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2062         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2063         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2064         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2065         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2066         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2067         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2068         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2069         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2070         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2071         (omp_remove_redundant_declare_simd_attrs): New prototype.
2072         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2073         GIMPLE_OMP_TEAMS): New codes.
2074         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2075         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2076         fields.
2077         (target_nesting_level): New variable.
2078         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2079         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2080         for collapse > 1 static schedule unless ordered.
2081         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2082         (determine_parallel_type): Adjust get_ws_args_for caller.
2083         (install_var_field): Handle mask & 4 for double indirection.
2084         (scan_sharing_clauses): Ignore shared clause on teams construct.
2085         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2086         (create_omp_child_function): If inside target or declare target
2087         constructs, set "omp declare target" attribute on the child function.
2088         (find_combined_for): New function.
2089         (scan_omp_parallel): Handle combined loops.
2090         (scan_omp_target, scan_omp_teams): New functions.
2091         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2092         restrictions and set ctx->cancellable for cancellable constructs.
2093         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2094         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2095         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2096         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2097         (omp_clause_aligned_alignment): New function.
2098         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2099         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2100         on teams constructs.  Handle user defined reductions and new
2101         OpenMP 4.0 clauses.
2102         (lower_reduction_clauses): Don't set placeholder to address of ref
2103         if it has already the right type.
2104         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2105         (expand_parallel_call): Use the new non-_start suffixed builtins,
2106         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2107         and GOMP_parallel_end after the call.
2108         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2109         (expand_omp_for_init_counts): Handle combined loops.
2110         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2111         loops.
2112         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2113         end of cancellable loops.
2114         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2115         Likewise.  Handle collapse > 1 loops.
2116         (expand_omp_simd): Handle combined loops.
2117         (expand_omp_for): Add inner_stmt argument, adjust callers of
2118         expand_omp_for* functions, use expand_omp_for_static*chunk even
2119         for collapse > 1 unless ordered.
2120         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2121         of cancellable sections.
2122         (expand_omp_single): Remove need_barrier variable, just rely on
2123         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2124         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2125         (expand_omp_atomic_load, expand_omp_atomic_store,
2126         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2127         (expand_omp_target): New function.
2128         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2129         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2130         (build_omp_regions_1): Immediately close region for
2131         GF_OMP_TARGET_KIND_UPDATE.
2132         (maybe_add_implicit_barrier_cancel): New function.
2133         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2134         cancellation.
2135         (lower_omp_single): Likewise.  Add clobber after the barrier.
2136         (lower_omp_taskgroup): New function.
2137         (lower_omp_for): Handle combined loops.  Adjust
2138         lower_rec_input_clauses caller.  Handle cancellation.
2139         (lower_depend_clauses): New function.
2140         (lower_omp_taskreg): Lower depend clauses.  Adjust
2141         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2142         cancellation.
2143         (lower_omp_target, lower_omp_teams): New functions.
2144         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2145         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2146         and GOMP_cancellation_point calls.
2147         (lower_omp): Fold stmts inside of target region.
2148         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2149         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2150         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2151         (BT_FN_VOID_OMPFN_PTR_UINT,
2152         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2153         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2154         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2155         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2156         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2157         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2158         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2159         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2160         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2161         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2162         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2163         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2164         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2165         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2166         BUILT_IN_GOMP_PARALLEL_END.
2167         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2168         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2169         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2170         GF_OMP_FOR_KIND_DISTRIBUTE.
2171         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2172         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2173         (dump_gimple_omp_return): Print lhs if it has any.
2174         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2175         gimple_omp_atomic_seq_cst_p.
2176         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2177         and GIMPLE_OMP_TEAMS.
2178         * langhooks.c (lhd_omp_mappable_type): New function.
2179         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2180         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2181         hook.
2182         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2183         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2184         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2185         (struct gimplify_omp_ctx): Add combined_loop field.
2186         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2187         on stmts inside of target region.
2188         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2189         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2190         ORT_TARGET and ORT_TARGET_DATA.
2191         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2192         Handle GOVD_MAP.
2193         (omp_notice_threadprivate_variable): Complain about threadprivate
2194         variables in target region.
2195         (omp_notice_variable): Complain about vars with non-mappable type
2196         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2197         (omp_check_private): Ignore ORT_TARGET* regions.
2198         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2199         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2200         (find_combined_omp_for): New function.
2201         (gimplify_omp_for): Handle gimplification of combined loops.
2202         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2203         OMP_TEAMS.
2204         (gimplify_omp_target_update): New function.
2205         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2206         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2207         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2208         (gimplify_body): If fndecl has "omp declare target" attribute, add
2209         implicit ORT_TARGET context around it.
2210         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2211         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2212         * tree-nested.c (convert_nonlocal_reference_stmt,
2213         convert_local_reference_stmt, convert_gimple_call): Handle
2214         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2215         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2216         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2217         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2218         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2219         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2220         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2221         BUILT_IN_GOMP_LOOP_END_CANCEL,
2222         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2223         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2224         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2225         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2226         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2227         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2228         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2229         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2230         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2231         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2232         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2233         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2234         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2235         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2236         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2237         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2238         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2239         gimple_build_omp_teams): New functions.
2240         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2241         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2242         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2243         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2244         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2245         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2246         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2247         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2248         GF_OMP_ATOMIC_SEQ_CST): New.
2249         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2250         gimple_build_omp_teams): New prototypes.
2251         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2252         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2253         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2254         GIMPLE_OMP_SINGLE as end of range.
2255         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2256         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2257         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2258         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2259         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2260         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2261         gimple_omp_target_kind, gimple_omp_target_set_kind,
2262         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2263         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2264         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2265         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2266         gimple_omp_teams_set_clauses): New inlines.
2267         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2268         and GIMPLE_OMP_TASKGROUP.
2269         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2270         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2271         enum omp_clause_proc_bind_kind): New.
2272         (union omp_clause_subcode): Add depend_kind, map_kind and
2273         proc_bind_kind fields.
2274         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2275         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2276         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2277         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2278         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2280 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2282         * predict.c (tree_estimate_probability): Add new parameter
2283         for estimate_bb_frequencies.
2284         (estimate_bb_frequencies): Add new parameter to force estimation.
2285         (rebuild_frequencies): When max frequency in function is small,
2286         recompute counts from frequencies.
2287         * predict.h (estimate_bb_frequencies): New parameter.
2289 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2291         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2292         optimizations are disabled.
2294 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2296         * coverage.c (coverage_finish): Fix leak of da_file_name.
2298 2013-10-10  Jan Hubicka  <jh@suse.cz>
2300         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2301         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2302         Bobcat and generic.
2304 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2306         PR middle-end/58670
2307         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2308         if any labels are in FALLTHRU_BB, use a special label emitted
2309         immediately after the asm goto insn rather than label_rtx
2310         of the LABEL_DECL.
2311         (expand_asm_stmt): Adjust caller.
2312         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2313         edge if the last insn in predecessor is a jump with single successor,
2314         but it isn't simplejump_p.
2316 2013-10-10  Richard Biener  <rguenther@suse.de>
2318         PR tree-optimization/58656
2319         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2321 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2323         * gimplify.c: Include expr.h and tm_p.h for targets with special
2324         va-arg padding requirements.
2326 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2328         * tree-flow.h: Move some prototypes to gimple.h.
2329         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2330         * gimple.h: Relocate some prototypes from tree-flow.h
2331         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2332         Move to gimplify.c.
2333         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2334         (build_va_arg_indirect_ref): Relocate and make static.
2335         (std_gimplify_va_arg_expr): Relocate here.
2336         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2337         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2339 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2341         * doc/md.texi: Document the mnemonic attribute.
2343 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2345         PR target/57377
2346         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2347         (set_attr_alternative x ...) when searching for user defined
2348         mnemonic attribute.
2350 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2352         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2354 2013-10-09  Easwaran Raman <eraman@google.com>
2356         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2357         * cfgexpand.c (defer_stack_allocation): ...use here
2358         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2360 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2362         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2363         (operand_equal_for_value_replacement): New function, extracted from
2364         value_replacement and enhanced to catch more cases.
2365         (value_replacement): Use operand_equal_for_value_replacement.
2367 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2369         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2370         get_max_loop_iterations.
2372 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2374         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2375         Place comment above function.
2377 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2379         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2380         are not related to tree-cfg.c.
2381         * tree-ssa-address.h: New file.  Relocate prototypes.
2382         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2383         (addr_for_mem_ref): New.  Combine call to get_address_description and
2384         return addr_for_mem_ref.
2385         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2386         * tree-ssa-live.h: Adjust prototypes.
2387         * passes.c: Include tree-ssa-live.h.
2388         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2389         * graphite.c (graphite_transform_loops): Make static.
2390         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2391         make_pass_graphite, pass_data_graphite_transforms,
2392         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2393         * ipa-pure-const.c (warn_function_noreturn): Make static.
2394         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2395         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2396         Relocate from tree-cfg.c
2397         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2398         static.
2399         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2400         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2401         Move to ipa-pure-const.c.
2402         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2403         Relocate from tree-optimize.c.
2404         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2405         make_pass_fixup_cfg): Move to tree-cfg.c.
2406         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2407         Relocate some prototypes.
2408         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2409         * tree-dump.c (dump_function_to_file): Remove prototype.
2410         Add tree-flow.h to the include file.
2411         * tree-dump.h: Remove prototype.
2412         * tree-parloops.h: New File.  Add prototypes.
2413         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2414         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2415         from tree-ssa-loop.c.
2416         * tree-predcom.c (run_tree_predictive_commoning,
2417         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2418         Relocate here from tree-ssa-loop.c.
2419         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2420         ssa_name_values.release ().
2421         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2422         (ssa_name_values): Relocate from tree-flow.h.
2423         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2424         * tree-ssa-loop.c (run_tree_predictive_commoning,
2425         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2426         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2427         make_pass_graphite, pass_data_graphite_transforms,
2428         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2429         tree_parallelize_loops, pass_data_parallelize_loops,
2430         make_pass_parallelize_loops): Move to other files.
2431         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2432         moved here.
2433         * tree.h: Remove prototypes from tree-address.c.
2435 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2437         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2438         (struct int_tree_map): Move to tree-hasher.h
2439         (SCALE, LABEL, PERCENT): Move to gimple.h
2440         * tree-flow-inline.h: Delete.  Move functions to other files.
2441         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2442         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2443         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2444         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2445         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2446         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2447         * tree-hasher.h: Don't include tree-flow.h.
2448         (struct int_tree_map): Relocate from tree-flow.h.
2449         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2450         tree-flow-inline.h and make static.
2451         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2452         tree-flow-inline.h.
2453         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2454         tree-flow-inline.h and make static.
2455         * tree.h (is_global_var, may_be_aliased): Relocate from
2456         tree-flow-inline.h.
2457         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2458         * value-prof.c: No longer include tree-flow-inline.h.
2459         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2461 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2463         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2464         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2465         (enum move_pos): Move to tree-ssa-loop-im.h
2466         * cfgloop.h: Move some prototypes.
2467         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2468         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2469         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2470         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2471         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2472         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2473         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2474         (enum move_pos): Relocate here.
2475         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2476         tree-ssa-loop.c.
2477         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2478         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2479         make_pass_lim): Relocate here from tree-ssa-loop.c.
2480         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2481         tree-ssa-loop.c.
2482         (loop_edge_to_cancel, unloop_loops): Make static.
2483         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2484         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2485         (tree_complete_unroll, gate_tree_complete_unroll,
2486         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2487         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2488         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2489         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2490         (stmt_invariant_in_loop_p): Remove unused function.
2491         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2492         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2493         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2494         (gcov_type_to_double_int): Move to cfgloop.h.
2495         (double_int_cmp, bound_index,
2496         estimate_numbers_of_iterations_loop): Make static.
2497         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2498         (max_loop_iterations): Factor out get_max_loop_iterations.
2499         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2500         cfgloop.c.
2501         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2502         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2503         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2504         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2505         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2506         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2507         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2508         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2509         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2510         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2511         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2512         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2513         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2514         pass_data_complete_unroll, make_pass_complete_unroll,
2515         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2516         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2517         tree-ssa-loop-ivcanon.c.
2518         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2519         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2520         tree-ssa-loop-prefetch.c.
2521         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2522         tree-ssa-loop-im.c.
2523         (get_lsm_tmp_name): Relocate and add suffix parameter.
2524         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2525         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2526         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2527         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2528         (get_estimated_loop_iterations): Factor out accessor from
2529         estimated_loop_iterations in tree-ssa-loop-niter.c.
2530         (get_max_loop_iterations): Factor out accessor from
2531         _max_loop_iterations in tree-ssa-niter.c.
2532         * loop-unroll.c (decide_unroll_constant_iterations,
2533         decide_unroll_runtime_iterations, decide_peel_simple,
2534         decide_unroll_stupid): Use new get_* accessors.
2536 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2538         PR tree-optimization/20318
2539         * doc/extend.texi (returns_nonnull): New function attribute.
2540         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2541         attribute.
2542         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2543         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2545 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2547         PR middle-end/58570
2548         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2549         false if both components are bitfields.
2551 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2553         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2554         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2555         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2556         * config/aarch64/aarch64.h
2557         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2558         * config/aarch64/aarch64-simd-builtins.def
2559         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2561 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2563         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2565 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2567         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2569 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2571         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2572         (vneg_f64): New intrinsic.
2573         (vneg_s8): Asm replaced with C.
2574         (vneg_s16): Likewise.
2575         (vneg_s32): Likewise.
2576         (vneg_s64): New intrinsic.
2577         (vnegq_f32): Asm replaced with C.
2578         (vnegq_f64): Likewise.
2579         (vnegq_s8): Likewise.
2580         (vnegq_s16): Likewise.
2581         (vnegq_s32): Likewise.
2582         (vnegq_s64): Likewise.
2584 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2586         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2588 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2590         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2591         packed stack special handling.
2592         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2593         back to fixed stack slots for FPRs saved due to stdarg.
2595 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2597         * config/s390/s390.c (s390_frame_info): Restructure function.
2599 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2601         * config/s390/s390.c (struct s390_frame_layout): New field
2602         gpr_save_slots.
2603         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2604         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2605         regs_ever_clobbered to char*.
2606         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2607         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2608         (s390_register_info): Enable FPR save slots.  Move/Copy some
2609         functionality into ...
2610         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2611         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2612         function.
2613         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2614         not imply a stack frame.
2615         (s390_init_frame_layout): Remove variable clobbered_regs.
2616         (s390_update_register_info): Remove function.
2617         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2618         cannot be used for register renaming.
2619         (s390_hard_regno_scratch_ok): New function.
2620         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2621         (s390_initial_elimination_offset): Change offset calculation of
2622         the return address pointer.
2623         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2624         (restore_gprs): Set frame related flag.
2625         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2626         (s390_emit_prologue): Call s390_register_info instead of
2627         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2628         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2629         (s390_optimize_prologue): Call s390_optimize_register_info.
2630         Try to remove also FPR slot save/restore INSNs.  Remove frame
2631         related flags from restore INSNs.
2633 2013-10-08  DJ Delorie  <dj@redhat.com>
2635         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2636         (movhi): Likewise.
2638         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2639         avoid conflict with the MI use of %c.
2640         * config/rl78/rl78-real.md: change %c to %C throughout.
2641         * config/rl78/rl78-virt.md: Likewise.
2643 2013-10-08  Jan Hubicka  <jh@suse.cz>
2645         * config/i386/i386.c (ix86_option_override_internal): Switch
2646         to SSE math for -ffast-math when target ISA supports SSE2.
2648 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2650         * tree-flow.h: Remove some prototypes.
2651         * tree.h: Remove some protypes, add a couple.
2652         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2653         using_eh_for_cleanups_p): Add interface routines for front ends.
2654         * tree-eh.h: New file.  Add protoptyes.
2655         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2656         (add_stmt_to_eh_lp_fn): Make static.
2657         (lower_try_finally): Use new using_eh_for_cleanups_p.
2658         * emit-rtl.c: Include tree-eh.h.
2659         * gimple.h: Include tree-eh.h.
2661 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2663         PR tree-optimization/58480
2664         * tree-vrp.c (infer_nonnull_range): New function.
2665         (infer_value_range): Call infer_nonnull_range.
2667 2013-10-08  Dehao Chen  <dehao@google.com>
2669         PR tree-optimization/58619
2670         * tree-inline.c (copy_phis_for_bb): Combine location data
2671         only if non-null.
2673 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2675         PR target/58423
2676         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2677         RTX_FRAME_RELATED_P on INSN.
2679 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2681         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2682         (altivec_expand_vec_perm_const): Call it.
2684 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2686         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2687         sequences for LE VSX loads and stores at expand time.
2688         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2689         prototype.
2690         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2691         (rs6000_gen_le_vsx_permute): New.
2692         (rs6000_gen_le_vsx_load): New.
2693         (rs6000_gen_le_vsx_store): New.
2694         (rs6000_gen_le_vsx_move): New.
2695         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2696         (*vsx_le_perm_load_v4si): New.
2697         (*vsx_le_perm_load_v8hi): New.
2698         (*vsx_le_perm_load_v16qi): New.
2699         (*vsx_le_perm_store_v2di): New.
2700         (*vsx_le_perm_store_v4si): New.
2701         (*vsx_le_perm_store_v8hi): New.
2702         (*vsx_le_perm_store_v16qi): New.
2703         (*vsx_xxpermdi2_le_<mode>): New.
2704         (*vsx_xxpermdi4_le_<mode>): New.
2705         (*vsx_xxpermdi8_le_V8HI): New.
2706         (*vsx_xxpermdi16_le_V16QI): New.
2707         (*vsx_lxvd2x2_le_<mode>): New.
2708         (*vsx_lxvd2x4_le_<mode>): New.
2709         (*vsx_lxvd2x8_le_V8HI): New.
2710         (*vsx_lxvd2x16_le_V16QI): New.
2711         (*vsx_stxvd2x2_le_<mode>): New.
2712         (*vsx_stxvd2x4_le_<mode>): New.
2713         (*vsx_stxvd2x8_le_V8HI): New.
2714         (*vsx_stxvd2x16_le_V16QI): New.
2716 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2718         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2720 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2722         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2723         loop to work also for 31bit ABI.
2724         Save the stack pointer for frame_size > 0.
2726 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2728         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2729         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2730         constraint letters from expanders.
2731         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2732         retry count to general_operand.
2733         ("tabort"): Give operand 0 a mode.
2734         ("tabort_1"): Add mode and constraint letter for operand 0.
2735         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2737 2013-10-04  Jeff Law  <law@redhat.com>
2739         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2741         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2742         out of ...
2743         (thread_across_edge): Here.  Call it.
2745 2013-10-04  Cary Coutant  <ccoutant@google.com>
2747         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2748         discarding a location list expression (or a piece of one).
2750 2013-10-03  Jan Hubicka  <jh@suse.cz>
2752         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2753         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2754         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2756 2013-10-03  Jan Hubicka  <jh@suse.cz>
2758         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2759         accumulate-outgoing-args when producing unwind info.
2761 2013-10-03  Wei Mi  <wmi@google.com>
2763         * lra-constraints.c (insert_move_for_subreg): New function
2764         extracted from simplify_operand_subreg.
2765         (simplify_operand_subreg): Add reload for paradoxical subreg.
2767 2013-10-03  Rong Xu  <xur@google.com>
2769         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2770         the candidate of builtin_expect such that we should fix the
2771         size/time estimation.
2772         (estimate_function_body_sizes): Do the acutally size/time fix-up
2773         for builtin_expect.
2775 2013-10-03  Rong Xu  <xur@google.com>
2777         * predict.c (tree_predict_by_opcode): Get the probability
2778         for builtin_expect from param builtin_expect_probability.
2779         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2780         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2781         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2783 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2785         PR c++/19476
2786         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2787         * calls.c (alloca_call_p): Use get_callee_fndecl.
2788         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2789         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2790         Likewise.
2791         (vrp_visit_stmt): Remove duplicated code.
2793 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2795         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2796         ceildf2, btruncdf2, instead of vsx_* name.
2798         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2799         iterators to only do V2DF and V4SF here.  Move the DF code to
2800         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2801         just 'v' since only vector operations are handled with these insns
2802         after moving the DF support to rs6000.md.
2803         (vsx_sub<mode>3): Likewise.
2804         (vsx_mul<mode>3): Likewise.
2805         (vsx_div<mode>3): Likewise.
2806         (vsx_fre<mode>2): Likewise.
2807         (vsx_neg<mode>2): Likewise.
2808         (vsx_abs<mode>2): Likewise.
2809         (vsx_nabs<mode>2): Likewise.
2810         (vsx_smax<mode>3): Likewise.
2811         (vsx_smin<mode>3): Likewise.
2812         (vsx_sqrt<mode>2): Likewise.
2813         (vsx_rsqrte<mode>2): Likewise.
2814         (vsx_fms<mode>4): Likewise.
2815         (vsx_nfma<mode>4): Likewise.
2816         (vsx_copysign<mode>3): Likewise.
2817         (vsx_btrunc<mode>2): Likewise.
2818         (vsx_floor<mode>2): Likewise.
2819         (vsx_ceil<mode>2): Likewise.
2820         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2821         (vsx_sminsf3): Likewise.
2822         (vsx_fmadf4): Likewise.
2823         (vsx_fmsdf4): Likewise.
2824         (vsx_nfmadf4): Likewise.
2825         (vsx_nfmsdf4): Likewise.
2826         (vsx_cmpdf_internal1): Likewise.
2828         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2829         simpler to select whether a target has SPE or traditional floating
2830         point support in iterators.
2831         (TARGET_DF_SPE): Likewise.
2832         (TARGET_SF_FPR): Likewise.
2833         (TARGET_DF_FPR): Likewise.
2834         (TARGET_SF_INSN): Macros to say whether floating point support
2835         exists for a given operation for expanders.
2836         (TARGET_DF_INSN): Likewise.
2838         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2839         combining of SF/DF mode operations, using both traditional and VSX
2840         registers.
2841         (Fvsx): Likewise.
2842         (Ff): Likewise.
2843         (Fv): Likewise.
2844         (Fs): Likewise.
2845         (Ffre): Likewise.
2846         (FFRE): Likewise.
2847         (abs<mode>2): Combine SF/DF modes using traditional floating point
2848         instructions.  Add support for using the upper DF registers with
2849         VSX support, and SF registers with power8-vector support.  Update
2850         expanders for operations supported by both the SPE and traditional
2851         floating point units.
2852         (abs<mode>2_fpr): Likewise.
2853         (nabs<mode>2): Likewise.
2854         (nabs<mode>2_fpr): Likewise.
2855         (neg<mode>2): Likewise.
2856         (neg<mode>2_fpr): Likewise.
2857         (add<mode>3): Likewise.
2858         (add<mode>3_fpr): Likewise.
2859         (sub<mode>3): Likewise.
2860         (sub<mode>3_fpr): Likewise.
2861         (mul<mode>3): Likewise.
2862         (mul<mode>3_fpr): Likewise.
2863         (div<mode>3): Likewise.
2864         (div<mode>3_fpr): Likewise.
2865         (sqrt<mode>3): Likewise.
2866         (sqrt<mode>3_fpr): Likewise.
2867         (fre<Fs>): Likewise.
2868         (rsqrt<mode>2): Likewise.
2869         (cmp<mode>_fpr): Likewise.
2870         (smax<mode>3): Likewise.
2871         (smin<mode>3): Likewise.
2872         (smax<mode>3_vsx): Likewise.
2873         (smin<mode>3_vsx): Likewise.
2874         (negsf2): Delete SF operations that are merged with DF.
2875         (abssf2): Likewise.
2876         (addsf3): Likewise.
2877         (subsf3): Likewise.
2878         (mulsf3): Likewise.
2879         (divsf3): Likewise.
2880         (fres): Likewise.
2881         (fmasf4_fpr): Likewise.
2882         (fmssf4_fpr): Likewise.
2883         (nfmasf4_fpr): Likewise.
2884         (nfmssf4_fpr): Likewise.
2885         (sqrtsf2): Likewise.
2886         (rsqrtsf_internal1): Likewise.
2887         (smaxsf3): Likewise.
2888         (sminsf3): Likewise.
2889         (cmpsf_internal1): Likewise.
2890         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2891         (negdf2): Delete DF operations that are merged with SF.
2892         (absdf2): Likewise.
2893         (nabsdf2): Likewise.
2894         (adddf3): Likewise.
2895         (subdf3): Likewise.
2896         (muldf3): Likewise.
2897         (divdf3): Likewise.
2898         (fred): Likewise.
2899         (rsqrtdf_internal1): Likewise.
2900         (fmadf4_fpr): Likewise.
2901         (fmsdf4_fpr): Likewise.
2902         (nfmadf4_fpr): Likewise.
2903         (nfmsdf4_fpr): Likewise.
2904         (sqrtdf2): Likewise.
2905         (smaxdf3): Likewise.
2906         (smindf3): Likewise.
2907         (cmpdf_internal1): Likewise.
2908         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2909         (btrunc<mode>2): Delete separate expander, and combine with the
2910         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2911         (btrunc<mode>2_fpr): Likewise.
2912         (ceil<mode>2): Likewise.
2913         (ceil<mode>2_fpr): Likewise.
2914         (floor<mode>2): Likewise.
2915         (floor<mode>2_fpr): Likewise.
2916         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2917         Add support for using the upper registers with VSX and
2918         power8-vector.  Move insns to be closer to the define_expands. On
2919         VSX systems, prefer the traditional form of FMA over the VSX
2920         version, since the traditional form allows the target not to
2921         overlap with the inputs.
2922         (fms<mode>4_fpr): Likewise.
2923         (nfma<mode>4_fpr): Likewise.
2924         (nfms<mode>4_fpr): Likewise.
2926 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2927             Richard Earnshaw  <richard.earnshaw@arm.com>
2929         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2930         (struct mult_cost_table): Likewise.
2931         (struct mem_cost_table): Likewise.
2932         (struct fp_cost_table): Likewise.
2933         (struct vector_cost_table): Likewise.
2934         (cpu_cost_table): Likewise.
2935         * config/arm/arm.opt (mold-rts-costs): New option.
2936         (mnew-generic-costs): Likewise.
2937         * config/arm/arm.c (generic_extra_costs): New table.
2938         (cortexa15_extra_costs): Likewise.
2939         (arm_slowmul_tune): Use NULL as new costs.
2940         (arm_fastmul_tune): Likewise.
2941         (arm_strongarm_tune): Likewise.
2942         (arm_xscale_tune): Likewise.
2943         (arm_9e_tune): Likewise.
2944         (arm_v6t2_tune): Likewise.
2945         (arm_cortex_a5_tune): Likewise.
2946         (arm_cortex_a9_tune): Likewise.
2947         (arm_v6m_tune): Likewise.
2948         (arm_fa726te_tune): Likewise.
2949         (arm_cortex_a15_tune): Use cortex15_extra_costs.
2950         (arm_cortex_tune): Use generict_extra_costs.
2951         (shifter_op_p): New function.
2952         (arm_unspec_cost): Likewise.
2953         (LIBCALL_COST): Define.
2954         (arm_new_rtx_costs): New function.
2955         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
2956         table is available. Use old costs otherwise unless mnew-generic-costs
2957         is specified.
2958         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
2959         (cpu_cost_table): Declare.
2961 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
2963         PR target/58460
2964         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
2965         (*subs_mul_imm_<mode>)
2966         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
2967         (*sub_<shift>_<mode>)
2968         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
2969         Remove k constraint.
2971 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
2973         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
2974         code.
2975         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
2977 2013-10-02  Teresa Johnson  <tejohnson@google.com>
2979         * predict.c (probably_never_executed): New function.
2980         (probably_never_executed_bb_p): Invoke probably_never_executed.
2981         (probably_never_executed_edge_p): Ditto.
2982         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
2983         Treat profile insanities conservatively.
2985 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
2987         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
2989 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
2991         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
2992         use it.  Use smaller increase for scratch.  Don't increase reject
2993         for early clobber scratch.
2994         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
2995         setting eliminated regs except setting fp from hfp.
2996         (lra_eliminate): Check lra_insn_recog_data on NULL.
2998 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3000         PR target/58587
3001         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3002         setting -mvsx-timode by default until the underlying problem is fixed.
3003         (RS6000_CPU, power7 defaults): Likewise.
3005 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3007         * config/x-linux (host-linux.o): Remove header dependencies.
3008         Use $(COMPILE) and $(POSTCOMPILE).
3009         * config/t-linux-android (linux-android.o): Ditto.
3011 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3013         * Makefile.in (expmed.o-warn): Remove.
3015 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3017         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3018         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3020 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3022         * dojump.c (do_jump_1): Divide probability between
3023         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3025 2013-10-02  Tom Tromey  <tromey@redhat.com>
3027         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3029 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3031         * tree-flow.h: Remove some prototypes.
3032         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3033         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3034         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3035         mark_virtual_phi_result_for_renaming): Relocate here.
3036         * tree-into-ssa.h: Add prototypes.
3037         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3038         single_pred_before_succ_order.
3039         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3040         (nonfreeing_call_p) Move to gimple.c.
3041         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3042         tree-ssa-phiopt.c.
3043         * basic-block.h (single_pred_before_succ_order): Add prototype.
3044         * gimple.c (nonfreeing_call_p): Relocate here.
3045         * gimple.h: Add prototype.
3046         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3047         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3048         * tree-ssa.h: Include tree-ssa-dom.h.
3050 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3052         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3053         Use $(COMPILE) and $(POSTCOMPILE).
3055         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3057 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3059         * tree-flow.h: Remove some prototypes.
3060         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3061         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3062         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3063         propagate_tree_value*): Move from here to...
3064         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3065         propagate_tree_value*): Relocate here.
3066         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3067         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3068         * gimple-fold.c: Remove gimple-fold.h from include list.
3069         * tree-vrp.c: Remove gimple-fold.h from include list.
3070         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3071         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3072         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3073         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3074         * sese.c: Add tree-ssa-propagate.h to include list.
3076 2013-10-02  Richard Biener  <rguenther@suse.de>
3078         * tree-loop-distribution.c: Include tree-vectorizer.h for
3079         find_loop_location.
3080         (enum partition_kind): Remove PKIND_REDUCTION.
3081         (struct partition_s): Remove has_writes member, add reduction_p member.
3082         (partition_alloc): Adjust.
3083         (partition_builtin_p): Likewise.
3084         (partition_has_writes): Remove.
3085         (partition_reduction_p): New function.
3086         (partition_merge_into): Likewise.
3087         (generate_code_for_partition): Commonize builtin partition
3088         handling tail.
3089         (rdg_cannot_recompute_vertex_p): Remove.
3090         (already_processed_vertex_p): Likewise.
3091         (rdg_flag_vertex): Do not set has_writes.
3092         (classify_partition): Adjust.
3093         (rdg_build_partitions): Do not set has_writes, treat all
3094         partitions as useful.
3095         (distribute_loop): Record number of library calls generated.  Adjust.
3096         (tree_loop_distribution): Report number of loops and library
3097         calls generated as opt-info.
3099 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3101         * tree-flow.h: Include new .h files.  Move prototypes.
3102         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3103         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3104         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3105         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3106         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3107         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3108         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3109         * tree.h (get_ref_base_and_extent): Move prototype out.
3110         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3111         tree-dfa.h.
3112         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3113         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3114         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3115         * tree-scalar-evolution.c: Include tree.h.
3116         * sese.c: Include tree.h.
3117         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3118         * dwarf2out.c: Include tree-dfa.h.
3119         * tree-chrec.c: Include tree.h.
3120         * tree-data-ref.c: Include tree.h.
3122 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3124         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3125         Fix whitespace.
3127 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3129         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3130         Use $(COMPILE) and $(POSTCOMPILE).
3131         (sol2-cxx.o): Likewise.
3132         (sol2-stubs.o): Likewise.
3133         (sol2.o): Likewise.
3134         * config/x-solaris (host-solaris.o): Likewise.
3136         * config/sparc/t-sparc (sparc.o): Remove.
3137         (sparc-c.o): Remove header dependencies.
3138         Use $(COMPILE) and $(POSTCOMPILE).
3139         * config/sparc/x-sparc: Likewise.
3141 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3143         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3144         * config/arc/arc700.md: Likewise.
3145         * config/arc/arc-modes.def: Likewise.
3146         * config/arc/arc-simd.h: Likewise.
3147         * config/arc/t-arc-uClibc: Likewise.
3148         * config/arc/t-arc-newlib: Likewise.
3150 2013-10-02  Renlin Li  <renlin.li@arm.com>
3152         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3153         plus_constant.
3154         (aarch64_expand_epilogue): Likewise.
3156 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3157             Yufeng Zhang  <yufeng.zhang@arm.com>
3159         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3160         declaration.
3161         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3162         'base_in' represent a conversion and legal_cast_p_1 holds; set
3163         'base_in' with the returned value from get_unwidened.
3165 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3167         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3168         plus_constant is not used.
3170 2013-10-01  Wei Mi  <wmi@google.com>
3172         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3173         * config/i386/i386.md: Add define_peephole2 to
3174         break partial reg stall for cvtss2sd/cvtsd2ss.
3176 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3178         * config/arc/arc.c (pass_arc_ifcvt::clone):
3179         Update for ctxt_ -> m_ctxt change.
3181 2013-10-01  Jeff Law  <law@redhat.com>
3183         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3184         outgoing_edge and intermediate_edge fields.  Instead store the path.
3185         (redirection_data::hash): Hash on the last edge's destination index.
3186         (redirection_data::equal): Check the entire thread path.
3187         (lookup_redirectio_data): Corresponding changes.
3188         (create_edge_and_update_destination_phis): Likewise.
3189         (thread_single_edge): Likewise.
3191 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3192             Diego Novillo <dnovillo@google.com>
3194         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3195         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3196         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3197         versions of these patterns.
3199         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3200         (__builtin_arc_aligned): Likewise.
3202         * config/arc/arc.md: Expand adc_0 comment stating the intended
3203         purpose and why it isn't ready.
3204         Replace commented out call_value_via_label_mixed with a
3205         plain comment about bl_s.
3207         * config/arc/arc.c (stdio.h): Don't include directly.
3208         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3209         Remove if (1) condition.
3210         (arc_encode_section_info): Fix comment.
3212 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3214         * config/arc/arc.c (arc_conditional_register_usage):
3215         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3216         Also set reg_alloc_order for DMA config regs.
3218 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3219             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3221         * doc/install.texi (--with-cpu): Mention ARC.
3222         (arc-*-elf32): New paragraph.
3223         (arc-linux-uclibc): Likewise.
3224         * doc/md.texi (Machine Constraints): Add ARC part.
3225         * doc/invoke.texi: (menu): Add ARC Options.
3226         (Machine Dependent Options) <ARC Options>: Add synopsis.
3227         (node ARC Options): Add.
3228         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3229         (ARC Built-in Functions): New section defining
3230         generic ARC built-in functions.
3231         (ARC SIMD Built-in Functions): New section defining SIMD specific
3232         built-in functions.
3233         (Declaring Attributes of Functions): Extended
3234         description of short_call and long_call attributes for ARC and
3235         added index entries.
3237 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3238             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3239             Joern Rennecke  <joern.rennecke@embecosm.com>
3240             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3241             Brendan Kehoe  <brendan@zen.org>
3242             Michael Eager  <eager@eagercon.com>
3243             Simon Cook  <simon.cook@embecosm.com>
3244             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3246         * config/arc, common/config/arc: New directories.
3248 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3249             Brendan Kehoe  <brendan@zen.org>
3250             Simon Cook  <simon.cook@embecosm.com>
3252         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3254 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3256         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3257         * tree-ssa-coalesce.h: New. Move prototype to here.
3258         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3259         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3260         (gimple_can_coalesce_p): Move to...
3261         * gimple.c (gimple_can_coalesce_p): Here.
3263 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3265         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3266         (dump_decl_set): Move to gimple.c.
3267         * gimple.h: Don't include tree-ssa-operands.h.
3268         (dump_decl_set): Add prototype.
3269         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3270         Move to gimple-ssa.h.
3271         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3272         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3273         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3274         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3275         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3276         than PHI_ARG_DEF.
3277         (dump_decl_set): Relocate here.
3278         * gimple-ssa.h: New file.
3279         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3280         Relocate from gimple.h.
3281         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3282         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3283         swap_tree_operands and remove non-ssa path.
3284         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3285         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3286         swap_ssa_operands.
3287         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3288         vect_is_simple_reduction_1): Use swap_ssa_operands.
3289         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3290         (enum need_phi_state): Move to tree-into-ssa.c.
3291         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3292         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3293         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3294         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3295         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3296         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3297         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3298         num_imm_uses): Move to ssa-iterators.h.
3299         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3300         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3301         tree-phinodes.h.
3302         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3303         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3304         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3305         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3306         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3307         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3308         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3309         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3310         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3311         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3312         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3313         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3314         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3315         (set_phi_nodes): Move to tree-phinodes.h.
3316         * tree-ssa-operands.h (enum ssa_op_iter_type,
3317         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3318         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3319         (dump_decl_set): Remove prototype.
3320         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3321         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3322         (set_phi_nodes): Relocate from tree-flow-inline.h.
3323         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3324         tree-flow-inline.h
3325         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3326         include list.  Temporarily add gimple.h to include list.
3327         * ssa-iterators.h: New file.
3328         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3329         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3330         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3331         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3332         Relocate from tree-ssa-operands.h.
3333         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3334         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3335         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3336         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3337         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3338         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3339         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3340         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3341         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3342         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3343         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3344         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3345         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3346         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3347         Relocate from tree-flow-inline.h.
3348         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3350 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3352         * aarch64-simd.md
3353         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3354         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3355         Insert '\t' to output template.
3356         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3357         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3358         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3359         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3361 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3363         * doc/install.texi (Host/target specific installation notes for GCC):
3364         Put @anchor before @heading.
3365         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3366         Use @email for email addresses.
3368 2013-10-01  Jeff Law  <law@redhat.com>
3370         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3371         a vec.  Only delete the path if we create one without successfully
3372         registering a jump thread.
3373         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3374         as a pointer.
3375         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3376         (paths): New vector of jump threading paths.
3377         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3378         (THREAD_PATH): New accessor macro for the entire thread path.
3379         (lookup_redirection_data): Get intermediate and final outgoing edge
3380         from the thread path.
3381         (create_edge_and_update_destination_phis): Copy the threading path.
3382         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3383         jump threading path.
3384         (ssa_redirect_edges): Get edges and block types from the jump threading
3385         path.  Free the path vector.
3386         (thread_block): Get edges from the jump threading path.  Look at the
3387         entire path to see if we thread to a loop exit.  If we cancel a jump
3388         thread request, then free the path vector.
3389         (thread_single_edge): Get edges and block types from the jump threading
3390         path.  Free the path vector.
3391         (thread_through_loop_header): Get edges and block types from the jump
3392         threading path.  Free the path vector.
3393         (mark_threaded_blocks): Iterate over the vector of paths and store
3394         the path on the appropriate edge.  Get edges and block types from the
3395         jump threading path.
3396         (mark_threaded_blocks): Get edges and block types from the jump
3397         threading path.  Free the path vector.
3398         (thread_through_all_blocks): Use the vector of paths rather than
3399         a vector of 3-edge sets.
3400         (register_jump_thread): Accept pointer to a path vector rather
3401         than the path vector itself.  Store the path vector for later use.
3402         Simplify.
3404 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3405             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3407         PR target/58574
3408         * config/s390/s390.c (s390_split_branches): Modify check for table
3409         jump insns.
3410         (s390_chunkify_start): Rearrange table jump insn check in order to
3411         deal with compare and branch insns correctly.
3413 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3415         PR target/58578
3416         Revert
3417         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3418         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3419         define_insn_and_split.
3420         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3421         (shiftsi3_compare): New pattern.
3422         (rrx): New pattern.
3423         * config/arm/unspecs.md (UNSPEC_RRX): New.
3425 2013-10-01  Alan Modra  <amodra@gmail.com>
3427         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3428         casing inout operands.
3430 2013-10-01  Richard Biener  <rguenther@suse.de>
3432         PR tree-optimization/58553
3433         * tree-loop-distribution.c (struct partition_s): Add niter member.
3434         (classify_partition): Populate niter member for the partition
3435         and properly identify whether the relevant store happens before
3436         or after the loop exit.
3437         (generate_memset_builtin): Use niter member from the partition.
3438         (generate_memcpy_builtin): Likewise.
3440 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3442         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3443         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3445 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3447         * basic-block.h (edge_list): Prefix member names with "m_".
3448         * context.h (context): Likewise.
3449         * domwalk.h (dom_walker): Likewise.
3450         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3451         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3452         * hash-table.h (hash_table): Likewise.
3453         * machmode.h (bit_field_mode_iterator): Likewise.
3454         * pass_manager.h (pass_list): Likewise.
3455         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3456         * tree-pass.h (pass_data): Likewise.
3457         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3458         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3459         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3460         * asan.c (pass_data_asan): Update accordingly.
3461         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3462         (control_dependences::control_dependences): Likewise.
3463         (control_dependences::~control_dependences): Likewise.
3464         (control_dependences::~control_dependences): Likewise.
3465         (control_dependences::get_edges_dependent_on): Likewise.
3466         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3467         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3468         * context.c (gcc::context::context): Likewise.
3469         * cprop.c (pass_rtl_cprop::clone): Likewise.
3470         * domwalk.c (dom_walker::walk): Likewise.
3471         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3472         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3473         * mode-switching.c (pass_mode_switching::clone): Likewise.
3474         * passes.c (opt_pass::opt_pass): Likewise.
3475         (pass_manager::pass_manager): Likewise.
3476         * predict.c (pass_strip_predict_hints::clone): Likewise.
3477         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3478         (pass_split_all_insns::clone): Likewise.
3479         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3480         Likewise.
3481         (bit_field_mode_iterator::next_mode): Likewise.
3482         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3483         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3484         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3485         * tree-complex.c (pass_lower_complex::clone): Likewise.
3486         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3487         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3488         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3489         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3490         (pass_fold_builtins::clone): Likewise.
3491         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3492         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3493         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3494         (pass_cd_dce::clone): Likewise.
3495         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3496         (pass_phi_only_cprop::clone): Likewise.
3497         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3498         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3499         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3500         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3501         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3502         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3503         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3504         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3505         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3506         * tree-vrp.c (pass_vrp::clone): Likewise.
3507         * tsan.c (pass_tsan::clone): Likewise.
3509 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3511         PR middle-end/58564
3512         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3513         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3515         PR middle-end/58564
3516         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3517         optimization, punt if sign_bit_p looked through any zero extension.
3519 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3521         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3522         Update redirected out edge count in joiner case.
3523         (ssa_redirect_edges): Common the joiner and non-joiner cases
3524         so that joiner case gets profile updates.
3526 2013-09-30  Richard Biener  <rguenther@suse.de>
3528         PR tree-optimization/58554
3529         * tree-loop-distribution.c (classify_partition): Require
3530         unconditionally executed stores for memcpy and memset recognition.
3531         (tree_loop_distribution): Calculate dominance info.
3533 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3535         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3536         (NO_PROFILE_COUNTERS): Likewise.
3537         (PROFILE_HOOK): Likewise.
3538         (FUNCTION_PROFILER): Likewise.
3539         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3541 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3543         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3544         calls and defines in TARGET_MACHO conditional.
3545         (load_macho_picbase_di): Likewise.
3546         (reload_macho_picbase): Likewise.
3547         (reload_macho_picbase_si): Likewise.
3548         (reload_macho_picbase_di): Likewise.
3549         (nonlocal_goto_receiver): Likewise.
3551 2013-09-30  Nick Clifton  <nickc@redhat.com>
3553         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3554         that use the MSP430X ISA.
3555         (msp430_option_override): Scan -mmcu command line option for any
3556         MCU name that supports the MSP430X ISA.
3557         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3558         -mmcu options which enable the MSP430X ISA.
3560 2013-09-30  Richard Biener  <rguenther@suse.de>
3562         PR middle-end/58532
3563         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3564         before looking for setjmp-like calls.
3566 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3568         PR target/10901
3569         * config/darwin-protos.h (machopic_get_function_picbase): New.
3570         * config/darwin.c (machopic_get_function_picbase): New.
3571         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3572         label for a new func.  (load_macho_picbase_di): Likewise.
3573         (reload_macho_picbase): New expand.
3574         (reload_macho_picbase_si): New insn.
3575         (reload_macho_picbase_di): New insn.
3576         (nonlocal_goto_receiver): New define and split.
3577         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3578         (unspecv enum): Add UNSPECV_NLGR.
3580 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3582         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3583         that altivec registers are correctly sized on Darwin.
3585 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3587         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3588         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3589         * config/x-darwin (host-darwin.o): Likewise.
3590         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3591         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3592         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3594 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3596         * doc/invoke.texi: Fix usage of @tie{} command.
3598 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3600         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3602 2013-09-28  Jan Hubicka  <jh@suse.cz>
3604         * config/i386/x86-tune.def: Add documentation for each of the options;
3605         add whitespace.
3607 2013-09-28  Jan Hubicka  <jh@suse.cz>
3609         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3610         generic.
3611         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3612         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3613         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3614         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3615         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3617 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3619         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3620         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3621         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3622         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3623         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3624         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3625         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3626         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3627         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3628         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3629         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3630         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3631         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3632         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3633         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3634         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3635         gimple.h, godump.c, graphite-clast-to-gimple.c,
3636         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3637         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3638         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3639         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3640         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3641         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3642         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3643         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3644         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3645         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3646         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3647         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3648         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3649         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3650         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3651         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3652         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3653         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3654         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3655         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3656         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3657         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3658         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3659         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3660         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3661         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3662         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3663         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3664         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3665         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3666         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3667         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3668         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3669         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3670         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3671         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3672         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3673         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3674         whitespace before "(".
3676 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3678         * expr.h (extract_bit_field): Remove packedp parameter.
3679         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3680         from forward declaration.
3681         (store_split_bit_field): Remove packedp arg from calls to
3682         extract_fixed_bit_field.
3683         (extract_bit_field_1): Remove packedp parameter and packedp
3684         argument from recursive calls and calls to extract_fixed_bit_field.
3685         (extract_bit_field): Remove packedp parameter and corresponding
3686         arg to extract_bit_field_1.
3687         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3688         to issue warnings.
3689         (extract_split_bit_field): Remove packedp arg from call to
3690         extract_fixed_bit_field.
3691         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3692         (copy_blkmode_from_reg): Likewise.
3693         (copy_blkmode_to_reg): Likewise.
3694         (read_complex_part): Likewise.
3695         (store_field): Likewise.
3696         (expand_expr_real_1): Likewise.
3697         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3698         to extract_bit_field.
3699         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3700         call to extract_bit_field.
3701         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3702         call to extract_bit_field.
3703         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3704         and special packedp behavior from -fstrict-volatile-bitfields
3705         documentation.
3707 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3709         * lra-eliminations.c (init_elim_table): Guard value_p.
3711 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3713         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3714         DFmode, DImode, and SFmode in the upper VSX registers based on the
3715         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3716         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3717         of the VSX register handling.
3719         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3720         (f32_sv): Likewise.
3721         (zero_extendsidi2_lfiwzx): Add support for loading into the
3722         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3723         only do VSX memory options on Altivec registers.
3724         (extendsidi2_lfiwax): Likewise.
3725         (extendsfdf2_fpr): Likewise.
3726         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3727         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3728         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3729         (movdi_internal64): Likewise.
3731 2013-09-27  Xinliang David Li  <davidxl@google.com>
3733         * opts.c (finish_options): Adjust parameters
3734         according to vect cost model.
3735         (common_handle_option): Set dynamic vect cost
3736         model for FDO.
3737         targhooks.c (default_add_stmt_cost): Compute stmt cost
3738         unconditionally.
3739         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3740         Use helper function.
3741         * tree-vectorizer.h (unlimited_cost_model): New function.
3742         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3743         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3744         function.
3745         (vect_enhance_data_refs_alignment): Ditto.
3746         * flag-types.h: New enum.
3747         * common/config/i386/i386-common.c (ix86_option_init_struct):
3748         No need to initialize vect_cost_model flag.
3749         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3750         unconditionally.
3752 2013-09-27  Diego Novillo  <dnovillo@google.com>
3754         * gimple.h (enum ssa_mode): Remove.
3756 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3758         * cfgloop.h (number_of_loops): Fix typo in check for null.
3760 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3762         PR middle-end/58551
3763         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3764         are children of outermost saved_cfun's loop, and set it up to
3765         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3766         if loop != loop0 and SESE region contains bbs that belong to loop0.
3768 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3770         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3771         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3772         (set_address_segment, set_address_base, set_address_index)
3773         (set_address_disp): Accept the argument unconditionally.
3774         (baseness): Remove must_be_base_p and must_be_index_p checks.
3775         (decompose_normal_address): Classify as much as possible in the
3776         main loop.
3778 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3780         * cse.c (count_reg_usage): Handle INT_LIST.
3781         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3782         * reginfo.c (reg_scan_mark_refs): Likewise.
3783         * reload1.c (eliminate_regs_1): Likewise.
3785 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3787         PR middle-end/58547
3788         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3789         signed.
3791 2013-09-27  Richard Biener  <rguenther@suse.de>
3793         PR tree-optimization/58459
3794         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3795         restriction not propagating into loops.
3797 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3799         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3800         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3801         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3803 2013-09-26  Richard Biener  <rguenther@suse.de>
3805         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3807 2013-09-26  Richard Biener  <rguenther@suse.de>
3809         * alias.h (component_uses_parent_alias_set): Rename to ...
3810         (component_uses_parent_alias_set_from): ... this.
3811         * alias.c (component_uses_parent_alias_set): Rename to ...
3812         (component_uses_parent_alias_set_from): ... this and return
3813         the desired parent.
3814         (reference_alias_ptr_type_1): Use the result from
3815         component_uses_parent_alias_set_from instead of stripping
3816         components one at a time.
3817         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3819 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3821         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3822         Move prototypes to...
3823         * tree-ssa-ter.h: New File.  Move prototypes here.
3824         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3825         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3826         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3827         from is_replaceable_p.
3828         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3829         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3830         (process_replaceable): Use ter_is_replaceable_p.
3831         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3832         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3833         newly refactored ssa_is_replaceable_p.
3834         * cfgexpand.c: Include tree-outof-ssa.h.
3835         * ssaexpand.h: Delete.
3837 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3839         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3840         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3841         tree-ssa.c
3842         (create_gimple_tmp): Delete.
3843         (get_expr_type, build_assign, build_type_cast): Move to...
3844         * gimple-builder.c: New File.
3845         (get_expr_type): Relocate from gimple.c.
3846         (build_assign, build_type_cast): Change to only create ssanames.
3847         * gimple.h: Move prototypes to...
3848         * gimple-builder.h: New File. Here.
3849         * tree-ssa.h: And here.
3850         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3851         count_uses_and_derefs): Relocate from gimple.c.
3852         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3853         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3854         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3855         gimple_replace_ssa_lhs.
3856         * asan.c: Include gimple-builder.h.
3857         * Makefile.in: Add gimple-builder.o.
3859 2013-09-26  Richard Biener  <rguenther@suse.de>
3861         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3862         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3863         (loe_visit_block): Use gcc_checking_assert.
3864         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3865         gimple_assign_ssa_name_copy_p.
3866         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3867         change.
3869 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3871         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3872         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3874 2013-09-26  Richard Biener  <rguenther@suse.de>
3876         PR tree-optimization/58539
3877         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3878         the fact that debug statements are not taking part in loop-closed
3879         SSA construction.
3881 2013-09-26  Nick Clifton  <nickc@redhat.com>
3883         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3884         time warning message.
3885         (msp430_print_operand_raw): Delete unused letter parameter.
3886         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3887         (msp430_print_operand_address): New function.
3888         (msp430_print_operand): Move address printing code from here to
3889         new function.
3890         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3891         assembler.
3892         (zero_extendpsisi2): Likewise.
3893         (extendpsisi2): New pattern.
3894         (andneghi3): New pattern.
3896 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3898         * config/aarch64/aarch64.opt (mlra): New option.
3899         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3900         (TARGET_LRA_P): Define.
3902 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3904         * expr.c (expand_assignment): Remove obsolete comment.
3906 2013-09-25  Jeff Law  <law@redhat.com>
3908         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3909         tree-ssa-threadupdate.h.
3910         (register_jump_thread): Similarly.
3911         * tree-ssa-threadupdate.h: New header file.
3912         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3913         * tree-vrp.c: Likewise.
3914         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3915         (thread_around_empty_blocks): Change type of path vector argument to
3916         an edge,type pair from just an edge.  Initialize both elements when
3917         appending to a jump threading path.  Tweak references to elements
3918         appropriately.
3919         (thread_across_edge): Similarly.  Release memory for the elements
3920         as needed.
3921         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3922         (dump_jump_thread_path): New function broken out from
3923         register_jump_thread.
3924         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3925         path vector entries.  Search the path for NULL edges and dump
3926         the path if one is found.  Tweak the conversion of path to 3-edge
3927         form to use the block copy type information embedded in the path.
3929 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3931         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3933 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3934             Vladimir Makarov  <vmakarov@redhat.com>
3936         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3937         from the least significant bit.
3938         (strip_address_mutations): Add bitfield operations handling.
3939         (must_be_index_p): Add shifting and rotate operations handling.
3940         (set_address_base): Use must_be_base_p predicate.
3941         (set_address_index): Use must_be_index_p predicate.
3943 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3944             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3945             Sergey Lega  <sergey.s.lega@intel.com>
3946             Anna Tikhonova  <anna.tikhonova@intel.com>
3947             Ilya Tocar  <ilya.tocar@intel.com>
3948             Andrey Turetskiy  <andrey.turetskiy@intel.com>
3949             Ilya Verbin  <ilya.verbin@intel.com>
3950             Kirill Yukhin  <kirill.yukhin@intel.com>
3951             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
3953         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
3954         Use new names.
3955         (ix86_expand_vector_move_misalign): Support new unaligned load and
3956         stores and use new names.
3957         (CODE_FOR_sse2_storedqu): Rename to ...
3958         (CODE_FOR_sse2_storedquv16qi): ... this.
3959         (CODE_FOR_sse2_loaddqu): Rename to ...
3960         (CODE_FOR_sse2_loaddquv16qi): ... this.
3961         (CODE_FOR_avx_loaddqu256): Rename to ...
3962         (CODE_FOR_avx_loaddquv32qi): ... this.
3963         (CODE_FOR_avx_storedqu256): Rename to ...
3964         (CODE_FOR_avx_storedquv32qi): ... this.
3965         * config/i386/i386.md (fpint_logic): New.
3966         * config/i386/sse.md (VMOVE): Extend for AVX512.
3967         (VF): Ditto.
3968         (VF_128_256): New.
3969         (VF_512): Ditto.
3970         (VI_UNALIGNED_LOADSTORE): Ditto.
3971         (sse2_avx_avx512f): Ditto.
3972         (sse2_avx2): Extend for AVX512.
3973         (sse4_1_avx2): Ditto.
3974         (avx2_avx512f): New.
3975         (sse): Extend for AVX512.
3976         (sse2): Ditto.
3977         (sse4_1): Ditto.
3978         (avxsizesuffix): Ditto.
3979         (sseintvecmode): Ditto.
3980         (ssePSmode): Ditto.
3981         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
3982         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
3983         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3984         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
3985         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
3986         (<sse2_avx_avx512f>_storedqu<mode): ... this.
3987         (<sse>_movnt<mode>): Replace constraint "x" with "v".
3988         (STORENT_MODE): Extend for AVX512.
3989         (*absneg<mode>2): Replace constraint "x" with "v".
3990         (*mul<mode>3): Ditto.
3991         (*ieee_smin<mode>3): Ditto.
3992         (*ieee_smax<mode>3): Ditto.
3993         (avx_cmp<mode>3): Replace VF with VF_128_256.
3994         (*<sse>_maskcmp<mode>3_comm): Ditto.
3995         (<sse>_maskcmp<mode>3): Ditto.
3996         (<sse>_andnot<mode>3): Extend for AVX512.
3997         (<code><mode>3, anylogic): Replace VF with VF_128_256.
3998         (<code><mode>3, fpint_logic): New.
3999         (*<code><mode>3): Extend for AVX512.
4000         (avx512flogicsuff): New.
4001         (avx512f_<logic><mode>): Ditto.
4002         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4003         VF_128_256.
4004         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4005         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4006         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4007         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4008         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4009         (xop_vpermil2<mode>3): Ditto.
4010         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4011         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4012         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4013         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4015 2013-09-25  Tom Tromey  <tromey@redhat.com>
4017         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4018         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4019         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4020         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4021         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4022         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4023         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4024         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4025         (GRAPHITE_HTAB_H): Remove.
4027 2013-09-25  Tom Tromey  <tromey@redhat.com>
4029         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4031 2013-09-25  Tom Tromey  <tromey@redhat.com>
4033         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4035 2013-09-25  Tom Tromey  <tromey@redhat.com>
4037         * config/i386/t-i386 (i386.o): Remove.
4038         (i386-c.o): Use COMPILE and POSTCOMPILE.
4040 2013-09-25  Tom Tromey  <tromey@redhat.com>
4042         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4044 2013-09-25  Tom Tromey  <tromey@redhat.com>
4046         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4047         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4048         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4049         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4050         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4051         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4052         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4053         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4054         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4055         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4056         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4057         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4058         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4059         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4060         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4061         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4062         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4063         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4064         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4065         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4066         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4067         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4068         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4069         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4070         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4071         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4072         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4073         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4074         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4075         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4076         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4077         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4078         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4079         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4080         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4081         (graphite-clast-to-gimple.o, graphite-dependences.o)
4082         (graphite-interchange.o, graphite-poly.o)
4083         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4084         (graphite-optimize-isl.o, tree-vect-loop.o)
4085         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4086         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4087         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4088         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4089         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4090         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4091         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4092         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4093         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4094         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4095         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4096         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4097         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4098         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4099         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4100         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4101         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4102         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4103         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4104         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4105         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4106         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4107         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4108         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4109         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4110         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4111         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4112         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4113         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4114         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4115         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4116         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4117         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4118         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4119         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4120         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4121         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4122         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4123         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4124         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4125         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4126         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4127         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4128         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4129         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4130         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4131         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4132         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4133         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4134         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4135         (gcov-dump.o): Remove.
4136         (default-c.o): Use COMPILE and POSTCOMPILE.
4137         (CFLAGS-gcc.o): New variable.
4138         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4140 2013-09-25  Tom Tromey  <tromey@redhat.com>
4142         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4143         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4144         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4145         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4146         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4147         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4148         (c-family/array-notation-common.o, c-family/stub-objc.o)
4149         (c-family/c-ubsan.o): Remove.
4151 2013-09-25  Tom Tromey  <tromey@redhat.com>
4153         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4155 2013-09-25  Tom Tromey  <tromey@redhat.com>
4157         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4158         to add -DENABLE_SHARED_LIBGCC.
4159         (gcc.o): Don't use subshell.
4161 2013-09-25  Tom Tromey  <tromey@redhat.com>
4163         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4164         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4165         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4166         * configure, config.in: Rebuild.
4168 2013-09-25  Tom Tromey  <tromey@redhat.com>
4170         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4171         (COMPILE, POSTCOMPILE): New variables.
4172         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4173         (DEPFILES): New variable.
4174         Include ".Po" files.
4175         * configure.ac: Add checks for dependency checking.
4176         * configure, aclocal.m4: Regenerate.
4178 2013-09-25  Tom Tromey  <tromey@redhat.com>
4180         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4181         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4183 2013-09-25  Tom Tromey  <tromey@redhat.com>
4185         * Makefile.in (generated_files): Add options.h,
4186         target-hooks-def.h, insn-opinit.h,
4187         common/common-target-hooks-def.h, pass-instances.def,
4188         c-family/c-target-hooks-def.h.
4190 2013-09-25  Jeff Law  <law@redhat.com>
4192         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4193         than foo[foo.length () - 1] to access last member in a vec.
4194         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4196 2013-09-25  Richard Biener  <rguenther@suse.de>
4198         PR middle-end/58521
4199         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4201 2013-09-25  Jan Hubicka  <jh@suse.cz>
4203         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4204         test.
4206 2013-09-25  Marek Polacek  <polacek@redhat.com>
4208         PR sanitizer/58420
4209         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4210         when determining the type name.
4212 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4214         * config/sh/sh.md: Fix formatting.
4216 2013-09-24  Xinliang David Li  <davidxl@google.com>
4218         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4219         max peel iterations parameter.
4220         * param.def: New parameter.
4221         * doc/invoke.texi: Document New parameter.
4223 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4225         * gimple-pretty-print.c: Various whitespace tweaks.
4226         * tree-core.h: Likewise.
4227         * tree-pretty-print.c: Likewise.
4228         * tree-ssa-alias.c: Likewise.
4229         * tree-ssa-copy.c: Likewise.
4230         * tree-ssanames.c: Likewise.
4231         * tree-ssanames.h: Likewise.
4232         * tree-vrp.c: Likewise.
4234 2013-09-24  Alan Modra  <amodra@gmail.com>
4236         PR middle-end/57134
4237         PR middle-end/57586
4238         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4239         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4240         inout operands.
4242 2013-09-24  Richard Biener  <rguenther@suse.de>
4244         PR middle-end/58513
4245         * tree.c (reference_alias_ptr_type): Move ...
4246         * alias.c (reference_alias_ptr_type): ... here and implement
4247         in terms of the new reference_alias_ptr_type_1.
4248         (ref_all_alias_ptr_type_p): New helper.
4249         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4250         use ref_all_alias_ptr_type_p.
4251         (get_deref_alias_set): Add flag_strict_aliasing check here.
4252         (reference_alias_ptr_type_1): New function, split out from ...
4253         (get_alias_set): ... here.
4254         (alias_ptr_types_compatible_p): New function.
4255         * alias.h (reference_alias_ptr_type): Declare.
4256         (alias_ptr_types_compatible_p): Likewise.
4257         * tree.h (reference_alias_ptr_type): Remove.
4258         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4259         to compare MEM_REF alias types.
4261 2013-09-24  Richard Biener  <rguenther@suse.de>
4263         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4265 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4267         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4268         reload helper function arrays into a single array reg_addr.
4269         (reload_fpr_gpr): Likewise.
4270         (reload_gpr_vsx): Likewise.
4271         (reload_vsx_gpr): Likewise.
4272         (struct rs6000_reg_addr): Likewise.
4273         (reg_addr): Likewise.
4274         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4275         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4276         (rs6000_init_hard_regno_mode_ok): Likewise.
4277         (rs6000_secondary_reload_direct_move): Likewise.
4278         (rs6000_secondary_reload): Likewise.
4280         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4281         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4282         power8 changes.  Put wg constraint in alphabetical order.
4284         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4285         for future work to add ISA 2.07 VSX single precision support.
4286         (-mvsx-scalar-double): Change default from -1 to 1, update
4287         documentation comment.
4288         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4289         (-mupper-regs-df): New debug switch to control whether DF values
4290         can go in the traditional Altivec registers.
4291         (-mupper-regs-sf): New debug switch to control whether SF values
4292         can go in the traditional Altivec registers.
4294         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4295         and wy constraints.
4296         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4297         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4298         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4299         (rs6000_debug_legitimate_address_p): Print if we are running
4300         before, during, or after reload.
4301         (rs6000_secondary_reload): Add a comment.
4302         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4304         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4305         constraints.  Update documentation string.
4306         (wd constraint): Likewise.
4307         (wf constraint): Likewise.
4308         (wg constraint): Likewise.
4309         (wn constraint): Likewise.
4310         (ws constraint): Likewise.
4311         (wt constraint): Likewise.
4312         (wx constraint): Likewise.
4313         (wz constraint): Likewise.
4314         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4315         instructions.
4316         (ww constraint): Likewise.
4317         (wy constraint): Likewise.
4318         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4319         in the previous submissions.
4320         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4322 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4324         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4325         (REG_BR_PROB): Say that the probability is stored in an int_list.
4326         * reg-notes.def: Update commentary to mention INT_LIST.
4327         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4328         (INT_LIST): New rtx.
4329         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4330         * rtlanal.c (int_reg_note_p): New function.
4331         (alloc_reg_note): Assert that the note does not have an int argument.
4332         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4333         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4334         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4335         rather than an INSN_LIST.  Handle INT_LIST.
4336         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4337         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4338         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4339         Manipulate them as ints rather than rtxes.
4340         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4341         * regmove.c (copy_src_to_dest): Likewise.
4342         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4344         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4345         into the cases that need it.
4346         * config/arm/arm.c (arm_unwind_emit): Likewise.
4348         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4349         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4350         * loop-doloop.c (add_test, doloop_modify): Likewise.
4351         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4352         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4353         * predict.c (combine_predictions_for_insn): Likewise.
4354         * print-rtl.c (print_rtx): Handle INT_LIST.
4355         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4356         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4357         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4358         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4359         (ix86_split_fp_branch, predict_jump): Likewise.
4360         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4361         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4362         * config/spu/spu.c (ea_load_store_inline): Likewise.
4364         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4365         value of a REG_BR_PROB note.
4366         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4367         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4368         * emit-rtl.c (try_split): Likewise.
4369         * predict.c (br_prob_note_reliable_p): Likewise.
4370         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4371         * reorg.c (mostly_true_jump): Likewise.
4372         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4373         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4374         * config/i386/i386.c (ix86_print_operand): Likewise.
4375         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4376         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4377         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4378         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4379         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4380         * config/sparc/sparc.c (output_cbranch): Likewise.
4381         * config/spu/spu.c (get_branch_target): Likewise.
4382         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4383         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4385 2013-09-23  Jan Hubicka  <jh@suse.cz>
4387         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4388         for ipa-devirt.
4389         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4390         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4391         of external calls
4392         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4393         (gimple_fold_call): Dump inconsistent devirtualizations; add
4394         sanity check for type based devirtualizations.
4395         * ipa-prop.c: Include ipa-utils.h
4396         (ipa_intraprocedural_devirtualization): Add sanity check.
4397         (try_make_edge_direct_virtual_call): Likewise.
4399 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4401         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4402         assignment statements.
4404 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4406         * gimple-pretty-print.c (dump_ssaname_info): New function.
4407         (dump_gimple_phi): Call it.
4408         (pp_gimple_stmt_1): Likewise.
4409         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4410         (range_info_def): Declare.
4411         * tree-pretty-print.c (pp_double_int): New function.
4412         (dump_generic_node): Call it.
4413         * tree-pretty-print.h (pp_double_int): Declare.
4414         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4415         * tree-ssanames.h (range_info_def): New structure.
4416         (value_range_type): Move definition here.
4417         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4418         Declare.
4419         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4420         initialization.
4421         (set_range_info): New function.
4422         (get_range_info): Likewise.
4423         (duplicate_ssa_name_range_info): Likewise.
4424         (duplicate_ssa_name_fn): Check pointer type and call
4425         duplicate_ssa_name_range_info.
4426         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4427         * tree-vrp.c (value_range_type): Remove definition, now in
4428         tree-ssanames.h.
4429         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4430         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4431         (SSA_NAME_RANGE_INFO): New macro.
4433 2013-09-23  Richard Biener  <rguenther@suse.de>
4435         PR tree-optimization/58464
4436         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4437         (phi_trans_add): Change to add conditionally on being not present.
4438         (phi_translate_1): Remove recursion detection here.
4439         (phi_translate): Pre-seed the cache with NULL to catch
4440         recursion here in a more generic way.
4441         (bitmap_find_leader): Adjust comment.
4442         (get_representative_for): Dump value-numbers.
4443         (create_expression_by_pieces): Likewise.
4444         (insert_into_preds_of_block): Likewise.
4446 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4448         PR target/58475
4449         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4450         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4452 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4454         Revert r202780:
4455         2013-09-20  Renlin Li  <renlin.li@arm.com>
4457         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4458         plus_constant.
4459         (aarch64_expand_epilogue): Likewise.
4460         (aarch64_legitimize_reload_address): Likewise.
4462 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4464         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4465         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4467 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4469         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4470         comparison operators when -fno-trapping-math is in effect.
4471         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4472         comparison operators in TFmode and assert that unsupported operators
4473         cannot reach here.
4474         (ia64_print_operand): Likewise.
4476 2013-09-21  Jan Hubicka  <jh@suse.cz>
4478         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4479         (sse_typeless_stores): Enable for core
4480         (sse_load0_by_pxor): Likewise.
4481         (four_jump_limit): Disable for core.
4482         (pad_returns): Likewise.
4483         (avoid_vector_decode): Likewise.
4484         (fuse_cmp_and_branch): Enable for cores.
4485         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4487 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4489         PR middle-end/56791
4490         * config/pa/pa.c (pa_option_override): Disable auto increment and
4491         decrement instructions until reload is completed.
4493         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4494         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4495         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4497 2013-09-20  DJ Delorie  <dj@redhat.com>
4498             Nick Clifton  <nickc@redhat.com>
4500         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4501         (need_to_save): Save bank 0 on interrupts.
4502         (characterize_address): Strip far address wrappers.
4503         (rl78_as_legitimate_address): Likewise.
4504         (transcode_memory_rtx): Likewise.
4505         (rl78_peep_movhi_p): Disable this peephole after devirt.
4506         (rl78_propogate_register_origins): Forget all origins when a
4507         CLOBBER is seen.
4508         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4509         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4510         peephole2's.
4511         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4512         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4513         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4514         relocating.
4515         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4517 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4519         * config/pa/pa.md: In "scc" insn patterns, change output template to
4520         handle const0_rtx in reg_or_0_operand operands.
4522 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4524         PR target/56875
4525         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4526         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4528 2013-09-20  Richard Biener  <rguenther@suse.de>
4530         PR middle-end/58484
4531         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4532         remembering SSA name version and block index.
4533         (new_scev_info_str): Adjust.
4534         (hash_scev_info): Likewise.  Also hash the block index.
4535         (eq_scev_info): Adjust.
4536         (find_var_scev_info): Likewise.
4537         (struct instantiate_cache_entry): Remove.
4538         (struct instantiate_cache_type): Use a htab to map name, block
4539         to chrec.
4540         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4541         (get_instantiated_value_entry): Likewise.
4542         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4543         (instantiate_scev_name): Adjust.
4545 2013-09-20  Jeff Law  <law@redhat.com>
4547         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4549 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4551         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4552         Call aarch64_simd_expand_args to update op[argc].
4554 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4556         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4557         plugin argument.
4559 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4561         * gengtype.c (file_rules): Added rule for *.cc files.
4562         (get_output_file_with_visibility): Give fatal message when no
4563         rules found.
4565 2013-09-20  Renlin Li  <renlin.li@arm.com>
4567         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4568         (aarch64_expand_epilogue): Likewise.
4569         (aarch64_legitimize_reload_address): Likewise.
4571 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4573         PR middle-end/57748
4574         * expr.c (expand_assignment): Remove misalignp code path.
4576 2013-09-20  Marek Polacek  <polacek@redhat.com>
4578         PR sanitizer/58413
4579         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4580         TYPE_PRECISION.  Add asserts.
4582 2013-09-20  Richard Biener  <rguenther@suse.de>
4584         PR tree-optimization/58453
4585         * tree-loop-distribution.c (distribute_loop): Apply the cost
4586         model for -ftree-loop-distribute-patterns, too.
4588 2013-09-20  Richard Biener  <rguenther@suse.de>
4590         PR middle-end/58473
4591         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4592         make type comparison less strict.
4594 2013-09-20  Alan Modra  <amodra@gmail.com>
4596         * configure: Regenerate.
4597         * aclocal.m4: Regenerate.
4599 2013-09-20  Marek Polacek  <polacek@redhat.com>
4601         PR other/58467
4602         * doc/extend.texi: Document that attribute used is meant to be used
4603         on variables with static storage duration.
4605 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4607         PR tree-optimization/58472
4608         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4609         simd_lane_access set inv_p = false.
4610         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4611         the simduid magic VAR_DECL.
4613 2013-09-19  Jan Hubicka  <jh@suse.cz>
4615         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4617 2013-09-17  Jeff Law  <law@redhat.com>
4619         * tree-ssa-dom.c (record_temporary_equivalences): New function
4620         split out of dom_opt_dom_walker::after_dom_children.
4621         (dom_opt_dom_walker::thread_across_edge): Move common code
4622         in here from dom_opt_dom_walker::after_dom_children.
4623         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4625 2013-09-19  Jan Hubicka  <jh@suse.cz>
4627         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4628         (TARGET_GENERIC): Use PROCESOR_GENERIC
4629         (enum processor_type): Unify generic32 and 64.
4630         * i386.md (cpu): Likewise.
4631         * x86-tune.def (use_leave): Enable for generic32.
4632         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4633         * athlon.md: Change generic64 to generic in all occurences.
4634         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4635         (ix86_target_macros_internal): Likewise.
4636         * driver-i386.c (host_detect_local_cpu): Likewise.
4637         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4638         to ..
4639         (generic_memcpy, generic_memset, generic_cost): This one.
4640         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4641         (m_GENERIC32, m_GENERIC64): Remove.
4642         (m_GENERIC): Turn into one flag.
4643         (processor_target): Unify generic tunnings.
4644         (ix86_option_override_internal): Replace generic32/64 by generic.
4645         (ix86_issue_rate): Likewise.
4646         (ix86_adjust_cost): Likewise.
4648 2013-09-19  Jan Hubicka  <jh@suse.cz>
4650         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4651         of speculative flag.
4653 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4655         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4656         GOMP_sections_start, even if !exit_reachable.
4658 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4660         * lra-constraints.c (need_for_all_save_p): Use macro
4661         HARD_REGNO_CALL_PART_CLOBBERED.
4662         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4663         set up pseudo conflict hard regs.
4665 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4667         PR target/58452
4668         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4669         operands.
4671 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4673         PR rtl-optimization/58438
4674         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4675         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4676         from previous subpasses.
4678 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4680         * arm.c (arm_get_frame_offsets): Validate architecture supports
4681         LDRD/STRD before accepting the tuning preference.
4682         (arm_expand_prologue): Likewise.
4683         (arm_expand_epilogue): Likewise.
4685 2013-09-18  Richard Biener  <rguenther@suse.de>
4687         PR tree-optimization/58417
4688         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4689         have chrecs with symbols defined in the loop as operands.
4690         (chrec_fold_multiply): Likewise.
4691         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4692         parameters before folding binary operations.
4693         (struct instantiate_cache_entry_hasher): Remove.
4694         (struct instantiate_cache_type): Use a pointer-map.
4695         (instantiate_cache_type::instantiate_cache_type): New function.
4696         (instantiate_cache_type::get): Likewise.
4697         (instantiate_cache_type::set): Likewise.
4698         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4699         (get_instantiated_value_entry): Likewise.
4700         (global_cache): New global.
4701         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4702         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4703         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4704         (instantiate_scev_name): Adjust.
4705         (instantiate_scev): Construct global instead of local cache.
4706         (resolve_mixers): Likewise.
4708 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4709             Paolo Carlini  <paolo.carlini@oracle.com>
4711         PR c++/58458
4712         * doc/implement-cxx.texi: Fix references to the C++ standards.
4714 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4716         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4717         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4718         simd_lane_access drs, update also DR_ALIGNED_TO.
4720 2013-09-18  Marek Polacek  <polacek@redhat.com>
4722         PR sanitizer/58411
4723         * doc/extend.texi: Document no_sanitize_undefined attribute.
4724         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4725         no_sanitize_undefined attribute.
4727 2013-09-18  Nick Clifton  <nickc@redhat.com>
4729         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4730         (ASM_DECLARE_FUNCTION_NAME): Define.
4732 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4734         * compare-elim.c (find_comparison_dom_walker): New class
4735         (find_comparisons_in_bb): Rename to
4736         find_comparison_dom_walker::before_dom_children
4737         (find_comparisons): Adjust
4738         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4739         adjust.
4740         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4741         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4742         (init_walk_dominator_tree): Remove declaration.
4743         (fini_walk_dominator_tree): Remove declaration.
4744         * fwprop.c (single_def_use_dom_walker): New class
4745         (single_def_use_enter_block): Convert to
4746         single_def_use_dom_walker::before_dom_children.
4747         (single_def_use_leave_block): Convert to
4748         single_def_use_dom_walker::after_dom_children.
4749         (build_single_def_use_links): Adjust.
4750         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4751         class.
4752         (find_candidates_in_block): Convert to
4753         find_candidates_dom_walker::before_dom_children.
4754         (execute_strength_reduction): Adjust.
4755         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4756         (sese_dom_walker): New class.
4757         (sese_dom_walker::sese_dom_walker): New constructor.
4758         (sese_dom_walker::~sese_dom_walker): New destructor.
4759         (build_sese_conditions_before): Convert to
4760         sese_dom_walker::before_dom_children.
4761         (build_sese_conditions_after): Convert to
4762         sese_dom_walker::after_dom_children.
4763         (build_poly_scop): Adjust
4764         * tree-into-ssa.c (rewrite_dom_walker): New class
4765         (rewrite_enter_block): Convert to
4766         rewrite_dom_walker::before_dom_children.
4767         (rewrite_leave_block): Convert to
4768         rewrite_dom_walker::after_dom_children.
4769         (rewrite_update_dom_walker): New class.
4770         (rewrite_update_enter_block): Convert to
4771         rewrite_update_dom_walker::before_dom_children.
4772         (rewrite_update_leave_block): Convert to
4773         rewrite_update_dom_walker::after_dom_children.
4774         (rewrite_blocks, rewrite_into_ssa): Adjust.
4775         (mark_def_dom_walker): New class.
4776         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4777         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4778         (mark_def_sites_blocks): Convert to
4779         mark_def_dom_walker::before_dom_children.
4780         (mark_def_site_blocks): Remove.
4781         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4782         (tree_ssa_dominator_optimize): Adjust.
4783         (dom_thread_across_edge): Convert to method
4784         dom_opt_dom_walker::thread_across_edge.
4785         (dom_opt_enter_block): Convert to member function
4786         dom_opt_dom_walker::before_dom_children.
4787         (dom_opt_leave_block): Convert to member function
4788         dom_opt_dom_walker::after_dom_children.
4789         * tree-ssa-dse.c (dse_dom_walker): New class.
4790         (dse_enter_block): Convert to member function
4791         dse_dom_walker::before_dom_children.
4792         (tree_ssa_dse): Adjust.
4793         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4794         (determine_invariantness_stmt): Convert to method
4795         invariantness_dom_walker::before_dom_children.
4796         (determine_invariantness): Remove
4797         (move_computations_dom_walker): New class.
4798         (move_computations_stmt): Convert to method
4799         move_computations_dom_walker::before_dom_children.
4800         (move_computations, tree_ssa_lim): Adjust.
4801         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4802         (nt_init_block): Make method
4803         notrappping_dom_walker::before_dom_children.
4804         (nt_fini_block): Make
4805         method nontrapping_dom_walker::after_dom_children.
4806         (get_non_trapping): Adjust.
4807         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4808         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4809         (eliminate_leave_block): Make method.
4810         eliminate_dom_walker::after_dom_children.
4811         (eliminate): Adjust
4812         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4813         (strlen_enter_block): Make method
4814         strlen_dom_walker::before_dom_children.
4815         (strlen_leave_block): Make
4816         method strlen_dom_walker::after_dom_children.
4817         (tree_ssa_strlen): Adjust.
4818         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4819         (tree_ssa_uncprop): Adjust.
4820         (uncprop_leave_block): Make method
4821         uncprop_dom_walker::after_dom_children.
4822         (uncprop_leave_block): Make method
4823         uncprop_dom_walker::before_dom_children.
4825 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4827         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4828         branch in same basic block.  Check both src and dest of the move insn.
4830 2013-09-17  Nick Clifton  <nickc@redhat.com>
4832         * config/rl78/rl78-real.md (bf): New pattern.
4833         (bt): New pattern.
4834         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4835         (rl78_print_operand): Do not put a # before a %B.
4836         * config/rl78/rl78.opt: Tweak doc strings.
4838 2013-09-17  DJ Delorie  <dj@redhat.com>
4840         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4841         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4842         (rl78_option_override): Likewise, if -mallregs.
4843         (is_virtual_register): Likewise.
4844         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4845         (REGNO_OK_FOR_BASE_P): Likewise.
4846         * config/rl78/rl78.opt (-mallregs): New.
4848 2013-09-17  Nick Clifton  <nickc@redhat.com>
4850         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4851         For interrupt functions: save all call clobbered registers if the
4852         interrupt handler is not a leaf function.
4853         (rl78_expand_prologue): Always recompute the frame information.
4854         For interrupt functions: only select bank 0 if one of the bank 0
4855         registers is going to be psuhed.
4857 2013-09-17  DJ Delorie  <dj@redhat.com>
4859         * config/rl78/constraints.md: For each W* constraint, rename to C*
4860         and create a W* constraint that checks for an optional ES: prefix
4861         pattern also.
4862         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4863         (es_addr): New.  Used to wrap far addresses.
4864         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4865         (rl78_es_base): New.
4866         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4867         wrapped far addresses.
4868         (rl78_print_operand_1): Unwrap far addresses before processing.
4869         (rl78_lo16): Wrap far addresses in unspecs.
4870         (rl78_es_addr): New.
4871         (rl78_es_base): New.
4872         (insn_ok_now): Check for not-yet-wrapped far addresses.
4873         (transcode_memory_rtx): Properly re-wrap far addresses.
4875 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4877         * config/sparc/t-rtems: Add leon3 multilibs.
4879 2013-09-17  Cong Hou  <congh@google.com>
4881         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4882         when checking the dot production pattern. The type of rhs operand
4883         of multiply is now checked correctly.
4885 2013-09-17  Jeff Law  <law@redhat.com>
4887         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4888         edge implied equivalences into successor phis.
4889         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4890         here from tree-ssa-threadedge.c.
4891         (mark_threaded_blocks): When threading through a joiner, if both
4892         successors of the joiner's clone reach the same block, verify the
4893         PHI arguments are equal.  If not, cancel the jump threading request.
4894         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4895         tree-ssa-threadupdate.c
4896         (thread_across_edge): Don't check PHI argument equality when
4897         threading through joiner block here.
4899 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4901         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4902         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4903         (warn_uninit, warn_uninitialized_vars,
4904         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4905         Move to tree-ssa-uninit.c.
4906         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4907         (has_undefined_value_p): New.  Pass dependant parts of
4908         ssa_undefined_value_p.
4909         (uninit_undefined_value_p): Use has_undefined_value_p.
4910         (warn_uninit, warn_uninitialized_vars,
4911         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4912         Move from tree-ssa.c.
4913         * tree-ssa.h: Adjust prototypes.
4915 2013-09-17  Jan Hubicka  <jh@suse.cz>
4917         PR middle-end/58332
4918         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4919         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4920         FUNCTION_NOT_OPTIMIZED.
4921         * ipa-inline-analysis.c (compute_inline_parameters): Function
4922         not optimized is not inlinable unless it is alwaysinline.
4923         (inline_analyze_function): Force calls in not optimized
4924         function not inlinable.
4926 2013-09-17  Jan Hubicka  <jh@suse.cz>
4928         PR middle-end/58329
4929         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4930         to return NULL.
4931         * ipa.c (function_and_variable_visibility): Likewise.
4932         * ipa-profile.c (ipa_profile): Likewise.
4934 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4936         PR ipa/58398
4937         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4938         attribute, and don't inline the resolver in this case.
4940 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4942         * coverage.c (get_coverage_counts): Add missing newline.
4944 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4946         PR tree-optimization/58088
4947         * fold-const.c (mask_with_trailing_zeros): New function.
4948         (fold_binary_loc): Make sure we don't recurse infinitely
4949         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
4950         Use mask_with_trailing_zeros where appropriate.
4952 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
4954         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
4955         of 'prev' var to get better distance estimation.
4957 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
4959         * tree-inline.h (struct copy_body_data): Add transform_parameter.
4960         * tree-inline.c (is_parameter_of): New predicate.
4961         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
4962         a parameter has been remapped.
4963         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
4964         (optimize_inline_calls): Initialize transform_parameter.
4965         (copy_gimple_seq_and_replace_locals): Likewise.
4966         (tree_function_versioning): Likewise.
4967         (maybe_inline_call_in_expr): Likewise.
4969 2013-09-17  Nick Clifton  <nickc@redhat.com>
4971         * config/msp430/msp430-protos.h: Add prototypes for new functions.
4972         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
4973         interrupt handlers.
4974         (is_attr_func): New function.
4975         (msp430_is_interrupt_func): New function.
4976         (is_naked_func): New function.
4977         (is_reentrant_func): New function.
4978         (is_critical_func): New function.
4979         (msp430_start_function): Add annotations for function attributes.
4980         (msp430_attr): New function.
4981         (msp430_attribute_table): New.
4982         (msp430_function_section): New function.
4983         (TARGET_ASM_FUNCTION_SECTION): Define.
4984         (msp430_builtin): New enum.
4985         (msp430_init_builtins): New function.
4986         (msp430_builtin_devl): New function.
4987         (msp430_expand_builtin): New function.
4988         (TARGET_INIT_BUILTINS): Define.
4989         (TARGET_EXPAND_BUILTINS): Define.
4990         (TARGET_BUILTIN_DECL): Define.
4991         (msp430_expand_prologue): Add support for naked, interrupt,
4992         critical and reentrant functions.
4993         (msp430_expand_epilogue): Likewise.
4994         (msp430_print_operand): Handle 'O' character.
4995         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
4996         NO_TRAMPOLINES.
4997         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
4998         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
4999         (pushm): Use a 'n' rather than an 'i' constraint.
5000         (msp_return): Add generation of the interrupt return instruction.
5001         (disable_interrupts): New pattern.
5002         (enable_interrupts): New pattern.
5003         (push_intr_state): New pattern.
5004         (pop_intr_state): New pattern.
5005         (bic_SR): New pattern.
5006         (bis_SR): New pattern.
5007         * doc/extend.texi: Document MSP430 function attributes and builtin
5008         functions.
5010 2013-09-17  Richard Biener  <rguenther@suse.de>
5012         PR tree-optimization/58432
5013         * tree-loop-distribution.c (tree_loop_distribution): Also
5014         scan PHIs for outside loop uses and seed a partition from them.
5016 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5018         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5019         (restructure_reference): Call backtrace_base_for_ref.
5021 2013-09-17  Alan Modra  <amodra@gmail.com>
5023         PR target/57589
5024         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5025         patch.
5027 2013-09-16  DJ Delorie  <dj@redhat.com>
5029         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5030         vregs location for RL78/G10.
5031         (rl78_expand_prologue): Avoid SEL on G10.
5032         (rl78_expand_epilogue): Likewise.
5033         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5034         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5035         __RL78_G10__ when appropriate.
5036         (ASM_SPEC): Pass -mg10 along to the assembler.
5037         * config/rl78/rl78.md (sel_rb): Disable for G10.
5038         * config/rl78/rl78.opt: Add -mg10 option.
5039         * config/rl78/t-rl78: Add -mg10 multilib.
5041 2013-09-16  Xinliang David Li  <davidxl@google.com>
5043         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5044         * omp-low.c (omp_max_vf): Ditto.
5045         (expand_omp_simd): Ditto.
5046         * tree-vectorizer.c (vectorize_loops): Ditto.
5047         (gate_vect_slp): Ditto.
5048         (gate_increase_alignment): Ditto.
5049         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5050         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5051         (gate_tree_vectorize): Name change.
5052         (tree_vectorize): Ditto.
5053         (pass_vectorize::gate): Call new function.
5054         (pass_vectorize::execute): Ditto.
5055         * opts.c: O3 default setting change.
5056         (finish_options): Check new flag.
5057         * doc/invoke.texi: Document new flags.
5058         * common.opt: New flags.
5060 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5062         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5063         (Condition Code): Don't reference it.
5064         * doc/tm.texi: Regenerate.
5066 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5068         PR middle-end/58418
5069         * lra-constraints.c (undo_optional_reloads): Consider all optional
5070         reload even if it did not get a hard reg.
5072 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5074         * dumpfile.c (dump_loc): Remove newline emission.
5075         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5076         emission to dump_printf_loc calls where missing.
5077         (vect_mark_for_runtime_alias_test): Ditto.
5078         (vect_analyze_data_ref_dependence): Ditto.
5079         (vect_analyze_data_ref_dependences): Ditto.
5080         (vect_slp_analyze_data_ref_dependence): Ditto.
5081         (vect_slp_analyze_data_ref_dependences): Ditto.
5082         (vect_compute_data_ref_alignment): Ditto.
5083         (vect_update_misalignment_for_peel): Ditto.
5084         (vect_verify_datarefs_alignment): Ditto.
5085         (vector_alignment_reachable_p): Ditto.
5086         (vect_get_data_access_cost): Ditto.
5087         (vect_enhance_data_refs_alignment): Ditto.
5088         (vect_find_same_alignment_drs): Ditto.
5089         (vect_analyze_data_refs_alignment): Ditto.
5090         (vect_analyze_group_access): Ditto.
5091         (vect_analyze_data_ref_access): Ditto.
5092         (vect_analyze_data_ref_accesses): Ditto.
5093         (vect_prune_runtime_alias_test_list): Ditto.
5094         (vect_analyze_data_refs): Ditto.
5095         (vect_create_addr_base_for_vector_ref): Ditto.
5096         (vect_create_data_ref_ptr): Ditto.
5097         (vect_grouped_store_supported): Ditto.
5098         (vect_grouped_load_supported): Ditto.
5099         * value-prof.c (check_counter): Ditto.
5100         (check_ic_target): Ditto.
5101         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5102         (vect_recog_widen_mult_pattern): Ditto.
5103         (vect_recog_widen_sum_pattern): Ditto.
5104         (vect_recog_over_widening_pattern): Ditto.
5105         (vect_recog_widen_shift_pattern): Ditto.
5106         (vect_recog_rotate_pattern): Ditto.
5107         (vect_recog_vector_vector_shift_pattern): Ditto.
5108         (vect_recog_divmod_pattern): Ditto.
5109         (vect_recog_mixed_size_cond_pattern): Ditto.
5110         (vect_recog_bool_pattern): Ditto.
5111         (vect_pattern_recog_1): Ditto.
5112         (vect_pattern_recog): Ditto.
5113         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5114         (vect_is_simple_iv_evolution): Ditto.
5115         (vect_analyze_scalar_cycles_1): Ditto.
5116         (vect_get_loop_niters): Ditto.
5117         (vect_analyze_loop_1): Ditto.
5118         (vect_analyze_loop_form): Ditto.
5119         (vect_analyze_loop_operations): Ditto.
5120         (vect_analyze_loop_2): Ditto.
5121         (vect_analyze_loop): Ditto.
5122         (report_vect_op): Ditto.
5123         (vect_is_slp_reduction): Ditto.
5124         (vect_is_simple_reduction_1): Ditto.
5125         (vect_get_known_peeling_cost): Ditto.
5126         (vect_estimate_min_profitable_iters): Ditto.
5127         (vect_model_reduction_cost): Ditto.
5128         (vect_model_induction_cost): Ditto.
5129         (get_initial_def_for_induction): Ditto.
5130         (vect_create_epilog_for_reduction): Ditto.
5131         (vectorizable_reduction): Ditto.
5132         (vectorizable_induction): Ditto.
5133         (vectorizable_live_operation): Ditto.
5134         (vect_loop_kill_debug_uses): Ditto.
5135         (vect_transform_loop): Ditto.
5136         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5137         (vect_stmt_relevant_p): Ditto.
5138         (process_use): Ditto.
5139         (vect_mark_stmts_to_be_vectorized): Ditto.
5140         (vect_model_simple_cost): Ditto.
5141         (vect_model_promotion_demotion_cost): Ditto.
5142         (vect_model_store_cost): Ditto.
5143         (vect_get_store_cost): Ditto.
5144         (vect_model_load_cost): Ditto.
5145         (vect_get_load_cost): Ditto.
5146         (vect_init_vector_1): Ditto.
5147         (vect_get_vec_def_for_operand): Ditto.
5148         (vect_finish_stmt_generation): Ditto.
5149         (vectorizable_call): Ditto.
5150         (vectorizable_conversion): Ditto.
5151         (vectorizable_assignment): Ditto.
5152         (vectorizable_shift): Ditto.
5153         (vectorizable_operation): Ditto.
5154         (vectorizable_store): Ditto.
5155         (vectorizable_load): Ditto.
5156         (vectorizable_condition): Ditto.
5157         (vect_analyze_stmt): Ditto.
5158         (vect_transform_stmt): Ditto.
5159         (vect_is_simple_use): Ditto.
5160         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5161         (vect_can_advance_ivs_p): Ditto.
5162         (vect_update_ivs_after_vectorizer): Ditto.
5163         (vect_do_peeling_for_loop_bound): Ditto.
5164         (vect_gen_niters_for_prolog_loop): Ditto.
5165         (vect_update_inits_of_drs): Ditto.
5166         (vect_create_cond_for_alias_checks): Ditto.
5167         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5168         (vect_build_slp_tree_1): Ditto.
5169         (vect_supported_load_permutation_p): Ditto.
5170         (vect_analyze_slp_instance): Ditto.
5171         (vect_analyze_slp): Ditto.
5172         (vect_make_slp_decision): Ditto.
5173         (vect_detect_hybrid_slp): Ditto.
5174         (vect_bb_vectorization_profitable_p): Ditto.
5175         (vect_slp_analyze_bb_1): Ditto.
5176         (vect_update_slp_costs_according_to_vf): Ditto.
5177         (vect_get_mask_element): Ditto.
5178         (vect_transform_slp_perm_load): Ditto.
5179         (vect_schedule_slp_instance): Ditto.
5180         (vect_schedule_slp): Ditto.
5181         (vect_slp_transform_bb): Ditto.
5182         * profile.c (read_profile_edge_counts): Ditto.
5183         (compute_branch_probabilities): Ditto.
5184         * coverage.c (get_coverage_counts): Ditto.
5186 2013-09-16  Diego Novillo  <dnovillo@google.com>
5188         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5190 2013-09-16  Jan Hubicka  <jh@suse.cz>
5192         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5193         abstract functions; for static functions check the presence of body.
5195 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5197         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5198         * config/aarch64/aarch64-simd.md
5199         (aarch64_mla_elt<mode>): New.
5200         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5201         (aarch64_mls_elt<mode>): Likewise.
5202         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5203         (aarch64_fma4_elt<mode>): Likewise.
5204         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5205         (aarch64_fma4_elt_to_128v2df): Likewise.
5206         (aarch64_fma4_elt_to_64df): Likewise.
5207         (fnma<mode>4): Likewise.
5208         (aarch64_fnma4_elt<mode>): Likewise.
5209         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5210         (aarch64_fnma4_elt_to_128v2df): Likewise.
5211         (aarch64_fnma4_elt_to_64df): Likewise.
5212         * config/aarch64/iterators.md (VDQSF): New.
5213         * config/aarch64/arm_neon.h
5214         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5215         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5217 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5219         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5220         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5221         (aarch64_mul3_elt_to_128df): Likewise.
5222         (aarch64_mul3_elt_to_64v2df): Likewise.
5223         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5224         (VMUL): New.
5225         (VMUL_CHANGE_NLANES) Likewise.
5226         (h_con): Likewise.
5227         (f): Likewise.
5228         * config/aarch64/arm_neon.h
5229         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5231 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5233         * config/aarch64/arm_neon.h
5234         (vcvtx_high_f32_f64): Fix parameters.
5236 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5237             Uros Bizjak  <ubizjak@gmail.com>
5239         * config/alpha.c: Include tree-ssanames.h.
5241 2013-09-16  Richard Biener  <rguenther@suse.de>
5243         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5244         (dot_rdg_1): Handle control_dd.
5245         (create_edge_for_control_dependence): New function.
5246         (create_rdg_edges): Add control dependences if asked for.
5247         (build_rdg): Likewise.
5248         (generate_loops_for_partition): If there are not necessary
5249         control stmts remove all their dependencies.
5250         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5251         (distribute_loop): Pass on control dependences.
5252         (tree_loop_distribution): Compute control dependences and remove
5253         restriction on number of loop nodes.
5255 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5257         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5258         for internal calls.
5260 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5262         * cse.c (try_const_anchors): Punt on CC modes.
5264 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5266         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5268 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5270         PR target/58382
5271         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5272         calls to word_mode.
5274 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5276         PR target/48094
5277         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5278         seen.
5279         (darwin_objc1_section): Likewise.
5280         (darwin_file_end): Emit Image Info section when required.
5282 2013-09-14  Jan Hubicka  <jh@suse.cz>
5284         * tree-into-ssa.c (gate_into_ssa): New.
5285         (pass_data_build_ssa): Use it.
5286         * cgraph.h (expand_thunk): Update prototype.
5287         * cgraphunit.c (analyze_function): Expand thunks early.
5288         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5289         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5290         set lowered flag; do not add new function.
5291         (assemble_thunks_and_aliases): Update.
5292         * tree-ssa.c (gate_init_datastructures): New gate.
5293         (pass_data_init_datastructures): Use it.
5295 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5297         PR target/58269
5298         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5299         xmm register set described in the psABI.
5301 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5303         * dwarf2out.c (should_emit_struct_debug): Add check
5304         for type_decl variable is not NULL.
5306 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5308         * config.gcc: Use new winnt-c.c target hooks
5309         * config/t-winnt: New file
5310         * config/winnt-c.c: New file
5311         * doc/tm.texi.in: Document new hook
5312         * doc/tm.texi: Regenerated
5314 2013-09-13  Jan Hubicka  <jh@suse.cz>
5316         PR middle-end/58094
5317         * ipa-inline.c (check_callers): New function.
5318         (check_caller_edge): Remove.
5319         (want_inline_function_to_all_callers_p): Also permit alises that are
5320         called dirrectly.
5321         (inline_to_all_callers): Terminate the walk when devirtualization
5322         introduce new calls.
5324 2013-09-13  Jan Hubicka  <jh@suse.cz>
5326         * ipa-inline-analysis.c (struct growth_data): Add node.
5327         (do_estimate_growth_1): Fix detection of recursion.
5329 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5331         PR tree-optimization/58392
5332         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5333         to avoid shadowing of outer loop variable.  If
5334         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5335         replace_by_duplicate_decl simduid of loops that have it set and
5336         set dest_cfun->has_simduid_loops and/or
5337         dest_cfun->has_force_vect_loops.
5338         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5339         instead of maybe_lookup_decl.
5340         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5341         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5342         call.  Copy over force_vect and copy and remap simduid.  Set
5343         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5344         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5345         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5346         caller.  Don't set cfun->has_simduid_loops and/or
5347         cfun->has_force_vect_loops here.
5348         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5349         caller.
5350         (expand_call_inline, tree_function_versioning): Adjust copy_body
5351         callers.
5353 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5355         PR bootstrap/58388
5356         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5357         the assert if the edge was a speculative one.
5359 2013-09-13  Richard Biener  <rguenther@suse.de>
5361         * tree-data-ref.h (known_dependences_p): Move here ...
5362         * tree-loop-distribution.c (known_dependences_p): ... from here.
5363         (dump_rdg_component, debug_rdg_component): Remove.
5364         (dump_rdg): Adjust.
5365         (generate_loops_for_partition): Use gimple_uid instead of
5366         relying on matching stmt visit order.
5367         (rdg_build_partitions): Take starting stmt vector.
5368         (ldist_gen): Merge into ...
5369         (distribute_loop): ... this function.  Do not compute starting
5370         vertices vector.
5371         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5373 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5375         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5376         Set type attribute correctly. Set predicable_short_it attribute.
5377         (cmpsi_shiftsi): Remove %? from output template.
5379 2013-09-13  Richard Biener  <rguenther@suse.de>
5381         * tree-loop-distribution.c (struct rdg_component,
5382         rdg_defs_used_in_other_loops_p, free_rdg_components,
5383         rdg_build_components): Remove.
5384         (stmts_from_loop): Do not record virtual PHIs.
5385         (generate_loops_for_partition): Skip virtual PHIs.
5386         (build_rdg_partition_for_component): Rename to ...
5387         (build_rdg_partition_for_vertex): ... this and adjust.
5388         (rdg_build_partitions): Take a vector of starting vertices
5389         instead of components.  Remove unnecessary leftover handling.
5390         (ldist_gen): Do not build components or record other stores.
5391         (distribute_loop): Do not distribute loops containing stmts
5392         with side-effects.
5394 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5396         PR target/58314
5397         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5399 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5401         * config.gcc: Separate cases for mingw and cygwin targets,
5402         and add 64-bit cygwin target case.
5404         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5405         dll-export inline-functions.
5406         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5409 2013-09-13  Jeff Law  <law@redhat.com>
5411         PR middle-end/58387
5412         Revert:
5413         2013-09-06  Jeff Law  <law@redhat.com>
5415         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5416         edge implied equivalences into successor phis.
5418 2013-09-12  DJ Delorie  <dj@redhat.com>
5420         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5421         separators.
5423 2013-09-12  Brooks Moses  <bmoses@google.com>
5425         PR driver/42955
5426         * Makefile.in: Do not install driver binaries in $(target)/bin.
5428 2013-09-12  DJ Delorie  <dj@redhat.com>
5430         * config/rl78/rl78.opt (mrelax): New.
5431         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5432         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5434         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5435         between SP and FP.
5436         (rl78_expand_epilogue): Likewise.
5438 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5440         PR middle-end/58335
5441         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5442         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5443         pointer elimination with using remove_reg_equal_offset_note.
5445 2013-09-12  DJ Delorie  <dj@redhat.com>
5447         * config/msp430/: New port.
5448         * config.gcc (msp430): Added.
5449         * doc/invoke.texi: Document MSP430 options.
5450         * doc/install.texi: Document msp430-elf
5451         * doc/md.texi: Document msp430-elf
5452         * doc/contrib.texi: Document msp430-elf
5454         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5455         PSImode.
5457 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5459         PR ipa/58389
5460         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5461         reference descriptor is NULL.
5462         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5463         edge to NULL.
5465 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5467         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5468         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5469         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5470         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5471         make_temp_ssa_name): move to tree-ssanames.h
5472         * tree-ssa-alias.h: Move prototype.
5473         * tree-ssa.h: Include tree-ssanames.h.
5474         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5475         * tree-ssanames.h: New.  Move items from tree-flow*.h
5476         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5478 2013-09-12  Richard Biener  <rguenther@suse.de>
5480         PR tree-optimization/58404
5481         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5482         propagate non-invariant addresses into dereferences wrapped
5483         in component references.
5485 2013-09-12  Richard Biener  <rguenther@suse.de>
5487         PR tree-optimization/58402
5488         * passes.def: Move pass_late_warn_uninitialized later.
5490 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5492         * tree-ssa.h: New.  Move content from tree-flow.h and
5493         tree-flow-inline.h.
5494         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5495         Move prototypes belonging to tree-ssa.c.
5496         * tree-flow-inline.h (redirect_edge_var_map_def,
5497         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5498         tree-ssa.h.
5499         * gimple.h: Adjust prototypes.
5500         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5501         to...
5502         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5503         * tree.h: Move prototype to tree-ssa.h.
5504         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5505         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5506         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5507         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5508         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5509         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5510         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5511         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5512         graphite-blocking.c, graphite-clast-to-gimple.c,
5513         graphite-dependences.c, graphite-interchange.c,
5514         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5515         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5516         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5517         ipa-reference.c, ipa-split.c, ipa-utils.c,
5518         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5519         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5520         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5521         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5522         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5523         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5524         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5525         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5526         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5527         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5528         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5529         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5530         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5531         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5532         value-prof.c, var-tracking.c,
5533         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5535 2013-09-12  Richard Biener  <rguenther@suse.de>
5537         PR tree-optimization/58396
5538         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5539         (build_rdg): Take a loop-nest parameter, fix memleaks.
5540         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5542 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5544         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5545         for SLM.
5547 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5549         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5550         The FMA instruction names should have a 'v' prefix.
5552 2013-09-12  Richard Biener  <rguenther@suse.de>
5554         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5555         (dot_rdg): Use popen instead of system in optional code.
5556         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5557         (already_processed_vertex_p): Adjust.
5558         (has_anti_or_output_dependence, predecessor_has_mem_write,
5559         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5560         rdg_flag_uses): Remove.
5561         (rdg_flag_vertex): Simplify.
5562         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5563         remove recursion.
5564         (build_rdg_partition_for_component): Process the first vertex
5565         of a component only.
5566         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5568 2013-09-12  Alan Modra  <amodra@gmail.com>
5570         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5572 2013-09-11  DJ Delorie  <dj@redhat.com>
5573             Nick Clifton  <nickc@redhat.com>
5575         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5576         (rl78_stack_based_mem): New.
5577         * config/rl78/constraints.md (Iv08): New.
5578         (Iv16): New.
5579         (Iv24): New.
5580         (Is09): New.
5581         (Is17): New.
5582         (Is25): New.
5583         (ISsi): New.
5584         (IShi): New.
5585         (ISqi): New.
5586         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5587         (movhi): Likewise.
5588         (movsi): Change from expand to insn-and-split.
5589         (ashrsi3): Clobber AX.
5590         (lshrsi3): New.
5591         (ashlsi3): New.
5592         (cbranchsi4): New.
5593         * config/rl78/rl78.md (CC_REG): Fix.
5594         (addsi3): Allow memory and immediate operands.
5595         (addsi3_internal): Split into...
5596         (addsi3_internal_virt): ...new, and ...
5597         (addsi3_internal_real): ...new.
5598         (subsi): New.
5599         (subsi3_internal_virt): New.
5600         (subsi3_internal_real): New.
5601         (mulsi3): Add memory operand.
5602         (mulsi3_rl78): Likewise.
5603         (mulsi3_g13): Likewise.
5604         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5605         (cbranchqi4_real): Add more constraint options.
5606         (cbranchhi4_real): Expand pattern.
5607         (cbranchhi4_real_signed): New.
5608         (cbranchhi4_real_inverted): New.
5609         (cbranchsi4_real_lt): New.
5610         (cbranchsi4_real_ge): New.
5611         (cbranchsi4_real_signed): New.
5612         (cbranchsi4_real): New.
5613         (peephole2): New.
5614         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5615         constant shifts.
5616         (lshrsi3_virt): Likewise.
5617         (ashlsi3_virt): Likewise.
5618         (cbranchqi4_virt_signed): New.
5619         (cbranchhi4_virt_signed): New.
5620         (cbranchsi4_virt): New.
5621         * config/rl78/rl78.c: Whitespace fixes throughout.
5622         (move_elim_pass): New.
5623         (pass_data_rl78_move_elim): New.
5624         (pass_rl78_move_elim): New.
5625         (make_pass_rl78_move_elim): New.
5626         (rl78_devirt_info): Run devirt earlier.
5627         (rl78_move_elim_info): New.
5628         (rl78_asm_file_start): Register it.
5629         (rl78_split_movsi): New.
5630         (rl78_as_legitimate_address): Allow virtual base registers when
5631         appropriate.
5632         (rl78_addr_space_convert): Remove spurious debug stuff.
5633         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5634         (rl78_print_operand): More cases for not printing '#'.
5635         (rl78_expand_compare): Remove most of the logic.
5636         (content_memory): New.
5637         (clear_content_memory): New.
5638         (get_content_index): New.
5639         (get_content_name): New.
5640         (display_content_memory): New.
5641         (update_content): New.
5642         (record_content): New.
5643         (already_contains): New.
5644         (insn_ok_now): Re-recog insns with virtual registers.
5645         (add_postponed_content_update): New.
5646         (process_postponed_content_update): New.
5647         (gen_and_emit_move): New.
5648         (transcode_memory_rtx): Record new location content.
5649         Use gen_and_emit_move.
5650         (force_into_acc): New.
5651         (move_to_acc): Use gen_and_emit_move.
5652         (move_from_acc): Likewise.
5653         (move_acc_to_reg): Likewise.
5654         (move_to_x): Likewise.
5655         (move_to_hl): Likewise.
5656         (move_to_de): Likewise.
5657         (rl78_alloc_physical_registers_op1): Record location content.
5658         (has_constraint): New.
5659         (rl78_alloc_physical_registers_op2): Record location content.
5660         Optimize use of HL.
5661         (rl78_alloc_physical_registers_ro1): Likewise.
5662         (rl78_alloc_physical_registers_cmp): Likewise.
5663         (rl78_alloc_physical_registers_umul): Likewise.
5664         (rl78_alloc_address_registers_macax): New.
5665         (rl78_alloc_physical_registers): Initialize and set location
5666         content memory as needed.
5667         (rl78_reorg): Make sure split2 is called.
5668         (rl78_rtx_costs): New.
5670 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5672         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5673         for (not (neg ...)) and (neg (not ...)) cases.
5675 2013-09-11  Richard Biener  <rguenther@suse.de>
5677         PR middle-end/58377
5678         * passes.def: Split critical edges before late uninit warning passes.
5679         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5681 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5683         PR tree-optimization/58385
5684         * fold-const.c (build_range_check): If both low and high are NULL,
5685         use omit_one_operand_loc to preserve exp side-effects.
5687 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5689         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5691 2013-09-11  Richard Biener  <rguenther@suse.de>
5693         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5694         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5695         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5696         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5697         stmts_from_loop, known_dependences_p, build_empty_rdg,
5698         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5699         * tree-loop-distribution.c: ... here.
5700         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5701         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5702         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5703         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5704         * tree-loop-distribution.c: ... here.
5705         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5706         (struct partition_s): Add loops member.
5707         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5708         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5709         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5711 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5712             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5713             Sergey Lega  <sergey.s.lega@intel.com>
5714             Anna Tikhonova  <anna.tikhonova@intel.com>
5715             Ilya Tocar  <ilya.tocar@intel.com>
5716             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5717             Ilya Verbin  <ilya.verbin@intel.com>
5718             Kirill Yukhin  <kirill.yukhin@intel.com>
5719             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5721         * config/i386/constraints.md (k): New.
5722         (Yk): Ditto.
5723         * config/i386/i386.c (const regclass_map): Add new mask registers.
5724         (dbx_register_map): Ditto.
5725         (dbx64_register_map): Ditto.
5726         (svr4_dbx_register_map): Ditto.
5727         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5728         disabled.
5729         (ix86_preferred_reload_class): Disable constants for mask registers.
5730         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5731         (ix86_hard_regno_mode_ok): Support new mask registers.
5732         (x86_order_regs_for_local_alloc): Ditto.
5733         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5734         (FIXED_REGISTERS): Add new mask registers.
5735         (CALL_USED_REGISTERS): Ditto.
5736         (REG_ALLOC_ORDER): Ditto.
5737         (VALID_MASK_REG_MODE): New.
5738         (FIRST_MASK_REG): Ditto.
5739         (LAST_MASK_REG): Ditto.
5740         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5741         (MAYBE_MASK_CLASS_P): New.
5742         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5743         (REG_CLASS_CONTENTS): Ditto.
5744         (MASK_REGNO_P): New.
5745         (ANY_MASK_REG_P): Ditto.
5746         (HI_REGISTER_NAMES): Add new mask registers.
5747         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5748         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5749         mask registers.
5750         (attribute "type"): Add mskmov, msklog.
5751         (attribute "length_immediate"): Support them.
5752         (attribute "memory"): Ditto.
5753         (attribute "prefix_0f"): Ditto.
5754         (*movhi_internal): Support new mask registers.
5755         (*movqi_internal): Ditto.
5756         (define_split): Split out clobber pattern is a logic
5757         insn on mask registers.
5758         (*k<logic><mode>): New.
5759         (*andhi_1): Extend to support mask regs.
5760         (*andqi_1): Extend to support mask regs.
5761         (kandn<mode>): New.
5762         (define_split): Split and-not to and and not if operands
5763         are not mask regs.
5764         (*<code><mode>_1): Separate HI mode to new pattern...
5765         (*<code>hi_1): This.
5766         (*<code>qi_1): Extend to support mask regs.
5767         (kxnor<mode>): New.
5768         (kortestzhi): Ditto.
5769         (kortestchi): Ditto.
5770         (kunpckhi): Ditto.
5771         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5772         (*one_cmplhi2_1): ...Here, now with mask registers support.
5773         (*one_cmplqi2_1): Support new mask registers.
5774         (HI/QImode arithmetics splitter): Don't split if mask registers
5775         are used.
5776         (HI/QImode not splitter): Ditto.
5777         * config/i386/predicated.md (mask_reg_operand): New.
5778         (general_reg_operand): Ditto.
5780 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5782         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5783         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5785 2013-09-10  Jeff Law  <law@redhat.com>
5787         PR tree-optimization/58380
5788         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5789         of threading through a buried loop header.
5791         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5792         return value for single successor case.
5794 2013-09-10  Jan Hubicka  <jh@suse.cz>
5796         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5798 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5800         PR target/58361
5801         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5802         support conditional execution.
5803         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5805 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5807         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5808         constraint pass.
5809         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5810         Check destination too to check move insn.
5811         (undo_optional_reloads): Add check that the original peudo did not
5812         changed its allocation and the optional reload was inherited on last
5813         inheritance pass.  Break loop after deciding to keep optional reload.
5814         (lra_undo_inheritance): Add check that inherited pseudo still in
5815         memory.
5817 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5819         * config/aarch64/aarch64.md (generic_sched): New.
5820         * config/aarch64/aarch64-generic.md (load): Make conditional
5821         on generic_sched attribute.
5822         (nonload): Likewise.
5824 2013-09-10  Jan Hubicka  <jh@suse.cz>
5826         * lto-cgraph.c: Include ipa-utils.h.
5827         (compute_ltrans_boundary): Also add possible targets into the boundary.
5829 2013-09-10  Jan Hubicka  <jh@suse.cz>
5831         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5832         VAR_DECL of vtable rather than full expression.
5834 2013-09-10  Jan Hubicka  <jh@suse.cz>
5835             Paolo Carlini  <paolo.carlini@oracle.com>
5837         * cgraphunit.c (analyze_functions): Save input_location, set it
5838         to UNKNOWN_LOCATION and restore it at the end.
5840 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5842         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5843         represented by a thunk.
5845 2013-09-10  Jeff Law  <law@redhat.com>
5847         PR tree-optimization/58343
5848         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5849         jump threading requests through loop headers buried in the middle
5850         of a jump threading path.
5852         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5853         in return value/type.
5855 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5857         PR rtl-optimization/58365
5858         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5859         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5860         it differs.
5862 2013-09-10  Richard Biener  <rguenther@suse.de>
5864         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5865         * tree-data-ref.c (create_rdg_vertices): Collect all data
5866         references, signal failure to the caller, use data-ref API.
5867         (build_rdg): Compute data references only once.  Maintain lifetime
5868         of data references and data dependences from within RDG.
5869         (free_rdg): Free dependence relations.
5870         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5871         inventing extra dependences.
5872         (distribute_loop): Update for RDG API changes.
5874 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5876         * doc/invoke.texi (fms-extensions): Document changed
5877         behavior for ms-abi targets.
5878         * config/i386/i386.c (ix86_option_override_internal):
5879         Set default value of option -fms-extension for ms-abi targets.
5881 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5883         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5885 2013-09-10  Alan Modra  <amodra@gmail.com>
5887         PR target/58330
5888         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5890 2013-09-10  Alan Modra  <amodra@gmail.com>
5892         * config/rs6000/predicates.md (add_cint_operand): New.
5893         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5894         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5895         using add_cint_operand.
5896         (largetoc_high_plus_aix): Likewise.
5898 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5900         PR tree-optimization/58364
5901         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5902         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5903         the current range can't be an unconditional true or false.
5905 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5907         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5909 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5911         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5913 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5915         PR c++/43452
5916         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5918 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5920         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5921         NO_REGS for immediate that can't be moved directly into FP_REGS.
5923 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5925         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5926         comparison with negated operand.
5927         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5928         RTL form.
5930 2013-09-09  Richard Biener  <rguenther@suse.de>
5932         PR middle-end/58326
5933         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5934         of a subloop record all its block as affecting loop-closed SSA form.
5936 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5938         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5939         of an rtx/bitpos pair.
5940         (store_fixed_bit_field): Update accordingly.
5942 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5944         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5945         GEN_INT.
5946         * builtins.c (expand_errno_check): Likewise.
5947         * dwarf2cfi.c (init_return_column_size): Likewise.
5948         * except.c (sjlj_mark_call_sites): Likewise.
5949         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
5950         * lra-constraints.c (emit_inc): Likewise.
5951         * ree.c (combine_set_extension): Likewise.
5952         * regmove.c (fixup_match_2): Likewise.
5953         * reload1.c (inc_for_reload): Likewise.
5955 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5957         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
5958         (force_to_mode, simplify_shift_const_1, simplify_comparison):
5959         Use gen_int_mode with the mode of the associated simplify_* call.
5960         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
5961         * expmed.c (expand_shift_1): Likewise.
5962         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5963         * loop-iv.c (iv_number_of_iterations): Likewise.
5964         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5965         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
5967 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5969         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
5970         of the associated expand_* call.
5971         (asan_emit_stack_protection): Likewise.
5972         * builtins.c (round_trampoline_addr): Likewise.
5973         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
5974         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
5975         (emit_store_flag): Likewise.
5976         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
5977         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
5978         Likewise.
5979         * function.c (instantiate_virtual_regs_in_insn): Likewise.
5980         * ifcvt.c (noce_try_store_flag_constants): Likewise.
5981         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
5982         * modulo-sched.c (generate_prolog_epilog): Likewise.
5983         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
5984         (expand_ctz, expand_ffs, expand_unop): Likewise.
5986 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5988         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
5989         of the associated gen_rtx_* call.
5990         * caller-save.c (init_caller_save): Likewise.
5991         * combine.c (find_split_point, make_extraction): Likewise.
5992         (make_compound_operation): Likewise.
5993         * dwarf2out.c (mem_loc_descriptor): Likewise.
5994         * explow.c (plus_constant, probe_stack_range): Likewise.
5995         * expmed.c (expand_mult_const): Likewise.
5996         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
5997         * reload1.c (init_reload): Likewise.
5998         * valtrack.c (cleanup_auto_inc_dec): Likewise.
5999         * var-tracking.c (adjust_mems): Likewise.
6000         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6001         rather than gen_rtx_fmt_ee.
6003 2013-09-09  Jan Hubicka  <jh@suse.cz>
6005         PR middle-end/58294
6006         * value-prof.c (gimple_ic): Copy also abnormal edges.
6008 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6010         * asan.c (asan_shadow_cst): Use gen_int_mode.
6012 2013-09-08  Jan Hubicka  <jh@suse.cz>
6014         * ipa-profile.c: Add toplevel comment.
6015         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6016         (contains_hot_call_p): New function.
6017         (ipa_propagate_frequency): Set frequencies based on counts when
6018         profile is read.
6019         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6020         profile; do not tamper with profile after inlining if it is read.
6022 2013-09-08  Jan Hubicka  <jh@suse.cz>
6024         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6025         speculative edges.
6027 2013-09-08  Jan Hubicka  <jh@suse.cz>
6029         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6030         summary generation.
6032 2013-09-08  Jeff Law  <law@redhat.com>
6034         PR bootstrap/58340
6035         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6036         of 'found'.
6038 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6040         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6042 2013-09-08  Jan Hubicka  <jh@suse.cz>
6044         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6046 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6048         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6049         for non-debug insns.
6050         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6051         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6052         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6053         accordingly.
6055 2013-09-08  Jan Hubicka  <jh@suse.cz>
6057         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6058         targets and devirtualize to BUILT_IN_UNREACHABLE.
6059         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6060         * ipa.c (walk_polymorphic_call_targets): New function.
6061         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6062         functions; use the new timevar.
6063         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6064         was removed from the program.
6065         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6066         not consider it in the walk when its vtable is dead.
6067         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6068         record_binfo.
6069         (devirt_variable_node_removal_hook): New function.
6070         (possible_polymorphic_call_targets): Also register
6071         devirt_variable_node_removal_hook.
6072         (ipa_devirt): Do not do non-speculative devirtualization.
6073         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6075 2013-09-08  Jan Hubicka  <jh@suse.cz>
6077         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6078         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6079         varpool_remove_variable_insertion_hook): Declare.
6080         * varpool.c (varpool_node_hook_list): New structure.
6081         (first_varpool_node_removal_hook,
6082         first_varpool_variable_insertion_hook): New variables.
6083         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6084         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6085         varpool_remove_variable_insertion_hook,
6086         varpool_call_variable_insertion_hooks): New functions.
6087         (varpool_remove_node): Use it.
6089 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6091         PR c++/54941
6092         * diagnostic.c (diagnostic_build_prefix): When s.file is
6093         "<built-in>" don't output line and column numbers.
6095 2013-09-06  Jan Hubicka  <jh@suse.cz>
6097         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6098         * lto-streamer-out.c: Stream thunks, too.
6099         * lto-streamer-in.c (input_function): Pop cfun here
6100         (lto_read_body): Instead of here.
6102 2013-09-06  Caroline Tice  <cmtice@google.com>
6104         * doc/install.texi: Add documentation for the --enable-vtable-verify
6105         and the --disable-libvtv configure options.
6107 2013-09-06  Jeff Law  <law@redhat.com>
6109         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6110         edge implied equivalences into successor phis.
6112 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6114         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6116 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6118         * resource.c (mark_target_live_regs): Compute resources taking
6119         into account if a call is predicated or not.
6121 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6123         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6124         the compiler in the function names.
6126 2013-09-06  Jan Hubicka  <jh@suse.cz>
6128         PR middle-end/58094
6129         * ipa-inline.c (has_caller_p): New function.
6130         (want_inline_function_to_all_callers_p): Use it.
6131         (sum_callers, inline_to_all_callers): Break out from ...
6132         (ipa_inline): ... here.
6134 2013-09-06  Jan Hubicka  <jh@suse.cz>
6136         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6137         only when AVX is enabled.
6139 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6141         * config/aarch64/aarch64.md
6142         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6143         is fpsimd_<load/store>2.
6144         (load_pair<mode>): Likewise.
6145         (store_pair<mode>): Likewise.
6147 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6149         * config/arm/types.md (type): Add "mrs" type.
6150         * config/aarch64/aarch64.md
6151         (aarch64_load_tp_hard): Make type "mrs".
6152         * config/arm/arm.md
6153         (load_tp_hard): Make type "mrs".
6154         * config/arm/cortex-a15.md: Update with new attributes.
6155         * config/arm/cortex-a5.md: Update with new attributes.
6156         * config/arm/cortex-a53.md: Update with new attributes.
6157         * config/arm/cortex-a7.md: Update with new attributes.
6158         * config/arm/cortex-a8.md: Update with new attributes.
6159         * config/arm/cortex-a9.md: Update with new attributes.
6160         * config/arm/cortex-m4.md: Update with new attributes.
6161         * config/arm/cortex-r4.md: Update with new attributes.
6162         * config/arm/fa526.md: Update with new attributes.
6163         * config/arm/fa606te.md: Update with new attributes.
6164         * config/arm/fa626te.md: Update with new attributes.
6165         * config/arm/fa726te.md: Update with new attributes.
6167 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6169         * config/aarch64/aarch64.md
6170         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6171         (*movtf_aarch64): Likewise.
6172         * config/arm/arm.md
6173         (thumb1_movdi_insn): Use "multiple" for type where more than one
6174         instruction is used for a move.
6175         (*arm32_movhf): Likewise.
6176         (*thumb_movdf_insn): Likewise.
6178 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6180         * config/arm/types.md (type): Rename fcpys to fmov.
6181         * config/arm/vfp.md
6182         (*arm_movsi_vfp): Rename type fcpys as fmov.
6183         (*thumb2_movsi_vfp): Likewise
6184         (*movhf_vfp_neon): Likewise
6185         (*movhf_vfp): Likewise
6186         (*movsf_vfp): Likewise
6187         (*thumb2_movsf_vfp): Likewise
6188         (*movsfcc_vfp): Likewise
6189         (*thumb2_movsfcc_vfp): Likewise
6190         * config/aarch64/aarch64-simd.md
6191         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6192         * config/aarch64/aarch64.md
6193         (*movsi_aarch64): Replace type mov_reg with fmovs.
6194         (*movdi_aarch64): Likewise
6195         (*movsf_aarch64): Likewise
6196         (*movdf_aarch64): Likewise
6197         * config/arm/arm.c
6198         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6199         * config/arm/iwmmxt.md
6200         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6201         * config/arm/arm1020e.md: Update with new attributes.
6202         * config/arm/cortex-a15-neon.md: Update with new attributes.
6203         * config/arm/cortex-a5.md: Update with new attributes.
6204         * config/arm/cortex-a53.md: Update with new attributes.
6205         * config/arm/cortex-a7.md: Update with new attributes.
6206         * config/arm/cortex-a8-neon.md: Update with new attributes.
6207         * config/arm/cortex-a9.md: Update with new attributes.
6208         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6209         * config/arm/cortex-r4f.md: Update with new attributes.
6210         * config/arm/marvell-pj4.md: Update with new attributes.
6211         * config/arm/vfp11.md: Update with new attributes.
6213 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6215         * config/aarch64/aarch64.md
6216         (*madd<mode>): Fix type attribute.
6217         (*maddsi_uxtw): Likewise.
6218         (*msub<mode>): Likewise.
6219         (*msubsi_uxtw): Likewise.
6220         (<su_optab>maddsidi4): Likewise.
6221         (<su_optab>msubsidi4): Likewise.
6223 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6225         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6226         * config/arm/arm.md (core_cycles): Remove fdiv.
6227         * config/arm/vfp.md:
6228         (*sqrtsf2_vfp): Update for attribute changes.
6229         (*sqrtdf2_vfp): Likewise.
6230         * config/aarch64/aarch64.md:
6231         (sqrt<mode>2): Update for attribute changes.
6232         * config/arm/arm1020e.md: Update with new attributes.
6233         * config/arm/cortex-a15-neon.md: Update with new attributes.
6234         * config/arm/cortex-a5.md: Update with new attributes.
6235         * config/arm/cortex-a53.md: Update with new attributes.
6236         * config/arm/cortex-a7.md: Update with new attributes.
6237         * config/arm/cortex-a8-neon.md: Update with new attributes.
6238         * config/arm/cortex-a9.md: Update with new attributes.
6239         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6240         * config/arm/cortex-r4f.md: Update with new attributes.
6241         * config/arm/marvell-pj4.md: Update with new attributes.
6242         * config/arm/vfp11.md: Update with new attributes.
6244 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6246         * config/arm/types.md
6247         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6248         * config/aarch64/aarch64.md
6249         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6250         new attributes.
6251         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6252         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6253         (float<GPI:mode><GPF:mode>2): Likewise.
6254         * config/arm/vfp.md
6255         (*truncsisf2_vfp): Update with new attributes.
6256         (*truncsidf2_vfp): Likewise.
6257         (fixuns_truncsfsi2): Likewise.
6258         (fixuns_truncdfsi2): Likewise.
6259         (*floatsisf2_vfp): Likewise.
6260         (*floatsidf2_vfp): Likewise.
6261         (floatunssisf2): Likewise.
6262         (floatunssidf2): Likewise.
6263         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6264         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6265         * config/arm/arm1020e.md: Update with new attributes.
6266         * config/arm/cortex-a15-neon.md: Update with new attributes.
6267         * config/arm/cortex-a5.md: Update with new attributes.
6268         * config/arm/cortex-a53.md: Update with new attributes.
6269         * config/arm/cortex-a7.md: Update with new attributes.
6270         * config/arm/cortex-a8-neon.md: Update with new attributes.
6271         * config/arm/cortex-a9.md: Update with new attributes.
6272         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6273         * config/arm/cortex-r4f.md: Update with new attributes.
6274         * config/arm/marvell-pj4.md: Update with new attributes.
6275         * config/arm/vfp11.md: Update with new attributes.
6277 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6279         * config/aarch64/arm_neon.h
6280         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6281         (vqtbx<1,2,3,4><q>_s8): Likewise.
6283 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6285         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6286         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6287         (add<mode>3): Add type attribute.
6288         (add<mode>3): Likewise.
6289         (usadd<mode>3): Likewise.
6290         (ssadd<mode>3): Likewise.
6291         (sub<mode>3): Likewise.
6292         (sub<mode>3): Likewise.
6293         (ussub<mode>3): Likewise.
6294         (sssub<mode>3): Likewise.
6295         (ssmulsa3): Likewise.
6296         (usmulusa3): Likewise.
6297         (arm_usatsihi): Likewise.
6298         * config/arm/vfp.md
6299         (*movdi_vfp): Add types for all instructions.
6300         (*movdi_vfp_cortexa8): Likewise.
6301         (*movhf_vfp_neon): Likewise.
6302         (*movhf_vfp): Likewise.
6303         (*movdf_vfp): Likewise.
6304         (*thumb2_movdf_vfp): Likewise.
6305         (*thumb2_movdfcc_vfp): Likewise.
6306         * config/arm/arm.md: Add type attribute to all insn patterns.
6307         (*thumb1_adddi3): Add type attribute.
6308         (*arm_adddi3): Likewise.
6309         (*adddi_sesidi_di): Likewise.
6310         (*adddi_zesidi_di): Likewise.
6311         (*thumb1_addsi3): Likewise.
6312         (addsi3_compare0): Likewise.
6313         (*addsi3_compare0_scratch): Likewise.
6314         (*compare_negsi_si): Likewise.
6315         (cmpsi2_addneg): Likewise.
6316         (*addsi3_carryin_<optab>): Likewise.
6317         (*addsi3_carryin_alt2_<optab>): Likewise.
6318         (*addsi3_carryin_clobercc_<optab>): Likewise.
6319         (*subsi3_carryin): Likewise.
6320         (*subsi3_carryin_const): Likewise.
6321         (*subsi3_carryin_compare): Likewise.
6322         (*subsi3_carryin_compare_const): Likewise.
6323         (*arm_subdi3): Likewise.
6324         (*thumb_subdi3): Likewise.
6325         (*subdi_di_zesidi): Likewise.
6326         (*subdi_di_sesidi): Likewise.
6327         (*subdi_zesidi_di): Likewise.
6328         (*subdi_sesidi_di): Likewise.
6329         (*subdi_zesidi_ze): Likewise.
6330         (thumb1_subsi3_insn): Likewise.
6331         (*arm_subsi3_insn): Likewise.
6332         (*anddi3_insn): Likewise.
6333         (*anddi_zesidi_di): Likewise.
6334         (*anddi_sesdi_di): Likewise.
6335         (*ne_zeroextracts): Likewise.
6336         (*ne_zeroextracts): Likewise.
6337         (*ite_ne_zeroextr): Likewise.
6338         (*ite_ne_zeroextr): Likewise.
6339         (*anddi_notdi_di): Likewise.
6340         (*anddi_notzesidi): Likewise.
6341         (*anddi_notsesidi): Likewise.
6342         (andsi_notsi_si): Likewise.
6343         (thumb1_bicsi3): Likewise.
6344         (*iordi3_insn): Likewise.
6345         (*iordi_zesidi_di): Likewise.
6346         (*iordi_sesidi_di): Likewise.
6347         (*thumb1_iorsi3_insn): Likewise.
6348         (*xordi3_insn): Likewise.
6349         (*xordi_zesidi_di): Likewise.
6350         (*xordi_sesidi_di): Likewise.
6351         (*arm_xorsi3): Likewise.
6352         (*andsi_iorsi3_no): Likewise.
6353         (*smax_0): Likewise.
6354         (*smax_m1): Likewise.
6355         (*arm_smax_insn): Likewise.
6356         (*smin_0): Likewise.
6357         (*arm_smin_insn): Likewise.
6358         (*arm_umaxsi3): Likewise.
6359         (*arm_uminsi3): Likewise.
6360         (*minmax_arithsi): Likewise.
6361         (*minmax_arithsi_): Likewise.
6362         (*satsi_<SAT:code>): Likewise.
6363         (arm_ashldi3_1bit): Likewise.
6364         (arm_ashrdi3_1bit): Likewise.
6365         (arm_lshrdi3_1bit): Likewise.
6366         (*arm_negdi2): Likewise.
6367         (*thumb1_negdi2): Likewise.
6368         (*arm_negsi2): Likewise.
6369         (*thumb1_negsi2): Likewise.
6370         (*negdi_extendsid): Likewise.
6371         (*negdi_zero_extend): Likewise.
6372         (*arm_abssi2): Likewise.
6373         (*thumb1_abssi2): Likewise.
6374         (*arm_neg_abssi2): Likewise.
6375         (*thumb1_neg_abss): Likewise.
6376         (one_cmpldi2): Likewise.
6377         (extend<mode>di2): Likewise.
6378         (*compareqi_eq0): Likewise.
6379         (*arm_extendhisi2addsi): Likewise.
6380         (*arm_movdi): Likewise.
6381         (*thumb1_movdi_insn): Likewise.
6382         (*arm_movt): Likewise.
6383         (*thumb1_movsi_insn): Likewise.
6384         (pic_add_dot_plus_four): Likewise.
6385         (pic_add_dot_plus_eight): Likewise.
6386         (tls_load_dot_plus_eight): Likewise.
6387         (*thumb1_movhi_insn): Likewise.
6388         (*thumb1_movsf_insn): Likewise.
6389         (*movdf_soft_insn): Likewise.
6390         (*thumb_movdf_insn): Likewise.
6391         (cbranchsi4_insn): Likewise.
6392         (cbranchsi4_scratch): Likewise.
6393         (*negated_cbranchsi4): Likewise.
6394         (*tbit_cbranch): Likewise.
6395         (*tlobits_cbranch): Likewise.
6396         (*tstsi3_cbranch): Likewise.
6397         (*cbranchne_decr1): Likewise.
6398         (*addsi3_cbranch): Likewise.
6399         (*addsi3_cbranch_scratch): Likewise.
6400         (*arm_cmpdi_insn): Likewise.
6401         (*arm_cmpdi_unsig): Likewise.
6402         (*arm_cmpdi_zero): Likewise.
6403         (*thumb_cmpdi_zero): Likewise.
6404         (*deleted_compare): Likewise.
6405         (*mov_scc): Likewise.
6406         (*mov_negscc): Likewise.
6407         (*mov_notscc): Likewise.
6408         (*cstoresi_eq0_thumb1_insn): Likewise.
6409         (cstoresi_nltu_thumb1): Likewise.
6410         (cstoresi_ltu_thu): Likewise.
6411         (thumb1_addsi3_addgeu): Likewise.
6412         (*arm_jump): Likewise.
6413         (*thumb_jump): Likewise.
6414         (*check_arch2): Likewise.
6415         (arm_casesi_internal): Likewise.
6416         (thumb1_casesi_dispatch): Likewise.
6417         (*arm_indirect_jump): Likewise.
6418         (*thumb1_indirect_jump): Likewise.
6419         (nop): Likewise.
6420         (*and_scc): Likewise.
6421         (*ior_scc): Likewise.
6422         (*compare_scc): Likewise.
6423         (*cond_move): Likewise.
6424         (*cond_arith): Likewise.
6425         (*cond_sub): Likewise.
6426         (*cmp_ite0): Likewise.
6427         (*cmp_ite1): Likewise.
6428         (*cmp_and): Likewise.
6429         (*cmp_ior): Likewise.
6430         (*ior_scc_scc): Likewise.
6431         (*ior_scc_scc_cmp): Likewise.
6432         (*and_scc_scc): Likewise.
6433         (*and_scc_scc_cmp): Likewise.
6434         (*and_scc_scc_nod): Likewise.
6435         (*negscc): Likewise.
6436         (movcond_addsi): Likewise.
6437         (movcond): Likewise.
6438         (*ifcompare_plus_move): Likewise.
6439         (*if_plus_move): Likewise.
6440         (*ifcompare_move_plus): Likewise.
6441         (*if_move_plus): Likewise.
6442         (*ifcompare_arith_arith): Likewise.
6443         (*if_arith_arith): Likewise.
6444         (*ifcompare_arith_move): Likewise.
6445         (*if_arith_move): Likewise.
6446         (*ifcompare_move_arith): Likewise.
6447         (*if_move_arith): Likewise.
6448         (*ifcompare_move_not): Likewise.
6449         (*if_move_not): Likewise.
6450         (*ifcompare_not_move): Likewise.
6451         (*if_not_move): Likewise.
6452         (*ifcompare_shift_move): Likewise.
6453         (*if_shift_move): Likewise.
6454         (*ifcompare_move_shift): Likewise.
6455         (*if_move_shift): Likewise.
6456         (*ifcompare_shift_shift): Likewise.
6457         (*ifcompare_not_arith): Likewise.
6458         (*ifcompare_arith_not): Likewise.
6459         (*if_arith_not): Likewise.
6460         (*ifcompare_neg_move): Likewise.
6461         (*if_neg_move): Likewise.
6462         (*ifcompare_move_neg): Likewise.
6463         (*if_move_neg): Likewise.
6464         (prologue_thumb1_interwork): Likewise.
6465         (*cond_move_not): Likewise.
6466         (*sign_extract_onebit): Likewise.
6467         (*not_signextract_onebit): Likewise.
6468         (stack_tie): Likewise.
6469         (align_4): Likewise.
6470         (align_8): Likewise.
6471         (consttable_end): Likewise.
6472         (consttable_1): Likewise.
6473         (consttable_2): Likewise.
6474         (consttable_4): Likewise.
6475         (consttable_8): Likewise.
6476         (consttable_16): Likewise.
6477         (*thumb1_tablejump): Likewise.
6478         (prefetch): Likewise.
6479         (force_register_use): Likewise.
6480         (thumb_eh_return): Likewise.
6481         (load_tp_hard): Likewise.
6482         (load_tp_soft): Likewise.
6483         (tlscall): Likewise.
6484         (*arm_movtas_ze): Likewise.
6485         (*arm_rev): Likewise.
6486         (*arm_revsh): Likewise.
6487         (*arm_rev16): Likewise.
6488         * config/arm/thumb2.md
6489         (*thumb2_smaxsi3): Likewise.
6490         (*thumb2_sminsi3): Likewise.
6491         (*thumb32_umaxsi3): Likewise.
6492         (*thumb2_uminsi3): Likewise.
6493         (*thumb2_negdi2): Likewise.
6494         (*thumb2_abssi2): Likewise.
6495         (*thumb2_neg_abss): Likewise.
6496         (*thumb2_movsi_insn): Likewise.
6497         (tls_load_dot_plus_four): Likewise.
6498         (*thumb2_movhi_insn): Likewise.
6499         (*thumb2_mov_scc): Likewise.
6500         (*thumb2_mov_negs): Likewise.
6501         (*thumb2_mov_negs): Likewise.
6502         (*thumb2_mov_nots): Likewise.
6503         (*thumb2_mov_nots): Likewise.
6504         (*thumb2_movsicc_): Likewise.
6505         (*thumb2_movsfcc_soft_insn): Likewise.
6506         (*thumb2_indirect_jump): Likewise.
6507         (*thumb2_and_scc): Likewise.
6508         (*thumb2_ior_scc): Likewise.
6509         (*thumb2_ior_scc_strict_it): Likewise.
6510         (*thumb2_cond_move): Likewise.
6511         (*thumb2_cond_arith): Likewise.
6512         (*thumb2_cond_ari): Likewise.
6513         (*thumb2_cond_sub): Likewise.
6514         (*thumb2_negscc): Likewise.
6515         (*thumb2_movcond): Likewise.
6516         (thumb2_casesi_internal): Likewise.
6517         (thumb2_casesi_internal_pic): Likewise.
6518         (*thumb2_alusi3_short): Likewise.
6519         (*thumb2_mov<mode>_shortim): Likewise.
6520         (*thumb2_addsi_short): Likewise.
6521         (*thumb2_subsi_short): Likewise.
6522         (thumb2_addsi3_compare0): Likewise.
6523         (*thumb2_cbz): Likewise.
6524         (*thumb2_cbnz): Likewise.
6525         (*thumb2_one_cmplsi2_short): Likewise.
6526         (*thumb2_negsi2_short): Likewise.
6527         (*orsi_notsi_si): Likewise.
6528         * config/arm/arm1020e.md: Update with new attributes.
6529         * config/arm/arm1026ejs.md: Update with new attributes.
6530         * config/arm/arm1136jfs.md: Update with new attributes.
6531         * config/arm/arm926ejs.md: Update with new attributes.
6532         * config/arm/cortex-a15.md: Update with new attributes.
6533         * config/arm/cortex-a5.md: Update with new attributes.
6534         * config/arm/cortex-a53.md: Update with new attributes.
6535         * config/arm/cortex-a7.md: Update with new attributes.
6536         * config/arm/cortex-a8.md: Update with new attributes.
6537         * config/arm/cortex-a9.md: Update with new attributes.
6538         * config/arm/cortex-m4.md: Update with new attributes.
6539         * config/arm/cortex-r4.md: Update with new attributes.
6540         * config/arm/fa526.md: Update with new attributes.
6541         * config/arm/fa606te.md: Update with new attributes.
6542         * config/arm/fa626te.md: Update with new attributes.
6543         * config/arm/fa726te.md: Update with new attributes.
6545 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6547         * config/aarch64/aarch64-simd.md
6548         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6549         <vwx> iterator to ensure correct register choice.
6550         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6551         (aarch64_sqdmull_n<mode>): Likewise.
6552         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6553         * config/aarch64/arm_neon.h
6554         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6555         (vml<as><q>_n_<su>16): Likewise.
6556         (vml<as>l_high_lane<q>_<su>16): Likewise.
6557         (vml<as>l_high_n_<su>16): Likewise.
6558         (vml<as>l_lane<q>_<su>16): Likewise.
6559         (vml<as>l_n_<su>16): Likewise.
6560         (vmul<q>_lane<q>_<su>16): Likewise.
6561         (vmul<q>_n_<su>16): Likewise.
6562         (vmull_lane<q>_<su>16): Likewise.
6563         (vmull_n_<su>16): Likewise.
6564         (vmull_high_lane<q>_<su>16): Likewise.
6565         (vmull_high_n_<su>16): Likewise.
6566         (vqrdmulh<q>_n_s16): Likewise.
6568 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6570         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6571         have the correct lane parameter.
6573 2013-09-06  Richard Biener <rguenther@suse.de>
6575         * cfganal.c (control_dependences::~control_dependences):
6576         Properly free all of the vector.
6578 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6580         PR target/58269
6581         * config/i386/i386.c (ix86_conditional_register_usage):
6582         Proper initialize extended SSE registers.
6584 2013-09-06  Jan Hubicka  <jh@suse.cz>
6586         PR tree-optimization/58311
6587         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6589 2013-09-06  Jan Hubicka  <jh@suse.cz>
6591         * Makefile.in (tree-sra.o): Update dependencies.
6592         * tree-sra.c: Include ipa-utils.h
6593         (scan_function): Use recursive_call_p.
6594         (has_caller_p): New function.
6595         (cgraph_for_node_and_aliases): Count also callers of aliases.
6597 2013-09-06  Jan Hubicka  <jh@suse.cz>
6599         PR middle-end/58094
6600         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6601         * tree-tailcall.c: Include ipa-utils.h.
6602         (find_tail_calls): Use it.
6603         * ipa-pure-const.c (check_call): Likewise.
6604         * ipa-utils.c (recursive_call_p): New function.
6605         * ipa-utils.h (recursive_call_p): Dclare.
6606         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6607         (symtab_semantically_equivalent_p): New function.
6608         * Makefile.in (tree-tailcall.o): Update dependencies.
6610 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6612         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6613         non-inlinable part.
6615 2013-09-06  Richard Biener  <rguenther@suse.de>
6617         * lto-streamer.h (lto_global_var_decls): Remove.
6618         * Makefile.in (OBJS): Remove lto-symtab.o.
6619         (lto-symtab.o): Remove.
6620         (GTFILES): Remove lto-symtab.c
6621         * lto-symtab.c: Move to lto/
6623 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6625         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6626         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6627         (UNSPEC_FPINT_RINT): New constant definitions.
6628         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6629         definition with 2 attributes.
6630         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6631         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6632         definitions.
6634 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6636         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6637         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6638         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6639         attribute to "z196_alone".
6640         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6641         "zEC12_simple".
6643 2013-09-06  Richard Biener  <rguenther@suse.de>
6645         * basic-block.h (class control_dependences): New.
6646         * tree-ssa-dce.c (control_dependence_map): Remove.
6647         (cd): New global.
6648         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6649         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6650         find_pdom, find_control_dependence, find_all_control_dependences):
6651         Move to cfganal.c.
6652         (mark_control_dependent_edges_necessary,
6653         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6654         tree_dce_done, perform_tree_ssa_dce): Adjust.
6655         * cfganal.c (set_control_dependence_map_bit,
6656         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6657         find_all_control_dependences): Move from tree-ssa-dce.c and
6658         implement as methods of control_dependences class.
6659         (control_dependences::control_dependences): New.
6660         (control_dependences::~control_dependences): Likewise.
6661         (control_dependences::get_edges_dependent_on): Likewise.
6662         (control_dependences::get_edge): Likewise.
6664 2013-09-04  Jan Hubicka  <jh@suse.cz>
6666         * tree.c (types_same_for_odr): Drop overactive check.
6667         * ipa-devirt.c (hash_type_name): Likewise.
6669 2013-09-04  Jan Hubicka  <jh@suse.cz>
6671         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6672         (analyze_functions): ... here.
6674 2013-09-04  Jan Hubicka  <jh@suse.cz>
6676         PR middle-end/58201
6677         * cgraphunit.c (analyze_functions): Clear AUX fields
6678         after processing; initialize assembler name has.
6680 2013-09-05  Jeff Law  <law@redhat.com>
6682         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6683         from thread_around_empty_block.  Record threading path into PATH.
6684         Recurse if threading through the initial block is successful.
6685         (thread_across_edge): Corresponding changes to slightly simplify.
6687 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6689         * config/aarch64/aarch64.md
6690         (type): Remove frecpe, frecps, frecpx.
6691         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6692         fix to be a TARGET_SIMD instruction.
6693         (aarch64_frecps): Remove.
6694         * config/aarch64/aarch64-simd.md
6695         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6696         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6698 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6699             Sofiane Naci  <sofiane.naci@arm.com>
6701         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6702         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6703         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6704         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6705         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6706         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6707         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6708         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6709         into "clz, "rbit".  Rename "shift" to "shift_imm".
6710         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6711         changes.  Update for attribute changes all occurrences of arlo_* and
6712         shift* types.
6713         * config/arm/arm-fixed.md: Update for attribute changes
6714         all occurrences of arlo_* types.
6715         * config/arm/thumb2.md: Update for attribute changes all occurrences
6716         of arlo_* types.
6717         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6718         (cortexa7_older_only): Likewise.
6719         (cortexa7_younger):  Likewise.
6720         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6721         (1020alu_shift_op): Likewise.
6722         (1020alu_shift_reg_op): Likewise.
6723         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6724         (alu_shift_op): Likewise.
6725         (alu_shift_reg_op): Likewise.
6726         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6727         (11_alu_shift_op): Likewise.
6728         (11_alu_shift_reg_op): Likewise.
6729         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6730         (9_alu_shift_reg_op): Likewise.
6731         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6732         attribute changes.
6733         (cortex_a15_alu_shift): Likewise.
6734         (cortex_a15_alu_shift_reg): Likewise.
6735         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6736         attribute changes.
6737         (cortex_a5_alu_shift): Likewise.
6738         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6739         attribute changes.
6740         (cortex_a53_alu_shift): Likewise.
6741         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6742         attribute changes.
6743         (cortex_a7_alu_reg): Likewise.
6744         (cortex_a7_alu_shift): Likewise.
6745         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6746         attribute changes.
6747         (cortex_a8_alu_shift): Likewise.
6748         (cortex_a8_alu_shift_reg): Likewise.
6749         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6750         (cortex_a9_dp_shift): Likewise.
6751         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6752         attribute changes.
6753         * config/arm/cortex-r4.md
6754         (cortex_r4_alu): Update for attribute changes.
6755         (cortex_r4_mov): Likewise.
6756         (cortex_r4_alu_shift_reg): Likewise.
6757         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6758         (526_alu_shift_op): Likewise.
6759         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6760         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6761         (626te_alu_shift_op): Likewise.
6762         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6763         (726te_alu_shift_op): Likewise.
6764         (726te_alu_shift_reg_op): Likewise.
6765         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6766         (mp626_alu_shift_op): Likewise.
6767         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6768         (pj4_alu_conds): Likewise.
6769         (pj4_shift): Likewise.
6770         (pj4_shift_conds): Likewise.
6771         (pj4_alu_shift): Likewise.
6772         (pj4_alu_shift_conds): Likewise.
6773         * config/aarch64/aarch64.md: Update for attribute change
6774         all occurrences of arlo_* and shift* types.
6776 2013-09-05  Mike Stump  <mikestump@comcast.net>
6778         * tree.h: Move documentation for tree_function_decl to tree-core.h
6779         with the declaration.
6781 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6783         PR target/58139
6784         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6785         looking for widest mode.
6787 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6789         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6791 2013-09-05  Richard Biener  <rguenther@suse.de>
6793         PR tree-optimization/58137
6794         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6795         Do not create vectors of pointers.
6796         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6797         types for the components of the vector initializer.
6798         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6799         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6801 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6803         * ipa-prop.c (remove_described_reference): Accept missing references,
6804         return false if that hppens, otherwise return true.
6805         (cgraph_node_for_jfunc): New function.
6806         (try_decrement_rdesc_refcount): Likewise.
6807         (try_make_edge_direct_simple_call): Use them.
6808         (ipa_edge_removal_hook): Remove references from rdescs.
6809         (ipa_edge_duplication_hook): Clone rdescs and their references
6810         when the new edge has the same caller as the old one.
6811         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6812         reference before removing any edges.
6814 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6816         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6817         initial store.
6818         * thumb2.md (thumb2_storewb_parisi): New pattern.
6820 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6822         * config/aarch64/aarch64-option-extensions.def: Add
6823         AARCH64_OPT_EXTENSION of 'crc'.
6824         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6825         (AARCH64_ISA_CRC): Ditto.
6826         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6827         description of the CRC extension.
6829 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6831         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6832         * config/rs6000/linux.h: Ditto.
6833         * alpha/linux.h: Ditto.
6834         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6835         no_c99_libc_has_function.
6836         * config/c6x/uclinux-elf.h: Ditto.
6837         * config/lm32/uclinux-elf.h: Ditto.
6838         * config/m68k/uclinux.h: Ditto.
6839         * config/moxie/uclinux.h: Ditto.
6840         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6841         (crisv32-*-linux*, cris-*-linux*): Ditto.
6842         * config/bfin/bfin.c: Include "tm_p.h".
6844 2013-09-05  Richard Biener  <rguenther@suse.de>
6846         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6847         check for a definition without a basic-block.
6849 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6850             Sofiane Naci <sofiane.naci@arm.com>
6852         * config/aarch64/aarch64.md
6853         (*movti_aarch64): Rename r_2_f and f_2_r.
6854         (*movsf_aarch64): Likewise.
6855         (*movdf_aarch64): Likewise.
6856         (*movtf_aarch64): Likewise.
6857         (aarch64_movdi_<mode>low): Likewise.
6858         (aarch64_movdi_<mode>high): Likewise.
6859         (aarch64_mov<mode>high_di): Likewise.
6860         (aarch64_mov<mode>low_di): Likewise.
6861         (aarch64_movtilow_tilow): Likewise.
6862         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6863         values to config/arm/types.md
6864         (attribute "conds"): Update for attribute change.
6865         (anddi3_insn): Likewise.
6866         (iordi3_insn): Likewise.
6867         (xordi3_insn): Likewise.
6868         (one_cmpldi2): Likewise.
6869         * config/arm/types.md (type): Add Neon types.
6870         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6871         use "type" attribute.
6872         (movmisalign<mode>_neon_store): Likewise.
6873         (movmisalign<mode>_neon_load): Likewise.
6874         (vec_set<mode>_internal): Likewise.
6875         (vec_setv2di_internal): Likewise.
6876         (vec_extract<mode>): Likewise.
6877         (vec_extractv2di): Likewise.
6878         (add<mode>3_neon): Likewise.
6879         (adddi3_neon): Likewise.
6880         (sub<mode>3_neon): Likewise.
6881         (subdi3_neon): Likewise.
6882         (mul<mode>3_neon): Likewise.
6883         (mul<mode>3add<mode>_neon): Likewise.
6884         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6885         (fma<VCVTF:mode>4)): Likewise.
6886         (fma<VCVTF:mode>4_intrinsic): Likewise.
6887         (fmsub<VCVTF:mode>4)): Likewise.
6888         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6889         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6890         (ior<mode>3): Likewise.
6891         (and<mode>3): Likewise.
6892         (anddi3_neon): Likewise.
6893         (orn<mode>3_neon): Likewise.
6894         (orndi3_neon): Likewise.
6895         (bic<mode>3_neon): Likewise.
6896         (bicdi3_neon): Likewise.
6897         (xor<mode>3): Likewise.
6898         (one_cmpl<mode>2): Likewise.
6899         (abs<mode>2): Likewise.
6900         (neg<mode>2): Likewise.
6901         (umin<mode>3_neon): Likewise.
6902         (umax<mode>3_neon): Likewise.
6903         (smin<mode>3_neon): Likewise.
6904         (smax<mode>3_neon): Likewise.
6905         (vashl<mode>3): Likewise.
6906         (vashr<mode>3_imm): Likewise.
6907         (vlshr<mode>3_imm): Likewise.
6908         (ashl<mode>3_signed): Likewise.
6909         (ashl<mode>3_unsigned): Likewise.
6910         (neon_load_count): Likewise.
6911         (ashldi3_neon_noclobber): Likewise.
6912         (signed_shift_di3_neon): Likewise.
6913         (unsigned_shift_di3_neon): Likewise.
6914         (ashrdi3_neon_imm_noclobber): Likewise.
6915         (lshrdi3_neon_imm_noclobber): Likewise.
6916         (widen_ssum<mode>3): Likewise.
6917         (widen_usum<mode>3): Likewise.
6918         (quad_halves_<code>v4si): Likewise.
6919         (quad_halves_<code>v4sf): Likewise.
6920         (quad_halves_<code>v8hi): Likewise.
6921         (quad_halves_<code>v16qi): Likewise.
6922         (reduc_splus_v2di): Likewise.
6923         (neon_vpadd_internal<mode>): Likewise.
6924         (neon_vpsmin<mode>): Likewise.
6925         (neon_vpsmax<mode>): Likewise.
6926         (neon_vpumin<mode>): Likewise.
6927         (neon_vpumax<mode>): Likewise.
6928         (ss_add<mode>_neon): Likewise.
6929         (us_add<mode>_neon): Likewise.
6930         (ss_sub<mode>_neon): Likewise.
6931         (us_sub<mode>_neon): Likewise.
6932         (neon_vadd<mode>_unspec): Likewise.
6933         (neon_vaddl<mode>): Likewise.
6934         (neon_vaddw<mode>): Likewise.
6935         (neon_vhadd<mode>): Likewise.
6936         (neon_vqadd<mode>): Likewise.
6937         (neon_vaddhn<mode>): Likewise.
6938         (neon_vmul<mode>): Likewise.
6939         (neon_vmla<mode>): Likewise.
6940         (neon_vmlal<mode>): Likewise.
6941         (neon_vmls<mode>): Likewise.
6942         (neon_vmlsl<mode>): Likewise.
6943         (neon_vqdmulh<mode>): Likewise.
6944         (neon_vqdmlal<mode>): Likewise.
6945         (neon_vqdmlsl<mode>): Likewise.
6946         (neon_vmull<mode>): Likewise.
6947         (neon_vqdmull<mode>): Likewise.
6948         (neon_vsub<mode>_unspec): Likewise.
6949         (neon_vsubl<mode>): Likewise.
6950         (neon_vsubw<mode>): Likewise.
6951         (neon_vqsub<mode>): Likewise.
6952         (neon_vhsub<mode>): Likewise.
6953         (neon_vsubhn<mode>): Likewise.
6954         (neon_vceq<mode>): Likewise.
6955         (neon_vcge<mode>): Likewise.
6956         (neon_vcgeu<mode>): Likewise.
6957         (neon_vcgt<mode>): Likewise.
6958         (neon_vcgtu<mode>): Likewise.
6959         (neon_vcle<mode>): Likewise.
6960         (neon_vclt<mode>): Likewise.
6961         (neon_vcage<mode>): Likewise.
6962         (neon_vcagt<mode>): Likewise.
6963         (neon_vtst<mode>): Likewise.
6964         (neon_vabd<mode>): Likewise.
6965         (neon_vabdl<mode>): Likewise.
6966         (neon_vaba<mode>): Likewise.
6967         (neon_vabal<mode>): Likewise.
6968         (neon_vmax<mode>): Likewise.
6969         (neon_vmin<mode>): Likewise.
6970         (neon_vpaddl<mode>): Likewise.
6971         (neon_vpadal<mode>): Likewise.
6972         (neon_vpmax<mode>): Likewise.
6973         (neon_vpmin<mode>): Likewise.
6974         (neon_vrecps<mode>): Likewise.
6975         (neon_vrsqrts<mode>): Likewise.
6976         (neon_vqabs<mode>): Likewise.
6977         (neon_vqneg<mode>): Likewise.
6978         (neon_vcls<mode>): Likewise.
6979         (clz<mode>2): Likewise.
6980         (popcount<mode>2): Likewise.
6981         (neon_vrecpe): Likewise.
6982         (neon_vrsqrte): Likewise.
6983         (neon_vget_lane<mode>_sext_internal): Likewise.
6984         (neon_vget_lane<mode>_zext_internal): Likewise.
6985         (neon_vdup_n<mode>): Likewise.
6986         (neon_vdup_nv2di): Likewise.
6987         (neon_vdpu_lane<mode>_internal): Likewise.
6988         (neon_vswp<mode>): Likewise.
6989         (float<mode><V_cvtto>2): Likewise.
6990         (floatuns<mode><V_cvtto>2): Likewise.
6991         (fix_trunc<mode><V_cvtto>)2): Likewise
6992         (fixuns_trunc<mode><V_cvtto)2): Likewise.
6993         (neon_vcvt<mode>): Likewise.
6994         (neon_vcvtv4sfv4hf): Likewise.
6995         (neon_vcvtv4hfv4sf): Likewise.
6996         (neon_vcvt_n<mode>): Likewise.
6997         (neon_vmovn<mode>): Likewise.
6998         (neon_vqmovn<mode>): Likewise.
6999         (neon_vqmovun<mode>): Likewise.
7000         (neon_vmovl<mode>): Likewise.
7001         (neon_vmul_lane<mode>): Likewise.
7002         (neon_vmull_lane<mode>): Likewise.
7003         (neon_vqdmull_lane<mode>): Likewise.
7004         (neon_vqdmulh_lane<mode>): Likewise.
7005         (neon_vmla_lane<mode>): Likewise.
7006         (neon_vmlal_lane<mode>): Likewise.
7007         (neon_vqdmlal_lane<mode>): Likewise.
7008         (neon_vmls_lane<mode>): Likewise.
7009         (neon_vmlsl_lane<mode>): Likewise.
7010         (neon_vqdmlsl_lane<mode>): Likewise.
7011         (neon_vext<mode>): Likewise.
7012         (neon_vrev64<mode>): Likewise.
7013         (neon_vrev32<mode>): Likewise.
7014         (neon_vrev16<mode>): Likewise.
7015         (neon_vbsl<mode>_internal): Likewise.
7016         (neon_vshl<mode>): Likewise.
7017         (neon_vqshl<mode>): Likewise.
7018         (neon_vshr_n<mode>): Likewise.
7019         (neon_vshrn_n<mode>): Likewise.
7020         (neon_vqshrn_n<mode>): Likewise.
7021         (neon_vqshrun_n<mode>): Likewise.
7022         (neon_vshl_n<mode>): Likewise.
7023         (neon_vqshl_n<mode>): Likewise.
7024         (neon_vqshlu_n<mode>): Likewise.
7025         (neon_vshll_n<mode>): Likewise.
7026         (neon_vsra_n<mode>): Likewise.
7027         (neon_vsri_n<mode>): Likewise.
7028         (neon_vsli_n<mode>): Likewise.
7029         (neon_vtbl1v8qi): Likewise.
7030         (neon_vtbl2v8qi): Likewise.
7031         (neon_vtbl3v8qi): Likewise.
7032         (neon_vtbl4v8qi): Likewise.
7033         (neon_vtbx1v8qi): Likewise.
7034         (neon_vtbx2v8qi): Likewise.
7035         (neon_vtbx3v8qi): Likewise.
7036         (neon_vtbx4v8qi): Likewise.
7037         (neon_vtrn<mode>_internal): Likewise.
7038         (neon_vzip<mode>_internal): Likewise.
7039         (neon_vuzp<mode>_internal): Likewise.
7040         (neon_vld1<mode>): Likewise.
7041         (neon_vld1_lane<mode>): Likewise.
7042         (neon_vld1_dup<mode>): Likewise.
7043         (neon_vld1_dupv2di): Likewise.
7044         (neon_vst1<mode>): Likewise.
7045         (neon_vst1_lane<mode>): Likewise.
7046         (neon_vld2<mode>): Likewise.
7047         (neon_vld2_lane<mode>): Likewise.
7048         (neon_vld2_dup<mode>): Likewise.
7049         (neon_vst2<mode>): Likewise.
7050         (neon_vst2_lane<mode>): Likewise.
7051         (neon_vld3<mode>): Likewise.
7052         (neon_vld3qa<mode>): Likewise.
7053         (neon_vld3qb<mode>): Likewise.
7054         (neon_vld3_lane<mode>): Likewise.
7055         (neon_vld3_dup<mode>): Likewise.
7056         (neon_vst3<mode>): Likewise.
7057         (neon_vst3qa<mode>): Likewise.
7058         (neon_vst3qb<mode>): Likewise.
7059         (neon_vst3_lane<mode>): Likewise.
7060         (neon_vld4<mode>): Likewise.
7061         (neon_vld4qa<mode>): Likewise.
7062         (neon_vld4qb<mode>): Likewise.
7063         (neon_vld4_lane<mode>): Likewise.
7064         (neon_vld4_dup<mode>): Likewise.
7065         (neon_vst4<mode>): Likewise.
7066         (neon_vst4qa<mode>): Likewise.
7067         (neon_vst4qb<mode>): Likewise.
7068         (neon_vst4_lane<mode>): Likewise.
7069         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7070         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7071         (neon_vec_<US>mult_lo_<mode>): Likewise.
7072         (neon_vec_<US>mult_hi_<mode>): Likewise.
7073         (neon_vec_<US>shiftl_<mode>): Likewise.
7074         (neon_unpack<US>_<mode>): Likewise.
7075         (neon_vec_<US>mult_<mode>): Likewise.
7076         (vec_pack_trunc_<mode>): Likewise.
7077         (neon_vec_pack_trunk_<mode>): Likewise.
7078         (neon_vabd<mode>_2): Likewise.
7079         (neon_vabd<mode>_3): Likewise.
7080         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7081         (thumb2_movsi_vfp): Likewise.
7082         (movdi_vfp): Likewise.
7083         (movdi_vfp_cortexa8): Likewise.
7084         (movhf_vfp_neon): Likewise.
7085         (movhf_vfp): Likewiwse.
7086         (movsf_vfp): Likewiwse.
7087         (thumb2_movsf_vfp): Likewiwse.
7088         (movdf_vfp): Likewise.
7089         (thumb2_movdf_vfp): Likewise.
7090         (movsfcc_vfp): Likewise.
7091         (thumb2_movsfcc_vfp): Likewise.
7092         (movdfcc_vfp): Likewise.
7093         (thumb2_movdfcc_vfp): Likewise.
7094         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7095         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7096         (v10_v2c): Likewise.
7097         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7098         attribute change.
7099         (cortex_a15_neon_int_2): Likewise.
7100         (cortex_a15_neon_int_3): Likewise.
7101         (cortex_a15_neon_int_4): Likewise.
7102         (cortex_a15_neon_int_5): Likewise.
7103         (cortex_a15_neon_vqneg_vqabs): Likewise.
7104         (cortex_a15_neon_vmov): Likewise.
7105         (cortex_a15_neon_vaba): Likewise.
7106         (cortex_a15_neon_vaba_qqq): Likewise.
7107         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7108         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7109         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7110         scalar_64_32_long_scalar): Likewise.
7111         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7112         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7113         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7114         lotype_qdd_64_32_long): Likewise.
7115         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7116         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7117         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7118         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7119         (cortex_a15_neon_shift_1): Likewise.
7120         (cortex_a15_neon_shift_2): Likewise.
7121         (cortex_a15_neon_shift_3): Likewise.
7122         (cortex_a15_neon_vshl_ddd): Likewise.
7123         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7124         (cortex_a15_neon_vsra_vrsra): Likewise.
7125         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7126         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7127         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7128         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7129         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7130         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7131         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7132         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7133         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7134         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7135         (cortex_a15_neon_bp_simple): Likewise.
7136         (cortex_a15_neon_bp_2cycle): Likewise.
7137         (cortex_a15_neon_bp_3cycle): Likewise.
7138         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7139         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7140         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7141         (cortex_a15_neon_vld2_4_regs): Likewise.
7142         (cortex_a15_neon_vld3_vld4): Likewise.
7143         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7144         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7145         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7146         (cortex_a15_neon_vst3_vst4): Likewise.
7147         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7148         (cortex_a15_neon_vld3_vld4_lane" 10
7149         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7150         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7151         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7152         (cortex_a15_neon_ldm_2): Likewise.0
7153         (cortex_a15_neon_stm_2): Likewise.
7154         (cortex_a15_neon_mcr): Likewise.
7155         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7156         (cortex_a15_neon_mrc): Likewise.
7157         (cortex_a15_neon_mrrc): Likewise.
7158         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7159         change.
7160         (cortex_a15_alu_shift): Likewise.
7161         (cortex_a15_alu_shift_reg): Likewise.
7162         (cortex_a15_mult32): Likewise.
7163         (cortex_a15_mult64): Likewise.
7164         (cortex_a15_block): Likewise.
7165         (cortex_a15_branch): Likewise.
7166         (cortex_a15_load1): Likewise.
7167         (cortex_a15_load3): Likewise.
7168         (cortex_a15_store1): Likewise.
7169         (cortex_a15_store3): Likewise.
7170         (cortex_a15_call): Likewise.
7171         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7172         (cortex_a5_f2r): Likewise.
7173         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7174         change.
7175         (cortex_a53_f2r): Likewise.
7176         * config/arm/cortex-a7.md
7177         (cortex_a7_branch): Update for attribute change.
7178         (cortex_a7_call): Likewise.
7179         (cortex_a7_alu_imm): Likewise.
7180         (cortex_a7_alu_reg): Likewise.
7181         (cortex_a7_alu_shift): Likewise.
7182         (cortex_a7_mul): Likewise.
7183         (cortex_a7_load1): Likewise.
7184         (cortex_a7_store1): Likewise.
7185         (cortex_a7_load2): Likewise.
7186         (cortex_a7_store2): Likewise.
7187         (cortex_a7_load3): Likewise.
7188         (cortex_a7_store3): Likewise.
7189         (cortex_a7_load4): Likewise.
7190         (cortex_a7_store4): Likewise.
7191         (cortex_a7_fpalu): Likewise.
7192         (cortex_a7_fconst): Likewise.
7193         (cortex_a7_fpmuls): Likewise.
7194         (cortex_a7_neon_mul): Likewise.
7195         (cortex_a7_fpmacs): Likewise.
7196         (cortex_a7_neon_mla: Likewise.
7197         (cortex_a7_fpmuld: Likewise.
7198         (cortex_a7_fpmacd: Likewise.
7199         (cortex_a7_fpfmad: Likewise.
7200         (cortex_a7_fdivs: Likewise.
7201         (cortex_a7_fdivd: Likewise.
7202         (cortex_a7_r2f: Likewise.
7203         (cortex_a7_f2r: Likewise.
7204         (cortex_a7_f_flags: Likewise.
7205         (cortex_a7_f_loads: Likewise.
7206         (cortex_a7_f_loadd: Likewise.
7207         (cortex_a7_f_stores: Likewise.
7208         (cortex_a7_f_stored: Likewise.
7209         (cortex_a7_neon): Likewise.
7210         * config/arm/cortex-a8-neon.md
7211         (cortex_a8_neon_mrc): Update for attribute change.
7212         (cortex_a8_neon_mrrc): Likewise.
7213         (cortex_a8_neon_int_1): Likewise.
7214         (cortex_a8_neon_int_2): Likewise.
7215         (cortex_a8_neon_int_3): Likewise.
7216         (cortex_a8_neon_int_4): Likewise.
7217         (cortex_a8_neon_int_5): Likewise.
7218         (cortex_a8_neon_vqneg_vqabs): Likewise.
7219         (cortex_a8_neon_vmov): Likewise.
7220         (cortex_a8_neon_vaba): Likewise.
7221         (cortex_a8_neon_vaba_qqq): Likewise.
7222         (cortex_a8_neon_vsma): Likewise.
7223         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7224         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7225         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7226         long_scalar): Likewise.
7227         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7228         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7229         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7230         long_scalar_qdd_64_32_long): Likewise.
7231         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7232         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7233         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7234         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7235         (cortex_a8_neon_shift_1): Likewise.
7236         (cortex_a8_neon_shift_2): Likewise.
7237         (cortex_a8_neon_shift_3): Likewise.
7238         (cortex_a8_neon_vshl_ddd): Likewise.
7239         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7240         (cortex_a8_neon_vsra_vrsra): Likewise.
7241         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7242         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7243         (cortex_a8_neon_fp_vsum): Likewise.
7244         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7245         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7246         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7247         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7248         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7249         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7250         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7251         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7252         (cortex_a8_neon_bp_simple): Likewise.
7253         (cortex_a8_neon_bp_2cycle): Likewise.
7254         (cortex_a8_neon_bp_3cycle): Likewise.
7255         (cortex_a8_neon_ldr): Likewise.
7256         (cortex_a8_neon_str): Likewise.
7257         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7258         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7259         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7260         (cortex_a8_neon_vld2_4_regs): Likewise.
7261         (cortex_a8_neon_vld3_vld4): Likewise.
7262         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7263         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7264         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7265         (cortex_a8_neon_vst3_vst4): Likewise.
7266         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7267         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7268         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7269         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7270         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7271         (cortex_a8_neon_mcr): Likewise.
7272         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7273         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7274         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7275         change.
7276         (ca9_neon_mrrc): Likewise.
7277         (cortex_a9_neon_int_1): Likewise.
7278         (cortex_a9_neon_int_2): Likewise.
7279         (cortex_a9_neon_int_3): Likewise.
7280         (cortex_a9_neon_int_4): Likewise.
7281         (cortex_a9_neon_int_5): Likewise.
7282         (cortex_a9_neon_vqneg_vqabs): Likewise.
7283         (cortex_a9_neon_vmov): Likewise.
7284         (cortex_a9_neon_vaba): Likewise.
7285         (cortex_a9_neon_vaba_qqq): Likewise.
7286         (cortex_a9_neon_vsma): Likewise.
7287         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7288         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7289         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7290         long_scalar): Likewise.
7291         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7292         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7293         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7294         long_scalar_qdd_64_32_long): Likewise.
7295         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7296         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7297         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7298         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7299         (cortex_a9_neon_shift_1): Likewise.
7300         (cortex_a9_neon_shift_2): Likewise.
7301         (cortex_a9_neon_shift_3): Likewise.
7302         (cortex_a9_neon_vshl_ddd): Likewise.
7303         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7304         (cortex_a9_neon_vsra_vrsra): Likewise.
7305         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7306         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7307         (cortex_a9_neon_fp_vsum): Likewise.
7308         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7309         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7310         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7311         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7312         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7313         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7314         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7315         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7316         (cortex_a9_neon_bp_simple): Likewise.
7317         (cortex_a9_neon_bp_2cycle): Likewise.
7318         (cortex_a9_neon_bp_3cycle): Likewise.
7319         (cortex_a9_neon_ldr): Likewise.
7320         (cortex_a9_neon_str): Likewise.
7321         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7322         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7323         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7324         (cortex_a9_neon_vld2_4_regs): Likewise.
7325         (cortex_a9_neon_vld3_vld4): Likewise.
7326         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7327         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7328         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7329         (cortex_a9_neon_vst3_vst4): Likewise.
7330         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7331         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7332         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7333         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7334         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7335         (cortex_a9_neon_mcr): Likewise.
7336         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7337         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7338         (cortex_a9_fps): Likewise.
7339         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7340         change.
7341         (cortex_m4_fmuls): Likewise.
7342         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7343         change.
7344         (cortex_r4_mrc): Likewise.
7345         * config/arm/iterators.md: Update comment referring to neon_type.
7346         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7347         (iwmmxt_movsi_insn): Likewise.
7348         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7349         attribute change.
7350         (pj4_core_to_vfp): Likewise.
7351         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7352         attribute change.
7353         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7354         (vfp_fstore): Likewise.
7355         * doc/md.texi: Change references to neon_type to refer to type.
7357 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7359         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7361 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7363         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7364         lglobl if not weak.
7366 2013-09-04  Easwaran Raman  <eraman@google.com>
7368         PR middle-end/57370
7369         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7370         (build_and_add_sum): Use it.
7371         (appears_later_in_bb): Simplify code.
7373 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7375         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7377 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7379         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7381 2013-09-04  Jan Hubicka  <jh@suse.cz>
7383         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7384         * ipa-devirt.c: Include diganostic.h
7385         (odr_type_d): Add types and types_set.
7386         (hash_type_name): Work for types with vtables during LTO.
7387         (odr_hasher::remove): Fix comment; destroy types_set.
7388         (add_type_duplicate): New function,
7389         (get_odr_type): Use it.
7390         (dump_type_inheritance_graph): Dump type duplicates.
7391         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7392         graph.
7393         * tree.c (types_same_for_odr): Give exact answers on types with
7394         virtual tables.
7396 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7398         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7399         explaining their differences.
7401 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7403         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7405 2013-09-03  Jeff Law  <law@redhat.com>
7407         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7408         when not threading through a joiner block.  Pass joiner/no joiner
7409         state to register_jump_thread.
7410         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7411         state from argument rather than implying on path length.
7412         Dump the entire jump thread path into debugging dump.
7413         * tree-flow.h (register_jump_thread): Update prototype.
7415 2013-08-29  Xinliang David Li  <davidxl@google.com>
7417         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7418         Remove a trivial gcc_assert.
7420 2013-08-29  Xinliang David Li  <davidxl@google.com>
7422         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7423         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7424         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7425         Delay base decl alignment adjustment.
7426         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7427         * tree-vectorizer.h: New data structure.
7428         (set_dr_misalignment): New function.
7429         (dr_misalignment): Ditto.
7430         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7431         (vectorizable_load): Ditto.
7432         (ensure_base_align): New function.
7433         (vectorize_loops): Add dbg_cnt support.
7434         (execute_vect_slp): Ditto.
7435         * dbgcnt.def: New debug counter.
7436         * Makefile: New dependency.
7438 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7440         Revert:
7442         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7444         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7446 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7448         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7449         function descriptor.
7451 2013-09-03  Richard Biener  <rguenther@suse.de>
7453         * tree-affine.c (add_elt_to_tree): Fix association issue,
7454         avoid useless converts and make sure to always return a
7455         properly typed result.
7457 2013-09-03  Richard Biener  <rguenther@suse.de>
7459         PR middle-end/57656
7460         * fold-const.c (negate_expr_p): Fix division case.
7461         (negate_expr): Likewise.
7463 2013-09-03  Richard Biener  <rguenther@suse.de>
7465         PR lto/58285
7466         * tree-streamer-out.c: Include tm.h.
7467         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7469 2013-09-03  Jan Hubicka  <jh@suse.cz>
7471         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7473 2013-09-03  Alan Modra  <amodra@gmail.com>
7475         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7476         multilibs to big-endian target and vice versa.
7477         * config/rs6000/t-linux64: Use := assignment on all vars.
7478         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7479         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7480         * config/rs6000/t-linux64le: New file.
7481         * config/rs6000/t-linux64bele: New file.
7482         * config/rs6000/t-linux64lebe: New file.
7484 2013-09-02  Jan Hubicka  <jh@suse.cz>
7486         * ipa-inline-transform.c (inline_transform): Do not
7487         optimize_inline_calls when not optimizing.
7489 2013-09-02  Jan Hubicka  <jh@suse.cz>
7491         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7492         duplicated nodes for assembler names.
7493         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7494         hard registers from assembler name hash.
7496 2013-09-02  Jan Hubicka  <jh@suse.cz>
7498         * ipa-split.c (execute_split_functions): Split externally visible
7499         functions called once.
7501 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7503         PR ipa/58106
7504         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7505         linked list.  When finding the correct duplicate, also consider also
7506         the caller in additon to its inlined_to node.
7508 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7510         * config/aarch64/aarch64-simd-builtins.def
7511         (dup_lane_scalar): Remove.
7512         * config/aarch64/aarch64-simd.md
7513         (aarch64_simd_dup): Add 'w->w' alternative.
7514         (aarch64_dup_lane<mode>): Allow for VALL.
7515         (aarch64_dup_lane_scalar<mode>): Remove.
7516         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7517         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7518         (aarch64_get_lane_unsigned<mode>): Likewise.
7519         (aarch64_get_lane<mode>): Likewise.
7520         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7521         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7522         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7523         (VCON): Change container of V2SF.
7524         (vswap_width_name): Likewise.
7525         * config/aarch64/arm_neon.h
7526         (__aarch64_vdup_lane_any): New.
7527         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7528         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7529         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7531 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7533         PR middle-end/56382
7534         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7535         the source or the destination is a single hard register.
7537 2013-09-02  Richard Biener  <rguenther@suse.de>
7539         PR middle-end/57511
7540         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7541         non-linear SCEVs.
7543 2013-09-02  Richard Biener  <rguenther@suse.de>
7545         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7546         arithmetic to sizetype.
7548 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7550         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7551         Find auto-increment use both before and after candidate.
7553 2013-09-02  Marek Polacek  <polacek@redhat.com>
7555         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7557 2013-09-01  Jan Hubicka  <jh@suse.cz>
7559         * Makefile.in: Add ipa-profile.o
7560         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7561         * cgraph.c (struct cgraph_propagate_frequency_data,
7562         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7563         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7564         * cgraph.h (cgraph_propagate_frequency): Remove.
7565         * ipa-inline-analysis.c: Include ipa-utils.h;
7566         drop duplicated cfgloop.h.
7567         (inline_update_callee_summaries): Update.
7568         * ipa-profile.c: New file.
7569         * ipa-utils.h (ipa_propagate_frequency): Declare.
7570         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7571         data-streamer.h, value-prof.h.
7572         (symtab_remove_unreachable_nodes): Update profile.
7573         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7574         account_time_size, cmp_counts, dump_histogram,
7575         ipa_profile_generate_summary, ipa_profile_write_summary,
7576         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7577         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7578         Move to ipa-profile.c.
7580 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7582         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7584 2013-09-01  Jan Hubicka  <jh@suse.cz>
7586         * common.opt (fdevirtualize-speculatively): New function.
7587         * invoke.texi (fdevirtualize-speculatively): Document.
7588         * ipa-devirt.c: Include ipa-inline.h
7589         (likely_target_p): New function.
7590         (ipa_devirt): New function.
7591         (gate_ipa_devirt): New function.
7592         (pass_data_ipa_devirt): New static var.
7593         (pass_ipa_devirt): Likewise.
7594         (make_pass_ipa_devirt): New function.
7595         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7596         (common_handle_option): Disable devirtualization when
7597         value range profiling is available.
7598         * passes.def (pass_ipa_devirt): Add.
7599         * timever.def (TV_IPA_DEVIRT): New timevar.
7600         * tree-pass.h (make_pass_ipa_devirt):
7602 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7604         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7605         include sanitize(undefined).
7607 2013-08-31  Diego Novillo  <dnovillo@google.com>
7609         * Makefile.in (TREE_CORE_H): Define.
7610         (TREE_H): Use.
7611         (GTFILES): Add tree-core.h.
7612         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7613         size the array.
7614         * tree-core.h: New file.
7615         Move all data structures, enum, typedefs, global
7616         declarations and constants from ...
7617         * tree.h: ... here.
7619 2013-08-31  Jan Hubicka  <jh@suse.cz>
7621         * bulitins.c (expand_builtin): Do not early exit for gcov
7622         instrumented functions.
7624 2013-08-31  Marek Polacek  <polacek@redhat.com>
7626         * ubsan.c: Include tm_p.h.
7628 2013-08-31  Jan Hubicka  <jh@suse.cz>
7630         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7631         warning.
7633         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7634         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7635         * tree-cfg.c (verify_gimple_label): ... here.
7636         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7637         (ipa_merge_profiles): New function.
7638         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7639         (lto_input_function_body): Likewise.
7640         * ipa-utils.h (ipa_merge_profiles): Declare.
7641         * lto-streamer.h (lto_input_function_body): Update prototype.
7642         (emit_label_in_global_context_p): Remove.
7643         * lto-symtab.c: Include ipa-utils.h
7644         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7646 2013-08-31  Jan Hubicka  <jh@suse.cz>
7648         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7650 2013-08-31  Jan Hubicka  <jh@suse.cz>
7652         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7654 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7656         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7657         "cmp" RTX before signed_comparison_operator check to account
7658         for "code" changes.
7660 2013-08-30  Jan Hubicka  <jh@suse.cz>
7662         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7663         (detect_type_change_1): Rename to ...
7664         (detect_type_change): ... this one; early return on non-polymorphic
7665         types.
7666         (detect_type_change_ssa): Add comp_type parameter; update
7667         use of detect_type_change.
7668         (compute_complex_assign_jump_func): Add param_type parameter;
7669         update use of detect_type_change_ssa.
7670         (compute_complex_ancestor_jump_func): Likewise.
7671         (ipa_get_callee_param_type): New function.
7672         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7673         update calls to the jump function computation functions.
7675 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7676             Steven Bosscher  <steven@gcc.gnu.org>
7678         * cfgrtl.c (fixup_new_cold_bb): New routine.
7679         (commit_edge_insertions): Invoke fixup_partitions.
7680         (find_partition_fixes): New routine.
7681         (fixup_partitions): Ditto.
7682         (verify_hot_cold_block_grouping): Update comments.
7683         (rtl_verify_edges): Invoke find_partition_fixes.
7684         (rtl_verify_bb_pointers): Update comments.
7685         (rtl_verify_bb_layout): Ditto.
7686         * basic-block.h (probably_never_executed_edge_p): Declare.
7687         (fixup_partitions): Ditto.
7688         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7689         * bb-reorder.c (sanitize_hot_paths): New function.
7690         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7691         sanitize_hot_paths.
7692         * predict.c (probably_never_executed_edge_p): New routine.
7693         * cfg.c (check_bb_profile): Add partition insanity warnings.
7695 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7697         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7699 2013-08-30  Marek Polacek  <polacek@redhat.com>
7701         * Makefile.in (ubsan.o): Add.
7702         (c-family/c-ubsan.o): Add.
7703         (builtins.o): Add ubsan.h dependency.
7704         * ubsan.h: New file.
7705         * ubsan.c: New file.
7706         * common.opt: Add -fsanitize=undefined option.
7707         (flag_sanitize): Add variable.
7708         (fsanitize=): Add option.  Add Driver.
7709         (fsanitize=thread): Remove option.
7710         (fsanitize=address): Likewise.
7711         (static-libubsan): New option.
7712         * doc/invoke.texi: Document the new flag and -static-libubsan.
7713         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7714         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7715         * builtin-attrs.def (ATTR_COLD): Define.
7716         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7717         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7718         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7719         * flag-types.h (sanitize_code): New enum.
7720         * opts.c (common_handle_option): Parse command line arguments
7721         of -fsanitize=.  Add -fsanitize=unreachable option.
7722         * varasm.c (get_variable_section): Adjust.
7723         (assemble_noswitch_variable): Likewise.
7724         (assemble_variable): Likewise.
7725         (output_constant_def_contents): Likewise.
7726         (categorize_decl_for_section): Likewise.
7727         (place_block_symbol): Likewise.
7728         (output_object_block): Likewise.
7729         * builtins.def: Likewise.
7730         * toplev.c (compile_file): Likewise.
7731         (process_options): Likewise.
7732         * cppbuiltin.c: Likewise.
7733         * tsan.c (tsan_pass): Likewise.
7734         (tsan_gate): Likewise.
7735         (tsan_gate_O0): Likewise.
7736         * cfgexpand.c (partition_stack_vars): Likewise.
7737         (expand_stack_vars): Likewise.
7738         (defer_stack_allocation): Likewise.
7739         (expand_used_vars): Likewise.
7740         * cfgcleanup.c (old_insns_match_p): Likewise.
7741         * asan.c (asan_finish_file): Likewise.
7742         (asan_instrument): Likewise.
7743         (gate_asan): Likewise.
7744         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7745         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7746         (asan_global_struct): Use pointer_sized_int_node instead
7747         calling build_nonstandard_integer_type.
7748         (initialize_sanitizer_builtins): Likewise.
7749         (asan_finish_file): Likewise.
7750         * gcc.c: Document %{%:function(args):X}.
7751         (static_spec_functions): Add sanitize.
7752         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7753         store funcval != NULL there.
7754         (do_spec_1): Adjust handle_spec_function caller.
7755         (handle_braces): Allow %:function(args) as condition.
7756         (sanitize_spec_function): New function.
7757         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7758         (LIBUBSAN_SPEC): Likewise.
7759         (LIBUBSAN_EARLY_SPEC): Likewise.
7760         (SANITIZER_SPEC): Handle libubsan.
7761         (SANITIZER_EARLY_SPEC): Likewise.
7762         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7763         instead of fsanitize=address.
7764         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7765         instead of fsanitize=address*.
7766         * builtins.c: Include ubsan.h.
7767         (fold_builtin_0): Instrument __builtin_unreachable.
7768         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7769         instead of flag_asan.
7770         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7771         (pointer_sized_int_node): Define.
7772         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7774 2013-08-30  Mike Stump  <mikestump@comcast.net>
7776         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7777         with RE patterns.
7779 2013-08-29  Jan Hubicka  <jh@suse.cz>
7781         * cgraph.c (cgraph_function_body_availability): Handle weakref
7782         correctly.
7783         * passes.def: Remove pass_fixup_cfg.
7784         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7785         track when we need to remove functions.
7786         (gate_ipa_inline): Execute inlining always; add comment why.
7787         (pass_data_ipa_inline): Remove TODO_remove_functions.
7788         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7789         do not produce summaries.
7790         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7791         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7792         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7793         constant pool and vtable.
7795 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7797         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7798         New arm_neon.h's internal macros to specify 64-bit constants.
7799         Avoid using stdint.h's macros.
7801 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7803         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7804         before applying REG_P.
7806 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7808         PR tree-optimization/58277
7809         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7810         after seeing too many stmts with vdef in between dombb and current
7811         bb, invalidate everything.
7813 2013-08-30  Richard Biener  <rguenther@suse.de>
7815         * fold-const.c (fold_single_bit_test): Fix overflow test.
7817 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7819         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7820         and which can live in a register, always retrieve the value on entry.
7821         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7822         passed by invisible reference specially.
7823         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7824         (vt_add_function_parameter): Correctly deal with a parameter passed by
7825         invisible reference.
7827 2013-08-30  Jan Hubicka  <jh@suse.cz>
7829         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7831 2013-08-30  Richard Biener  <rguenther@suse.de>
7833         PR tree-optimization/58228
7834         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7835         allow invariant loads in nested loop vectorization.
7837 2013-08-30  Richard Biener  <rguenther@suse.de>
7839         PR tree-optimization/58223
7840         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7841         (has_anti_or_output_dependence): ... this and adjust to also
7842         look for output dependences.
7843         (mark_nodes_having_upstream_mem_writes): Adjust.
7844         (rdg_flag_uses): Likewise.
7846 2013-08-30  Richard Biener  <rguenther@suse.de>
7848         PR tree-optimization/58010
7849         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7850         assert that we have a loop-closed PHI.
7852 2013-08-29  Jan Hubicka  <jh@suse.cz>
7854         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7855         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7856         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7857         (lto_free_function_in_decl_state_for_node): New function.
7859 2013-08-29  Xinliang David Li  <davidxl@google.com>
7861         * loop-unroll.c (report_unroll_peel): Minor message change.
7862         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7863         Emit alignment peeling message with default -fopt-info.
7864         (vect_loop_versioning): Emit loop version info message.
7865         * tree-vectorizer.c (vectorize_loops): Minor message change.
7866         (execute_vect_slp): Ditto.
7868 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7870         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7871         of the clone from the DECL_NAME of the original function.
7873 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7875         * passes.c (register_pass): Add overload.
7876         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7878 2013-08-29  Jan Hubicka  <jh@suse.cz>
7880         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7881         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7882         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7883         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7884         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7885         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7886         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7887         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7889 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7891         * dumpfile.c (dump_loc): Output column number.
7892         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7893         * doc/invoke.texi: Document optall -fopt-info flag.
7894         * profile.c (read_profile_edge_counts): Use new dump framework.
7895         (compute_branch_probabilities): Ditto.
7896         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7897         when pass not in any opt group.
7898         * pass_manager.h (pass_manager::get_pass_profile): New method.
7899         * value-prof.c (check_counter): Use new dump framework.
7900         (check_ic_target): Ditto.
7901         * coverage.c (get_coverage_counts): Ditto.
7902         (coverage_init): Setup new dump framework.
7904 2013-08-29  Richard Biener  <rguenther@suse.de>
7906         PR tree-optimization/58246
7907         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7908         handle the dominance check inside a basic-block.
7910 2013-08-29  Richard Biener  <rguenther@suse.de>
7912         PR middle-end/57287
7913         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7914         of default defs that appear in abnormal PHI nodes.
7916 2013-08-29  Richard Biener  <rguenther@suse.de>
7918         PR tree-optimization/57685
7919         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7920         single-use operands to avoid exponential complexity.
7922 2013-08-28  Dehao Chen  <dehao@google.com>
7924         * ipa-inline.c (edge_badness): Fix integer underflow.
7926 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7928         * gtm-builtins.def (_ITM_free): Declare leaf.
7930 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7932         PR target/58067
7933         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7934         (*tls_local_dynamic_base_64_largepic): Likewise.
7935         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7936         Remove predicate from call operand.
7937         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7938         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7940 2013-08-28  Jeff Law  <law@redhat.com>
7942         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7943         checks for the number of predecessors and successors allowed.
7944         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7945         which require copying a joiner block if there is a request which
7946         is a subpath that requires no joiner block copying.
7948 2013-08-28  Jan Hubicka  <jh@suse.cz>
7950         * lto-streamer-out.c (DFS_write_tree_body): Drop
7951         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
7952         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
7953         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
7954         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
7955         Do not read DECL_ERROR_ISSUED.
7956         (unpack_ts_decl_with_vis_value_fields): Do not read
7957         DECL_DEFER_OUTPUT.
7958         (lto_input_ts_binfo_tree_pointers): Do not read
7959         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
7960         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
7961         write DECL_ERROR_ISSUED..
7962         (pack_ts_decl_with_vis_value_fields): Do not write
7963         DECL_DEFER_OUTPUT.
7964         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
7965         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
7966         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
7967         * tree.h (tree_decl_common): Update comment.
7968         (DECL_ERROR_ISSUED): Remove.
7970 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7972         PR middle-end/58257
7973         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
7975 2013-08-28  Jan Hubicka  <jh@suse.cz>
7977         * builtins.def (free): Declare leaf.
7979 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
7981         * gdbhooks.py: New.
7982         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
7983         * configure: Regenerate.
7985 2013-08-27  Martin Jambor  <mjambor@suse.cz>
7987         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
7988         (ipa_ancestor_jf_data): Likewise.
7989         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
7990         (ipa_get_jf_pass_through_type_preserved): New function.
7991         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
7992         (ipa_get_jf_ancestor_type_preserved): New function.
7993         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
7994         (ipa_get_jf_ancestor_result): Likewise.
7995         (propagate_vals_accross_pass_through): Use
7996         ipa_get_jf_pass_through_result to do all the value mappings.
7997         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
7998         type_preserved flag.
7999         (ipa_set_jf_cst_copy): New function.
8000         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8001         (ipa_set_jf_arith_pass_through): Likewise.
8002         (ipa_set_ancestor_jf): Likewise.
8003         (compute_complex_assign_jump_func): Set type_preserved instead of
8004         punting.
8005         (ipa_compute_jump_functions_for_edge): Likewise.
8006         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8007         (update_jump_functions_after_inlining): Update type_preserved.
8008         Explicitely create jump functions when combining one with pass_through.
8009         (ipa_write_jump_function): Stream the type_preserved flags.
8010         (ipa_read_jump_function): Likewise.
8012 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8013             Aldy Hernandez  <aldyh@redhat.com>
8015         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8016         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8017         * function.h (struct function): Add has_force_vect_loops and
8018         has_simduid_loops.
8019         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8020         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8021         handle it.
8022         * gimple.def: Update CLAUSES comments.
8023         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8024         (gimple_build_omp_for): Add argument to prototype.
8025         (gimple_omp_for_kind): New.
8026         (gimple_omp_for_set_kind): New.
8027         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8028         GOVD_DATA_SHARE_CLASS.
8029         (enum omp_region_type): Add ORT_SIMD.
8030         (gimple_add_tmp_var): Handle ORT_SIMD.
8031         (gimplify_var_or_parm_decl): Same.
8032         (is_gimple_stmt): Same.
8033         (omp_firstprivatize_variable): Same.
8034         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8035         (omp_notice_variable): Handle ORT_SIMD.
8036         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8037         (omp_check_private): Handle ORT_SIMD.
8038         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8039         OMP_CLAUSE_SAFELEN.
8040         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8041         Handle OMP_CLAUSE_LASTPRIVATE.
8042         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8043         OMP_CLAUSE_SAFELEN.
8044         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8045         (gimplify_expr): Handle OMP_SIMD.
8046         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8047         (expand_GOMP_SIMD_VF): New.
8048         (expand_GOMP_SIMD_LAST_LANE): New.
8049         * internal-fn.def (GOMP_SIMD_LANE): New.
8050         (GOMP_SIMD_VF): New.
8051         (GOMP_SIMD_LAST_LANE): New.
8052         * omp-low.c: Include target.h.
8053         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8054         OMP_CLAUSE_SAFELEN.
8055         (check_omp_nesting_restrictions): Same.
8056         (omp_max_vf): New.
8057         (lower_rec_simd_input_clauses): New.
8058         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8059         OMP_CLAUSE_LINEAR.
8060         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8061         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8062         (expand_omp_build_assign): New.
8063         (expand_omp_for_init_counts): New.
8064         (expand_omp_for_init_vars): New.
8065         (extract_omp_for_update_vars): New.
8066         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8067         and rewrite accordingly.
8068         (expand_omp_simd): New.
8069         (expand_omp_for): Use expand_omp_simd.
8070         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8071         (lower_omp_for): Do not lower the body.
8072         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8073         in their own loops.
8074         * tree-flow.h (find_omp_clause): Remove prototype.
8075         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8076         forcing vectorization of the loop, or if flag_tree_vectorize.
8077         (gate_tree_if_conversion): Similarly.
8078         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8079         gimple_build_omp_for.
8080         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8081         * tree-parloops (create_parallel_loop): Pass kind argument to
8082         gimple_build_omp_for.
8083         * tree-pretty-print.c (dump_omp_clause): Add cases for
8084         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8085         OMP_CLAUSE__SIMDUID_.
8086         (dump_generic_node): Handle OMP_SIMD.
8087         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8088         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8089         unroll OMP_SIMD loops here.
8090         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8091         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8092         loop->safelen.
8093         (vect_analyze_data_refs): Handle simd loops.
8094         * tree-vect-loop.c (vectorizable_live_operation): Handle
8095         IFN_GOMP_SIMD*.
8096         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8097         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8098         (vectorizable_load): Same.
8099         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8100         (struct simduid_to_vf): New.
8101         (simduid_to_vf::hash): New.
8102         (simduid_to-vf::equal): New.
8103         (struct simd_array_to_simduid): New.
8104         (simd_array_to_simduid::hash): New.
8105         (simd_array_to_simduid::equal): New.
8106         (adjust_simduid_builtins): New.
8107         (struct note_simd_array_uses_struct): New.
8108         (note_simd_array_uses_cb): New.
8109         (note_simd_array_uses): New.
8110         (vectorize_loops): Handle simd hints and adjust simd builtins
8111         accordingly.
8112         * tree-vectorizer.h (struct _stmt_vec_info): Add
8113         simd_lane_access_p field.
8114         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8115         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8116         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8117         (omp_clause_code_name): Same.
8118         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8119         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8120         * tree.def (OMP_SIMD): New entry.
8121         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8122         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8123         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8124         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8125         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8126         (OMP_CLAUSE_LINEAR_STEP): New.
8127         (OMP_CLAUSE_SAFELEN_EXPR): New.
8128         (OMP_CLAUSE__SIMDUID__DECL): New.
8129         (find_omp_clause): New prototype.
8131 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8133         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8134         Haswell processor detection.
8136 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8138         PR target/57927
8139         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8140         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8141         AVX2 capable processors.
8143 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8145         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8146         of vget_low_* with implementations in terms of other intrinsics.
8148 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8150         PR middle-end/57219
8151         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8152         values to -1, 0 and 1.
8154 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8156         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8157         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8158         (<optab><mode>3_insn): Remove.
8159         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8160         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8161         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8162         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8163         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8164         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8165         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8166         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8167         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8168         (ror<mode>3_insn): Likewise.
8169         * config/aarch64/predicates.md (aarch64_simd_register): New.
8171 2013-08-27  Richard Biener  <rguenther@suse.de>
8173         PR tree-optimization/57521
8174         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8175         one edge is non-critical.
8176         (find_phi_replacement_condition): Make sure to use a non-critical
8177         edge.  Cleanup and remove old bug workarounds.
8178         (bb_postdominates_preds): Remove.
8179         (if_convertible_loop_p_1): Do not compute post-dominators.
8180         (combine_blocks): Do not free post-dominators.
8181         (main_tree_if_conversion): Likewise.
8182         (pass_data_if_conversion): Add TODO_verify_ssa.
8184 2013-08-27  DJ Delorie  <dj@redhat.com>
8186         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8188 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8190         * function.c (assign_parm_find_data_types): Set passed_mode and
8191         nominal_mode to the TYPE_MODE of nominal_type for the built
8192         pointer type in case of the struct-pass-by-reference.
8194 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8196         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8197         (UINT16_TYPE): Change default to "unsigned int".
8199         * config/avr/avr.opt (mfract-convert-truncate): New option.
8200         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8201         is set, round negative fractional integers according to n1169
8202         when converting to integer types.
8204 2013-08-26  Jan Hubicka  <jh@suse.cz>
8206         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8207         methods can not be called indirectly when their address is not taken.
8209 2013-08-26  Jan Hubicka  <jh@suse.cz>
8211         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8212         ctor_for_folding.
8214 2013-08-26  Jan Hubicka  <jh@suse.cz>
8216         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8217         can be unshared.
8219 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8221         * reload.c (find_valid_class): Allow classes that do not include
8222         FIRST_PSEUDO_REGISTER - 1.
8224 2013-08-26  Jan Hubicka  <jh@suse.cz>
8226         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8227         fix edge count/frequency when speculation failed; fix type check
8228         for the direct call.
8230 2013-08-26  Jan Hubicka  <jh@suse.cz>
8232         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8234 2013-08-26  Jan Hubicka  <jh@suse.cz>
8236         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8237         to be changed by edge redirection.
8239 2013-08-26  Jan Hubicka  <jh@suse.cz>
8241         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8242         formating; add sanity check.
8243         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8244         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8245         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8246         (resolve_noninline_speculation): Update callee keys, too.
8248 2013-08-26  Jan Hubicka  <jh@suse.cz>
8250         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8251         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8253 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8255         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8256         into proper place.
8258 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8260         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8261         (x86_64_elf_select_section): Ditto.
8262         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8263         arguments.
8264         (ix86_pass_by_reference): Ditto.
8265         (output_set_got): Ditto.
8266         (ix86_unary_operator_ok): Ditto.
8267         (ix86_expand_builtin): Ditto.
8269 2013-08-23  Jan Hubicka  <jh@suse.cz>
8271         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8273 2013-08-23  Jan Hubicka  <jh@suse.cz>
8275         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8276         (tree_decl_with_vis): Add FINAL field.
8278 2013-08-23  Jeff Law  <law@redhat.com>
8280         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8281         the debugging dump when the expression is fully redundant.
8283 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8285         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8286         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8287         * pretty-print.c (pp_formatted_text_data): Likewise.
8288         (pp_write_text_to_stream): Likewise.
8289         (pp_write_text_as_dot_label_to_stream): Likewise.
8290         (pp_append_r): Likewise.
8291         (pp_format): Likewise.
8292         (pp_flush): Likewise.
8293         (pp_clear_output_area): Likewise.
8294         (pp_append_text): Likewise.
8295         (pp_formatted_text): Likewise.
8296         (pp_remaining_character_count_for_line): Likewise.
8297         (pp_newline): Likewise.
8298         (pp_character): Likewise.
8299         (output_buffer::~output_buffer): Define.
8300         (pretty_printer::~pretty_printer): Destruct output buffer.
8301         * pretty-print.h (output_buffer::~output_buffer): Declare.
8302         (pretty_printer::~pretty_printer): Declare virtual.
8304 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8306         PR other/57324
8307         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8308         HOST_WIDE_INT_M1U): New macros.
8309         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8310         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8311         unsigned -1 for lshift.
8312         * cse.c (cse_insn): Likewise.
8313         * double-int.c (rshift_double, lshift_double): Likewise.
8314         * builtins.c (fold_builtin_bitop): Likewise.
8315         * combine.c (force_to_mode): Likewise.
8316         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8317         * simplify-rtx.c (simplify_const_unary_operation,
8318         simplify_const_binary_operation): Likewise.
8319         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8320         check_va_list_escapes): Likewise.
8321         * rtlanal.c (nonzero_bits1): Likewise.
8322         * expmed.c (expand_smod_pow2): Likewise.
8323         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8325 2013-08-23  Jan Hubicka  <jh@suse.cz>
8327         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8328         as having address taken.
8330 2013-08-23  Jan Hubicka  <jh@suse.cz>
8332         * ipa-utils.h (method_class_type): Declare.
8333         * ipa-devirt.c (method_class_type): Export.
8335         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8336         do not walk base classes of anonymous types.
8338 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8340         PR rtl-optimization/58220
8341         PR regression/58221
8342         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8343         handle SEQUENCE insns properly.
8345 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8347         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8348         definition.
8349         (pp_newline_and_indent): Likewise.
8350         (pp_separate_with): Likewise.
8351         * pretty-print.c (pp_newline_and_flush): Define.
8352         (pp_newline_and_indent): Likewise.
8353         (pp_separate_with): Likewise.
8355 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8357         PR target/58218
8358         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8359         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8361 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8363         * config/i386/predicates.md (ext_sse_reg_operand): New.
8364         * config/i386/i386.md (*movti_internal): Use
8365         predicate to determine if EVEX is needed.
8366         (*movsi_internal): Ditto.
8367         (*movdf_internal): Ditto.
8368         (*movsf_internal): Ditto.
8369         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8371 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8373         PR tree-optimization/58209
8374         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8375         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8376         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8377         emit POINTER_PLUS_EXPR.
8378         (create_tailcall_accumulator): For pointer result type accumulate in
8379         sizetype type.
8381 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8383         * configure.ac: Add backslashes missing from the last change.
8384         * configure: Regenerate.
8386 2013-08-22  Jan Hubicka  <jh@susue.cz>
8388         * ipa.c (function_and_variable_visibility): First remember function
8389         was global and then make it local.
8391 2013-08-22  Julian Brown  <julian@codesourcery.com>
8393         * configure.ac: Add aarch64 to list of arches which use "nop" in
8394         debug_line test.
8395         * configure: Regenerate.
8397 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8399         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8400         gnu_libc_has_function.
8401         * config/s390/tpf.h: Likewise.
8403 2013-08-22  Jan Hubicka  <jh@susue.cz>
8405         * timevar.c (validate_phases): Add cast.
8407 2013-08-22  Jan Hubicka  <jh@susue.cz>
8409         * timevar.c (validate_phases): Use size_t for memory.
8410         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8412 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8414         * pretty-print.h (output_buffer::output_buffer): Declare.
8415         (pretty_printer::pretty_printer): Likewise.
8416         (pp_construct): Remove.
8417         * pretty-print.c (output_buffer::output_buffer): Define.
8418         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8419         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8420         (print_gimple_expr): Likewise.
8421         (print_gimple_seq): Likewise.
8422         (gimple_dump_bb): Likewise.
8423         * sched-vis.c (dump_value_slim): Likewise.
8424         (dump_insn_slim): Likewise.
8425         (dump_rtl_slim): Likewise.
8426         (str_pattern_slim): Likewise.
8427         * tree-mudflap.c (mf_varname_tree): Likewise.
8428         * graph.c (print_graph_cfg): Likewise.
8429         (start_graph_dump): Likewise.
8430         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8431         placement-new.
8432         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8433         pretty printer initialization.
8434         * coretypes.h (diagnostic_context): Remove superflous type alias
8435         declaration.
8436         (pretty_printer): Likewise.  Declare directly as a class.
8437         (pretty_print_info): Remove declaration as class.
8438         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8439         and pp_clear_output_area.
8440         (asan_add_global): Likewise.
8442 2013-08-22  Jan Hubicka  <jh@suse.cz>
8444         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8445         * ipa-utils.h (update_type_inheritance_graph): Declare.
8446         (possible_polymorphic_call_target_p): Declare.
8447         (possible_polymorphic_call_target_p): New.
8448         * ipa-devirt.c: Update toplevel comments.
8449         (cached_polymorphic_call_targets): Move up.
8450         (odr_type_d): Move ID down.
8451         (polymorphic_type_binfo_p): Update comment.
8452         (odr_hasher::remove): Likewise;
8453         (get_odr_type): Set anonymous_namespace.
8454         (dump_odr_type): Dump it.
8455         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8456         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8457         (record_binfo): Add comment.
8458         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8459         built.
8460         (devirt_node_removal_hook): Do not iCE when cache is freed.
8461         (possible_polymorphic_call_target_p): New predicate.
8462         (update_type_inheritance_graph): New function.
8464 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8465             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8466             Sergey Lega  <sergey.s.lega@intel.com>
8467             Anna Tikhonova  <anna.tikhonova@intel.com>
8468             Ilya Tocar  <ilya.tocar@intel.com>
8469             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8470             Ilya Verbin  <ilya.verbin@intel.com>
8471             Kirill Yukhin  <kirill.yukhin@intel.com>
8472             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8474         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8475         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8476         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8477         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8478         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8479         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8480         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8481         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8482         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8483         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8484         OPT_mavx512pf, OPT_mavx512er cases.
8485         * config/i386/constraints.md (v): New constraint.
8486         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8487         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8488         (bit_AVX512CD): New.
8489         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8490         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8491         * config/i386/i386-c.c (ix86_target_macros_internal):
8492         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8493         __AVX512PF__.
8494         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8495         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8496         * config/i386/i386.c (regclass_map, dbx_register_map)
8497         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8498         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8499         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8500         -mavx512pf options.
8501         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8502         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8503         -mavx512pf options.  Fix formatting.
8504         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8505         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8506         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8507         -mavx512cd, -mavx512pf options.
8508         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8509         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8510         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8511         ALL_SSE_REGS.
8512         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8513         (ix86_set_reg_reg_cost): Ditto.
8514         (x86_order_regs_for_local_alloc): Ditto.
8515         (MAX_VECT_LEN): Extend to 64-byte.
8516         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8517         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8518         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8519         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8520         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8521         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8522         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8523         (SSE_REG_MODE_P): Support new modes.
8524         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8525         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8526         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8527         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8528         (REG_CLASS_CONTENTS): Add new registers.
8529         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8530         (EXT_REX_SSE_REGNO_P): New.
8531         (HI_REGISTER_NAMES): Add new registers.
8532         * config/i386/i386.md: Define constants for new registers.
8533         (mode): Add new 512-bit modes.
8534         (prefix): Support evex prefix.
8535         (isa): Support avx512f, noavx512f, fma_avx512f.
8536         (ssemodesuffix): Add new 512-bit modes.
8537         (movxi): New.
8538         (*movxi_internal_avx512f): Ditto.
8539         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8540         Support MODE_XI.
8541         (*movsi_internal): Likewise.
8542         (*movdf_internal): Likewise.
8543         (*movsf_internal): Likewise.
8544         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8545         (<code><mode>3): Likewise.
8546         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8547         New.
8548         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8549         with the new constraint "v".
8550         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8551         modes.
8552         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8553         with the new constraint "v".
8554         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8555         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8556         (*<plusminus_insn><mode>3): Likewise.
8557         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8558         (*mul<mode>3): Likewise.
8559         (<sse>_vmmul<mode>3): Likewise.
8560         (<sse>_div<mode>3): Likewise.
8561         (<sse>_vmdiv<mode>3): Likewise.
8562         (<sse>_sqrt<mode>2): Likewise.
8563         (<sse>_vmsqrt<mode>2): Likewise.
8564         (*<code><mode>3_finite): Likewise.
8565         (*<code><mode>3) <smaxmin>: Likewise.
8566         (<sse>_vm<code><mode>3): Likewise.
8567         (*<code><mode>3) <any_logic>: Likewise.
8568         (*fma_fmadd_<mode>): Likewise.
8569         (*fma_fmsub_<mode>): Likewise.
8570         (*fma_fnmadd_<mode>): Likewise.
8571         (*fma_fnmsub_<mode>): Likewise.
8572         (*fma_fmaddsub_<mode>): Likewise.
8573         (*fma_fmsubadd_<mode>): Likewise.
8574         (*fmai_fmadd_<mode>): Likewise.
8575         (*fmai_fmsub_<mode>): Likewise.
8576         (*fmai_fnmadd_<mode>): Likewise.
8577         (*fmai_fnmsub_<mode>): Likewise.
8578         (sse_cvtsi2ss): Likewise.
8579         (sse_cvtsi2ssq): Likewise.
8580         (sse_cvtss2si): Likewise.
8581         (sse_cvtss2si_2): Likewise.
8582         (sse_cvtss2siq): Likewise.
8583         (sse_cvtss2siq_2): Likewise.
8584         (sse_cvttss2si): Likewise.
8585         (sse_cvtss2siq_2): Likewise.
8586         (float<sseintvecmodelower><mode>2): Likewise.
8587         (sse2_cvtsd2si_2): Likewise.
8588         (sse2_cvtsd2siq_2): Likewise.
8589         (*<plusminus_insn><mode>3): Likewise.
8590         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8591         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8592         (ashr<mode>3): Likewise.
8593         (<shift_insn><mode>3): Likewise.
8594         (avx2_<code><mode>3): Likewise.
8595         (*avx2_<code><mode>3): Likewise.
8596         (*andnot<mode>3): Likewise.
8597         (*<code><mode>3) <any_logic>: Likewise.
8598         (abs<mode>2): Likewise.
8599         (avx2_permvar<mode>): Likewise.
8600         (avx2_perm<mode>_1): Likewise.
8601         (*avx_vpermilp<mode>): Likewise.
8602         (avx_vpermilvar<mode>3): Likewise.
8603         (avx2_ashrv<mode>): Likewise.
8604         (avx2_<shift_insn>v<mode>): Likewise.
8605         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8606         -mavx512cd.
8607         * doc/rtl.texi: Document XImode.
8609 2013-08-21  Jeff Law  <law@redhat.com>
8611         * tree-flow.h (register_jump_thread): Pass vector of edges
8612         instead of each important edge.
8613         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8614         thread path into a vector and pass that to register_jump_thread.
8615         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8616         passed in edge vector to the current 3-edge form.
8618         Revert:
8619         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8621         * dce.c (fini_dce): Call df_analyze again just in case
8622         delete_unmarked_insns removed anything.
8624 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8626         * reload.h (struct reg_equivs): Rename to ..
8627         (struct reg_equivs_s): .. this.
8629 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8631         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8633 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8635         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8637 2013-08-21  Jeff Law  <law@redhat.com>
8639         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8640         simplify assignments too.  If the RHS collapses to a singleton
8641         range, then return the value for the range.
8643 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8645         * config/i386/sse.md (V16): Rename to...
8646         (VMOVE): this.
8647         (mov<mode>): Update iterator name.
8648         (*mov<mode>_internal): Ditto.
8649         (push<mode>1): Ditto.
8650         (movmisalign<mode>): Ditto.
8652 2013-08-20  Jan Hubicka  <jh@suse.cz>
8654         PR bootstrap/58186
8655         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8656         entry for direct edges.
8657         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8659 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8661         Revert my last two changes, r201865 and r201864:
8663         Revert r201865:
8664         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8666         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8667         instances can own GC refs.
8669         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8670         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8671         (gcc::context::gt_pch_nx): Likewise.
8672         (gcc::context::gt_pch_nx):  Likewise.
8673         * ggc.h (gt_ggc_mx <T>): New.
8674         (gt_pch_nx_with_op <T>): New.
8675         (gt_pch_nx <T>): New.
8676         * passes.c (opt_pass::gt_ggc_mx): New.
8677         (opt_pass::gt_pch_nx): New.
8678         (opt_pass::gt_pch_nx_with_op): New.
8679         (pass_manager::gt_ggc_mx): New.
8680         (pass_manager::gt_pch_nx): New.
8681         (pass_manager::gt_pch_nx_with_op): New.
8682         (pass_manager::operator new): Use
8683         ggc_internal_cleared_alloc_stat rather than xcalloc.
8684         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8685         (pass_manager::gt_ggc_mx): New.
8686         (pass_manager::gt_pch_nx): New.
8687         (pass_manager::gt_pch_nx_with_op): New.
8688         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8689         (opt_pass::operator new): New.
8690         (opt_pass::gt_ggc_mx): New.
8691         (opt_pass::gt_pch_nx): New.
8692         (opt_pass::gt_pch_nx_with_op): New.
8694         Revert r201864:
8695         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8697         * Makefile.in (GTFILES): Add context.h.
8698         * context.c (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         * context.h (gcc::context): Add GTY((user)) marking.
8703         (gcc::context::operator new): New.
8704         (gcc::context::gt_ggc_mx): New.
8705         (gcc::context::gt_pch_nx): New.
8706         (gcc::context::gt_pch_nx): New.
8707         (g): Add GTY marking.
8708         (gt_ggc_mx (gcc::context *)): New.
8709         (gt_pch_nx (gcc::context *)): New.
8710         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8711         void *cookie)): New.
8712         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8714 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8716         * dce.c (fini_dce): Call df_analyze again just in case
8717         delete_unmarked_insns removed anything.
8719 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8721         PR rtl-optimizations/57451
8722         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8723         from crossing split section boundaries.
8725 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8727         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8728         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8729         with MULTLIB_DEFAULTS.
8731 2013-08-20  Nick Clifton  <nickc@redhat.com>
8733         * target.def (narrow_volatile_bitfield): Note that the default
8734         value is false, not !TARGET_STRICT_ALIGN.
8735         * doc/tm.texi: Regenerate.
8737 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8739         Fix LIB_SPEC for systems without libpthread.
8741         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8742         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8743         for Android.
8744         * config/i386/linux-common.h: Likewise.
8745         * config/mips/linux-common.h: Likewise.
8747 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8749         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8750         checks.
8752 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8754         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8755         instances can own GC refs.
8757         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8758         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8759         (gcc::context::gt_pch_nx): Likewise.
8760         (gcc::context::gt_pch_nx):  Likewise.
8761         * ggc.h (gt_ggc_mx <T>): New.
8762         (gt_pch_nx_with_op <T>): New.
8763         (gt_pch_nx <T>): New.
8764         * passes.c (opt_pass::gt_ggc_mx): New.
8765         (opt_pass::gt_pch_nx): New.
8766         (opt_pass::gt_pch_nx_with_op): New.
8767         (pass_manager::gt_ggc_mx): New.
8768         (pass_manager::gt_pch_nx): New.
8769         (pass_manager::gt_pch_nx_with_op): New.
8770         (pass_manager::operator new): Use
8771         ggc_internal_cleared_alloc_stat rather than xcalloc.
8772         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8773         (pass_manager::gt_ggc_mx): New.
8774         (pass_manager::gt_pch_nx): New.
8775         (pass_manager::gt_pch_nx_with_op): New.
8776         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8777         (opt_pass::operator new): New.
8778         (opt_pass::gt_ggc_mx): New.
8779         (opt_pass::gt_pch_nx): New.
8780         (opt_pass::gt_pch_nx_with_op): New.
8782 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8784         * Makefile.in (GTFILES): Add context.h.
8785         * context.c (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         * context.h (gcc::context): Add GTY((user)) marking.
8790         (gcc::context::operator new): New.
8791         (gcc::context::gt_ggc_mx): New.
8792         (gcc::context::gt_pch_nx): New.
8793         (gcc::context::gt_pch_nx): New.
8794         (g): Add GTY marking.
8795         (gt_ggc_mx (gcc::context *)): New.
8796         (gt_pch_nx (gcc::context *)): New.
8797         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8798         void *cookie)): New.
8799         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8801 2013-08-20  Alan Modra  <amodra@gmail.com>
8803         PR target/57865
8804         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8805         (rs6000_emit_epilogue): Likewise.
8807 2013-08-19  Dehao Chen  <dehao@google.com>
8809         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8811 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8812             Jakub Jelinek  <jakub@redhat.com>
8814         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8815         (BUILT_IN_FABSD64): Likewise.
8816         (BUILT_IN_FABSD128): Likewise.
8817         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8818         (fold_builtin_1): Likewise.
8819         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8820         destination and source operands.
8821         (*abstd2_fpr): Likewise.
8822         (*nabstd2_fpr): Likewise.
8824 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8826         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8827         JUMP_P and INSN_P.
8829 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8831         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8832         incomplete.
8834 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8836         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8837         * builtins.c (default_libc_has_function): New.
8838         (gnu_libc_has_function): Ditto.
8839         (no_c99_libc_has_function): Ditto.
8840         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8841         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8842         (fold_builtin_sincos): Likewise.
8843         (fold_builtin_cexp): Likewise.
8844         * builtins.def (DEF_C94_BUILTIN): Likewise.
8845         (DEF_C99_BUILTIN): Likewise.
8846         (DEF_C99_C90RES_BUILTIN): Likewise.
8847         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8848         definitions to using this define.
8849         * config/darwin-protos.h (darwin_libc_has_function): New.
8850         * config/darwin.c (darwin_libc_has_function): Ditto.
8851         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8852         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8853         * config/darwin.h: Ditto.
8854         * config/elfos.h: Ditto.
8855         * config/freebsd.h: Ditto.
8856         * config/i386/cygming.h: Ditto.
8857         * config/i386/djgpp.h: Ditto.
8858         * config/i386/i386-interix.h: Ditto.
8859         * config/microblaze/microblaze.h: Ditto.
8860         * config/mmix/mmix.h: Ditto.
8861         * config/gnu-user.h: Ditto.
8862         * config/ia64/hpux.h: Ditto.
8863         * config/pa/pa-hpux.h: Ditto.
8864         * config/pdp11/pdp11.h: Ditto.
8865         * config/picochip/picochip.h: Ditto.
8866         * config/linux.h: Ditto.
8867         * config/netbsd.h: Ditto.
8868         * config/openbsd.h: Ditto.
8869         * config/rs6000/aix43.h: Ditto.
8870         * config/rs6000/aix51.h: Ditto.
8871         * config/rs6000/aix52.h: Ditto.
8872         * config/rs6000/aix53.h: Ditto.
8873         * config/rs6000/aix61.h: Ditto.
8874         * config/rs6000/darwin.h: Ditto.
8875         * config/rs6000/linux.h: Ditto.
8876         * config/rs6000/linux64.h: Ditto.
8877         * config/s390/tpf.h: Ditto.
8878         * config/sol2-10.h: Ditto.
8879         * config/sol2.h: Ditto.
8880         * config/vms/vms.h: Ditto.
8881         * config/vxworks.h: Ditto.
8882         * config/linux-android.c (linux_android_libc_has_function):
8883         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8884         * config/linux-protos.h (linux_android_libc_has_function):
8885         New declaration.
8886         * config/i386/i386.c (ix86_libc_has_function): New.
8887         * config/i386/i386-protos.h
8888         (ix86_libc_has_function): New declaration.
8889         * config/i386/i386.md
8890         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8891         ("isinf<mode>2): Likewise.
8892         * convert.c (convert_to_integer): Using new target hook
8893         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8894         TARGET_C99_FUNCTIONS.
8895         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8896         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8897         * coretypes.h (function_class): New enum for different
8898         classes of functions.
8899         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8900         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8901         (TARGET_HAS_SINCOS): Likewise.
8902         (TARGET_LIBC_HAS_FUNCTION): New.
8903         * doc/tm.texi: Regenerated.
8904         * targhooks.h (default_libc_has_function): New declaration.
8905         (no_c99_libc_has_function): Ditto.
8906         (gnu_libc_has_function): Ditto.
8907         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8908         and TARGET_HAS_SINCOS.
8910 2013-08-18  Jan Hubicka  <jh@suse.cz>
8912         * Makeifle-in (ipa-devirt.o): New.
8913         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8914         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8915         (analyze_functions): Look into possible targets of polymorphic call.
8916         * dumpfile.c (dump_files): Add type-inheritance dump.
8917         * dumpfile.h (TDI_inheritance): New.
8918         * ipa-devirt.c: New file.
8919         * ipa-utils.h (odr_type_d): Forward declare.
8920         (odr_type): New type.
8921         (build_type_inheritance_graph): Declare.
8922         (possible_polymorphic_call_targets): Declare and introduce inline
8923         variant when only edge is pased.
8924         (dump_possible_polymorphic_call_targets): Likewise.
8925         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8926         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8927         (types_same_for_odr): ... here.
8928         * tree.h (type_in_anonymous_namespace_p): Declare.
8930 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8932         PR tree-optimization/58006
8933         * tree-parloops.c (take_address_of): Don't ICE if get_name
8934         returns NULL.
8935         (eliminate_local_variables_stmt): Remove clobber stmts.
8937 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8939         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8940         error message is issued for an alias to undefined symbol.
8942 2013-08-18  Jan Hubicka  <jh@suse.cz>
8944         * cgraph.c (cgraph_create_indirect_edge): Discover
8945         polymorphic calls and record basic info into indirect_info.
8946         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8947         devirtualization, ignore objc function calls.
8948         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
8949         call with no parm index info.
8950         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
8951         * tree.c (virtual_method_call_p): New function.
8952         * tree.h (virtual_method_call_p): Declare.
8954 2013-08-16  Jan Hubicka  <jh@suse.cz>
8956         PR middle-end/58179
8957         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
8959 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
8961         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
8962         attribute.
8964 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
8966         * gengtype.c (type_for_name): Add special-case support for
8967         locating types within the "gcc::" namespace.
8968         (open_base_files): Emit a "using namespace gcc" directive.
8970 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
8972         PR target/58160
8973         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
8974         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
8976         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
8977         array instead of each individual operand as a separate argument.
8978         (emit_fusion_gpr_load): Likewise.
8979         (expand_fusion_gpr_load): Add new function declaration.
8981         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
8982         signature to have the operands passed as an array, instead of as
8983         separate arguments.  Allow ZERO_EXTEND to be in the memory
8984         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
8985         depend on the register live/dead flags when peepholes are run.
8986         (expand_fusion_gpr_load): New function to be called from the
8987         peephole2 pass, to change the register that addis sets to be the
8988         target register.
8989         (emit_fusion_gpr_load): Change the calling signature to have the
8990         operands passed as an array, instead of as separate arguments.
8991         Allow ZERO_EXTEND to be in the memory address, and also
8992         SIGN_EXTEND if -mpower8-fusion-sign.
8994         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
8995         unspec enumeration.
8996         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
8997         adjust the register addis loads up in the peephole2 pass.  Do not
8998         depend on the register live/dead state when the peephole pass is done.
9000 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9002         * gengtype.c (create_user_defined_type): Ensure that the kind
9003         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9004         declaration is seen before the GTY((user)) marking.
9006 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9008         PR target/58105
9009         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9011 2013-08-16  Jan Hubicka  <jh@suse.cz>
9013         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9014         arugment expected_type.
9015         (gimple_fold_call): Use it.
9016         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9017         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9018         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9019         (try_make_edge_direct_virtual_call): Likewise.
9020         * tree.c (obj_type_ref_class): New.
9021         * tree.h (obj_type_ref_class): Use it.
9023 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9025         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9026         (rtl_slim_pp): Likewise.
9027         (init_rtl_slim_pretty_print): Likewise.
9028         (dump_value_slim):  Don't call it.  Use local pretty printer.
9029         (dump_insn_slim): Likewise.
9030         (dump_rtl_slim): Likewise.
9031         (str_pattern_slim): Likewise.
9032         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9033         Simplify.
9035 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9037         PR tree-optimization/58164
9038         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9039         walk gimple_goto_dest of GIMPLE_GOTO.
9041         PR tree-optimization/58165
9042         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9043         bi_call must be the last stmt in a bb, don't split_block, instead
9044         use fallthru edge from it and give up if there is none.
9045         Release conds vector when returning early.
9047 2013-08-14  Xinliang David Li  <davidxl@google.com>
9049         * config/i386/i386.c (ix86_option_override_internal):
9050         Remove unused variable and field.
9052 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9054         PR target/57949
9055         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9056         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9057         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9058         and Linux, correct BLKmode alignment when 128-bit alignment is
9059         required and compatibility flag is not set.
9060         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9061         for zero-size arguments when compatibility flag is not set.
9063 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9065         PR tree-optimization/58145
9066         * tree-sra.c (build_ref_for_offset): If prev_base has
9067         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9069 2013-08-14  Xinliang David Li  <davidxl@google.com>
9071         * config/i386/i386.c (ix86_option_override_internal):
9072         Fix uninitialized variable error.
9074 2013-08-14  Xinliang David Li  <davidxl@google.com>
9076         * config/i386/i386.opt: Define two new options.
9077         * config/i386/x86-tune.def: Add arch selector field in macros.
9078         * config/i386/i386.h: Adjust macro definition.
9079         * config/i386/i386.c (ix86_option_override_internal):
9080         Refactor the code.
9081         (parse_mtune_ctrl_str): New function.
9082         (set_ix86_tune_features): New function.
9083         (ix86_function_specific_restore): Call the new helper function.
9085 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9087         PR rtl-optimization/57662
9088         * sel-sched.c (code_motion_process_successors): When the current insn
9089         is removed after the recursive traversal, break from the loop.
9090         Add comments and debug printouts.
9092 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9093             Alexandre Oliva  <aoliva@redhat.com>
9095         PR target/58067
9096         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9097         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9098         there also UNSPEC_PLTOFF.
9100 2013-08-14  Marek Polacek  <polacek@redhat.com>
9102         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9103         NUM_CONDITIONS bit positions.
9105 2013-08-13  Cary Coutant  <ccoutant@google.com>
9107         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9108         (attr_checksum): Hash vector contents instead of pointer.
9109         (attr_checksum_ordered): Likewise.
9111 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9113         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9114         when Pmode != word_mode.  Add length_address attribute.
9115         (sse3_monitor_<mode>): Merge from sse3_monitor and
9116         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9117         Pmode != word_mode.  Update insn length attribute.
9118         * config/i386/i386.c (ix86_option_override_internal): Update
9119         ix86_gen_monitor selection for merged sse3_monitor insn.
9121 2013-08-13  Julian Brown  <julian@codesourcery.com>
9123         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9124         perform invalid legitimization on greater-than-word-size modes for
9125         TARGET_E500_DOUBLE.
9127 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9129         * ira.c (setup_class_translate_array): Use aclass instead of cl
9130         for classes not fully covered by allocno classes.
9132 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9134         PR tree-optimization/57661
9135         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9136         * tree-inline.c (tree_function_versioning): Initialize it.
9137         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9138         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9139         that is not being copied.
9141         PR sanitizer/56417
9142         * asan.c (instrument_strlen_call): Fix typo in comment.
9143         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9145 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9147         * config/mips/mips.md (prefetch): Use lw instead of ld on
9148         loongson in 32bit mode.
9150 2013-08-13  Nick Clifton  <nickc@redhat.com>
9152         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9154 2013-08-13  Jan Hubicka  <jh@suse.cz>
9156         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9157         introduced edge; fix typo in sanity check.
9158         (cgraph_resolve_speculation): Export; improve diagnostic.
9159         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9160         speculation at type mismatch.
9161         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9162         (cgraph_resolve_speculation): Declare.
9163         (symtab_can_be_discarded): New function.
9164         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9165         * ipa-inline-transform.c (speculation_removed): New global var.
9166         (clone_inlined_nodes): See if speculation can be removed.
9167         (inline_call): If speculations was removed, we growths may not match.
9168         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9169         (speculation_useful_p): New function.
9170         (resolve_noninline_speculation): New function.
9171         (inline_small_functions): Resolve useless speculations.
9172         * ipa-inline.h (speculation_useful_p): Declare
9173         * ipa.c (can_replace_by_local_alias): Simplify.
9174         (ipa_profile): Produce speculative calls in non-lto, too;
9175         add simple cost model; produce local aliases.
9177 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9179         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9180         PASS_MANAGER_H.
9182 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9184         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9186 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9188         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9189         iterator instead of X87MODEF.
9191 2013-08-12  Perez Read  <netfirewall@gmail.com>
9193         PR target/58132
9194         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9195         operand 0 for intel asm alternative.
9196         (*movabs<mode>_2): Ditto for operand 1.
9198 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9200         * config/aarch64/arm_none.h
9201         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9203 2013-08-12  Nick Clifton  <nickc@redhat.com>
9205         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9207 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9209         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9210         expand for QI/HImode operand to produce more effictive code for
9211         unsigned char(short) --> float(double) conversion.
9213 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9215         * doc/invoke.texi: Mention that -ftls-model does not force the final
9216         model.
9218 2013-08-12  Marek Polacek  <polacek@redhat.com>
9219             Marc Glisse  <marc.glisse@inria.fr>
9221         PR tree-optimization/57980
9222         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9223         when creating -1 constant.
9225 2013-08-10  Jan Hubicka  <jh@suse.cz>
9227         Workaround binutils PR14342.
9228         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9229         (gimple_init_edge_profiler): Likewise.
9230         (gimple_gen_ic_func_profiler): Likewise.
9232 2013-08-09  Jan Hubicka  <jh@suse.cz>
9234         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9236 2013-08-09  Xinliang David Li  <davidxl@google.com>
9238         * config/i386/stringop.def: New file.
9239         * config/i386/stringop.opt: New file.
9240         * config/i386/i386-opts.h: Include stringopt.def.
9241         * config/i386/i386.opt: Include stringopt.opt.
9242         * config/i386/i386.c (ix86_option_override_internal):
9243         Override default size based stringop inline strategies with options.
9244         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9245         New function.
9247 2013-08-09  Jan Hubicka  <jh@suse.cz>
9249         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9251 2013-08-09  Jan Hubicka  <jh@suse.cz>
9253         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9254         CGRAPH_FREQ_MAX.
9255         (dump_cgraph_node): Dump profile-id.
9256         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9257         and common_target_probability.
9258         * lto-cgraph.c (lto_output_edge): Stream common targets.
9259         (lto_output_node): Stream profile ids.
9260         (input_node): Stream profile ids.
9261         (input_edge): Stream common targets.
9262         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9263         * ipa.c: Include value-prof.h
9264         (ipa_profile_generate_summary): Turn indirect call statement histograms
9265         into common targets.
9266         (ipa_profile): Turn common targets into speculative edges.
9268 2013-08-09  Jan Hubicka  <jh@suse.cz>
9270         * cgraph.h (cgraph_node): Add profile_id.
9271         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9272         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9273         (del_node_map): Update.
9274         (find_func_by_funcdef_no): Replace by ...
9275         (find_func_by_profile_id): ... this one.
9276         (gimple_ic_transform): Do not remove useful histograms when
9277         speculation is not done; dump info when indirect call removal
9278         can happen at LTO.
9279         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9280         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9281         (__gcov_indirect_call_profiler_v2): .. this one.
9282         * profile.h (init_node_map): Update.
9283         * coverage.c (coverage_compute_profile_id): New function.
9284         * coverage.h (coverage_compute_profile_id): Declare.
9285         * tree-profile.c (init_ic_make_global_vars): Make
9286         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9287         (gimple_init_edge_profiler): Update prototype of
9288         __gcov_indirect_call_profiler.
9289         (gimple_gen_ic_func_profiler): Simplify.
9290         (tree_profiling): Use init_node_map
9292 2013-08-09  Jan Hubicka  <jh@suse.cz>
9294         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9295         non-speculative refs.
9296         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9297         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9298         (cgraph_set_call_stmt): Likewise.
9299         (cgraph_create_edge_1): Fix release checking compilatoin;
9300         clear lto_stmt_uid.
9301         (cgraph_free_edge): Free indirect info.
9302         (cgraph_turn_edge_to_speculative): New function.
9303         (cgraph_speculative_call_info): New function.
9304         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9305         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9306         (dump_cgraph_node): Dump speculation.
9307         (verify_edge_count_and_frequency): Accept speculative edges.
9308         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9309         (verify_cgraph_node): Handle speculation.
9310         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9311         (cgraph_set_call_stmt): Update prototype.
9312         (cgraph_make_edge_direct): Update prototype.
9313         (cgraph_speculative_call_info): Declare.
9314         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9315         to change; update call of ipa_find_references.
9316         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9317         lto_stmt_uid and speculative flags.
9318         (ipa_dump_references): Dump speculation.
9319         (ipa_clone_references): Clone speculative flag.
9320         (ipa_clone_referring): Likewise.
9321         (ipa_clone_ref): New function.
9322         (ipa_find_reference): Look into lto_stmt_uids
9323         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9324         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9325         (ipa_find_reference): Update declaration.
9326         (ipa_clone_ref): Declare.
9327         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9328         stream speculative flag.
9329         (lto_output_ref): Stream statements uids and speculation.
9330         (input_ref): Likewise.
9331         (input_edge): Stream speuclation.
9332         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9333         (cgraph_set_call_stmt_including_clones): Handle speculation.
9334         * ipa-inline.c (heap_edge_removal_hook): New function.
9335         (inline_small_functions): Register it.
9336         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9337         also initialize refs.
9338         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9339         edge to change.
9340         (try_make_edge_direct_simple_call): Likewise.
9341         (try_make_edge_direct_simple_call): Likewise.
9342         (update_indirect_edges_after_inlining): Likewise.
9343         (remove_described_reference): Look proper lto_stmt_uid.
9344         (propagate_controlled_uses): Likewise.
9345         (propagate_controlled_uses): Liekwise.
9346         * tree-inline.c (copy_bb): Copy speculative edges.
9347         (redirect_all_calls): New function.
9348         (copy_cfg_body): Do redirection after loop info is updated.
9349         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9351 2013-08-09  Jan Hubicka  <jh@suse.cz>
9353         * lto-streamer-out.c (output_function): Renumber PHIs.
9354         * lto-streamer-in.c (input_function): Likewise.
9356 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9358         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9359         (get_lane_unsigned): Likewise.
9360         (dup_lane_scalar): Likewise.
9361         (get_lane): enable for VALL.
9362         * config/aarch64/aarch64-simd.md
9363         (aarch64_dup_lane_scalar<mode>): Remove.
9364         (aarch64_get_lane_signed<mode>): Likewise.
9365         (aarch64_get_lane_unsigned<mode>): Likewise.
9366         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9367         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9368         (aarch64_get_lane<mode>): Enable for all vector modes.
9369         (aarch64_get_lanedi): Remove misleading constraints.
9370         * config/aarch64/arm_neon.h
9371         (__aarch64_vget_lane_any): Define.
9372         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9373         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9374         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9375         * config/aarch64/iterators.md (VDQQH): New.
9376         (VDQQHS): Likewise.
9377         (vwcore): Likewise.
9379 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9381         * configure.ac: Add GAS check for LEON instructions on SPARC.
9382         * configure: Regenerate.
9383         * config.in: Likewise.
9384         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9385         sparc*-*-* block.
9386         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9387         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9388         for LEON or LEON3.
9389         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9390         (AS_LEON_FLAG): New macro.
9391         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9392         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9393         Deal with LEON and LEON3 for the memory model.
9394         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9395         (atomic_compare_and_swap<mode>_1): Likewise.
9396         (*atomic_compare_and_swap<mode>_1): Likewise.
9398 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9400         * config/arm/neon.md (vcond): Fix floating-point vector
9401         comparisons against 0.
9403 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9405         * lra-constraints.c (emit_spill_move): Remove assert.
9406         (process_alt_operands): Add more debugging
9407         output.  Increase reject for spilling into memory.  Decrease
9408         reject for reloading scratch.
9409         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9411 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9413         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9414         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9415         micromips incompatible.  Add nan2008.
9416         (MULTILIB_DIRNAMES): Add nan2008.
9417         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9418         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9419         and micromips incompatible.  Add nan2008.
9420         (MULTILIB_DIRNAMES): Add nan2008.
9421         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9423 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9425         PR rtl-optimization/58079
9426         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9427         simplify_comparison has widened a comparison with an integer.
9429 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9431         * config/arm/neon.md (movmisalign<mode>): Disable when we
9432         don't allow unaligned accesses.
9433         (*movmisalign<mode>_neon_store): Likewise.
9434         (*movmisalign<mode>_neon_load): Likewise.
9435         (*movmisalign<mode>_neon_store): Likewise.
9436         (*movmisalign<mode>_neon_load): Likewise.
9438 2013-08-08  Jan Hubicka  <jh@suse.cz>
9440         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9441         (make_pass_rebuild_cgraph_edges): Also clear references.
9442         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9443         * ipa-inline-transform.c (inline_transform): Remove all references
9444         after inlining.
9445         * cgraphunit.c (expand_function): Remove all references after
9446         expansion.
9447         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9448         (ipa_find_reference): Rewrite to iterator.
9449         (remove_stmt_references): Likewise.
9450         (ipa_clear_stmts_in_references): New function.
9451         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9452         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9453         clear references.
9454         * ipa-split.c (split_function): Remove references in split function.
9456 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9458         PR target/57431
9459         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9460         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9462 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9464         PR target/56979
9465         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9466         suggested mode for the assignment isn't compatible with the
9467         registers required.
9469 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9471         PR target/58065
9472         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9474 2013-08-07  Xinliang David Li  <davidxl@google.com>
9476         * config/i386/i386.opt: New option -mtune-ctrl=.
9477         * config/i386/x86-tune.def: New file.
9478         * config/i386/i386.h: include x86-tune.def.
9479         * config/i386/i386.c (ix86_option_override_internal):
9480         Parsing -mtune-ctrl= option and set tune features.
9482 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9484         PR other/12081
9485         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9486         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9487         to gen_2arg_fn_t.
9489 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9491         * rtl.h (update_alignments): Declare.
9492         * final.c (grow_label_align): New function extracted from...
9493         (shorten_branches): ...here.  Call it.
9494         (update_alignments): New function.
9495         * reorg.c (sibling_labels): New variable.
9496         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9497         the new label along with it onto the sibling_labels vector.
9498         (fill_simple_delay_slots): Adjust call to get_label_before.
9499         (fill_slots_from_thread): Likewise.
9500         (relax_delay_slots): Likewise.
9501         (make_return_insns): Likewise.
9502         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9504 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9506         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9507         document its semantics.
9508         (diagnostic_report_diagnostic): Adjust accordingly.
9510 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9512         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9513         (sparc_option_override): ...and port to new C++ pass API.
9514         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9516 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9518         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9520 2013-08-06  Caroline Tice  <cmtice@google.com>
9522         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9523         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9524         * tree-pass.h: Add pass_vtable_verify.
9525         * varasm.c (assemble_variable): Add code to properly set the comdat
9526         section and name for the .vtable_map_vars section.
9527         (assemble_vtyv_preinit_initializer): New function.
9528         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9529         LINK_ONCE flag.
9530         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9531         * vtable-verify.c: New file.
9532         * vtable-verify.h: New file.
9533         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9534         initialiation levels.
9535         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9536         * passes.def: Insert pass_vtable_verify.
9537         * aclocal.m4: Reorder includes.
9538         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9539         -fvtv-counts options.
9540         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9541         as appropriate, if -fvtable-verify=... is used.
9542         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9543         -fvtable-verify=... is used.
9544         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9545         (vtable-verify.o): Add new build rule.
9546         (GTFILES): Add vtable-verify.c to list.
9547         * common.opt (fvtable-verify=): New flag.
9548         (vtv_priority): Values for fvtable-verify= flag.
9549         (fvtv-counts): New flag.
9550         (fvtv-debug): New flag.
9551         * tree.h (save_vtable_map_decl): New extern function decl.
9553 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9555         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9556         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9557         (pass_data_rl78_devirt): ...new pass_data instance and...
9558         (make_pass_rl78_devirt): ...new function.
9559         (rl78_asm_file_start): Port pass registration to new C++ API.
9561 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9563         * coretypes.h (rtl_opt_pass): Add.
9564         (gcc::context): Add.
9565         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9566         (epiphany_init): Port to new C++ pass API.
9567         (epiphany_optimize_mode_switching): Likewise.
9568         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9569         (pass_manager::get_pass_mode_switching): New.
9570         (pass_manager::get_pass_peephole2): New.
9571         * mode-switching.c (pass_mode_switching): Add clone method.
9572         * recog.c (pass_peephole2): Add clone method.
9573         (pass_split_all_insns): Add clone method.
9575 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9577         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9578         (mips_option_override): ...here, porting to new C++ API for passes.
9580 2013-08-06  Jan Hubicka  <jh@suse.cz>
9582         * cgraph.c (cgraph_get_body): New function based on lto.c
9583         implementation.
9584         * cgraph.h (cgraph_get_body): Declare.
9585         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9586         LTO paths.
9587         * cgraphunit.c (expand_function): Get body prior expanding.
9588         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9589         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9590         really need.
9591         * passes.c (do_per_function_toporder): Get body.
9592         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9593         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9595 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9597         PR fortran/57987
9598         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9599         is not re-finalized.  Rename second parameter to no_collect.
9601 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9603         PR middle-end/58041
9604         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9605         MEM_REF has proper alignment information.
9607 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9609         PR other/12081
9610         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9611         class insn_gen_fn.
9612         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9613         rtx (*) (rtx, ...) with insn_gen_fn.
9614         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9615         insn_gen_fn::stored_funcptr.  Add initializer braces.
9617 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9619         Rewrite how instances of passes are cloned to remove assumptions
9620         about their sizes (thus allowing pass subclasses to have
9621         additional data fields, albeit non-GC-managed ones at this point).
9623         * passes.c (make_pass_instance): Now that passes have clone
9624         methods, rewrite this function to eliminate XNEW and memcpy
9625         calls that used hardcoded sizes.  Since this function no longer
9626         creates pass instances, rename it to...
9627         (add_pass_instance): ...this.  Document the old way that passes were
9628         numbered and flagged, and rework this function to continue using it.
9629         (next_pass_1): Add an initial_pass argument for use by
9630         add_pass_instance.
9631         (position_pass): When adding multiple instances of a pass, use
9632         the pass's clone method, rather than relying on the XNEW/memcpy
9633         within the former make_pass_instance (now add_pass_instance).
9634         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9635         the initial instance of the current pass within the pass manager.
9637 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9639         This is the automated part of the conversion of passes from C
9640         structs to C++ classes.
9642         Patch autogenerated by refactor_passes.py from
9643         https://github.com/davidmalcolm/gcc-refactoring-scripts
9644         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9646         * asan.c (pass_asan): Convert from a global struct to a subclass of
9647         gimple_opt_pass along with...
9648         (pass_data_asan): ...new pass_data instance and...
9649         (make_pass_asan): ...new function.
9650         (pass_asan_O0): Convert from a global struct to a subclass of
9651         gimple_opt_pass along with...
9652         (pass_data_asan_O0): ...new pass_data instance and...
9653         (make_pass_asan_O0): ...new function.
9654         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9655         subclass of rtl_opt_pass along with...
9656         (pass_data_inc_dec): ...new pass_data instance and...
9657         (make_pass_inc_dec): ...new function.
9658         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9659         a subclass of rtl_opt_pass along with...
9660         (pass_data_reorder_blocks): ...new pass_data instance and...
9661         (make_pass_reorder_blocks): ...new function.
9662         (pass_duplicate_computed_gotos): Convert from a global struct to a
9663         subclass of rtl_opt_pass along with...
9664         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9665         (make_pass_duplicate_computed_gotos): ...new function.
9666         (pass_partition_blocks): Convert from a global struct to a subclass of
9667         rtl_opt_pass along with...
9668         (pass_data_partition_blocks): ...new pass_data instance and...
9669         (make_pass_partition_blocks): ...new function.
9670         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9671         struct to a subclass of rtl_opt_pass along with...
9672         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9673         and...
9674         (make_pass_branch_target_load_optimize1): ...new function.
9675         (pass_branch_target_load_optimize2): Convert from a global struct to a
9676         subclass of rtl_opt_pass along with...
9677         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9678         and...
9679         (make_pass_branch_target_load_optimize2): ...new function.
9680         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9681         of rtl_opt_pass along with...
9682         (pass_data_jump): ...new pass_data instance and...
9683         (make_pass_jump): ...new function.
9684         (pass_jump2): Convert from a global struct to a subclass of
9685         rtl_opt_pass along with...
9686         (pass_data_jump2): ...new pass_data instance and...
9687         (make_pass_jump2): ...new function.
9688         * cfgexpand.c (pass_expand): Convert from a global struct to a
9689         subclass of rtl_opt_pass along with...
9690         (pass_data_expand): ...new pass_data instance and...
9691         (make_pass_expand): ...new function.
9692         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9693         of rtl_opt_pass along with...
9694         (pass_data_free_cfg): ...new pass_data instance and...
9695         (make_pass_free_cfg): ...new function.
9696         (pass_into_cfg_layout_mode): Convert from a global struct to a
9697         subclass of rtl_opt_pass along with...
9698         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9699         (make_pass_into_cfg_layout_mode): ...new function.
9700         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9701         subclass of rtl_opt_pass along with...
9702         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9703         (make_pass_outof_cfg_layout_mode): ...new function.
9704         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9705         struct to a subclass of gimple_opt_pass along with...
9706         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9707         (make_pass_build_cgraph_edges): ...new function.
9708         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9709         subclass of gimple_opt_pass along with...
9710         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9711         (make_pass_rebuild_cgraph_edges): ...new function.
9712         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9713         subclass of gimple_opt_pass along with...
9714         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9715         and...
9716         (make_pass_remove_cgraph_callee_edges): ...new function.
9717         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9718         struct to a subclass of rtl_opt_pass along with...
9719         (pass_data_stack_adjustments): ...new pass_data instance and...
9720         (make_pass_stack_adjustments): ...new function.
9721         * combine.c (pass_combine): Convert from a global struct to a subclass
9722         of rtl_opt_pass along with...
9723         (pass_data_combine): ...new pass_data instance and...
9724         (make_pass_combine): ...new function.
9725         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9726         global struct to a subclass of rtl_opt_pass along with...
9727         (pass_data_compare_elim_after_reload): ...new pass_data instance
9728         and...
9729         (make_pass_compare_elim_after_reload): ...new function.
9730         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9731         of rtl_opt_pass along with...
9732         (pass_data_rtl_cprop): ...new pass_data instance and...
9733         (make_pass_rtl_cprop): ...new function.
9734         * cse.c (pass_cse): Convert from a global struct to a subclass of
9735         rtl_opt_pass along with...
9736         (pass_data_cse): ...new pass_data instance and...
9737         (make_pass_cse): ...new function.
9738         (pass_cse2): Convert from a global struct to a subclass of
9739         rtl_opt_pass along with...
9740         (pass_data_cse2): ...new pass_data instance and...
9741         (make_pass_cse2): ...new function.
9742         (pass_cse_after_global_opts): Convert from a global struct to a
9743         subclass of rtl_opt_pass along with...
9744         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9745         (make_pass_cse_after_global_opts): ...new function.
9746         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9747         of rtl_opt_pass along with...
9748         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9749         (make_pass_ud_rtl_dce): ...new function.
9750         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9751         rtl_opt_pass along with...
9752         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9753         (make_pass_fast_rtl_dce): ...new function.
9754         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9755         a subclass of rtl_opt_pass along with...
9756         (pass_data_df_initialize_opt): ...new pass_data instance and...
9757         (make_pass_df_initialize_opt): ...new function.
9758         (pass_df_initialize_no_opt): Convert from a global struct to a
9759         subclass of rtl_opt_pass along with...
9760         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9761         (make_pass_df_initialize_no_opt): ...new function.
9762         (pass_df_finish): Convert from a global struct to a subclass of
9763         rtl_opt_pass along with...
9764         (pass_data_df_finish): ...new pass_data instance and...
9765         (make_pass_df_finish): ...new function.
9766         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9767         rtl_opt_pass along with...
9768         (pass_data_rtl_dse1): ...new pass_data instance and...
9769         (make_pass_rtl_dse1): ...new function.
9770         (pass_rtl_dse2): Convert from a global struct to a subclass of
9771         rtl_opt_pass along with...
9772         (pass_data_rtl_dse2): ...new pass_data instance and...
9773         (make_pass_rtl_dse2): ...new function.
9774         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9775         subclass of rtl_opt_pass along with...
9776         (pass_data_dwarf2_frame): ...new pass_data instance and...
9777         (make_pass_dwarf2_frame): ...new function.
9778         * except.c (pass_set_nothrow_function_flags): Convert from a global
9779         struct to a subclass of rtl_opt_pass along with...
9780         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9781         and...
9782         (make_pass_set_nothrow_function_flags): ...new function.
9783         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9784         subclass of rtl_opt_pass along with...
9785         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9786         and...
9787         (make_pass_convert_to_eh_region_ranges): ...new function.
9788         * final.c (pass_compute_alignments): Convert from a global struct to a
9789         subclass of rtl_opt_pass along with...
9790         (pass_data_compute_alignments): ...new pass_data instance and...
9791         (make_pass_compute_alignments): ...new function.
9792         (pass_final): Convert from a global struct to a subclass of
9793         rtl_opt_pass along with...
9794         (pass_data_final): ...new pass_data instance and...
9795         (make_pass_final): ...new function.
9796         (pass_shorten_branches): Convert from a global struct to a subclass of
9797         rtl_opt_pass along with...
9798         (pass_data_shorten_branches): ...new pass_data instance and...
9799         (make_pass_shorten_branches): ...new function.
9800         (pass_clean_state): Convert from a global struct to a subclass of
9801         rtl_opt_pass along with...
9802         (pass_data_clean_state): ...new pass_data instance and...
9803         (make_pass_clean_state): ...new function.
9804         * function.c (pass_instantiate_virtual_regs): Convert from a global
9805         struct to a subclass of rtl_opt_pass along with...
9806         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9807         (make_pass_instantiate_virtual_regs): ...new function.
9808         (pass_leaf_regs): Convert from a global struct to a subclass of
9809         rtl_opt_pass along with...
9810         (pass_data_leaf_regs): ...new pass_data instance and...
9811         (make_pass_leaf_regs): ...new function.
9812         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9813         subclass of rtl_opt_pass along with...
9814         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9815         and...
9816         (make_pass_thread_prologue_and_epilogue): ...new function.
9817         (pass_match_asm_constraints): Convert from a global struct to a
9818         subclass of rtl_opt_pass along with...
9819         (pass_data_match_asm_constraints): ...new pass_data instance and...
9820         (make_pass_match_asm_constraints): ...new function.
9821         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9822         subclass of rtl_opt_pass along with...
9823         (pass_data_rtl_fwprop): ...new pass_data instance and...
9824         (make_pass_rtl_fwprop): ...new function.
9825         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9826         rtl_opt_pass along with...
9827         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9828         (make_pass_rtl_fwprop_addr): ...new function.
9829         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9830         rtl_opt_pass along with...
9831         (pass_data_rtl_pre): ...new pass_data instance and...
9832         (make_pass_rtl_pre): ...new function.
9833         (pass_rtl_hoist): Convert from a global struct to a subclass of
9834         rtl_opt_pass along with...
9835         (pass_data_rtl_hoist): ...new pass_data instance and...
9836         (make_pass_rtl_hoist): ...new function.
9837         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9838         subclass of gimple_opt_pass along with...
9839         (pass_data_lower_cf): ...new pass_data instance and...
9840         (make_pass_lower_cf): ...new function.
9841         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9842         from a global struct to a subclass of gimple_opt_pass along with...
9843         (pass_data_strength_reduction): ...new pass_data instance and...
9844         (make_pass_strength_reduction): ...new function.
9845         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9846         of rtl_opt_pass along with...
9847         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9848         (make_pass_rtl_ifcvt): ...new function.
9849         (pass_if_after_combine): Convert from a global struct to a subclass of
9850         rtl_opt_pass along with...
9851         (pass_data_if_after_combine): ...new pass_data instance and...
9852         (make_pass_if_after_combine): ...new function.
9853         (pass_if_after_reload): Convert from a global struct to a subclass of
9854         rtl_opt_pass along with...
9855         (pass_data_if_after_reload): ...new pass_data instance and...
9856         (make_pass_if_after_reload): ...new function.
9857         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9858         a subclass of rtl_opt_pass along with...
9859         (pass_data_initialize_regs): ...new pass_data instance and...
9860         (make_pass_initialize_regs): ...new function.
9861         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9862         of ipa_opt_pass_d along with...
9863         (pass_data_ipa_cp): ...new pass_data instance and...
9864         (make_pass_ipa_cp): ...new function.
9865         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9866         global struct to a subclass of gimple_opt_pass along with...
9867         (pass_data_inline_parameters): ...new pass_data instance and...
9868         (make_pass_inline_parameters): ...new function.
9869         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9870         subclass of gimple_opt_pass along with...
9871         (pass_data_early_inline): ...new pass_data instance and...
9872         (make_pass_early_inline): ...new function.
9873         (pass_ipa_inline): Convert from a global struct to a subclass of
9874         ipa_opt_pass_d along with...
9875         (pass_data_ipa_inline): ...new pass_data instance and...
9876         (make_pass_ipa_inline): ...new function.
9877         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9878         struct to a subclass of gimple_opt_pass along with...
9879         (pass_data_local_pure_const): ...new pass_data instance and...
9880         (make_pass_local_pure_const): ...new function.
9881         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9882         ipa_opt_pass_d along with...
9883         (pass_data_ipa_pure_const): ...new pass_data instance and...
9884         (make_pass_ipa_pure_const): ...new function.
9885         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9886         to a subclass of ipa_opt_pass_d along with...
9887         (pass_data_ipa_reference): ...new pass_data instance and...
9888         (make_pass_ipa_reference): ...new function.
9889         * ipa-split.c (pass_split_functions): Convert from a global struct to
9890         a subclass of gimple_opt_pass along with...
9891         (pass_data_split_functions): ...new pass_data instance and...
9892         (make_pass_split_functions): ...new function.
9893         (pass_feedback_split_functions): Convert from a global struct to a
9894         subclass of gimple_opt_pass along with...
9895         (pass_data_feedback_split_functions): ...new pass_data instance and...
9896         (make_pass_feedback_split_functions): ...new function.
9897         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9898         global struct to a subclass of simple_ipa_opt_pass along with...
9899         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9900         instance and...
9901         (make_pass_ipa_function_and_variable_visibility): ...new function.
9902         (pass_ipa_free_inline_summary): Convert from a global struct to a
9903         subclass of simple_ipa_opt_pass along with...
9904         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9905         (make_pass_ipa_free_inline_summary): ...new function.
9906         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9907         subclass of ipa_opt_pass_d along with...
9908         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9909         and...
9910         (make_pass_ipa_whole_program_visibility): ...new function.
9911         (pass_ipa_profile): Convert from a global struct to a subclass of
9912         ipa_opt_pass_d along with...
9913         (pass_data_ipa_profile): ...new pass_data instance and...
9914         (make_pass_ipa_profile): ...new function.
9915         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9916         ipa_opt_pass_d along with...
9917         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9918         (make_pass_ipa_cdtor_merge): ...new function.
9919         * ira.c (pass_ira): Convert from a global struct to a subclass of
9920         rtl_opt_pass along with...
9921         (pass_data_ira): ...new pass_data instance and...
9922         (make_pass_ira): ...new function.
9923         (pass_reload): Convert from a global struct to a subclass of
9924         rtl_opt_pass along with...
9925         (pass_data_reload): ...new pass_data instance and...
9926         (make_pass_reload): ...new function.
9927         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9928         subclass of rtl_opt_pass along with...
9929         (pass_data_cleanup_barriers): ...new pass_data instance and...
9930         (make_pass_cleanup_barriers): ...new function.
9931         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9932         of rtl_opt_pass along with...
9933         (pass_data_loop2): ...new pass_data instance and...
9934         (make_pass_loop2): ...new function.
9935         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9936         rtl_opt_pass along with...
9937         (pass_data_rtl_loop_init): ...new pass_data instance and...
9938         (make_pass_rtl_loop_init): ...new function.
9939         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9940         rtl_opt_pass along with...
9941         (pass_data_rtl_loop_done): ...new pass_data instance and...
9942         (make_pass_rtl_loop_done): ...new function.
9943         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9944         subclass of rtl_opt_pass along with...
9945         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9946         (make_pass_rtl_move_loop_invariants): ...new function.
9947         (pass_rtl_unswitch): Convert from a global struct to a subclass of
9948         rtl_opt_pass along with...
9949         (pass_data_rtl_unswitch): ...new pass_data instance and...
9950         (make_pass_rtl_unswitch): ...new function.
9951         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
9952         subclass of rtl_opt_pass along with...
9953         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
9954         and...
9955         (make_pass_rtl_unroll_and_peel_loops): ...new function.
9956         (pass_rtl_doloop): Convert from a global struct to a subclass of
9957         rtl_opt_pass along with...
9958         (pass_data_rtl_doloop): ...new pass_data instance and...
9959         (make_pass_rtl_doloop): ...new function.
9960         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
9961         a subclass of rtl_opt_pass along with...
9962         (pass_data_lower_subreg): ...new pass_data instance and...
9963         (make_pass_lower_subreg): ...new function.
9964         (pass_lower_subreg2): Convert from a global struct to a subclass of
9965         rtl_opt_pass along with...
9966         (pass_data_lower_subreg2): ...new pass_data instance and...
9967         (make_pass_lower_subreg2): ...new function.
9968         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
9969         struct to a subclass of ipa_opt_pass_d along with...
9970         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
9971         (make_pass_ipa_lto_gimple_out): ...new function.
9972         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
9973         of ipa_opt_pass_d along with...
9974         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
9975         (make_pass_ipa_lto_finish_out): ...new function.
9976         * mode-switching.c (pass_mode_switching): Convert from a global struct
9977         to a subclass of rtl_opt_pass along with...
9978         (pass_data_mode_switching): ...new pass_data instance and...
9979         (make_pass_mode_switching): ...new function.
9980         * modulo-sched.c (pass_sms): Convert from a global struct to a
9981         subclass of rtl_opt_pass along with...
9982         (pass_data_sms): ...new pass_data instance and...
9983         (make_pass_sms): ...new function.
9984         * omp-low.c (pass_expand_omp): Convert from a global struct to a
9985         subclass of gimple_opt_pass along with...
9986         (pass_data_expand_omp): ...new pass_data instance and...
9987         (make_pass_expand_omp): ...new function.
9988         (pass_lower_omp): Convert from a global struct to a subclass of
9989         gimple_opt_pass along with...
9990         (pass_data_lower_omp): ...new pass_data instance and...
9991         (make_pass_lower_omp): ...new function.
9992         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
9993         of gimple_opt_pass along with...
9994         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
9995         (make_pass_diagnose_omp_blocks): ...new function.
9996         * passes.c (pass_early_local_passes): Convert from a global struct to
9997         a subclass of simple_ipa_opt_pass along with...
9998         (pass_data_early_local_passes): ...new pass_data instance and...
9999         (make_pass_early_local_passes): ...new function.
10000         (pass_all_early_optimizations): Convert from a global struct to a
10001         subclass of gimple_opt_pass along with...
10002         (pass_data_all_early_optimizations): ...new pass_data instance and...
10003         (make_pass_all_early_optimizations): ...new function.
10004         (pass_all_optimizations): Convert from a global struct to a subclass
10005         of gimple_opt_pass along with...
10006         (pass_data_all_optimizations): ...new pass_data instance and...
10007         (make_pass_all_optimizations): ...new function.
10008         (pass_all_optimizations_g): Convert from a global struct to a subclass
10009         of gimple_opt_pass along with...
10010         (pass_data_all_optimizations_g): ...new pass_data instance and...
10011         (make_pass_all_optimizations_g): ...new function.
10012         (pass_rest_of_compilation): Convert from a global struct to a subclass
10013         of rtl_opt_pass along with...
10014         (pass_data_rest_of_compilation): ...new pass_data instance and...
10015         (make_pass_rest_of_compilation): ...new function.
10016         (pass_postreload): Convert from a global struct to a subclass of
10017         rtl_opt_pass along with...
10018         (pass_data_postreload): ...new pass_data instance and...
10019         (make_pass_postreload): ...new function.
10020         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10021         subclass of rtl_opt_pass along with...
10022         (pass_data_gcse2): ...new pass_data instance and...
10023         (make_pass_gcse2): ...new function.
10024         * postreload.c (pass_postreload_cse): Convert from a global struct to
10025         a subclass of rtl_opt_pass along with...
10026         (pass_data_postreload_cse): ...new pass_data instance and...
10027         (make_pass_postreload_cse): ...new function.
10028         * predict.c (pass_profile): Convert from a global struct to a subclass
10029         of gimple_opt_pass along with...
10030         (pass_data_profile): ...new pass_data instance and...
10031         (make_pass_profile): ...new function.
10032         (pass_strip_predict_hints): Convert from a global struct to a subclass
10033         of gimple_opt_pass along with...
10034         (pass_data_strip_predict_hints): ...new pass_data instance and...
10035         (make_pass_strip_predict_hints): ...new function.
10036         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10037         of rtl_opt_pass along with...
10038         (pass_data_peephole2): ...new pass_data instance and...
10039         (make_pass_peephole2): ...new function.
10040         (pass_split_all_insns): Convert from a global struct to a subclass of
10041         rtl_opt_pass along with...
10042         (pass_data_split_all_insns): ...new pass_data instance and...
10043         (make_pass_split_all_insns): ...new function.
10044         (pass_split_after_reload): Convert from a global struct to a subclass
10045         of rtl_opt_pass along with...
10046         (pass_data_split_after_reload): ...new pass_data instance and...
10047         (make_pass_split_after_reload): ...new function.
10048         (pass_split_before_regstack): Convert from a global struct to a
10049         subclass of rtl_opt_pass along with...
10050         (pass_data_split_before_regstack): ...new pass_data instance and...
10051         (make_pass_split_before_regstack): ...new function.
10052         (pass_split_before_sched2): Convert from a global struct to a subclass
10053         of rtl_opt_pass along with...
10054         (pass_data_split_before_sched2): ...new pass_data instance and...
10055         (make_pass_split_before_sched2): ...new function.
10056         (pass_split_for_shorten_branches): Convert from a global struct to a
10057         subclass of rtl_opt_pass along with...
10058         (pass_data_split_for_shorten_branches): ...new pass_data instance
10059         and...
10060         (make_pass_split_for_shorten_branches): ...new function.
10061         * ree.c (pass_ree): Convert from a global struct to a subclass of
10062         rtl_opt_pass along with...
10063         (pass_data_ree): ...new pass_data instance and...
10064         (make_pass_ree): ...new function.
10065         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10066         subclass of rtl_opt_pass along with...
10067         (pass_data_stack_regs): ...new pass_data instance and...
10068         (make_pass_stack_regs): ...new function.
10069         (pass_stack_regs_run): Convert from a global struct to a subclass of
10070         rtl_opt_pass along with...
10071         (pass_data_stack_regs_run): ...new pass_data instance and...
10072         (make_pass_stack_regs_run): ...new function.
10073         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10074         subclass of rtl_opt_pass along with...
10075         (pass_data_cprop_hardreg): ...new pass_data instance and...
10076         (make_pass_cprop_hardreg): ...new function.
10077         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10078         subclass of rtl_opt_pass along with...
10079         (pass_data_reginfo_init): ...new pass_data instance and...
10080         (make_pass_reginfo_init): ...new function.
10081         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10082         of rtl_opt_pass along with...
10083         (pass_data_regmove): ...new pass_data instance and...
10084         (make_pass_regmove): ...new function.
10085         * regrename.c (pass_regrename): Convert from a global struct to a
10086         subclass of rtl_opt_pass along with...
10087         (pass_data_regrename): ...new pass_data instance and...
10088         (make_pass_regrename): ...new function.
10089         * reorg.c (pass_delay_slots): Convert from a global struct to a
10090         subclass of rtl_opt_pass along with...
10091         (pass_data_delay_slots): ...new pass_data instance and...
10092         (make_pass_delay_slots): ...new function.
10093         (pass_machine_reorg): Convert from a global struct to a subclass of
10094         rtl_opt_pass along with...
10095         (pass_data_machine_reorg): ...new pass_data instance and...
10096         (make_pass_machine_reorg): ...new function.
10097         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10098         of rtl_opt_pass along with...
10099         (pass_data_sched): ...new pass_data instance and...
10100         (make_pass_sched): ...new function.
10101         (pass_sched2): Convert from a global struct to a subclass of
10102         rtl_opt_pass along with...
10103         (pass_data_sched2): ...new pass_data instance and...
10104         (make_pass_sched2): ...new function.
10105         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10106         to a subclass of rtl_opt_pass along with...
10107         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10108         (make_pass_stack_ptr_mod): ...new function.
10109         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10110         to a subclass of rtl_opt_pass along with...
10111         (pass_data_rtl_store_motion): ...new pass_data instance and...
10112         (make_pass_rtl_store_motion): ...new function.
10113         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10114         of gimple_opt_pass along with...
10115         (pass_data_tracer): ...new pass_data instance and...
10116         (make_pass_tracer): ...new function.
10117         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10118         to a subclass of gimple_opt_pass along with...
10119         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10120         (make_pass_diagnose_tm_blocks): ...new function.
10121         (pass_lower_tm): Convert from a global struct to a subclass of
10122         gimple_opt_pass along with...
10123         (pass_data_lower_tm): ...new pass_data instance and...
10124         (make_pass_lower_tm): ...new function.
10125         (pass_tm_init): Convert from a global struct to a subclass of
10126         gimple_opt_pass along with...
10127         (pass_data_tm_init): ...new pass_data instance and...
10128         (make_pass_tm_init): ...new function.
10129         (pass_tm_mark): Convert from a global struct to a subclass of
10130         gimple_opt_pass along with...
10131         (pass_data_tm_mark): ...new pass_data instance and...
10132         (make_pass_tm_mark): ...new function.
10133         (pass_tm_edges): Convert from a global struct to a subclass of
10134         gimple_opt_pass along with...
10135         (pass_data_tm_edges): ...new pass_data instance and...
10136         (make_pass_tm_edges): ...new function.
10137         (pass_tm_memopt): Convert from a global struct to a subclass of
10138         gimple_opt_pass along with...
10139         (pass_data_tm_memopt): ...new pass_data instance and...
10140         (make_pass_tm_memopt): ...new function.
10141         (pass_ipa_tm): Convert from a global struct to a subclass of
10142         simple_ipa_opt_pass along with...
10143         (pass_data_ipa_tm): ...new pass_data instance and...
10144         (make_pass_ipa_tm): ...new function.
10145         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10146         subclass of gimple_opt_pass along with...
10147         (pass_data_call_cdce): ...new pass_data instance and...
10148         (make_pass_call_cdce): ...new function.
10149         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10150         subclass of gimple_opt_pass along with...
10151         (pass_data_build_cfg): ...new pass_data instance and...
10152         (make_pass_build_cfg): ...new function.
10153         (pass_split_crit_edges): Convert from a global struct to a subclass of
10154         gimple_opt_pass along with...
10155         (pass_data_split_crit_edges): ...new pass_data instance and...
10156         (make_pass_split_crit_edges): ...new function.
10157         (pass_warn_function_return): Convert from a global struct to a
10158         subclass of gimple_opt_pass along with...
10159         (pass_data_warn_function_return): ...new pass_data instance and...
10160         (make_pass_warn_function_return): ...new function.
10161         (pass_warn_function_noreturn): Convert from a global struct to a
10162         subclass of gimple_opt_pass along with...
10163         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10164         (make_pass_warn_function_noreturn): ...new function.
10165         (pass_warn_unused_result): Convert from a global struct to a subclass
10166         of gimple_opt_pass along with...
10167         (pass_data_warn_unused_result): ...new pass_data instance and...
10168         (make_pass_warn_unused_result): ...new function.
10169         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10170         a subclass of gimple_opt_pass along with...
10171         (pass_data_merge_phi): ...new pass_data instance and...
10172         (make_pass_merge_phi): ...new function.
10173         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10174         a subclass of gimple_opt_pass along with...
10175         (pass_data_lower_complex): ...new pass_data instance and...
10176         (make_pass_lower_complex): ...new function.
10177         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10178         gimple_opt_pass along with...
10179         (pass_data_lower_complex_O0): ...new pass_data instance and...
10180         (make_pass_lower_complex_O0): ...new function.
10181         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10182         subclass of gimple_opt_pass along with...
10183         (pass_data_lower_eh): ...new pass_data instance and...
10184         (make_pass_lower_eh): ...new function.
10185         (pass_refactor_eh): Convert from a global struct to a subclass of
10186         gimple_opt_pass along with...
10187         (pass_data_refactor_eh): ...new pass_data instance and...
10188         (make_pass_refactor_eh): ...new function.
10189         (pass_lower_resx): Convert from a global struct to a subclass of
10190         gimple_opt_pass along with...
10191         (pass_data_lower_resx): ...new pass_data instance and...
10192         (make_pass_lower_resx): ...new function.
10193         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10194         of gimple_opt_pass along with...
10195         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10196         (make_pass_lower_eh_dispatch): ...new function.
10197         (pass_cleanup_eh): Convert from a global struct to a subclass of
10198         gimple_opt_pass along with...
10199         (pass_data_cleanup_eh): ...new pass_data instance and...
10200         (make_pass_cleanup_eh): ...new function.
10201         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10202         to a subclass of simple_ipa_opt_pass along with...
10203         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10204         (make_pass_ipa_lower_emutls): ...new function.
10205         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10206         a subclass of gimple_opt_pass along with...
10207         (pass_data_if_conversion): ...new pass_data instance and...
10208         (make_pass_if_conversion): ...new function.
10209         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10210         subclass of gimple_opt_pass along with...
10211         (pass_data_build_ssa): ...new pass_data instance and...
10212         (make_pass_build_ssa): ...new function.
10213         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10214         global struct to a subclass of gimple_opt_pass along with...
10215         (pass_data_loop_distribution): ...new pass_data instance and...
10216         (make_pass_loop_distribution): ...new function.
10217         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10218         subclass of gimple_opt_pass along with...
10219         (pass_data_mudflap_1): ...new pass_data instance and...
10220         (make_pass_mudflap_1): ...new function.
10221         (pass_mudflap_2): Convert from a global struct to a subclass of
10222         gimple_opt_pass along with...
10223         (pass_data_mudflap_2): ...new pass_data instance and...
10224         (make_pass_mudflap_2): ...new function.
10225         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10226         subclass of gimple_opt_pass along with...
10227         (pass_data_mudflap_1): ...new pass_data instance and...
10228         (make_pass_mudflap_1): ...new function.
10229         (pass_mudflap_2): Convert from a global struct to a subclass of
10230         gimple_opt_pass along with...
10231         (pass_data_mudflap_2): ...new pass_data instance and...
10232         (make_pass_mudflap_2): ...new function.
10233         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10234         gimple_opt_pass along with...
10235         (pass_data_nrv): ...new pass_data instance and...
10236         (make_pass_nrv): ...new function.
10237         (pass_return_slot): Convert from a global struct to a subclass of
10238         gimple_opt_pass along with...
10239         (pass_data_return_slot): ...new pass_data instance and...
10240         (make_pass_return_slot): ...new function.
10241         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10242         to a subclass of gimple_opt_pass along with...
10243         (pass_data_object_sizes): ...new pass_data instance and...
10244         (make_pass_object_sizes): ...new function.
10245         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10246         global struct to a subclass of gimple_opt_pass along with...
10247         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10248         and...
10249         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10250         (pass_fixup_cfg): Convert from a global struct to a subclass of
10251         gimple_opt_pass along with...
10252         (pass_data_fixup_cfg): ...new pass_data instance and...
10253         (make_pass_fixup_cfg): ...new function.
10254         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10255         (make_pass_mudflap_1): ...new function.
10256         (pass_mudflap_2): Replace declaration with that of...
10257         (make_pass_mudflap_2): ...new function.
10258         (pass_asan): Replace declaration with that of...
10259         (make_pass_asan): ...new function.
10260         (pass_asan_O0): Replace declaration with that of...
10261         (make_pass_asan_O0): ...new function.
10262         (pass_tsan): Replace declaration with that of...
10263         (make_pass_tsan): ...new function.
10264         (pass_tsan_O0): Replace declaration with that of...
10265         (make_pass_tsan_O0): ...new function.
10266         (pass_lower_cf): Replace declaration with that of...
10267         (make_pass_lower_cf): ...new function.
10268         (pass_refactor_eh): Replace declaration with that of...
10269         (make_pass_refactor_eh): ...new function.
10270         (pass_lower_eh): Replace declaration with that of...
10271         (make_pass_lower_eh): ...new function.
10272         (pass_lower_eh_dispatch): Replace declaration with that of...
10273         (make_pass_lower_eh_dispatch): ...new function.
10274         (pass_lower_resx): Replace declaration with that of...
10275         (make_pass_lower_resx): ...new function.
10276         (pass_build_cfg): Replace declaration with that of...
10277         (make_pass_build_cfg): ...new function.
10278         (pass_early_tree_profile): Replace declaration with that of...
10279         (make_pass_early_tree_profile): ...new function.
10280         (pass_cleanup_eh): Replace declaration with that of...
10281         (make_pass_cleanup_eh): ...new function.
10282         (pass_sra): Replace declaration with that of...
10283         (make_pass_sra): ...new function.
10284         (pass_sra_early): Replace declaration with that of...
10285         (make_pass_sra_early): ...new function.
10286         (pass_early_ipa_sra): Replace declaration with that of...
10287         (make_pass_early_ipa_sra): ...new function.
10288         (pass_tail_recursion): Replace declaration with that of...
10289         (make_pass_tail_recursion): ...new function.
10290         (pass_tail_calls): Replace declaration with that of...
10291         (make_pass_tail_calls): ...new function.
10292         (pass_tree_loop): Replace declaration with that of...
10293         (make_pass_tree_loop): ...new function.
10294         (pass_tree_loop_init): Replace declaration with that of...
10295         (make_pass_tree_loop_init): ...new function.
10296         (pass_lim): Replace declaration with that of...
10297         (make_pass_lim): ...new function.
10298         (pass_tree_unswitch): Replace declaration with that of...
10299         (make_pass_tree_unswitch): ...new function.
10300         (pass_predcom): Replace declaration with that of...
10301         (make_pass_predcom): ...new function.
10302         (pass_iv_canon): Replace declaration with that of...
10303         (make_pass_iv_canon): ...new function.
10304         (pass_scev_cprop): Replace declaration with that of...
10305         (make_pass_scev_cprop): ...new function.
10306         (pass_empty_loop): Replace declaration with that of...
10307         (make_pass_empty_loop): ...new function.
10308         (pass_record_bounds): Replace declaration with that of...
10309         (make_pass_record_bounds): ...new function.
10310         (pass_graphite): Replace declaration with that of...
10311         (make_pass_graphite): ...new function.
10312         (pass_graphite_transforms): Replace declaration with that of...
10313         (make_pass_graphite_transforms): ...new function.
10314         (pass_if_conversion): Replace declaration with that of...
10315         (make_pass_if_conversion): ...new function.
10316         (pass_loop_distribution): Replace declaration with that of...
10317         (make_pass_loop_distribution): ...new function.
10318         (pass_vectorize): Replace declaration with that of...
10319         (make_pass_vectorize): ...new function.
10320         (pass_slp_vectorize): Replace declaration with that of...
10321         (make_pass_slp_vectorize): ...new function.
10322         (pass_complete_unroll): Replace declaration with that of...
10323         (make_pass_complete_unroll): ...new function.
10324         (pass_complete_unrolli): Replace declaration with that of...
10325         (make_pass_complete_unrolli): ...new function.
10326         (pass_parallelize_loops): Replace declaration with that of...
10327         (make_pass_parallelize_loops): ...new function.
10328         (pass_loop_prefetch): Replace declaration with that of...
10329         (make_pass_loop_prefetch): ...new function.
10330         (pass_iv_optimize): Replace declaration with that of...
10331         (make_pass_iv_optimize): ...new function.
10332         (pass_tree_loop_done): Replace declaration with that of...
10333         (make_pass_tree_loop_done): ...new function.
10334         (pass_ch): Replace declaration with that of...
10335         (make_pass_ch): ...new function.
10336         (pass_ccp): Replace declaration with that of...
10337         (make_pass_ccp): ...new function.
10338         (pass_phi_only_cprop): Replace declaration with that of...
10339         (make_pass_phi_only_cprop): ...new function.
10340         (pass_build_ssa): Replace declaration with that of...
10341         (make_pass_build_ssa): ...new function.
10342         (pass_build_alias): Replace declaration with that of...
10343         (make_pass_build_alias): ...new function.
10344         (pass_build_ealias): Replace declaration with that of...
10345         (make_pass_build_ealias): ...new function.
10346         (pass_dominator): Replace declaration with that of...
10347         (make_pass_dominator): ...new function.
10348         (pass_dce): Replace declaration with that of...
10349         (make_pass_dce): ...new function.
10350         (pass_dce_loop): Replace declaration with that of...
10351         (make_pass_dce_loop): ...new function.
10352         (pass_cd_dce): Replace declaration with that of...
10353         (make_pass_cd_dce): ...new function.
10354         (pass_call_cdce): Replace declaration with that of...
10355         (make_pass_call_cdce): ...new function.
10356         (pass_merge_phi): Replace declaration with that of...
10357         (make_pass_merge_phi): ...new function.
10358         (pass_split_crit_edges): Replace declaration with that of...
10359         (make_pass_split_crit_edges): ...new function.
10360         (pass_pre): Replace declaration with that of...
10361         (make_pass_pre): ...new function.
10362         (pass_profile): Replace declaration with that of...
10363         (make_pass_profile): ...new function.
10364         (pass_strip_predict_hints): Replace declaration with that of...
10365         (make_pass_strip_predict_hints): ...new function.
10366         (pass_lower_complex_O0): Replace declaration with that of...
10367         (make_pass_lower_complex_O0): ...new function.
10368         (pass_lower_complex): Replace declaration with that of...
10369         (make_pass_lower_complex): ...new function.
10370         (pass_lower_vector): Replace declaration with that of...
10371         (make_pass_lower_vector): ...new function.
10372         (pass_lower_vector_ssa): Replace declaration with that of...
10373         (make_pass_lower_vector_ssa): ...new function.
10374         (pass_lower_omp): Replace declaration with that of...
10375         (make_pass_lower_omp): ...new function.
10376         (pass_diagnose_omp_blocks): Replace declaration with that of...
10377         (make_pass_diagnose_omp_blocks): ...new function.
10378         (pass_expand_omp): Replace declaration with that of...
10379         (make_pass_expand_omp): ...new function.
10380         (pass_expand_omp_ssa): Replace declaration with that of...
10381         (make_pass_expand_omp_ssa): ...new function.
10382         (pass_object_sizes): Replace declaration with that of...
10383         (make_pass_object_sizes): ...new function.
10384         (pass_strlen): Replace declaration with that of...
10385         (make_pass_strlen): ...new function.
10386         (pass_fold_builtins): Replace declaration with that of...
10387         (make_pass_fold_builtins): ...new function.
10388         (pass_stdarg): Replace declaration with that of...
10389         (make_pass_stdarg): ...new function.
10390         (pass_early_warn_uninitialized): Replace declaration with that of...
10391         (make_pass_early_warn_uninitialized): ...new function.
10392         (pass_late_warn_uninitialized): Replace declaration with that of...
10393         (make_pass_late_warn_uninitialized): ...new function.
10394         (pass_cse_reciprocals): Replace declaration with that of...
10395         (make_pass_cse_reciprocals): ...new function.
10396         (pass_cse_sincos): Replace declaration with that of...
10397         (make_pass_cse_sincos): ...new function.
10398         (pass_optimize_bswap): Replace declaration with that of...
10399         (make_pass_optimize_bswap): ...new function.
10400         (pass_optimize_widening_mul): Replace declaration with that of...
10401         (make_pass_optimize_widening_mul): ...new function.
10402         (pass_warn_function_return): Replace declaration with that of...
10403         (make_pass_warn_function_return): ...new function.
10404         (pass_warn_function_noreturn): Replace declaration with that of...
10405         (make_pass_warn_function_noreturn): ...new function.
10406         (pass_cselim): Replace declaration with that of...
10407         (make_pass_cselim): ...new function.
10408         (pass_phiopt): Replace declaration with that of...
10409         (make_pass_phiopt): ...new function.
10410         (pass_forwprop): Replace declaration with that of...
10411         (make_pass_forwprop): ...new function.
10412         (pass_phiprop): Replace declaration with that of...
10413         (make_pass_phiprop): ...new function.
10414         (pass_tree_ifcombine): Replace declaration with that of...
10415         (make_pass_tree_ifcombine): ...new function.
10416         (pass_dse): Replace declaration with that of...
10417         (make_pass_dse): ...new function.
10418         (pass_nrv): Replace declaration with that of...
10419         (make_pass_nrv): ...new function.
10420         (pass_rename_ssa_copies): Replace declaration with that of...
10421         (make_pass_rename_ssa_copies): ...new function.
10422         (pass_sink_code): Replace declaration with that of...
10423         (make_pass_sink_code): ...new function.
10424         (pass_fre): Replace declaration with that of...
10425         (make_pass_fre): ...new function.
10426         (pass_check_data_deps): Replace declaration with that of...
10427         (make_pass_check_data_deps): ...new function.
10428         (pass_copy_prop): Replace declaration with that of...
10429         (make_pass_copy_prop): ...new function.
10430         (pass_vrp): Replace declaration with that of...
10431         (make_pass_vrp): ...new function.
10432         (pass_uncprop): Replace declaration with that of...
10433         (make_pass_uncprop): ...new function.
10434         (pass_return_slot): Replace declaration with that of...
10435         (make_pass_return_slot): ...new function.
10436         (pass_reassoc): Replace declaration with that of...
10437         (make_pass_reassoc): ...new function.
10438         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10439         (make_pass_rebuild_cgraph_edges): ...new function.
10440         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10441         (make_pass_remove_cgraph_callee_edges): ...new function.
10442         (pass_build_cgraph_edges): Replace declaration with that of...
10443         (make_pass_build_cgraph_edges): ...new function.
10444         (pass_local_pure_const): Replace declaration with that of...
10445         (make_pass_local_pure_const): ...new function.
10446         (pass_tracer): Replace declaration with that of...
10447         (make_pass_tracer): ...new function.
10448         (pass_warn_unused_result): Replace declaration with that of...
10449         (make_pass_warn_unused_result): ...new function.
10450         (pass_diagnose_tm_blocks): Replace declaration with that of...
10451         (make_pass_diagnose_tm_blocks): ...new function.
10452         (pass_lower_tm): Replace declaration with that of...
10453         (make_pass_lower_tm): ...new function.
10454         (pass_tm_init): Replace declaration with that of...
10455         (make_pass_tm_init): ...new function.
10456         (pass_tm_mark): Replace declaration with that of...
10457         (make_pass_tm_mark): ...new function.
10458         (pass_tm_memopt): Replace declaration with that of...
10459         (make_pass_tm_memopt): ...new function.
10460         (pass_tm_edges): Replace declaration with that of...
10461         (make_pass_tm_edges): ...new function.
10462         (pass_split_functions): Replace declaration with that of...
10463         (make_pass_split_functions): ...new function.
10464         (pass_feedback_split_functions): Replace declaration with that of...
10465         (make_pass_feedback_split_functions): ...new function.
10466         (pass_strength_reduction): Replace declaration with that of...
10467         (make_pass_strength_reduction): ...new function.
10468         (pass_ipa_lower_emutls): Replace declaration with that of...
10469         (make_pass_ipa_lower_emutls): ...new function.
10470         (pass_ipa_function_and_variable_visibility): Replace declaration with
10471         that of...
10472         (make_pass_ipa_function_and_variable_visibility): ...new function.
10473         (pass_ipa_tree_profile): Replace declaration with that of...
10474         (make_pass_ipa_tree_profile): ...new function.
10475         (pass_early_local_passes): Replace declaration with that of...
10476         (make_pass_early_local_passes): ...new function.
10477         (pass_ipa_whole_program_visibility): Replace declaration with that
10478         of...
10479         (make_pass_ipa_whole_program_visibility): ...new function.
10480         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10481         (make_pass_ipa_lto_gimple_out): ...new function.
10482         (pass_ipa_increase_alignment): Replace declaration with that of...
10483         (make_pass_ipa_increase_alignment): ...new function.
10484         (pass_ipa_inline): Replace declaration with that of...
10485         (make_pass_ipa_inline): ...new function.
10486         (pass_ipa_free_lang_data): Replace declaration with that of...
10487         (make_pass_ipa_free_lang_data): ...new function.
10488         (pass_ipa_free_inline_summary): Replace declaration with that of...
10489         (make_pass_ipa_free_inline_summary): ...new function.
10490         (pass_ipa_cp): Replace declaration with that of...
10491         (make_pass_ipa_cp): ...new function.
10492         (pass_ipa_reference): Replace declaration with that of...
10493         (make_pass_ipa_reference): ...new function.
10494         (pass_ipa_pure_const): Replace declaration with that of...
10495         (make_pass_ipa_pure_const): ...new function.
10496         (pass_ipa_pta): Replace declaration with that of...
10497         (make_pass_ipa_pta): ...new function.
10498         (pass_ipa_lto_finish_out): Replace declaration with that of...
10499         (make_pass_ipa_lto_finish_out): ...new function.
10500         (pass_ipa_tm): Replace declaration with that of...
10501         (make_pass_ipa_tm): ...new function.
10502         (pass_ipa_profile): Replace declaration with that of...
10503         (make_pass_ipa_profile): ...new function.
10504         (pass_ipa_cdtor_merge): Replace declaration with that of...
10505         (make_pass_ipa_cdtor_merge): ...new function.
10506         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10507         of...
10508         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10509         (pass_init_datastructures): Replace declaration with that of...
10510         (make_pass_init_datastructures): ...new function.
10511         (pass_fixup_cfg): Replace declaration with that of...
10512         (make_pass_fixup_cfg): ...new function.
10513         (pass_expand): Replace declaration with that of...
10514         (make_pass_expand): ...new function.
10515         (pass_instantiate_virtual_regs): Replace declaration with that of...
10516         (make_pass_instantiate_virtual_regs): ...new function.
10517         (pass_rtl_fwprop): Replace declaration with that of...
10518         (make_pass_rtl_fwprop): ...new function.
10519         (pass_rtl_fwprop_addr): Replace declaration with that of...
10520         (make_pass_rtl_fwprop_addr): ...new function.
10521         (pass_jump): Replace declaration with that of...
10522         (make_pass_jump): ...new function.
10523         (pass_jump2): Replace declaration with that of...
10524         (make_pass_jump2): ...new function.
10525         (pass_lower_subreg): Replace declaration with that of...
10526         (make_pass_lower_subreg): ...new function.
10527         (pass_cse): Replace declaration with that of...
10528         (make_pass_cse): ...new function.
10529         (pass_fast_rtl_dce): Replace declaration with that of...
10530         (make_pass_fast_rtl_dce): ...new function.
10531         (pass_ud_rtl_dce): Replace declaration with that of...
10532         (make_pass_ud_rtl_dce): ...new function.
10533         (pass_rtl_dce): Replace declaration with that of...
10534         (make_pass_rtl_dce): ...new function.
10535         (pass_rtl_dse1): Replace declaration with that of...
10536         (make_pass_rtl_dse1): ...new function.
10537         (pass_rtl_dse2): Replace declaration with that of...
10538         (make_pass_rtl_dse2): ...new function.
10539         (pass_rtl_dse3): Replace declaration with that of...
10540         (make_pass_rtl_dse3): ...new function.
10541         (pass_rtl_cprop): Replace declaration with that of...
10542         (make_pass_rtl_cprop): ...new function.
10543         (pass_rtl_pre): Replace declaration with that of...
10544         (make_pass_rtl_pre): ...new function.
10545         (pass_rtl_hoist): Replace declaration with that of...
10546         (make_pass_rtl_hoist): ...new function.
10547         (pass_rtl_store_motion): Replace declaration with that of...
10548         (make_pass_rtl_store_motion): ...new function.
10549         (pass_cse_after_global_opts): Replace declaration with that of...
10550         (make_pass_cse_after_global_opts): ...new function.
10551         (pass_rtl_ifcvt): Replace declaration with that of...
10552         (make_pass_rtl_ifcvt): ...new function.
10553         (pass_into_cfg_layout_mode): Replace declaration with that of...
10554         (make_pass_into_cfg_layout_mode): ...new function.
10555         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10556         (make_pass_outof_cfg_layout_mode): ...new function.
10557         (pass_loop2): Replace declaration with that of...
10558         (make_pass_loop2): ...new function.
10559         (pass_rtl_loop_init): Replace declaration with that of...
10560         (make_pass_rtl_loop_init): ...new function.
10561         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10562         (make_pass_rtl_move_loop_invariants): ...new function.
10563         (pass_rtl_unswitch): Replace declaration with that of...
10564         (make_pass_rtl_unswitch): ...new function.
10565         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10566         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10567         (pass_rtl_doloop): Replace declaration with that of...
10568         (make_pass_rtl_doloop): ...new function.
10569         (pass_rtl_loop_done): Replace declaration with that of...
10570         (make_pass_rtl_loop_done): ...new function.
10571         (pass_web): Replace declaration with that of...
10572         (make_pass_web): ...new function.
10573         (pass_cse2): Replace declaration with that of...
10574         (make_pass_cse2): ...new function.
10575         (pass_df_initialize_opt): Replace declaration with that of...
10576         (make_pass_df_initialize_opt): ...new function.
10577         (pass_df_initialize_no_opt): Replace declaration with that of...
10578         (make_pass_df_initialize_no_opt): ...new function.
10579         (pass_reginfo_init): Replace declaration with that of...
10580         (make_pass_reginfo_init): ...new function.
10581         (pass_inc_dec): Replace declaration with that of...
10582         (make_pass_inc_dec): ...new function.
10583         (pass_stack_ptr_mod): Replace declaration with that of...
10584         (make_pass_stack_ptr_mod): ...new function.
10585         (pass_initialize_regs): Replace declaration with that of...
10586         (make_pass_initialize_regs): ...new function.
10587         (pass_combine): Replace declaration with that of...
10588         (make_pass_combine): ...new function.
10589         (pass_if_after_combine): Replace declaration with that of...
10590         (make_pass_if_after_combine): ...new function.
10591         (pass_ree): Replace declaration with that of...
10592         (make_pass_ree): ...new function.
10593         (pass_partition_blocks): Replace declaration with that of...
10594         (make_pass_partition_blocks): ...new function.
10595         (pass_match_asm_constraints): Replace declaration with that of...
10596         (make_pass_match_asm_constraints): ...new function.
10597         (pass_regmove): Replace declaration with that of...
10598         (make_pass_regmove): ...new function.
10599         (pass_split_all_insns): Replace declaration with that of...
10600         (make_pass_split_all_insns): ...new function.
10601         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10602         (make_pass_fast_rtl_byte_dce): ...new function.
10603         (pass_lower_subreg2): Replace declaration with that of...
10604         (make_pass_lower_subreg2): ...new function.
10605         (pass_mode_switching): Replace declaration with that of...
10606         (make_pass_mode_switching): ...new function.
10607         (pass_sms): Replace declaration with that of...
10608         (make_pass_sms): ...new function.
10609         (pass_sched): Replace declaration with that of...
10610         (make_pass_sched): ...new function.
10611         (pass_ira): Replace declaration with that of...
10612         (make_pass_ira): ...new function.
10613         (pass_reload): Replace declaration with that of...
10614         (make_pass_reload): ...new function.
10615         (pass_clean_state): Replace declaration with that of...
10616         (make_pass_clean_state): ...new function.
10617         (pass_branch_prob): Replace declaration with that of...
10618         (make_pass_branch_prob): ...new function.
10619         (pass_value_profile_transformations): Replace declaration with that
10620         of...
10621         (make_pass_value_profile_transformations): ...new function.
10622         (pass_postreload_cse): Replace declaration with that of...
10623         (make_pass_postreload_cse): ...new function.
10624         (pass_gcse2): Replace declaration with that of...
10625         (make_pass_gcse2): ...new function.
10626         (pass_split_after_reload): Replace declaration with that of...
10627         (make_pass_split_after_reload): ...new function.
10628         (pass_branch_target_load_optimize1): Replace declaration with that
10629         of...
10630         (make_pass_branch_target_load_optimize1): ...new function.
10631         (pass_thread_prologue_and_epilogue): Replace declaration with that
10632         of...
10633         (make_pass_thread_prologue_and_epilogue): ...new function.
10634         (pass_stack_adjustments): Replace declaration with that of...
10635         (make_pass_stack_adjustments): ...new function.
10636         (pass_peephole2): Replace declaration with that of...
10637         (make_pass_peephole2): ...new function.
10638         (pass_if_after_reload): Replace declaration with that of...
10639         (make_pass_if_after_reload): ...new function.
10640         (pass_regrename): Replace declaration with that of...
10641         (make_pass_regrename): ...new function.
10642         (pass_cprop_hardreg): Replace declaration with that of...
10643         (make_pass_cprop_hardreg): ...new function.
10644         (pass_reorder_blocks): Replace declaration with that of...
10645         (make_pass_reorder_blocks): ...new function.
10646         (pass_branch_target_load_optimize2): Replace declaration with that
10647         of...
10648         (make_pass_branch_target_load_optimize2): ...new function.
10649         (pass_leaf_regs): Replace declaration with that of...
10650         (make_pass_leaf_regs): ...new function.
10651         (pass_split_before_sched2): Replace declaration with that of...
10652         (make_pass_split_before_sched2): ...new function.
10653         (pass_compare_elim_after_reload): Replace declaration with that of...
10654         (make_pass_compare_elim_after_reload): ...new function.
10655         (pass_sched2): Replace declaration with that of...
10656         (make_pass_sched2): ...new function.
10657         (pass_stack_regs): Replace declaration with that of...
10658         (make_pass_stack_regs): ...new function.
10659         (pass_stack_regs_run): Replace declaration with that of...
10660         (make_pass_stack_regs_run): ...new function.
10661         (pass_df_finish): Replace declaration with that of...
10662         (make_pass_df_finish): ...new function.
10663         (pass_compute_alignments): Replace declaration with that of...
10664         (make_pass_compute_alignments): ...new function.
10665         (pass_duplicate_computed_gotos): Replace declaration with that of...
10666         (make_pass_duplicate_computed_gotos): ...new function.
10667         (pass_variable_tracking): Replace declaration with that of...
10668         (make_pass_variable_tracking): ...new function.
10669         (pass_free_cfg): Replace declaration with that of...
10670         (make_pass_free_cfg): ...new function.
10671         (pass_machine_reorg): Replace declaration with that of...
10672         (make_pass_machine_reorg): ...new function.
10673         (pass_cleanup_barriers): Replace declaration with that of...
10674         (make_pass_cleanup_barriers): ...new function.
10675         (pass_delay_slots): Replace declaration with that of...
10676         (make_pass_delay_slots): ...new function.
10677         (pass_split_for_shorten_branches): Replace declaration with that of...
10678         (make_pass_split_for_shorten_branches): ...new function.
10679         (pass_split_before_regstack): Replace declaration with that of...
10680         (make_pass_split_before_regstack): ...new function.
10681         (pass_convert_to_eh_region_ranges): Replace declaration with that
10682         of...
10683         (make_pass_convert_to_eh_region_ranges): ...new function.
10684         (pass_shorten_branches): Replace declaration with that of...
10685         (make_pass_shorten_branches): ...new function.
10686         (pass_set_nothrow_function_flags): Replace declaration with that of...
10687         (make_pass_set_nothrow_function_flags): ...new function.
10688         (pass_dwarf2_frame): Replace declaration with that of...
10689         (make_pass_dwarf2_frame): ...new function.
10690         (pass_final): Replace declaration with that of...
10691         (make_pass_final): ...new function.
10692         (pass_rtl_seqabstr): Replace declaration with that of...
10693         (make_pass_rtl_seqabstr): ...new function.
10694         (pass_release_ssa_names): Replace declaration with that of...
10695         (make_pass_release_ssa_names): ...new function.
10696         (pass_early_inline): Replace declaration with that of...
10697         (make_pass_early_inline): ...new function.
10698         (pass_inline_parameters): Replace declaration with that of...
10699         (make_pass_inline_parameters): ...new function.
10700         (pass_update_address_taken): Replace declaration with that of...
10701         (make_pass_update_address_taken): ...new function.
10702         (pass_convert_switch): Replace declaration with that of...
10703         (make_pass_convert_switch): ...new function.
10704         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10705         to a subclass of simple_ipa_opt_pass along with...
10706         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10707         (make_pass_ipa_tree_profile): ...new function.
10708         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10709         subclass of gimple_opt_pass along with...
10710         (pass_data_sra_early): ...new pass_data instance and...
10711         (make_pass_sra_early): ...new function.
10712         (pass_sra): Convert from a global struct to a subclass of
10713         gimple_opt_pass along with...
10714         (pass_data_sra): ...new pass_data instance and...
10715         (make_pass_sra): ...new function.
10716         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10717         gimple_opt_pass along with...
10718         (pass_data_early_ipa_sra): ...new pass_data instance and...
10719         (make_pass_early_ipa_sra): ...new function.
10720         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10721         subclass of gimple_opt_pass along with...
10722         (pass_data_ccp): ...new pass_data instance and...
10723         (make_pass_ccp): ...new function.
10724         (pass_fold_builtins): Convert from a global struct to a subclass of
10725         gimple_opt_pass along with...
10726         (pass_data_fold_builtins): ...new pass_data instance and...
10727         (make_pass_fold_builtins): ...new function.
10728         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10729         subclass of gimple_opt_pass along with...
10730         (pass_data_copy_prop): ...new pass_data instance and...
10731         (make_pass_copy_prop): ...new function.
10732         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10733         global struct to a subclass of gimple_opt_pass along with...
10734         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10735         (make_pass_rename_ssa_copies): ...new function.
10736         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10737         subclass of gimple_opt_pass along with...
10738         (pass_data_dce): ...new pass_data instance and...
10739         (make_pass_dce): ...new function.
10740         (pass_dce_loop): Convert from a global struct to a subclass of
10741         gimple_opt_pass along with...
10742         (pass_data_dce_loop): ...new pass_data instance and...
10743         (make_pass_dce_loop): ...new function.
10744         (pass_cd_dce): Convert from a global struct to a subclass of
10745         gimple_opt_pass along with...
10746         (pass_data_cd_dce): ...new pass_data instance and...
10747         (make_pass_cd_dce): ...new function.
10748         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10749         subclass of gimple_opt_pass along with...
10750         (pass_data_dominator): ...new pass_data instance and...
10751         (make_pass_dominator): ...new function.
10752         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10753         gimple_opt_pass along with...
10754         (pass_data_phi_only_cprop): ...new pass_data instance and...
10755         (make_pass_phi_only_cprop): ...new function.
10756         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10757         subclass of gimple_opt_pass along with...
10758         (pass_data_dse): ...new pass_data instance and...
10759         (make_pass_dse): ...new function.
10760         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10761         a subclass of gimple_opt_pass along with...
10762         (pass_data_forwprop): ...new pass_data instance and...
10763         (make_pass_forwprop): ...new function.
10764         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10765         struct to a subclass of gimple_opt_pass along with...
10766         (pass_data_tree_ifcombine): ...new pass_data instance and...
10767         (make_pass_tree_ifcombine): ...new function.
10768         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10769         subclass of gimple_opt_pass along with...
10770         (pass_data_ch): ...new pass_data instance and...
10771         (make_pass_ch): ...new function.
10772         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10773         subclass of gimple_opt_pass along with...
10774         (pass_data_tree_loop): ...new pass_data instance and...
10775         (make_pass_tree_loop): ...new function.
10776         (pass_tree_loop_init): Convert from a global struct to a subclass of
10777         gimple_opt_pass along with...
10778         (pass_data_tree_loop_init): ...new pass_data instance and...
10779         (make_pass_tree_loop_init): ...new function.
10780         (pass_lim): Convert from a global struct to a subclass of
10781         gimple_opt_pass along with...
10782         (pass_data_lim): ...new pass_data instance and...
10783         (make_pass_lim): ...new function.
10784         (pass_tree_unswitch): Convert from a global struct to a subclass of
10785         gimple_opt_pass along with...
10786         (pass_data_tree_unswitch): ...new pass_data instance and...
10787         (make_pass_tree_unswitch): ...new function.
10788         (pass_predcom): Convert from a global struct to a subclass of
10789         gimple_opt_pass along with...
10790         (pass_data_predcom): ...new pass_data instance and...
10791         (make_pass_predcom): ...new function.
10792         (pass_vectorize): Convert from a global struct to a subclass of
10793         gimple_opt_pass along with...
10794         (pass_data_vectorize): ...new pass_data instance and...
10795         (make_pass_vectorize): ...new function.
10796         (pass_graphite): Convert from a global struct to a subclass of
10797         gimple_opt_pass along with...
10798         (pass_data_graphite): ...new pass_data instance and...
10799         (make_pass_graphite): ...new function.
10800         (pass_graphite_transforms): Convert from a global struct to a subclass
10801         of gimple_opt_pass along with...
10802         (pass_data_graphite_transforms): ...new pass_data instance and...
10803         (make_pass_graphite_transforms): ...new function.
10804         (pass_check_data_deps): Convert from a global struct to a subclass of
10805         gimple_opt_pass along with...
10806         (pass_data_check_data_deps): ...new pass_data instance and...
10807         (make_pass_check_data_deps): ...new function.
10808         (pass_iv_canon): Convert from a global struct to a subclass of
10809         gimple_opt_pass along with...
10810         (pass_data_iv_canon): ...new pass_data instance and...
10811         (make_pass_iv_canon): ...new function.
10812         (pass_scev_cprop): Convert from a global struct to a subclass of
10813         gimple_opt_pass along with...
10814         (pass_data_scev_cprop): ...new pass_data instance and...
10815         (make_pass_scev_cprop): ...new function.
10816         (pass_record_bounds): Convert from a global struct to a subclass of
10817         gimple_opt_pass along with...
10818         (pass_data_record_bounds): ...new pass_data instance and...
10819         (make_pass_record_bounds): ...new function.
10820         (pass_complete_unroll): Convert from a global struct to a subclass of
10821         gimple_opt_pass along with...
10822         (pass_data_complete_unroll): ...new pass_data instance and...
10823         (make_pass_complete_unroll): ...new function.
10824         (pass_complete_unrolli): Convert from a global struct to a subclass of
10825         gimple_opt_pass along with...
10826         (pass_data_complete_unrolli): ...new pass_data instance and...
10827         (make_pass_complete_unrolli): ...new function.
10828         (pass_parallelize_loops): Convert from a global struct to a subclass
10829         of gimple_opt_pass along with...
10830         (pass_data_parallelize_loops): ...new pass_data instance and...
10831         (make_pass_parallelize_loops): ...new function.
10832         (pass_loop_prefetch): Convert from a global struct to a subclass of
10833         gimple_opt_pass along with...
10834         (pass_data_loop_prefetch): ...new pass_data instance and...
10835         (make_pass_loop_prefetch): ...new function.
10836         (pass_iv_optimize): Convert from a global struct to a subclass of
10837         gimple_opt_pass along with...
10838         (pass_data_iv_optimize): ...new pass_data instance and...
10839         (make_pass_iv_optimize): ...new function.
10840         (pass_tree_loop_done): Convert from a global struct to a subclass of
10841         gimple_opt_pass along with...
10842         (pass_data_tree_loop_done): ...new pass_data instance and...
10843         (make_pass_tree_loop_done): ...new function.
10844         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10845         struct to a subclass of gimple_opt_pass along with...
10846         (pass_data_cse_reciprocals): ...new pass_data instance and...
10847         (make_pass_cse_reciprocals): ...new function.
10848         (pass_cse_sincos): Convert from a global struct to a subclass of
10849         gimple_opt_pass along with...
10850         (pass_data_cse_sincos): ...new pass_data instance and...
10851         (make_pass_cse_sincos): ...new function.
10852         (pass_optimize_bswap): Convert from a global struct to a subclass of
10853         gimple_opt_pass along with...
10854         (pass_data_optimize_bswap): ...new pass_data instance and...
10855         (make_pass_optimize_bswap): ...new function.
10856         (pass_optimize_widening_mul): Convert from a global struct to a
10857         subclass of gimple_opt_pass along with...
10858         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10859         (make_pass_optimize_widening_mul): ...new function.
10860         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10861         subclass of gimple_opt_pass along with...
10862         (pass_data_phiopt): ...new pass_data instance and...
10863         (make_pass_phiopt): ...new function.
10864         (pass_cselim): Convert from a global struct to a subclass of
10865         gimple_opt_pass along with...
10866         (pass_data_cselim): ...new pass_data instance and...
10867         (make_pass_cselim): ...new function.
10868         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10869         subclass of gimple_opt_pass along with...
10870         (pass_data_phiprop): ...new pass_data instance and...
10871         (make_pass_phiprop): ...new function.
10872         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10873         subclass of gimple_opt_pass along with...
10874         (pass_data_pre): ...new pass_data instance and...
10875         (make_pass_pre): ...new function.
10876         (pass_fre): Convert from a global struct to a subclass of
10877         gimple_opt_pass along with...
10878         (pass_data_fre): ...new pass_data instance and...
10879         (make_pass_fre): ...new function.
10880         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10881         subclass of gimple_opt_pass along with...
10882         (pass_data_reassoc): ...new pass_data instance and...
10883         (make_pass_reassoc): ...new function.
10884         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10885         subclass of gimple_opt_pass along with...
10886         (pass_data_sink_code): ...new pass_data instance and...
10887         (make_pass_sink_code): ...new function.
10888         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10889         subclass of gimple_opt_pass along with...
10890         (pass_data_strlen): ...new pass_data instance and...
10891         (make_pass_strlen): ...new function.
10892         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10893         struct to a subclass of gimple_opt_pass along with...
10894         (pass_data_build_alias): ...new pass_data instance and...
10895         (make_pass_build_alias): ...new function.
10896         (pass_build_ealias): Convert from a global struct to a subclass of
10897         gimple_opt_pass along with...
10898         (pass_data_build_ealias): ...new pass_data instance and...
10899         (make_pass_build_ealias): ...new function.
10900         (pass_ipa_pta): Convert from a global struct to a subclass of
10901         simple_ipa_opt_pass along with...
10902         (pass_data_ipa_pta): ...new pass_data instance and...
10903         (make_pass_ipa_pta): ...new function.
10904         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10905         subclass of gimple_opt_pass along with...
10906         (pass_data_uncprop): ...new pass_data instance and...
10907         (make_pass_uncprop): ...new function.
10908         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10909         global struct to a subclass of gimple_opt_pass along with...
10910         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10911         (make_pass_late_warn_uninitialized): ...new function.
10912         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10913         to a subclass of gimple_opt_pass along with...
10914         (pass_data_init_datastructures): ...new pass_data instance and...
10915         (make_pass_init_datastructures): ...new function.
10916         (pass_early_warn_uninitialized): Convert from a global struct to a
10917         subclass of gimple_opt_pass along with...
10918         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10919         (make_pass_early_warn_uninitialized): ...new function.
10920         (pass_update_address_taken): Convert from a global struct to a
10921         subclass of gimple_opt_pass along with...
10922         (pass_data_update_address_taken): ...new pass_data instance and...
10923         (make_pass_update_address_taken): ...new function.
10924         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10925         struct to a subclass of gimple_opt_pass along with...
10926         (pass_data_release_ssa_names): ...new pass_data instance and...
10927         (make_pass_release_ssa_names): ...new function.
10928         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10929         subclass of gimple_opt_pass along with...
10930         (pass_data_stdarg): ...new pass_data instance and...
10931         (make_pass_stdarg): ...new function.
10932         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10933         global struct to a subclass of gimple_opt_pass along with...
10934         (pass_data_convert_switch): ...new pass_data instance and...
10935         (make_pass_convert_switch): ...new function.
10936         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10937         to a subclass of gimple_opt_pass along with...
10938         (pass_data_tail_recursion): ...new pass_data instance and...
10939         (make_pass_tail_recursion): ...new function.
10940         (pass_tail_calls): Convert from a global struct to a subclass of
10941         gimple_opt_pass along with...
10942         (pass_data_tail_calls): ...new pass_data instance and...
10943         (make_pass_tail_calls): ...new function.
10944         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10945         struct to a subclass of gimple_opt_pass along with...
10946         (pass_data_lower_vector): ...new pass_data instance and...
10947         (make_pass_lower_vector): ...new function.
10948         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
10949         gimple_opt_pass along with...
10950         (pass_data_lower_vector_ssa): ...new pass_data instance and...
10951         (make_pass_lower_vector_ssa): ...new function.
10952         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
10953         to a subclass of gimple_opt_pass along with...
10954         (pass_data_slp_vectorize): ...new pass_data instance and...
10955         (make_pass_slp_vectorize): ...new function.
10956         (pass_ipa_increase_alignment): Convert from a global struct to a
10957         subclass of simple_ipa_opt_pass along with...
10958         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
10959         (make_pass_ipa_increase_alignment): ...new function.
10960         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
10961         gimple_opt_pass along with...
10962         (pass_data_vrp): ...new pass_data instance and...
10963         (make_pass_vrp): ...new function.
10964         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
10965         subclass of simple_ipa_opt_pass along with...
10966         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
10967         (make_pass_ipa_free_lang_data): ...new function.
10968         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
10969         gimple_opt_pass along with...
10970         (pass_data_tsan): ...new pass_data instance and...
10971         (make_pass_tsan): ...new function.
10972         (pass_tsan_O0): Convert from a global struct to a subclass of
10973         gimple_opt_pass along with...
10974         (pass_data_tsan_O0): ...new pass_data instance and...
10975         (make_pass_tsan_O0): ...new function.
10976         * var-tracking.c (pass_variable_tracking): Convert from a global
10977         struct to a subclass of rtl_opt_pass along with...
10978         (pass_data_variable_tracking): ...new pass_data instance and...
10979         (make_pass_variable_tracking): ...new function.
10980         * web.c (pass_web): Convert from a global struct to a subclass of
10981         rtl_opt_pass along with...
10982         (pass_data_web): ...new pass_data instance and...
10983         (make_pass_web): ...new function.
10984         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
10985         declaration with that of...
10986         (make_pass_mode_switch_use): ...new function.
10987         (pass_resolve_sw_modes): Replace declaration with that of...
10988         (make_pass_resolve_sw_modes): ...new function.
10989         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
10990         from a global struct to a subclass of rtl_opt_pass along with...
10991         (pass_data_mode_switch_use): ...new pass_data instance and...
10992         (make_pass_mode_switch_use): ...new function.
10993         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
10994         from a global struct to a subclass of rtl_opt_pass along with...
10995         (pass_data_resolve_sw_modes): ...new pass_data instance and...
10996         (make_pass_resolve_sw_modes): ...new function.
10997         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
10998         struct to a subclass of rtl_opt_pass along with...
10999         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11000         (make_pass_insert_vzeroupper): ...new function.
11001         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11002         global struct to a subclass of rtl_opt_pass along with...
11003         (pass_data_work_around_errata): ...new pass_data instance and...
11004         (make_pass_work_around_errata): ...new function.
11005         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11006         struct to a subclass of rtl_opt_pass along with...
11007         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11008         (make_pass_mips_machine_reorg2): ...new function.
11010 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11012         * passes.c (pass_manager::operator new): New.
11014 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11016         Handwritten part of conversion of passes to C++ classes.
11018         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11019         (toplev.o): Add dep on PASS_MANAGER_H.
11020         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11021         of early local pases to reflect this moving from a global to a
11022         member of gcc::pass_manager.
11023         (cgraph_add_new_function): Likewise.
11024         * lto-cgraph.c (lto_output_node): Update for conversion of
11025         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11026         * passes.c (opt_pass::clone): New.
11027         (opt_pass::gate): New.
11028         (opt_pass::execute): New.
11029         (opt_pass::opt_pass): New.
11030         (pass_manager::execute_early_local_passes): New.
11031         (pass_manager::execute_pass_mode_switching): new.
11032         (finish_optimization_passes): Convert to...
11033         (pass_manager::finish_optimization_passes): ...this.
11034         (finish_optimization_passes): Update for conversion of passes to
11035         C++ classes.
11036         (register_dump_files_1): Use has_gate since we cannot portably
11037         check a vtable entry against NULL.
11038         (dump_one_pass): Likewise.
11039         (ipa_write_summaries_2): Likewise.
11040         (ipa_write_optimization_summaries_1): Likewise.
11041         (ipa_read_summaries_1): Likewise.
11042         (ipa_read_optimization_summaries_1): Likewise.
11043         (execute_ipa_stmt_fixups): Likewise.
11044         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11045         pass-creation functions rather than wiring up globals, and
11046         storing the results in fields of pass_manager generated using
11047         pass-instances.def.
11048         (pass_manager::dump_profile_report): Update for conversion of
11049         passes to C++ classes.
11050         (pass_manager::execute_ipa_summary_passes): Likewise.
11051         (execute_one_ipa_transform_pass): Likewise.
11052         (execute_one_pass): Use has_gate and has_execute since we cannot
11053         portably check a vtable entry against NULL.
11054         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11055         (pass_manager): Use pass-instances.def to add fields for the
11056         various pass instances.
11057         * toplev.c (finalize): Update for move of
11058         finish_optimization_passes to a method of gcc::pass_manager.
11059         * toplev.h (finish_optimization_passes): Move to method of class
11060         pass_manager.
11061         * tree-pass.h (struct pass_data): New.
11062         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11063         (opt_pass::gate): Convert to virtual function.
11064         (opt_pass::~opt_pass): New.
11065         (opt_pass::clone): New.
11066         (opt_pass::execute): Convert to virtual function.
11067         (opt_pass::opt_pass): New.
11068         (opt_pass::ctxt_): new.
11069         (gimple_opt_pass): Convert to subclass of opt_pass.
11070         (gimple_opt_pass::gimple_opt_pass): New.
11071         (rtl_opt_pass): Convert to subclass of opt_pass.
11072         (rtl_opt_pass::rtl_opt_pass): New.
11073         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11074         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11075         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11076         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11077         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11078         invocation of pass_mode_switching to reflect this moving from a
11079         global to a member of gcc::pass_manager.
11080         (ix86_option_override): Rework how pass_insert_vzeroupper is
11081         added to the pass_manager to reflect autogenerated changes.
11082         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11084 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11086         PR rtl-optimization/57708
11087         * recog.c (peep2_find_free_register): Validate all regs in a
11088         multi-reg mode.
11090 2013-08-05  Jan Hubicka  <jh@suse.cz>
11092         PR lto/57602
11093         * cgraph.c (verify_cgraph_node): Accept local flags from other
11094         partitions.
11095         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11096         (function_and_variable_visibility): Likewise.
11097         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11099 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11101         * graph.c (init_graph_slim_pretty_print): Remove.
11102         (print_graph_cfg): Do not call it.  Use local pretty printer.
11103         (start_graph_dump): Likewise.
11105 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11107         * gimple-pretty-print.c (buffer): Remove.
11108         (initialized): Likewise.
11109         (maybe_init_pretty_print): Likewise.
11110         (print_gimple_stmt): Do not call it.  Use non-static local
11111         pretty_printer variable.
11112         (print_gimple_expr): Likewise.
11113         (print_gimple_seq): Likewise.
11114         (gimple_dump_bb): Likewise.
11116 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11118         * asan.c (asan_pp): Remove.
11119         (asan_pp_initialized): Likewise.
11120         (asan_pp_initialize): Likewise.
11121         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11122         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11123         (asan_add_global): Likewise.
11125 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11127         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11128         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11129         * pretty-print.c (pp_formatted_text_data): Likewise.
11130         (pp_indent): Rename from pp_base_indent.
11131         (pp_format): Rename from pp_base_format.
11132         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11133         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11134         (pp_flush): Rename from pp_base_flush.
11135         (pp_set_line_maximum_length): Rename from
11136         pp_base_set_line_maximum_length.
11137         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11138         (pp_set_prefix): Rename from pp_base_set_prefix.
11139         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11140         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11141         (pp_append_text): Rename from pp_base_append_text.
11142         (pp_formatted_text): Rename from pp_base_formatted_text.
11143         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11144         (pp_remaining_character_count_for_line): Rename from
11145         pp_base_remaining_character_count_for_line.
11146         (pp_newline): Rename from pp_base_newline.
11147         (pp_character): Rename from pp_base_character.
11148         (pp_string): Rename from pp_base_string.
11149         (pp_maybe_space): Rename from pp_base_maybe_space.
11150         * asan.c (asan_pp_string): Adjust.
11151         (asan_emit_stack_protection): Likewise.
11152         (asan_add_global): Likewise.
11153         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11154         * tree-mudflap.c (mf_varname_tree): Likewise.
11155         * tree-pretty-print.c (pp_tree_identifier): Rename from
11156         pp_base_tree_identifier.
11157         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11158         Declare as function.
11160 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11162         * pretty-print.h (pp_bar_bar): New.
11163         (pp_ampersand_ampersand): Likewise.
11164         (pp_less_equal): Likewise.
11165         (pp_greater_equal): Likewise.
11166         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11167         printer functions instead of pp_string or operators and punctuators.
11168         (dump_gimple_call): Likewise.
11169         (dump_gimple_omp_for): Likewise.
11170         (dump_gimple_transaction): Likewise.
11171         (dump_gimple_phi): Likewise.
11172         (pp_gimple_stmt_1): Likewise.
11173         * sched-vis.c (print_insn): Likewise.
11174         * tree-mudflap.c (mf_varname_tree): Likewise.
11175         * tree-pretty-print.c (dump_block_node): Likewise.
11176         (dump_generic_node): Likewise.
11178 2013-08-02  Jan Hubicka  <jh@suse.cz>
11180         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11181         boundaries.
11182         * lto-streamer-out.c (tree_is_indexable): Results decls and
11183         parm decls are not indexable.
11184         (DFS_write_tree_body): Do not follow args and results.
11185         (hash_tree): Likewise.
11186         (output_functions): Rearrange so struct function is needed
11187         only when real body is output; be able to also ouptut abstract
11188         functions; output DECL_ARGUMENTS and DECL_RESULT.
11189         (lto_output): When not in WPA, ale store abstract functions.
11190         (write_symbol): Do not care about RESULT_DECL.
11191         (output_symbol_p): Handle correctly sbtract decls.
11192         * lto-streamer-in.c (input_function): Rearrange so struct
11193         function can be NULL at entry; allow streaming of
11194         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11195         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11196         sanity check during LTO.
11197         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11198         RESULT_DECl and DECL_ARGUMENTS.
11199         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11200         Likewise.
11202 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11204         * pretty-print.h (pp_underscore): New.
11205         (pp_comma): Tidy.
11206         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11207         printer functions instead of pp_character.
11208         (dump_binary_rhs): Likewise.
11209         (dump_ternary_rhs): Likewise.
11210         (dump_gimple_call_args): Likewise.
11211         (pp_points_to_solution): Likewise.
11212         (dump_gimple_call): Likewise.
11213         (dump_gimple_switch): Likewise.
11214         (dump_gimple_cond): Likewise.
11215         (dump_gimple_bind): Likewise.
11216         (dump_gimple_try): Likewise.
11217         (dump_gimple_omp_for): Likewise.
11218         (dump_gimple_omp_continue): Likewise.
11219         (dump_gimple_omp_single): Likewise.
11220         (dump_gimple_omp_sections): Likewise.
11221         (dump_gimple_omp_block): Likewise.
11222         (dump_gimple_omp_critical): Likewise.
11223         (dump_gimple_transaction): Likewise.
11224         (dump_gimple_asm): Likewise.
11225         (dump_gimple_phi): Likewise.
11226         (dump_gimple_omp_parallel): Likewise.
11227         (dump_gimple_omp_task): Likewise.
11228         (dump_gimple_omp_atomic_load): Likewise.
11229         (dump_gimple_omp_atomic_store): Likewise.
11230         (dump_gimple_mem_ops): Likewise.
11231         (pp_gimple_stmt_1): Likewise.
11232         (pp_cfg_jump): Likewise.
11233         (dump_implicit_edges): Likewise.
11234         (gimple_dump_bb_for_graph): Likewise.
11235         * graph.c (draw_cfg_node): Likewise.
11236         * langhooks.c (lhd_print_error_function): Likewise.
11237         * sched-vis.c (print_exp): Likewise.
11238         (print_value): Likewise.
11239         (print_pattern): Likewise.
11240         (print_insn): Likewise.
11241         (rtl_dump_bb_for_graph): Likewise.
11242         * tree-pretty-print.c (dump_function_declaration): Likewise.
11243         (dump_array_domain): Likewise.
11244         (dump_omp_clause): Likewise.
11245         (dump_location): Likewise.
11246         (dump_generic_node): Likewise.
11247         (print_struct_decl): Likewise.
11248         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11250 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11252         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11253         candidate table when replacing a candidate statement.
11254         (replace_rhs_if_not_dup): Likewise.
11255         (replace_one_candidate): Likewise.
11257 2013-08-02  Jan Hubicka  <jh@suse.cz>
11258             Martin Liska  <marxin.liska@gmail.com>
11260         * cgraphunit.c (add_new_function): Fix logic when adding from
11261         late IPA pass.
11262         (assemble_thunk): Rename to ...
11263         (expand_thunk); .. this one; export; get it working with
11264         general functions; make produced gimple valid.
11265         * cgraph.h (expand_thunk): Declare.
11267 2013-08-02  Jan Hubicka  <jh@suse.cz>
11269         * ipa-cp.c (gather_context_independent_values): Use
11270         ipa_get_param_move_cost.
11271         (get_replacement_map): Remove PARAM; move parameter folding
11272         into tree-inline.c
11273         (create_specialized_node): Update.
11274         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11275         assert that we have gimple body; update move_cost.
11276         (count_formal_params): Assert that we have gimple body.
11277         (ipa_dump_param): New function.
11278         (ipa_alloc_node_params): Break out from ...
11279         (ipa_initialize_node_params): ... here.
11280         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11281         (ipa_write_node_info): Stream move costs.
11282         (ipa_read_node_info): Read move costs.
11283         (ipa_update_after_lto_read): Do not recompute node params.
11284         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11285         (ipa_get_param): Check we are not in WPA.
11286         (ipa_get_param_move_cost): New.
11287         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11288         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11289         parm numbers to be present.
11291 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11293         PR rtl-optimization/58048
11294         * lra-constraints.c (process_alt_operands): Don't check asm
11295         operand on register.
11297 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11299         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11300         the implied StoreLoad barrier for atomic operations if before.
11302 2013-08-02  Jan Hubicka  <jh@suse.cz>
11303             Martin Liska  <marxin.liska@gmail.com>
11305         * cgraph.c (cgraph_function_body_availability): Do not check
11306         cgraph flags.
11307         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11308         symtab_node_availability): Declare.
11309         * ipa.c (can_replace_by_local_alias): New.
11310         (function_and_variable_visibility): Use it.
11311         * symtab.c (symtab_for_node_and_aliases,
11312         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11314 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11316         PR rtl-optimization/57963
11317         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11318         (lra_constraints): Use them.
11320 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11322         * config/arm/types.md (define_attr "type"): Add "load_acq"
11323         and "store_rel".
11324         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11325         changes.
11326         (cortex_a53_store1): Likewise.
11328 2013-08-01  Jan Hubicka  <jh@suse.cz>
11330         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11331         partitions are not needed.
11333 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11335         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11336         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11337         MAYBE_NON_Q_CLASS_P where appropriate.
11339 2013-08-01  Jan Hubicka  <jh@suse.cz>
11341         * cgraph.h (release_function_body): Declare.
11342         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11343         of unused delcarations.
11345 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11347         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11348         RTL form when subtracting a constant.
11350 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11352         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11353         Generate canonical plus rtx with negated immediate instead of minus
11354         where appropriate.
11355         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11357 2013-08-01  Jan Hubicka  <jh@suse.cz>
11359         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11360         (cgraph_release_function_body): Likewise.
11361         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11362         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11363         to used_as_abstract_origin.
11364         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11365         symbols used as abstract origins.
11366         * cgraphunit.c (analyze_functions): Update.
11367         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11368         used_as_abstract_origin.
11369         * tree-inline.c (tree_function_versioning): Update
11370         used_as_abstract_origin; be ready for DECL_RESULT and
11371         DECL_ARGUMENTS to be NULL.
11373         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11374         for abstract functions.
11375         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11376         real symbols.
11378 2013-08-01  Jan Hubicka  <jh@suse.cz>
11380         * profile.c (compute_value_histograms): Fix thinko.
11382 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11384         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11385         aarch-common-protos.h to extra_headers.
11386         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11387         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11388         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11390 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11392         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11393         Include "../arm/types.md".  Define "type" attribute for all patterns.
11394         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11395         attribute changes.
11397 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11399         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11400         to support power8 load fusion.
11401         (fusion_gpr_mem_load): Likewise.
11403         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11405         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11406         declarations for power8 load fusion.
11407         (emit_fusion_gpr_load): Likewise.
11409         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11410         tuning for power8, turn on fusion mode by default.  Turn on sign
11411         extending fusion mode if normal fusion mode is on, and we are at
11412         -O2 or -O3.
11413         (fusion_gpr_load_p): New function, return true if we can fuse an
11414         addis instruction with a dependent load to a GPR.
11415         (emit_fusion_gpr_load): Emit the instructions for power8 load
11416         fusion to GPRs.
11418         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11419         (VSX load fusion peepholes): New peepholes to fuse together an
11420         addi instruction with a VSX load instruction.
11422         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11423         peepholes to fuse an addis instruction with a load to a GPR base
11424         register.  If we are supporting sign extending fusions, convert
11425         sign extending loads to zero extending loads and add an explicit
11426         sign extension.
11428 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11430         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11431         aarch-common-protos.h to extra_headers.
11432         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11433         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11434         (arm_early_store_addr_dep): Likewise.
11435         (arm_no_early_alu_shift_dep): Likewise.
11436         (arm_no_early_alu_shift_value_dep): Likewise.
11437         (arm_no_early_mul_dep): Likewise.
11438         (arm_no_early_store_addr_dep): Likewise.
11439         (arm_mac_accumulator_is_mul_result): Likewise.
11440         (arm_mac_accumulator_is_result): Likewise.
11441         * config/arm/aarch-common.c: ... here.  New file.
11442         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11443         here to ...
11444         (arm_early_store_addr_dep): Likewise.
11445         (arm_no_early_alu_shift_dep): Likewise.
11446         (arm_no_early_alu_shift_value_dep): Likewise.
11447         (arm_no_early_mul_dep): Likewise.
11448         (arm_no_early_store_addr_dep): Likewise.
11449         (arm_mac_accumulator_is_mul_result): Likewise.
11450         (arm_mac_accumulator_is_result): Likewise.
11451         * config/arm/aarch-common-protos.h: ... here.  New file.
11452         * config/arm/t-arm (aarch-common.o): Define.
11454 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11456         * config/arm/arm.md: Include new file "types.md".
11457         (define_attr "type"): Move from here to ...
11458         (define_attr "mul32"): Likewise.
11459         (define_attr "mul64"): Likewise.
11460         * config/arm/types.md: ... here.  New file.
11462 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11464         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11465         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11467 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11469         * gen-pass-instances.awk: Fix offset of substr().
11471 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11473         * Makefile.in (pass-instances.def): New.
11474         (passes.o): Replace dependency on passes.def with one on
11475         pass-instances.def
11477         * gen-pass-instances.awk: New.
11479         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11480         rather than passes.def, updating local definition of NEXT_PASS
11481         macro to add an extra NUM parameter (currently unused).
11483 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11485         * Makefile.in (PASS_MANAGER_H): New.
11486         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11487         (passes.o): Likewise.
11488         (statistics.o): Likewise.
11489         (cgraphunit.o): Likewise.
11490         (context.o): Depend on PASS_MANAGER_H.
11492         * pass_manager.h: New.
11494         * cgraphunit.c (cgraph_add_new_function): Update for moves
11495         of globals to fields of pass_manager.
11496         (analyze_function): Likewise.
11497         (expand_function): Likewise.
11498         (ipa_passes): Likewise.
11499         (compile): Likewise.
11501         * context.c (context::context): New.
11502         * context.h  (context::context): New.
11503         (context::get_passes): New.
11504         (context::passes_): New.
11506         * lto-cgraph.c (input_node): Update for moves of globals to
11507         fields of pass_manager.
11509         * passes.c (all_passes): Remove, in favor of a field of the
11510         same name within the new class pass_manager.
11511         (all_small_ipa_passes): Likewise.
11512         (all_lowering_passes): Likewise.
11513         (all_regular_ipa_passes): Likewise.
11514         (all_late_ipa_passes): Likewise.
11515         (all_lto_gen_passes): Likewise.
11516         (passes_by_id): Likewise.
11517         (passes_by_id_size): Likewise.
11518         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11519         the new class pass_manager.
11520         (set_pass_for_id): Convert to...
11521         (pass_manager::set_pass_for_id): ...method.
11522         (get_pass_for_id): Convert to...
11523         (pass_manager::get_pass_for_id): ...method.
11524         (register_one_dump_file): Move body of implementation into...
11525         (pass_manager::register_one_dump_file): ...here.
11526         (register_dump_files_1): Convert to...
11527         (pass_manager::register_dump_files_1): ...method.
11528         (register_dump_files): Convert to...
11529         (pass_manager::register_dump_files): ...method.
11530         (create_pass_tab): Update for moves of globals to fields of
11531         pass_manager.
11532         (dump_passes): Move body of implementation into...
11533         (pass_manager::dump_passes): ...here.
11534         (register_pass): Move body of implementation into...
11535         (pass_manager::register_pass): ...here.
11536         (init_optimization_passes): Convert into...
11537         (pass_manager::pass_manager): ...constructor for new
11538         pass_manager class, and initialize the pass_lists array.
11539         (check_profile_consistency): Update for moves of globals to
11540         fields of pass_manager.
11541         (dump_profile_report): Move body of implementation into...
11542         (pass_manager::dump_profile_report): ...here.
11543         (ipa_write_summaries_1): Update for moves of pass lists from
11544         being globals to fields of pass_manager.
11545         (ipa_write_optimization_summaries): Likewise.
11546         (ipa_read_summaries):  Likewise.
11547         (ipa_read_optimization_summaries): Likewise.
11548         (execute_all_ipa_stmt_fixups): Likewise.
11550         * statistics.c (statistics_fini): Update for moves of globals to
11551         fields of pass_manager.
11553         * toplev.c (general_init): Replace call to
11554         init_optimization_passes with construction of the pass_manager
11555         instance.
11557         * tree-pass.h (all_passes): Remove, in favor of a field of the
11558         same name within the new class pass_manager.
11559         (all_small_ipa_passes): Likewise.
11560         (all_lowering_passes): Likewise.
11561         (all_regular_ipa_passes): Likewise.
11562         (all_lto_gen_passes): Likewise.
11563         (all_late_ipa_passes): Likewise.
11564         (passes_by_id): Likewise.
11565         (passes_by_id_size): Likewise.
11566         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11567         the new class pass_manager.
11568         (get_pass_for_id): Remove.
11570 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11572         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11573         configs.
11575 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11577         * arm.md (mulhi3): New expand pattern.
11579 2013-07-30  Jan Hubicka  <jh@suse.cz>
11580             Martin Liska  <marxin.liska@gmail.com>
11582         * profile.c (compute_value_histograms): Do not ICE when
11583         there is mismatch only on some counters.
11585 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11587         PR rtl-optimization/57637
11588         * function.c (move_insn_for_shrink_wrap): Also check the
11589         GEN set of the LIVE problem for the liveness analysis
11590         if it exists, otherwise give up.
11592 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11594         PR tree-optimization/57993
11595         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11596         replaced statement in the candidate table.
11597         (phi_add_costs): Return infinite cost when the hidden basis does
11598         not dominate all phis on which the candidate is dependent.
11599         (replace_one_candidate): Record replaced statement in the
11600         candidate table.
11602 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11604         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11605         (ashlv2si3): New expander.
11606         (*ashlv2si3_i): New define_insn_and_split.
11607         * predicates.md (float_operation): Allow patterns with three
11608         basic sub-patterns.
11610         PR rtl-optimization/58021
11611         * mode-switching.c (create_pre_exit): Always split off preceding
11612         insns if we are not at the basic block head.
11614 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11616         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11617         (UCLIBC_DYNAMIC_LINKER): New macro.
11618         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11619         `-mnan=2008'.
11620         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11621         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11622         `-mnan=2008'.
11623         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11624         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11625         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11626         for SF and DF modes.  Use ieee_quad_format for TF mode.
11627         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11628         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11629         (mips_option_override): Handle `-mnan=legacy'.
11630         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11631         `-mabs=2008' and `-mnan=2008'.
11632         (OPTION_DEFAULT_SPECS): Add "nan" default.
11633         (ASM_SPEC): Handle `-mnan='.
11634         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11635         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11636         comment accordingly.
11637         (neg<mode>2): Likewise.
11638         * config/mips/mips.opt (mabs, mnan): New options.
11639         * doc/install.texi (Configuration): Document `--with-nan=' option.
11640         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11641         `-mnan=' options.
11642         (MIPS Options): Document them.
11643         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11644         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11645         * configure: Regenerate.
11646         * config.in: Regenerate.
11648 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11650         * config/i386/i386.md (float post-reload splitters): Do not check
11651         for subregs of SSE registers.
11653 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11654             H.J. Lu  <hongjiu.lu@intel.com>
11656         PR target/57954
11657         PR target/57988
11658         * config/i386/i386.md (post-reload splitter
11659         to avoid partial SSE reg dependency stalls): New pattern.
11661 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11663         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11665 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11667         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11668         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11670 2013-07-26  Cary Coutant  <ccoutant@google.com>
11672         * dwarf2out.c (die_checksum_ordered): Don't include template
11673         instantiations in signature.
11674         (is_template_parameter): New function.
11675         (is_template_instantiation): New function.
11676         (generate_skeleton_bottom_up): Don't include template instantiations
11677         in type unit DIE.
11678         (generate_skeleton): Likewise.
11679         (break_out_comdat_types): Move recursive call to break out nested
11680         types earlier.
11681         (prune_unused_types_mark_generic_parms_dies): Call
11682         is_template_parameter.
11684 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11686         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11687         uses vector registers.
11688         * config/aarch64/iterators.md: Add attributes rtn and vas.
11690 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11691             Richard Earnshaw  <richard.earnshaw@arm.com>
11693         * combine.c (simplify_comparison): Re-canonicalize operands
11694         where appropriate.
11695         * config/arm/arm.md (movcond_addsi): New splitter.
11697 2013-07-25  Sterling Augustine  <saugustine@google.com>
11699         * dwarf2out.c (size_of_pubnames): Move code to...
11700         (include_pubname_in_output): ...here.  New.
11701         (want_pubnames): Rearrange.
11702         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11704 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11706         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11708 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11710         PR target/38836
11711         * doc/extend.texi: Remove obsolete builtins. Fix
11712         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11714 2013-07-25  Jan Hubicka  <jh@suse.cz>
11716         * cgraph.c (release_function_body): Break out from ...
11717         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11718         and DECL_ARGUMENTS.
11719         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11720         old_tree in the map.
11721         (create_specialized_node): Update.
11722         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11723         into index.
11724         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11725         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11726         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11727         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11728         DECL_RESULT.
11730 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11732         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11733         addsi3_carryin_alt2_<optab>): Correct output template.
11735 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11737         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11738         Adjust for arm_restrict_it.
11739         Remove trailing whitespace.
11741 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11743         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11744         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11746         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11748 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11750         PR rtl-optimization/57960
11751         * lra-constraints.c (process_alt_operands): Use the right mode
11752         when checking strict_low.
11754 2013-07-25  Jan Hubicka  <jh@suse.cz>
11756         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11757         * cgraph.c (cgraph_remove_node): Do not release function body
11758         when in cgraph streaming.
11759         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11760         in other partitions are not considered reachable; fix handling of
11761         clones.
11763 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11765         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11767 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11769         PR target/19599
11770         PR target/57731
11771         PR target/57837
11772         * config/arm/arm.md ("*sibcall_insn): Replace use of
11773         Ss with US. Adjust output for v5 and v4t.
11774         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11776         * config/arm/constraints.md ("Ss"): Rename to US.
11778 2013-07-25  Terry Guo  <terry.guo@arm.com>
11780         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11781         shift_add/shift_sub0/shift_sub1 RTXs.
11783 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11784             Anton Blanchard  <anton@au1.ibm.com>
11786         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11787         (altivec_vpks<VI_char>ss): Likewise.
11788         (altivec_vpks<VI_char>us): Likewise.
11789         (altivec_vpku<VI_char>us): Likewise.
11790         (altivec_vpku<VI_char>um): Likewise.
11792 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11794         Introduce context class.
11796         * Makefile.in (CONTEXT_H): New.
11797         (OBJS): Add context.o.
11798         (toplev.o): Add CONTEXT_H to dependencies.
11799         (context.o): New.
11801         * toplev.c (general_init): Create the singleton gcc::context instance.
11803         * context.c: New.
11805         * context.h: New.
11807 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11809         PR rtl-optimization/57968
11810         * mode-switching.c (create_pre_exit): Allow instructions that
11811         don't set a return register to need a non-exit mode.
11813 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11814             Anton Blanchard  <anton@au1.ibm.com>
11816         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11817         operands to vperm for little endian.
11818         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11819         of lvsl to create the control mask for a vperm for little endian.
11821 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11822             Anton Blanchard  <anton@au1.ibm.com>
11824         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11825         two operands for little-endian.
11827 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11829         * config/mips/mips.c (mips_case_values_threshold): New.
11830         (TARGET_CASE_VALUES_THRESHOLD): Define.
11832 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11833             Anton Blanchard  <anton@au1.ibm.com>
11835         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11836         selection of field for vector splat in little endian mode.
11838 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11840         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11841         expanders to rs6000.md.
11842         (ior<mode>3): Likewise.
11843         (and<mode>3): Likewise.
11844         (one_cmpl<mode>2): Likewise.
11845         (nor<mode>3): Likewise.
11846         (andc<mode>3): Likewise.
11847         (eqv<mode>3): Likewise.
11848         (nand<mode>3): Likewise.
11849         (orc<mode>3): Likewise.
11851         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11852         declaration.
11854         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11855         to split multi-word logical operations.
11856         (rs6000_split_logical_di): Likewise.
11857         (rs6000_split_logical): Likewise.
11859         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11860         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11861         and allow TImode operations in 32-bit.
11862         (vsx_and<mode>3_64bit): Likewise.
11863         (vsx_ior<mode>3_32bit): Likewise.
11864         (vsx_ior<mode>3_64bit): Likewise.
11865         (vsx_xor<mode>3_32bit): Likewise.
11866         (vsx_xor<mode>3_64bit): Likewise.
11867         (vsx_one_cmpl<mode>2_32bit): Likewise.
11868         (vsx_one_cmpl<mode>2_64bit): Likewise.
11869         (vsx_nor<mode>3_32bit): Likewise.
11870         (vsx_nor<mode>3_64bit): Likewise.
11871         (vsx_andc<mode>3_32bit): Likewise.
11872         (vsx_andc<mode>3_64bit): Likewise.
11873         (vsx_eqv<mode>3_32bit): Likewise.
11874         (vsx_eqv<mode>3_64bit): Likewise.
11875         (vsx_nand<mode>3_32bit): Likewise.
11876         (vsx_nand<mode>3_64bit): Likewise.
11877         (vsx_orc<mode>3_32bit): Likewise.
11878         (vsx_orc<mode>3_64bit): Likewise.
11880         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11881         logical types in GPRs.
11883         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11884         logical insns to rs6000.md, and allow TImode operations in
11885         32-bit.
11886         (altivec_ior<mode>3): Likewise.
11887         (altivec_xor<mode>3): Likewise.
11888         (altivec_one_cmpl<mode>2): Likewise.
11889         (altivec_nor<mode>3): Likewise.
11890         (altivec_andc<mode>3): Likewise.
11892         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11893         attributes for moving the 128-bit logical operations into
11894         rs6000.md.
11895         (BOOL_REGS_OUTPUT): Likewise.
11896         (BOOL_REGS_OP1): Likewise.
11897         (BOOL_REGS_OP2): Likewise.
11898         (BOOL_REGS_UNARY): Likewise.
11899         (BOOL_REGS_AND_CR0): Likewise.
11900         (one_cmpl<mode>2): Add support for DI logical operations on
11901         32-bit, splitting the operations to 32-bit.
11902         (anddi3): Likewise.
11903         (iordi3): Likewise.
11904         (xordi3): Likewise.
11905         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11906         changes to combine the 32/64-bit code, allow logical operations on
11907         TI mode in 32-bit, and to use similar match_operator patterns like
11908         scalar mode uses.  Combine the Altivec and VSX code for logical
11909         operations, and move it here.
11910         (ior<mode>3, 128-bit types): Likewise.
11911         (xor<mode>3, 128-bit types): Likewise.
11912         (one_cmpl<mode>3, 128-bit types): Likewise.
11913         (nor<mode>3, 128-bit types): Likewise.
11914         (andc<mode>3, 128-bit types): Likewise.
11915         (eqv<mode>3, 128-bit types): Likewise.
11916         (nand<mode>3, 128-bit types): Likewise.
11917         (orc<mode>3, 128-bit types): Likewise.
11918         (and<mode>3_internal): Likewise.
11919         (bool<mode>3_internal): Likewise.
11920         (boolc<mode>3_internal1): Likewise.
11921         (boolc<mode>3_internal2): Likewise.
11922         (boolcc<mode>3_internal1): Likewise.
11923         (boolcc<mode>3_internal2): Likewise.
11924         (eqv<mode>3_internal1): Likewise.
11925         (eqv<mode>3_internal2): Likewise.
11926         (one_cmpl1<mode>3_internal): Likewise.
11928 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11930         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11931         Rename flag_stack_usage to flag_stack_usage_info.
11933 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11935         * config/microblaze/sync.md: New file.
11936         * config/microblaze/microblaze.md: Include sync.md
11937         * config/microblaze/microblaze.c: Add print_operand 'y'.
11938         * config/microblaze/constraints.md: Add memory_contraint
11939         'Q' which is a single register.
11941 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11943         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11945 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11947         * reload.c (find_reloads): Exit loop once we find this operand
11948         cannot be reloaded somehow for this alternative.
11950         * reload.c (find_reloads): Exit loop once we find a hard register.
11952         * rtlanal.c (computed_jump_p): Exit loop once we find label
11953         reference is used.
11955         * i386.c (ix86_pad_returns): Exit loop after setting replace.
11957         * cfgloopmanip.c (remove_path): Exit loop after setting
11958         irred_invalidated.
11960         * gensupport.c (subst_dup): Avoid loop if code is not
11961         MATCH_DUP nor MATCH_OP_DUP.
11963 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
11965         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
11967 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11969         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
11970         true for SP_REGNUM if mode == ptr_mode.
11971         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
11972         with value R0_REGNUM + 31.
11974 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11976         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
11977         pad pointer-typed argument downward.
11979 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11981         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
11982         and __ILP32__ when the ILP32 model is in use.
11984 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
11986         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
11987         (aarch64_load_symref_appropriately): In the case of
11988         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
11989         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
11990         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
11991         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
11992         if 'mode' doesn't equal to 'ptr_mode'.
11993         (aarch64_output_mi_thunk): Add an assertion on the alignment of
11994         'vcall_offset'; change to call aarch64_emit_move differently depending
11995         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
11996         to calculate the upper bound of 'vcall_offset'.
11997         (aarch64_cannot_force_const_mem): Change to also return true if
11998         mode != ptr_mode.
11999         (aarch64_legitimize_reload_address): In the case of large
12000         displacements, add new local variable 'xmode' and an assertion
12001         based on it; change to use 'xmode' to generate the new rtx and
12002         reload.
12003         (aarch64_asm_trampoline_template): Change to generate the template
12004         differently depending on TARGET_ILP32 or not; change to use
12005         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12006         (aarch64_trampoline_size): Removed.
12007         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12008         and replace immediate literals with it.  Change to use 'ptr_mode'
12009         instead of 'DImode' and call convert_memory_address if the mode
12010         of 'fnaddr' doesn't equal to 'ptr_mode'.
12011         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12012         to output symbol.
12013         (aarch64_elf_asm_destructor): Likewise.
12014         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12015         on TARGET_ILP32 instead of aarch64_trampoline_size.
12016         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12017         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12018         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12019         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12020         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12021         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12022         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12023         depending on the value of 'mode'.
12024         (add_losym_<mode>): New.
12025         (ldr_got_small_<mode>): New, based on ldr_got_small.
12026         (ldr_got_small): Remove.
12027         (ldr_got_small_sidi): New.
12028         * config/aarch64/iterators.md (P): New.
12029         (PTR): Change to 'ptr_mode' in the condition.
12031 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12033         * config.gcc (aarch64*-*-*): Support --with-abi.
12034         (aarch64*-*-elf): Support --with-multilib-list.
12035         (aarch64*-*-linux*): Likewise.
12036         (supported_defaults): Add abi to aarch64*-*-*.
12037         * configure.ac: Mention AArch64 for --with-multilib-list.
12038         * configure: Re-generated.
12039         * config/aarch64/biarchilp32.h: New file.
12040         * config/aarch64/biarchlp64.h: New file.
12041         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12042         (ABI_SPEC): Ditto.
12043         (MULTILIB_DEFAULTS): Ditto.
12044         (DRIVER_SELF_SPECS): Ditto.
12045         (ASM_SPEC): Update to also substitute -mabi.
12046         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12047         file whose name depends on -mabi= and -mbig-endian.
12048         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12049         TARGET_ILP32.
12050         (POINTER_SIZE): New define.
12051         (POINTERS_EXTEND_UNSIGNED): Ditto.
12052         (enum aarch64_abi_type): New enumeration tag.
12053         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12054         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12055         (TARGET_ILP32): New define.
12056         * config/aarch64/aarch64.opt (mabi): New.
12057         (aarch64_abi): New.
12058         (ilp32, lp64): New values for -mabi.
12059         * config/aarch64/t-aarch64 (comma): New define.
12060         (MULTILIB_OPTIONS): Ditto.
12061         (MULTILIB_DIRNAMES): Ditto.
12062         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12063         * doc/invoke.texi: Document -mabi for AArch64.
12065 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12067         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12069 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12070             Anton Blanchard  <anton@au1.ibm.com>
12072         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12073         endianness when selecting field to splat.
12075 2013-07-22  Eric Christopher  <echristo@gmail.com>
12077         * dwarf2out.c (die_odr_checksum): New function to use
12078         CHECKSUM_ macros and ULEB128 for DIE tag.
12079         (generate_type_signature): Use.
12081 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12083         * config.gcc (sparc*-*-*): Accept leon3 processor.
12084         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12085         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12086         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12087         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12088         (mfix-ut699): Adjust comment.
12089         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12090         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12091         (CPP_CPU_SPEC): Likewise.
12092         (ASM_CPU_SPEC): Likewise.
12093         * config/sparc/sparc.c (leon3_cost): New constant.
12094         (sparc_option_override): Add leon3 support.
12095         (mem_ref): New function.
12096         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12097         (sparc_do_work_around_errata): Look into the instruction in the delay
12098         slot and adjust accordingly.  Add fix for the data cache nullify issues
12099         of the UT699.  Change insertion position for the NOP.
12100         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12101         (leon3_load): New reservation.
12102         (leon_store): Bump latency to 2.
12103         (grfpu): New automaton.
12104         (grfpu_alu): New unit.
12105         (grfpu_ds): Likewise.
12106         (leon_fp_alu): Adjust.
12107         (leon_fp_mult): Delete.
12108         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12109         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12110         * config/sparc/sparc.md (cpu): Add leon3.
12111         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12112         (swapsi): Likewise.
12113         (atomic_test_and_set): Likewise.
12114         (ldstub): Likewise.
12116 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12118         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12119         default for R5900 targets.
12120         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12121         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12122         * config/mips/mips.c (mips_option_override): Report an error for
12123         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12124         for -march=r5900 -mhard-float.
12126 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12128         * df-problems.c (can_move_insns_across): Exit loop once we
12129         find a non-fixed, non-global register.
12131         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12132         setting can_throw.
12134         * omega.c (omega_eliminate_red): Break after setting red_found.
12135         (omega_problem_has_red_equations): Similarly after setting found.
12136         (omega_query_variable): Similarly after setting coupled.
12138 2013-07-22  Marek Polacek  <polacek@redhat.com>
12140         * gimplify.c: Don't include gimple.h twice.
12142 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12144         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12145         instead of TARGET_THUMB1.
12146         (Pz): New constraint.
12147         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12148         encodings.
12149         (compare_negsi_si): Likewise.
12150         (compare_addsi2_op0): Likewise.
12151         (compare_addsi2_op1): Likewise.
12152         (addsi3_carryin_<optab>): Likewise.
12153         (addsi3_carryin_alt2_<optab>): Likewise.
12154         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12155         for arm_restrict_it.
12156         (subsi3_carryin): Likewise.
12157         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12158         (minmax_arithsi): Disable for arm_restrict_it.
12159         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12160         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12161         (satsi_<SAT:code>_shift): Likewise.
12162         (arm_shiftsi3): Add alternative for 16-bit encoding.
12163         (arm32_movhf): Disable for arm_restrict_it.
12164         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12165         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12167 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12169         * config/arm/arm.md (attribute "insn"): Delete.
12170         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12171         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12172         (not_shiftsi): Update for attribute change.
12173         (not_shiftsi_compare0): Likewise.
12174         (not_shiftsi_compare0_scratch): Likewise.
12175         (arm_one_cmplsi2): Likewise.
12176         (thumb1_one_cmplsi2): Likewise.
12177         (notsi_compare0): Likewise.
12178         (notsi_compare0_scratch): Likewise.
12179         (thumb1_movdi_insn): Likewise.
12180         (arm_movsi_insn): Likewise.
12181         (movhi_insn_arch4): Likewise.
12182         (movhi_bytes): Likewise.
12183         (arm_movqi_insn): Likewise.
12184         (thumb1_movqi_insn): Likewise.
12185         (arm32_movhf): Likewise.
12186         (thumb1_movhf): Likewise.
12187         (arm_movsf_soft_insn): Likewise.
12188         (thumb1_movsf_insn): Likewise.
12189         (thumb_movdf_insn): Likewise.
12190         (movsicc_insn): Likewise.
12191         (movsfcc_soft_insn): Likewise.
12192         (and_scc): Likewise.
12193         (cond_move): Likewise.
12194         (if_move_not): Likewise.
12195         (if_not_move): Likewise.
12196         (if_shift_move): Likewise.
12197         (if_move_shift): Likewise.
12198         (if_shift_shift): Likewise.
12199         (if_not_arith): Likewise.
12200         (if_arith_not): Likewise.
12201         (cond_move_not): Likewise.
12202         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12203         (neon_mov<mode>): Likewise.
12204         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12205         (thumb2_movsi_vfp): Likewise.
12206         (movsf_vfp): Likewise.
12207         (thumb2_movsf_vfp): Likewise.
12208         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12209         change.
12210         (cortexa7_older_only): Likewise.
12211         (cortexa7_younger): Likewise.
12212         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12213         (1020alu_shift_op): Likewise.
12214         (1020alu_shift_reg_op): Likewise.
12215         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12216         (alu_shift_op): Likewise.
12217         (alu_shift_reg_op): Likewise.
12218         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12219         (11_alu_shift_op): Likewise.
12220         (11_alu_shift_reg_op): Likewise.
12221         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12222         (9_alu_shift_reg_op): Likewise.
12223         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12224         change.
12225         (cortex_a15_alu_shift): Likewise.
12226         (cortex_a15_alu_shift_reg): Likewise.
12227         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12228         (cortex_a5_alu_shift): Likewise.
12229         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12230         change.
12231         (cortex_a53_alu_shift): Likewise.
12232         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12233         change.
12234         (cortex_a7_alu_reg): Likewise.
12235         (cortex_a7_alu_shift): Likewise.
12236         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12237         (cortex_a8_alu_shift): Likewise.
12238         (cortex_a8_alu_shift_reg): Likewise.
12239         (cortex_a8_mov): Likewise.
12240         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12241         (cortex_a9_dp_shift): Likewise.
12242         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12243         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12244         (cortex_r4_mov): Likewise.
12245         (cortex_r4_alu_shift): Likewise.
12246         (cortex_r4_alu_shift_reg): Likewise.
12247         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12248         (526_alu_shift_op): Likewise.
12249         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12250         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12251         (626te_alu_shift_op): Likewise.
12252         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12253         (726te_alu_op): Likewise.
12254         (726te_alu_shift_op): Likewise.
12255         (726te_alu_shift_reg_op): Likewise.
12256         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12257         (mp626_alu_shift_op): Likewise.
12258         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12259         (pj4_alu_e1_conds): Likewise.
12260         (pj4_alu): Likewise.
12261         (pj4_alu_conds): Likewise.
12262         (pj4_shift): Likewise.
12263         (pj4_shift_conds): Likewise.
12264         (pj4_alu_shift): Likewise.
12265         (pj4_alu_shift_conds): Likewise.
12267 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12269         * config/arm/predicates.md (shiftable_operator_strict_it):
12270         New predicate.
12271         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12272         Disable cond_exec version for arm_restrict_it.
12273         (thumb2_smaxsi3): Convert to generate cond_exec.
12274         (thumb2_sminsi3): Likewise.
12275         (thumb32_umaxsi3): Likewise.
12276         (thumb2_uminsi3): Likewise.
12277         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12278         (thumb2_neg_abssi2): Likewise.
12279         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12280         (thumb2_movsicc_insn): Adjust alternatives.
12281         (thumb2_mov_negscc): Disable for arm_restrict_it.
12282         (thumb2_mov_negscc_strict_it): New pattern.
12283         (thumb2_mov_notscc_strict_it): New pattern.
12284         (thumb2_mov_notscc): Disable for arm_restrict_it.
12285         (thumb2_ior_scc): Likewise.
12286         (thumb2_ior_scc_strict_it): New pattern.
12287         (thumb2_cond_move): Adjust for arm_restrict_it.
12288         (thumb2_cond_arith): Disable for arm_restrict_it.
12289         (thumb2_cond_arith_strict_it): New pattern.
12290         (thumb2_cond_sub): Adjust for arm_restrict_it.
12291         (thumb2_movcond): Likewise.
12292         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12293         (thumb2_zero_extendhisi2_v6): Likewise.
12294         (thumb2_zero_extendqisi2_v6): Likewise.
12295         (orsi_notsi_si): Likewise.
12296         (orsi_not_shiftsi_si): Likewise.
12298 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12300         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12301         instruction sequence is 1 byte shorter.
12303 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12305         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12306         it is not needed after split.
12308 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12310         PR target/51784
12311         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12312         second label for nonlocal goto receivers. Don't output pic base labels
12313         unless we're producing PIC; mark that action unreachable().
12314         (ix86_save_reg): If the function contains a nonlocal label, save the
12315         PIC base reg.
12316         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12317         * config/darwin.c (emitted_pic_label_num): New GTY.
12318         (update_pic_label_number_if_needed): New.
12319         (machopic_output_function_base_name): Adjust for nonlocal receiver
12320         case.
12321         (machopic_should_output_picbase_label): New.
12322         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12323         (nonlocal_goto_receiver): New insn and split.
12325 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12327         * config/aarch64/aarch64-builtins.c
12328         (aarch64_fold_builtin): Fold abs in all modes.
12329         * config/aarch64/aarch64-simd-builtins.def
12330         (abs): Enable for all modes.
12331         * config/aarch64/arm_neon.h
12332         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12333         (vabs_f64): Add missing intrinsic.
12335 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12337         * config/aarch64/arm_neon.h (vabs_s64): New function
12339 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12341         PR target/57516
12342         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12343         * config/avr/avr.md (adjust_len): Add `round'.
12344         * config/avr/avr-protos.h (avr_out_round): New prototype.
12345         (avr_out_plus): Add `out_label' argument.
12346         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12347         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12348         Handle the case where `insn' is just a pattern.
12349         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12350         (avr_out_round): New function.
12351         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12353 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12355         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12356         Add check for flag_stack_usage to handle -fstack-usage support
12358 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12360         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12361         interaction for new Power8 flags and VSX.
12363 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12365         PR middle-end/57698
12366         * tree-inline.c (expand_call_inline): Emit errors during
12367         early_inlining only if optimization is not turned on.
12369 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12371         * passes.def: New.
12373         * passes.c (init_optimization_passes): Move the construction of
12374         the pass hierarchy into a new passes.def file.
12376         * Makefile.in (passes.o): Add dependency on passes.def.
12378 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12380         * passes.c (init_optimization_passes): Introduce macros for
12381         constructing the tree of passes (INSERT_PASSES_AFTER,
12382         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12383         TERMINATE_PASS_LIST).
12385 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12386             Wei Mi  <wmi@google.com>
12388         PR rtl-optimization/57878
12389         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12390         top.
12391         (reload_pseudo_compare_func): Check nregs first for reload
12392         pseudos.
12394 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12396         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12398 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12400         * read-rtl.c (validate_const_int): Once an invalid character is
12401         seen, quit the loop.
12403         * gengtype.c (write_roots): Similarly once we find the "deletable"
12404         or "if_marked" option.
12406 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12408         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12409         "xtab" and "sat".  Move value "clz" from here to ...
12410         (attriubte "type"): ... here.
12411         (satsi_<SAT:code>): Delete "insn" attribute.
12412         (satsi_<SAT:code>_shift): Likewise.
12413         (arm_zero_extendqisi2addsi): Likewise.
12414         (arm_extendqisi2addsi): Likewise.
12415         (clzsi2): Update for attribute changes.
12416         (rbitsi2): Likewise.
12417         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12418         attribute.
12419         (arm_usatsihi): Likewise.
12420         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12422 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12424         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12425         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12426         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12427         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12428         in alphabetical order.
12429         (attribute "core_cycles"): Update for attribute changes.
12430         (arm_addsi3): Likewise.
12431         (addsi3_compare0): Likewise.
12432         (addsi3_compare0_scratch): Likewise.
12433         (addsi3_compare_op1): Likewise.
12434         (addsi3_compare_op2): Likewise.
12435         (compare_addsi2_op0): Likewise.
12436         (compare_addsi2_op1): Likewise.
12437         (addsi3_carryin_shift_<optab>): Likewise.
12438         (subsi3_carryin_shift): Likewise.
12439         (rsbsi3_carryin_shift): Likewise.
12440         (arm_subsi3_insn): Likewise.
12441         (subsi3_compare0): Likewise.
12442         (subsi3_compare): Likewise.
12443         (arm_andsi3_insn): Likewise.
12444         (thumb1_andsi3_insn): Likewise.
12445         (andsi3_compare0): Likewise.
12446         (andsi3_compare0_scratch): Likewise.
12447         (zeroextractsi_compare0_scratch
12448         (andsi_not_shiftsi_si): Likewise.
12449         (iorsi3_insn): Likewise.
12450         (iorsi3_compare0): Likewise.
12451         (iorsi3_compare0_scratch): Likewise.
12452         (arm_xorsi3): Likewise.
12453         (thumb1_xorsi3_insn): Likewise.
12454         (xorsi3_compare0): Likewise.
12455         (xorsi3_compare0_scratch): Likewise.
12456         (satsi_<SAT:code>_shift): Likewise.
12457         (rrx): Likewise.
12458         (arm_shiftsi3): Likewise.
12459         (shiftsi3_compare0): Likewise.
12460         (not_shiftsi): Likewise.
12461         (not_shiftsi_compare0): Likewise.
12462         (not_shiftsi_compare0_scratch): Likewise.
12463         (arm_one_cmplsi2): Likewise.
12464         (thumb_one_complsi2): Likewise.
12465         (notsi_compare0): Likewise.
12466         (notsi_compare0_scratch): Likewise.
12467         (thumb1_zero_extendhisi2): Likewise.
12468         (arm_zero_extendhisi2): Likewise.
12469         (arm_zero_extendhisi2_v6): Likewise.
12470         (arm_zero_extendhisi2addsi): Likewise.
12471         (thumb1_zero_extendqisi2): Likewise.
12472         (thumb1_zero_extendqisi2_v6): Likewise.
12473         (arm_zero_extendqisi2): Likewise.
12474         (arm_zero_extendqisi2_v6): Likewise.
12475         (arm_zero_extendqisi2addsi): Likewise.
12476         (thumb1_extendhisi2): Likewise.
12477         (arm_extendhisi2): Likewise.
12478         (arm_extendhisi2_v6): Likewise.
12479         (arm_extendqisi): Likewise.
12480         (arm_extendqisi_v6): Likewise.
12481         (arm_extendqisi2addsi): Likewise.
12482         (thumb1_extendqisi2): Likewise.
12483         (thumb1_movdi_insn): Likewise.
12484         (arm_movsi_insn): Likewise.
12485         (movsi_compare0): Likewise.
12486         (movhi_insn_arch4): Likewise.
12487         (movhi_bytes): Likewise.
12488         (arm_movqi_insn): Likewise.
12489         (thumb1_movqi_insn): Likewise.
12490         (arm32_movhf): Likewise.
12491         (thumb1_movhf): Likewise.
12492         (arm_movsf_soft_insn): Likewise.
12493         (thumb1_movsf_insn): Likewise.
12494         (movdf_soft_insn): Likewise.
12495         (thumb_movdf_insn): Likewise.
12496         (arm_cmpsi_insn): Likewise.
12497         (cmpsi_shiftsi): Likewise.
12498         (cmpsi_shiftsi_swp): Likewise.
12499         (arm_cmpsi_negshiftsi_si): Likewise.
12500         (movsicc_insn): Likewise.
12501         (movsfcc_soft_insn): Likewise.
12502         (arith_shiftsi): Likewise.
12503         (arith_shiftsi_compare0
12504         (arith_shiftsi_compare0_scratch
12505         (sub_shiftsi): Likewise.
12506         (sub_shiftsi_compare0
12507         (sub_shiftsi_compare0_scratch
12508         (and_scc): Likewise.
12509         (cond_move): Likewise.
12510         (if_plus_move): Likewise.
12511         (if_move_plus): Likewise.
12512         (if_move_not): Likewise.
12513         (if_not_move): Likewise.
12514         (if_shift_move): Likewise.
12515         (if_move_shift): Likewise.
12516         (if_shift_shift): Likewise.
12517         (if_not_arith): Likewise.
12518         (if_arith_not): Likewise.
12519         (cond_move_not): Likewise.
12520         (thumb1_ashlsi3): Set type attribute.
12521         (thumb1_ashrsi3): Likewise.
12522         (thumb1_lshrsi3): Likewise.
12523         (thumb1_rotrsi3): Likewise.
12524         (shiftsi3_compare0_scratch): Likewise.
12525         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12526         (neon_mov<mode>): Likewise.
12527         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12528         attribute changes.
12529         (thumb2_movsi_insn): Likewise.
12530         (thumb2_cmpsi_neg_shiftsi): Likewise.
12531         (thumb2_extendqisi_v6): Likewise.
12532         (thumb2_zero_extendhisi2_v6): Likewise.
12533         (thumb2_zero_extendqisi2_v6): Likewise.
12534         (thumb2_shiftsi3_short): Likewise.
12535         (thumb2_addsi3_compare0_scratch): Likewise.
12536         (orsi_not_shiftsi_si): Likewise.
12537         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12538         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12539         changes.
12540         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12541         (1020alu_shift_op): Likewise.
12542         (1020alu_shift_reg_op): Likewise.
12543         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12544         (alu_shift_op): Likewise.
12545         (alu_shift_reg_op): Likewise.
12546         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12547         (11_alu_shift_op): Likewise.
12548         (11_alu_shift_reg_op): Likewise.
12549         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12550         (9_alu_shift_reg_op): Likewise.
12551         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12552         changes.
12553         (cortex_a15_alu_shift): Likewise.
12554         (cortex_a15_alu_shift_reg): Likewise.
12555         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12556         changes.
12557         (cortex_a5_alu_shift): Likewise.
12558         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12559         changes.
12560         (cortex_a53_alu_shift): Likewise.
12561         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12562         changes.
12563         (cortex_a7_alu_reg): Likewise.
12564         (cortex_a7_alu_shift): Likewise.
12565         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12566         changes.
12567         (cortex_a8_alu_shift): Likewise.
12568         (cortex_a8_alu_shift_reg): Likewise.
12569         (cortex_a8_mov): Likewise.
12570         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12571         (cortex_a9_dp_shift): Likewise.
12572         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12573         changes.
12574         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12575         changes.
12576         (cortex_r4_mov): Likewise.
12577         (cortex_r4_alu_shift): Likewise.
12578         (cortex_r4_alu_shift_reg): Likewise.
12579         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12580         (526_alu_shift_op): Likewise.
12581         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12582         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12583         (626te_alu_shift_op): Likewise.
12584         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12585         (726te_alu_op): Likewise.
12586         (726te_alu_shift_op): Likewise.
12587         (726te_alu_shift_reg_op): Likewise.
12588         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12589         (mp626_alu_shift_op): Likewise.
12590         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12591         (pj4_alu_e1_conds): Likewise.
12592         (pj4_alu): Likewise.
12593         (pj4_alu_conds): Likewise.
12594         (pj4_shift): Likewise.
12595         (pj4_shift_conds): Likewise.
12596         (pj4_alu_shift): Likewise.
12597         (pj4_alu_shift_conds): Likewise.
12598         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12599         changes.
12600         (cortexa7_older_only): Likewise.
12601         (cortexa7_younger): Likewise.
12603 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12605         * ipa-pure-const.c (generate_summary): Rename to...
12606         (pure_const_generate_summary): ... this.
12608 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12610         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12612 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12614         PR target/57909
12615         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12616         usage in HI mode.
12618 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12620         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12621         enabled without -march=zEC12.
12622         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12623         flags to be set.
12625 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12627         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12628         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12629         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12630         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12631         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12632         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12633         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12634         ISA_HAS_FP4.
12635         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12636         and ISA_HAS_NMADD3_NMSUB3.
12637         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12638         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12639         (nmsub4<mode>, nmsub3<mode>): Likewise.
12640         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12642 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12644         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12645         TARGET_MIPS5400 checking.
12647 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12648             Peter Bergner  <bergner@vnet.ibm.com>
12650         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12651         registers in the comment.
12652         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12653         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12654         rather than FIRST_PSEUDO_REGISTERS.
12656 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12658         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12659         enable extra ISA flags with TARGET_HTM.
12661 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12663         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12664         Fix comment typos.
12666 2013-07-15  Cong Hou  <congh@google.com>
12668         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12669         in compare function for sorting.
12671 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12673         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12674         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12675         * config/rs6000/rs6000.opt: Add -mhtm option.
12676         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12677         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12678         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12679         __HTM__ if the HTM instructions are available.
12680         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12681         htm_spr_reg_operand): New define_predicates.
12682         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12683         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12684         Include htm.md.
12685         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12686         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12687         HTM builtin functions.
12688         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12689         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12690         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12691         (rs6000_builtin_mask_calculate): Likewise.
12692         (rs6000_option_override_internal): Likewise.
12693         (bdesc_htm): Add new HTM builtin support.
12694         (htm_spr_num): New function.
12695         (htm_spr_regno): Likewise.
12696         (rs6000_htm_spr_icode): Likewise.
12697         (htm_expand_builtin): Likewise.
12698         (htm_init_builtins): Likewise.
12699         (rs6000_expand_builtin): Add support for HTM builtin functions.
12700         (rs6000_init_builtins): Likewise.
12701         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12702         option.
12703         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12704         (TARGET_HTM, MASK_HTM): Define macros.
12705         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12706         (FIXED_REGISTERS): Likewise.
12707         (CALL_USED_REGISTERS): Likewise.
12708         (CALL_REALLY_USED_REGISTERS): Likewise.
12709         (REG_ALLOC_ORDER): Likewise.
12710         (enum reg_class): Likewise.
12711         (REG_CLASS_NAMES): Likewise.
12712         (REG_CLASS_CONTENTS): Likewise.
12713         (REGISTER_NAMES): Likewise.
12714         (ADDITIONAL_REGISTER_NAMES): Likewise.
12715         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12716         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12717         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12718         * config/rs6000/htm.md: New file.
12719         * config/rs6000/htmintrin.h: New file.
12720         * config/rs6000/htmxlintrin.h: New file.
12722 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12724         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12725         Define SYMBOL_TINY_GOT, update comment.
12726         * config/aarch64/aarch64.c
12727         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12728         (aarch64_expand_mov_immediate): Likewise.
12729         (aarch64_print_operand): Likewise.
12730         (aarch64_classify_symbol): Likewise.
12731         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12732         (ldr_got_tiny): Define.
12734 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12736         PR tree-optimization/54094
12737         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12738           scheduling dimension for the parallelism check from the polyhedral
12739           information in the AST.
12740         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12741           in 2D + 1 form.
12743 2013-07-13  Jason Merrill  <jason@redhat.com>
12745         * print-tree.c (debug_vec_tree): Use debug_raw.
12746         (debug_raw (vec<tree, va_gc> &)): New.
12747         (debug_raw (vec<tree, va_gc> *)): New.
12748         * tree.h: Declare them.
12750 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12752         * ifcvt.c (ifcvt_after_combine): New static variable.
12753         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12754         for size.
12755         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12756         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12757         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12759 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12761         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12763 2013-07-12  Michael Matz  <matz@suse.de>
12765         PR middle-end/55771
12766         * convert.c (convert_to_real): Reject non-float inner types.
12768 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12770         * config/aarch64/aarch64-protos.h
12771         (aarch64_simd_immediate_valid_for_move): Remove.
12772         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12773         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12774         cases.
12775         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12777 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12779         * config/mips/mips.c (mips_conditional_register_usage): Do not
12780         use t[0-7] registers in MIPS16 mode when optimizing for size.
12782 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12784         * config/i386/i386.c (dispatch_function_versions): Fix array
12785         indexing of function_version_info to match actual_versions.
12787 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12789         * vec.h (struct va_gc): Move release out-of-line.
12790         (va_gc::release): Call ggc_free on released vec.
12792 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12794         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12795         Require GOT register as additional operand in UNSPEC.
12796         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12797         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12798         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12799         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12800         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12801         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12802         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12804 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12806         PR target/57631
12807         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12808         name seen by assembler/linker rather if available.
12810 2013-07-11  Andreas Schwab  <schwab@suse.de>
12812         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12814 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12816         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12818 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12820         * doc/tm.texi.in: Move hook documentation to ....
12821         * target.def: ... here.
12823         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12824         text on @hook line.
12825         * doc/tm.texi: Regenerate.
12827 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12829         PR c++/57869
12830         * doc/invoke.texi: Document -Wconditionally-supported.
12832 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12834         PR target/57844
12835         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12836         of my_fp.
12838 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12840         PR target/57506
12841         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12842         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12843         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12844         Remove duplicate devices.
12845         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12846         * config/avr/t-multilib: Regenerate.
12847         * config/avr/avr-tables.opt: Regenerate.
12848         * doc/avr-mmcu.texi: Regenerate.
12850 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12852         PR target/56987
12853         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12855 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12857         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12858         the cost of MULT when optimizing for size.
12860 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12862         * config/cr16/cr16-protos.h: Don't include target.h.
12864 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12866         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12867         adjust register size for TDmode and TFmode for VSX registers.
12869 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12871         PR target/56892
12872         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12873         hook_bool_const_tree_true.
12875 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12877         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12878         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12879         * config/s390/s390.md: Define FPR*_REGNUM constants.
12880         Fix FPR2_REGNUM constant (18 -> 17).
12881         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12882         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12883         ("trunc<BFP:mode><DFP_ALL:mode>2")
12884         ("trunc<DFP_ALL:mode><BFP:mode>2")
12885         ("*extend<BFP:mode><DFP_ALL:mode>2")
12886         ("*extend<DFP_ALL:mode><BFP:mode>2")
12887         ("extend<BFP:mode><DFP_ALL:mode>2")
12888         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12889         FPR4_REGNUM.
12891 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12893         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12895 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12897         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12898         and cfun_fpr_bit_p to cfun_fpr_save_p.
12899         (s390_frame_area, s390_register_info, s390_frame_info)
12900         (s390_emit_prologue, s390_emit_epilogue)
12901         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12902         register numbers.
12903         * config/s390/s390.h: Define *_REGNUM macros for floating point
12904         register numbers.
12906 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12908         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12910 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12912         PR rtl-optimization/57786
12913         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12914         and break out of the loop when it is set to false.
12916 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12918         PR target/57819
12919         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12920         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12921         (const_int 63)) 0)).
12922         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12923         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12924         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12926         PR rtl-optimization/57829
12927         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12928         mask bits outside of mode are just sign-extension from mode to HWI.
12930 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12932         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12933         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12934         adjust_address instead of change_address to keep info about alignment.
12935         (emit_strmov): Remove.
12936         (emit_memmov): New function.
12937         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12938         (expand_movmem_epilogue): Likewise and return updated rtx for
12939         destination.
12940         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12941         destination and source.
12942         (decide_alignment): Refactor, handle vector_loop.
12943         (ix86_expand_movmem): Likewise.
12944         (ix86_expand_setmem): Likewise.
12945         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12947 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
12949         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
12950         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
12952 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12954         * config/i386/sse.md (sse_movlhps): Change alternative 3
12955         of operand 2 to "m".
12957 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
12959         PR target/57807
12960         * config/i386/sse.md (iptr): New mode attribute.
12961         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
12962         (<sse>_vm<plusminus_insn><mode>3): Ditto.
12963         (<sse>_vmmul<mode>3): Ditto.
12964         (<sse>_vmdiv<mode>3): Ditto.
12965         (sse_vmrcpv4sf2): Ditto.
12966         (<sse>_vmsqrt<mode>2): Ditto.
12967         (sse_vmrsqrtv4sf2): Ditto.
12968         (<sse>_vm<code><mode>3): Ditto.
12969         (avx_vmcmp<mode>3): Ditto.
12970         (<sse>_vmmaskcmp<mode>3): Ditto.
12971         (<sse>_comi): Ditto.
12972         (<sse>_ucomi): Ditto.
12973         (*xop_vmfrcz_<mode>): Ditto.
12974         (*fmai_fmadd_<mode>): Ditto.
12975         (*fmai_fmsub_<mode>): Ditto.
12976         (*fmai_fnmadd_<mode>): Ditto.
12977         (*fmai_fnmsub_<mode>): Ditto.
12978         (*fma4i_vmfmadd_<mode>): Ditto.
12979         (*fma4i_vmfmsub_<mode>): Ditto.
12980         (*fma4i_vmfnmadd_<mode>): Ditto.
12981         (*fma4i_vmfnmsub_<mode>): Ditto.
12982         (*xop_vmfrcz_<mode>): Ditto.
12983         (sse_cvtps2pi): Ditto.
12984         (sse_cvttps2pi): Ditto.
12985         (sse_cvtss2si): Ditto.
12986         (sse_cvtss2si_2): Ditto.
12987         (sse_cvtss2siq_2): Ditto.
12988         (sse_cvttss2si): Ditto.
12989         (sse_cvttss2siq): Ditto.
12990         (sse_cvtsd2si): Ditto.
12991         (sse_cvtsd2si_2): Ditto.
12992         (sse_cvtsd2siq_2): Ditto.
12993         (sse_cvttsd2si): Ditto.
12994         (sse_cvttsd2siq): Ditto.
12995         (sse_cvtsd2ss): Ditto.
12996         (sse_cvtss2sd): Ditto.
12997         (avx2_pbroadcast<mode>): Ditto.
12998         (avx2_pbroadcast<mode>_1): Ditto.
12999         (*avx_vperm_broadcast_v4sf): Ditto.
13001         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13002         (sse_movlhps): Ditto.
13003         (sse_storehps): Ditto.
13004         (sse_loadhps): Ditto.
13005         (sse_storelps): Ditto.
13006         (sse_loadlps): Ditto.
13007         (*vec_concatv4sf): Ditto.
13008         (*vec_interleave_highv2df): Ditto.
13009         (*vec_interleave_lowv2df): Ditto.
13010         (*vec_extractv2df_1_sse): Ditto.
13011         (*vec_extractv2df_0_sse): Ditto.
13012         (sse2_storelpd): Ditto.
13013         (sse2_loadlpd): Ditto.
13014         (sse2_movsd): Ditto.
13015         (*vec_concatv4si): Ditto.
13016         (vec_concatv2di): Ditto.
13018         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13019         for Intel asm dialect.
13020         (mmx_punpcklwd): Ditto.
13021         (mmx_punpckldq): Ditto.
13023         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13024         for intel assembler dialect.
13026 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13028         PR target/29776
13029         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13030         for BUILT_IN_C{LZ,LRSB}*.
13031         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13032         * tree-vrp.c (extract_range_basic): Handle
13033         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13034         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13035         fall thru to code calling set_value*.
13036         * builtins.c (expand_builtin): Remove *IMAX cases.
13037         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13038         if width is bigger than 2*HWI.
13040 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13042         PR rtl-optimization/55342
13043         * lra-int.h (lra_subreg_reload_pseudos): New.
13044         * lra.c: Add undoing optional reloads to the block diagram.
13045         (lra_subreg_reload_pseudos): New.
13046         (lra_optional_reload_pseudos): Change comments.
13047         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13048         lra_optional_reload_pseudos after undo transformations.
13049         * lra-assigns.c (pseudo_prefix_title): New.
13050         (lra_setup_reg_renumber): Use it.
13051         (spill_for): Ditto.  Check subreg reload pseudos too.
13052         (assign_by_spills): Consider subreg reload pseudos too.
13053         * lra-constraints.c (simplify_operand_subreg): Use
13054         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13055         (curr_insn_transform): Recognize and do optional reloads.
13056         (undo_optional_reloads): New.
13057         (lra_undo_inheritance): Call undo_optional_reloads.
13059 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13061         * tree-complex.c (expand_complex_operations_1): Fix typo.
13063 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13065         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13066         (tune_params): New member 'const vec_costs'.
13067         * config/aarch64/aarch64.c (generic_vector_cost): New.
13068         (generic_tunings): New member 'generic_vector_cost'.
13069         (aarch64_builtin_vectorization_cost): New.
13070         (aarch64_add_stmt_cost): New.
13071         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13072         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13074 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13076         PR target/57777
13077         * config/i386/predicates.md (vsib_address_operand): Disallow
13078         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13080 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13082         PR middle-end/55030
13083         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13084         expand_builtin_setjmp_receiver.
13085         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13086         with NULL for the label parameter.
13087         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13088         the frame-pointer.  Adjust comments.
13089         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13090         only if LABEL is non-NULL.
13092 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13094         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13095         (ARM_ABI_AAPCS64): Ditto.
13096         (arm_abi): Ditto.
13097         (ARM_DEFAULT_ABI): Ditto.
13099 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13101         * config/aarch64/aarch64-builtins.c
13102         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13103         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13104         (st1): Likewise.
13105         * config/aarch64/aarch64-simd.md
13106         (aarch64_ld1<VALL:mode>): New.
13107         (aarch64_st1<VALL:mode>): Likewise.
13108         * config/aarch64/arm_neon.h
13109         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13111 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13113         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13114         target ISA is AVX.
13115         (ix86_option_override_internal):Turn on all -mavx target flags by
13116         default as they are dependent on AVX anyway.
13118 2013-07-02  Cary Coutant  <ccoutant@google.com>
13120         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13121         deterministic hash.
13122         (loc_checksum_ordered): Likewise.
13123         (hash_loc_operands): Remove inline keyword.
13125 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13127         PR tree-optimization/57741
13128         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13129         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13130         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13131         Allow REAL_CST step_exprs if flag_associative_math.
13132         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13134 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13136         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13138 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13140         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13142 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13144         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13145         encoding.
13146         (iorsi3_insn): Likewise.
13147         (arm_xorsi3): Likewise.
13149 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13151         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13152         to ...
13153         (attribute "type"): ... here, and prefix with "wmmx_".
13154         (attribute "core_cycles"): Update for attribute changes.
13155         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13156         (tbcstv4hi): Likewise.
13157         (tbcstv2si): Likewise.
13158         (iwmmxt_iordi3): Likewise.
13159         (iwmmxt_xordi3): Likewise.
13160         (iwmmxt_anddi3): Likewise.
13161         (iwmmxt_nanddi3): Likewise.
13162         (iwmmxt_arm_movdi): Likewise.
13163         (iwmmxt_movsi_insn): Likewise.
13164         (mov<mode>_internal): Likewise.
13165         (and<mode>3_iwmmxt): Likewise.
13166         (ior<mode>3_iwmmxt): Likewise.
13167         (xor<mode>3_iwmmxt): Likewise.
13168         (add<mode>3_iwmmxt): Likewise.
13169         (ssaddv8qi3): Likewise.
13170         (ssaddv4hi3): Likewise.
13171         (ssaddv2si3): Likewise.
13172         (usaddv8qi3): Likewise.
13173         (usaddv4hi3): Likewise.
13174         (usaddv2si3): Likewise.
13175         (sub<mode>3_iwmmxt): Likewise.
13176         (sssubv8qi3): Likewise.
13177         (sssubv4hi3): Likewise.
13178         (sssubv2si3): Likewise.
13179         (ussubv8qi3): Likewise.
13180         (ussubv4hi3): Likewise.
13181         (ussubv2si3): Likewise.
13182         (mulv4hi3_iwmmxt): Likewise.
13183         (smulv4hi3_highpart): Likewise.
13184         (umulv4hi3_highpart): Likewise.
13185         (iwmmxt_wmacs): Likewise.
13186         (iwmmxt_wmacsz): Likewise.
13187         (iwmmxt_wmacu): Likewise.
13188         (iwmmxt_wmacuz): Likewise.
13189         (iwmmxt_clrdi): Likewise.
13190         (iwmmxt_clrv8qi): Likewise.
13191         (iwmmxt_clr4hi): Likewise.
13192         (iwmmxt_clr2si): Likewise.
13193         (iwmmxt_uavgrndv8qi3): Likewise.
13194         (iwmmxt_uavgrndv4hi3): Likewise.
13195         (iwmmxt_uavgv8qi3): Likewise.
13196         (iwmmxt_uavgv4hi3): Likewise.
13197         (iwmmxt_tinsrb): Likewise.
13198         (iwmmxt_tinsrh): Likewise.
13199         (iwmmxt_tinsrw): Likewise.
13200         (iwmmxt_textrmub): Likewise.
13201         (iwmmxt_textrmsb): Likewise.
13202         (iwmmxt_textrmuh): Likewise.
13203         (iwmmxt_textrmsh): Likewise.
13204         (iwmmxt_textrmw): Likewise.
13205         (iwmxxt_wshufh): Likewise.
13206         (eqv8qi3): Likewise.
13207         (eqv4hi3): Likewise.
13208         (eqv2si3): Likewise.
13209         (gtuv8qi3): Likewise.
13210         (gtuv4hi3): Likewise.
13211         (gtuv2si3): Likewise.
13212         (gtv8qi3): Likewise.
13213         (gtv4hi3): Likewise.
13214         (gtv2si3): Likewise.
13215         (smax<mode>3_iwmmxt): Likewise.
13216         (umax<mode>3_iwmmxt): Likewise.
13217         (smin<mode>3_iwmmxt): Likewise.
13218         (umin<mode>3_iwmmxt): Likewise.
13219         (iwmmxt_wpackhss): Likewise.
13220         (iwmmxt_wpackwss): Likewise.
13221         (iwmmxt_wpackdss): Likewise.
13222         (iwmmxt_wpackhus): Likewise.
13223         (iwmmxt_wpackwus): Likewise.
13224         (iwmmxt_wpackdus): Likewise.
13225         (iwmmxt_wunpckihb): Likewise.
13226         (iwmmxt_wunpckihh): Likewise.
13227         (iwmmxt_wunpckihw): Likewise.
13228         (iwmmxt_wunpckilb): Likewise.
13229         (iwmmxt_wunpckilh): Likewise.
13230         (iwmmxt_wunpckilw): Likewise.
13231         (iwmmxt_wunpckehub): Likewise.
13232         (iwmmxt_wunpckehuh): Likewise.
13233         (iwmmxt_wunpckehuw): Likewise.
13234         (iwmmxt_wunpckehsb): Likewise.
13235         (iwmmxt_wunpckehsh): Likewise.
13236         (iwmmxt_wunpckehsw): Likewise.
13237         (iwmmxt_wunpckelub): Likewise.
13238         (iwmmxt_wunpckeluh): Likewise.
13239         (iwmmxt_wunpckeluw): Likewise.
13240         (iwmmxt_wunpckelsb): Likewise.
13241         (iwmmxt_wunpckelsh): Likewise.
13242         (iwmmxt_wunpckelsw): Likewise.
13243         (ror<mode>3): Likewise.
13244         (ashr<mode>3_iwmmxt): Likewise.
13245         (lshr<mode>3_iwmmxt): Likewise.
13246         (ashl<mode>3_iwmmxt): Likewise.
13247         (ror<mode>3_di): Likewise.
13248         (ashr<mode>3_di): Likewise.
13249         (lshr<mode>3_di): Likewise.
13250         (ashl<mode>3_di): Likewise.
13251         (iwmmxt_wmadds): Likewise.
13252         (iwmmxt_wmaddu): Likewise.
13253         (iwmmxt_tmia): Likewise.
13254         (iwmmxt_tmiaph): Likewise.
13255         (iwmmxt_tmiabb): Likewise.
13256         (iwmmxt_tmiatb): Likewise.
13257         (iwmmxt_tmiabt): Likewise.
13258         (iwmmxt_tmiatt): Likewise.
13259         (iwmmxt_tmovmskb): Likewise.
13260         (iwmmxt_tmovmskh): Likewise.
13261         (iwmmxt_tmovmskw): Likewise.
13262         (iwmmxt_waccb): Likewise.
13263         (iwmmxt_wacch): Likewise.
13264         (iwmmxt_waccw): Likewise.
13265         (iwmmxt_waligni): Likewise.
13266         (iwmmxt_walignr): Likewise.
13267         (iwmmxt_walignr0): Likewise.
13268         (iwmmxt_walignr1): Likewise.
13269         (iwmmxt_walignr2): Likewise.
13270         (iwmmxt_walignr3): Likewise.
13271         (iwmmxt_wsadb): Likewise.
13272         (iwmmxt_wsadh): Likewise.
13273         (iwmmxt_wsadbz): Likewise.
13274         (iwmmxt_wsadhz): Likewise.
13275         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13276         (iwmmxt_wabsdiffb): Likewise.
13277         (iwmmxt_wabsdiffh): Likewise.
13278         (iwmmxt_wabsdiffw): Likewise.
13279         (iwmmxt_waddsubhx): Likewise
13280         (iwmmxt_wsubaddhx): Likewise.
13281         (addc<mode>3): Likewise.
13282         (iwmmxt_avg4): Likewise.
13283         (iwmmxt_avg4r): Likewise.
13284         (iwmmxt_wmaddsx): Likewise.
13285         (iwmmxt_wmaddux): Likewise.
13286         (iwmmxt_wmaddsn): Likewise.
13287         (iwmmxt_wmaddun): Likewise.
13288         (iwmmxt_wmulwsm): Likewise.
13289         (iwmmxt_wmulwum): Likewise.
13290         (iwmmxt_wmulsmr): Likewise.
13291         (iwmmxt_wmulumr): Likewise.
13292         (iwmmxt_wmulwsmr): Likewise.
13293         (iwmmxt_wmulwumr): Likewise.
13294         (iwmmxt_wmulwl): Likewise.
13295         (iwmmxt_wqmulm): Likewise.
13296         (iwmmxt_wqmulwm): Likewise.
13297         (iwmmxt_wqmulmr): Likewise.
13298         (iwmmxt_wqmulwmr): Likewise.
13299         (iwmmxt_waddbhusm): Likewise.
13300         (iwmmxt_waddbhusl): Likewise.
13301         (iwmmxt_wqmiabb): Likewise.
13302         (iwmmxt_wqmiabt): Likewise.
13303         (iwmmxt_wqmiatb): Likewise.
13304         (iwmmxt_wqmiatt): Likewise.
13305         (iwmmxt_wqmiabbn): Likewise.
13306         (iwmmxt_wqmiabtn): Likewise.
13307         (iwmmxt_wqmiatbn): Likewise.
13308         (iwmmxt_wqmiattn): Likewise.
13309         (iwmmxt_wmiabb): Likewise.
13310         (iwmmxt_wmiabt): Likewise.
13311         (iwmmxt_wmiatb): Likewise.
13312         (iwmmxt_wmiatt): Likewise.
13313         (iwmmxt_wmiabbn): Likewise.
13314         (iwmmxt_wmiabtn): Likewise.
13315         (iwmmxt_wmiatbn): Likewise.
13316         (iwmmxt_wmiattn): Likewise.
13317         (iwmmxt_wmiawbb): Likewise.
13318         (iwmmxt_wmiawbt): Likewise.
13319         (iwmmxt_wmiawtb): Likewise.
13320         (iwmmxt_wmiawtt): Likewise.
13321         (iwmmxt_wmiawbbn): Likewise.
13322         (iwmmxt_wmiawbtn): Likewise.
13323         (iwmmxt_wmiawtbn): Likewise.
13324         (iwmmxt_wmiawttn): Likewise.
13325         (iwmmxt_wmerge): Likewise.
13326         (iwmmxt_tandc<mode>3): Likewise.
13327         (iwmmxt_torc<mode>3): Likewise.
13328         (iwmmxt_torvsc<mode>3): Likewise.
13329         (iwmmxt_textrc<mode>3): Likewise.
13330         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13331         (wmmxt_pack): Likewise.
13332         (wmmxt_mult_c1): Likewise.
13333         (wmmxt_mult_c2): Likewise.
13334         (wmmxt_alu_c1): Likewise.
13335         (wmmxt_alu_c2): Likewise.
13336         (wmmxt_alu_c3): Likewise.
13337         (wmmxt_transfer_c1): Likewise.
13338         (wmmxt_transfer_c2): Likewise.
13339         (wmmxt_transfer_c3): Likewise.
13340         (marvell_f_iwmmxt_wstr): Likewise.
13341         (marvell_f_iwmmxt_wldr): Likewise.
13343 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13345         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13347 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13349         Revert:
13350         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13351         * lra-constraints.c (need_for_split_p): Check call used hard regs
13352         living through calls.
13354         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13355         call used regs for call insn.
13357 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13359         PR target/57736
13360         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13361         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13363 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13365         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13367 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13369         * lra-constraints.c (need_for_split_p): Check call used hard regs
13370         living through calls.
13372 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13374         PR target/57744
13375         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13376         to tie with any other modes.  Eliminate Altivec vector mode tests,
13377         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13378         code, to return 0 if testing MODE2 for a condition, if we've
13379         already tested MODE1 for the same condition.
13381 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13383         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13384         layout.
13386 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13388         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13389         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13391 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13393         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13394         Define.
13395         (aarch64_symbolic_constant_p): Remove.
13396         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13397         static.  Fix line length and white space.
13398         (aarch64_symbolic_constant_p): Remove.
13399         * config/aarch64/predicates.md (aarch64_valid_symref):
13400         Use aarch64_classify_symbol_expression.
13402 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13404         * config/arm/constraints.md (Ts): New constraint.
13405         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13406         16-bit encodings.
13407         (compare_scc): Use "Ts" constraint for operand 0.
13408         (ior_scc_scc): Likewise.
13409         (and_scc_scc): Likewise.
13410         (and_scc_scc_nodom): Likewise.
13411         (ior_scc_scc_cmp): Likewise for operand 7.
13412         (and_scc_scc_cmp): Likewise.
13413         * config/arm/thumb2.md (thumb2_movsi_insn):
13414         Add alternatives for 16-bit encodings.
13415         (thumb2_movhi_insn): Likewise.
13416         (thumb2_movsicc_insn): Likewise.
13417         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13418         (thumb2_negscc): Use "Ts" constraint.
13419         Move mvn instruction outside cond_exec block.
13420         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13421         for 16-bit encodings.
13423 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13425         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13426         encoding.
13427         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13428         (mulsi3subsi): Likewise.
13429         (mulsidi3adddi): Likewise.
13430         (mulsidi3_v6): Likewise.
13431         (umulsidi3_v6): Likewise.
13432         (umulsidi3adddi_v6): Likewise.
13433         (smulsi3_highpart_v6): Likewise.
13434         (umulsi3_highpart_v6): Likewise.
13435         (mulhisi3tb): Likewise.
13436         (mulhisi3bt): Likewise.
13437         (mulhisi3tt): Likewise.
13438         (maddhisi4): Likewise.
13439         (maddhisi4tb): Likewise.
13440         (maddhisi4tt): Likewise.
13441         (maddhidi4): Likewise.
13442         (maddhidi4tb): Likewise.
13443         (maddhidi4tt): Likewise.
13444         (zeroextractsi_compare0_scratch): Likewise.
13445         (insv_zero): Likewise.
13446         (insv_t2): Likewise.
13447         (anddi_notzesidi_di): Likewise.
13448         (anddi_notsesidi_di): Likewise.
13449         (andsi_notsi_si): Likewise.
13450         (iordi_zesidi_di): Likewise.
13451         (xordi_zesidi_di): Likewise.
13452         (andsi_iorsi3_notsi): Likewise.
13453         (smax_0): Likewise.
13454         (smax_m1): Likewise.
13455         (smin_0): Likewise.
13456         (not_shiftsi): Likewise.
13457         (unaligned_loadsi): Likewise.
13458         (unaligned_loadhis): Likewise.
13459         (unaligned_loadhiu): Likewise.
13460         (unaligned_storesi): Likewise.
13461         (unaligned_storehi): Likewise.
13462         (extv_reg): Likewise.
13463         (extzv_t2): Likewise.
13464         (divsi3): Likewise.
13465         (udivsi3): Likewise.
13466         (arm_zero_extendhisi2addsi): Likewise.
13467         (arm_zero_extendqisi2addsi): Likewise.
13468         (compareqi_eq0): Likewise.
13469         (arm_extendhisi2_v6): Likewise.
13470         (arm_extendqisi2addsi): Likewise.
13471         (arm_movt): Likewise.
13472         (thumb2_ldrd): Likewise.
13473         (thumb2_ldrd_base): Likewise.
13474         (thumb2_ldrd_base_neg): Likewise.
13475         (thumb2_strd): Likewise.
13476         (thumb2_strd_base): Likewise.
13477         (thumb2_strd_base_neg): Likewise.
13478         (arm_negsi2): Add alternative for 16-bit encoding.
13479         (arm_one_cmplsi2): Likewise.
13481 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13483         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13484         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13485         (movdfcc): Likewise.
13486         * config/arm/vfp.md (*thumb2_movsf_vfp):
13487         Disable predication for arm_restrict_it.
13488         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13489         (*thumb2_movdfcc_vfp): Likewise.
13490         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13491         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13492         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13493         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13494         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13495         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13496         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13497         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13498         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13499         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13500         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13501         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13502         Disable predication for arm_restrict_it.
13504 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13506         * config/i386/bmiintrin.h (_bextr_u32): New.
13507         (_bextr_u64): Ditto.
13509 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13511         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13512         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13513         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13514         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13515         * config/mips/n32-elf.h: ...this new file.
13517 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13519         PR target/57224
13520         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13521         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13523 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13525         * config/mips/mips-tables.opt: Regenerate.
13526         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13527         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13528         * doc/invoke.texi: Add -m14kc.
13530 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13532         PR target/57623
13533         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13534         constraints of operand 1 and 2.
13536         PR target/57623
13537         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13538         to match RTL canonicalization.  Swap predicates and
13539         constraints of operand 1 and 2.
13541 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13543         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13544         Process OP_INOUT regs for splitting too.
13546 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13548         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13549         decl before the loop, initialize to NULL.
13550         (vectorizable_load): Initialize ptr_incr to NULL.
13552 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13554         PR lto/57208
13555         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13556         * ipa-ref.c (ipa_maybe_record_reference): New function.
13557         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13558         * ipa-cp.c (create_specialized_node): Record potential references from
13559         aggvals.
13560         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13562 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13564         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13565         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13566         'mode' to force_reg.
13567         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13568         (aarch64_expand_mov_immediate): Likewise.
13570 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13572         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13573         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13575 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13577         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13578         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13579         (struct machine_function): Add tbegin_p.
13580         (s390_canonicalize_comparison): Fold CC mode compares to
13581         conditional jump if possible.
13582         (s390_emit_jump): Return the emitted jump.
13583         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13584         Handle CCRAWmode compares.
13585         (s390_option_override): Default to -mhtm if available.
13586         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13587         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13588         FPRs instead of df_regs_ever_live_p.
13589         (s390_optimize_nonescaping_tx): New function.
13590         (s390_init_frame_layout): Extend clobbered_regs array to cover
13591         FPRs as well.
13592         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13593         (s390_expand_tbegin): New function.
13594         (enum s390_builtin): New enum definition.
13595         (code_for_builtin): New array definition.
13596         (s390_init_builtins): New function.
13597         (s390_expand_builtin): New function.
13598         (TARGET_INIT_BUILTINS): Define.
13599         (TARGET_EXPAND_BUILTIN): Define.
13600         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13601         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13602         (s390_alc_comparison): Likewise.
13603         * config/s390/s390-modes.def: Add CCRAWmode.
13604         * config/s390/s390.h (processor_flags): Add PF_TX.
13605         (TARGET_CPU_HTM): Define macro.
13606         (TARGET_HTM): Define macro.
13607         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13608         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13609         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13610         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13611         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13612         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13613         ("*cc_to_int"): Move up.
13614         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13615         constants other than 0.
13616         ("*ccraw_to_int"): New insn and splitter definition.
13617         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13618         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13619         ("tx_assist"): New expander.
13620         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13621         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13622         * config/s390/s390.opt: Add -mhtm option.
13623         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13624         * config/s390/htmxlintrin.h: New file.
13625         * config/s390/htmintrin.h: New file.
13626         * config/s390/s390intrin.h: New file.
13627         * doc/extend.texi: Document htm builtins.
13628         * config.gcc: Add the new header files to extra_headers.
13630 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13632         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13633         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13635 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13636             Pat Haugen  <pthaugen@us.ibm.com>
13637             Peter Bergner  <bergner@vnet.ibm.com>
13639         * config/rs6000/power8.md: New.
13640         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13641         setting for power8 entry.
13642         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13643         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13644         test for Power4/Power5 only.
13645         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13646         support.
13647         (force_new_group): Adjust comment.
13648         * config/rs6000/rs6000.md: Include power8.md.
13650 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13652         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13653         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13654         declaration.
13655         (tune_params): Update comment.
13656         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13657         (arm_max_conditional_execute): New function.
13658         (thumb2_final_prescan_insn): Use max_insn_skipped and
13659         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13661 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13663         PR tree-optimization/57705
13664         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13665         SSA_NAME step, provided that it is not defined inside the loop.
13666         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13667         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13669 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13671         PR middle-end/57670
13672         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13673         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13674         calls in the dump.
13675         (ipa_note_param_call): Initialize member_ptr flag.
13676         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13677         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13678         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13679         (ipa_read_indirect_edge_info): Likewise.
13681 2013-06-25  Richard Biener  <rguenther@suse.de>
13683         PR middle-end/56977
13684         * passes.c (init_optimization_passes): Move pass_fold_builtins
13685         and pass_dce earlier with -Og.
13687 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13689         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13690         <BIT_FIELD_REF>: Remove trailing TAB.
13691         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13692         remove blank line.
13694 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13696         PR tree-optimization/57358
13697         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13698         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13699         (ipa_analyze_params_uses): Generate pessimistic info when true.
13701 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13703         PR tree-optimization/57539
13704         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13705         global.inlined_to of the new node to it.  All callers changed.
13706         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13707         inlining_into, pass it to cgraph_clone_node.
13708         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13709         ipa_free_edge_args_substructures.
13710         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13711         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13712         Assert we have found an rdesc in the rdesc list.
13714 2013-06-24  Richard Biener  <rguenther@suse.de>
13716         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13717         (pointer_set_lookup): Declare.
13718         (class pointer_map): New template class implementing a
13719         generic pointer to T map.
13720         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13721         pointer_map<T>::contains, pointer_map<T>::insert,
13722         pointer_map<T>::traverse): New functions.
13723         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13724         (pointer_set_lookup): New function.
13725         (pointer_set_contains): Use pointer_set_lookup.
13726         (pointer_set_insert): Likewise.
13727         (insert_aux): Remove.
13728         (struct pointer_map_t): Embed a pointer_set_t.
13729         (pointer_map_create): Adjust.
13730         (pointer_map_destroy): Likewise.
13731         (pointer_map_contains): Likewise.
13732         (pointer_map_insert): Likewise.
13733         (pointer_map_traverse): Likewise.
13734         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13735         pointer_map<unsigned> instead of a pointer_map_t.
13736         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13737         (streamer_tree_cache_lookup): Likewise.
13738         (streamer_tree_cache_create): Likewise.
13739         (streamer_tree_cache_delete): Likewise.
13740         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13741         pointer_map<unsigned> instead of a pointer_map_t.
13742         (lto_init_tree_ref_encoder): Adjust.
13743         (lto_destroy_tree_ref_encoder): Likewise.
13744         * lto-section-out.c (lto_output_decl_index): Likewise.
13745         (lto_record_function_out_decl_state): Likewise.
13746         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13748 2013-06-24  Richard Biener  <rguenther@suse.de>
13750         PR tree-optimization/57488
13751         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13753 2013-06-24  Alan Modra  <amodra@gmail.com>
13755         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13756         (gen_easy_altivec_constant): Likewise.
13757         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13758         easy_vector_constant_msb): Likewise.
13760 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13762         PR target/57688
13763         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13764         add missing return true.
13766 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13768         PR target/52483
13769         * config/sh/predicates.md (general_extend_operand): Invoke
13770         general_movsrc_operand for memory operands.
13771         (general_movsrc_operand): Allow reg+reg addressing, do not use
13772         general_operand for memory operands.
13774 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13776         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13777         when current target options does not apply.
13778         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13779         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13780         * config/i386/bmiintrin.h: Pass appropriate target
13781         attributes to header.
13782         * config/i386/mmintrin.h: Ditto.
13783         * config/i386/nmmintrin.h: Ditto.
13784         * config/i386/avx2intrin.h: Ditto.
13785         * config/i386/fxsrintrin.h: Ditto.
13786         * config/i386/tbmintrin.h: Ditto.
13787         * config/i386/xsaveintrin.h: Ditto.
13788         * config/i386/f16cintrin.h: Ditto.
13789         * config/i386/xtestintrin.h: Ditto.
13790         * config/i386/xsaveoptintrin.h: Ditto.
13791         * config/i386/bmi2intrin.h: Ditto.
13792         * config/i386/lzcntintrin.h: Ditto.
13793         * config/i386/smmintrin.h: Ditto.
13794         * config/i386/wmmintrin.h: Ditto.
13795         * config/i386/x86intrin.h: Remove all header include guards.
13796         * config/i386/prfchwintrin.h: Ditto.
13797         * config/i386/pmmintrin.h: Ditto.
13798         * config/i386/tmmintrin.h: Ditto.
13799         * config/i386/xmmintrin.h: Ditto.
13800         * config/i386/popcntintrin.h: Ditto.
13801         * config/i386/rdseedintrin.h: Ditto.
13802         * config/i386/ammintrin.h: Ditto.
13803         * config/i386/emmintrin.h: Ditto.
13804         * config/i386/immintrin.h: Remove all header include guards.
13805         * config/i386/fma4intrin.h: Ditto.
13806         * config/i386/lwpintrin.h: Ditto.
13807         * config/i386/xopintrin.h: Ditto.
13808         * config/i386/ia32intrin.h: Ditto.
13809         * config/i386/avxintrin.h: Ditto.
13810         * config/i386/rtmintrin.h: Ditto.
13811         * config/i386/fmaintrin.h: Ditto.
13812         * config/i386/mm3dnow.h: Ditto.
13814 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13816         * common/config/i386/i386-common.c: Handle LZCNT.
13818 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13820         * doc/extend.texi: Use __atomic_store_n instead of
13821         __atomic_store in HLE example.
13823 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13825         * config/sh/sh.c: Remove <cstdlib> workaround.
13827 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13829         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13831 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13833         * doc/extend.texi: Document that __atomic_clear and
13834           __atomic_test_and_set should only be used with bool.
13836 2013-06-20  Jan Hubicka  <jh@suse.cz>
13838         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13839         types_same_for_odr.
13840         * tree.c (decls_same_for_odr): New function.
13841         (same_for_edr): New function.
13842         (types_same_for_odr): New function.
13843         (get_binfo_at_offset): Use it.
13844         * tree.h (types_same_for_odr): Declare.
13846 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13847             Jason Merrill  <jason@redhat.com>
13849         * system.h: Include <cstdlib> as well as <stdlib.h>.
13851 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13853         PR target/57655
13854         * config/i386/i386.c (construct_container): Report error if
13855         long double is used with disabled x87 float returns.
13857 2013-06-20  Jan Hubicka  <jh@suse.cz>
13859         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13861 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13863         PR rtl-optimization/57425
13864         PR rtl-optimization/57569
13865         * alias.c (write_dependence_p): Remove parameters mem_mode and
13866         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13867         Changed all callers.
13868         (canon_anti_dependence): Get comments and semantics in sync.
13869         Add parameter mem_canonicalized.  Changed all callers.
13870         * rtl.h (canon_anti_dependence): Update prototype.
13872 2013-06-20  Richard Biener  <rguenther@suse.de>
13874         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13875         byte case, inline streamer_read_uchar and defer section
13876         overrun check.
13878 2013-06-20  Richard Biener  <rguenther@suse.de>
13880         PR tree-optimization/57584
13881         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13882         SSA names into the expanded expression that take part in
13883         abnormal coalescing.
13885 2013-06-19  Sharad Singhai  <singhai@google.com>
13887         * gcov.c (print_usage): Handle new option.
13888         (process_args): Ditto.
13889         (get_gcov_intermediate_filename): New function.
13890         (output_intermediate_file): New function.
13891         (output_gcov_file): New function
13892         (generate_results): Handle new option.
13893         (release_function): Relase demangled name.
13894         (read_graph_file): Handle demangled name.
13895         (output_lines): Ditto.
13896         * doc/gcov.texi: Document gcov intermediate format.
13898 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13900         PR bootstrap/57604
13901         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13902         (lra_emit_add): Use the functions.  Add comment about Y as an
13903         address segment.
13905 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13907         PR driver/57652
13908         * collect2.c (collect_atexit): New.
13909         (collect_exit): Delete.
13910         (main): Register collect_atexit with atexit.
13911         (collect_wait): Change collect_exit to exit.
13912         (do_wait): Same.
13913         * collect2.h (collect_exit): Delete.
13914         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13916 2013-06-19  Wei Mi  <wmi@google.com>
13918         PR rtl-optimization/57518
13919         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13920         if regno is used in paradoxical subreg.
13921         (update_equiv_regs): Check pdx_subregs[regno] before
13922         set a reg to be equivalent with a mem.
13924 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13926         PR driver/57651
13927         * file-find.h (find_a_file): Add a mode parameter.
13928         * file-find.c (find_a_file): Likewise.
13929         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13930         with X_OK for the executables.
13931         * collect2.c (main): Call find_a_file with X_OK.
13933 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13935         PR target/56942
13936         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13937         Use NEXT_INSN instead of next_real_insn.
13939 2013-06-19  Jan Hubicka  <jh@suse.cz>
13941         * cgraph.h (const_value_known_p): Replace by ...
13942         (ctor_for_folding): .. this one.
13943         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13944         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13945         * expr.c (expand_expr_real_1): Likewise.
13946         (string_constant): Likewise.
13947         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
13948         * ipa.c (process_references): Likewise.
13949         (symtab_remove_unreachable_nodes): Likewise.
13950         * ipa-inline-analysis.c (param_change_prob): Likewise.
13951         * gimple-fold.c (canonicalize_constructor_val): Likewise.
13952         (get_base_constructor): Likwise.
13953         * varpool.c (varpool_remove_node): Likewise.
13954         (varpool_remove_initializer): LIkewise.
13955         (dump_varpool_node): LIkwise.
13956         (const_value_known_p): Rewrite to ...
13957         (ctor_for_folding): ... this one.
13959 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
13961         PR driver/57651
13962         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
13963         PERSONALITY in $PATH derived prefixes.
13965 2013-06-19  Jeff Law  <law@redhat.com>
13967         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
13968         in comment.
13970         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
13971         (simplify_bitwise_binary): Use it to simpify certain binary ops on
13972         booleans.
13974 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
13976         * config/arm/vfp.md: Move VFP instruction classification documentation
13977         to ...
13978         * config/arm/arm.md: ... here.  Update instruction classification
13979         documentation.
13981 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
13983         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
13984         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
13985         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
13987 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
13989         PR target/57609
13990         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
13991         with NEXT_INSN.  Use tablejump_p to check for jump table data
13992         insns.
13994 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
13996         PR c++/56544
13997         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
13998         that now in C++ the value is correct per the C++ standards.
14000 2013-06-19  Richard Biener  <rguenther@suse.de>
14002         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14003         for global context.
14005 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14007         Revert:
14008         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14010         PR target/57609
14011         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14012         with next_active_insn.
14014 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14016         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14017         functions are inlined during failures to flag an error.
14018         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14019         in early inline pass.
14021 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14023         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14024         in comments.
14026 2013-06-18  Julian Brown  <julian@codesourcery.com>
14028         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14029         Permit virtual register pre-reload if !strict.
14030         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14031         change.
14032         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14033         prototype.
14034         * config/arm/neon.md (movmisalign<mode>): Use
14035         neon_perm_struct_or_reg_operand instead of
14036         neon_struct_or_register_operand.
14037         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14038         neon_permissive_struct_operand instead of neon_struct_operand.
14039         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14040         neon_vector_mem_operand.
14041         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14042         neon_vector_mem_operand.
14043         (neon_permissive_struct_operand): New.
14044         (neon_struct_or_register_operand): Rename to...
14045         (neon_perm_struct_or_reg_operand): This. Adjust call to
14046         neon_vector_mem_operand.
14048 2013-06-18  Richard Biener  <rguenther@suse.de>
14050         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14051         * lto-streamer.h: Include pointer-set.h.
14052         (struct lto_decl_slot): Remove.
14053         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14054         Remove next_index entry.
14055         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14056         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14057         (lto_init_tree_ref_encoder): Adjust.
14058         (lto_destroy_tree_ref_encoder): Likewise.
14059         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14060         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14061         (lto_output_decl_index): Adjust.
14062         (lto_new_out_decl_state): Likewise.
14063         (lto_record_function_out_decl_state): Likewise.
14064         * lto-streamer-out.c (copy_function): Likewise.
14066 2013-06-18  Richard Biener  <rguenther@suse.de>
14068         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14069         * cgraphunit.c: Include cfgloop.h.
14070         (init_lowered_empty_function): Initialize the loop tree.
14071         (assemble_thunk): Insert new BBs into loops.
14073 2013-06-18  Richard Biener  <rguenther@suse.de>
14075         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14076         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14077         the map from cache entry to cache index optional.
14078         (streamer_tree_cache_replace_tree): Adjust accordingly.
14079         (streamer_tree_cache_append): Likewise.
14080         (streamer_tree_cache_delete): Likewise.
14081         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14082         streamer cache map from cache entry to cache index.
14083         * lto-streamer-out.c (create_output_block): Adjust.
14085 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14087         * config/arm/arm.md (attribute "insn"): Move multiplication and
14088         division attributes to...
14089         (attribute "type"): ... here.  Remove mult.
14090         (attribute "mul32"): New attribute.
14091         (attribute "mul64"): Add umaal.
14092         (*arm_mulsi3): Update attributes.
14093         (*arm_mulsi3_v6): Likewise.
14094         (*thumb_mulsi3): Likewise.
14095         (*thumb_mulsi3_v6): Likewise.
14096         (*mulsi3_compare0): Likewise.
14097         (*mulsi3_compare0_v6): Likewise.
14098         (*mulsi_compare0_scratch): Likewise.
14099         (*mulsi_compare0_scratch_v6): Likewise.
14100         (*mulsi3addsi): Likewise.
14101         (*mulsi3addsi_v6): Likewise.
14102         (*mulsi3addsi_compare0): Likewise.
14103         (*mulsi3addsi_compare0_v6): Likewise.
14104         (*mulsi3addsi_compare0_scratch): Likewise.
14105         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14106         (*mulsi3subsi): Likewise.
14107         (*mulsidi3adddi): Likewise.
14108         (*mulsi3addsi_v6): Likewise.
14109         (*mulsidi3adddi_v6): Likewise.
14110         (*mulsidi3_nov6): Likewise.
14111         (*mulsidi3_v6): Likewise.
14112         (*umulsidi3_nov6): Likewise.
14113         (*umulsidi3_v6): Likewise.
14114         (*umulsidi3adddi): Likewise.
14115         (*umulsidi3adddi_v6): Likewise.
14116         (*smulsi3_highpart_nov6): Likewise.
14117         (*smulsi3_highpart_v6): Likewise.
14118         (*umulsi3_highpart_nov6): Likewise.
14119         (*umulsi3_highpart_v6): Likewise.
14120         (mulhisi3): Likewise.
14121         (*mulhisi3tb): Likewise.
14122         (*mulhisi3bt): Likewise.
14123         (*mulhisi3tt): Likewise.
14124         (maddhisi4): Likewise.
14125         (*maddhisi4tb): Likewise.
14126         (*maddhisi4tt): Likewise.
14127         (maddhidi4): Likewise.
14128         (*maddhidi4tb): Likewise.
14129         (*maddhidi4tt): Likewise.
14130         (divsi3): Likewise.
14131         (udivsi3): Likewise.
14132         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14133         (thumb2_mulsi_short_compare0): Likewise.
14134         (thumb2_mulsi_short_compare0_scratch): Likewise.
14135         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14136         (1020mult2): Likewise.
14137         (1020mult3): Likewise.
14138         (1020mult4): Likewise.
14139         (1020mult5): Likewise.
14140         (1020mult6): Likewise.
14141         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14142         change.
14143         (cortex_a15_mult64): Likewise.
14144         (cortex_a15_sdiv): Likewise.
14145         (cortex_a15_udiv): Likewise.
14146         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14147         (mult2): Likewise.
14148         (mult3): Likewise.
14149         (mult4): Likewise.
14150         (mult5): Likewise.
14151         (mult6): Likewise.
14152         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14153         (pj4_ir_div): Likewise.
14154         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14155         (11_mult2): Likewise.
14156         (11_mult3): Likewise.
14157         (11_mult4): Likewise.
14158         (11_mult5): Likewise.
14159         (11_mult6): Likewise.
14160         (11_mult7): Likewise.
14161         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14162         (cortex_a8_mla): Likewise.
14163         (cortex_a8_mull): Likewise.
14164         (cortex_a8_smulwy): Likewise.
14165         (cortex_a8_smlald): Likewise.
14166         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14167         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14168         (cortex_r4_mul_3): Likewise.
14169         (cortex_r4_mla_4): Likewise.
14170         (cortex_r4_mla_3): Likewise.
14171         (cortex_r4_smlald): Likewise.
14172         (cortex_r4_mull): Likewise.
14173         (cortex_r4_sdiv): Likewise.
14174         (cortex_r4_udiv): Likewise.
14175         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14176         (cortex_a7_idiv): Likewise.
14177         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14178         (9_mult2): Likewise.
14179         (9_mult3): Likewise.
14180         (9_mult4): Likewise.
14181         (9_mult5): Likewise.
14182         (9_mult6): Likewise.
14183         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14184         (cortex_a53_sdiv): Likewise.
14185         (cortex_a53_udiv): Likewise.
14186         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14187         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14188         (mp626_mult2): Likewise.
14189         (mp626_mult3): Likewise.
14190         (mp626_mult4): Likewise.
14191         * config/arm/fa526.md (526_mult1): Update attribute change.
14192         (526_mult2): Likewise.
14193         * config/arm/arm-generic.md (mult): Update attribute change.
14194         (mult_ldsched_strongarm): Likewise.
14195         (mult_ldsched): Likewise.
14196         (multi_cycle): Likewise.
14197         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14198         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14199         (606te_mult2): Likewise.
14200         (606te_mult3): Likewise.
14201         (606te_mult4): Likewise.
14202         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14203         (cortex_a9_mac16): Likewise.
14204         (cortex_a9_multiply): Likewise.
14205         (cortex_a9_mac): Likewise.
14206         (cortex_a9_multiply_long): Likewise.
14207         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14208         (626te_mult2): Likewise.
14209         (626te_mult3): Likewise.
14210         (626te_mult4): Likewise.
14212 2013-06-18  Richard Biener  <rguenther@suse.de>
14214         PR lto/57334
14215         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14217 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14219         PR target/57609
14220         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14221         with next_active_insn.
14223 2013-06-18  Alan Modra  <amodra@gmail.com>
14225         * config/rs6000/rs6000.h (enum data_align): New.
14226         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14227         (DATA_ABI_ALIGNMENT): Define.
14228         (CONSTANT_ALIGNMENT): Correct comment.
14229         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14230         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14232 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14234         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14235         ATTRIBUTE_UNUSED marking.
14237 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14239         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14240         alternative and update.
14241         (aarch64_dup_lanedi): Delete.
14242         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14243         * config/aarch64/aarch64-simd-builtins.def: Update.
14245 2013-06-17  Richard Biener  <rguenther@suse.de>
14247         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14248         (lto_input_scc): Declare.
14249         (lto_input_tree_1): Likewise.
14250         (struct lto_stats_d): Add num_tree_bodies_output and
14251         num_pickle_refs_output.
14252         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14253         (lto_read_tree_1): Split out from ...
14254         (lto_read_tree): ... this.
14255         (lto_input_scc): New function.
14256         (lto_input_tree_1): Split out from ...
14257         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14258         (lto_data_in_create): Create the streamer cache without hashes.
14259         * lto-streamer-out.c (create_output_block): Create the streamer
14260         cache with hashes when not doing WPA.
14261         (lto_write_tree_1): Split out from ...
14262         (lto_write_tree): ... this.
14263         (get_symbol_initial_value): New function.
14264         (lto_output_tree_1): Split out from ...
14265         (lto_output_tree): ... this.  Write trees as series of SCCs
14266         using a DFS walk via DFS_write_tree.
14267         (struct sccs, struct scc_entry): New types.
14268         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14269         (DFS_write_tree_body): New function.
14270         (DFS_write_tree): Likewise.
14271         (hash_tree): Likewise.
14272         (scc_entry_compare): Likewise.
14273         (hash_scc): Likewise.
14274         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14275         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14276         TREE_CHAIN as regular reference.
14277         (streamer_read_integer_cst): Remove.
14278         (streamer_get_pickled_tree): Adjust.
14279         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14280         of DECL_EXTERNALs in BLOCK_VARS for now.
14281         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14282         reference.
14283         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14284         Add hash value argument and record that if hashes are recorded
14285         in the cache.
14286         (streamer_tree_cache_insert_1): Adjust.
14287         (streamer_tree_cache_insert): Likewise.
14288         (streamer_tree_cache_insert_at): Rename to ...
14289         (streamer_tree_cache_replace_tree): ... this and adjust.
14290         (streamer_tree_cache_append): Adjust.
14291         (record_common_node): Likewise.
14292         (streamer_tree_cache_create): Add argument whether to
14293         record hash values together with trees.
14294         (streamer_tree_cache_delete): Adjust.
14295         * tree-streamer.h (struct streamer_tree_cache_d): Add
14296         vector of hashes.
14297         (streamer_read_integer_cst): Remove.
14298         (streamer_tree_cache_insert): Adjust.
14299         (streamer_tree_cache_append): Likewise.
14300         (streamer_tree_cache_insert_at): Rename to ...
14301         (streamer_tree_cache_replace_tree): ... this and adjust.
14302         (streamer_tree_cache_create): Add argument whether to record hashes.
14303         (streamer_tree_cache_get): Rename to ...
14304         (streamer_tree_cache_get_tree): ... this.
14305         (streamer_tree_cache_get_hash): New function.
14306         * tree.c (cache_integer_cst): New function.
14307         * tree.h (cache_integer_cst): Declare.
14308         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14309         * lto-symtab.c (lto_varpool_replace_node): Only release
14310         DECL_INITIAL of non-prevailing decls.
14311         * varpool.c (varpool_remove_initializer): Do not release
14312         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14314 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14316         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14317         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14318         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14319         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14320         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14321         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14322         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14323         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14324         instead of TARGET_64BIT.
14325         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14326         Require ISA_HAS_<D>DIV.
14328 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14330         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14331         (mips*-*-linux*): Move default with_llsc setting to where other
14332         defaults are set.
14333         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14334         with_arch block.
14335         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14336         Likewise.  Remove default with_tune setting.  Move default float
14337         setting to its own block.  Handle with_llsc in the same block as above.
14339 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14341         PR rtl-optimization/57425
14342         PR rtl-optimization/57569
14343         * alias.c (write_dependence_p): Add new parameters mem_mode,
14344         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14345         Changed all callers.
14346         (canon_anti_dependence): New function.
14347         * cse.c (check_dependence): Use canon_anti_dependence.
14348         * cselib.c (cselib_invalidate_mem): Likewise.
14349         * rtl.h (canon_anti_dependence): Declare.
14351 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14353         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14354         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14355         ".set mips3" for 64-bit targets.
14357 2013-06-15  Dehao Chen  <dehao@google.com>
14359         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14360         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14361         (gimple_check_call_args): Likewise.
14362         * value-prof.c (check_ic_target): Likewise.
14363         * ipa-inline.c (early_inliner): Likewise.
14364         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14365         * cgraph.c (cgraph_create_edge_1): Likewise.
14366         (cgraph_make_edge_direct): Likewise.
14368 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14370         PR target/57615
14371         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14372         rs6000_output_move_128bit to handle emitting quad memory
14373         operations.  Set attribute length to 8 bytes.
14375 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14377         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14378         New pattern.
14379         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14380         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14381         (aarch64_<su>mlsl<mode>): Likewise.
14383 2013-06-14  Mike Stump  <mikestump@comcast.net>
14385         * Makefile.in (TARGET_H): Add insn-codes.h.
14387 2013-06-14  Alan Modra  <amodra@gmail.com>
14389         PR middle-end/57134
14390         PR middle-end/57586
14391         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14392         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14393         bitfield expansion when EXPAND_MEMORY.
14394         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14396 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14398         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14399         test for clearing quad memory on 32-bit later.
14401 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14403         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14404         (fold_negate_expr): Likewise.
14405         (fold_real_zero_addition_p): Handle vectors.
14406         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14408 2013-06-14  Alan Modra  <amodra@gmail.com>
14410         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14412 2013-06-13  Jan Hubicka  <jh@suse.cz>
14414         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14415         Local comdats are not externally visible.
14416         * symtab.c (dump_symtab_base): Dump externally visible.
14417         (verify_symtab_base): Verify back links in the symtab hash.
14419 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14421         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14422         CONVERT_EXPR as equal nodes.
14424 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14426         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14428 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14430         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14431         Generalize to complex and vector.
14432         * tree.c (build_all_ones_cst): New function.
14433         * tree.h (build_all_ones_cst): Declare it.
14435 2013-06-13  Alan Modra  <amodra@gmail.com>
14437         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14438         * config/rs6000/rs6000.md (signbittf2): New insn.
14439         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14440         (abstf2_internal, cmptf_internal2): Likewise.
14441         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14443 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14444             Pat Haugen  <pthaugen@us.ibm.com>
14445             Peter Bergner  <bergner@vnet.ibm.com>
14447         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14448         power8 byte, half-word, and quad-word atomic instructions.
14449         (emit_store_conditional): Likewise.
14450         (rs6000_expand_atomic_compare_and_swap): Likewise.
14451         (rs6000_expand_atomic_op): Likewise.
14453         * config/rs6000/sync.md (larx): Add new modes for power8.
14454         (stcx): Likewise.
14455         (AINT): New mode iterator to include TImode as well as normal
14456         integer modes on power8.
14457         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14458         that VSX registers are not considered.  Use AINT mode iterator
14459         instead of INT1 to allow inclusion of quad word atomic operations
14460         on power8.
14461         (load_locked<mode>): Likewise.
14462         (store_conditional<mode>): Likewise.
14463         (atomic_compare_and_swap<mode>): Likewise.
14464         (atomic_exchange<mode>): Likewise.
14465         (atomic_nand<mode>): Likewise.
14466         (atomic_fetch_<fetchop_name><mode>): Likewise.
14467         (atomic_nand_fetch<mode>): Likewise.
14468         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14469         each type.
14470         (ATOMIC): On power8, add QImode, HImode modes.
14471         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14472         modes that promote to SImode.
14473         (load_lockedti): Convert TImode arguments to PTImode, so that we
14474         get a guaranteed even/odd register pair.
14475         (load_lockedpti): Likewise.
14476         (store_conditionalti): Likewise.
14477         (store_conditionalpti): Likewise.
14479         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14480         atomic load/store instructions.
14481         (HSI): Likewise.
14483 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14485         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14486         loads.
14487         (insn_count): New attribute, with most cases extracted from...
14488         (length): ...here.  Redefine most cases in terms of insn_count.
14489         (single_insn): Delete.
14490         (can_delay): Use insn_count to check for single instructions.
14491         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14492         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14493         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14494         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14495         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14496         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14497         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14498         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14499         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14500         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14501         rather than "length".
14502         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14503         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14504         Use "insn_count" rather than "length".
14505         * config/mips/mips-dsp.md
14506         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14507         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14508         length attributes.
14510 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14512         PR tree-optimization/57361
14513         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14515 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14517         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14518         to split.
14519         (aarch64_simd_combine<mode>): New instruction expansion.
14520         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14521         function prototype.
14522         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14523         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14525 2013-06-12  Jan Hubicka  <jh@suse.cz>
14527         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14528         decl has when in streaming stage.
14529         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14530         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14532 2013-06-12  Roland Stigge  <stigge@antcom.de>
14534         PR target/57578
14535         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14537 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14539         PR tree-optimization/57537
14540         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14541         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14543 2013-06-12  Richard Biener  <rguenther@suse.de>
14545         * data-streamer.h (streamer_write_char_stream): CSE
14546         obs->current_pointer.
14547         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14548         streamer_write_char_stream manually and optimize the resulting loop.
14549         (streamer_write_hwi_stream): Likewise.
14551 2013-06-12  Jan Hubicka  <jh@suse.cz>
14553         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14554         * cgraph.h (varpool_create_empty_node): Declare.
14555         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14556         duplicated nodes.
14557         * symtab.c (symtab_unregister_node): Be lax about missin entries
14558         in node hash.
14559         (symtab_get_node): Update comment.
14560         * varpool.c (varpool_create_empty_node): Break out from ...
14561         (varpool_node_for_decl): ... here.
14562         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14564 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14566         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14567         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14568         part.  Use straight-line flow at the end.
14569         <COMPONENT_REF>: Remove superfluous else.
14570         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14572 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14574         PR target/56564
14575         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14576         target hook even for !TREE_PUBLIC decls.  If no resolution info
14577         is available, return false for common and external decls.
14579 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14581         * config/rl78/constraints.md (U): New constraint.
14582         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14583         valloc attribute.
14585 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14587         PR target/57589
14588         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14589         to allow returning address to AT_PLATFORM name.
14591 2013-06-11  Jan Hubicka  <jh@suse.cz>
14593         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14594         * cgraph.h (symtab_node_base): Add weakref flag.
14595         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14596         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14597         (output_weakrefs): Use weakref flag.
14598         * fold-const.c (simple_operand_p): Handle WEAK.
14599         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14600         * ipa.c (varpool_externally_visible_p): Drop weakref.
14601         (function_and_variable_visibility): Update comment; fix weakref
14602         sanity checks; do not clear DECL_WEAK on them.
14603         * lto-cgraph.c (lto_output_node): update.
14604         (lto_output_varpool_node): Update.
14605         (input_overwrite_node): Update.
14606         (input_node): Update.
14607         (input_varpool_node): Update.
14608         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14609         (lto_symtab_merge_symbols): Add sanity check.
14610         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14611         * passes.c (rest_of_decl_compilation): Set static flag, too.
14612         * symtab.c (dump_symtab_base): Dump weakref.
14613         (verify_symtab_base): Sanity check weakrefs.
14614         (symtab_make_decl_local): Remove duplicated code.
14615         (symtab_alias_ultimate_target): Simplify.
14616         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14618 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14620         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14621         sequence_vect == NULL.
14623 2013-06-11  DJ Delorie  <dj@redhat.com>
14625         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14626         (rl78_unwind_word_mode): New.
14628 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14630         * final.c (debug_prefix_maps): Make static.
14632 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14634         * function.c (initial_trampoline): Remove stray copy.
14636 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14638         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14640 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14642         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14643         within bounds at the beginning of the function.
14645 2013-06-11  Alan Modra  <amodra@gmail.com>
14647         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14648         reporting.
14649         (get_named_section): Don't NULL !DECL_P decl.
14651 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14653         * doc/invoke.texi (core-avx2): Document.
14654         (slm): Likewise.
14655         (atom): Updated with MOVBE.
14657 2013-06-11  Richard Biener  <rguenther@suse.de>
14659         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14661 2013-06-11  Anton Blanchard  <anton@samba.org>
14663         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14664         correct shift value in little-endian mode.
14666 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14668         PR target/56564
14669         * varasm.c (get_variable_align): Move #endif to the right place.
14671 2013-06-10  Cary Coutant  <ccoutant@google.com>
14673         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14674         for hash so that hash table traversal order is deterministic.
14676 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14677             Pat Haugen  <pthaugen@us.ibm.com>
14678             Peter Bergner  <bergner@vnet.ibm.com>
14680         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14681         of vectors in GPRS if they are direct moves or quad word load or
14682         store moves.
14684         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14685         declaration.
14686         (direct_move_p): Likewise.
14687         (quad_load_store_p): Likewise.
14689         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14690         classes into bins based on the physical register type.
14691         (reg_class_to_reg_type): Likewise.
14692         (IS_STD_REG_TYPE): Likewise.
14693         (IS_FP_VECT_REG_TYPE): Likewise.
14694         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14695         use direct move instructions.
14696         (reload_gpr_vsx): Likewise.
14697         (reload_vsx_gpr): Likewise.
14698         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14699         information that is a simplification of register classes.  Also
14700         precalculate direct move reload helpers.
14701         (direct_move_p): New function to return true if the operation can
14702         be done as a direct move instruciton.
14703         (quad_load_store_p): New function to return true if the operation
14704         is a quad memory operation.
14705         (rs6000_legitimize_address): If quad memory, only allow register
14706         indirect for TImode addresses.
14707         (rs6000_legitimate_address_p): Likewise.
14708         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14709         (rs6000_reload_register_type): Likewise.
14710         (register_to_reg_type): Return register type.
14711         (rs6000_secondary_reload_simple_move): New helper function for
14712         secondary reload and secondary memory needed to identify anything
14713         that is a simple move, and does not need reloading.
14714         (rs6000_secondary_reload_direct_move): New helper function for
14715         secondary reload to identify cases that can be done with several
14716         instructions via the direct move instructions.
14717         (rs6000_secondary_reload_move): New helper function for secondary
14718         reload to identify moves between register types that can be done.
14719         (rs6000_secondary_reload): Add support for quad memory operations
14720         and for direct move.
14721         (rs6000_secondary_memory_needed): Likewise.
14722         (rs6000_debug_secondary_memory_needed): Change argument names.
14723         (rs6000_output_move_128bit): New function to return the move to
14724         use for 128-bit moves, including knowing about the various
14725         limitations of quad memory operations.
14727         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14728         memory operations.  call rs6000_output_move_128bit for the actual
14729         instruciton(s) to generate.
14730         (vsx_movti_64bit): Likewise.
14732         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14733         (UNSPEC_P8V_MTVSRWZ): Likewise.
14734         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14735         (UNSPEC_P8V_MTVSRD): Likewise.
14736         (UNSPEC_P8V_XXPERMDI): Likewise.
14737         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14738         (UNSPEC_FUSION_GPR): Likewise.
14739         (FMOVE128_GPR): New iterator for direct move.
14740         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14741         (f32_sv): Likewise.
14742         (f32_dm): Likewise.
14743         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14744         loads and direct move instructions.
14745         (zero_extendsidi2_lfiwzx): Likewise.
14746         (extendsidi2_lfiwax): Likewise.
14747         (extendsidi2_nocell): Likewise.
14748         (floatsi<mode>2_lfiwax): Likewise.
14749         (lfiwax): Likewise.
14750         (floatunssi<mode>2_lfiwzx): Likewise.
14751         (lfiwzx): Likewise.
14752         (fix_trunc<mode>_stfiwx): Likewise.
14753         (fixuns_trunc<mode>_stfiwx): Likewise.
14754         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14755         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14756         (parity<mode>2_cmpb): Set length/type attr.
14757         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14758         for 'mr.' to fast_compare.
14759         (bpermd_<mode>): Change type attr to popcnt.
14760         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14761         (p8_mtvsrwz_1): Likewise.
14762         (p8_mtvsrwz_2): Likewise.
14763         (reload_fpr_from_gpr<mode>): Likewise.
14764         (p8_mtvsrd_1): Likewise.
14765         (p8_mtvsrd_2): Likewise.
14766         (p8_xxpermdi_<mode>): Likewise.
14767         (reload_vsx_from_gpr<mode>): Likewise.
14768         (reload_vsx_from_gprsf): Likewise.
14769         (p8_mfvsrd_3_<mode>): LIkewise.
14770         (reload_gpr_from_vsx<mode>): Likewise.
14771         (reload_gpr_from_vsxsf): Likewise.
14772         (p8_mfvsrd_4_disf): Likewise.
14773         (multi-word GPR splits): Do not split direct moves or quad memory
14774         operations.
14776 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14778         * tree-into-ssa.c (interesting_blocks): Make static.
14780 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14782         PR target/56564
14783         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14784         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14785         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14786         (get_variable_align): New function.
14787         (get_variable_section, emit_bss, emit_common,
14788         assemble_variable_contents, place_block_symbol): Use
14789         get_variable_align instead of DECL_ALIGN.
14790         (assemble_noswitch_variable): Add align argument, use it
14791         instead of DECL_ALIGN.
14792         (assemble_variable): Adjust caller.  Use get_variable_align
14793         instead of DECL_ALIGN.
14794         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14795         caller.
14796         (DATA_ABI_ALIGNMENT): Define.
14797         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14798         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14799         opt is false, only return the psABI mandated alignment increase.
14800         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14801         (DATA_ABI_ALIGNMENT): ... this.
14802         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14803         (DATA_ABI_ALIGNMENT): ... this.
14804         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14805         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14806         (DATA_ABI_ALIGNMENT): ... this.
14807         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14808         * doc/tm.texi: Regenerated.
14810 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14812         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14813         cmp_code to construct REG_EQUAL note.
14815 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14817         PR target/57568
14818         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14819         that operands[2] doesn't overlap with operands[0].
14821 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14822             Jan Hubicka  <jh@suse.cz>
14824         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14825         hack to mark symbols as used.
14827 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14829         PR rtl-optimization/57559
14830         * lra-constraints.c (process_alt_operands): Don't discourage
14831         memory with known offset for offsetable memory constraint.
14832         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14834 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14836         * varasm.c (struct oc_local_state): Reorder fields.
14837         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14838         and adjust accordingly.
14839         (output_constructor): Reorder initialization code and adjust call to
14840         output_constructor_bitfield.
14842 2013-06-07  Jan Hubicka  <jh@suse.cz>
14844         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14846 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14848         * tree-object-size.c (unknown): Make const.
14850 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14852         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14853         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14854         for last alternative in the cpu_facility attribute.
14856 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14858         PR target/56315
14859         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14860         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14861         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14862         * config/arm/constraints.md (Dg): New constraint.
14863         * config/arm/neon.md (xordi3_neon): Remove.
14864         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14865         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14867 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14869         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14870         Clean up alternatives.
14872 2013-06-07  Alan Modra  <amodra@gmail.com>
14874         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14875         va_list_gpr_size.
14877 2013-06-07  Alan Modra  <amodra@gmail.com>
14879         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14881 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14883         * config/arm/constraints.md (Df): New constraint.
14884         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14885         Correct length attribute for last two alternatives.
14887 2013-06-07  Alan Modra  <amodra@gmail.com>
14889         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14890         override user -mfp-in-toc.
14891         (offsettable_ok_by_alignment): Consider just the current access
14892         rather than the whole object, unless BLKmode.  Handle
14893         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14894         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14895         for -mcmodel=medium.
14896         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14897         override user -mfp-in-toc or -msum-in-toc.  Default to
14898         -mno-fp-in-toc for -mcmodel=medium.
14900 2013-06-06  DJ Delorie  <dj@redhat.com>
14902         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14903         TARGET_VALID_POINTER_MODE.
14905 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14906             Pat Haugen  <pthaugen@us.ibm.com>
14907             Peter Bergner  <bergner@vnet.ibm.com>
14909         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14910         Document new power8 builtins.
14912         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14913         condition code register, to allow 128-bit logical operations to be
14914         done in the VSX or GPR registers.
14915         (nor<mode>3): Use the canonical form for nor.
14916         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14917         vclz*, and vpopcnt* vector instructions.
14918         (nand<mode>3): Likewise.
14919         (orc<mode>3): Likewise.
14920         (clz<mode>2): LIkewise.
14921         (popcount<mode>2): Likewise.
14923         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14924         that only the GPRs are recognized.
14926         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14927         support for new power8 builtins.
14929         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14930         builtin functions.
14931         (xscvdpspn): Likewise.
14932         (vclz): Likewise.
14933         (vclzb): Likewise.
14934         (vclzh): Likewise.
14935         (vclzw): Likewise.
14936         (vclzd): Likewise.
14937         (vpopcnt): Likewise.
14938         (vpopcntb): Likewise.
14939         (vpopcnth): Likewise.
14940         (vpopcntw): Likewise.
14941         (vpopcntd): Likewise.
14942         (vgbbd): Likewise.
14943         (vmrgew): Likewise.
14944         (vmrgow): Likewise.
14945         (eqv): Likewise.
14946         (eqv_v16qi3): Likewise.
14947         (eqv_v8hi3): Likewise.
14948         (eqv_v4si3): Likewise.
14949         (eqv_v2di3): Likewise.
14950         (eqv_v4sf3): Likewise.
14951         (eqv_v2df3): Likewise.
14952         (nand): Likewise.
14953         (nand_v16qi3): Likewise.
14954         (nand_v8hi3): Likewise.
14955         (nand_v4si3): Likewise.
14956         (nand_v2di3): Likewise.
14957         (nand_v4sf3): Likewise.
14958         (nand_v2df3): Likewise.
14959         (orc): Likewise.
14960         (orc_v16qi3): Likewise.
14961         (orc_v8hi3): Likewise.
14962         (orc_v4si3): Likewise.
14963         (orc_v2di3): Likewise.
14964         (orc_v4sf3): Likewise.
14965         (orc_v2df3): Likewise.
14967         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
14968         allow power8 quad mode in 64-bit.
14969         (rs6000_builtin_vectorized_function): Add support to vectorize
14970         ISA 2.07 count leading zeros, population count builtins.
14971         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
14972         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
14973         (builtin_function_type): Add vgbbd builtin function which takes an
14974         unsigned argument.
14975         (altivec_expand_vec_perm_const): Add support for new power8 merge
14976         instructions.
14978         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
14979         that does not include TImdoe for use with 32-bit.
14980         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
14981         instructions.
14982         (UNSPEC_VSX_CVDPSPN): Likewise.
14983         (vsx_xscvdpspn): Likewise.
14984         (vsx_xscvspdpn): Likewise.
14985         (vsx_xscvdpspn_scalar): Likewise.
14986         (vsx_xscvspdpn_directmove): Likewise.
14987         (vsx_and<mode>3): Split logical operations into 32-bit and
14988         64-bit. Add support to do logical operations on TImode as well as
14989         VSX vector types.  Allow logical operations to be done in either
14990         VSX registers or in general purpose registers in 64-bit mode.  Add
14991         splitters if GPRs were used. For AND, add clobber of CCmode to
14992         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
14993         encoding.
14994         (vsx_and<mode>3_32bit): Likewise.
14995         (vsx_and<mode>3_64bit): Likewise.
14996         (vsx_ior<mode>3): Likewise.
14997         (vsx_ior<mode>3_32bit): Likewise.
14998         (vsx_ior<mode>3_64bit): Likewise.
14999         (vsx_xor<mode>3): Likewise.
15000         (vsx_xor<mode>3_32bit): Likewise.
15001         (vsx_xor<mode>3_64bit): Likewise.
15002         (vsx_one_cmpl<mode>2): Likewise.
15003         (vsx_one_cmpl<mode>2_32bit): Likewise.
15004         (vsx_one_cmpl<mode>2_64bit): Likewise.
15005         (vsx_nor<mode>3): Likewise.
15006         (vsx_nor<mode>3_32bit): Likewise.
15007         (vsx_nor<mode>3_64bit): Likewise.
15008         (vsx_andc<mode>3): Likewise.
15009         (vsx_andc<mode>3_32bit): Likewise.
15010         (vsx_andc<mode>3_64bit): Likewise.
15011         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15012         and xxlorc instructions.
15013         (vsx_eqv<mode>3_64bit): Likewise.
15014         (vsx_nand<mode>3_32bit): Likewise.
15015         (vsx_nand<mode>3_64bit): Likewise.
15016         (vsx_orc<mode>3_32bit): Likewise.
15017         (vsx_orc<mode>3_64bit): Likewise.
15019         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15021         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15022         instruction.
15023         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15024         (p8_vmrgow): Likewise.
15025         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15026         GPRs to be split under VSX.
15027         (p8v_clz<mode>2): Add power8 count leading zero support.
15028         (p8v_popcount<mode>2): Add power8 population count support.
15029         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15030         support.
15032         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15033         instruction.
15035         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15036         builtin functions.
15037         (vec_nand): Likewise.
15038         (vec_vclz): Likewise.
15039         (vec_vclzb): Likewise.
15040         (vec_vclzd): Likewise.
15041         (vec_vclzh): Likewise.
15042         (vec_vclzw): Likewise.
15043         (vec_vgbbd): Likewise.
15044         (vec_vmrgew): Likewise.
15045         (vec_vmrgow): Likewise.
15046         (vec_vpopcnt): Likewise.
15047         (vec_vpopcntb): Likewise.
15048         (vec_vpopcntd): Likewise.
15049         (vec_vpopcnth): Likewise.
15050         (vec_vpopcntw): Likewise.
15052 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15054         PR rtl-optimization/57468
15055         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15056         spilled pseudos.
15058 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15060         PR rtl-optimization/57459
15061         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15062         type when setting live regs.
15064 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15066         * config/s390/s390.opt (mlra): New option.
15067         * config/s390/s390.c (s390_decompose_address): Check displacement
15068         for all registers for LRA.
15069         (s390_secondary_reload): Don't used secondary reloads for LRA.
15070         (s390_lra_p): New function.
15071         (TARGET_LRA_P): Define.
15072         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15073         of attribute cpu_facility to zarch for the last alternative.
15074         (*cmpmem_short): Ditto.
15076 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15078         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15079         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15080         (arm_expand_prologue): Likewise.
15082 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15084         PR c++/53743
15085         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15086         as this is now done by redirect_edge_and_branch_force.
15087         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15088         barriers, and fix interaction with splitting.
15089         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15090         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15091         reflect changes made in the routine.
15092         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15093         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15094         since this is called in cfglayout mode, and replace partition fixup
15095         with assert as that is now done by force_nonfallthru_and_redirect.
15096         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15097         already be marked with region crossing note.
15098         (insert_section_boundary_note): Make non-static, gate on flag
15099         has_bb_partition, rewrite to also check for multiple partitions.
15100         (rest_of_handle_reorder_blocks): Remove call to
15101         insert_section_boundary_note, now done later during free_cfg.
15102         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15103         * bb-reorder.h (insert_section_boundary_note): Declare.
15104         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15105         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15106         invoke insert_section_boundary_note.
15107         (try_redirect_by_replacing_jump): Remove unnecessary
15108         check for region crossing note.
15109         (fixup_partition_crossing): New function.
15110         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15111         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15112         in non-cfglayout mode.
15113         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15114         remove old code that tried to do this. Emit barrier correctly
15115         when we are in cfglayout mode.
15116         (last_bb_in_partition): New function.
15117         (rtl_split_edge): Correctly fixup partition boundaries.
15118         (commit_one_edge_insertion): Remove old code that tried to
15119         fixup region crossing edge since this is now handled in
15120         split_block, and set up insertion point correctly since
15121         block may now end in a jump.
15122         (verify_hot_cold_block_grouping): Guard against checking when not in
15123         linearized RTL mode.
15124         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15125         notes.
15126         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15127         rtl_verify_flow_info, so not called in cfglayout mode.
15128         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15129         (fixup_reorder_chain): Remove old code that attempted to fixup region
15130         crossing note as this is now handled in force_nonfallthru_and_redirect.
15131         (duplicate_insn_chain): Don't duplicate switch section notes.
15132         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15133         note.
15134         * basic-block.h (emit_barrier_after_bb): Declare.
15136 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15138         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15139         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15140         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15142 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15144         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15145         where appropriate.
15146         * config/arm/ldmstm.md: Regenerate.
15148 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15150         * config/arm/sync.md (atomic_loaddi_1):
15151         Disable predication for arm_restrict_it.
15152         (arm_load_exclusive<mode>): Likewise.
15153         (arm_load_exclusivesi): Likewise.
15154         (arm_load_exclusivedi): Likewise.
15155         (arm_load_acquire_exclusive<mode>): Likewise.
15156         (arm_load_acquire_exclusivesi): Likewise.
15157         (arm_load_acquire_exclusivedi): Likewise.
15158         (arm_store_exclusive<mode>): Likewise.
15159         (arm_store_exclusive<mode>): Likewise.
15160         (arm_store_release_exclusivedi): Likewise.
15161         (arm_store_release_exclusive<mode>): Likewise.
15163 2013-06-06  Richard Biener  <rguenther@suse.de>
15165         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15166         after LTO_null.
15167         (lto_tag_is_tree_code_p): Adjust.
15168         (lto_tag_is_gimple_code_p): Likewise.
15169         (lto_gimple_code_to_tag): Likewise.
15170         (lto_tag_to_gimple_code): Likewise.
15171         (lto_tree_code_to_tag): Likewise.
15172         (lto_tag_to_tree_code): Likewise.
15173         * data-streamer.h (streamer_write_hwi_in_range): Use
15174         uhwi streaming to stream the normalized range.
15175         (streamer_read_hwi_in_range): Likewise.
15177 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15179         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15180         (predicable_short_it): Likewise.
15181         (predicated): Likewise.
15182         (enabled): Handle above.
15183         (define_cond_exec): Set predicated attribute to yes.
15185 2013-06-05  Mike Stump  <mikestump@comcast.net>
15187         * gdbinit.in (__FUNCTION__): Add.
15189 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15191         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15192         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15194 2013-06-05  Jan Hubicka  <jh@suse.cz>
15196         * varasm.c (mark_decl_referenced): Revert the removal until targets
15197         are fixed.
15199 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15201         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15202         instead of mark_decl_referenced.
15204 2013-06-05  Jan Hubicka  <jh@suse.cz>
15206         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15207         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15208         and symtab_used_from_object_file_p.
15209         (cgraph_make_node_local_1): Clear forced_by_abi.
15210         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15211         * cgraph.h (symtab_node_base): Add forced_by_abi.
15212         (decide_is_variable_needed): Remove.
15213         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15214         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15215         (decide_is_symbol_needed): ... this one; handle symbols in general;
15216         always analyze virtuals; honnor forced_by_abi.
15217         (cgraph_finalize_function): Update.
15218         (varpool_finalize_decl): Update.
15219         (symbol_defined_and_needed): Remove.
15220         (analyze_functions): Update.
15221         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15222         output_refs, input_overwrite_node): Handle forced_by_abi.
15223         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15224         (address_taken_from_non_vtable_p): ... this one.
15225         (comdat_can_be_unshared_p_1): New function.
15226         (cgraph_comdat_can_be_unshared_p): Rename to ...
15227         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15228         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15229         (function_and_variable_visibility): Clear forced_by_abi as needed.
15230         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15231         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15232         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15233         * varpool.c (decide_is_variable_needed): Remove.
15235 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15237         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15238         (arm_option_override): Override arm_restrict_it where appropriate.
15239         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15240         * config/arm/arm.opt (mrestrict-it): New command-line option.
15241         * doc/invoke.texi: Document -mrestrict-it.
15243 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15245         * tsan.c (tsan_atomic_table): Make const.
15247 2013-06-05  Richard Biener  <rguenther@suse.de>
15249         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15250         index associated with the tree we are supposed to replace.
15251         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15252         TREE_ASM_WRITTEN as zero for everything but SSA names.
15254 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15256         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15258 2013-06-04  Jan Hubicka  <jh@suse.cz>
15260         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15261         (input_node, input_varpool_node): Handle correctly external same
15262         body aliases.
15263         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15264         nodes at ltrans stage.
15266 2013-06-04  Jan Hubicka  <jh@suse.cz>
15268         * ipa-inline.c (update_caller_keys): Fix availability test.
15269         (update_callee_keys): Likewise.
15270         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15271         to follow ELF standard.
15273 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15275         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15276         (mips64r5900el-*-elf*): New configurations.
15277         * config/mips/mips-cpus.def (r5900): New processor.
15278         * config/mips/mips-tables.opt: Regenerate.
15279         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15280         (mips_issue_rate): Handle PROCESSOR_R5900.
15281         (mips_reorg_process_insns): Force reorder mode for the R5900.
15282         * config/mips/mips.h (TARGET_MIPS5900): Define.
15283         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15284         TARGET_MIPS5900.
15285         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15286         TARGET_MIPS5900.
15287         * config/mips/mips.md (processor): Add r5900.
15288         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15290 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15292         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15293         into function to generate MOVI instruction.
15294         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15295         (aarch64_preferred_simd_mode): Turn into wrapper.
15296         (aarch64_output_scalar_simd_mov_immediate): New function.
15297         * config/aarch64/aarch64-protos.h: Add prototype for above.
15299 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15301         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15302         element_char member.
15303         (sizetochar): Return signed char.
15304         (aarch64_simd_valid_immediate): Remove elchar and other
15305         unnecessary variables.
15306         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15307         Calculate element_char as required.
15308         * config/aarch64/aarch64-protos.h: Update and move prototype
15309         for aarch64_output_simd_mov_immediate.
15310         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15311         Update arguments.
15313 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15315         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15316         information completed by aarch64_simd_valid_immediate.
15317         (aarch64_legitimate_constant_p): Update arguments.
15318         (aarch64_simd_valid_immediate): Work with struct rather than many
15319         pointers.
15320         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15321         (aarch64_simd_make_constant): Update arguments.
15322         (aarch64_output_simd_mov_immediate): Work with struct rather than
15323         many pointers.  Output immediate directly rather than as operand.
15324         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15325         Update prototype.
15326         * config/aarch64/constraints.md (Dn): Update arguments.
15328 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15330         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15331         longer static.
15332         (aarch64_simd_immediate_valid_for_move): Remove.
15333         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15334         (aarch64_simd_make_constant): Update call.
15335         (aarch64_output_simd_mov_immediate): Update call.
15336         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15337         Add prototype.
15338         * config/aarch64/constraints.md (Dn): Update call.
15340 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15342         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15343         return type to bool for prototype.
15344         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15345         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15346         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15348 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15350         * config/mips/mips.opt (meva): New.
15351         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15352         (ASM_SPEC): Handle -meva.
15353         * doc/invoke.texi (meva):  Document.
15355 2013-06-04  Alan Modra  <amodra@gmail.com>
15357         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15358         constant output.
15360 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15362         * rtl.def: Add extra fourth optional field to define_cond_exec.
15363         * gensupport.c (process_one_cond_exec): Process attributes from
15364         define_cond_exec.
15365         * doc/md.texi: Document fourth field in define_cond_exec.
15367 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15369         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15370         out the processing order as in store_bit_field_1.
15372 2013-06-04  Jan Hubicka  <jh@suse.cz>
15374         PR middle-end/57500
15375         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15376         non-VAR_DECL node if it does not exist yet.
15378 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15380         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15381         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15382         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15383         target_cpu_default setting.
15385 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15387         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15388         default to -fopt-info=optimized instead of all.
15389         * doc/invoke.texi: Ditto.
15390         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15391         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15392         (execute_vect_slp): Emit BB vectorization success under
15393         MSG_OPTIMIZED_LOCATIONS.
15394         * tree-vect-slp.c (vect_slp_transform_bb): Change
15395         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15396         * tree-vect-loop.c (vect_transform_loop): Ditto.
15398 2013-06-03  Jason Merrill  <jason@redhat.com>
15400         PR c++/57415
15401         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15402         Use TARGET_EXPR for C++.
15404 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15406         PR rtl-optimization/57268
15407         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15408         if DEBUG_INSN_P (insn).
15410         Reapply
15411         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15413         PR rtl-optimization/57268
15414         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15415         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15417 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15419         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15420         (ix86_avoid_lea_for_addr): Likewise.
15421         (exact_dependency_1): Likewise.
15422         (ix86_adjust_cost): Likewise.
15423         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15424         removed.
15425         (do_reorder_for_imul): Fix typo, formatting and
15426         !reload_completed check removed.
15427         (ix86_sched_reorder): Fix typo and formatting.
15428         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15429         list.
15431 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15433         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15435 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15437         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15438         <STRING_CST>: Likewise.
15439         <VECTOR_CST>: Likewise.
15441 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15442             Mikael Morin  <mikael@gcc.gnu.org>
15444         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15445         * config.in: Regenerated.
15446         * configure: Regenerated.
15448 2013-06-01  Jan Hubicka  <jh@suse.cz>
15450         PR middle-end/57366
15451         * cgraphunit.c (compile): When weakref is not supported,
15452         set up transparent aliases before final output pass.
15453         * varasm.c (assemble_alias): Do not try to do it here.
15455 2013-06-01  Jan Hubicka  <jh@suse.cz>
15457         PR middle-end/57467
15458         * passes.c (for_per_function): Skip unanalyzed functions.
15460 2013-06-01  Jan Hubicka  <jh@suse.cz>
15462         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15463         (lto_symtab_merge_symbols_1): ... this one.
15464         (lto_symtab_merge_cgraph_nodes): Rename to ...
15465         (lto_symtab_merge_symbols): ... this one; simplify.
15466         * cgraph.c (same_body_aliases_done): Rename to ...
15467         (cpp_implicit_aliases_done): ... this one.
15468         (cgraph_create_function_alias): Update.
15469         (cgraph_same_body_alias): Update.
15470         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15471         (verify_edge_corresponds_to_fndecl): Simplify.
15472         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15473         (cgraph_node): Remove same_body_alias.
15474         (varpool_node): Remove alias_of and extra_name_alias.
15475         (same_body_aliases_done): Rename to ..
15476         (cpp_implicit_aliases_done): ... this one.
15477         (symtab_alias_ultimate_target): Add default parameter.
15478         (symtab_resolve_alias): New function.
15479         (fixup_same_cpp_alias_visibility): Declare.
15480         (cgraph_function_node): Add default parameter.
15481         (cgraph_node_asm_name): Likewise.
15482         (cgraph_function_or_thunk_node): Add default parameter; do
15483         not ICE when it is NULL.
15484         (varpool_variable_node): Likewise.
15485         * tree-emutls.c (create_emultls_var): Update.
15486         (ipa_lower_emutls): Update.
15487         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15488         (cgraph_reset_node): Reset alias info.
15489         (cgraph_finalize_function): Update.
15490         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15491         (analyze_function): Simplify.
15492         (cgraph_process_same_body_aliases): Simplify.
15493         (analyze_functions): Fixup same body aliases.
15494         (handle_alias_pairs): Simplify.
15495         (assemble_thunk): Update.
15496         (assemble_thunks_and_aliases): Update.
15497         (output_weakrefs): Rewrite.
15498         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15499         (lto_output_varpool_node): Likewise.
15500         (compute_ltrans_boundary): Remve assert.
15501         (get_alias_symbol): New functoin.
15502         (input_node): Rewrite alias handling.
15503         (input_varpool_node): Likewise.
15504         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15505         * ipa.c (process_references): Handle weakrefs correctly.
15506         (symtab_remove_unreachable_nodes): Likewise.
15507         * trans-mem.c (get_cg_data): Update.
15508         (ipa_tm_create_version_alias): Update.
15509         (ipa_tm_execute): Update.
15510         * symtab.c (dump_symtab_base): Dump aliases.
15511         (verify_symtab_base): Verify aliases.
15512         (symtab_node_availability): New function.
15513         (symtab_alias_ultimate_target): Simplify.
15514         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15515         handle all the fixup cases.
15516         (symtab_resolve_alias): New function.
15517         * passes.c (ipa_write_summaries): Handle weakrefs.
15518         * varpool.c (varpool_analyze_node): Simplify.
15519         (assemble_aliases): Update.
15520         (varpool_create_variable_alias): Simplify.
15521         (varpool_extra_name_alias): Simplify.
15522         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15523         (lto_symtab_merge_symbols): ... this one.
15525 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15527         Revert
15528         PR rtl-optimization/57268
15529         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15530         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15532 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15534         Partially reverted:
15535         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15537         PR middle-end/57073
15538         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15539         further up.
15541 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15543         PR rtl-optimization/57268
15544         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15545         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15547 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15549         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15550         unordered comparison operators when -fno-trapping-math is in effect
15551         on the e500.
15552         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15553         and implement unordered comparison operators properly on the e500.
15555 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15557         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15558         for constant scalar integers.
15559         (simplify_relational_operation_1): Likewise.
15561 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15563         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15564         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15565         Fix comment.
15567 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15568             Igor Zamyatin  <igor.zamyatin@intel.com>
15570         Silvermont (SLM) architecture performance tuning.
15571         * config/i386/i386.h (enum ix86_tune_indices): Add
15572         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15573         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15575         * config/i386/i386.c (initial_ix86_tune_features)
15576         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15577         (ix86_lea_outperforms): Handle Silvermont tuning.
15578         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15579         call.
15580         (ix86_use_lea_for_mov): Likewise.
15581         (ix86_avoid_lea_for_addr): Likewise.
15582         (ix86_lea_for_add_ok): Likewise.
15583         (exact_dependency_1): New function.
15584         (exact_store_load_dependency): Likewise.
15585         (ix86_adjust_cost): Handle Silvermont tuning.
15586         (do_reoder_for_imul): Likewise.
15587         (swap_top_of_ready_list): New function.
15588         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15590         * config/i386/i386.md (peepholes that split memory operand in fp
15591         converts): New.
15593 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15595         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15596         Remove un-necessary braces.
15598 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15600         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15601         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15603 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15605         PR middle-end/57073
15606         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15608 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15610         PR target/56315
15611         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15612         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15613         * config/arm/neon.md (iordi3_neon): Remove.
15614         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15615         * config/arm/predicates.md (imm_for_neon_logic_operand):
15616         Move to earlier in the file.
15617         (neon_logic_op2): Likewise.
15618         (arm_iordi_operand_neon): New predicate.
15620 2013-05-31  Richard Biener  <rguenther@suse.de>
15622         PR tree-optimization/57478
15623         PR tree-optimization/57453
15624         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15625         are life as well.
15627 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15629         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15630         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15632 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15633             Thomas Koenig  <tkoenig@gcc.gnu.org>
15635         PR middle-end/57073
15636         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15637         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15639 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15641         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15643 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15645         * target.def (register_usage_leveling_p): New hook.
15646         * targhooks.c (default_register_usage_leveling_p): New.
15647         * targhooks.h (default_register_usage_leveling_p): New prototype.
15648         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15649         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15650         * doc/tm.texi: Update.
15651         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15653 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15655         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15656         (*insv_reg<mode>): New define_insn.
15658 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15660         PR rtl-optimization/57439
15661         * postreload.c (move2add_valid_value_p): Check that we have
15662         a zero subreg_regno_offset when accessing the register in
15663         the requested mode.
15665 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15666             Igor Zamyatin  <igor.zamyatin@intel.com>
15668         Silvermont (SLM) architecture pipeline model, tuning and
15669         insn selection.
15670         * config.gcc: Add slm config options and target.
15672         * config/i386/slm.md: New.
15674         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15676         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15677         PROCESSOR_SLM.
15678         (ix86_target_macros_internal): Likewise.
15680         * config/i386/i386.c (slm_cost): New cost.
15681         (m_SLM): New macro flag.
15682         (initial_ix86_tune_features): Set m_SLM.
15683         (x86_accumulate_outgoing_args): Likewise.
15684         (x86_arch_always_fancy_math_387): Likewise.
15685         (processor_target_table): Add slm cost.
15686         (cpu_names): Add slm cpu name.
15687         (x86_option_override_internal): Set SLM ISA.
15688         (ix86_issue_rate): New case PROCESSOR_SLM.
15689         (ia32_multipass_dfa_lookahead): Likewise.
15690         (fold_builtin_cpu): Add slm.
15692         * config/i386/i386.h (TARGET_SLM): New target macro.
15693         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15694         (processor_type): Add PROCESSOR_SLM.
15696         * config/i386/i386.md (cpu): Add new value "slm".
15697         (slm.md): Include slm.md.
15699 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15700             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15702         * config/arm/arm-protos.h: Add and update function protos.
15703         * config/arm/arm.c (use_simple_return_p): New added.
15704         (thumb2_expand_return): Check simple_return flag.
15705         * config/arm/arm.md: Add simple_return and conditional simple_return.
15706         * config/arm/iterators.md: Add iterator for return and simple_return.
15708 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15710         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15711         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15712         (arm_emit_vfp_multi_reg_pop): Likewise.
15713         (thumb2_emit_ldrd_pop): Likewise.
15714         (arm_expand_epilogue): Add misc REG_CFA notes.
15715         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15717 2013-05-29  Lawrence Crowl  <crowl@google.com>
15719         * config/arm/t-arm: Update for below.
15721         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15722         Change type to hash_table.  Update dependent calls and types.
15724         * config/i386/t-cygming: Update for below.
15726         * config/i386/t-interix: Update for below.
15728         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15729         Change type to hash_table.  Update dependent calls and types.
15730         (i386_find_on_wrapper_list::wrappers): Likewise.
15732         * config/ia64/t-ia64: Update for below.
15734         * config/ia64/ia64.c (bundle_state_table):
15735         Change type to hash_table.  Update dependent calls and types.
15737         * config/mips/mips.c (mips_reorg_process_insns::htab):
15738         Change type to hash_table.  Update dependent calls and types.
15740         * config/sol2.c (solaris_comdat_htab):
15741         Change type to hash_table.  Update dependent calls and types.
15743         * config/t-sol2: Update for above.
15745 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15747         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15748         functions are not yet marked as defined.
15750 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15751             Pat Haugen  <pthaugen@us.ibm.com>
15752             Peter Bergner  <bergner@vnet.ibm.com>
15754         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15755         instructions.
15756         (VEC_A): Likewise.
15757         (VEC_C): Likewise.
15758         (vrotl<mode>3): Likewise.
15759         (vashl<mode>3): Likewise.
15760         (vlshr<mode>3): Likewise.
15761         (vashr<mode>3): Likewise.
15763         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15764         support for power8 V2DI builtins.
15766         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15767         power8 V2DI builtins.
15768         (vupkhsw): Likewise.
15769         (vupklsw): Likewise.
15770         (vaddudm): Likewise.
15771         (vminsd): Likewise.
15772         (vmaxsd): Likewise.
15773         (vminud): Likewise.
15774         (vmaxud): Likewise.
15775         (vpkudum): Likewise.
15776         (vpksdss): Likewise.
15777         (vpkudus): Likewise.
15778         (vpksdus): Likewise.
15779         (vrld): Likewise.
15780         (vsld): Likewise.
15781         (vsrd): Likewise.
15782         (vsrad): Likewise.
15783         (vsubudm): Likewise.
15784         (vcmpequd): Likewise.
15785         (vcmpgtsd): Likewise.
15786         (vcmpgtud): Likewise.
15787         (vcmpequd_p): Likewise.
15788         (vcmpgtsd_p): Likewise.
15789         (vcmpgtud_p): Likewise.
15790         (vupkhsw): Likewise.
15791         (vupklsw): Likewise.
15792         (vaddudm): Likewise.
15793         (vmaxsd): Likewise.
15794         (vmaxud): Likewise.
15795         (vminsd): Likewise.
15796         (vminud): Likewise.
15797         (vpksdss): Likewise.
15798         (vpksdus): Likewise.
15799         (vpkudum): Likewise.
15800         (vpkudus): Likewise.
15801         (vrld): Likewise.
15802         (vsld): Likewise.
15803         (vsrad): Likewise.
15804         (vsrd): Likewise.
15805         (vsubudm): Likewise.
15807         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15808         support for power8 V2DI instructions.
15810         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15811         power8 V2DI instructions.  Combine pack and unpack insns to use an
15812         iterator for each mode.  Check whether a particular mode supports
15813         Altivec instructions instead of just checking TARGET_ALTIVEC.
15814         (UNSPEC_VPKUWUM): Likewise.
15815         (UNSPEC_VPKSHSS): Likewise.
15816         (UNSPEC_VPKSWSS): Likewise.
15817         (UNSPEC_VPKUHUS): Likewise.
15818         (UNSPEC_VPKSHUS): Likewise.
15819         (UNSPEC_VPKUWUS): Likewise.
15820         (UNSPEC_VPKSWUS): Likewise.
15821         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15822         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15823         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15824         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15825         (UNSPEC_VUPKHSB): Likewise.
15826         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15827         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15828         (UNSPEC_VUPKHSH): Likewise.
15829         (UNSPEC_VUPKLSB): Likewise.
15830         (UNSPEC_VUPKLSH): Likewise.
15831         (VI2): Likewise.
15832         (VI_char): Likewise.
15833         (VI_scalar): Likewise.
15834         (VI_unit): Likewise.
15835         (VP): Likewise.
15836         (VP_small): Likewise.
15837         (VP_small_lc): Likewise.
15838         (VU_char): Likewise.
15839         (add<mode>3): Likewise.
15840         (altivec_vaddcuw): Likewise.
15841         (altivec_vaddu<VI_char>s): Likewise.
15842         (altivec_vadds<VI_char>s): Likewise.
15843         (sub<mode>3): Likewise.
15844         (altivec_vsubcuw): Likewise.
15845         (altivec_vsubu<VI_char>s): Likewise.
15846         (altivec_vsubs<VI_char>s): Likewise.
15847         (altivec_vavgs<VI_char>): Likewise.
15848         (altivec_vcmpbfp): Likewise.
15849         (altivec_eq<mode>): Likewise.
15850         (altivec_gt<mode>): Likewise.
15851         (altivec_gtu<mode>): Likewise.
15852         (umax<mode>3): Likewise.
15853         (smax<mode>3): Likewise.
15854         (umin<mode>3): Likewise.
15855         (smin<mode>3): Likewise.
15856         (altivec_vpkuhum): Likewise.
15857         (altivec_vpkuwum): Likewise.
15858         (altivec_vpkshss): Likewise.
15859         (altivec_vpkswss): Likewise.
15860         (altivec_vpkuhus): Likewise.
15861         (altivec_vpkshus): Likewise.
15862         (altivec_vpkuwus): Likewise.
15863         (altivec_vpkswus): Likewise.
15864         (altivec_vpks<VI_char>ss): Likewise.
15865         (altivec_vpks<VI_char>us): Likewise.
15866         (altivec_vpku<VI_char>us): Likewise.
15867         (altivec_vpku<VI_char>um): Likewise.
15868         (altivec_vrl<VI_char>): Likewise.
15869         (altivec_vsl<VI_char>): Likewise.
15870         (altivec_vsr<VI_char>): Likewise.
15871         (altivec_vsra<VI_char>): Likewise.
15872         (altivec_vsldoi_<mode>): Likewise.
15873         (altivec_vupkhsb): Likewise.
15874         (altivec_vupkhs<VU_char>): Likewise.
15875         (altivec_vupkls<VU_char>): Likewise.
15876         (altivec_vupkhsh): Likewise.
15877         (altivec_vupklsb): Likewise.
15878         (altivec_vupklsh): Likewise.
15879         (altivec_vcmpequ<VI_char>_p): Likewise.
15880         (altivec_vcmpgts<VI_char>_p): Likewise.
15881         (altivec_vcmpgtu<VI_char>_p): Likewise.
15882         (abs<mode>2): Likewise.
15883         (vec_unpacks_hi_v16qi): Likewise.
15884         (vec_unpacks_hi_v8hi): Likewise.
15885         (vec_unpacks_lo_v16qi): Likewise.
15886         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15887         (vec_unpacks_lo_v8hi): Likewise.
15888         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15889         (vec_pack_trunc_v8h): Likewise.
15890         (vec_pack_trunc_v4si): Likewise.
15891         (vec_pack_trunc_<mode>): Likewise.
15893         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15894         V2DI builtins.
15895         (vec_vmaxsd): Likewise.
15896         (vec_vmaxud): Likewise.
15897         (vec_vminsd): Likewise.
15898         (vec_vminud): Likewise.
15899         (vec_vpksdss): Likewise.
15900         (vec_vpksdus): Likewise.
15901         (vec_vpkudum): Likewise.
15902         (vec_vpkudus): Likewise.
15903         (vec_vrld): Likewise.
15904         (vec_vsld): Likewise.
15905         (vec_vsrad): Likewise.
15906         (vec_vsrd): Likewise.
15907         (vec_vsubudm): Likewise.
15908         (vec_vupkhsw): Likewise.
15909         (vec_vupklsw): Likewise.
15911 2013-05-29  Jan Hubicka  <jh@suse.cz>
15913         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15914         flags; reorder rest of fields in more consistent way.
15915         (varpool_node): Remove analyzed, finalized and alias.
15916         (cgraph_ndoe): Likewise.
15917         (symtab_alias_ultimate_target): New function.
15918         (cgraph_function_node): Move offline.
15919         (cgraph_reset_node): Declare.
15920         (cgraph_comdat_can_be_unshared_p): Remove.
15921         (varpool_remove_initializer): Declare.
15922         (varpool_first_defined_variable, varpool_next_defined_variable
15923         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15924         (cgraph_function_with_gimple_body_p): Update.
15925         (varpool_all_refs_explicit_p): Update.
15926         (symtab_alias_target): New function.
15927         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15928         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15929         (cgraph_function_or_thunk_node): Simplify using
15930         symtab_alias_ultimate_target.
15931         (varpool_variable_node): Likewise.
15932         * cgraph.c (cgraph_create_function_alias): Update.
15933         (cgraph_add_thunk): Update.
15934         (cgraph_remove_node): Update.
15935         (dump_cgraph_node): Do not dump removed flags.
15936         (cgraph_function_body_availability): Update.
15937         (cgraph_propagate_frequency): Update.
15938         (verify_cgraph_node): Check sanity of local flag.
15939         (cgraph_function_node): Move here from cgraph.h; revamp for
15940         cgraph_function_or_thunk_node.
15941         * lto-symtab.c (lto_varpool_replace_node): Update.
15942         (lto_symtab_resolve_can_prevail_p): Update.
15943         (lto_symtab_merge_cgraph_nodes): Update.
15944         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15945         propagate_constants_accross_call, devirtualization_time_bonus,
15946         ipcp_propagate_stage): Update.
15947         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
15948         * ipa-inline-transform.c (clone_inlined_nodes,
15949         preserve_function_body_p): Update.
15950         * ipa-reference.c (propagate): Update.
15951         (write_node_summary_p): Update.
15952         * toplev.c (wrapup_global_declaration_2): Update.
15953         * cgraphunit.c (cgraph_analyze_function): Rename to ...
15954         (analyze_function) ... this one.
15955         (cgraph_process_new_functions): Update.
15956         (cgraph_reset_node): Export.
15957         (cgraph_finalize_function): Update.
15958         (cgraph_add_new_function): Update.
15959         (process_function_and_variable_attributes): Update.
15960         (varpool_finalize_decl): Update.
15961         (symbol_finalized): Remove.
15962         (symbol_finalized_and_needed): Rename to ...
15963         (symbol_defined_and_needed): ... update.
15964         (cgraph_analyze_functions): Update.
15965         (handle_alias_pairs): Update.
15966         (mark_functions_to_output): Update.
15967         (assemble_thunk): Update.
15968         (output_in_order): Update.
15969         (output_weakrefs): Update.
15970         (finalize_compilation_unit): Update.
15971         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
15972         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
15973         input_node, input_varpool_node): Update.
15974         * dbxout.c (dbxout_expand_expr): Update.
15975         * cgraphclones.c (cgraph_clone_node): Update.
15976         (cgraph_copy_node_for_versioning): Update.
15977         (cgraph_materialize_clone): Update.
15978         (cgraph_materialize_all_clones): Update.
15979         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
15980         propagate_pure_const, propagate_nothrow): Update.
15981         * lto-streamer-out.c (lto_output, write_symbol): Update.
15982         * ipa-utils.c (ipa_reverse_postorder): Update.
15983         * ipa-inline.c (can_inline_edge_p): Update.
15984         (update_caller_keys, ipa_inline): Update.
15985         * dwarf2out.c (reference_to_unused,
15986         premark_types_used_by_global_vars_helper): Update.
15987         * tree-eh.c (tree_could_trap_p): Update.
15988         * ipa-split.c (consider_split, execute_split_functions): Update.
15989         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
15990         has_addr_references_p): Update; move ahead in file for better
15991         readability.
15992         (process_references): Simplify.
15993         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
15994         bodies are removed.
15995         (cgraph_comdat_can_be_unshared_p): Make static.
15996         (cgraph_externally_visible_p): Update.
15997         (varpool_externally_visible_p): Update.
15998         (function_and_variable_visibility): Update.
15999         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16000         ipa_tm_mark_force_output_node): Update.
16001         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16002         estimate_edge_devirt_benefit, inline_generate_summary,
16003         inline_write_summary): Update.
16004         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16005         * ipa-prop.c (ipa_compute_jump_functions): Update.
16006         (ipa_print_node_params, ipa_prop_read_section,
16007         ipa_update_after_lto_read, read_replacements_section): Update.
16008         * varasm.c (mark_decl_referenced): Update.
16009         (assemble_alias, dump_tm_clone_pairs): Update.
16010         * tree-inline.c (copy_bb): Update.
16011         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16012         Update.
16013         * symtab.c (dump_symtab_base): Print new flags.
16014         (verify_symtab_base): Verify new flags.
16015         (symtab_alias_ultimate_target): New function.
16016         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16017         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16018         Update.
16019         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16020         Update.
16021         * i386.c (ix86_get_function_versions_dispatcher,
16022         ix86_generate_version_dispatcher_body): Update.
16023         (fold_builtin_cpu): Use varpool_add_new_variable.
16024         * varpool.c (varpool_remove_initializer): Break out from ...
16025         (varpool_remove_node): ... this one.
16026         (dump_varpool_node, varpool_node_for_asm,
16027         cgraph_variable_initializer_availability, varpool_analyze_node,
16028         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16029         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16030         Update.
16032 2013-05-29  Jan Hubicka  <jh@suse.cz>
16034         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16036 2013-05-29  Easwaran Raman  <eraman@google.com>
16038         PR tree-optimization/57442
16039         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16040         when control exits the main loop.
16042 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16044         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16045         and RX600.
16046         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16047         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16048         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16049         option.
16051 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16053         PR tree-optimization/57441
16054         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16055         Don't limit size of incr_vec to number of candidates.
16057 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16059         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16060         and mips16 directories.
16061         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16062         (MULTILIB_DIRNAMES): Ditto.
16063         (MULTILIB_EXCEPTIONS): Add new exceptions.
16064         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16065         (MULTILIB_DIRNAMES): Ditto.
16066         (MULTILIB_EXCEPTIONS): Add new exceptions.
16068 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16069             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16071         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16072         SYMBOL_TINY_ABSOLUTE.
16073         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16074         SYMBOL_TINY_ABSOLUTE.
16075         (aarch64_expand_mov_immediate): Likewise.
16076         (aarch64_classify_symbol): Likewise.
16077         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16078         Permit SYMBOL_TINY_ABSOLUTE.
16079         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16081 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16082             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16084         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16085         Refactor if/switch.  Replace gcc_assert with if.
16087 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16089         * config/i386/i386.c (initial_ix86_tune_features): Enable
16090         FP Reassociation for AMD bdver1 and bdver2.
16092 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16094         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16095         and IMAGPART_EXPR do not occur within other handled_components.
16097 2013-05-29  Richard Biener  <rguenther@suse.de>
16099         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16100         access on whether the use is in the BB we currently try to
16101         vectorize.
16102         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16103         vectorize to vect_bb_slp_scalar_cost.
16105 2013-05-29  Richard Biener  <rguenther@suse.de>
16107         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16108         computing scalar cost offsetted by stmts that are kept live
16109         by scalar uses.
16110         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16111         for computation of scalar cost.
16113 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16115         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16117 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16119         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16120         array notation built-in reduction functions.
16121         * doc/passes.texi (Passes): Added documentation about changes done
16122         for Cilk Plus.
16123         * doc/invoke.texi (C Dialect Options): Added documentation about
16124         the -fcilkplus flag.
16125         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16126         (BUILTINS_DEF): Depend on cilkplus.def.
16127         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16128         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16129         * cilkplus.def: New file.
16131 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16133         PR rtl-optimization/57439
16134         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16136 2013-05-28  Easwaran Raman  <eraman@google.com>
16138         PR tree-optimization/57337
16139         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16140         (find_insert_point): Correctly identify the insertion point
16141         when two statements with the same UID is compared.
16143 2013-05-28  Richard Biener  <rguenther@suse.de>
16145         PR tree-optimization/56787
16146         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16147         from the list of data references.
16148         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16149         clobbers.
16150         (vect_analyze_loop_operations): Likewise.
16151         (vect_transform_loop): Remove clobbers.
16153 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16155         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16156         and REALPART_EXPRs have scalar type.
16158 2013-05-28  Richard Biener  <rguenther@suse.de>
16160         PR tree-optimization/57411
16161         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16162         virtual operands.
16163         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16164         virtual operand propagation.
16166 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16168         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16169         destination register for bmasksi_vis.
16170         (vector_init_bshuffle): Likewise.
16171         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16173 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16175         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16176         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16177         mode if the instruction isn't available in the original mode.
16178         * config/sparc/sparc.opt (mfix-ut699): New option.
16179         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16180         (divdf3): Turn into expander.
16181         (divdf3_nofix): New insn.
16182         (divdf3_fix): Likewise.
16183         (divsf3): Disable if -mfix-ut699.
16184         (sqrtdf2): Turn into expander.
16185         (sqrtdf2_nofix): New insn.
16186         (sqrtdf2_fix): Likewise.
16187         (sqrtsf2): Disable if -mfix-ut699.
16189 2013-05-27  Richard Biener  <rguenther@suse.de>
16191         PR middle-end/57412
16192         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16193         block for the new loop.
16195 2013-05-27  Richard Biener  <rguenther@suse.de>
16197         PR tree-optimization/57343
16198         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16199         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16200         (number_of_iterations_cond): Do not build the folded tree.
16202 2013-05-27  Richard Biener  <rguenther@suse.de>
16204         Revert
16205         PR middle-end/57381
16206         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16207         OEP_CONSTANT_ADDRESS_OF retained.
16209         PR tree-optimization/57417
16210         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16211         for unchanged base.
16212         (set_ssa_val_to): Compare addresses using
16213         get_addr_base_and_unit_offset.
16215 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16217         PR rtl-optimization/56833
16218         * postreload.c (move2add_record_mode): New function.
16219         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16220         (move2add_use_add2_insn): Use move2add_record_sym_value.
16221         (move2add_use_add3_insn): Likewise.
16222         (reload_cse_move2add): Use move2add_valid_value_p and
16223         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16224         affected regs by setting reg_mode to VOIDmode.
16225         (move2add_note_store): Don't pretend the inside of a SUBREG is
16226         the actual destination.  Invalidate single/leading registers by
16227         setting reg_mode to VOIDmode.
16228         Use move2add_record_sym_value, move2add_valid_value_p and
16229         move2add_record_mode.
16231 2013-05-27  Richard Biener  <rguenther@suse.de>
16233         PR tree-optimization/57396
16234         * tree-affine.c (double_int_constant_multiple_p): Properly
16235         return false for val == 0 and div != 0.
16237 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16239         * config/mips/mips.h: Use #elif in preprocessor conditions.
16241 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16243         PR target/53916
16244         * config/mips/constraints.md (kl): New constraint.
16245         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16246         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16247         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16248         from LO for MIPS16.
16249         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16251 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16253         PR target/55777
16254         * config/mips/mips.c (mips_can_inline_p): New function.
16255         (TARGET_CAN_INLINE_P): Define.
16257 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16259         * sched-int.h (ds_t, dw_t): Make unsigned int.
16260         Fix documentation that describes how all the ds_t bits are used.
16261         Reserve the last bit for delayed-branch scheduling.
16262         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16263         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16264         (gen_dep_weak_1): Remove prototype.
16265         * sched-deps.c (get_dep_weak_1): Make static.
16266         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16267         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16268         * doc/tm.texi: Regenerate.
16269         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16271 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16273         PR debug/56950
16274         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16276 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16277             Sandra Loosemore  <sandra@codesourcery.com>
16279         * config.gcc (powerpc-*): Allow native for with-cpu.
16281 2013-05-24  Jeff Law  <law@redhat.com>
16283         PR tree-optimization/57124
16284         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16285         conversion feeding a condition if the range has an overflow
16286         if -fstrict-overflow.  Add warnings for when we do make the
16287         transformation.
16289 2013-05-24  Dehao Chen  <dehao@google.com>
16291         * tree-cfg.c (locus_discrim_map): Fix the typo.
16292         (locus_discrim_hasher): Likewise.
16293         (locus_discrim_hasher::hash): Likewise.
16294         (locus_discrim_hasher::equal): Likewise.
16296 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16298         PR tree-optimization/57294
16299         * cgraph.h (ipa_record_stmt_references): Declare.
16300         * cgraphbuild.c (ipa_record_stmt_references): New function.
16301         (build_cgraph_edges): Use ipa_record_stmt_references.
16302         (rebuild_cgraph_edges): Likewise.
16303         (cgraph_rebuild_references): Likewise.
16304         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16305         associated with the old statement and build references from the
16306         newly built statements.
16307         * ipa-ref.c (ipa_remove_stmt_references): New function.
16308         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16310 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16312         * lra-constraints.c (emit_spill_move): Use smaller mode for
16313         mem-mem moves.
16314         (check_and_process_move): Consider mem-reg moves for secondary
16315         too.
16316         (curr_insn_transform): Don't lose insns emitted before for
16317         secondary memory moves.
16318         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16319         reg set up in the current insn.
16321 2013-05-24  Dehao Chen  <dehao@google.com>
16323         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16324         hash function.
16325         (locus_descrim_hasher::equal): Likewise.
16326         (build_gimple_cfg): New discriminator assignment algorithm.
16327         (make_edges): Likewise.
16328         (next_discriminator_for_locus): Likewise.
16329         (same_line_p): Likewise.
16330         (assign_discriminators): Likewise.
16331         (make_cond_expr_edges): Likewise.
16332         (make_gimple_switch_edges): Likewise.
16333         (make_goto_expr_edges): Likewise.
16334         (make_gimple_asm_edges): Likewise.
16336 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16338         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16339         X format specifier to only display bottom 16 bits.
16340         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16341         immediate to match for operand 2, since it will be masked.
16343 2013-05-24  Richard Biener  <rguenther@suse.de>
16345         PR tree-optimization/57287
16346         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16347         all SSA names that occur in abnormal PHIs.
16349 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16351         PR tree-ssa/57385
16352         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16353         that index is not negative.
16355 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16357         PR rtl-optimization/55177
16358         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16359         (simplify_byte_swapping_operation): New.
16360         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16361         (simplify_relational_operation_1): Deal with BSWAP.
16363 2013-05-23  Richard Henderson  <rth@redhat.com>
16365         PR target/56742
16366         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16367         (ix86_reorg): Call it.
16369 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16371         PR target/57379
16372         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16373         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16374         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16376 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16378         PR debug/57351
16379         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16381 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16382             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16384         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16385         * config/aarch64/constraints.md (Usa): Remove.
16386         * doc/md.texi (AArch64 Usa): Remove.
16388 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16389             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16391         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16392         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16393         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16394         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16396 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16398         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16399         instruction (AdvSIMD).
16400         * config/aarch64/aarch64-builtins.c
16401         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16402         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16404 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16406         PR middle-end/57347
16407         * tree.h (contains_bitfld_component_ref_p): Declare.
16408         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16409         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16410         caller.
16411         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16412         not access a bit-field.  Assert all final offsets are byte-aligned.
16414 2013-05-23  Richard Biener  <rguenther@suse.de>
16416         PR tree-optimization/57380
16417         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16418         least one invariant or re-used load.
16419         * passes.c (init_optimization_passes): Move pass_phiprop before
16420         pass_forwprop.
16422 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16424         * config/aarch64/aarch64-simd.md
16425         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16427 2013-05-23  Richard Biener  <rguenther@suse.de>
16429         PR middle-end/57381
16430         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16431         OEP_CONSTANT_ADDRESS_OF retained.
16433 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16435         PR middle-end/57344
16436         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16437         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16439 2013-05-23  Richard Biener  <rguenther@suse.de>
16441         PR rtl-optimization/57341
16442         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16443         instead of true_dependence.
16445 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16447         * bb-reorder.c (branch_threshold): Make const.
16448         (exec_threshold): Ditto.
16450 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16451             Pat Haugen  <pthaugen@us.ibm.com>
16452             Peter Bergner  <bergner@vnet.ibm.com>
16454         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16455         documentation for the power8 crypto builtins.
16457         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16459         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16460         macros for defining power8 builtin functions.
16461         (BU_P8V_AV_2): Likewise.
16462         (BU_P8V_AV_P): Likewise.
16463         (BU_P8V_VSX_1): Likewise.
16464         (BU_P8V_OVERLOAD_1): Likewise.
16465         (BU_P8V_OVERLOAD_2): Likewise.
16466         (BU_CRYPTO_1): Likewise.
16467         (BU_CRYPTO_2): Likewise.
16468         (BU_CRYPTO_3): Likewise.
16469         (BU_CRYPTO_OVERLOAD_1): Likewise.
16470         (BU_CRYPTO_OVERLOAD_2): Likewise.
16471         (XSCVSPDP): Fix typo, point to the correct instruction.
16472         (VCIPHER): Add power8 crypto builtins.
16473         (VCIPHERLAST): Likewise.
16474         (VNCIPHER): Likewise.
16475         (VNCIPHERLAST): Likewise.
16476         (VPMSUMB): Likewise.
16477         (VPMSUMH): Likewise.
16478         (VPMSUMW): Likewise.
16479         (VPERMXOR_V2DI): Likewise.
16480         (VPERMXOR_V4SI: Likewise.
16481         (VPERMXOR_V8HI: Likewise.
16482         (VPERMXOR_V16QI: Likewise.
16483         (VSHASIGMAW): Likewise.
16484         (VSHASIGMAD): Likewise.
16485         (VPMSUM): Likewise.
16486         (VPERMXOR): Likewise.
16487         (VSHASIGMA): Likewise.
16489         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16490         __CRYPTO__ if the crypto instructions are available.
16491         (altivec_overloaded_builtins): Add support for overloaded power8
16492         builtins.
16494         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16495         support for power8 crypto builtins.
16496         (builtin_function_type): Likewise.
16497         (altivec_init_builtins): Add support for builtins that take vector
16498         long long (V2DI) arguments.
16500         * config/rs6000/crypto.md: New file, define power8 crypto
16501         instructions.
16503 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16504             Pat Haugen  <pthaugen@us.ibm.com>
16505             Peter Bergner  <bergner@vnet.ibm.com>
16507         * doc/invoke.texi (Option Summary): Add power8 options.
16508         (RS/6000 and PowerPC Options): Likewise.
16510         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16511         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16512         wm, wn, wr documentation.
16514         * config/rs6000/constraints.md (wm): New constraint for VSX
16515         registers if direct move instructions are enabled.
16516         (wn): New constraint for no registers.
16517         (wq): New constraint for quad word even GPR registers.
16518         (wr): New constraint if 64-bit instructions are enabled.
16519         (wv): New constraint if power8 vector instructions are enabled.
16520         (wQ): New constraint for quad word memory locations.
16522         * config/rs6000/predicates.md (const_0_to_15_operand): New
16523         constraint for 0..15 for crypto instructions.
16524         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16525         as GPR and floating point registers.
16526         (int_reg_operand): New predicate to match only GPR registers.
16527         (base_reg_operand): New predicate to match base registers.
16528         (quad_int_reg_operand): New predicate to match even GPR registers
16529         for quad memory operations.
16530         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16531         operations in both GPR and VSX registers.
16532         (quad_memory_operand): New predicate for quad memory operations.
16533         (reg_or_indexed_operand): New predicate for direct move support.
16535         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16536         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16537         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16538         (POWERPC_MASKS): Add power8 options.
16539         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16540         various options.
16542         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16543         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16545         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16546         (-mpower8-fusion): New power8 options.
16547         (-mpower8-fusion-sign): Likewise.
16548         (-mpower8-vector): Likewise.
16549         (-mcrypto): Likewise.
16550         (-mdirect-move): Likewise.
16551         (-mquad-memory): Likewise.
16553         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16554         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16555         registers.
16556         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16557         (rs6000_debug_vector_unit): Add p8_vector.
16558         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16559         definitions.  Also print fusion state.
16560         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16561         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16562         (rs6000_option_override_internal): Add support for power8.
16563         (rs6000_common_init_builtins): Add debugging for skipped builtins
16564         if -mdebug=builtin.
16565         (rs6000_adjust_cost): Add power8 support.
16566         (rs6000_issue_rate): Likewise.
16567         (insn_must_be_first_in_group): Likewise.
16568         (insn_must_be_last_in_group): Likewise.
16569         (force_new_group): Likewise.
16570         (rs6000_register_move_cost): Likewise.
16571         (rs6000_opt_masks): Likewise.
16573         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16574         power8 capable assembler, default to power7 options.
16575         (TARGET_DIRECT_MOVE): Likewise.
16576         (TARGET_CRYPTO): Likewise.
16577         (TARGET_P8_VECTOR): Likewise.
16578         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16579         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16580         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16581         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16582         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16583         (TARGET_XSCVDPSPN): Likewise.
16584         (TARGET_XSCVSPDPN): Likewsie.
16585         (TARGET_SYNC_HI_QI): Likewise.
16586         (TARGET_SYNC_TI): Likewise.
16587         (MASK_CRYPTO): Likewise.
16588         (MASK_DIRECT_MOVE): Likewise.
16589         (MASK_P8_FUSION): Likewise.
16590         (MASK_P8_VECTOR): Likewise.
16591         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16592         temporary used by some of the direct move instructions to get two FP
16593         temporary registers does not force creation of a stack frame.
16594         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16595         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16596         that any VSX registers are tieable, even if they are also an
16597         Altivec vector mode.
16598         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16599         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16600         (RS6000_BTM_CRYPTO): Likewise.
16601         (RS6000_BTM_COMMON): Likewise.
16603         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16604         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16605         (enum rs6000_vector): Add power8 vector support.
16607 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16609         PR target/19599
16610         PR target/57340
16611         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16612         (any_sibcall_could_use_r3): this and handle indirect calls.
16613         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16615 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16617         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16619 2013-05-22  Richard Biener  <rguenther@suse.de>
16621         PR middle-end/57349
16622         * profile.c (branch_prob): Do not split blocks that are
16623         abnormally receiving from ECF_RETURNS_TWICE functions.
16625 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16627         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16628         address mode.  Move pointer mode initialization to the same place.
16630 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16632         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16633         while it has any effect.
16635 2013-05-21  Easwaran Raman  <eraman@google.com>
16637         PR tree-optimization/57322
16638         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16639         UID of the statement added to the BB to be 1.
16641 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16643         PR tree-optimization/57331
16644         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16645         of conversion from pointer type to integral type with integer.
16647 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16649         PR lto/57289
16650         * ipa-prop.c (ipa_read_node_info): Process param_used and
16651         controlled_uses in the same order as when writing.
16653 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16655         PR plugins/56754
16656         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16658 2013-05-21  Richard Biener  <rguenther@suse.de>
16660         PR tree-optimization/57318
16661         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16662         estimate stmts with side-effects as likely eliminated.
16664 2013-05-21  Richard Biener  <rguenther@suse.de>
16666         PR tree-optimization/57330
16667         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16668         preserve the call stmts fntype.
16670 2013-05-21  Richard Biener  <rguenther@suse.de>
16672         PR tree-optimization/57303
16673         * tree-ssa-sink.c (statement_sink_location): Improve killing
16674         stmt detection and properly handle self-assignments.
16676 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16678         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16679         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16680         registers. Set register size out of the PARALLEL loop.
16682 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16684         PR target/56547
16685         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16686         (*fmasf4, *fmasf4_media): New insns.
16688 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16690         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16691         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16692         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16693         (mips_idiv_insns): Update the comments to say that the returned
16694         instruction counts are in units of BASE_INSN_LENGTH.
16695         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16696         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16697         using 2 rather than 4 as the length of indirect MIPS16 and
16698         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16699         length of a NOP.  Don't divide MIPS16 lengths by 2.
16700         (mips16_split_long_branches): Assume a branch is long if the
16701         length is greater than 4 rather than 8.
16702         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16703         rather than multiplying them by 2.  Multiply instruction counts
16704         by BASE_INSN_LENGTH rather than 4.
16705         (*jump_mips16, tls_get_tp_mips16_<mode>)
16706         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16708 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16710         * config/mips/mips.md (extended_mips16): Remove branch case.
16711         (length): Remove duplicated extended_mips16 test.
16713 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16715         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16717 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16719         * recog.h (Recog_data): Rename to...
16720         (recog_data_d): ...this.
16721         (recog_data): Update accordingly.
16722         * recog.c (recog_data): Likewise.
16723         * reload.c (save_recog_data): Likewise.
16724         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16725         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16727 2013-05-17  Julian Brown  <julian@codesourcery.com>
16729         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16730         found in a REG_EQUAL note, invalidate it.
16732 2013-05-17  Easwaran Raman  <eraman@google.com>
16734         * tree-ssa-reassoc.c (find_insert_point): New function.
16735         (insert_stmt_after): Likewise.
16736         (get_def_stmt): Likewise.
16737         (ensure_ops_are_available): Likewise.
16738         (not_dominated_by): Likewise.
16739         (rewrite_expr_tree): Do not move statements beyond what is
16740         necessary. Remove call to swap_ops_for_binary_stmt...
16741         (reassociate_bb): ... and move it here.
16742         (build_and_add_sum): Assign UIDs for new statements.
16743         (linearize_expr): Likewise.
16744         (do_reassoc): Renumber gimple statement UIDs.
16746 2013-05-17  Jan Hubicka  <jh@suse.cz>
16748         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16749         weakrefs.
16750         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16751         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16752         weakrefs.
16753         (output_weakrefs): Update.
16755 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16756             Martin Jambor  <mjambor@suse.cz>
16758         PR middle-end/57276
16759         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16760         value that corresponds to the given aggval is found in values vector.
16762 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16764         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16765         sse, sse2, sse3, ssse3 and sse4a flags to options.
16767 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16769         * gengtype-state.c: (s_expr_writer): New class, to handle
16770         prettifying of output layout of s-expressions.
16771         (state_writer): New class, to write out gtype.state.
16772         (state_written_type_count): Move this variable into member data of
16773         state_writer.
16774         (s_expr_writer::s_expr_writer): New code: constructor for new class
16775         (state_writer::state_writer(): ditto
16776         (s_expr_writer::write_new_line): New function
16777         (s_expr_writer::write_any_indent): ditto
16778         (s_expr_writer::begin_s_expr): ditto
16779         (s_expr_writer::end_s_expr): ditto
16780         (write_state_fileloc): convert to method of state_writer...
16781         (state_writer:: write_state_fileloc): ...and use methods of
16782         s_expr_writer to write indentation into the gtype.state output file
16783         to visually represent the hierarchical structure of the list
16784         structures
16785         (write_state_fields): ditto, renaming to...
16786         (state_writer::write_state_fields)
16787         (write_state_a_string): ditto, renaming to...
16788         (state_writer::write_state_a_string)
16789         (write_state_string_option): ditto, renaming to...
16790         (state_writer::write_state_string_option)
16791         (write_state_type_option): ditto, renaming to...
16792         (state_writer::write_state_type_option)
16793         (write_state_nested_option): ditto, renaming to...
16794         (state_writer::write_state_nested_option)
16795         (write_state_option): ditto, renaming to...
16796         (state_writer::write_state_option)
16797         (write_state_options): ditto, renaming to...
16798         (state_writer::write_state_options)
16799         (write_state_lang_bitmap): ditto, renaming to...
16800         (state_writer::write_state_lang_bitmap)
16801         (write_state_version): ditto, renaming to...
16802         (state_writer::write_state_version)
16803         (write_state_scalar_type): ditto, renaming to...
16804         (state_writer::write_state_scalar_type)
16805         (write_state_string_type): ditto, renaming to...
16806         (state_writer::write_state_string_type)
16807         (write_state_undefined_type): ditto, renaming to...
16808         (state_writer::write_state_undefined_type)
16809         (write_state_struct_union_type): ditto, renaming to...
16810         (state_writer::write_state_struct_union_type)
16811         (write_state_struct_type): ditto, renaming to...
16812         (state_writer::write_state_struct_type)
16813         (write_state_user_struct_type): ditto, renaming to...
16814         (state_writer::write_state_user_struct_type)
16815         (write_state_lang_struct_type): ditto, renaming to...
16816         (state_writer::write_state_lang_struct_type)
16817         (write_state_param_struct_type): ditto, renaming to...
16818         (state_writer::write_state_param_struct_type)
16819         (write_state_pointer_type): ditto, renaming to...
16820         (state_writer::write_state_pointer_type)
16821         (write_state_array_type): ditto, renaming to...
16822         (state_writer::write_state_array_type)
16823         (write_state_gc_used): ditto, renaming to...
16824         (state_writer::write_state_gc_used)
16825         (write_state_common_type_content): ditto, renaming to...
16826         (state_writer::write_state_common_type_content)
16827         (write_state_type): ditto, renaming to...
16828         (state_writer::write_state_type)
16829         (write_state_pair_list): ditto, renaming to...
16830         (state_writer::write_state_pair_list)
16831         (write_state_pair): ditto, renaming to...
16832         (state_writer::write_state_pair)
16833         (write_state_typedefs): ditto, renaming to...
16834         (state_writer::write_state_typedefs)
16835         (write_state_structures): ditto, renaming to...
16836         (state_writer::write_state_structures)
16837         (write_state_param_structs): ditto, renaming to...
16838         (state_writer::write_state_param_structs)
16839         (write_state_variables): ditto, renaming to...
16840         (state_writer::write_state_variables)
16841         (write_state_srcdir): ditto, renaming to...
16842         (state_writer::write_state_srcdir)
16843         (write_state_files_list): ditto, renaming to...
16844         (state_writer::write_state_files_list)
16845         (write_state_languages): ditto, renaming to...
16846         (state_writer::write_state_languages)
16847         (write_state): create a state_writer instance and use it when
16848         writing out the state file
16850 2013-05-17  Mike Stump  <mikestump@comcast.net>
16852         PR rtl-optimization/57304
16853         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16854         accessing DF_REF_REAL_LOC.
16856 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16858         PR rtl-optimization/57281
16859         PR rtl-optimization/57300
16860         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16861         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16862         what the other splitter did if the registers are dead.
16864 2013-05-17  Richard Biener  <rguenther@suse.de>
16866         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16867         MEM_REF offsets.
16869 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16871         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16872         linking.
16874 2013-05-17  Marek Polacek  <polacek@redhat.com>
16876         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16877         length when doing non-zero store of storing '\0' to '\0'.
16879 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16881         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16882         vect_external_def oprnd1 with loop_vinfo, try to emit
16883         optional cast, negation and and stmts on the loop preheader
16884         edge instead of into the pattern def seq.
16886         PR tree-optimization/57051
16887         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16888         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16890 2013-05-16  Nick Clifton  <nickc@redhat.com>
16892         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16893         (rl78_is_naked_func): New function.
16894         (rl78_expand_prologue): Skip prologue generation for naked functions.
16895         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16896         * doc/extend.texi (naked): Add RL78 to the list of processors
16897         that supports this attribute.
16899 2013-05-16  Jeff Law  <law@redhat.com>
16901         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16903 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16905         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16906         cache parameters using detect_caches_amd also for CYRIX,
16907         NSC and TM2 signatures.
16909 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16910             Dzianis Kahanovich  <mahatma@eu.by>
16912         PR target/45359
16913         PR target/46396
16914         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16915         VIA/Centaur processors and determine their cache parameters
16916         using detect_caches_amd.
16918 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16920         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16921         (rtl_verify_edges): New function.
16922         (rtl_verify_bb_insns): Ditto.
16923         (rtl_verify_bb_pointers): Ditto.
16924         (rtl_verify_bb_insn_chain): Ditto.
16925         (rtl_verify_fallthru): Ditto.
16926         (rtl_verify_bb_layout): Ditto.
16927         (rtl_verify_flow_info_1): Outline checks into new functions.
16928         (rtl_verify_flow_info): Ditto.
16930 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16932         * cfghooks.c (copy_bbs): Add update_dominance argument.
16933         * cfghooks.h (copy_bbs): Update prototype.
16934         * tree-cfg.c (gimple_duplicate_sese_region):
16935         Add update_dominance argument.
16936         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16937         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16938         gimple_duplicate_sese_region call.
16939         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16940         Update copy_bbs call.
16941         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16942         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16944 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16946         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16947         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
16948         vect_recog_rotate_pattern.
16949         (vect_recog_rotate_pattern): New function.
16951 2013-05-16  Jason Merrill  <jason@redhat.com>
16953         * Makefile.in (LLINKER): New variable.
16954         (mostlyclean): Remove link mutex.
16955         * configure.ac: Handle --enable-link-mutex.
16956         * lock-and-run.sh: New script.
16958 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16960         PR target/19599
16961         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
16962         for NULL decl.
16964 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16966         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
16968 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
16970         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
16971         * config/arm/arm.c (next_consecutive_mem): New function.
16972         (gen_movmem_ldrd_strd): Likewise.
16973         * config/arm/arm.md (movmemqi): Update condition and code.
16974         (unaligned_loaddi, unaligned_storedi): New patterns.
16976 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
16978         * config.gcc: Obsolete *-*-solaris2.9*.
16979         * doc/install.texi (Specific, *-*-solaris2*): Document it.
16981 2013-05-16  Richard Biener  <rguenther@suse.de>
16983         * passes.c (init_optimization_passes): Move pass_parallelize_loops
16984         earlier, after GRAPHITE transforms and IV canonicalization.
16986 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16988         * omp-low.c (extract_omp_for_data): For collapsed loops,
16989         if at least one of the loops is known at compile time to
16990         iterate zero times, set count to 0.
16991         (expand_omp_regimplify_p): New function.
16992         (expand_omp_for_generic): For collapsed loops, if at least
16993         one of the loops isn't known to iterate at least once,
16994         add runtime check with setting count to 0.
16995         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
16996         For unsigned types if it isn't known at compile time that
16997         the loop will iterate at least once, add runtime check to bypass
16998         the whole loop if initial condition isn't true.
17000 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17002         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17004 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17006         PR middle-end/57286
17007         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17008         transformations to avoid an infinite loop.
17010 2013-05-16  Marek Polacek  <polacek@redhat.com>
17012         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17014 2013-05-15  Leif Ekblad  <leif@rdos.net>
17016         * config/i386/i386.c (ix86_decompose_address): Use
17017         DEFAULT_TLS_SEG_REG to access TLS segment register.
17018         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17019         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17020         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17022 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17024         PR target/57260
17025         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17026         sibling calls to functions that would normally be lazily bound,
17027         unless $gp is call-clobbered.
17029 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17031         * config/i386/i386.c (ix86_option_override_internal): Update
17032         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17033         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17034         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17035         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17036         of TARGET_3DNOW.
17037         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17039 2013-05-15  Andreas Schwab  <schwab@suse.de>
17041         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17042         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17043         third operand.
17045 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17047         * loop-unroll.c (report_unroll_peel): Check decision before
17048         emitting unroll/peel message.
17050 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17052         * function.h (has_bb_partition): New rtl_data flag.
17053         (bb_reorder_complete): Ditto.
17054         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17055         instead of flag_reorder_blocks_and_partition.
17056         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17057         with some enhancements.
17058         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17059         * bb-reorder.c (connect_traces): Check for has_bb_partition
17060         instead of flag_reorder_blocks_and_partition.
17061         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17062         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17063         verify_hot_cold_block_grouping.
17064         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17066 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17068         PR target/19599
17069         * config/arm/predicates.md (call_insn_operand): New predicate.
17070         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17071         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17072         if insn is not a tail call.
17073         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17074         registers.
17075         * config/arm/arm.h (enum reg_class): New caller save register class.
17076         (REG_CLASS_NAMES): Likewise.
17077         (REG_CLASS_CONTENTS): Likewise.
17078         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17079         without decls.
17081 2013-05-15  Richard Biener  <rguenther@suse.de>
17083         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17084         of MSG_OPTIMIZED_LOCATIONS.
17085         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17086         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17087         message.
17088         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17089         of MSG_OPTIMIZED_LOCATIONS.
17090         (execute_vect_slp): Likewise.
17091         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17092         (vect_create_cond_for_alias_checks): Likewise.
17093         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17094         (vect_recog_widen_mult_pattern): Likewise.
17095         (vect_recog_widen_sum_pattern): Likewise.
17096         (vect_recog_over_widening_pattern): Likewise.
17097         (vect_recog_widen_shift_pattern): Likewise.
17098         (vect_recog_vector_vector_shift_pattern): Likewise.
17099         (vect_recog_divmod_pattern): Likewise.
17100         (vect_recog_mixed_size_cond_pattern): Likewise.
17101         (vect_recog_bool_pattern): Likewise.
17102         (vect_pattern_recog_1): Likewise.
17104 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17106         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17107         non-functions to builtin_unreachable.
17108         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17109         correct when new direct edges were discovered.
17111 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17113         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17114         header, print symbol order instead of node uid, print more information
17115         about indirect edge targets.
17116         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17117         uids.
17118         (ipa_make_edge_direct_to_target): Likewise.
17119         (remove_described_reference): Likewise.
17120         (propagate_controlled_uses): Likewise.
17121         (ipa_print_node_params): Also print symbol order.
17122         (ipcp_transform_function): Print symbol order instead of node uids.
17123         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17124         (cgraph_get_create_real_symbol_node): Likewise.
17125         * ipa-cp.c (print_lattice): Likewise.
17126         (print_all_lattices): Likewise.
17127         (determine_versionability): Likewise.
17128         (initialize_node_lattices): Likewise.
17129         (estimate_local_effects): Likewise.
17130         (update_profiling_info): Likewise.
17131         (create_specialized_node): Likewise.
17132         (perhaps_add_new_callers): Likewise.
17133         (decide_about_value): Likewise.
17134         (decide_whether_version_node): Likewise.
17135         (identify_dead_nodes): Likewise.
17136         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17137         (dump_inline_summary): Likewise.
17138         (estimate_node_size_and_time): Likewise.
17139         (inline_analyze_function): Likewise.
17140         * ipa-inline.c (report_inline_failed_reason): Likewise.
17141         (want_early_inline_function_p): Likewise.
17142         (edge_badness): Likewise.
17143         (update_edge_key): Likewise.
17144         (inline_small_functions): Likewise.  Add dumping of order to two other
17145         dumps.
17146         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17147         instead of node uids.
17148         (propagate_pure_const): Likewise.
17149         (propagate_pure_const): Likewise.
17150         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17151         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17152         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17153         of node uids.
17154         * tree-pretty-print.c (dump_function_header): Likewise.
17155         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17156         Print symbol order instead of node uids.
17158 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17160         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17161         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17163 2013-05-15  Richard Biener  <rguenther@suse.de>
17165         PR tree-optimization/57275
17166         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17167         return value for fail to do runtime alias checks for gather loads.
17169 2013-05-15  Jan Hubicka  <jh@suse.cz>
17171         PR lto/57038
17172         PR lto/47375
17173         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17174         weakrefs are not external.
17175         (lto_symtab_merge_decls): Fix thinko when dealing with
17176         non-lto_symtab decls.
17177         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17178         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17179         * varpool.c (dump_varpool_node): Dump more flags.
17181 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17183         * config/i386/i386.c (processor_alias_table): Add instruction
17184         FSGSBASE for AMD bdver3 architecture.
17186 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17188         * tree.c (warn_deprecated_use): Print file:line using locus color.
17189         * diagnostic.c (diagnostic_report_current_module): Print file:line
17190         and file:line:column using locus color.
17192 2013-05-14  Mike Stump  <mikestump@comcast.net>
17194         * gdbinit.in: Add __null.
17196 2013-05-14  Mike Stump  <mikestump@comcast.net>
17198         * recog.h: Rename struct recog_data to Recog_data.
17199         * recog.c: Likewise.
17200         * reload.c (can_reload_into): Likewise.
17201         * config/picochip/picochip.c: Likewise.
17203 2013-05-14  Mike Stump  <mikestump@comcast.net>
17205         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17207 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17209         * resource.h (struct resources): Remove unch_memory member.
17210         (CLEAR_RESOURCE): Don't clear unch_memory.
17211         * resource.c (mark_referenced_resources): Don't set it.
17212         (mark_set_resources): Likewise.
17213         (mark_target_live_regs): Don't clear it.
17214         (init_resource_info): Likewise.
17215         * reorg.c (resource_conflicts_p): Don't compare it.
17216         (redundant_insn): Don't set it.
17218         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17219         Remove prototypes.
17220         * emit-rtl.c (next_label): Remove unused function.
17221         (skip_consecutive_labels, link_cc0_insns): Move to ...
17222         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17223         only place where these functions are used, and make them static.
17225 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17227         * fold-const.c (fold_negate_expr): Handle vectors.
17228         (fold_truth_not_expr): Make it static.
17229         (fold_invert_truthvalue): New static function.
17230         (invert_truthvalue_loc): Handle vectors. Do not call
17231         fold_truth_not_expr directly.
17232         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17233         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17234         (fold_comparison): Handle vector constants.
17235         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17236         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17237         * tree.h (fold_truth_not_expr): Remove declaration.
17239 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17241         * config/aarch64/aarch64-simd.md
17242         (aarch64_vcond_internal<mode>): Rename to...
17243         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17244         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17245         float modes. Clarify all iterator modes.
17246         (vcond<mode><mode>): Use new name for vcond expanders.
17247         (vcond<v_cmp_result><mode>): Likewise.
17248         (vcondu<mode><mode>: Likewise.
17249         * config/aarch64/iterators.md (VDQF_COND): New.
17251 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17253         PR bootstrap/57266
17254         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17255         variable for the shift amount. Check that we shift by non-negative
17256         amounts.
17258 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17260         PR target/42017
17261         * config/arm/arm.h (EPILOGUE_USES): Only return true
17262         for LR_REGNUM after epilogue_completed.
17264 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17266         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17267         is error_mark_node.
17269 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17271         PR target/57261
17272         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17273         and Solaris 11+/x86 with gld.
17274         * configure: Regenerate.
17276 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17278         * expmed.c (expand_shift_1): Canonicalize rotates by
17279         constant bitsize / 2 to bitsize - 1.
17280         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17281         case ROTATERT>: Likewise.
17283         Revert:
17284         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17286         * config/i386/i386.md (rotateinv): New code attr.
17287         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17288         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17289         roll $31, %eax, etc.
17291 2013-05-14  Richard Biener  <rguenther@suse.de>
17293         PR middle-end/57235
17294         * tree-eh.c (sink_clobbers): Give up for successors with
17295         multiple predecessors and no virtual uses.
17297 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17299         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17300         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17302 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17304         PR middle-end/57251
17305         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17306         the case when both op0 and op1 have VOIDmode.
17308 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17310         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17311         in multiply-accumulate mode.
17313 2013-05-13  Guozhi Wei  <carrot@google.com>
17315         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17317 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17319         PR target/56975
17320         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17321         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17322         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17323         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17324         (ix86_expand_prologue): Likewise.
17325         (ix86_expand_split_stack_prologue): Likewise.
17326         (legitimate_pic_address_disp_p): Likewise.
17327         (legitimize_pic_address): Likewise.
17328         (legitimize_tls_address): Likewise.
17329         (legitimize_pe_coff_symbol): Likewise.
17330         (output_pic_addr_const): Likewise.
17331         (construct_plt_address): Likewise.
17332         (ix86_expand_call): Likewise.
17333         (x86_output_mi_thunk): Likewise.
17334         (x86_function_profiler): Likewise.
17336 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17338         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17339         similar switch cases.
17340         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17341         (aarch64_simd_mov_to_<mode>low): Delete.
17342         (aarch64_simd_mov_to_<mode>high): Delete.
17343         (move_lo_quad_<mode>): Add w<-r alternative.
17344         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17345         (aarch64_simd_mov_from_*): Update type attribute.
17346         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17347         statement.
17349 2013-05-13  Jan Hubicka  <jh@suse.cz>
17351         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17352         * config/i386/i386.c (ix86_compute_frame_layout,
17353         ix86_expand_epilogue, emit_i387_cw_initialization,
17354         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17355         ix86_local_alignment): Fix use of size/speed predicates.
17357 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17359         PR tree-optimization/45216
17360         PR tree-optimization/57157
17361         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17362         the (-Y) & (B - 1) variant if OP is |.
17363         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17364         return shifted.  Use (-op1) & (prec - 1) as other_amount
17365         instead of prec - op1.
17367 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17369         PR middle-end/42371
17370         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17371         (ipa_constant_data): New type.
17372         (ipa_jump_func): Use ipa_constant_data to hold information about
17373         constant jump functions.
17374         (ipa_get_jf_constant): Adjust to jump function type changes.
17375         (ipa_get_jf_constant_rdesc): New function.
17376         (ipa_param_descriptor): New field controlled_uses.
17377         (ipa_get_controlled_uses): New function.
17378         (ipa_set_controlled_uses): Likewise.
17379         * ipa-ref.h (ipa_find_reference): Declare.
17380         * ipa-prop.c (ipa_cst_ref_desc): New type.
17381         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17382         changes.
17383         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17384         New parameter cs.  Adjust all callers.
17385         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17386         (remove_described_reference): New function.
17387         (jfunc_rdesc_usable): Likewise.
17388         (try_make_edge_direct_simple_call): Decrement controlled use count,
17389         attempt to remove reference if it hits zero.
17390         (combine_controlled_uses_counters): New function.
17391         (propagate_controlled_uses): Likewise.
17392         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17393         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17394         (ipa_print_node_params): Print described use counter.
17395         (ipa_write_jump_function): Adjust to jump function type changes.
17396         (ipa_read_jump_function): New parameter CS, pass it to
17397         ipa_set_jf_constant.  Adjust caller.
17398         (ipa_write_node_info): Stream controlled use count
17399         (ipa_read_node_info): Likewise.
17400         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17401         asserting.
17402         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17403         count.  Remove cloning-added reference if it reaches zero.
17404         * ipa-ref.c (ipa_find_reference): New function.
17406 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17408         * config/i386/i386.c (processor_target_table): Modified default
17409         alignment values for AMD BD and BT architectures.
17411 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17413         * tree-vect-generic.c (uniform_vector_p): Move ...
17414         * tree.c (uniform_vector_p): ... here.
17415         * tree.h (uniform_vector_p): Declare it.
17416         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17417         into a scalar.
17419 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17421         PR tree-optimization/57230
17422         * tree-ssa-strlen.c (handle_char_store): Record length for
17423         array store from STRING_CST.
17425         PR tree-optimization/57230
17426         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17427         check.
17429 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17431         * config/epiphany/epiphany.c (epiphany_init): Check size of
17432         NUM_MODES_FOR_MODE_SWITCHING.
17433         (epiphany_expand_prologue):
17434         Remove CONFIG_REGNUM initial value handling code.
17435         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17436         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17437         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17438         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17439         Don't return 1 for FP_MODE_NONE.
17440         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17441         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17442         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17443         * config/epiphany/epiphany.md (save_config): New pattern.
17445 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17447         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17449 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17451         * config/i386/i386.md (memory): Handle sseishft1.
17452         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17453         (*vec_extractv2di_1): Ditto.
17455 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17457         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17458         saved registers.
17460 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17462         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17463         Add mthumb/march=armv7-a multilib.
17464         Add mthumb/march=armv7-r multilib.
17465         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17467 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17469         * config/v850/t-rtems: Add more multilibs.
17471 2013-05-10  Richard Biener  <rguenther@suse.de>
17473         PR tree-optimization/57214
17474         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17475         not propagate from SSA names that occur in abnormal PHI nodes.
17477 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17479         * stor-layout.c (element_precision): New function.
17480         * machmode.h (element_precision): Declare it.
17481         * tree.c (build_minus_one_cst): New function.
17482         (element_precision): Likewise.
17483         * tree.h (build_minus_one_cst): Declare new function.
17484         (element_precision): Likewise.
17485         * fold-const.c (operand_equal_p): Use element_precision.
17486         (fold_binary_loc): Handle vector types.
17487         * convert.c (convert_to_integer): Use element_precision.
17488         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17489         separately.
17491 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17493         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17494         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17495         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17496         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17497         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17498         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17499         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17500         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17501         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17502         (Uuw8): New constraints.
17503         (Usb4): Move into alphabetical order.
17504         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17505         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17506         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17507         previously unnamed patterns.
17508         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17509         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17510         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17511         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17512         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17513         of specific lengths.
17515 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17517         * config/i386/i386.md (rotateinv): New code attr.
17518         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17519         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17520         roll $31, %eax, etc.
17522         PR tree-optimization/45216
17523         PR tree-optimization/57157
17524         * tree-ssa-forwprop.c (simplify_rotate): New function.
17525         (ssa_forward_propagate_and_combine): Call it.
17527 2013-05-10  Richard Biener  <rguenther@suse.de>
17529         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17530         disable peeling when we version for aliasing.
17531         (vector_alignment_reachable_p): Honor explicit user alignment.
17532         (vect_supportable_dr_alignment): Likewise.
17533         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17534         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17535         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17536         then peeling to arrange for the cost-model check to come first.
17538 2013-05-10  Alan Modra  <amodra@gmail.com>
17540         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17541         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17542         * configure: Regenerate.
17544 2013-05-10  Alan Modra  <amodra@gmail.com>
17546         PR target/55033
17547         * varasm.c (default_elf_select_section): Move !DECL_P check..
17548         (get_named_section): ..to here before calling get_section_name.
17549         Adjust assertion.
17550         (default_section_type_flags): Add DECL_P check.
17551         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17552         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17554 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17556         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17557         When using gen_stack_adjust_str with a register offset, add a
17558         REG_FRAME_RELATED_EXPR note.
17560 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17562         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17563         (*vec_extractv4si_zext_mem): Ditto.
17564         (*vec_extractv2di): Add 0->x and x->x alternatives.
17565         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17566         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17568 2013-05-09  Jason Merrill  <jason@redhat.com>
17570         N3639 C++1y VLA support
17571         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17572         DECL_VALUE_EXPR.
17574         * tree.c (build_constructor_va): New.
17575         * tree.h: Declare it.
17577 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17579         PR lto/57084
17580         * gimple-fold.c (canonicalize_constructor_val): Call
17581         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17583 2013-05-09  Jan Hubicka  <jh@suse.cz>
17584             Richard Biener  <rguenther@suse.de>
17586         PR lto/54095
17587         * symtab.c (symtab_make_decl_local): Do not add private names.
17589 2013-05-09  Jan Hubicka  <jh@suse.cz>
17591         PR lto/54095
17592         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17593         (unlink_from_assembler_name_hash): Likewise.
17594         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17595         symtab_unregister_node, symtab_initialize_asm_name_hash,
17596         change_decl_assembler_name): Update.
17598 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17600         * config/aarch64/aarch64.md: New movtf split.
17601         (*movtf_aarch64): Update.
17602         (aarch64_movdi_tilow): Handle TF modes and rename to
17603         aarch64_movdi_<mode>low.
17604         (aarch64_movdi_tihigh): Handle TF modes and rename to
17605         aarch64_movdi_<mode>high
17606         (aarch64_movtihigh_di): Handle TF modes and rename to
17607         aarch64_mov<mode>high_di
17608         (aarch64_movtilow_di): Handle TF modes and rename to
17609         aarch64_mov<mode>low_di
17610         (aarch64_movtilow_tilow): Remove spurious whitespace.
17611         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17612         splits.
17613         (aarch64_print_operand): Update.
17615 2013-05-09  Alan Modra  <amodra@gmail.com>
17617         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17618         powerpc64le.
17619         * configure: Regenerate.
17621 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17623         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17624         splitter preparation statements.
17625         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17626         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17627         adjust_address_nv.
17629 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17631         * gimple-ssa-strength-reduction.c (count_candidates): Change
17632         return value to int.
17633         (analyze_candidates_and_replace): Change type of length to int.
17635 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17637         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17638         (*vec_extract<mode>): Use VI12_128 mode iterator.
17639         (*vec_extract<mode>_mem): Ditto.
17640         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17641         attribute.
17643 2013-05-08  Diego Novillo  <dnovillo@google.com>
17645         PR bootstrap/54659
17647         Revert:
17648         2012-08-17  Diego Novillo  <dnovillo@google.com>
17650         PR bootstrap/54281
17651         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17652         * config.in: Regenerate.
17653         * configure: Regenerate.
17654         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17656 2013-05-08  Jan Hubicka  <jh@suse.cz>
17658         PR lto/54095
17659         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17660         * cgraph.h (symtab_node_base): Add unique_name.
17661         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17662         input_overwrite_node, input_varpool_node): Stream unique_name.
17663         * cgraphclones.c (cgraph_create_virtual_clone,
17664         cgraph_function_versioning): Set unique_name.
17665         * ipa.c (function_and_variable_visibility): Set unique_name.
17667 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17669         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17670         (alloc_cand_and_find_basis): Restrict conditional candidate
17671         processing to CAND_MULTs.
17673 2013-05-08  Jan Hubicka  <jh@suse.cz>
17675         PR lto/54095
17676         lto-symtab.c (lto_symtab_symbol_p): New function.
17677         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17678         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17679         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17680         Skip static symbols.
17682 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17684         PR tree-optimization/57200
17685         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17686         Only call inform if the preceding warning_at returns true.
17688 2013-05-07  Han Shen  <shenhan@google.com>
17690         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17691         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17692         * common.opt (fstack-protector-strong): New option.
17693         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17694         * doc/invoke.texi (Optimization Options): Document
17695         "-fstack-protector-strong".
17696         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17698 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17700         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17702 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17704         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17705         it is not lhs of the insn.
17706         (setup_reg_equiv): Remove insn having equiv memory which it is not
17707         lhs of the insn.
17708         * lra-constraints.c (process_address): Try to improve generation
17709         code for address base + disp.
17710         (lra_constraints): Make correct the code for checking insn setting
17711         up backward equivalence.  Remove insn only if it is in the init
17712         insn list.
17713         * lra-eliminations.c (update_reg_eliminate): Change return value.
17714         (lra_eliminate): Use the result.
17716 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17718         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17719         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17720         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17721         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17722         register target operands.
17723         (*vec_extractv8hi_sse2): New pattern.
17724         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17725         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17726         (*vec_extract<mode>_mem): New insn and split pattern.
17728 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17730         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17731         (TARGET_ASAN_SHADOW_OFFSET): Define.
17732         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17733         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17735 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17737         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17738         (incr_vec_index): Return -1 if increment not found.
17739         (create_add_on_incoming_edge): Assert if increment not found.
17740         (record_increment): Limit number of increments recorded.
17741         (all_phi_incrs_profitable): Return false if an increment not found.
17742         (replace_profitable_candidates): Don't process increments that were
17743         not recorded.
17744         (analyze_candidates_and_replace): Limit size of incr_vec.
17746 2013-05-07  Richard Biener  <rguenther@suse.de>
17748         * calls.c (special_function_p): setjmp-like functions are leaf.
17749         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17750         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17752 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17754         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17755         (aarch64_simd_mov<mode>): New expander.
17756         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17757         (aarch64_simd_mov_to_<mode>high): Likewise.
17758         (aarch64_simd_mov_from_<mode>low): Likewise.
17759         (aarch64_simd_mov_from_<mode>high): Likewise.
17760         (aarch64_dup_lane<mode>): Update.
17761         (aarch64_dup_lanedi): New instruction pattern.
17762         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17763         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17765 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17767         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17768         (replace_mult_candidate): Remove unnecessary argument; remove
17769         unnecessary parameter from call to introduce_cast_before_cand.
17770         (replace_unconditional_candidate): Remove unnecessary parameter
17771         from call to replace_mult_candidate.
17772         (replace_conditional_candidate): Likewise.
17773         (insert_initializers): Use make_temp_ssa_name.
17774         (introduce_cast_before_cand): Remove unnecessary argument; use
17775         make_temp_ssa_name.
17776         (replace_one_candidate): Remove unnecessary argument; remove
17777         unnecessary parameter from calls to introduce_cast_before_cand.
17778         (replace_profitable_candidates): Remove unnecessary parameters
17779         from calls to replace_one_candidate.
17781 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17783         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17784         phi def as possibly hiding a basis for a CAND_ADD whose operands
17785         have been commuted in the analysis.
17786         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17788 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17790         * config/aarch64/aarch64.md
17791         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17792         shift value between 0-4.
17794 2013-05-07  Richard Biener  <rguenther@suse.de>
17796         * double-int.h (rshift): New overload.
17797         * double-int.c (rshift): New function.
17798         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17799         (create_reference_ops_from_ref): Remove.
17800         (vn_reference_insert): Use shared ops for constructing the
17801         reference and copy it.
17803 2013-05-07  Richard Biener  <rguenther@suse.de>
17805         PR middle-end/57190
17806         * tree-eh.c (sink_clobbers): Properly propagate
17807         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17809 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17811         PR tree-optimization/57149
17812         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17813         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17814         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17815         uninit_undefined_value_p instead of ssa_undefined_value_p.
17817         PR debug/57184
17818         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17819         for modifier == EXPAND_INITIALIZER.
17821 2013-05-07  Anton Blanchard  <anton@samba.org>
17823         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17824         for powerpc64 little endian.
17825         * configure: Regenerate.
17827 2013-05-06  Graham Stott  <grahams@btinternet.com>
17829         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17830         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17831         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17832         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17834 2013-05-06  Graham Stott  <grahams@btinternet.com>
17836         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17837         codes which allow non-lvalues.
17839 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17841         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17842         components are all 1s.
17843         (integer_minus_onep): New function.
17844         * tree.h (integer_minus_onep): Declare it.
17845         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17846         integer_minus_onep instead of integer_all_onesp.
17848 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17850         PR target/52933
17851         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17852         variations of these patterns.
17854 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17856         * config/i386/i386.md (isa): Add x64_sse4 member.
17857         (enabled): Handle x64_sse4.
17858         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17859         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17860         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17861         instruction for SSE4_1 targets.  Update insn attributes.
17862         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17863         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17864         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17865         const_1 selector.
17866         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17867         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17868         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17870 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17872         PR target/57108
17873         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17875 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17877         * final.c (do_assembler_dialects): Don't handle curly braces and
17878         vertical bar escaped by % as dialect delimiters.
17879         (output_asm_insn): Print curly braces and vertical bar if escaped
17880         by % and ASSEMBLER_DIALECT defined.
17881         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17882         * doc/tm.texi: Regenerated.
17884 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17886         * config/mips/mips.c: Include tree-pass.h.
17887         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17888         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17889         (pass_mips_machine_reorg2): New machine specific pass.
17890         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17891         (mips_option_override): Register the new pass.
17892         * rtl.h (cleanup_barriers): Remove prototype.
17893         (dbr_schedule): Likewise.
17894         * jump.c (cleanup_barriers): Make static.
17895         * reorg.c (dbr_schedule): Likewise.
17897 2013-05-06  Richard Biener  <rguenther@suse.de>
17899         PR tree-optimization/57185
17900         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17901         SSA names properly.
17903 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17905         PR target/57106
17906         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17907         between "first_arg" and "insn", not "last" and "insn".
17909 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17911         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17912         (find_candidates_in_block): Re-enable slsr_process_phi.
17913         (create_phi_basis): Fix double counting of candidate adjustment.
17915 2013-05-06  Richard Biener  <rguenther@suse.de>
17917         PR middle-end/57147
17918         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17919         the edge is also fallthru, preserve it and just clear the
17920         abnormal flag.
17921         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17922         also complex, preserve that and just clear the fallthru flag.
17923         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17924         update virtual operands.
17926 2013-05-06  Alan Modra  <amodra@gmail.com>
17928         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17929         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17930         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17931         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17932         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17933         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17935 2013-05-06  Alan Modra  <amodra@gmail.com>
17937         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17938         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17939         (DEFAULT_ASM_ENDIAN): Define.
17940         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17941         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17942         Update -K PIC clause from sysv4.h.
17943         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17944         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17946 2013-05-06  Alan Modra  <amodra@gmail.com>
17948         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
17949         twice for little-endian.
17950         (ashrdi3_no_power, ashrdi3): Support little-endian.
17952 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17954         PR target/55303
17955         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
17956         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
17957         related expanders.
17958         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
17959         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
17960         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
17961         New predicates.
17963 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
17964             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
17966         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
17967         * config/pa/pa.opt: Make mbig-switch a no-op.
17968         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
17969         (CASE_VECTOR_MODE): Always return SImode.
17970         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
17971         for the !TARGET_BIG_SWITCH case.
17972         * config/pa/pa-linux.h: Likewise.
17973         * config/pa/pa-openbsd.h: Likewise.
17974         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
17975         * config/pa/pa.md (short_jump): Remove define_insn.
17976         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
17977         (casesi0): Remove define_insn.
17978         (type): Remove btable_branch.
17979         (pa_combine_type): Likewise.
17980         (in_nullified_branch_delay): Likewise.
17981         (in_call_delay): Likewise.
17982         (define_delay): Likewise.
17983         (define_insn_reservation "Z3"): Likewise.
17984         (define_insn_reservation "Z4"): Likewise.
17985         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
17986         (pa_adjust_insn_length): Remove adjustment for btable branches.
17987         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
17988         and mno-big-switch
17990 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
17992         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17993         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
17994         Add m->r,x alternatives.
17995         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
17996         splitters using SWI48x mode iterator.
17997         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
17998         TARGET_64BIT.  Add m->x alternative.
17999         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18000         Add o->x alternative.  Enable for TARGET_SSE.
18001         (sse_storeq): Remove expander.
18002         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18003         with memory input operand.
18004         (*vec_extractv2di_1 splitter): New.
18005         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18006         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18008 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18010         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18011         (INT_LOWPART): Delete.
18012         (extract_MB): Adjust.
18013         (extract_ME): Adjust.
18014         (print_operand): Adjust.
18016 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18018         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18019         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18020         (reg_or_logical_cint_operand, easy_fp_constant,
18021         logical_const_operand): Delete "CONST_DOUBLE" case.
18022         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18023         "HOST_BITS_PER_WIDE_INT == 64" test.
18024         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18025         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18026         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18027         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18028         test.
18029         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18030         CONST_DOUBLE DImode/VOIDmode case.
18031         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18032         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18033         CONST_DOUBLE VOIDmode case.
18034         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18035         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18036         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18037         Delete CONST_DOUBLE case.
18038         (splitters for mov FMOVE64 const_double): Delete
18039         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18040         "HOST_BITS_PER_WIDE_INT >= 64" test.
18041         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18042         case.
18043         (mov DI const_double): Delete.
18045 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18047         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18048         on op shows all bits zero in mode of a lowpart subreg, return zero.
18050 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18052         PR target/57150
18053         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18054         to save TFmode registers and DImode to save TImode registers for
18055         caller save operations.
18056         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18057         mark being partially clobbered since they only use the first
18058         double word.
18060         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18061         and TDmode only use the upper 64-bits of each VSX register.
18063 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18065         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18066         (find_candidates_in_block): Disable slsr_process_phi.
18068 2013-05-03  Guozhi Wei  <carrot@google.com>
18070         * coverage.c (coverage_obj_init): Move the construction of gcov
18071         constructor to ...
18072         (build_init_ctor): ... here.
18074 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18076         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18077         (slsr_cand_d): Redefine def_phi.
18078         (stride_status, phi_adjust_status, count_phis_status): New enums.
18079         (find_phi_def): New.
18080         (find_basis_for_base_expr): New.
18081         (find_basis_for_candidate): Handle hidden bases.
18082         (alloc_cand_and_find_basis): Handle phi candidates.
18083         (slsr_process_phi): New.
18084         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18085         (create_mul_imm_cand): Likewise.
18086         (create_add_ssa_cand): Exclude phi base candidates.
18087         (create_add_imm_cand): Likewise.
18088         (slsr_process_cast): Likewise.
18089         (slsr_process_copy): Likewise.
18090         (find_candidates_in_block): Handle phi candidates.
18091         (dump_candidate): Likewise.
18092         (unconditional_cands): Delete.
18093         (unconditional_cands_with_known_stride_p): Delete.
18094         (phi_dependent_cand_p): New.
18095         (cand_increment): Handle phi-dependent candidates.
18096         (replace_dependent): Delete.
18097         (replace_mult_candidate): New.
18098         (replace_unconditional_candidate): New.
18099         (incr_vec_index): Move to avoid forward reference.
18100         (create_add_on_incoming_edge): New.
18101         (create_phi_basis): New.
18102         (replace_dependents): Delete.
18103         (replace_conditional_candidate): New.
18104         (phi_add_costs): New.
18105         (replace_uncond_cands_and_profitable_phis): New.
18106         (record_increment): Handle phi adjustments.
18107         (record_phi_increments): New.
18108         (record_increments): Handle phi adjustments.
18109         (phi_incr_cost): New.
18110         (lowest_cost_path): Handle phis.
18111         (total_savings): Likewise.
18112         (analyze_increments): Likewise.
18113         (ncd_with_phi): New.
18114         (ncd_of_cand_and_phis): New.
18115         (nearest_common_dominator_for_cands): Handle phi increments.
18116         (all_phi_incrs_profitable): New.
18117         (replace_profitable_candidates): Handle phi-dependent candidates.
18118         (analyze_candidates_and_replace): Likewise.
18120 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18122         PR bootstrap/57154
18123         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18124         do not exceed REG_BR_PROB_BASE.
18126 2013-05-03  Jeff Law  <law@redhat.com>
18128         PR tree-optimization/57144
18129         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18130         operand of the condition will bit into the new type when eliminating
18131         a cast feeding a condition.
18133 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18135         PR rtl-optimization/57130
18136         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18137         of COMPARE as in_code to the recursive call if needed.
18139 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18141         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18142         (enabled): Handle new members.
18143         * config/i386/sse.md (*vec_concatv2si): Merge from
18144         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18145         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18147 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18149         PR tree-optimization/57027
18150         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18151         for fnms opportunity, check we got the prerequisite kind
18152         of tree / gimple before using accessor functions.
18154 2013-05-03  Richard Biener  <rguenther@suse.de>
18156         * double-int.h (lshift): New overload without precision
18157         and arith argument.
18158         (operator *=, operator +=, operator -=): Move ...
18159         * double-int.c (operator *=, operator +=, operator -=): ... here
18160         and implement more efficiently.
18161         (mul_double_with_sign): Remove.
18162         (lshift_double): Adjust to take unsinged shift argument, push
18163         dispatching code to callers.
18164         (mul_double_wide_with_sign): Add early out for callers that
18165         are not interested in high parts or overflow.
18166         (lshift): New function.
18167         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18168         dispatch code here.
18169         (lrotate, rrotate): Use logical shifts.
18170         * expr.c (get_inner_reference): Use lshift.
18171         * fixed-value.c (do_fixed_divide): Likewise.
18172         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18173         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18174         (indirect_refs_may_alias_p): Likewise.
18175         (stmt_kills_ref_p_1): Likewise.
18177 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18179         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18181 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18183         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18184         scalar form of FABD instruction.
18186 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18188         * lra-constraints.c (process_alt_operands): Add checking alt
18189         number to choose the best alternative.
18191 2013-05-02  Richard Biener  <rguenther@suse.de>
18193         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18194         bitmap and its handling.
18195         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18197 2013-05-02  Richard Biener  <rguenther@suse.de>
18199         PR middle-end/57140
18200         * tree-inline.c (copy_loops): Properly handle removed loops.
18201         (copy_cfg_body): Mark destination loops for fixup if source
18202         loops needed fixup.
18204 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18206         PR target/56732
18207         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18208         generating simple_return for naked functions.
18210 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18212         PR middle-end/56988
18213         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18214         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18215         flags match.
18216         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18217         ipa_agg_replacement_value structures.
18218         (known_aggs_to_agg_replacement_list): Likewise.
18219         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18220         (read_agg_replacement_chain): Likewise.
18221         (ipcp_transform_function): Also check that by_ref flags match.
18223 2013-05-02  Richard Biener  <rguenther@suse.de>
18225         * graphds.h (struct graph): Add obstack member.
18226         * graphds.c (new_graph): Initialize obstack and allocate
18227         vertices from it.
18228         (add_edge): Allocate edge from the obstack.
18229         (free_graph): Free the obstack instead of all edges and vertices.
18231 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18233         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18234         divides.
18235         * cfg.c (update_bb_profile_for_threading): Ditto.
18236         * tree-inline.c (copy_bb): Ditto.
18237         (copy_edges_for_bb): Ditto.
18238         (initialize_cfun): Ditto.
18239         (copy_cfg_body): Ditto.
18240         (expand_call_inline): Ditto.
18241         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18242         (estimate_node_size_and_time): Ditto.
18243         (inline_merge_summary): Ditto.
18244         * cgraphclones.c (cgraph_clone_edge): Ditto.
18245         (cgraph_clone_node): Ditto.
18246         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18247         (compute_trg_info): Ditto.
18249 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18251         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18252         S reg when fp attribute set.
18253         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18255 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18257         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18258         New pattern.
18259         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18260         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18261         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18263 2013-05-02  Richard Biener  <rguenther@suse.de>
18265         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18266         (struct instantiate_cache_entry): New type.
18267         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18268         (struct instantiate_cache_type): New type.
18269         (set_instantiated_value, get_instantiated_value): Remove.
18270         (get_instantiated_value_entry): New function.
18271         (instantiate_scev_name): Use the new cache and adjust.
18272         (instantiate_scev_poly): Adjust.
18273         (instantiate_scev_binary): Likewise.
18274         (instantiate_array_ref): Likewise.
18275         (instantiate_scev_convert): Likewise.
18276         (instantiate_scev_not): Likewise.
18277         (instantiate_scev_3): Likewise.
18278         (instantiate_scev_2): Likewise.
18279         (instantiate_scev_r): Likewise.
18280         (instantiate_scev): Likewise.
18281         (resolve_mixers): Likewise.
18283 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18285         PR target/57091
18286         * lra-constraints.c (best_small_class_operands_num): Remove.
18287         (process_alt_operands): Remove small_class_operands_num.  Take
18288         small classes operands into losers and only if the operand is not
18289         matched.  Modify debugging output.
18290         (curr_insn_transform): Remove best_small_class_operands_num.
18291         Print insn name.
18293 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18295         * config/aarch64/aarch64-builtins.c
18296         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18297         * config/aarch64/aarch64-simd-builtins.def
18298         (reduc_splus_): Add new modes.
18299         (reduc_uplus_): New.
18300         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18301         (reduc_uplus_v4sf): Likewise.
18302         (reduc_splus_v4sf): Likewise.
18303         (aarch64_addv<mode>): Likewise.
18304         (reduc_uplus_<mode>): Likewise.
18305         (reduc_splus_<mode>): Likewise.
18306         (aarch64_addvv2di): Likewise.
18307         (reduc_uplus_v2di): Likewise.
18308         (reduc_splus_v2di): Likewise.
18309         (aarch64_addvv2si): Likewise.
18310         (reduc_uplus_v2si): Likewise.
18311         (reduc_splus_v2si): Likewise.
18312         (reduc_<sur>plus_<mode>): New.
18313         (reduc_<sur>plus_v2di): Likewise.
18314         (reduc_<sur>plus_v2si): Likewise.
18315         (reduc_<sur>plus_v4sf): Likewise.
18316         (aarch64_addpv4sf): Likewise.
18317         * config/aarch64/arm_neon.h
18318         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18319         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18320         add UNSPEC_SADDV, UNSPEC_UADDV.
18321         (SUADDV): New.
18322         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18324 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18326         * config/aarch64/arm_neon.h
18327         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18329 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18331         * config/aarch64/aarch64-builtins
18332         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18334 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18336         * config/aarch64/aarch64-simd-builtins.def
18337         (reduc_smax_): New.
18338         (reduc_smin_): Likewise.
18339         (reduc_umax_): Likewise.
18340         (reduc_umin_): Likewise.
18341         (reduc_smax_nan_): Likewise.
18342         (reduc_smin_nan_): Likewise.
18343         (fmax): Remove.
18344         (fmin): Likewise.
18345         (smax): Update for V2SF, V4SF and V2DF modes.
18346         (smin): Likewise.
18347         (smax_nan): New.
18348         (smin_nan): Likewise.
18349         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18350         (<su><maxmin><mode>3): ...This, refactor.
18351         (s<maxmin><mode>3): New.
18352         (<maxmin_uns><mode>3): Likewise.
18353         (reduc_<maxmin_uns>_<mode>): Refactor.
18354         (reduc_<maxmin_uns>_v4sf): Likewise.
18355         (reduc_<maxmin_uns>_v2si): Likewise.
18356         (aarch64_<fmaxmin><mode>: Remove.
18357         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18358         new builtin names.
18359         (vmin<q>_f<32,64>): Likewise.
18360         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18361         (FMAXMIN): New.
18362         (su): Add mappings for smax, smin, umax, umin.
18363         (maxmin): New.
18364         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18365         (FMAXMIN): Rename as...
18366         (FMAXMIN_UNS): ...This.
18367         (maxminv): Remove.
18368         (fmaxminv): Likewise.
18369         (fmaxmin): Likewise.
18370         (maxmin_uns): New.
18371         (maxmin_uns_op): Likewise.
18373 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18375         * config/aarch64/arm_neon.h
18376         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18377         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18378         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18380 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18382         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18383         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18385 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18387         * config/aarch64/aarch64-simd.md
18388         (vcond<mode>_internal): Handle special cases for constant masks.
18389         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18390         (vcondu<mode><mode>): Likewise.
18391         (vcond<v_cmp_result><mode>): New.
18393 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18395         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18396         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18397         * config/aarch64/aarch64-simd-builtins.def
18398         (cmeq): Update to BUILTIN_VALLDI.
18399         (cmgt): Likewise.
18400         (cmge): Likewise.
18401         (cmle): Likewise.
18402         (cmlt): Likewise.
18403         * config/aarch64/arm_neon.h
18404         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18405         to builtins or C as appropriate.
18407 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18409         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18410         (cmgeu): ...This.
18411         (cmhi): Rename to...
18412         (cmgtu): ...This.
18413         * config/aarch64/aarch64-simd.md
18414         (simd_mode): Add SF.
18415         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18416         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18417         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18418         (cstore<mode>_neg): ...This.
18419         * config/aarch64/iterators.md
18420         (VALLF): new.
18421         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18422         (COMPARISONS): New.
18423         (UCOMPARISONS): Likewise.
18424         (optab): Add missing comparisons.
18425         (n_optab): New.
18426         (cmp_1): Likewise.
18427         (cmp_2): Likewise.
18428         (CMP): Likewise.
18429         (cmp): Remove.
18430         (VCMP_S): Likewise.
18431         (VCMP_U): Likewise.
18432         (V_cmp_result): Add DF, SF modes.
18433         (v_cmp_result): Likewise.
18434         (v): Likewise.
18435         (vmtype): Likewise.
18436         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18438 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18440         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18441         define_insn to define_insn_and_split.
18442         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18443         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18444         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18445         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18446         (thumb2_negscc): Likewise.
18448 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18450         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18452 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18454         * config/arm/thumb2.md: Remove trailing whitespaces.
18456 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18458         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18459         Use gen_int_mode rather than GEN_INT.
18461 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18463         * value-prof.c (stream_in_histogram_value): Remove the strayed
18464         debug_gimple_stmt.
18466 2013-04-30  Richard Biener  <rguenther@suse.de>
18468         PR middle-end/57122
18469         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18471 2013-04-30  Richard Biener  <rguenther@suse.de>
18473         PR middle-end/57107
18474         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18476 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18478         PR rtl-optimization/56957
18479         PR rtl-optimization/57105
18480         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18481         variable.  Use just INSN_UID for determining whether an insn
18482         should be only disconnected from the insn stream.
18483         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18485 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18487         PR tree-optimization/57104
18488         * tsan.c (instrument_expr): Don't instrument accesses to
18489         DECL_HARD_REGISTER VAR_DECLs.
18491 2013-04-30  Richard Biener  <rguenther@suse.de>
18493         * function.h (loops_for_fn): New inline function.
18494         (set_loops_for_fn): Likewise.
18495         * cfgloop.h (place_new_loop): Add struct function parameter.
18496         (get_loop): Likewise.
18497         (get_loops): Likewise.
18498         (number_of_loops): Likewise.
18499         (fel_next): Adjust.
18500         (fel_init): Likewise.
18501         * cfg.c (get_loop_copy): Adjust.
18502         * cfgloop.c (flow_loops_dump): Likewise.
18503         (record_loop_exits): Likewise.
18504         (verify_loop_structure): Likewise.
18505         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18506         (estimate_reg_pressure_cost): Likewise.
18507         (mark_loop_exit_edges): Likewise.
18508         * cfgloopmanip.c (place_new_loop): Likewise.
18509         (add_loop): Likewise.
18510         (duplicate_loop): Likewise.
18511         * graph.c (draw_cfg_nodes): Likewise.
18512         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18513         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18514         (extract_affine_chrec): Likewise.
18515         (build_scop_iteration_domain): Likewise.
18516         * graphite.c (graphite_initialize): Likewise.
18517         * ira-build.c (create_loop_tree_nodes): Likewise.
18518         (more_one_region_p): Likewise.
18519         (rebuild_regno_allocno_maps): Likewise.
18520         (mark_loops_for_removal): Likewise.
18521         (mark_all_loops_for_removal): Likewise.
18522         (remove_unnecessary_regions): Likewise.
18523         (ira_build): Likewise.
18524         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18525         * loop-init.c (fix_loop_structure): Likewise.
18526         (gate_rtl_move_loop_invariants): Likewise.
18527         (gate_rtl_unswitch): Likewise.
18528         (gate_rtl_unroll_and_peel_loops): Likewise.
18529         (rtl_doloop): Likewise.
18530         * lto-streamer-in.c (input_cfg): Likewise.
18531         * lto-streamer-out.c (output_cfg): Likewise.
18532         * modulo-sched.c (sms_schedule): Likewise.
18533         * predict.c (tree_estimate_probability): Likewise.
18534         (tree_estimate_probability_driver): Likewise.
18535         (estimate_loops): Likewise.
18536         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18537         (move_sese_region_to_fn): Likewise.
18538         (debug_loop_num): Likewise.
18539         * tree-chrec.c (chrec_evaluate): Likewise.
18540         (hide_evolution_in_other_loops_than_loop): Likewise.
18541         (chrec_component_in_loop_num): Likewise.
18542         (reset_evolution_in_loop): Likewise.
18543         (evolution_function_is_invariant_rec_p): Likewise.
18544         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18545         * tree-inline.c (copy_loops): Likewise.
18546         (copy_cfg_body): Likewise.
18547         (tree_function_versioning): Likewise.
18548         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18549         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18550         Likewise.
18551         (add_to_evolution_1): Likewise.
18552         (scev_const_prop): Likewise.
18553         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18554         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18555         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18556         (tree_ssa_lim_initialize): Likewise.
18557         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18558         (verify_loop_closed_ssa): Likewise.
18559         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18560         (tree_ssa_loop_im): Likewise.
18561         (tree_ssa_loop_unswitch): Likewise.
18562         (tree_vectorize): Likewise.
18563         (check_data_deps): Likewise.
18564         (tree_ssa_loop_ivcanon): Likewise.
18565         (tree_ssa_loop_bounds): Likewise.
18566         (tree_complete_unroll): Likewise.
18567         (tree_complete_unroll_inner): Likewise.
18568         (tree_parallelize_loops): Likewise.
18569         (tree_ssa_loop_prefetch): Likewise.
18570         (tree_ssa_loop_ivopts): Likewise.
18571         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18572         * tree-vectorizer.c (vectorize_loops): Likewise.
18574 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18576         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18577         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18578         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18579         with EABI_LINK_SPEC.
18581 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18583         PR target/44578
18584         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18585         alternative.
18587 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18589         PR target/57097
18590         * lra-constraints.c (process_alt_operands): Discourage a bit more
18591         using memory for pseudos.  Print cost dump for alternatives.
18592         Modify cost values for conflicts with early clobbers.
18593         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18595 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18597         PR target/57098
18598         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18600 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18602         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18603         from/to S register.
18604         (movdi_aarch64): Support LDR/STR from/to D register.
18606 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18608         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18609         or higher by default.
18611 2013-04-29  Richard Biener  <rguenther@suse.de>
18613         PR middle-end/57075
18614         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18615         even if not adding abnormal edges for calls that can make
18616         abnormal gotos.
18618 2013-04-29  Richard Biener  <rguenther@suse.de>
18620         PR middle-end/57103
18621         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18622         TREE_BLOCK.
18623         (move_stmt_r): Remove redundant checking.
18625 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18627         PR bootstrap/57077
18628         * basic-block.h (apply_scale): New function.
18629         (apply_probability): Use apply_scale.
18630         * gimple-streamer-in.c (input_bb): Ditto.
18631         * lto-streamer-in.c (input_cfg): Ditto.
18632         * lto-cgraph.c (merge_profile_summaries): Ditto.
18633         * tree-optimize.c (execute_fixup_cfg): Ditto.
18634         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18635         (copy_edges_for_bb): Ditto.
18636         (copy_cfg_body): Ditto.
18638 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18640         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18641         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18642         (tail_merge_optimize): Handle current_loops == NULL.
18644 2013-04-26  Jeff Law  <law@redhat.com>
18646         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18647         (simplify_cond_using_ranges): Generalize code to simplify
18648         COND_EXPRs where one argument is a constant and the other
18649         is an SSA_NAME created by an integral type conversion.
18651 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18653         * config/arm/arm.md (store_minmaxsi): Use only when
18654         optimize_insn_for_size_p.
18656 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18658         PR target/57108
18659         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18661 2013-04-29  Richard Biener  <rguenther@suse.de>
18663         PR middle-end/57089
18664         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18665         loop tree make sure to schedule a fixup for the child as well.
18666         (expand_omp_for_generic): Properly add loops.
18667         (expand_omp_for_static_nochunk): Likewise.
18668         (expand_omp_for_static_chunk): Likewise.
18669         (expand_omp_for): For the degenerate case fixup loops.
18670         (expand_omp_sections): Fix default bb placement in loops.
18671         (expand_omp_atomic_pipeline): Properly add loops.
18673 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18675         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18677 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18679         * tree-ssa-tail-merge.c: Update header comment.
18681 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18683         * config/aarch64/arm_neon.h
18684         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18685         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18686         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18687         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18688         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18689         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18690         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18691         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18693 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18695         * config/aarch64/aarch64-simd.md
18696         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18697         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18698         fix_trunc, fixuns_trunc.
18699         (ftrunc<VDQF:mode>2): New.
18700         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18701         (fix_trunc_optab): New.
18703 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18705         * config/aarch64/aarch64-builtins.c
18706         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18707         iceilf, lround, iroundf.
18709 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18711         PR target/54349
18712         * config/i386/i386.h (enum ix86_tune_indices)
18713         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18714         New, split from X86_TUNE_INTER_UNIT_MOVES.
18715         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18716         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18717         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18718         (TARGET_INTER_UNIT_MOVES): Remove.
18719         * config/i386/i386.c (initial_ix86_tune_features): Update.
18720         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18721         (ix86_expand_convert_uns_didf_sse): Use
18722         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18723         (ix86_expand_vector_init_one_nonzero): Ditto.
18724         (ix86_expand_vector_init_interleave): Ditto.
18725         (inline_secondary_memory_needed): Return true for moves from SSE class
18726         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18727         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18728         * config/i386/constraints.md (Yi, Ym): Depend on
18729         TARGET_INTER_UNIT_MOVES_TO_VEC.
18730         (Yj, Yn): New constraints.
18731         * config/i386/i386.md (*movdi_internal): Change constraints of
18732         operand 1 from Yi to Yj and from Ym to Yn.
18733         (*movsi_internal): Ditto.
18734         (*movdf_internal): Ditto.
18735         (*movsf_internal): Ditto.
18736         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18737         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18738         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18739         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18740         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18741         * config/i386/sse.md (movdi_to_sse): Ditto.
18742         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18743         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18744         TARGET_INTER_UNIT_MOVES.
18745         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18746         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18747         instead of TARGET_INTER_UNIT_MOVES.
18748         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18749         operand 1 from Yi to Yj and from Ym to Yn.
18751 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18753         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18754         (float_truncate_hi_): Likewise.
18755         (float_extend_lo_): Likewise.
18756         (float_truncate_lo_): Likewise.
18757         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18758         (aarch64_float_extend_lo_v2df): Likewise.
18759         (vec_unpacks_hi_v4sf): Likewise.
18760         (aarch64_float_truncate_lo_v2sf): Likewise.
18761         (aarch64_float_truncate_hi_v4sf): Likewise.
18762         (vec_pack_trunc_v2df): Likewise.
18763         (vec_pack_trunc_df): Likewise.
18765 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18767         * config/aarch64/aarch64-builtins.c
18768         (aarch64_fold_builtin): Fold float conversions.
18769         * config/aarch64/aarch64-simd-builtins.def
18770         (floatv2si, floatv4si, floatv2di): New.
18771         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18772         * config/aarch64/aarch64-simd.md
18773         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18774         * config/aarch64/iterators.md (FLOATUORS): New.
18775         (optab): Add float, floatuns.
18776         (su_optab): Likewise.
18778 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18780         * config/aarch64/aarch64-builtins.c
18781         (aarch64_builtin_vectorized_function): Use new names for
18782         fcvt builtins.
18783         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18784         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18785         (fcvtzu): Split as...
18786         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18787         (fcvtas): Split as...
18788         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18789         (fcvtau): Split as...
18790         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18791         (fcvtps): Split as...
18792         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18793         (fcvtpu): Split as...
18794         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18795         (fcvtms): Split as...
18796         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18797         (fcvtmu): Split as...
18798         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18799         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18800         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18801         (lfrintnusf, lfrintnudf): Likewise.
18802         * config/aarch64/aarch64-simd.md
18803         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18804         define_insn.
18805         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18806         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18807         (fcvt_pattern): Likewise.
18809 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18811         * config/aarch64/aarch64-simd.md
18812         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18813         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18815 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18817         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18818         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18819         (vrnd<a,m,n,p>_f32): Implement using builtins.
18820         (vrnd<i,x><q>_f<32, 64>): New.
18822 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18824         * config/aarch64/aarch64-builtins.c
18825         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18826         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18827         (frintz): Rename to...
18828         (btrunc): ...this.
18829         (frintp): Rename to...
18830         (ceil): ...this.
18831         (frintm): Rename to...
18832         (floor): ...this.
18833         (frinti): Rename to...
18834         (nearbyint): ...this.
18835         (frintx): Rename to...
18836         (rint): ...this.
18837         (frinta): Rename to...
18838         (round): ...this.
18839         * config/aarch64/aarch64-simd.md
18840         (aarch64_frint<frint_suffix><mode>): Delete.
18841         (<frint_pattern><mode>2): Convert to insn.
18842         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18843         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18844         (frint_pattern): Likewise.
18845         (frint_suffix): Likewise.
18847 2013-04-29  Richard Biener  <rguenther@suse.de>
18849         PR tree-optimization/57081
18850         * loop-init.c: Include tree-flow.h.
18851         (loop_optimizer_finalize): Free number of iteration estimates.
18852         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18854 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18856         PR tree-optimization/57083
18857         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18858         non-singleton shift count range, zero extend low_bound for uns case.
18860         * config/i386/predicates.md (general_vector_operand): New predicate.
18861         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18862         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18863         if they aren't nonimmediate operands.  If their original values
18864         satisfy const_vector_equal_evenodd_p, don't shift them.
18865         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18866         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18867         if not nonimmediate_operand.
18868         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18869         instead of register_operand.
18870         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18872 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18874         * stor-layout.c (finalize_size_functions): Allocate a structure and
18875         reset cfun before dumping the functions.
18877 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18879         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18881         PR target/56866
18882         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18883         use xop_pmacsdqh if uns_p.
18884         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18885         the immediate rotate count.
18887 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18889         * rtl.h (struct rtx_def): Add comment for field jump.
18890         (LRA_SUBREG_P): New macro.
18891         * recog.c (register_operand): Check LRA_SUBREG_P.
18892         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18893         stack frame.
18894         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18895         (lra_final_code_change): Skip subreg change for operators.
18896         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18897         if there are no operand changes.
18898         * lra-constraints.c (curr_insn_set): New.
18899         (match_reload): Set LRA_SUBREG_P.
18900         (emit_spill_move): Ditto.
18901         (check_and_process_move): Use curr_insn_set. Process only single
18902         set insns.  Don't initialize sec_mem_p and change_p.
18903         (simplify_operand_subreg): Use LRA_SUBREG_P.
18904         (reg_in_class_p): New function.
18905         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18906         of #ifdef.  Add code to remove cycling.
18907         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18908         non-null disp.  Reload inner instead of disp when base and index
18909         are null.  Try to put lo_sum into register.
18910         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18911         (check_and_process_move): Move code for move cost check to
18912         simple_move_p.  Remove equiv_substitution.
18913         (simple_move_p): New function.
18914         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18915         curr_insn_set.  Call check_and_process_move only for single set
18916         insns.  Use the new function.  Move call of check_and_process_move
18917         after operand equiv substitution and address process.
18919 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18921         PR go/57045
18922         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18923         with nonlocal goto receivers or returns twice calls, ignore
18924         unininitialized values from abnormal edges to nl goto receiver
18925         or returns twice call.
18927 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18929         PR tree-optimization/57051
18930         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18931         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18932         bitsize.
18934 2013-04-26  Richard Biener  <rguenther@suse.de>
18936         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18937         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18938         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18939         (fixup_loop_arrays_after_move): New function.
18940         (move_sese_region_to_fn): Properly outline the loop tree parts
18941         of the SESE region.
18943 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18945         * config/i386/i386.md (type, unit): Fix long lines.
18947 2013-04-26  Richard Biener  <rguenther@suse.de>
18949         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
18950         (lto-streamer-out.o): Likewise.
18951         * cfgloop.c (init_loops_structure): Export, add struct function
18952         argument and adjust.
18953         (flow_loops_find): Adjust.
18954         * cfgloop.h (enum loop_estimation): Add EST_LAST.
18955         (init_loops_structure): Declare.
18956         * lto-streamer-in.c: Include cfgloop.h.
18957         (input_cfg): Input the loop tree.
18958         * lto-streamer-out.c: Include cfgloop.h.
18959         (output_cfg): Output the loop tree.
18960         (output_struct_function_base): Do not drop PROP_loops.
18962 2013-03-26  Richard Biener  <rguenther@suse.de>
18964         * tree-cfg.c (execute_build_cfg): Build the loop tree.
18965         (pass_build_cfg): Provide PROP_loops.
18966         (move_sese_region_to_fn): Remove loops that are outlined into fn
18967         for now.
18968         * tree-inline.c: Include cfgloop.h.
18969         (initialize_cfun): Do not drop PROP_loops.
18970         (copy_loops): New function.
18971         (copy_cfg_body): Copy loop structure.
18972         (tree_function_versioning): Initialize destination loop tree.
18973         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
18974         (pass_parallelize_loops): Do IL verification.
18975         * loop-init.c (loop_optimizer_init): Fixup loops if required.
18976         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
18977         the CFG make sure we fixup loops as well.
18978         * tree-ssa-tail-merge.c: Include cfgloop.h.
18979         (replace_block_by): When merging loop latches mark loops for fixup.
18980         * lto-streamer-out.c (output_struct_function_base): Drop
18981         PROP_loops for now.
18982         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
18983         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
18984         * ipa-split.c: Include cfgloop.h.
18985         (split_function): Add the new return block to the loop tree root.
18986         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
18987         whether we have removed the forwarder block.
18988         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
18989         * cfgloop.h (place_new_loop): Declare.
18990         * cfgloopmanip.c (place_new_loop): Export.
18991         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
18992         (tree-switch-conversion.o): Likewise.
18993         (tree-complex.o): Likewise.
18994         (tree-inline.o): Likewise.
18995         (tree-ssa-tailmerge.o): Likewise.
18996         (ipa-split.o): Likewise.
18997         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
18998         (tree-ssa-copy.o): Likewise.
18999         * tree-switch-conversion.c: Include cfgloop.h
19000         (process_switch): If we emit a bit-test cascade, schedule loops
19001         for fixup.
19002         * tree-complex.c: Include cfgloop.h.
19003         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19004         * asan.c: Include cfgloop.h.
19005         (create_cond_insert_point): Properly add new basic-blocks to
19006         loops, schedule loop fixup.
19007         * cfgloop.c (verify_loop_structure): Check that looks are not
19008         marked for fixup.
19009         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19010         to loops.
19011         (expand_omp_for_generic): Likewise.
19012         (expand_omp_sections): Likewise.
19013         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19014         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19015         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19016         is initialized, not when loops are present.
19017         * tree-parloops.c (parallelize_loops): Remove checking here.
19018         * passes.c (init_optimization_passes): Schedule a copy-propagation
19019         pass before complete unrolling of inner loops.
19021 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19023         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19024         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19025         (colorize_init): Add argument to _WIN32 version.
19026         * toplev.c: Include diagnostic-color.h.
19027         (process_options): Default to -fdiagnostics-color=auto if
19028         GCC_COLORS env var is in the environment.
19029         * common.opt (fdiagnostics-color=): Add Var and Init.
19030         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19031         env var is in the environment, the default is auto rather than never.
19033         * diagnostic.h (file_name_as_prefix): Add context argument.
19034         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19035         the string as locus.
19036         * langhooks.c (lhd_print_error_function): Adjust caller.
19038 2013-04-25  Lawrence Crowl  <crowl@google.com>
19040         * var-tracking.c (shared_hash_def::htab):
19041         Change type to hash_table.  Update dependent calls and types.
19043 2013-04-25  Lawrence Crowl  <crowl@google.com>
19045         * Makefile.in: Update as needed below.
19047         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19048         Move declaration to after the type's method definitons.
19050         * attribs.c (htab_t scoped_attributes::attribute_hash):
19051         Change type to hash_table.  Update dependent calls and types.
19053         * bitmap.c (htab_t bitmap_desc_hash):
19054         Change type to hash_table.  Update dependent calls and types.
19056         * cselib.c (htab_t cselib_hash_table):
19057         Change type to hash_table.  Update dependent calls and types.
19059         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19060         (hash_string_slot_node): Move implementation into lto-streamer.h
19061         struct string_slot_hasher.
19062         (eq_string_slot_node): Likewise.
19064         * data-streamer-out.c: Update output_block::string_hash_table
19065         dependent calls and types.
19067         * dwarf2cfi.c (htab_t trace_index):
19068         Change type to hash_table.  Update dependent calls and types.
19070         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19071         Change type to hash_table.  Update dependent calls and types.
19072         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19073         (htab_t optimize_external_refs::map): Likewise.
19074         (htab_t output_comp_unit::extern_map): Likewise.
19075         (htab_t output_comdat_type_unit::extern_map): Likewise.
19076         (htab_t output_macinfo::macinfo_htab): Likewise.
19077         (htab_t optimize_location_lists::htab): Likewise.
19078         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19080         * except.c (htab_t ehspec_hash_type):
19081         Change type to hash_table.  Update dependent calls and types.
19082         (assign_filter_values::ttypes): Likewise.
19083         (assign_filter_values::ehspec): Likewise.
19084         (sjlj_assign_call_site_values::ar_hash): Likewise.
19085         (convert_to_eh_region_ranges::ar_hash): Likewise.
19087         * gcse.c (htab_t pre_ldst_table):
19088         Change type to hash_table.  Update dependent calls and types.
19090         * ggc-common.c (htab_t saving_htab):
19091         Change type to hash_table.  Update dependent calls and types.
19092         (htab_t loc_hash): Likewise.
19093         (htab_t ptr_hash): Likewise.
19094         (call_count): Rename ggc_call_count.
19095         (call_alloc): Rename ggc_call_alloc.
19096         (loc_descriptor): Rename make_loc_descriptor.
19097         (add_statistics): Rename ggc_add_statistics.
19099         * ggc-common.c (saving_htab):
19100         Change type to hash_table.  Update dependent calls and types.
19102         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19103         (push_gimplify_context): Likewise.
19104         (pop_gimplify_context): Likewise.
19105         (struct gimple_temp_hash_elt): Added.
19106         (struct gimplify_hasher): Likewise.
19107         (struct gimplify_ctx.temp_htab):
19108         Change type to hash_table.  Update dependent calls and types.
19110         * gimple-fold.c: Include gimplify-ctx.h.
19112         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19113         Change type to hash_table.  Update dependent calls and types.
19114         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19115         avoid potential global name collision.
19117         * gimplify.c: Include gimplify-ctx.h.
19118         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19119         (htab_t gimplify_ctx::temp_htab):
19120         Update dependent calls and types for new type hash_table.
19121         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19122         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19124         * gimplify-ctx.h: New.
19125         (struct gimple_temp_hash_elt): Move from gimplify.c.
19126         (class gimplify_hasher): New.
19127         (struct gimplify_ctx): Move from gimple.h.
19128         (htab_t gimplify_ctx::temp_htab):
19129         Change type to hash_table.  Update dependent calls and types.
19131         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19132         (htab_t ivs_params::newivs_index):
19133         Change type to hash_table.  Update dependent calls and types.
19134         (htab_t ivs_params::params_index): Likewise.
19135         (htab_t print_generated_program::params_index): Likewise.
19136         (htab_t gloog::newivs_index): Likewise.
19137         (htab_t gloog::params_index): Likewise.
19139         * graphite.c: Include graphite-htab.h.
19140         4htab_t graphite_transform_loops::bb_pbb_mapping):
19141         Change type to hash_table.  Update dependent calls and types.
19143         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19144         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19145         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19147         * graphite-dependences.c: Include graphite-htab.h.
19148         (loop_is_parallel_p): Change hash table type of parameter.
19150         * graphite-htab.h: New.
19151         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19152         (extern find_pbb_via_hash): Move from graphite-poly.h.
19153         (extern loop_is_parallel_p): Move from graphite-poly.h.
19154         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19156         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19157         (extern loop_is_parallel_p): Move to graphite-htab.h.
19158         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19160         * haifa-sched.c (htab_t delay_htab):
19161         Change type to hash_table.  Update dependent calls and types.
19162         (htab_t delay_htab_i2): Likewise.
19164         * ira-color.c (htab_t allocno_hard_regs_htab):
19165         Change type to hash_table.  Update dependent calls and types.
19167         * ira-costs.c (htab_t cost_classes_htab):
19168         Change type to hash_table.  Update dependent calls and types.
19170         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19171         Change type to hash_table.  Update dependent calls and types.
19173         * loop-iv.c (htab_t bivs):
19174         Change type to hash_table.  Update dependent calls and types.
19176         * loop-unroll.c (htab_t opt_info::insns_to_split):
19177         Change type to hash_table.  Update dependent calls and types.
19178         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19180         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19181         (struct string_slot_hasher): New.
19182         (htab_t output_block::string_hash_table):
19183         Change type to hash_table.  Update dependent calls and types.
19185         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19186         (htab_t file_name_hash_table):
19187         Change type to hash_table.  Update dependent calls and types.
19189         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19190         calls and types.
19192         * lto-streamer.c (htab_t tree_htab):
19193         Change type to hash_table.  Update dependent calls and types.
19195         * omp-low.c: Include gimplify-ctx.h.
19197         * passes.c (htab_t name_to_pass_map):
19198         Change type to hash_table.  Update dependent calls and types.
19199         (pass_traverse): Rename to passes_pass_traverse.
19201         * plugin.c (htab_t event_tab):
19202         Change type to hash_table.  Update dependent calls and types.
19204         * postreload-gcse.c (htab_t expr_table):
19205         Change type to hash_table.  Update dependent calls and types.
19206         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19208         * sese.c (debug_rename_map_1): Make extern.
19209         (htab_t copy_bb_and_scalar_dependences::rename_map):
19210         Change type to hash_table.  Update dependent calls and types.
19212         * sese.h (extern debug_rename_map): Move to .c file.
19214         * store-motion.c (htab_t store_motion_mems_table):
19215         Change type to hash_table.  Update dependent calls and types.
19217         * trans-mem.c (htab_t tm_new_mem_hash):
19218         Change type to hash_table.  Update dependent calls and types.
19220         * tree-browser.c (htab_t TB_up_ht):
19221         Change type to hash_table.  Update dependent calls and types.
19223         * tree-cfg.c (htab_t discriminator_per_locus):
19224         Change type to hash_table.  Update dependent calls and types.
19226         * tree-complex.c: Include tree-hasher.h
19227         (htab_t complex_variable_components):
19228         Change type to hash_table.  Update dependent calls and types.
19230         * tree-eh.c (htab_t finally_tree):
19231         Change type to hash_table.  Update dependent calls and types.
19233         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19234         struct int_tree_hasher.
19235         (extern int_tree_map_eq): Likewise.
19236         (uid_decl_map_hash): Removed.
19237         (extern decl_tree_map_eq): Likewise.
19239         * tree-hasher.h: New.
19240         (struct int_tree_hasher): New.
19241         (typedef int_tree_htab_type): New.
19243         * tree-inline.c: Include gimplify-ctx.h.
19245         * tree-mudflap.c: Include gimplify-ctx.h.
19247         * tree-parloops.c: Include tree-hasher.h.
19248         (htab_t eliminate_local_variables_stmt::decl_address):
19249         Change type to hash_table.  Update dependent calls and types.
19250         (htab_t separate_decls_in_region::decl_copies): Likewise.
19252         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19253         Change type to hash_table.  Update dependent calls and types.
19255         * tree-sra.c (candidates):
19256         Change type to hash_table.  Update dependent calls and types.
19258         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19259         in tree-flow.h.
19260         (int_tree_map_hash): Likewise.
19262         * tree-ssa-dom.c (htab_t avail_exprs):
19263         Change type to hash_table.  Update dependent calls and types.
19265         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19266         Change type to hash_table.  Update dependent calls and types.
19268         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19269         Change type to hash_table.  Update dependent calls and types.
19271         * tree-ssa-phiopt.c (seen_ssa_names):
19272         Change type to hash_table.  Update dependent calls and types.
19274         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19275         Change type to hash_table.  Update dependent calls and types.
19277         * tree-ssa-uncprop.c (equiv):
19278         Change type to hash_table.  Update dependent calls and types.
19280 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19282         PR rtl-optimization/57003
19283         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19284         call note_stores with kill_clobbered_value callback again after
19285         killing regs_invalidated_by_call.
19287 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19289         * config/aarch64/aarch64-simd.md
19290         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19291         (aarch64_simd_bsl<mode>): Likewise.
19292         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19294 2013-04-25  Marek Polacek  <polacek@redhat.com>
19296         PR tree-optimization/57066
19297         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19299 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19301         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19303 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19305         * config/aarch64/aarch64-builtins.c
19306         (aarch64_fold_builtin): New.
19307         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19308         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19309         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19310         * config/aarch64/arm_neon.h
19311         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19313 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19314             Tejas Belagod  <tejas.belagod@arm.com>
19316         * config/aarch64/aarch64-builtins.c
19317         (aarch64_gimple_fold_builtin): New.
19318         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19319         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19320         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19321         (addvv4sf): Update.
19322         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19324 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19326         * config/aarch64/aarch64.md
19327         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19329 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19331         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19332         (*ngcsi_uxtw): New pattern.
19334 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19335             Julian Brown  <julian@codesourcery.com>
19337         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19338         (TB_DREG): Add T_V4HF.
19339         (v4hf_UP): New macro.
19340         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19341         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19342         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19343         built-ins.
19344         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19345         (arm_vector_mode_supported_p): Handle V4HF.
19346         (arm_mangle_map): Handle V4HFmode.
19347         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19348         * config/arm/arm_neon_builtins.def: Add entries for
19349         vcvtv4hfv4sf, vcvtv4sfv4hf.
19350         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19351         (neon_vcvtv4hfv4sf): Likewise.
19352         * config/arm/neon-gen.ml: Handle half-precision floating point
19353         features.
19354         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19355         * config/arm/arm_neon.h: Regenerate.
19356         * config/arm/neon.ml (type elts): Add F16.
19357         (type vectype): Add T_float16x4, T_floatHF.
19358         (type vecmode): Add V4HF.
19359         (type features): Add Requires_FP_bit feature.
19360         (elt_width): Handle F16.
19361         (elt_class): Likewise.
19362         (elt_of_class_width): Likewise.
19363         (mode_of_elt): Refactor.
19364         (type_for_elt): Handle F16, fix error messages.
19365         (vectype_size): Handle T_float16x4.
19366         (vcvt_sh): New function.
19367         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19368         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19369         (string_of_mode): Handle V4HF.
19370         * doc/arm-neon-intrinsics.texi: Regenerate.
19372 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19374         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19375         format specifier in 'X' case.
19377 2013-04-25  Alan Modra  <amodra@gmail.com>
19379         PR target/57052
19380         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19381         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19382         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19383         Repeat for many other rotate/shift and mask patterns using subregs.
19384         Name lshiftrt insns.
19385         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19386         on WORDS_BIG_ENDIAN.
19388 2013-04-25  Alan Modra  <amodra@gmail.com>
19390         * config.gcc: Support little-endian powerpc-linux targets.
19391         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19392         (LINK_OS_LINUX_SPEC): Define.
19393         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19394         Preserve MASK_LITTLE_ENDIAN.
19395         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19396         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19397         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19398         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19399         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19400         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19401         Correct fp word order for little-endian.  Don't shift toc entries
19402         smaller than a word for little-endian.
19403         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19404         (bswapdi2 splits): Correct low-part subreg for little-endian.
19405         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19406         low/high where such is correct only for be.
19407         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19408         little-endian for -mcall-aixdesc.
19410 2013-04-25  Alan Modra  <amodra@gmail.com>
19412         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19413         replace_equiv_address_nv.
19415 2013-04-25  Alan Modra  <amodra@gmail.com>
19417         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19419 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19421         Revert:
19422         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19423         * rtl.h (struct rtx_def): ...
19425 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19427         PR rtl-optimizations/57046
19428         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19429         for multi-reg splits.
19431 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19433         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19435 2013-04-24  Sterling Augustine  <saugustine@google.com>
19437         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19438         (comp_dir_string, debug_str_dwo_section): New.
19439         (DEBUG_STR_DWO_SECTION): Rename to ...
19440         (DEBUG_DWO_STR_SECTION): ... this.
19441         (DEBUG_NORM_STR_SECTION): Delete.
19442         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19443         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19444         (find_AT_string): Move most logic to ...
19445         (find_AT_string_in_table): ... here.  New.
19446         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19447         add_skeleton_AT_string.  Delete logic.
19448         (output_skeleton_debug_sections): Remove call to
19449         add_top_level_skeleton_die_attrs.
19450         (add_comp_dir_attribute): Move logic to comp_dir_string.
19451         (dwarf2out_init): Initialize debug_str_dwo_section.
19452         (output_indirect_string): Call find_string_form.
19453         (output_indirect_strings): Rewrite.
19454         (prune_unused_types): Empty skeleton_debug_str_hash.
19455         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19456         (dwarf2out_finish):  Call output_indirect_strings.
19458 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19460         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19462 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19464         * rtl.h (struct rtx_def): Add comment for field jump.
19465         (LRA_SUBREG_P): New macro.
19466         * recog.c (register_operand): Check LRA_SUBREG_P.
19467         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19468         stack frame.
19469         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19470         (lra_final_code_change): Skip subreg change for operators.
19471         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19472         if there are no operand changes.
19473         * lra-constraints.c (curr_insn_set): New.
19474         (match_reload): Set LRA_SUBREG_P.
19475         (emit_spill_move): Ditto.
19476         (check_and_process_move): Use curr_insn_set. Process only single
19477         set insns.  Don't initialize sec_mem_p and change_p.
19478         (simplify_operand_subreg): Use LRA_SUBREG_P.
19479         (reg_in_class_p): New function.
19480         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19481         of #ifdef.  Add code to remove cycling.
19482         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19483         non-null disp.  Reload inner instead of disp when base and index
19484         are null.  Try to put lo_sum into register.
19485         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19486         (check_and_process_move): Move code for move cost check to
19487         simple_move_p.  Remove equiv_substitution.
19488         (simple_move_p): New function.
19489         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19490         curr_insn_set.  Call check_and_process_move only for single set
19491         insns.  Use the new function.  Move call of check_and_process_move
19492         after operand equiv substitution and address process.
19494 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19496         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19497         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19498         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19500 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19502         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19504 2013-04-24  Marek Polacek  <polacek@redhat.com>
19506         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19507         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19508         (select_loops_exit_conditions): Likewise.
19509         (number_of_iterations_for_all_loops): Likewise.
19510         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19511         (scev_analysis): Likewise.
19513 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19514             Chao-ying Fu  <fu@mips.com>
19516         * config/mips/micromips.md (jraddiusp): New pattern.
19517         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19518         instruction if possible.
19520 2013-04-24  Alan Modra  <amodra@gmail.com>
19522         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19524 2013-04-24  Julian Brown  <julian@codesourcery.com>
19525             Chung-Lin Tang  <cltang@codesourcery.com>
19527         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19528         dependency behavior in enumeration type DIE generation. Add TODO note
19529         to comments about future DW_FORM_sdata/udata re-work of related code.
19531 2013-04-23  Lawrence Crowl  <crowl@google.com>
19533         * Makefile.in: Update as needed below.
19535         * hash-table.h (class hash_table):
19536         Correct many methods with parameter types compare_type to the correct
19537         value_type.  (Correct code was unlikely to notice the change.)
19538         (hash_table::elements_with_deleted) New.
19539         (class hashtable::iterator): New.
19540         (hashtable::begin()): New.
19541         (hashtable::end()): New.
19542         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19544         * statistics.c (statistics_hashes):
19545         Change type to hash_table.  Update dependent calls and types.
19547         * tree-into-ssa.c (var_infos):
19548         Change type to hash_table.  Update dependent calls and types.
19550         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19551         Change type to hash_table.  Update dependent calls and types.
19553         * tree-ssa-loop-im.c (struct mem_ref.refs):
19554         Change type to hash_table.  Update dependent calls and types.
19556         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19557         Change type to hash_table.  Update dependent calls and types.
19559         * tree-ssa-sccvn.c (vn_tables_s::nary):
19560         Change type to hash_table.  Update dependent calls and types.
19561         (vn_tables_s::phis): Likewise.
19562         (vn_tables_s::references): Likewise.
19564         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19565         (vn_reference_eq): Update parameter and return types.
19567         * tree-ssa-structalias.c (pointer_equiv_class_table):
19568         Change type to hash_table.  Update dependent calls and types.
19569         (location_equiv_class_table): Likewise.
19571         * tree-vect-data-refs.c: Consequential changes for making
19572         peeling a hash_table.
19574         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19575         (destroy_loop_vec_info): Dependent hash_table update.
19577         * tree-vectorizer.h (peeling_htab):
19578         Change type to hash_table.  Update dependent calls and types.
19580 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19582         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19583         to check the register content is equal or not.
19584         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19585         to assign register content record.
19586         * lra-eliminations.c (update_reg_eliminate): Use
19587         lra_update_reg_val_offset to update register content offset.
19588         * lra-int.h (struct lra_reg): Add offset member.
19589         (lra_reg_val_equal_p): New static inline function.
19590         (lra_update_reg_val_offset): New static inline function.
19591         (lra_assign_reg_val): New static inline function.
19592         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19593         to assign register content record.
19594         (initialize_lra_reg_info_element): Initial offset to zero.
19596 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19598         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19599         operands.  Record compression.
19601 2013-04-23  Xinliang David Li  <davidxl@google.com>
19603         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19605 2013-04-23  Richard Biener  <rguenther@suse.de>
19607         PR middle-end/57036
19608         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19609         parameter, only add abnormal goto edges from the copied body
19610         if the call could perform abnormal gotos.
19611         (copy_cfg_body): Adjust.
19613 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19615         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19617 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19619         * coretypes.h (gimple_stmt_iterator): Add struct to make
19620         compatible with C.
19622 2013-04-23  Richard Biener  <rguenther@suse.de>
19624         PR tree-optimization/57026
19625         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19626         from SSA names occuring in abnormal PHI nodes.
19628 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19630         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19632 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19634         * lto/lto.c (print_lto_report_1): Declare early.
19635         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19637 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19639         * common.opt (-flto-report-wpa): Add.
19640         * doc/invoke.texi (-flto-report-wpa): Add.
19641         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19642         (lto_main): dito.
19644 2013-04-22  Xinliang David Li  <davidxl@google.com>
19646         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19647         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19648         * Makefile.in: New dependency
19650             David Daney <ddaney.cavm@gmail.com>
19652         * configure.ac (gcc_cv_as_micromips_support):  Use the
19653         --fatal-warnings option.
19654         * configure: Regenerate.
19656 2013-04-22  Marek Polacek  <polacek@redhat.com>
19658         PR sanitizer/56990
19659         * tsan.c (instrument_expr): Don't instrument expression
19660         in case its size is zero.
19662 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19664         PR target/57032
19665         Revert:
19666         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19668         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19670 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19672         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19673         (gimple_stmt_iterator): New typedef.
19674         * gimple.h (gimple_stmt_iterator): Rename to...
19675         (gimple_stmt_iterator_d): ... This.
19676         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19677         trees be valid for GIMPLE and GENERIC.
19678         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19679         * gimple-fold.c (gimple_fold_call): Call target hook
19680         gimple_fold_builtin.
19681         * hooks.c (hook_bool_gsiptr_false): New.
19682         * hooks.h (hook_bool_gsiptr_false): New.
19683         * target.def (fold_stmt): New.
19684         * doc/tm.texi: Regenerate.
19686 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19688         PR target/57018
19689         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19690         a set sp if no stack realignment.
19692 2013-04-22  Nick Clifton  <nickc@redhat.com>
19694         * config.gcc (tilegx-linux): Extend extra_objs rather than
19695         overwriting it.
19696         (tilepro-linux): Likewise.
19698 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19700         * config/aarch64/aarch64-builtins.c
19701         (CF): Remove.
19702         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19703         (VAR<1-12>): Add MAP parameter.
19704         (BUILTIN_*): Likewise.
19705         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19706         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19707         (aarch64_ushl_n<mode>): Likewise.
19708         (aarch64_sshr_n<mode>): Likewise.
19709         (aarch64_ushr_n<mode>): Likewise.
19710         (aarch64_<maxmin><mode>): Likewise.
19711         (aarch64_sqrt<mode>): Likewise.
19712         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19713         (vshr<q>_n_*): Likewise.
19715 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19717         * config/aarch64/aarch64-builtins.c
19718         (aarch64_simd_builtin_type_mode): Handle SF types.
19719         (sf_UP): Define.
19720         (BUILTIN_GPF): Define.
19721         (aarch64_init_simd_builtins): Handle SF types.
19722         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19723         (frecps): Likewise.
19724         (frecpx): Likewise.
19725         * config/aarch64/aarch64-simd.md
19726         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19727         (aarch64_frecpe<mode>): New.
19728         (aarch64_frecps<mode>): Likewise.
19729         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19730         (v8type): Add frecp<esx>.
19731         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19732         (aarch64_frecps<mode>): Likewise.
19733         * config/aarch64/iterators.md (FRECP): New.
19734         (frecp_suffix): Likewise.
19735         * config/aarch64/arm_neon.h
19736         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19738 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19740         PR target/56995
19741         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19742         (REG_CLASS_NAMES): Idem.
19743         (REG_CLASS_CONTENTS): Idem.
19744         (REGCLASS_HAS_FP_REG): Idem.
19745         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19746         (sh_conditional_register_usage): Idem.
19748 2013-04-21  Jeff Law  <law@redhat.com>
19750         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19751         (ssa_forward_propagate_and_combine): Use it.
19753 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19755         * lra.c: Update the flow chart diagram.
19757 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19759         PR rtl-optimization/56847
19760         * lra-constraints.c (process_alt_operands): Discourage alternative
19761         with non-matche doffsettable memory constraint fro memory with
19762         known offset.
19764 2013-04-19  Richard Biener  <rguenther@suse.de>
19766         PR tree-optimization/56982
19767         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19768         function.
19769         * gimplify.c (gimplify_call_expr): Notice special calls.
19770         (gimplify_modify_expr): Likewise.
19771         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19772         abnormal control flow receivers.
19773         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19774         in the same way as cfun->has_nonlocal_labels.
19775         (gimple_purge_dead_abnormal_call_edges): Likewise.
19776         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19777         receivers start a basic-block.
19779 2013-04-19  Richard Biener  <rguenther@suse.de>
19781         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19782         member ...
19783         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19784         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19785         (SLP_TREE_LOAD_PERMUTATION): Add.
19786         (vect_transform_slp_perm_load): Adjust prototype.
19787         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19788         (vect_free_slp_instance): Likewise.
19789         (vect_create_new_slp_node): Likewise.
19790         (vect_supported_slp_permutation_p): Remove.
19791         (vect_slp_rearrange_stmts): Adjust.
19792         (vect_supported_load_permutation_p): Likewise.  Inline
19793         vect_supported_slp_permutation_p here.
19794         (vect_analyze_slp_instance): Compute load permutations per
19795         slp node instead of per instance.
19796         (vect_get_slp_defs): Adjust.
19797         (vect_transform_slp_perm_load): Likewise.
19798         (vect_schedule_slp_instance): Remove redundant code.
19799         (vect_schedule_slp): Remove hack for PR56270, add it ...
19800         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19801         CSE loads for SLP.  Adjust.
19803 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19805         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19806         spelling in two comments.
19808 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19810         PR target/56797
19811         * config/arm/arm.c (load_multiple_sequence): Require SP
19812         as base register for loads if SP is in the register list.
19814 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19816         PR tree-optimization/56718
19817         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19818         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19819         and made public.  Adjusted all callers.
19820         (ipa_intraprocedural_devirtualization): New function.
19821         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19822         (ipa_intraprocedural_devirtualization): Likewise.
19823         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19825 2013-04-19  Richard Biener  <rguenther@suse.de>
19827         PR tree-optimization/57000
19828         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19830 2013-04-19  Terry Guo  <terry.guo@arm.com>
19832         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19833         Replace with ...
19834         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19835         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19836         (cortex_m4_fmacs): Use new reservations.
19837         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19839 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19841         PR rtl-optimization/56999
19842         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19843         related code.
19844         (lra_coalesce): Remove split_origin_bitmap and related code.
19845         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19846         ranges if necessary.
19848 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19850         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19851         New array.
19852         (ix86_expand_call): Remove clobbered_registers array and use
19853         x86_64_ms_sysv_extra_clobbered_registers instead.
19854         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19855         Declare here.
19856         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19857         predicate.
19858         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19859         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19860         (*call_value_rex64_ms_sysv): Ditto.
19862 2013-04-18  Cary Coutant  <ccoutant@google.com>
19864         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19865         parent instead of die_mark.
19867 2013-04-18  Diego Novillo  <dnovillo@google.com>
19869         * gimple.c (create_gimple_tmp): New.
19870         (get_expr_type): New.
19871         (build_assign): New.
19872         (build_type_cast): New.
19873         * gimple.h (enum ssa_mode): Define.
19874         (gimple_seq_set_location): New.
19875         * asan.c (build_check_stmt): Change some gimple_build_* calls
19876         to use build_assign and build_type_cast.
19878 2013-04-18  Richard Biener  <rguenther@suse.de>
19880         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19881         handle negative step.  Remove redundant checks.
19882         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19883         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19884         for negative step and grouped loads fail to vectorize.
19886 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19888         * emit-rtl.c (reset_insn_used_flags): New function.
19889         (reset_all_used_flags): Use it.
19890         (verify_insn_sharing): New function.
19891         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19893 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19895         PR tree-optimization/56984
19896         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19897         and (x >> M) >= N don't register any assertion if N << M is the
19898         minimum value.
19900 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19902         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19903         do not delete_insn insns that have not yet been emitted, only
19904         unlink them with remove_insn.
19905         * df-scan.c (df_insn_delete): Revert r197492.
19907 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19909         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19910         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19912 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19914         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19915         define_insn_and_split.
19916         (and_scc,ior_scc,negscc): Likewise.
19917         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19919 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19921         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19922         can benefit from using a sequence of LDRD instructions in epilogue
19923         instead of a single LDM instruction.
19925 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19927         PR 45688
19928         * doc/extend.texi: Fix typo.
19930 2013-04-17  Richard Biener  <rguenther@suse.de>
19932         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19933         (vect_build_slp_tree): ... here.
19934         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19935         match.  Remove special-casing of mismatched complex loads.
19936         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19937         re-try the match with swapped commutative operands.
19938         (vect_supported_load_permutation_p): Remove special-casing of
19939         mismatched complex loads.
19940         (vect_analyze_slp_instance): Adjust.
19942 2013-04-17  Richard Biener  <rguenther@suse.de>
19944         PR rtl-optimization/56921
19945         * cfgloop.h (struct loop): Add simple_loop_desc member.
19946         (struct niter_desc): Mark with GTY(()).
19947         (simple_loop_desc): Do not use aux field but simple_loop_desc.
19948         * loop-iv.c (get_simple_loop_desc): Likewise.
19949         (free_simple_loop_desc): Likewise.
19951         Revert
19952         2013-04-16  Richard Biener  <rguenther@suse.de>
19954         PR rtl-optimization/56921
19955         * loop-init.c (pass_rtl_move_loop_invariants): Add
19956         TODO_do_not_ggc_collect to todo_flags_finish.
19957         (pass_rtl_unswitch): Same.
19958         (pass_rtl_unroll_and_peel_loops): Same.
19959         (pass_rtl_doloop): Same.
19961 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
19963         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
19964         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
19965         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
19966         references.
19967         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
19968         * tree-streamer.c (record_common_node): Adjust reference in comment.
19970 2013-04-17  Terry Guo  <terry.guo@arm.com>
19972         * config/arm/cortex-m4.md: Add a new bypass.
19974 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19976         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
19977         New pattern.
19978         (*subs_<optab><mode>_multp2): New pattern.
19979         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19980         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
19982 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19984         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
19985         (*subs_mul_imm_<mode>): New pattern.
19987 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
19989         PR target/56948
19990         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
19991         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
19992         (vsx_movti_32bit): Change j->wa to O->wa.
19994 2013-04-16  Richard Biener  <rguenther@suse.de>
19996         PR rtl-optimization/56921
19997         * loop-init.c (pass_rtl_move_loop_invariants): Add
19998         TODO_do_not_ggc_collect to todo_flags_finish.
19999         (pass_rtl_unswitch): Same.
20000         (pass_rtl_unroll_and_peel_loops): Same.
20001         (pass_rtl_doloop): Same.
20003 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20005         * config/arm/arm.c (emit_multi_reg_push): New declaration
20006         for an existing function.
20007         (arm_emit_strd_push): New function.
20008         (arm_expand_prologue): Used here.
20009         (arm_emit_ldrd_pop): New function.
20010         (arm_expand_epilogue): Used here.
20011         (arm_get_frame_offsets): Update condition.
20012         (arm_emit_multi_reg_pop): Add a special case for load of a single
20013         register with writeback.
20015 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20017         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20018         description.
20020 2013-04-16  Richard Biener  <rguenther@suse.de>
20022         PR tree-optimization/56756
20023         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20024         (first_mem_ref_loc): New.
20025         (execute_sm): Place the load temporarily before a previous
20026         access instead of in the latch edge to ensure its SSA dependencies
20027         are defined at points dominating the load.
20029 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20031         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20032         correct fix by moving header and footer insn to the footer of
20033         the merged basic block.  Clear BB_END of the merged-away block.
20035         PR middle-end/43631
20036         * emit-rtl.c (make_note_raw): New function.
20037         (link_insn_into_chain): New static inline function.
20038         (add_insn): Use it.
20039         (add_insn_before, add_insn_after): Factor insn chain linking code...
20040         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20041         using link_insn_into_chain.
20042         (note_outside_basic_block_p): New helper function for emit_note_after
20043         and emit_note_before.
20044         (emit_note_after): Use nobb variant of add_insn_after if the note
20045         should not be contained in a basic block.
20046         (emit_note_before): Use nobb variant of add_insn_before if the note
20047         should not be contained in a basic block.
20048         (emit_note_copy): Use make_note_raw.
20049         (emit_note): Likewise.
20050         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20051         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20052         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20053         the moved barrier the tail of the basic block it follows.
20054         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20056 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20058         PR tree-optimization/56962
20059         * gimple-ssa-strength-reduction.c (record_increment): Only set
20060         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20061         either rhs1 or rhs2 is equal to c->base_expr.
20063 2013-04-15  Richard Biener  <rguenther@suse.de>
20065         PR tree-optimization/56933
20066         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20067         member.
20068         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20069         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20070         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20071         dependence check ...
20072         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20073         ... here.
20074         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20075         GROUP_READ_WRITE_DEPENDENCE.
20077 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20079         * emit-rtl.c (reset_all_used_flags): New function.
20080         (verify_rtl_sharing): Call reset_all_used_flags before and after
20081         performing the checks.
20083 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20085         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20086         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20087         * config/arm/constraints.md (De): New constraint.
20088         * config/arm/neon.md (anddi3_neon): Delete.
20089         (neon_vand<mode>): Expand to standard anddi3 pattern.
20090         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20091         Move earlier in the file.
20092         (neon_inv_logic_op2): Likewise.
20093         (arm_anddi_operand_neon): New predicate.
20095 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20097         * configure.ac (gcc_cv_ld_as_needed): Set
20098         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20099         Use -z ignore, -z record on *-*-solaris2*.
20100         (HAVE_LD_AS_NEEDED): Update comment.
20101         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20102         * configure: Regenerate.
20103         * config.in: Regenerate.
20104         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20105         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20106         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20107         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20108         equivalents.  Fix markup.
20109         * doc/tm.texi: Regenerate.
20111 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20113         * config/i386/i386.opt: New option mstack-protector-guard=.
20114         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20115         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20116         TARGET_SSP_TLS_GUARD.
20117         * config/i386/i386.c (ix86_option_override_internal): Set
20118         ix86_stack_protector_guard.
20119         * config/i386/i386.md (stack_protect_set): Enable for
20120         TARGET_SSP_TLS_GUARD only.
20121         (stack_protect_set_<mode>): Ditto.
20122         (stack_protect_test): Ditto.
20123         (stack_protect_test_<mode>): Ditto.
20124         * doc/invoke.texi (i386 Option): Document.
20126 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20128         PR target/56890
20129         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20130         (S_MODES): Set H_MODE bit.
20131         (SF_MODES): Set only S_MODE and SF_MODE bits.
20132         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20133         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20134         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20135         <MODE_FLOAT>: Likewise.
20137 2013-04-15  Joey Ye  <joey.ye@arm.com>
20139         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20141 2013-04-15  Joey Ye  <joey.ye@arm.com>
20143         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20144         for real far jump.
20145         (thumb_far_jump_used_p): Count instruction size and set
20146         far_jump_used.
20148 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20150         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20151         * resource.c (mark_target_live_regs): Reformat conditional block.
20153 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20155         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20156         notes, they are emitted only just before final.
20157         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20159 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20161         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20162         * cfgrtl.c (delete_insn): Call it here instead.
20163         * lra-spills.c (lra_final_code_change): Use delete_insn.
20164         * haifa-sched.c (sched_remove_insn): Likewise.
20165         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20166         returning to the nop pool.
20167         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20168         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20170 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20172         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20173         * doc/tm.texi: Regenerated.
20175 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20177         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20178         QImode checks.
20180 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20182         * df-core.c (df_find_def): Compare register numbers.
20183         (df_find_use): Likewise.
20185 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20187         PR target/56903
20188         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20189         lra_in_progress for return.
20191 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20193         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20194         define_insn into define_insn_and_split and emit movsicc patterns.
20196 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20198         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20200 2013-04-12  Richard Biener  <rguenther@suse.de>
20202         * tree-pass.h (TODO_do_not_ggc_collect): New.
20203         * passes.c (execute_one_ipa_transform_pass): Honor
20204         TODO_do_not_ggc_collect.
20205         (execute_one_pass): Likewise.
20207         Revert
20208         2013-04-10  Richard Biener  <rguenther@suse.de>
20210         * passes.c (init_optimization_passes): Remove reload pass.
20211         * ira.c (do_reload): Merge into ...
20212         (ira): ... this.
20213         (rest_of_handle_reload): Remove.
20214         (pass_reload): Likewise.
20215         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20216         of reload for vzeroupper pass placement.
20218 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20220         PR tree-optimization/56918
20221         PR tree-optimization/56920
20222         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20223         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20224         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20225         use wide_mul_with_sign method.
20227 2013-04-12  Richard Biener  <rguenther@suse.de>
20229         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20230         not be considered a gimple constant.
20232 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20234         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20235         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20236         operations.
20238 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20239             Jakub Jelinek  <jakub@redhat.com>
20241         * opts.c: Include diagnostic-color.h.
20242         (common_handle_option): Handle OPT_fdiagnostics_color_.
20243         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20244         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20245         (diagnostic-color.o): New.
20246         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20247         (diagnostic_color_rule): New enum.
20248         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20249         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20250         the location string.
20251         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20252         either NULL, or color kind.
20253         * diagnostic-color.c: New file.
20254         * diagnostic-color.h: New file.
20255         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20256         arguments.
20257         * doc/invoke.texi (-fdiagnostics-color): Document.
20258         * pretty-print.h (pp_show_color): Define.
20259         (struct pretty_print_info): Add show_color field.
20260         * diagnostic.c: Include diagnostic-color.h.
20261         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20262         macros.  Colorize error:, warning: etc. strings and also the location
20263         string.
20264         (diagnostic_show_locus): Colorize the caret line.
20265         * pretty-print.c: Include diagnostic-color.h.
20266         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20267         inside of %< %> quotes or quoted through q format modifier.
20269 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20271         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20273 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20275         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20276         code in CC_NZ mode.
20277         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20278         pattern.
20280 2013-04-11  Marek Polacek  <polacek@redhat.com>
20282         PR tree-optimization/48184
20283         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20285 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20287         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20288         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20289         (skip_simple_arithmetic): Tidy up.
20290         * tree.h (skip_simple_constant_arithmetic): Declare.
20292 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20294         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20296 2013-04-11  Richard Biener  <rguenther@suse.de>
20298         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20299         generate vector constants.
20301 2013-04-11  Richard Biener  <rguenther@suse.de>
20303         PR tree-optimization/56878
20304         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20305         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20306         New function.
20307         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20308         Prefer to align the DR with the most invariant base address.
20310 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20312         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20313         comment.
20315 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20317         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20318         floating-point vector comparisons against 0.
20320 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20322         PR tree-optimization/56899
20323         * fold-const.c (extract_muldiv_1): Apply distributive law
20324         only if TYPE_OVERFLOW_WRAPS (ctype).
20326 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20328         PR target/56124
20329         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20330         loading has side effect.
20332 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20334         * config/sparc/sparc.c: Include tree-pass.h.
20335         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20336         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20337         head of file.  Change return type.  Split off gate function.
20338         (sparc_gate_work_around_errata): New function.
20339         (pass_work_around_errata): New pass definition.
20340         (insert_pass_work_around_errata) New pass insert definition to
20341         insert pass_work_around_errata just after delayed-branch scheduling.
20342         (sparc_option_override): Insert the pass.
20343         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20345 2013-04-10  David S. Miller  <davem@davemloft.net>
20347         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20348         or -mcpu=hypersparc.
20350         * target.def (cstore_mode): New hook.
20351         * target.h: Include insn-codes.h
20352         * targhooks.c: Likewise.
20353         (default_cstore_mode): New function.
20354         * targhooks.h: Declare it.
20355         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20356         * doc/tm.texi: Rebuild.
20357         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20358         target hook, rather than inspecting the insn_data.
20359         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20360         (TARGET_CSTORE_MODE): Redefine.
20361         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20362         result patterns.
20363         * config/sparc/predicates.md (cstore_result_operand): New special
20364         predicate.
20365         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20366         Use it for operand 0.
20367         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20368         (*snesi_special): Likewise.
20369         (*snesi_zero): Likewise.
20370         (*seqsi_zero): Likewise.
20371         (*sltu_insn): Likewise.
20372         (*sgeu_insn): Likewise.
20373         (*seqdi_special): Make operand 0 and comparison operation be of
20374         DImode.
20375         (*snedi_special): Likewise.
20376         (*snedi_special_vis3): Likewise.
20377         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20378         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20379         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20380         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20381         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20382         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20383         (*sltu_extend_sp64): Likewise.
20384         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20385         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20386         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20387         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20388         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20390 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20392         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20393         (aarch64_start_file): Use the new function.
20395 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20396             Jason Merrill  <jason@redhat.com>
20398         * common.opt: Add -gdwarf.
20399         * opts.c (common_handle_option): Handle it.
20400         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20402 2013-04-10  Richard Biener  <rguenther@suse.de>
20404         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20405         (execute_one_ipa_transform_pass): But unconditionally here.
20406         (execute_one_pass): And here.
20407         (init_optimization_passes): Remove reload pass.
20408         * tree-pass.h (TODO_ggc_collect): Remove.
20409         (pass_reload): Likewise.
20410         * ira.c (do_reload): Merge into ...
20411         (ira): ... this.
20412         (rest_of_handle_reload): Remove.
20413         (pass_reload): Likewise.
20414         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20415         of reload for vzeroupper pass placement.
20416         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20417         and todo_flags_finish of all passes.
20419 2013-04-10  Richard Biener  <rguenther@suse.de>
20421         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20422         first_const_oprnd field, rename first_def_type to first_op_type.
20423         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20424         (vect_get_and_check_slp_defs): Always use the type of the
20425         operand.  Allow mixed vect_external_def, vect_constant_def types.
20426         (vect_get_constant_vectors): Handle mixed vect_external_def,
20427         vect_constant_def types.
20429 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20431         PR tree-optimization/55524
20432         * tree-ssa-math-opts.c
20433         (convert_mult_to_fma): Don't use an fms construct
20434         when we don't have an fms operation, but fnma, and it looks
20435         likely that we'll be able to use the latter.
20437 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20439         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20440         function.
20441         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20442         inline fail caused by overwritable functions.
20444 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20446         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20447         unnecessary bits in the constant power of two case.
20449 2013-04-10  Richard Biener  <rguenther@suse.de>
20451         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20452         broken code swapping operands.
20453         (vect_build_slp_tree): Do not compute load permutations here.
20454         (vect_analyze_slp_instance): Compute load permutations here,
20455         after building the SLP tree.
20457 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20459         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20460         of next/prev_real_insn.
20462 2013-04-09  Jan Hubicka  <jh@suse.cz>
20464         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20465         Drop aliased parameter.
20466         (function_and_variable_visibility): Do not handle alias pairs.
20467         * cgraph.c (varpool_externally_visible_p): Update prototype.
20468         * varpool.c (varpool_add_new_variable): Update.
20470 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20472         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20474 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20476         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20478         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20480 2013-04-09  Marek Polacek  <polacek@redhat.com>
20482         PR tree-optimization/48762
20483         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20485 2013-04-09  Richard Biener  <rguenther@suse.de>
20487         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20488         dealing with cost.
20489         (vect_build_slp_tree): Likewise.
20490         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20491         calculating the cost of a SLP instance.
20492         (vect_analyze_slp_instance): Use it from here, after building
20493         the SLP tree.
20495 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20497         PR middle-end/56883
20498         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20499         expand_omp_for_static_chunk): Use simple_p = true in
20500         force_gimple_operand_gsi calls when assigning to addressable decls.
20502 2013-04-09  Jeff Law  <law@redhat.com>
20504         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20505         when the boolean was created by converting a wider object which
20506         had a boolean range.
20508 2013-04-09  Richard Biener  <rguenther@suse.de>
20510         * tree-vectorizer.h (slp_void_p): Remove.
20511         (slp_tree): Typedef before _slp_tree declaration.
20512         (struct _slp_tree): Use a vector of slp_tree as children.
20513         (vect_get_place_in_interleaving_chain): Remove.
20514         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20515         Move ...
20516         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20517         and make static.
20518         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20519         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20520         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20521         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20522         Use slp_node instead of slp_void_p and adjust.
20524 2013-04-09  Richard Biener  <rguenther@suse.de>
20526         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20527         work that is not necessary.
20529 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20531         PR tree-optimization/56854
20532         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20533         forward into clobber stmts if it would change MEM_REF lhs into
20534         non-MEM_REF.
20536 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20538         * tree.c (type_hash_lookup, type_hash_add): Make static.
20539         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20541 2013-04-09  Richard Biener  <rguenther@suse.de>
20543         * tree.h (unsave_expr_now): Remove.
20544         * tree-inline.c (mark_local_for_remap_r): Remove.
20545         (unsave_expr_1): Likewise.
20546         (unsave_r): Likewise.
20547         (unsave_expr_now): Likewise.
20548         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20549         (propagate_tree_value): Likewise.
20551 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20553         * doc/rtl.texi (sequence): Rewrite documentation to match the
20554         current use of SEQUENCE rtl objects.
20555         * rtl.def (SEQUENCE): Likewise.
20557         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20558         Update documentation.
20559         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20560         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20562         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20564 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20566         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20567         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20568         methods.
20569         (estimate_edge_size_and_time): Add comment to suggest using rounding
20570         methods.
20571         (estimate_node_size_and_time): Ditto.
20572         (remap_edge_change_prob): Use helper rounding divide methods.
20573         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20574         (gimple_mod_pow2_value_transform): Ditto.
20575         (gimple_mod_subtract_transform): Ditto.
20576         (gimple_ic_transform): Ditto.
20577         (gimple_stringops_transform): Ditto.
20578         * stmt.c (conditional_probability): Ditto.
20579         (emit_case_dispatch_table): Ditto.
20580         * lto-cgraph.c (merge_profile_summaries): Ditto.
20581         * tree-optimize.c (execute_fixup_cfg): Ditto.
20582         * cfgcleanup.c (try_forward_edges): Ditto.
20583         * cfgloopmanip.c (scale_loop_profile): Ditto.
20584         (loopify): Ditto.
20585         (duplicate_loop_to_header_edge): Ditto.
20586         (lv_adjust_loop_entry_edge): Ditto.
20587         * tree-vect-loop.c (vect_transform_loop): Ditto.
20588         * profile.c (compute_branch_probabilities): Ditto.
20589         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20590         * lto-streamer-in.c (input_cfg): Ditto.
20591         * gimple-streamer-in.c (input_bb): Ditto.
20592         * ipa-cp.c (update_profiling_info): Ditto.
20593         (update_specialized_profile): Ditto.
20594         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20595         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20596         rounding methods.
20597         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20598         (compute_trg_info): Ditto.
20599         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20600         (purge_dead_edges): Ditto.
20601         * loop-unswitch.c (unswitch_loop): Ditto.
20602         * cgraphclones.c (cgraph_clone_edge): Ditto.
20603         (cgraph_clone_node): Ditto.
20604         * tree-inline.c (copy_bb): Ditto.
20605         (copy_edges_for_bb): Ditto.
20606         (initialize_cfun): Ditto.
20607         (copy_cfg_body): Ditto.
20608         (expand_call_inline): Ditto.
20610 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20612         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20613         TARGET_CYGWIN64 by TARGET_64BIT.
20615 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20617         * config/epiphany/epiphany.md (GPR_1): New constant.
20618         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20619         * config/epiphany/epiphany.c (gen_compare_reg):
20620         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20621         is already in place.
20622         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20623         Don't require being called during rtl expansion; If y operlaps r0,
20624         return 0.
20625         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20626         (epiphany_expand_epilogue): Likewise.
20628         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20629         Don't use CC_FPmode for ORDERED / UNORDERED.
20630         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20632         * config/epiphany/constraints.md (CnL): New constraint.
20633         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20634         * config/epiphany/predicates.md (add_operand): Allow 1024.
20636         * config/epiphany/epiphany.md (logical_op): New code iterator.
20637         (op_mnc): New code attribute.
20638         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20639         (mov_f+1, mov_f+2): New peephole2 patterns.
20641         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20642         (cstoresi4): Also allow re-use of zero result when doing a NE
20643         comparison to a non-zero operand.
20644         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20646         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20647         Use gen_addsi3_i / gen_subsi3_i.
20649 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20651         PR c++/34949
20652         PR c++/50243
20653         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20654         contain anything but clobbers, at most one __builtin_stack_restore,
20655         optionally debug stmts and final resx, and if it has at least one
20656         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20657         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20658         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20659         which isn't defaut definition, remove them.
20660         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20661         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20662         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20663         with MEM_REF LHS with SSA_NAME address.
20665 2013-04-08  Jeff Law  <law@redhat.com>
20667         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20669 2013-04-08  Richard Biener  <rguenther@suse.de>
20671         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20672         extra newline.
20673         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20674         determined vector type.
20675         (vect_analyze_data_refs): Likewise.
20676         (vect_get_new_vect_var): Adjust.
20677         (vect_create_destination_var): Preserve SSA name versions.
20678         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20679         not dump anything here.
20681 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20683         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20684         Add member lr_slot_known.
20685         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20686         if necessary.
20687         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20688         Remove code that sets lr_slot_offset according to what a previous
20689         version of epiphany_emit_save_restore used to do.
20690         (epiphany_emit_save_restore): When doing an lr save or restore,
20691         set/verify lr_slot_known and lr_slot_offset.
20693 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20695         PR target/54338
20696         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20697         in ALL_REGS.
20699 2013-04-08  Richard Biener  <rguenther@suse.de>
20701         * alias.c (find_base_term): Fix thinko in previous change.
20703 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20705         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20706         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20707         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20708         if possible to compute val.
20709         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20710         For QImode integers don't require anything about precision.  Use
20711         const_with_all_bytes_same to find out if the constant doesn't have
20712         repeated bytes in it.
20714 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20716         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20717         within mode size.
20719 2013-04-08  Marek Polacek  <polacek@redhat.com>
20721         PR rtl-optimization/48182
20722         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20723         value to 1.
20725 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20727         PR target/55487
20728         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20729         nuses, make sure we have a label.
20731 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20733         PR target/56843
20734         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20735         (rs6000_emit_swdiv_low_precision): Remove.
20736         (rs6000_emit_swdiv): Rewrite to handle between one and four
20737         iterations of Newton-Raphson generally; modify required number of
20738         iterations for some cases.
20739         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20741 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20743         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20744         set-but-unused variable.
20746         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20747         basic blocks of released function bodies garbage-collectable.
20749         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20750         (struct rtl_opt_pass): Add TODO_df_finish.
20752         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20754 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20756         * config/arm/constraints.md (q): New constraint.
20757         * config/arm/ldrdstrd.md: New file.
20758         * config/arm/arm.md (ldrdstrd.md) New include.
20759         (arm_movdi): Use "q" instead of "r" constraint
20760         for double-word memory access.
20761         (movdf_soft_insn): Likewise.
20762         * config/arm/vfp.md (movdi_vfp): Likewise.
20763         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20764         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20765         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20766         (mem_ok_for_ldrd_strd): Likewise.
20767         (output_move_double): Update assertion.
20769 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20771         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20773 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20775         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20776         define_insn_and_split.
20777         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20779 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20781         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20782         define_insn_and_split.
20783         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20784         (shiftsi3_compare): New pattern.
20785         (rrx): New pattern.
20786         * config/arm/unspecs.md (UNSPEC_RRX): New.
20788 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20790         * config/arm/arm.md (negdi_extendsidi): New pattern.
20791         (negdi_zero_extendsidi): Likewise.
20793 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20795         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20796         define_insn_and_split.
20797         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20798         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20800 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20802         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20803         define_insn_and_split.
20804         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20805         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20807 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20809         * config/arm/arm.md (subsi3_carryin): New pattern.
20810         (subsi3_carryin_const): Likewise.
20811         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20812         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20814 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20816         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20818 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20820         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20821         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20823 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20825         * config/arm/arm.c (arm_expand_builtin): Change fcode
20826         type to unsigned int.
20828 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20830         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20832 2013-04-04  Ian Lance Taylor  <iant@google.com>
20834         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20835         language standard.
20837 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20839         PR middle-end/56729
20840         * df-scan.c (df_insn_delete): Disable failing assert.
20842 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20844         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20845         New function prototype.
20846         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20847         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20848         (arm_builtin_vectorized_function): New function.
20850 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20852         * config/arm/arm_neon_builtins.def: New file.
20853         * config/arm/arm.c (neon_builtin_data): Move contents to
20854         arm_neon_builtins.def.
20855         (enum arm_builtins): Include neon builtin definitions.
20856         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20857         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20859 2013-04-04  Marek Polacek  <polacek@redhat.com>
20861         PR tree-optimization/48186
20862         * predict.c (maybe_hot_frequency_p): Return false if
20863         HOT_BB_FREQUENCY_FRACTION is 0.
20864         (cgraph_maybe_hot_edge_p): Likewise.
20866 2013-04-04  Richard Biener  <rguenther@suse.de>
20868         PR tree-optimization/56826
20869         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20870         more accurately.
20872 2013-04-04  Richard Biener  <rguenther@suse.de>
20874         PR tree-optimization/56213
20875         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20876         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20878 2013-04-04  Richard Biener  <rguenther@suse.de>
20880         PR tree-optimization/56837
20881         * tree-loop-distribution.c (classify_partition): For non-zero
20882         values require that the value has the same precision as its
20883         mode to be useful as memset value.
20885 2013-04-03  Nick Clifton  <nickc@redhat.com>
20887         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20888         (fmssf4): Use fmsf.s on E3V5 architectures.
20889         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20890         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20892 2013-04-03  Jeff Law  <law@redhat.com>
20894         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20895         (lra-eliminations.o): Likewise.
20897 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20899         * gcov-io.c (compute_working_sets): Moved most of body of old
20900         compute_working_sets here from profile.c.
20901         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20902         (gcov_working_set_t): Moved typedef here from basic-block.h
20903         (compute_working_set): Declare.
20904         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20905         (get_working_sets): Renamed from compute_working_set,
20906         replace most of body with call to new compute_working_sets.
20907         (get_exec_counts): Replace call to compute_working_sets
20908         to get_working_sets.
20909         * profile.h (get_working_sets): Renamed from compute_working_set.
20910         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20911         to get_working_sets.
20912         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20913         * gcov-dump.c (dump_working_sets): New function.
20915 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20917         * hwint.c (sext_hwi, zext_hwi): New functions.
20918         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20919         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20920         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20921         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20922         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20923         (sext_hwi, zext_hwi): New functions.
20925 2013-04-03  Jeff Law  <law@redhat.com>
20927         PR tree-optimization/56799
20928         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20929         back test for widening conversion erroneously dropped in prior change.
20931 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20933         PR target/56809
20934         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20935         instead of next_real_insn.
20937 2013-04-03  Marek Polacek  <polacek@redhat.com>
20939         PR sanitizer/55702
20940         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20942 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20944         PR target/56809
20945         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20946         next_real_insn.
20947         (thumb1_output_casesi): Likewise.
20948         (thumb2_output_casesi): Likewise.
20950 2013-04-03  Richard Biener  <rguenther@suse.de>
20952         PR tree-optimization/56817
20953         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
20954         Split out ...
20955         (tree_unroll_loops_completely_1): ... new function to manually
20956         walk the loop tree, properly defering outer loops of unrolled
20957         loops to later iterations.
20959 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20961         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
20962         (vectorizable_load): Likewise.
20963         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
20964         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
20966 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20968         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
20969         BIT_FIELD_REF.
20971 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
20973         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
20975 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
20977         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
20979 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20981         PR tree-optimization/56790
20982         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
20983         folding.
20985 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
20987         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
20988         Handle VEC_MERGE.
20989         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
20990         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
20991         equal arguments.
20993 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
20995         PR c/19449
20996         * tree.h (force_folding_builtin_constant_p): New decl.
20997         * builtins.c (force_folding_builtin_constant_p): New variable.
20998         (fold_builtin_constant_p): Fold immediately also if
20999         force_folding_builtin_constant_p.
21001 2013-04-03  Richard Biener  <rguenther@suse.de>
21003         PR tree-optimization/56812
21004         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21005         DRs of the same interleaving chain are independent.
21007 2013-04-02  Jason Merrill  <jason@redhat.com>
21009         * gdbinit.in (pbb): Use debug fn.
21011 2013-04-02  Lawrence Crowl  <crowl@google.com>
21013         * sese.h (struct ivtype_map_elt_s): Remove unused.
21014         (extern debug_ivtype_map): Remove unused.
21015         (extern eq_ivtype_map_elts): Remove unused.
21016         * sese.c (debug_ivtype_map): Removed unused.
21017         (debug_ivtype_map_1): Removed unused.
21018         (debug_ivtype_elt): Remove unused.
21019         (eq_ivtype_map_elts): Remove unused.
21022 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21024         PR target/52790
21025         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21026         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21027         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21028         function.
21029         (legitimize_pe_coff_symbol): Likewise.
21030         (is_imported_p): New helper-function.
21031         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21032         for Windows x64 targets.
21033         (ix86_expand_prologue): Optimize for pe-coff targets.
21034         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21035         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21036         medium/large code-model.
21037         (legitimize_pic_address): Likewise.
21038         (legitimize_tls_address): Likewise.
21039         (ix86_expand_call): Likewise.
21040         (x86_output_mi_thunk): Likewise.
21041         (get_dllimport_decl): Add new beimport argument.
21042         (construct_plt_address): Don't assert for x64 pe-coff targets.
21043         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21044         targets.
21045         (SYMBOL_FLAG_STUBVAR): New macro.
21046         (SYMBOL_REF_STUBVAR_P): Likewise.
21047         * config/i386/winnt.c (stub_list): New structure.
21048         (stub_head): New local variable.
21049         (i386_pe_record_stub): New function.
21050         (i386_pe_file_end): Emit refptr-stubs.
21052 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21054         PR rtl-optimization/56745
21055         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21056         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21058         PR c++/34949
21059         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21060         and both of them are MEM_REFs, just compare first argument for
21061         equality and attempt to deal even with differing offsets.
21063         PR c++/34949
21064         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21065         of gimple_clobber_p to be MEM_REF.
21066         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21067         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21068         after gimplification.
21069         * asan.c (get_mem_ref_of_assignment): Don't instrument
21070         gimple_clobber_p stmts.
21071         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21072         gimple_clobber_p stmt if they have MEM_REF lhs and
21073         are dead because of another gimple_clobber_p stmt.
21074         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21075         gimple_clobber_p stmts like debug stmts.
21076         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21077         that refer to unused VAR_DECLs or uninitialized values.
21078         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21079         gimple_clobber_p stmts if they refer to removed parameters.
21080         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21081         formatting.
21083 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21085         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21086         using SWI48 mode attribute.
21088 2013-04-02  Wei Mi  <wmi@google.com>
21090         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21091         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21092         *<rotate_insn><mode>3_mask in i386.md.
21094 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21096         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21098 2013-04-02  Richard Biener  <rguenther@suse.de>
21100         PR tree-optimization/56778
21101         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21102         Runtime alias tests are not supported for gather loads.
21103         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21104         stmts referenced from SSA operands before updating SSA form.
21106 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21107             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21109         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21110         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21111         * config/arm/cortex-a53.md: New file.
21112         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21113         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21114         * config/arm/arm.c (arm_issue_rate): Likewise.
21115         * config/arm/arm-tune.md: Regenerate
21116         * config/arm/arm-tables.opt: Regenerate.
21117         * config/arm/arm-cores.def: Add cortex-a53.
21119 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21121         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21122         non-static link.
21124 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21126         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21127         scalar load/store operations using B/H registers.
21128         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21130 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21132         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21133         scalar move.
21134         * config/aarch64/aarch64.c
21135         (aarch64_simd_scalar_immediate_valid_for_move): New.
21136         * config/aarch64/aarch64-protos.h
21137         (aarch64_simd_scalar_immediate_valid_for_move): New.
21138         * config/aarch64/constraints.md (Dh, Dq): New.
21139         * config/aarch64/iterators.md (hq): New.
21141 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21143         * reorg.c (get_branch_condition): Deal with conditional returns.
21144         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21146 2013-04-01  Wei Mi  <wmi@google.com>
21148         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21149         Truncate operand 2 using %b asm operand modifier.
21150         (*<shift_insn><mode>3_mask): Ditto.
21151         (*<rotate_insn><mode>3_mask): Ditto.
21153 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21155         PR middle-end/56798
21156         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21158 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21160         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21161         of next_real_insn.
21162         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21164 2013-03-30  Lawrence Crowl  <crowl@google.com>
21166         * dse.c (clear_alias_sets): Remove never set.
21167         (disqualified_clear_alias_sets): Remove never set.
21168         (clear_alias_mode_pool): Remove never set.
21169         (dse_step0): Remove condition that is never true.
21170         (canon_address): Remove condition that is never true.
21171         (dse_step7): Remove condition that is never true.
21172         (rest_of_handle_dse): Remove condition that is never true.
21173         (rest_of_handle_dse::did_global): Remove never read from above.
21174         (dse_step2_spill): Remove never called from above.
21175         (dse_step5_spill): Remove never called from above.
21177 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21179         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21180         JUMP_TABLE_DATA changes.
21181         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21182         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21183         (Insns) <jump_table_data>: New entry.
21184         * doc/tm.texi: Regenerate.
21186         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21188         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21189         for table jump at the end of a basic block using tablejump_p.
21190         * targhooks.c (default_invalid_within_doloop): Likewise.
21191         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21192         target hook implementation that is identical to the default hook.
21193         (rs6000_invalid_within_doloop): Remove.
21195         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21196         unused variable from tablejump_p call.
21198         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21199         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21200         (INSN_DELETED_P): Likewise.
21201         (emit_jump_table_data): New prototype.
21202         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21203         after 4th as unused.
21204         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21205         * sched-vis.c (print_insn): Likewise.
21206         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21207         insn for compatibility with back ends that use next_active_insn to
21208         identify jump table data.
21209         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21210         (remove_insn): Likewise.
21211         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21212         to be emitted.
21213         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21214         (emit_jump_table_data): New function.
21216         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21217         basic block, a JUMP_TABLE_DATA never is.
21218         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21219         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21220         off from code handling real insns.
21221         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21222         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21223         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21224         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21225         is not a NONDEBUG_INSN_P.
21226         * ira-costs.c (scan_one_insn): Likewise.
21227         * jump.c (mark_all_labels): Likewise.
21228         (mark_jump_label_1): Likewise.
21229         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21230         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21231         a basic block.
21232         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21233         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21234         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21235         JUMP_TABLE_DATA_P insns.
21236         (calculate_elim_costs_all_insns): Likewise.
21237         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21238         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21239         (delete_output_reload): Code style fixups.
21240         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21241         insn flags on this non-insn.
21242         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21243         as scheduling barriers, for pre-change compatibility.
21244         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21245         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21247         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21248         redundant JUMP_TABLE_DATA_P test.
21249         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21250         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21251         (frv_for_each_packet): Likewise.
21252         * config/i386/i386.c (min_insn_size): Likewise.
21253         (ix86_avoid_jump_mispredicts): Likewise.
21254         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21255         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21256         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21257         (mips16_insn_length): Robustify.
21258         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21259         (mips16_split_long_branches): Likewise.
21260         * config/pa/pa.c (pa_combine_instructions): Likewise.
21261         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21262         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21263         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21264         as contributing to pool range lengths.
21265         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21266         Remove redundant JUMP_TABLE_DATA_P test.
21267         (sh_loop_align): Likewise.
21268         (split_branches): Likewise.
21269         (sh_insn_length_adjustment): Likewise.
21270         * config/spu/spu.c (get_branch_target): Likewise.
21272 2013-03-29  Jan Hubicka  <jh@suse.cz>
21274         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21275         gcov streaming; stream hot bb threshold to ltrans.
21276         * predict.c (get_hot_bb_threshold): Break out from ....
21277         (maybe_hot_count_p): ... here.
21278         (set_hot_bb_threshold): New function.
21279         * lto-section-in.c (lto_section_name): Add profile.
21280         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21281         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21282         and data-streamer.h
21283         (histogram_entry): New structure.
21284         (histogram, histogram_pool): New global vars.
21285         (histogram_hash): New structure.
21286         (histogram_hash::hash): New method.
21287         (histogram_hash::equal): Likewise.
21288         (account_time_size): New function.
21289         (cmp_counts): New function.
21290         (dump_histogram): New function.
21291         (ipa_profile_generate_summary): New function.
21292         (ipa_profile_write_summary): New function.
21293         (ipa_profile_read_summary): New function.
21294         (ipa_profile): Decide on threshold.
21295         (pass_ipa_profile): Add ipa_profile_write_summary and
21296         ipa_profile_read_summary.
21297         * Makefile.in (ipa.o): Update dependencies.
21298         * lto-streamer.h (LTO_section_ipa_profile): New section.
21300 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21302         * tree.h (VAR_P): New.
21304 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21306         PR lto/56777
21307         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21309 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21311         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21312         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21313         (control_flow_insn_p): Likewise.
21314         * cfgrtl.c (duplicate_insn_chain): Likewise.
21315         * final.c (get_attr_length_1): Likewise.
21316         (shorten_branches): Likewise.
21317         (final_scan_insn): Likewise.
21318         * function.c (instantiate_virtual_regs): Likewise.
21319         * gcse.c (insert_insn_end_basic_block): Likewise.
21320         * ira-costs.c (scan_one_insn): Likewise.
21321         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21322         * lra.c (check_rtl): Likewise.
21323         * reload1.c (elimination_costs_in_insn): Likewise.
21324         * reorg.c (follow_jumps): Likewise.
21326         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21327         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21328         (thumb_far_jump_used_p): Likewise.
21329         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21330         (workaround_speculation): Likewise.
21331         (add_sched_insns_for_speculation): Likewise.
21332         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21333         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21334         (frv_for_each_packet): Likewise.
21335         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21336         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21337         (final_emit_insn_group_barriers): Likewise.
21338         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21339         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21340         (mips16_insn_length): Likewise.
21341         * config/pa/pa.c (pa_reorg): Likewise.
21342         (pa_combine_instructions): Likewise.
21343         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21344         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21345         (sh_reorg): Likewise.
21346         (split_branches): Likewise.
21347         * config/spu/spu.c (get_branch_target): Likewise.
21349         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21350         JUMP_TABLE_DATA_P.
21352 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21354         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21355         Fix declaration name.
21357 2013-03-28  Lawrence Crowl  <crowl@google.com>
21359         * graphds.h (struct graph.indicies): Remove unused.
21360         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21361         (SCOP_ORIGINAL_PDDRS): Remove unused.
21362         * sese.h (extern insert_loop_close_phis): Removed unused.
21363         (extern insert_guard_phis): Removed unused.
21364         (extern ivtype_map_elt_info): Removed unused.
21365         (new_ivtype_map_elt): Removed unused.
21366         * sese.c (ivtype_map_elt_info): Removed unused.
21368 2013-03-28  Lawrence Crowl  <crowl@google.com>
21370         * Makefile.in: Add several missing include dependences.
21371         (DUMPFILE_H): New.
21372         (test-dump.o): New.  This object is not added to any executable,
21373         but is present for ad-hoc testing.
21374         * bitmap.c
21375         (debug (const bitmap_head_def &)): New.
21376         (debug (const bitmap_head_def *)): New.
21377         * bitmap.h
21378         (extern debug (const bitmap_head_def &)): New.
21379         (extern debug (const bitmap_head_def *)): New.
21380         * cfg.c
21381         (debug (edge_def &)): New.
21382         (debug (edge_def *)): New.
21383         * cfghooks.c
21384         (debug (basic_block_def &)): New.
21385         (debug (basic_block_def *)): New.
21386         * dumpfile.h
21387         (dump_node (const_tree, int, FILE *)): Correct source file.
21388         * dwarf2out.c
21389         (debug (die_struct &)): New.
21390         (debug (die_struct *)): New.
21391         * dwarf2out.h
21392         (extern debug (die_struct &)): New.
21393         (extern debug (die_struct *)): New.
21394         * gimple-pretty-print.c
21395         (debug (gimple_statement_d &)): New.
21396         (debug (gimple_statement_d *)): New.
21397         * gimple-pretty-print.h
21398         (extern debug (gimple_statement_d &)): New.
21399         (extern debug (gimple_statement_d *)): New.
21400         * ira-build.c
21401         (debug (ira_allocno_copy &)): New.
21402         (debug (ira_allocno_copy *)): New.
21403         (debug (ira_allocno &)): New.
21404         (debug (ira_allocno *)): New.
21405         * ira-int.h
21406         (extern debug (ira_allocno_copy &)): New.
21407         (extern debug (ira_allocno_copy *)): New.
21408         (extern debug (ira_allocno &)): New.
21409         (extern debug (ira_allocno *)): New.
21410         * ira-lives.c
21411         (debug (live_range &)): New.
21412         (debug (live_range *)): New.
21413         * lra-int.h
21414         (debug (lra_live_range &)): New.
21415         (debug (lra_live_range *)): New.
21416         * lra-lives.c
21417         (debug (lra_live_range &)): New.
21418         (debug (lra_live_range *)): New.
21419         * omega.c
21420         (debug (omega_pb_d &)): New.
21421         (debug (omega_pb_d *)): New.
21422         * omega.h
21423         (extern debug (omega_pb_d &)): New.
21424         (extern debug (omega_pb_d *)): New.
21425         * print-rtl.c
21426         (debug (const rtx_def &)): New.
21427         (debug (const rtx_def *)): New.
21428         * print-tree.c
21429         (debug_tree (tree): Move within file.
21430         (debug_raw (const tree_node &)): New.
21431         (debug_raw (const tree_node *)): New.
21432         (dump_tree_via_hooks (const tree_node *, int)): New.
21433         (debug (const tree_node &)): New.
21434         (debug (const tree_node *)): New.
21435         (debug_verbose (const tree_node &)): New.
21436         (debug_verbose (const tree_node *)): New.
21437         (debug_head (const tree_node &)): New.
21438         (debug_head (const tree_node *)): New.
21439         (debug_body (const tree_node &)): New.
21440         (debug_body (const tree_node *)): New.
21441         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21442         (debug (vec<tree, va_gc> &)): New.
21443         (debug (vec<tree, va_gc> *)): New.
21444         * rtl.h
21445         (extern debug (const rtx_def &)): New.
21446         (extern debug (const rtx_def *)): New.
21447         * sbitmap.c
21448         (debug_raw (simple_bitmap_def &)): New.
21449         (debug_raw (simple_bitmap_def *)): New.
21450         (debug (simple_bitmap_def &)): New.
21451         (debug (simple_bitmap_def *)): New.
21452         * sbitmap.h
21453         (extern debug (simple_bitmap_def &)): New.
21454         (extern debug (simple_bitmap_def *)): New.
21455         (extern debug_raw (simple_bitmap_def &)): New.
21456         (extern debug_raw (simple_bitmap_def *)): New.
21457         * sel-sched-dump.c
21458         (debug (vinsn_def &)): New.
21459         (debug (vinsn_def *)): New.
21460         (debug_verbose (vinsn_def &)): New.
21461         (debug_verbose (vinsn_def *)): New.
21462         (debug (expr_def &)): New.
21463         (debug (expr_def *)): New.
21464         (debug_verbose (expr_def &)): New.
21465         (debug_verbose (expr_def *)): New.
21466         (debug (vec<rtx> &)): New.
21467         (debug (vec<rtx> *)): New.
21468         * sel-sched-dump.h
21469         (extern debug (vinsn_def &)): New.
21470         (extern debug (vinsn_def *)): New.
21471         (extern debug_verbose (vinsn_def &)): New.
21472         (extern debug_verbose (vinsn_def *)): New.
21473         (extern debug (expr_def &)): New.
21474         (extern debug (expr_def *)): New.
21475         (extern debug_verbose (expr_def &)): New.
21476         (extern debug_verbose (expr_def *)): New.
21477         (extern debug (vec<rtx> &)): New.
21478         (extern debug (vec<rtx> *)): New.
21479         * sel-sched-ir.h
21480         (_list_iter_cond_expr): Make inline instead of static.
21481         * sreal.c
21482         (debug (sreal &)): New.
21483         (debug (sreal *)): New.
21484         * sreal.h
21485         (extern debug (sreal &)): New.
21486         (extern debug (sreal *)): New.
21487         * tree.h
21488         (extern debug_raw (const tree_node &)): New.
21489         (extern debug_raw (const tree_node *)): New.
21490         (extern debug (const tree_node &)): New.
21491         (extern debug (const tree_node *)): New.
21492         (extern debug_verbose (const tree_node &)): New.
21493         (extern debug_verbose (const tree_node *)): New.
21494         (extern debug_head (const tree_node &)): New.
21495         (extern debug_head (const tree_node *)): New.
21496         (extern debug_body (const tree_node &)): New.
21497         (extern debug_body (const tree_node *)): New.
21498         (extern debug (vec<tree, va_gc> &)): New.
21499         (extern debug (vec<tree, va_gc> *)): New.
21500         * tree-cfg.c
21501         (debug (struct loop &)): New.
21502         (debug (struct loop *)): New.
21503         (debug_verbose (struct loop &)): New.
21504         (debug_verbose (struct loop *)): New.
21505         * tree-dump.c: Add header dependence.
21506         * tree-flow.h
21507         (extern debug (struct loop &)): New.
21508         (extern debug (struct loop *)): New.
21509         (extern debug_verbose (struct loop &)): New.
21510         (extern debug_verbose (struct loop *)): New.
21511         * tree-data-ref.c
21512         (debug (data_reference &)): New.
21513         (debug (data_reference *)): New.
21514         (debug (vec<data_reference_p> &)): New.
21515         (debug (vec<data_reference_p> *)): New.
21516         (debug (vec<ddr_p> &)): New.
21517         (debug (vec<ddr_p> *)): New.
21518         * tree-data-ref.h
21519         (extern debug (data_reference &)): New.
21520         (extern debug (data_reference *)): New.
21521         (extern debug (vec<data_reference_p> &)): New.
21522         (extern debug (vec<data_reference_p> *)): New.
21523         (extern debug (vec<ddr_p> &)): New.
21524         (extern debug (vec<ddr_p> *)): New.
21525         * tree-ssa-alias.c
21526         (debug (pt_solution &)): New.
21527         (debug (pt_solution *)): New.
21528         * tree-ssa-alias.h
21529         (extern debug (pt_solution &)): New.
21530         (extern debug (pt_solution *)): New.
21531         * tree-ssa-alias.c
21532         (debug (_var_map &)): New.
21533         (debug (_var_map *)): New.
21534         (debug (tree_live_info_d &)): New.
21535         (debug (tree_live_info_d *)): New.
21536         * tree-ssa-alias.h
21537         (extern debug (_var_map &)): New.
21538         (extern debug (_var_map *)): New.
21539         (extern debug (tree_live_info_d &)): New.
21540         (extern debug (tree_live_info_d *)): New.
21542 2013-03-28  Jan Hubicka  <jh@suse.cz>
21544         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21546 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21548         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21549         record only when desired or required.
21551 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21553         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21554         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21556 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21558         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21559         (*andsi3_compare0_uxtw): New pattern.
21560         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21561         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21563 2013-03-28  Jan Hubicka  <jh@suse.cz>
21565         * data-streamer-in.c (streamer_read_gcov_count): New function.
21566         * gimple-streamer-out.c: Include value-prof.h.
21567         (output_gimple_stmt): Output histogram.
21568         (output_bb): Use streamer_write_gcov_count.
21569         * value-prof.c: Include data-streamer.h
21570         (dump_histogram_value): Add HIST_TYPE_MAX.
21571         (stream_out_histogram_value): New function.
21572         (stream_in_histogram_value): New function.
21573         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21574         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21575         * data-streamer-out.c (streamer_write_gcov_count): New function.
21576         (streamer_write_gcov_count_stream): New function.
21577         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21578         (lto_output_node): Likewise.
21579         (input_node, input_edge): Likewise.
21580         * lto-streamer-out.c (output_cfg): Update streaming.
21581         * lto-streamer-in.c (input_cfg): Likewise.
21582         * data-streamer.h (streamer_write_gcov_count,
21583         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21584         * gimple-streamer-in.c: Include value-prof.h
21585         (input_gimple_stmt): Input histograms.
21586         (input_bb): Update profile streaming.
21588 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21590         * genmodes.c (emit_max_int): New function.
21591         (emit_insn_modes_h): Added call to emit_max_function.
21592         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21593         Added doc.
21594         * machmode.def: Fixed comment.
21596 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21598         * combine.c (try_combine): Removed useless assert.
21599         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21601 2013-03-28  Marek Polacek  <polacek@redhat.com>
21602             Richard Biener  <rguenther@suse.de>
21604         PR tree-optimization/56695
21605         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21606         build signed result of a vector comparison.
21607         * tree-cfg.c (verify_gimple_comparison): Check that a result
21608         of a vector comparison has signed type.
21610 2013-03-28  Richard Biener  <rguenther@suse.de>
21612         PR tree-optimization/37021
21613         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21614         do not restrict gaps between groups.
21615         * tree-vect-stmts.c (vectorizable_load): Properly account for
21616         a gap between groups.
21618 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21620         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21621         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21622         is not enabled.
21624 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21626         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21627         * doc/extend.texi (Named Address Spaces): Ditto.
21628         (Variable Attributes): Ditto.
21630 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21632         * config.build: Add support for cygwin x64 target.
21633         * config.gcc: Likewise.
21634         * config.host: Likewise.
21635         * configure.ac: Likewise
21636         * configure: Regenerated.
21638 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21640         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21641         * config/i386/t-cygwin-w64: New file.
21642         * config/i386/cygwin-w64.h: New file.
21643         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21644         and add support for x64-cygwin target.
21645         (CPP_SPEC): Likewise.
21646         (CXX_WRAP_SPEC_LIST): Undefine before define.
21647         (LIBGCJ_SONAME): Use 15 as version.
21649 2013-03-27  Richard Biener  <rguenther@suse.de>
21651         PR tree-optimization/56716
21652         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21653         dumping for ref nodes.
21655 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21657         PR tree-optimization/55334
21658         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21659         restricted pointers to arrays.
21661 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21663         * Makefile.in (.SUFFIXES): Add .cc.
21664         (.c.o): Apply same recipe for implicit rule .cc.o.
21666 2013-03-27  Richard Biener  <rguenther@suse.de>
21668         PR tree-optimization/37021
21669         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21670         REALPART/IMAGPART_EXPRs around the supported refs.
21671         * tree-ssa-structalias.c (find_func_aliases): Assume that
21672         floating-point values are not used to transfer pointers.
21674 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21676         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21677         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21678         * doc/tm.texi: Regenerate.
21679         * targhooks.h (default_has_ifunc_p): New.
21680         * targhooks.c (default_has_ifunc_p): Ditto.
21681         * config/linux-protos.h: New file.
21682         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21683         hook for linux which disables support of indirect functions in android.
21684         * config/linux-android.c: New file.
21685         * config/t-linux-android.c: Ditto.
21686         * config.gcc: Added new object file linux-android.o.
21687         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21688         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21689         * varasm.c (do_assemble_alias): Likewise.
21690         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21691         doesn't support indirect functions.
21692         * configure: Regenerate.
21694 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21696         PR target/56102
21697         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21698         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21699         mult-word mode.
21701 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21703         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21705 2013-03-27  Terry Guo  <terry.guo@arm.com>
21707         * config/arm/arm-cores.def: Added core cortex-r7.
21708         * config/arm/arm-tune.md: Regenerated.
21709         * config/arm/arm-tables.opt: Regenerated.
21710         * doc/invoke.texi: Added entry for core cortex-r7.
21712 2013-03-27  Walter Lee  <walt@tilera.com>
21714         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21715         double-decrement of next_scratch_regno.
21717 2013-03-27  Walter Lee  <walt@tilera.com>
21719         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21720         input operands.
21721         (insn_v1mulus): Ditto.
21722         (insn_v2muls): Ditto.
21724 2013-03-27  Walter Lee  <walt@tilera.com>
21726         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21727         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21729 2013-03-27  Walter Lee  <walt@tilera.com>
21731         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21732         (*sibcall_value): Ditto.
21734 2013-03-27  Walter Lee  <walt@tilera.com>
21736         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21737         (insn_mnz_v8qi): ... this ...
21738         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21739         vector equivalent.
21740         (insn_v<n>mnz): Replaced by ...
21741         (insn_v1mnz): ... this ...
21742         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21743         equivalent.
21744         (insn_mz_<mode>): Replaced by ...
21745         (insn_mz_v8qi): ... this ...
21746         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21747         vector equivalent.
21748         (insn_v<n>mz): Replaced by ...
21749         (insn_v1mz): ... this ...
21750         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21751         equivalent.
21753 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21755         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21757 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21759         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21760         than fprintf with a non-constant, non-format string.
21762 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21764         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21765         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21766         operand 0 predicate.
21767         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21768         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21769         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21770         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21771         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21772         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21773         (mov<mode>_insv_1): Remove expander.  Merge insn with
21774         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21775         Use general_x64nomem_operand as operand 1 predicate.
21776         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21777         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21778         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21779         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21780         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21781         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21782         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21783         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21784         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21785         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21786         (general_x64nomem_operand): Ditto.
21788 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21790         * config/rtems.opt: Add -pthread option.
21792 2013-03-26  Richard Biener  <rguenther@suse.de>
21794         * alias.c (find_base_term): Avoid redundant and not used recursion.
21795         (base_alias_check): Get the initial base term from the caller.
21796         (true_dependence_1): Compute and pass base terms to base_alias_check.
21797         (write_dependence_p): Likewise.
21798         (may_alias_p): Likewise.
21800 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21802         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21803         PC-relative load in SI modes and above only.
21805 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21807         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21808         * config/arm/iwmmxt.md (WCGR0): Update.
21809         (WCGR1, WCGR2, WCGR3): Likewise.
21811 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21813         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21814         Use x64 and nox64 isa attributes.
21816 2013-03-26  Richard Biener  <rguenther@suse.de>
21818         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21819         alignment computations and rely on get_object_alignment_1
21820         for the !TYPE_P case.
21821         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21823 2013-03-26  Walter Lee  <walt@tilera.com>
21825         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21826         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21828 2013-03-25  Jeff Law  <law@redhat.com>
21830         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21831         check for INTEGRAL_TYPE_P that was missing due to checking in
21832         wrong version of prior patch.
21834 2013-03-25  Walter Lee  <walt@tilera.com>
21836         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21837         TILEGX_INSN_SHUFFLEBYTES1.
21838         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21839         shufflebytes1.
21840         (tilegx_builtins): Ditto.
21841         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21843 2013-03-25  Walter Lee  <walt@tilera.com>
21845         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21846         (floatunssisf2): New pattern.
21847         (floatsidf2): New pattern.
21848         (floatunssidf2): New pattern.
21850 2013-03-25  Walter Lee  <walt@tilera.com>
21852         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21853         tests for constraint J, K, N, P.
21855 2013-03-25  Walter Lee  <walt@tilera.com>
21857         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21858         Use indirect/pcrel encoding.
21859         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21860         Ditto.
21862 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21864         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21865         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21866         * config/mips/mips.c (mips_option_override): Set IMADD default.
21867         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21868         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21869         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21870         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21871         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21873 2013-03-25  Jeff Law  <law@redhat.com>
21875         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21876         slightly to avoid creating and folding useless trees.  Simplify
21877         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21879 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21881         * config/i386/i386.md (*zero_extendsidi2): Merge with
21882         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21883         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21884         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21885         targets.  Clarify comment.
21887 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21889         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21890         pass-through jump functions differently.
21891         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21892         up jump functions.
21894 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21896         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21897         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21898         process it.
21899         (ipa_get_indirect_edge_target): New function.
21900         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21901         ipa_get_indirect_edge_target.  Update all callers.
21902         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21903         ipa_get_indirect_edge_target_1 instead of calling
21904         ipa_get_indirect_edge_target.
21905         (create_specialized_node): Pass aggvlas to
21906         ipcp_discover_new_direct_edges.
21908 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21910         * config/arm/arm.md (f_sels, f_seld): New types.
21911         (*cmov<mode>): New pattern.
21912         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21913         predicate.
21915 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21917         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21918         POSIX-printf for mingw-hosted  builds.
21920 2013-03-25  Richard Biener  <rguenther@suse.de>
21922         PR middle-end/56694
21923         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21924         must-not-throw stmt location.
21926 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21928         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21929         Emit load-acquire versions when acq is true.
21930         (arm_emit_store_exclusive): Add rel parameter.
21931         Emit store-release versions when rel is true.
21932         (arm_split_compare_and_swap): Use acquire-release instructions
21933         instead.
21934         of barriers when appropriate.
21935         (arm_split_atomic_op): Likewise.
21936         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21937         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21938         (VUNSPEC_SLX): Likewise.
21939         (VUNSPEC_LDA): Likewise.
21940         (VUNSPEC_STL): Likewise.
21941         * config/arm/sync.md (atomic_load<mode>): New pattern.
21942         (atomic_store<mode>): Likewise.
21943         (arm_load_acquire_exclusive<mode>): Likewise.
21944         (arm_load_acquire_exclusivesi): Likewise.
21945         (arm_load_acquire_exclusivedi): Likewise.
21946         (arm_store_release_exclusive<mode>): Likewise.
21948 2013-03-25  Catherine Moore  <clm@codesourcery.com>
21950         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
21951         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
21952         * config/mip/predicates.md (lwsp_swsp_operand,
21953         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
21954         sb16_operand, db4_operand, db7_operand, ib3_operand,
21955         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
21956         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
21957         andi16_operand): New predicates.
21958         * config/mips/mips.md (compression): New attribute.
21959         (enabled): New attribute.
21960         (length): Consider compression in computing length.
21961         (shift_compression): New code attribute.
21962         (*add<mode>3): New operands. Record compression.
21963         (sub<mode>3): Likewise.
21964         (one_cmpl<mode>2): Likewise.
21965         (*and<mode>3): Likewise.
21966         (*ior<mode>3): Likewise.
21967         (unnamed pattern for xor): Likewise.
21968         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
21969         (*<optab><mode>3): Likewise.
21970         (*mov<mode>_internal: Likewise.
21971         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
21972         (mips_unsigned_immediate_p): New.
21973         (umips_lwsp_swsp_address_p): New.
21974         (m16_based_address_p): New.
21975         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
21976         (mips_unsigned_immediate_p): New prototype.
21977         (lwsp_swsp_address_p): New prototype.
21978         (m16_based_address_p): New prototype.
21979         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
21980         (mips_signed_immediate_p): New function.
21981         (m16_based_address_p): New function.
21982         (lwsp_swsp_address_p): New function.
21983         (mips_print_operand_punctuation): Recognize short delay slot insns
21984         for microMIPS.add<mode>3"
21986 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21988         PR target/56720
21989         * config/arm/iterators.md (v_cmp_result): New mode attribute.
21990         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
21992 2013-03-25  Richard Biener  <rguenther@suse.de>
21994         PR tree-optimization/56689
21995         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
21996         any edge.
21998 2013-03-25  Richard Biener  <rguenther@suse.de>
22000         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22001         of bitmap.
22002         (memory_references): Likewise.
22003         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22004         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22005         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22006         (gather_mem_refs_in_loops): Fold into ...
22007         (analyze_memory_references): ... this.  Move initialization
22008         to tree_ssa_lim_initialize.
22009         (fill_always_executed_in): Rename to ...
22010         (fill_always_executed_in_1): ... this.
22011         (fill_always_executed_in): Move contains_call computation to
22012         this new function from ...
22013         (tree_ssa_lim_initialize): ... here.
22014         (tree_ssa_lim): Call fill_always_executed_in.
22016 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22018         * postreload.c (reload_combine): Fix code detecting returns.
22020 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22022         * function.c (emit_use_return_register_into_block): On cc0 targets,
22023         do not emit the sequence between cc0 setter and user.
22025 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22027         * config/i386/predicates.md (local_symbolic_operand): Interpret
22028         dll-imported symbols as none-local.
22030 2013-03-25  Richard Biener  <rguenther@suse.de>
22032         * tree-ssa-loop-im.c (struct depend): Remove.
22033         (struct lim_aux_data): Make depends a vec of gimples.
22034         (free_lim_aux_data): Adjust.
22035         (add_dependency): Likewise.
22036         (set_level): Likewise.
22038 2013-03-25  Richard Biener  <rguenther@suse.de>
22040         PR middle-end/56434
22041         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22042         the pointer returned by calls with ECF_MALLOC set.
22044 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22046         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22048 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22050         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22051         using MMXMODE mode iterator.
22052         (*move<mode>_internal): Merge with *movv2sf_internal and
22053         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22055 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22057         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22058         (record_last_mem_set_info): Likewise.
22060         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22061         of XNEWVEC followed by memset.
22062         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22064 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22066         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22067         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22068         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22069         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22070         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22071         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22072         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22073         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22074         BARRIER_P instead of GET_CODE.
22076 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22078         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22079         inaccuracy in the probing code.
22081         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22082         (ctrapdi4): Likewise.
22084 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22086         * calls.c (expand_call): Add missing guard to code handling return
22087         of non-BLKmode structures in MSB.
22088         * function.c (expand_function_end): Likewise.
22090 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22092         * combine.c (try_combine): Adjust comment.  Do not add the set of
22093         insn #0 if the destination indirectly is set or dies in insn #2.
22094         Tidy up code to distribute a new note.
22096 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22098         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22099         also for alternatives 16 and 17.
22101 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22103         * config/i386/sse.md (*mov<mode>_internal): Merge with
22104         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22105         Emit insn template depending on type attribute.  Use
22106         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22107         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22108         attribute calculation.  Remove unit attribute calculation.
22109         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22110         Set prefix_data16 attribute for DImode ssemov types.
22111         Use Ym instead of y for SSE-MMX conversion alternatives.
22112         Reorder operand constraints.
22114 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22116         * df.h (df_insn_delete): Adjust prototype.
22117         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22118         and let it decide whether mark the basic block dirty.
22119         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22120         * df-scan.c (df_insn_info_delete): New helper function, split
22121         off from df_insn_delete.
22122         (df_scan_free_bb_info): Use it.
22123         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22124         Likewise.
22125         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22126         that the insn is actually an insn and it has a non-NULL basic block.
22127         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22129 2013-03-22  Richard Biener  <rguenther@suse.de>
22131         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22132         dep_ref members.
22133         (mem_ref_alloc): Do not allocate them.
22134         (refs_independent_p): Do not query or maintain a cache.
22136 2013-03-22  Richard Biener  <rguenther@suse.de>
22138         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22139         (gather_mem_refs_in_loops): Do not compute it.
22140         (analyze_memory_references): Do not allocate it.
22141         (tree_ssa_lim_finalize): Do not free it.
22142         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22144 2013-03-22  Richard Biener  <rguenther@suse.de>
22146         * is-a.h (as_a): Use gcc_checking_assert.
22148 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22150         * config/aarch64/aarch64.c (aarch64_print_operand): New
22151         format specifier for printing a constant in hex.
22152         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22153         format specifier for printing second operand.
22155 2013-03-22  Richard Biener  <rguenther@suse.de>
22157         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22158         bitmaps.
22159         (gather_mem_refs_in_loops): Perform store accumulation here.
22160         (create_vop_ref_mapping_loop): Remove.
22161         (create_vop_ref_mapping): Likewise.
22162         (analyze_memory_references): Initialize refs_stored_in_loop.
22163         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22164         (record_indep_loop): Remove.
22165         (record_dep_loop): New function.
22166         (ref_indep_loop_p_1): Adjust to only walk over references
22167         in the loop, not its subloops.
22168         (ref_indep_loop_p): Rename to ...
22169         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22170         maintaining a more fine-grained cache.
22171         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22172         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22174 2013-03-22  Richard Biener  <rguenther@suse.de>
22176         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22177         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22178         aggregate mem_ref_loc.
22179         (free_mem_ref_locs): Inline into ...
22180         (memref_free): ... this and adjust.
22181         (mem_ref_alloc): Adjust.
22182         (mem_ref_locs_alloc): Remove.
22183         (record_mem_ref_loc): Adjust.
22184         (get_all_locs_in_loop): Rewrite into ...
22185         (for_all_locs_in_loop): ... this iterator.
22186         (rewrite_mem_ref_loc): New functor.
22187         (rewrite_mem_refs): Use for_all_locs_in_loop.
22188         (sm_set_flag_if_changed): New functor.
22189         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22190         (ref_always_accessed): New functor.
22191         (ref_always_accessed_p): Use for_all_locs_in_loop.
22193 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22195         * tree-pass.h (PROP_gimple_lvec): New.
22196         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22197         (init_optimization_passes): Move pass_lower_vector.
22198         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22199         PROP_gimple_lvec.
22200         (pass_lower_vector): Provide PROP_gimple_lvec.
22201         (pass_lower_vector_ssa): Likewise.
22202         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22204 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22206         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22208 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22210         * config/i386/i386.md (*movdi_internal): Disparage slightly
22211         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22212         conversion alternatives.
22214 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22216         PR middle-end/48087
22217         * diagnostic.def (DK_WERROR): New kind.
22218         * diagnostic.h (werrorcount): Define.
22219         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22220         promoted to DK_ERROR, increment DK_WERROR counter instead of
22221         DK_ERROR counter.
22222         * toplev.c (toplev_main): Call print_ignored_options even if
22223         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22224         even if just werrorcount is non-zero.
22226         PR debug/55608
22227         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22228         on failure.
22229         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22230         (string_cst_pool_decl): New function.
22231         (optimize_one_addr_into_implicit_ptr): New function.
22232         (resolve_addr_in_expr): Optimize DWARF location expression
22233         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22234         which doesn't live in memory, but has DW_AT_location or
22235         DW_AT_const_value, or refers to a string literal, into
22236         DW_OP_GNU_implicit_pointer.
22237         (optimize_location_into_implicit_ptr): New function.
22238         (resolve_addr): If removing DW_AT_location of a variable because
22239         it was DW_OP_addr of address of the variable, but the variable doesn't
22240         live in memory, try to emit const value attribute for the initializer.
22242 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22244         * tree.h (VECTOR_TYPE_P): New macro.
22245         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22246         TYPE_MODE): Use it.
22247         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22248         VEC_COND_EXPR cannot be lvalues.
22249         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22251 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22253         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22254         Restrict the transformation to equal modes.
22256 2013-03-21  Richard Biener  <rguenther@suse.de>
22258         PR tree-optimization/39326
22259         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22260         (MEM_ANALYZABLE): Adjust.
22261         (record_mem_ref_loc): Move bitmap ops ...
22262         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22263         unanalyzable refs, do not record locations for it.
22264         (analyze_memory_references): Allocate ref zero as shared
22265         unanalyzable ref.
22266         (refs_independent_p): Do not test for unanalyzed mems here.
22267         (ref_indep_loop_p_1): Special-case disambiguation against
22268         the unanalyzed ref.
22269         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22271 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22273         * config/arm/arm-protos.h (tune_params): Add
22274         prefer_neon_for_64bits field.
22275         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22276         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22277         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22278         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22279         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22280         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22281         (arm_option_override): Handle -mneon-for-64bits new option.
22282         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22283         (prefer_neon_for_64bits): Declare new variable.
22284         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22285         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22286         (arch_enabled): Handle new arch types. Remove support for onlya8
22287         and nota8.
22288         (one_cmpldi2): Use new arch names.
22289         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22290         * config/arm/arm.opt (mneon-for-64bits): Add option.
22291         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22292         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22293         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22294         of onlya8.
22295         * doc/invoke.texi (-mneon-for-64bits): Document.
22297 2013-03-21  Richard Biener  <rguenther@suse.de>
22299         PR tree-optimization/39326
22300         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22301         (sort_bbs_in_loop_postorder_cmp): New function.
22302         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22304 2013-03-21  Richard Biener  <rguenther@suse.de>
22306         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22307         (vect_insert_into_interleaving_chain): Likewise.
22308         (vect_drs_dependent_in_basic_block): Inline ...
22309         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22310         split out from ...
22311         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22312         (vect_check_interleaving): Simplify.
22313         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22314         (vect_slp_analyze_data_ref_dependences): ... this new function.
22315         (dr_group_sort_cmp): New function.
22316         (vect_analyze_data_ref_accesses): Compute data-reference groups
22317         here instead of in vect_analyze_data_ref_dependence.  Use
22318         a more efficient algorithm.
22319         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22320         vect_slp_analyze_data_ref_dependences.  Call
22321         vect_analyze_data_ref_accesses earlier.
22322         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22323         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22324         (vect_slp_analyze_data_ref_dependences): New prototype.
22326 2013-03-21  Richard Biener  <rguenther@suse.de>
22328         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22329         ref is stored in the loop.
22330         (find_refs_for_sm): Walk only over all stores.
22331         (store_motion_loop): Allocate from lim_bitmap_obstack.
22332         (store_motion): Likewise.
22334 2013-03-21  Richard Biener  <rguenther@suse.de>
22336         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22337         Update virtual SSA form.
22339 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22341         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22342         * configure: Regenerate.
22343         * config.in: Regenerate.
22344         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22345         if !HAVE_LD_EH_FRAME_CIEV3.
22347 2013-03-21  Richard Biener  <rguenther@suse.de>
22349         * tree-cfg.c (verify_expr_no_block): New function.
22350         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22351         nor DECL_VALUE_EXPR have locations with associated blocks.
22352         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22353         (clear_unused_block_pointer): Remove code dealing with
22354         blocks in DECL_DEBUG_EXPR locations.
22356 2013-03-21  Richard Biener  <rguenther@suse.de>
22358         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22359         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22360         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22361         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22362         instead of DECL_DEBUG_EXPR_IS_FROM.
22363         * gimplify.c (gimplify_modify_expr): Likewise.
22364         * tree-cfg.c (verify_expr_location_1): Likewise.
22365         * tree-complex.c (create_one_component_var): Likewise.
22366         * tree-sra.c (create_access_replacement): Likewise.
22367         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22368         (clear_unused_block_pointer): Likewise.
22369         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22370         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22371         * var-tracking.c (var_debug_decl): Likewise.
22372         (track_expr_p): Likewise.
22373         * tree-inline.c (add_local_variables): Likewise.  Set
22374         DECL_HAS_DEBUG_EXPR_P after copying it.
22375         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22376         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22378 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22380         PR bootstrap/56656
22381         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22382         * configure: Regenerate.
22383         * config.in: Regenerate.
22384         * config/i386/i386.md (*movdf_internal): Use
22385         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22386         movd instead of movq mnemonic for interunit moves.
22387         (*movdi_internal): Ditto.
22389 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22391         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22392         (abd<mode>_3): New pattern.
22393         (aba<mode>_3): New pattern.
22394         (fabd<mode>_3): New pattern.
22396 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22398         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22399         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22400         occurrence of REGISTER_PREFIX as its empty string.
22402 2013-03-20  Jeff Law  <law@redhat.com>
22404         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22405         addititional equivalences for equality comparisons between an SSA_NAME
22406         and a constant where the SSA_NAME was set from a widening conversion.
22408 2013-03-20  Walter Lee  <walt@tilera.com>
22410         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22412 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22414         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22415         depending on type attribute.
22416         (*movti_internal): Ditto.
22417         (*movtf_internal): Ditto.
22418         (*movxf_internal): Ditto.
22419         (*movdf_internal): Ditto.
22420         (*movsf_internal): Ditto.
22422 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22424         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22425         maybe_vex for sselog1 and ssemov types.
22426         (*movdi_internal): Reorder operand constraints.
22427         (*movsi_internal): Ditto.  Set prefix attribute to
22428         maybe_vex for sselog1 and ssemov types.
22429         (*movtf_internal): Set prefix attribute to maybe_vex
22430         for sselog1 and ssemov types.
22431         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22432         DImode ssemov types.  Reorder operand constraints.
22433         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22434         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22435         attribute for SImode ssemov types.  Reorder operand constraints.
22437 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22439         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22440         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22442 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22444         * config/rs6000/predicates.md (indexed_address, update_address_mem
22445         update_indexed_address_mem): New predicates.
22446         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22447         attribute for load/store instructions.
22448         * config/rs6000/dfp.md (movsd_store): Likewise.
22449         (movsd_load): Likewise.
22450         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22451         (unnamed HI->DI extend define_insn): Likewise.
22452         (unnamed SI->DI extend define_insn): Likewise.
22453         (unnamed QI->SI extend define_insn): Likewise.
22454         (unnamed QI->HI extend define_insn): Likewise.
22455         (unnamed HI->SI extend define_insn): Likewise.
22456         (unnamed HI->SI extend define_insn): Likewise.
22457         (extendsfdf2_fpr): Likewise.
22458         (movsi_internal1): Likewise.
22459         (movsi_internal1_single): Likewise.
22460         (movhi_internal): Likewise.
22461         (movqi_internal): Likewise.
22462         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22463         attribute for load/store instructions.
22464         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22465         instructions.
22466         (mov<mode>_softfloat): Likewise.
22467         (mov<mode>_hardfloat32): Likewise.
22468         (mov<mode>_hardfloat64): Likewise.
22469         (mov<mode>_softfloat64): Likewise.
22470         (movdi_internal32): Likewise.
22471         (movdi_internal64): Likewise.
22472         (probe_stack_<mode>): Likewise.
22474 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22476         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22477         floating point, and decimal floating point to reload iterator.
22479         * config/rs6000/constraints.md (wl constraint): New constraints to
22480         return FLOAT_REGS if certain options are used to reduce the number
22481         of separate patterns that exist in the file.
22482         (wx constraint): Likewise.
22483         (wz constraint): Likewise.
22485         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22486         -mdebug=reg, print wg, wl, wx, and wz constraints.
22487         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22488         Initialize the reload functions for 64-bit binary/decimal floating
22489         point types.
22490         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22491         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22492         create the buffer on the stack to overcome not having a 32-bit
22493         load and store.
22494         (rs6000_emit_move): Likewise.
22495         (rs6000_secondary_memory_needed_rtx): Likewise.
22496         (rs6000_alloc_sdmode_stack_slot): Likewise.
22497         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22498         via xxlxor, just like DFmode 0.0.
22500         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22501         define as 1 if we are running on a power7 or newer.
22502         (enum r6000_reg_class_enum): Add new constraints.
22504         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22505         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22506         with other moves by using conditional constraits (wg).  Use LFIWZX
22507         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22508         (movsd splitter): Likewise.
22509         (movsd_hardfloat): Likewise.
22510         (movsd_softfloat): Likewise.
22512         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22513         binary and decimal floating point moves.
22514         (fmove_ok): New attributes to combine binary and decimal floating
22515         point moves, and to combine power6x (mfpgpr) moves along normal
22516         floating moves.
22517         (real_value_to_target): Likewise.
22518         (f32_lr): Likewise.
22519         (f32_lm): Likewise.
22520         (f32_li): Likewise.
22521         (f32_sr): Likewise.
22522         (f32_sm): Likewise.
22523         (f32_si): Likewise.
22524         (movsf): Combine binary and decimal floating point moves.  Combine
22525         power6x (mfpgpr) moves with other moves by using conditional
22526         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22527         (mov<mode> for SFmode/SDmode); Likewise.
22528         (SFmode/SDmode splitters): Likewise.
22529         (movsf_hardfloat): Likewise.
22530         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22531         (movsf_softfloat): Likewise.
22532         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22534         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22535         wx and wz constraints.
22537         * config/rs6000/constraints.md (wg constraint): New constraint to
22538         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22540         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22541         constraint.
22543         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22544         -mdebug=reg, print wg, wl, wx, and wz constraints.
22545         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22546         Initialize the reload functions for 64-bit binary/decimal floating
22547         point types.
22548         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22549         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22550         create the buffer on the stack to overcome not having a 32-bit
22551         load and store.
22552         (rs6000_emit_move): Likewise.
22553         (rs6000_secondary_memory_needed_rtx): Likewise.
22554         (rs6000_alloc_sdmode_stack_slot): Likewise.
22555         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22556         via xxlxor, just like DFmode 0.0.
22558         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22559         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22560         with other moves by using conditional constraits (wg).  Use LFIWZX
22561         and STFIWX for loading SDmode on power7.
22562         (movdd splitters): Likewise.
22563         (movdd_hardfloat32): Likewise.
22564         (movdd_softfloat32): Likewise.
22565         (movdd_hardfloat64_mfpgpr): Likewise.
22566         (movdd_hardfloat64): Likewise.
22567         (movdd_softfloat64): Likewise.
22569         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22570         64-bit binary and decimal floating point moves.
22571         (FMOVE64X): Likewise.
22572         (movdf): Combine 64-bit binary and decimal floating point moves.
22573         Combine power6x (mfpgpr) moves with other moves by using
22574         conditional constraits (wg).
22575         (mov<mode> for DFmode/DDmode): Likewise.
22576         (DFmode/DDmode splitters): Likewise.
22577         (movdf_hardfloat32): Likewise.
22578         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22579         (movdf_softfloat32): Likewise.
22580         (movdf_hardfloat64_mfpgpr): Likewise.
22581         (movdf_hardfloat64): Likewise.
22582         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22583         (movdf_softfloat64): Likewise.
22584         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22585         (reload_<mode>_load): Move to later in the file so they aren't in
22586         the middle of the floating point move insns.
22587         (reload_<mode>_store): Likewise.
22589         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22590         constraint.
22592         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22593         constraint if -mdebug=reg.
22594         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22595         Enable using dd reload support if needed.
22597         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22598         binary and decimal floating point moves in rs6000.md.
22599         (movtd_internal): Likewise.
22601         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22602         decimal floating point moves.
22603         (movtf): Likewise.
22604         (movtf_internal): Likewise.
22605         (mov<mode>_internal, TDmode/TFmode): Likewise.
22606         (movtf_softfloat): Likewise.
22607         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22609         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22610         movdi_internal64, using wg constraint for move direct operations.
22611         (movdi_internal64): Likewise.
22613         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22614         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22615         the various virtual registers. Use GPR/FPR first/last values,
22616         instead of hard coding the register numbers.  Print which modes
22617         have reload functions registered.
22618         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22619         settings before/after setting cpu, target and subtarget settings.
22620         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22621         and for secondary reload failures in rs6000_secondary_reload_inner.
22622         (rs6000_secondary_reload_fail): Likewise.
22623         (rs6000_secondary_reload_inner): Likewise.
22625         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22626         macros for first/last GPR and FPR registers.
22627         (LAST_GPR_REGNO): Likewise.
22628         (FIRST_FPR_REGNO): Likewise.
22629         (LAST_FPR_REGNO): Likewise.
22631         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22632         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22633         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22634         (vcond<mode><mode>): Likewise.
22635         (vcondu<mode><mode>): Likewise.
22636         (vector_gtu<mode>): Likewise.
22637         (vector_gte<mode>): Likewise.
22638         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22639         to prevent the compiler from converting DImode operations to TImode.
22640         (ior<mode>3): Likewise.
22641         (and<mode>3): Likewise.
22642         (one_cmpl<mode>2): Likewise.
22643         (nor<mode>3): Likewise.
22644         (andc<mode>3): Likewise.
22646         * config/rs6000/constraints.md (wt constraint): New constraint
22647         that returns VSX_REGS if TImode is allowed in VSX registers.
22649         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22650         constant under VSX.
22652         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22653         similar to TImode, but it is restricted to being in the GPRs.
22655         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22656         TImode to occupy a single VSX register.
22658         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22659         -mvsx-timode for power7/power8.
22660         (power7 cpu): Likewise.
22661         (power8 cpu): Likewise.
22663         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22664         sure that TFmode/TDmode take up two registers if they are ever
22665         allowed in the upper VSX registers.
22666         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22667         registers.
22668         (rs6000_init_hard_regno_mode_ok): Likewise.
22669         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22670         constraint.  Print if LRA is turned on.
22671         (rs6000_option_override_internal): Give an error if -mvsx-timode
22672         and VSX is not enabled.
22673         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22674         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22675         to reg+offset addressing.  Use PTImode when checking offset
22676         addresses for validity.
22677         (reg_offset_addressing_ok_p): Likewise.
22678         (rs6000_legitimate_offset_address_p): Likewise.
22679         (rs6000_legitimize_address): Likewise.
22680         (rs6000_legitimize_reload_address): Likewise.
22681         (rs6000_legitimate_address_p): Likewise.
22682         (rs6000_eliminate_indexed_memrefs): Likewise.
22683         (rs6000_emit_move): Likewise.
22684         (rs6000_secondary_reload): Likewise.
22685         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22686         reloads to fpr registers to continue to use reg+offset addressing,
22687         but 64-bit reloads to altivec registers need reg+reg addressing.
22688         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22689         it.  Treat LO_SUM like a PLUS operation.
22690         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22691         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22692         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22693         registers to share a register with a smaller sized type, since VSX
22694         puts scalars in the upper 64-bits.
22695         (print_operand): Add support for PTImode.
22696         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22697         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22698         registers, but don't have arithmetic support.
22699         (rs6000_memory_move_cost): Add test for VSX.
22700         (rs6000_opt_masks): Add -mvsx-timode.
22702         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22703         for TImode.
22704         (VSs): Likewise.
22705         (VSr): Use wt constraint for TImode.
22706         (VSv): Drop TImode support.
22707         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22708         (vsx_movti_64bit): Likewise.
22709         (vsx_movti_32bit): Likewise.
22710         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22711         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22712         one '?' on the appropriate output constraint.  Do not allow TImode
22713         logical operations on 32-bit systems.
22714         (vsx_ior<mode>3): Likewise.
22715         (vsx_xor<mode>3): Likewise.
22716         (vsx_one_cmpl<mode>2): Likewise.
22717         (vsx_nor<mode>3): Likewise.
22718         (vsx_andc<mode>3): Likewise.
22719         (vsx_concat_<mode>): Likewise.
22720         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22722         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22723         OPTION_MASK_VSX_TIMODE.
22724         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22725         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22727         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22728         (TI2 iterator): New iterator for TImode, PTImode.
22729         (wd mode attribute): Add values for vector types.
22730         (movti_string): Replace TI move operations with operations for TImode
22731         and PTImode.  Add support for TImode being allowed in VSX registers.
22732         (mov<mode>_string, TImode/PTImode): Likewise.
22733         (movti_ppc64): Likewise.
22734         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22735         (TI mode splitters): Likewise.
22737         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22738         constraint.
22740 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22742         PR tree-optimization/56355
22743         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22744         Also handle integers with undefined overflow.
22746 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22747             Maciej W. Rozycki  <macro@codesourcery.com>
22748             Tom de Vries  <tom@codesourcery.com>
22749             Nathan Sidwell  <nathan@codesourcery.com>
22750             Iain Sandoe  <iain@codesourcery.com>
22751             Nathan Froyd  <froydnj@codesourcery.com>
22752             Chao-ying Fu  <fu@mips.com>
22754         * doc/extend.texi: (micromips, nomicromips, nocompression):
22755         Document new function attributes.
22756         * doc/invoke.texi (minterlink-compressed, mmicromips,
22757         m14k, m14ke, m14kec): Document new options.
22758         (minterlink-mips16): Update documentation.
22759         * doc/md.texi (ZC, ZD): Document new constraints.
22760         * configure.ac (gcc_cv_as_micromips): Check if linker
22761         supports the .set micromips directive.
22762         * configure: Regenerate.
22763         * config.in: Regenerate.
22764         * config/mips/mips-tables.opt: Regenerate.
22765         * config/mips/micromips.md: New file.
22766         * constraints.md (ZC, ZD): New constraints.
22767         * config/mips/predicates.md (movep_src_register): New predicate.
22768         (movep_src_operand): New predicate.
22769         (non_volatile_mem_operand): New predicate.
22770         * config/mips/mips.md (multimem): New type.
22771         (length): Differentiate between 17-bit and 18-bit branch offsets.
22772         (MOVEP1, MOVEP2): New mode iterator.
22773         (mov_<load>l): Use ZC constraint.
22774         (mov_<load>r): Likewise.
22775         (mov_<store>l): Likewise.
22776         (mov_<store>r): Likewise.
22777         (*branch_equality<mode>_inverted): Add microMIPS support.
22778         (*branch_equality<mode>): Likewise.
22779         (*jump_absolute): Likewise.
22780         (indirect_jump_<mode>): Likewise.
22781         (tablejump_<mode>): Likewise.
22782         (<optab>_internal): Likewise.
22783         (sibcall_internal): Likewise.
22784         (sibcall_value_internal): Likewise.
22785         (prefetch): Use constraint ZD.
22786         * config/mips/mips.opt (minterlink-compressed): New option.
22787         (minterlink-mips16): Now an alias for minterlink-compressed.
22788         (mmicromips): New option.
22789         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22790         (compare_and_swap_12): Likewise.
22791         (sync_add<mode>): Likewise.
22792         (sync_<optab>_12): Likewise.
22793         (sync_old_<optab>_12): Likewise.
22794         (sync_new_<optab>_12): Likewise.
22795         (sync_nand_12): Likewise.
22796         (sync_old_nand_12): Likewise.
22797         (sync_new_nand_12): Likewise.
22798         (sync_sub<mode>): Likewise.
22799         (sync_old_add<mode>): Likewise.
22800         (sync_old_sub<mode>): Likewise.
22801         (sync_new_add<mode>): Likewise.
22802         (sync_new_sub<mode>): Likewise.
22803         (sync_<optab><mode>): Likewise.
22804         (sync_old_<optab><mode>): Likewise.
22805         (sync_new_<optab><mode>): Likewise.
22806         (sync_nand<mode>): Likewise.
22807         (sync_old_nand<mode>): Likewise.
22808         (sync_new_nand<mode>): Likewise.
22809         (sync_lock_test_and_set<mode>): Likewise.
22810         (test_and_set_12): Likewise.
22811         (atomic_compare_and_swap<mode>): Likewise.
22812         (atomic_exchange<mode>_llsc): Likewise.
22813         (atomic_fetch_add<mode>_llsc): Likewise.
22814         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22815         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22816         (umips_save_restore_pattern_p): Likewise.
22817         (umips_load_store_pair_p): Likewise.
22818         (umips_output_load_store_pair): Likewise.
22819         (umips_movep_target_p): Likewise.
22820         (umips_12bit_offset_address_p): Likewise.
22821         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22822         (mips_base_mips16): Rename this...
22823         (mips_base_compression_flags): ...to this. Update all uses.
22824         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22825         (mips_mips16_decl_p): Delete.
22826         (mips_nomips16_decl_p): Delete.
22827         (mips_get_compress_on_flags): New function.
22828         (mips_get_compress_off_flags): New function.
22829         (mips_get_compress_mode): New function.
22830         (mips_get_compress_on_name): New function.
22831         (mips_get_compress_off_name): New function.
22832         (mips_insert_attributes): Support multiple compression types.
22833         (mips_merge_decl_attributes): Likewise.
22834         (umips_12bit_offset_address_p): New function.
22835         (mips_start_function_definition): Emit .set micromips directive.
22836         (mips_call_may_need_jalx_p): New function.
22837         (mips_function_ok_for_sibcall): Add microMIPS support.
22838         (mips_print_operand_punctuation): Support short delay slots and
22839         compact jumps.
22840         (umips_swm_mask, umips_swm_encoding): New.
22841         (umips_build_save_restore): New function.
22842         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22843         (was_mips16_p): Remove.
22844         (old_compression_mode): New.
22845         (mips_set_compression_mode): New function.
22846         (mips_set_current_function): Add microMIPS support.
22847         (mips_option_override): Likewise.
22848         (umips_save_restore_pattern_p): New function.
22849         (umips_output_save_restore): New function.
22850         (umips_load_store_pair_p_1): New function.
22851         (umips_load_store_pair_p): New function.
22852         (umips_output_load_store_pair_1): New function.
22853         (umips_output_load_store_pair): New function.
22854         (umips_movep_target_p) New function.
22855         (mips_prepare_pch_save): Add microMIPS support.
22856         * config/mips/mips.h (TARGET_COMPRESSION): New.
22857         (TARGET_CPU_CPP_BUILTINS): Update macro
22858         to use new compression flags and to support microMIPS.
22859         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22860         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22861         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22862         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22863         (ASM_SPEC): Support mmicromips and mno-micromips.
22864         (M16STORE_REG_P): New macro.
22865         (MIPS_CALL): Support TARGET_MICROMIPS.
22866         (MICROMIPS_J): New macro.
22867         (mips_base_mips16): Rename this...
22868         (mips_base_compression_flags): ...to this.
22869         (UMIPS_12BIT_OFFSET_P): New macro.
22870         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22871         (MULTILIB_DIRNAMES): Likewise.
22872 2013-03-20  Richard Biener  <rguenther@suse.de>
22874         PR tree-optimization/56661
22875         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22876         the result does not have to be distinct.
22878 2013-03-20  Richard Biener  <rguenther@suse.de>
22880         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22881         remap_gimple_op_r.
22883 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22884             Steven Bosscher  <steven@gcc.gnu.org>
22886         PR rtl-optimization/56605
22887         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22889 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22891         PR bootstrap/56656
22892         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22893         that require movd instead of movq.
22895 2013-03-20  Richard Biener  <rguenther@suse.de>
22897         * tree-ssa-structalias.c (struct variable_info): Add pointer
22898         to the first field of an aggregate with sub-vars.  Make
22899         this and the pointer to the next subfield its ID.
22900         (vi_next): New function.
22901         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22902         storedanything_id, integer_id): Increment by one.
22903         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22904         get_call_clobber_vi): Adjust.
22905         (solution_set_expand): Simplify and speedup.
22906         (solution_set_add): Inline into ...
22907         (set_union_with_increment): ... this.  Adjust accordingly.
22908         (do_sd_constraint): Likewise.
22909         (do_ds_constraint): Likewise.
22910         (do_complex_constraint): Simplify.
22911         (build_pred_graph): Adjust.
22912         (solve_graph): Likewise.  Simplify and speedup.
22913         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22914         get_constraint_for_component_ref, get_constraint_for_1,
22915         first_vi_for_offset, first_or_preceding_vi_for_offset,
22916         create_function_info_for, create_variable_info_for_1,
22917         create_variable_info_for, intra_create_variable_infos): Adjust.
22918         (init_base_vars): Push NULL for ID zero.
22919         (compute_points_to_sets): Adjust.
22921 2013-03-20  Richard Biener  <rguenther@suse.de>
22923         * cfgloop.c (verify_loop_structure): Streamline and avoid
22924         ICEing on corrupt loop tree.
22925         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22926         loop tree.
22928 2013-03-20  Richard Biener  <rguenther@suse.de>
22930         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22931         check whether an SSA update is needed.
22933 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22935         * config/mips/constraints.md (T): Rename to...
22936         (Yf): ...this.
22937         (U): Rename to...
22938         (Yd): ...this.
22939         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22940         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22942 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22944         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22945         (*subsi3_carryin_uxtw): Likewise.
22947 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22949         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
22950         (*rorsi3_insn_uxtw): Likewise.
22952 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22954         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
22955         (*extrsi5_insn_uxtw): Likewise.
22957 2013-03-19  Richard Biener  <rguenther@suse.de>
22959         PR tree-optimization/56273
22960         * passes.c (init_optimization_passes): Move second VRP after DOM.
22962 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
22964         * config/i386/i386.md (*movti_internal): Merge from
22965         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
22966         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
22967         nox64 isa attributes.
22969 2013-03-18  Richard Biener  <rguenther@suse.de>
22971         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
22972         (unite): Likewise.
22973         (merge_node_constraints): Likewise.
22974         (build_succ_graph): Likewise.
22975         (valid_graph_edge): Inline into single caller.
22976         (unify_nodes): Likewise.  Use bitmap_set_bit return value
22977         and cache varinfo.
22978         (scc_visit): Fix formatting and variable use.
22979         (do_sd_constraint): Use gcc_checking_assert.
22980         (do_ds_constraint): Likewise.
22981         (do_complex_constraint): Likewise.
22982         (condense_visit): Likewise.  Cleanup.
22983         (dump_pred_graph): New function.
22984         (perform_var_substitution): Dump the pred-graph before
22985         variable substitution.
22986         (find_equivalent_node): Use gcc_checking_assert.
22987         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
22989 2013-03-18  Richard Biener  <rguenther@suse.de>
22991         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
22992         Remove cond_expr_stmt_list argument and do not gimplify the
22993         built expression.
22994         (vect_loop_versioning): Adjust.
22995         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
22996         Cleanup to use less temporaries.
22997         (vect_create_data_ref_ptr): Cleanup.
22999 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23001         PR tree-optimization/56635
23002         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23003         require types_compatible_p types.
23005 2013-03-18  Nick Clifton  <nickc@redhat.com>
23007         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23008         spurious backslash.
23010         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23011         Add missing line to comment describing function.
23013 2013-03-18  Richard Biener  <rguenther@suse.de>
23015         PR tree-optimization/56210
23016         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23017         Handle string / character search functions.
23018         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23020 2013-03-18  Richard Biener  <rguenther@suse.de>
23022         PR middle-end/56483
23023         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23024         and implement properly.
23025         * gimple.h (gimple_cond_single_var_p): Remove.
23027 2013-03-18  Richard Biener  <rguenther@suse.de>
23029         * tree-data-ref.h (find_data_references_in_loop): Declare.
23030         * tree-data-ref.c (get_references_in_stmt): Use a stack
23031         vector pre-allocated in the callers.
23032         (find_data_references_in_stmt): Adjust.
23033         (graphite_find_data_references_in_stmt): Likewise.
23034         (create_rdg_vertices): Likewise.
23035         (find_data_references_in_loop): Export.
23036         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23037         Compute dependences here...
23038         (vect_analyze_data_refs): ...not here.  When we encounter
23039         a non-vectorizable data reference in basic-block vectorization
23040         truncate the data reference vector.  Do not bother to
23041         fixup data-dependence information for gather loads.
23042         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23043         of data references, as reported.
23045 2013-03-18  Richard Biener  <rguenther@suse.de>
23047         PR tree-optimization/3713
23048         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23049         has_constants and expr.
23050         (stmt_has_constants): Properly valueize SSA names when deciding
23051         whether the stmt has constants.
23053 2013-03-18  Richard Biener  <rguenther@suse.de>
23055         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23056         whole function when there is nothing to do.
23057         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23058         * tree-vectorizer.c (vectorize_loops): Update virtual and
23059         loop-closed SSA once.
23060         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23062 2013-03-18  Richard Biener  <rguenther@suse.de>
23064         PR middle-end/56113
23065         * domwalk.c (bb_postorder): New global static.
23066         (cmp_bb_postorder): New function.
23067         (walk_dominator_tree): Replace scheme imposing an order for
23068         visiting dominator sons by one sorting them at the time they
23069         are pushed on the stack.
23071 2013-03-18  Richard Biener  <rguenther@suse.de>
23073         PR tree-optimization/39326
23074         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23075         (struct mem_ref): Replace mem member with ao_ref typed member.
23076         (MEM_ANALYZABLE): Adjust.
23077         (memref_eq): Likewise.
23078         (mem_ref_alloc): Likewise.
23079         (gather_mem_refs_stmt): Likewise.
23080         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23081         (execute_sm_if_changed_flag_set): Adjust.
23082         (execute_sm): Likewise.
23083         (ref_always_accessed_p): Likewise.
23084         (refs_independent_p): Likewise.
23085         (can_sm_ref_p): Likewise.
23087 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23089         PR c/56566
23090         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23091         return 1 even for !unsignedp.
23093 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23095         * config/i386/i386.md (isa): Add x64 and nox64.
23096         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23097         (*pushtf): Enable *roF alternative for x64 isa only.
23098         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23099         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23100         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23101         (*movtf_internal): Merge from *movtf_internal_rex64 and
23102         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23103         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23104         nox64 isa attributes.
23105         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23106         nox64 isa attributes.
23107         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23109 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23111         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23113 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23115         PR target/56640
23116         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23117         class names.  Remove trailing comma after "ALL_REGS".
23119 2013-03-16  Jan Hubicka  <jh@suse.cz>
23121         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23122         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23123         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23124         of cgraph_get_create_node.
23125         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23127 2013-03-16  Jason Merrill  <jason@redhat.com>
23129         PR debug/49090
23130         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23131         with DW_AT_default_value.
23133 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23135         * BASE-VER: Set to 4.9.0.
23137 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23139         PR target/56619
23140         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23141         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23142         Document _x* TSX intrinsics.
23144 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23145             David Holsgrove  <david.holsgrove@xilinx.com>
23147         * configure.ac: Add MicroBlaze TLS support detection.
23148         * configure: Regenerate.
23149         * config/microblaze/microblaze-protos.h
23150         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23151         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23152         * config/microblaze/microblaze.c (microblaze_address_type): Add
23153         ADDRESS_TLS and tls_reloc address types.
23154         (microblaze_address_info): Add tls_reloc.
23155         (TARGET_HAVE_TLS): Define.
23156         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23157          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23158          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23159         load_tls_operand,  microblaze_call_tls_get_addr,
23160         microblaze_legitimize_tls_address): New functions.
23161         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23162         (get_base_reg): Use microblaze_tls_symbol_p.
23163         (microblaze_classify_address): Handle TLS.
23164         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23165         label_mentioned_p and microblaze_tls_referenced_p.
23166         (microblaze_legitimize_address): Handle TLS.
23167         (microblaze_address_insns): Handle ADDRESS_TLS.
23168         (pic_address_needs_scratch): Handle TLS.
23169         (print_operand_address): Handle TLS.
23170         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23171         (microblaze_expand_move): Handle TLS.
23172         (microblaze_legitimate_constant_p): Check
23173         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23174         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23175         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23176         (PIC_OFFSET_TABLE_REGNUM): Set.
23177         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23178         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23179         (addsi3, movsi_internal2, movdf_internal): Update constraints
23180         * config/microblaze/predicates.md (arith_plus_operand): Define
23181         (move_operand): Redefine as move_src_operand,
23182         check microblaze_tls_referenced_p.
23184 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23186         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23187         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23189 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23191         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23192         CC mode for AND.
23194 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23196         PR tree-optimization/53265
23197         * common.opt (Waggressive-loop-optimizations): New option.
23198         * tree-ssa-loop-niter.c: Include tree-pass.h.
23199         (do_warn_aggressive_loop_optimizations): New function.
23200         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23201         if number_of_latch_executions returned constant.
23202         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23203         early.  If number_of_latch_executions returned constant, set
23204         nb_iterations_upper_bound back to it.
23205         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23206         field.
23207         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23208         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23210         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23211         (MULTILIB_OSDIRNAMES): Set.
23212         * genmultilib: If defaultosdirname doesn't start with :: , set
23213         defaultosdirname2 instead, clear it and emit two . multilib_raw
23214         entries instead of just one.
23216 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23218         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23219         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23220         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23221         (SUBTARGET_OVERRIDE_OPTIONS): New.
23223 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23225         PR target/49880
23226         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23227         (musermode): Convert to Var(TARGET_USERMODE).
23228         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23229         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23230         * config/sh/sh.c (sh_option_override): Use
23231         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23232         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23233         condition.
23234         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23235         TARGET_SH4.
23236         (udivsi3_i4_single, divsi3_i4_single): Use
23237         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23239 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23241         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23242         default setting.
23244 2013-03-13  Richard Biener  <rguenther@suse.de>
23246         PR tree-optimization/56608
23247         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23248         calls when vectorizing basic-blocks.
23250 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23252         PR plugins/45078
23253         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23254         tm_file.
23256 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23258         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23260 2013-03-11  Jan Hubicka  <jh@suse.cz>
23262         PR lto/56557
23263         * lto-streamer-out.c (output_symbol_p): Skip references from
23264         constructors of external variables.
23266 2013-03-11  Jan Hubicka  <jh@suse.cz>
23268         PR middle-end/56571
23269         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23270         from pseudos.
23271         * emit-rtl.c (verify_rtx_sharing): Likewise.
23272         (copy_insn_1): Likewise.
23273         * rtl.c (copy_rtx): Likewise.
23275 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23277         PR target/56591
23278         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23279         output_operand_lossage message.
23281 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23283         PR target/56470
23284         * arm.c (shift_op): Validate RTL pattern on the fly.
23285         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23286         the RTL.
23288 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23290         PR target/56347
23291         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23292         new call patterns that clobber %fr12.
23293         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23294         split and postreload patterns.
23295         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23296         registers %fr12 and %fr12R as call used.
23298 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23300         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23301         (canon_address, record_store, replace_read, check_mem_read_rtx,
23302         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23303         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23304         rest_of_handle_dse): Likewise.
23306 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23308         PR middle-end/56524
23309         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23310         Add base_optabs.
23311         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23312         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23313         (save_optabs_if_changed): Replace with...
23314         (init_tree_optimization_optabs): ...this.
23315         * optabs.c (save_optabs_if_changed): Rename to...
23316         (init_tree_optimization_optabs): ...this.  Take the optimization node
23317         as argument.  Do nothing if the base optabs are already correct.
23318         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23319         to recompute optabs.
23320         * function.h (function): Remove optabs field.
23321         * function.c (invoke_set_current_function_hook): Call
23322         init_tree_optimization_optabs.  Use the result to initialize
23323         this_fn_optabs.
23325 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23327         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23328         if GTMA_HAS_NO_INSTRUMENTATION.
23329         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23330         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23331         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23332         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23333         GTMA_HAS_NO_INSTRUMENTATION.
23335 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23337         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23338         libasan_preinit.o.
23340 2013-03-08  Marek Polacek  <polacek@redhat.com>
23341             Jakub Jelinek  <jakub@redhat.com>
23343         PR tree-optimization/56478
23344         * predict.c (is_comparison_with_loop_invariant_p): Change the
23345         type of loop_step to tree.
23346         (predict_loops): Adjust.
23347         (predict_iv_comparison): Perform the computations on double_ints.
23349 2013-03-08  Richard Biener  <rguenther@suse.de>
23351         PR tree-optimization/56570
23352         * tree-cfg.c (verify_expr_location_1): Verify locations for
23353         DECL_DEBUG_EXPR.
23354         * tree-sra.c (create_access_replacement): Strip locations
23355         from DECL_DEBUG_EXPRs.
23357 2013-03-08  Richard Biener  <rguenther@suse.de>
23359         * tree-inline.c (expand_call_inline): Do not associate
23360         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23361         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23363 2013-03-08  Richard Biener  <rguenther@suse.de>
23365         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23366         or block changes with -Og.  Fix for location / block encoding
23367         changes and PHI arguments with locations.
23369 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23371         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23372         for all counters.
23373         (struct output_info): Likewise.
23374         (register_overhead): Remove bad gcc_assert.
23375         (bitmap_find_bit): If there is only a single bitmap element, do not
23376         count a miss as a search.
23377         (print_statistics): Update for counter type changes.
23378         (dump_bitmap_statistics): Likewise.  Print headers such that they
23379         are properly lined up with the printed counters.
23381 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23383         PR tree-optimization/56559
23384         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23385         check that it has only a single use.
23387 2013-03-07  Richard Biener  <rguenther@suse.de>
23389         * doc/invoke.texi (fwhole-program): Discourage use in combination
23390         with -flto.
23392 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23394         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23396         PR tree-optimization/56539
23397         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23398         instead of GSI_CONTINUE_LINKING as last argument to
23399         force_gimple_operand_gsi.  Adjust function comment.
23401         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23402         aarch64-cores.def.
23404         PR middle-end/56548
23405         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23406         promoted mode, convert the result back to the original mode.
23408 2013-03-06  Richard Biener  <rguenther@suse.de>
23410         PR middle-end/56294
23411         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23412         (insert_updated_phi_nodes_compare_uids): New function.
23413         (update_ssa): Sort symbols_to_rename after UID before
23414         traversing it to insert PHI nodes.
23416 2013-03-06  Richard Biener  <rguenther@suse.de>
23418         PR middle-end/50494
23419         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23420         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23422         Revert
23423         2013-02-13  Richard Biener  <rguenther@suse.de>
23425         PR lto/50494
23426         * varasm.c (output_constant_def_1): Get the decl representing
23427         the constant as argument.
23428         (output_constant_def): Wrap output_constant_def_1.
23429         (make_decl_rtl): Use output_constant_def_1 with the decl
23430         representing the constant.
23431         (build_constant_desc): Optionally re-use a decl already
23432         representing the constant.
23433         (tree_output_constant_def): Adjust.
23435 2013-03-06  Joey Ye  <joey.ye@arm.com>
23437         PR lto/50293
23438         * gcc.c (convert_white_space): New function.
23439         (main): Handles white space in function name.
23441 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23443         PR target/56529
23444         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23445         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23446         to SH_DIV_CALL_TABLE for TARGET_SH2.
23447         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23448         list.
23449         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23450         call-table options.
23452 2013-03-05  Sterling Augustine  <saugustine@google.com>
23453             Cary Coutant  <ccoutant@google.com>
23455         PR debug/55364
23456         * dwarf2out.c (resolve_addr): Don't call
23457         remove_loc_list_addr_table_entries a second time for the same
23458         expression.
23460 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23462         PR debug/56510
23463         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23464         (avoid_complex_debug_insns): New function.
23465         (expand_debug_locations): Call it.
23467         PR rtl-optimization/56484
23468         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23469         lifetimes of hard registers on small register class machines.
23471 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23473         * config/microblaze/microblaze-protos.h: Rename
23474         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23475         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23476         fast_interrupt.
23477         (microblaze_fast_interrupt_function_p): New function.
23478         (microblaze_is_interrupt_handler): Rename to
23479         microblaze_is_interrupt_variant and add fast_interrupt check.
23480         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23481         (save_restore_insns): Likewise.
23482         (compute_frame_size): Likewise.
23483         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23484         (microblaze_globalize_label): Likewise.
23485         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23486         * config/microblaze/microblaze.md: Use wrapper
23487         microblaze_is_interrupt_variant.
23489 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23491         * sdbout.c (sdbout_one_type): Switch to current function's section
23492         supporting cold/hot.
23494 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23496         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23497         -mxl-reorder.
23499 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23501         PR middle-end/56461
23502         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23503         if VALGRIND_GET_VBITS is defined, temporarily make object
23504         memory all defined, and restore previous valgrind addressability
23505         and definability afterwards.  Free this_object at the end.
23507         PR middle-end/56461
23508         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23509         right before calling lra_create_live_ranges, also call it
23510         when clearing live_p.  Only call lra_clear_live_ranges
23511         at the end if live_p.
23513         PR middle-end/56461
23514         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23516 2013-03-05  Richard Biener  <rguenther@suse.de>
23518         PR tree-optimization/56521
23519         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23520         value-id.
23522 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23524         PR c++/55135
23525         * except.h (remove_unreachable_eh_regions): New prototype.
23526         * except.c (remove_eh_handler_splicer): New function, split out
23527         of remove_eh_handler.
23528         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23529         warning about running it on many EH regions one at a time.
23530         (remove_unreachable_eh_regions_worker): New function, walk the
23531         EH tree in depth-first order and remove non-marked regions.
23532         (remove_unreachable_eh_regions): New function.
23533         * tree-eh.c (mark_reachable_handlers): New function, split out
23534         from remove_unreachable_handlers.
23535         (remove_unreachable_handlers): Use mark_reachable_handlers and
23536         remove_unreachable_eh_regions.
23537         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23538         and remove_unreachable_eh_regions.
23540 2013-03-05  Richard Biener  <rguenther@suse.de>
23542         PR middle-end/56525
23543         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23544         not freeing them until the end.
23546 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23548         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23550 2013-03-05  Richard Biener  <rguenther@suse.de>
23552         PR tree-optimization/56270
23553         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23554         of loads after scheduling an SLP instance.
23556 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23558         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23559         tic6x.exp.
23560         (check_gcc_parallelize): Run guality.exp as a separate job from
23561         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23562         struct-layout-1.exp with stackalign.exp.
23564         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23566         PR middle-end/56461
23567         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23568         load_index sbitmap even if some bit in it isn't set.
23570         PR middle-end/56461
23571         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23572         (discover_iteration_bound_by_body_walk): Change queues to
23573         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23574         spelling in comment.  Call safe_push on queues[bound_index] directly.
23575         Release queues[queue_index] in every iteration unconditionally.
23576         Release bounds vector.
23578         PR middle-end/56461
23579         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23580         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23581         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23582         inner_phis vector.
23584 2013-03-05  Richard Biener  <rguenther@suse.de>
23586         PR lto/56515
23587         * tree-inline.c (remap_blocks_to_null): New function.
23588         (expand_call_inline): When expanding a call stmt without
23589         an associated block inline remap all callee blocks to NULL.
23591 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23593         PR rtl-optimization/56494
23594         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23595         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23596         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23598         PR middle-end/56461
23599         * sel-sched-ir.c (free_sched_pools): Release
23600         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23601         if succs_info_pool.max_top isn't -1.
23603         PR bootstrap/56509
23604         * opts.c (opts_obstack, opts_concat): Moved to...
23605         * opts-common.c (opts_obstack, opts_concat): ... here.
23607 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23609         PR middle-end/56461
23610         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23612 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23614         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23615         all appropriate places.
23617 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23619         PR tree-optimization/56424
23620         * ipa-split.c (split_function): Do not set the RSO flag if result is
23621         not by reference and its type is a register type.
23623 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23625         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23626         (microblaze_legitimate_pic_operand): Likewise
23627         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23628         new function microblaze_legitimate_pic_operand
23629         * config/microblaze/microblaze-protos.h
23630         (microblaze_legitimate_pic_operand): Declare.
23632 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23634         * config/microblaze/predicates.md (call_insn_simple_operand):
23635         New predicate for supported rtx code types.
23636         * config/microblaze/microblaze.md (call_internal1): Use
23637         call_insn_simple_operand predicate.
23639 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23641         PR middle-end/56461
23642         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23643         partitions.ordered_remove.
23645         PR middle-end/56461
23646         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23647         vec_oprnds0.create (1) for modifier == NONE.
23649         PR middle-end/56461
23650         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23651         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23652         vec_oprnds1 right before pushing anything to it for
23653         scalar_shift_arg.
23655         PR middle-end/56461
23656         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23657         set nbbs to 0 instead of having separate code path.
23658         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23659         instead of false as last argument if returning NULL.
23661 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23663         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23664         the attribute is now called "target" instead of "option".
23665         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23666         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23667         attribute/pragma name for TARGET_OPTION_VALID_P and
23668         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23669         * doc/tm.texi: Regenerated.
23671 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23673         * config/microblaze/microblaze.c:
23674         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23675         * config/microblaze/microblaze.h: Add -mxl-reorder to
23676         DRIVER_SELF_SPECS.
23677         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23678         instructions emitted if TARGET_REORDER.
23679         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23680         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23681         separately.
23683 2013-03-01  Xinliang David Li  <davidxl@google.com>
23685         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23686         walk length.
23688 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23690         PR middle-end/56461
23691         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23692         vector even when returning true.  Fix up function comment formatting.
23694         PR middle-end/56461
23695         * ira-build.c (ira_loop_nodes_count): New variable.
23696         (create_loop_tree_nodes): Initialize it.
23697         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23699         PR middle-end/56461
23700         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23701         method on dr_chain and result_chain.
23702         * tree-vect-stmts.c (vectorizable_store): Only call
23703         result_chain.create if j == 0.
23705         PR middle-end/56461
23706         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23707         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23708         before overwriting it.
23710 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23712         * doc/extended.texi (C Extensions): Change order in @menu
23713         to match @node.
23714         (Other MIPS Built-in Functions): Move last MIPS entry before
23715         "picoChip Built-in Functions".
23716         (SH Built-in Functions): Move after RX Built-in Functions.
23717         * doc/gcc.texi (Introduction): Change order in @menu
23718         to match @node.
23719         * doc/md.texi (Constraints): Ditto.
23720         * gty.texi (Type Information): Ditto.
23721         (User-provided marking routines for template types): Make
23722         subsection.
23723         * doc/invoke.texi (AArch64 Options): Move before
23724         "Adapteva Epiphany Options".
23726 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23727             Jakub Jelinek  <jakub@redhat.com>
23729         PR sanitizer/56454
23730         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23731         no_address_safety_analysis attribute.
23732         * doc/extend.texi (no_address_safety_attribute): Rename to
23733         no_sanitize_address attribute, mention no_address_safety_analysis
23734         attribute as deprecated alias.
23736 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23738         PR middle-end/56461
23739         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23740         type to vec<vec<tree> > *.
23741         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23742         to be vec<tree> instead of vec<tree> *, set vec_defs
23743         to vNULL and call vec_defs.create (number_of_vects), adjust other
23744         uses of vec_defs.
23745         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23746         vectorizable_condition): Adjust vect_get_slp_defs callers.
23748 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23750         * config/aarch64/aarch64.c
23751         (aarch64_float_const_representable): Remove unused variable.
23753 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23755         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23757 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23759         * config/aarch64/aarch64-builtins.c
23760         (aarch64_init_simd_builtins): Make static.
23762 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23764         * config/aarch64/aarch64.c
23765         (aarch64_simd_make_constant): Make static.
23767 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23769         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23770         with no initialization to the RHS of debug statements.
23772 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23774         PR tree-optimization/56294
23775         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23776         Adjust dumping.
23777         (get_access_replacement): Do not call create_access_replacement.
23778         Assert a replacement exists.
23779         (get_repl_default_def_ssa_name): Create the replacement declaration
23780         itself.
23782 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23784         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23785         final_end_function.
23787 2013-02-28  Marek Polacek  <polacek@redhat.com>
23789         PR rtl-optimization/56466
23790         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23791         if we're changing a loop.
23792         (peel_loops_completely): Likewise.
23794 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23796         PR c++/55813
23797         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23799 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23801         PR target/56445
23802         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23803         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23804         INTX_FTYPE_FX, FX_FTYPE_INTX.
23805         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23807 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23809         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23810         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23811         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23812         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23813         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23814         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23815         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23816         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23817         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23818         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23819         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23820         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23821         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23822         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23823         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23824         (avrxmega6): Increase max flash segments from 5 to 6.
23825         * config/avr/t-multilib: Regenerate.
23826         * config/avr/avr-tables.opt: Regenerate.
23827         * doc/avr-mmcu.texi: Regenerate.
23829 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23831         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23832         (avr_device_to_arch): Rename to avr_device_to_ld.
23833         (avr_device_to_as): New prototype.
23834         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23835         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23836         * config/avr/driver-avr.c (avr_device_to_as): New.
23837         (avr_device_to_arch): Rename to avr_device_to_ld.
23839 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23841         PR middle-end/56461
23842         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23843         method on dr_chain and result_chain.
23845         PR middle-end/56461
23846         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23847         pointer_set_destroy on not_executed_last_iteration.
23849         PR middle-end/56461
23850         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23852         PR middle-end/56461
23853         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23854         FOR_EACH_DEFINED_FUNCTION when freeing state.
23856         PR middle-end/56461
23857         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23858         pool_free.
23859         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23860         overwriting it.
23862         PR middle-end/56461
23863         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23864         known_aggs[i].items and release known_aggs vector.
23866         PR middle-end/56461
23867         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23869 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23871         * config/microblaze/microblaze.c (microblaze_emit_compare):
23872         Use xor for EQ/NE comparisions.
23873         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23874         (cbranchsf4): Adjust operator to comparison_operator.
23876 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23878         PR middle-end/56461
23879         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23880         vector.
23881         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23882         vec_safe_push, always update *slot.
23883         (redirect_edge_var_map_clear): Use vec_free.
23884         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23885         (free_var_map_entry): Use vec_free.
23886         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23887         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23889 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23891         PR middle-end/45472
23892         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23893         when the may_trap_p bit of the exprs being merged differs.
23894         Reorder tests for speculativeness in the logical and operator.
23896 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23898         * incpath.c (add_standard_paths): Use reconcat instead of concat
23899         where appropriate and avoid leaking memory.
23901         * opts.h: Include obstack.h.
23902         (opts_concat): New prototype.
23903         (opts_obstack): New declaration.
23904         * opts.c (opts_concat): New function.
23905         (opts_obstack): New variable.
23906         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23907         (finish_options): Use opts_concat instead of concat
23908         and XOBNEWVEC instead of XNEWVEC.
23909         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23910         generate_option): Likewise.
23911         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23912         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23914         PR target/56455
23915         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23916         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23918 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23920         PR middle-end/56461
23921         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23923 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23925         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23926         (arm_block_move_unaligned_straight): Likewise.
23927         (arm_adjust_block_mem): Likewise.
23929 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23931         PR target/48901
23932         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23933         temp, cond and label.
23934         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23936         PR target/52500
23937         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23938         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23940         PR target/52501
23941         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23942         of prologue/epilogue functions.
23944         PR target/52550
23945         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23946         Remove unused variable cfa_offset.
23947         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
23949         PR target/54639
23950         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
23951         type promotion to unsigned.
23953         PR target/54640
23954         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
23955         for HOST_WIDE_INT of 32 bit / same size as int.
23956         (arm_block_move_unaligned_straight): Likewise.
23957         (arm_adjust_block_mem): Likewise.
23959         PR target/54662
23960         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
23961         ALL_CFLAGS.
23963 2013-02-26  Marek Polacek  <polacek@redhat.com>
23965         PR tree-optimization/56426
23966         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
23968 2013-02-26  Richard Biener  <rguenther@suse.de>
23970         PR target/56444
23971         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
23972         unused variable loops.
23974 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23976         PR tree-optimization/56448
23977         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
23978         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
23979         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
23980         later operands of the references, or even first operand for
23981         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
23983         PR tree-optimization/56443
23984         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
23985         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
23986         to type_for_mode langhook.
23988 2013-02-25  Matt Turner  <mattst88@gmail.com>
23990         * doc/invoke.texi: Document r4700.
23992 2013-02-25  Richard Biener  <rguenther@suse.de>
23994         PR tree-optimization/56175
23995         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
23996         split out from ...
23997         (simplify_bitwise_binary): ... here.  Also guard the conversion
23998         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24000 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24002         Revert:
24003         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24004             Maciej W. Rozycki  <macro@codesourcery.com>
24005             Tom de Vries  <tom@codesourcery.com>
24006             Nathan Sidwell  <nathan@codesourcery.com>
24007             Iain Sandoe  <iain@codesourcery.com>
24008             Nathan Froyd  <froydnj@codesourcery.com>
24009             Chao-ying Fu  <fu@mips.com>
24011         * doc/extend.texi: (micromips, nomicromips, nocompression):
24012         Document new function attributes.
24013         * doc/invoke.texi (minterlink-compressed, mmicromips,
24014         m14k, m14ke, m14kec): Document new options.
24015         (minterlink-mips16): Update documentation.
24016         * doc/md.texi (ZC, ZD): Document new constraints.
24017         * configure.ac (gcc_cv_as_micromips): Check if linker
24018         supports the .set micromips directive.
24019         * configure: Regenerate.
24020         * config.in: Regenerate.
24021         * config/mips/mips-tables.opt: Regenerate.
24022         * config/mips/micromips.md: New file.
24023         * constraints.md (ZC, AD): New constraints.
24024         * config/mips/predicates.md (movep_src_register): New predicate.
24025         (movep_src_operand): New predicate.
24026         (non_volatile_mem_operand): New predicate.
24027         * config/mips/mips.md (multimem): New type.
24028         (length): Differentiate between 17-bit and 18-bit branch offsets.
24029         (MOVEP1, MOVEP2): New mode iterator.
24030         (mov_<load>l): Use ZC constraint.
24031         (mov_<load>r): Likewise.
24032         (mov_<store>l): Likewise.
24033         (mov_<store>r): Likewise.
24034         (*branch_equality<mode>_inverted): Add microMIPS support.
24035         (*branch_equality<mode>): Likewise.
24036         (*jump_absolute): Likewise.
24037         (indirect_jump_<mode>): Likewise.
24038         (tablejump_<mode>): Likewise.
24039         (<optab>_internal): Likewise.
24040         (sibcall_internal): Likewise.
24041         (sibcall_value_internal): Likewise.
24042         (prefetch): Use constraint ZD.
24043         * config/mips/mips.opt (minterlink-compressed): New option.
24044         (minterlink-mips16): Now an alias for minterlink-compressed.
24045         (mmicromips): New option.
24046         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24047         (compare_and_swap_12): Likewise.
24048         (sync_add<mode>): Likewise.
24049         (sync_<optab>_12): Likewise.
24050         (sync_old_<optab>_12): Likewise.
24051         (sync_new_<optab>_12): Likewise.
24052         (sync_nand_12): Likewise.
24053         (sync_old_nand_12): Likewise.
24054         (sync_new_nand_12): Likewise.
24055         (sync_sub<mode>): Likewise.
24056         (sync_old_add<mode>): Likewise.
24057         (sync_old_sub<mode>): Likewise.
24058         (sync_new_add<mode>): Likewise.
24059         (sync_new_sub<mode>): Likewise.
24060         (sync_<optab><mode>): Likewise.
24061         (sync_old_<optab><mode>): Likewise.
24062         (sync_new_<optab><mode>): Likewise.
24063         (sync_nand<mode>): Likewise.
24064         (sync_old_nand<mode>): Likewise.
24065         (sync_new_nand<mode>): Likewise.
24066         (sync_lock_test_and_set<mode>): Likewise.
24067         (test_and_set_12): Likewise.
24068         (atomic_compare_and_swap<mode>): Likewise.
24069         (atomic_exchange<mode>_llsc): Likewise.
24070         (atomic_fetch_add<mode>_llsc): Likewise.
24071         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24072         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24073         (umips_save_restore_pattern_p): Likewise.
24074         (umips_load_store_pair_p): Likewise.
24075         (umips_output_load_store_pair): Likewise.
24076         (umips_movep_target_p): Likewise.
24077         (umips_12bit_offset_address_p): Likewise.
24078         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24079         (mips_base_mips16): Rename this...
24080         (mips_base_compression_flags): ...to this. Update all uses.
24081         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24082         (mips_mips16_decl_p): Delete.
24083         (mips_nomips16_decl_p): Delete.
24084         (mips_get_compress_on_flags): New function.
24085         (mips_get_compress_off_flags): New function.
24086         (mips_get_compress_mode): New function.
24087         (mips_get_compress_on_name): New function.
24088         (mips_get_compress_off_name): New function.
24089         (mips_insert_attributes): Support multiple compression types.
24090         (mips_merge_decl_attributes): Likewise.
24091         (umips_12bit_offset_address_p): New function.
24092         (mips_start_function_definition): Emit .set micromips directive.
24093         (mips_call_may_need_jalx_p): New function.
24094         (mips_function_ok_for_sibcall): Add microMIPS support.
24095         (mips_print_operand_punctuation): Support short delay slots and
24096         compact jumps.
24097         (umips_swm_mask, umips_swm_encoding): New.
24098         (umips_build_save_restore): New function.
24099         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24100         (was_mips16_p): Remove.
24101         (old_compression_mode): New.
24102         (mips_set_compression_mode): New function.
24103         (mips_set_current_function): Add microMIPS support.
24104         (mips_option_override): Likewise.
24105         (umips_save_restore_pattern_p): New function.
24106         (umips_output_save_restore): New function.
24107         (umips_load_store_pair_p_1): New function.
24108         (umips_load_store_pair_p): New function.
24109         (umips_output_load_store_pair_1): New function.
24110         (umips_output_load_store_pair): New function.
24111         (umips_movep_target_p) New function.
24112         (mips_prepare_pch_save): Add microMIPS support.
24113         * config/mips/mips.h (TARGET_COMPRESSION): New.
24114         (TARGET_CPU_CPP_BUILTINS): Update macro
24115         to use new compression flags and to support microMIPS.
24116         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24117         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24118         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24119         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24120         (ASM_SPEC): Support mmicromips and mno-micromips.
24121         (M16STORE_REG_P): New macro.
24122         (MIPS_CALL): Support TARGET_MICROMIPS.
24123         (MICROMIPS_J): New macro.
24124         (mips_base_mips16): Rename this...
24125         (mips_base_compression_flags): ...to this.
24126         (UMIPS_12BIT_OFFSET_P): New macro.
24127         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24128         (MULTILIB_DIRNAMES): Likewise.
24130 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24132         PR rtl-optimization/56131
24133         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24134         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24135         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24137 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24139         * doc/invoke.texi (-fsanitize=): Move from optimization
24140         to debugging options.
24142 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24144         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24146 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24147             Alexander Monakov  <amonakov@ispras.ru>
24149         PR middle-end/56077
24150         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24151         flush pending lists also on non-jumps.  Adjust comment.
24153 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24154             Maciej W. Rozycki  <macro@codesourcery.com>
24155             Tom de Vries  <tom@codesourcery.com>
24156             Nathan Sidwell  <nathan@codesourcery.com>
24157             Iain Sandoe  <iain@codesourcery.com>
24158             Nathan Froyd  <froydnj@codesourcery.com>
24159             Chao-ying Fu  <fu@mips.com>
24161         * doc/extend.texi: (micromips, nomicromips, nocompression):
24162         Document new function attributes.
24163         * doc/invoke.texi (minterlink-compressed, mmicromips,
24164         m14k, m14ke, m14kec): Document new options.
24165         (minterlink-mips16): Update documentation.
24166         * doc/md.texi (ZC, ZD): Document new constraints.
24167         * configure.ac (gcc_cv_as_micromips): Check if linker
24168         supports the .set micromips directive.
24169         * configure: Regenerate.
24170         * config.in: Regenerate.
24171         * config/mips/mips-tables.opt: Regenerate.
24172         * config/mips/micromips.md: New file.
24173         * constraints.md (ZC, AD): New constraints.
24174         * config/mips/predicates.md (movep_src_register): New predicate.
24175         (movep_src_operand): New predicate.
24176         (non_volatile_mem_operand): New predicate.
24177         * config/mips/mips.md (multimem): New type.
24178         (length): Differentiate between 17-bit and 18-bit branch offsets.
24179         (MOVEP1, MOVEP2): New mode iterator.
24180         (mov_<load>l): Use ZC constraint.
24181         (mov_<load>r): Likewise.
24182         (mov_<store>l): Likewise.
24183         (mov_<store>r): Likewise.
24184         (*branch_equality<mode>_inverted): Add microMIPS support.
24185         (*branch_equality<mode>): Likewise.
24186         (*jump_absolute): Likewise.
24187         (indirect_jump_<mode>): Likewise.
24188         (tablejump_<mode>): Likewise.
24189         (<optab>_internal): Likewise.
24190         (sibcall_internal): Likewise.
24191         (sibcall_value_internal): Likewise.
24192         (prefetch): Use constraint ZD.
24193         * config/mips/mips.opt (minterlink-compressed): New option.
24194         (minterlink-mips16): Now an alias for minterlink-compressed.
24195         (mmicromips): New option.
24196         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24197         (compare_and_swap_12): Likewise.
24198         (sync_add<mode>): Likewise.
24199         (sync_<optab>_12): Likewise.
24200         (sync_old_<optab>_12): Likewise.
24201         (sync_new_<optab>_12): Likewise.
24202         (sync_nand_12): Likewise.
24203         (sync_old_nand_12): Likewise.
24204         (sync_new_nand_12): Likewise.
24205         (sync_sub<mode>): Likewise.
24206         (sync_old_add<mode>): Likewise.
24207         (sync_old_sub<mode>): Likewise.
24208         (sync_new_add<mode>): Likewise.
24209         (sync_new_sub<mode>): Likewise.
24210         (sync_<optab><mode>): Likewise.
24211         (sync_old_<optab><mode>): Likewise.
24212         (sync_new_<optab><mode>): Likewise.
24213         (sync_nand<mode>): Likewise.
24214         (sync_old_nand<mode>): Likewise.
24215         (sync_new_nand<mode>): Likewise.
24216         (sync_lock_test_and_set<mode>): Likewise.
24217         (test_and_set_12): Likewise.
24218         (atomic_compare_and_swap<mode>): Likewise.
24219         (atomic_exchange<mode>_llsc): Likewise.
24220         (atomic_fetch_add<mode>_llsc): Likewise.
24221         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24222         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24223         (umips_save_restore_pattern_p): Likewise.
24224         (umips_load_store_pair_p): Likewise.
24225         (umips_output_load_store_pair): Likewise.
24226         (umips_movep_target_p): Likewise.
24227         (umips_12bit_offset_address_p): Likewise.
24228         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24229         (mips_base_mips16): Rename this...
24230         (mips_base_compression_flags): ...to this. Update all uses.
24231         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24232         (mips_mips16_decl_p): Delete.
24233         (mips_nomips16_decl_p): Delete.
24234         (mips_get_compress_on_flags): New function.
24235         (mips_get_compress_off_flags): New function.
24236         (mips_get_compress_mode): New function.
24237         (mips_get_compress_on_name): New function.
24238         (mips_get_compress_off_name): New function.
24239         (mips_insert_attributes): Support multiple compression types.
24240         (mips_merge_decl_attributes): Likewise.
24241         (umips_12bit_offset_address_p): New function.
24242         (mips_start_function_definition): Emit .set micromips directive.
24243         (mips_call_may_need_jalx_p): New function.
24244         (mips_function_ok_for_sibcall): Add microMIPS support.
24245         (mips_print_operand_punctuation): Support short delay slots and
24246         compact jumps.
24247         (umips_swm_mask, umips_swm_encoding): New.
24248         (umips_build_save_restore): New function.
24249         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24250         (was_mips16_p): Remove.
24251         (old_compression_mode): New.
24252         (mips_set_compression_mode): New function.
24253         (mips_set_current_function): Add microMIPS support.
24254         (mips_option_override): Likewise.
24255         (umips_save_restore_pattern_p): New function.
24256         (umips_output_save_restore): New function.
24257         (umips_load_store_pair_p_1): New function.
24258         (umips_load_store_pair_p): New function.
24259         (umips_output_load_store_pair_1): New function.
24260         (umips_output_load_store_pair): New function.
24261         (umips_movep_target_p) New function.
24262         (mips_prepare_pch_save): Add microMIPS support.
24263         * config/mips/mips.h (TARGET_COMPRESSION): New.
24264         (TARGET_CPU_CPP_BUILTINS): Update macro
24265         to use new compression flags and to support microMIPS.
24266         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24267         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24268         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24269         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24270         (ASM_SPEC): Support mmicromips and mno-micromips.
24271         (M16STORE_REG_P): New macro.
24272         (MIPS_CALL): Support TARGET_MICROMIPS.
24273         (MICROMIPS_J): New macro.
24274         (mips_base_mips16): Rename this...
24275         (mips_base_compression_flags): ...to this.
24276         (UMIPS_12BIT_OFFSET_P): New macro.
24277         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24278         (MULTILIB_DIRNAMES): Likewise.
24280 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24282         PR target/52555
24283         * target-globals.c (save_target_globals): For init_reg_sets and
24284         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24286 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24288         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24289         * config/aarch64/t-aarch64
24290         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24292 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24294         PR inline-asm/56148
24295         * lra-constraints.c (process_alt_operands): Reload operand
24296         conflicting with earlier clobber only if no more other conflicting
24297         operands.
24299 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24301         PR sanitizer/56393
24302         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24303         if not linking a shared library.
24305 2013-02-22  Seth LaForge  <sethml@google.com>
24307         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24309 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24311         * config/arm/arm.md (split for extendsidi): Update condition.
24312         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24313         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24314         (qhs_zextenddi_cstr): Likewise.
24316 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24318         PR middle-end/56420
24319         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24320         avoid signed wrapping.
24321         (expand_mult): Handle properly multiplication by
24322         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24323         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24324         in the compiler if coeff is HOST_WIDE_INT_MIN.
24325         (expand_divmod): Don't make ext_op1 static, change it's type to
24326         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24328         PR rtl-optimization/50339
24329         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24330         field.
24331         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24332         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24333         into splitting_ashiftrt field.
24334         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24335         ASHIFTRT.
24336         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24337         choices.
24339 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24341         PR middle-end/56108
24342         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24343         are sure to go irrevocable.
24345 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24347         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24348         scalars are valid operands.
24350 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24352         PR tree-optimization/56310
24353         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24354         only matching indices and non-negative final offsets.
24355         (intersect_aggregates_with_edge): Pass src_idx to
24356         agg_replacements_to_vector.  Pass src_idx insstead of index to
24357         intersect_with_agg_replacements.
24359 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24361         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24362         instead of hard-wired defaults.
24364 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24366         * doc/invoke.texi (MIPS Options): Update documentation of the
24367         floating-point multiply-accumulate instruction restrictions.
24369 2013-02-21  Kostya Serebryany  <kcc@google.com>
24371         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24372         asan_shadow_offset on x86_64 linux.
24374 2013-02-21  Richard Biener  <rguenther@suse.de>
24376         PR tree-optimization/56415
24377         Revert
24378         2013-02-11  Richard Biener  <rguenther@suse.de>
24380         PR tree-optimization/56273
24381         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24382         first VRP run.
24384 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24386         PR bootstrap/56258
24387         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24388         instead of @itemx.
24390         PR inline-asm/56405
24391         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24392         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24394 2013-02-20  Jan Hubicka  <jh@suse.cz>
24396         PR tree-optimization/56265
24397         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24398         when target is referenced for first time.
24400 2013-02-20  Richard Biener  <rguenther@suse.de>
24402         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24403         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24404         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24405         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24406         not return anything.
24407         (rename_ssa_copies): Do not remove unused locals.
24408         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24409         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24410         * passes.c (execute_function_todo): Do not schedule unused locals
24411         removal if cleanup_tree_cfg did something.
24412         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24413         about the number of removed locals.
24415 2013-02-20  Richard Biener  <rguenther@suse.de>
24417         PR tree-optimization/56398
24418         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24420 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24422         PR tree-optimization/55334
24423         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24424         restricted pointers to arrays.
24426 2013-02-20  Richard Biener  <rguenther@suse.de>
24427             Jakub Jelinek  <jakub@redhat.com>
24429         PR tree-optimization/56396
24430         * tree-ssa-ccp.c (n_const_val): New static variable.
24431         (get_value): Return NULL for SSA names we don't have a lattice
24432         entry for.
24433         (ccp_initialize): Initialize n_const_val.
24434         * tree-ssa-copy.c (n_copy_of): New static variable.
24435         (init_copy_prop): Initialize n_copy_of.
24436         (get_value): Return NULL_TREE for SSA names we don't have a
24437         lattice entry for.
24439 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24441         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24443 2013-02-20  Richard Biener  <rguenther@suse.de>
24445         * genpreds.c (write_lookup_constraint): Do not compare first
24446         letter of the constraint again.
24448 2013-02-20  Richard Biener  <rguenther@suse.de>
24450         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24451         and ceil_log2.
24452         (get_use_iv_cost): Terminate hashtable walk when coming across
24453         an empty entry.
24455 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24457         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24458         reassociation for avx2 targets.
24460 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24462         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24463         Add version check for v8.10.a to enable microblaze_has_clz
24464         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24465         version and TARGET_PATTERN_COMPARE check
24466         * config/microblaze/microblaze.md: New clzsi2 instruction
24468 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24470         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24471         function before branching.
24473 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24475         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24476         DUMP_INSN_RTX_UID.
24477         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24479 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24481         PR middle-end/55889
24482         * sel-sched.c: Include ira.h.
24483         (implicit_clobber_conflict_p): New function.
24484         (moveup_expr): Use it.
24485         * Makefile.in (sel-sched.o): Depend on ira.h.
24487 2013-02-19  Richard Biener  <rguenther@suse.de>
24489         PR tree-optimization/56384
24490         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24491         (vn_hash_type): Split out from ...
24492         (vn_hash_constant_with_type): ... here.
24493         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24494         (vn_phi_eq): Compare types from vn_phi_s structure.
24495         (vn_phi_lookup): Populate vn_phi_s type.
24496         (vn_phi_insert): Likewise.
24498 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24500         PR tree-optimization/56350
24501         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24502         if haven't found reduction or nested cycle operand, rather than
24503         asserting we must find it.
24505         PR tree-optimization/56381
24506         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24507         to fold_build3.
24509 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24510             Jakub Jelinek  <jakub@redhat.com>
24512         PR target/52555
24513         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24514         (swap_optab_enable): Same.
24515         (init_all_optabs): Use argument instead of global.
24516         * tree.h (struct tree_optimization_option): New field target_optabs.
24517         * expr.h (init_all_optabs): Add argument to prototype.
24518         (TREE_OPTIMIZATION_OPTABS): New.
24519         (save_optabs_if_changed): Protoize.
24520         * optabs.h: Declare this_fn_optabs.
24521         * optabs.c (save_optabs_if_changed): New.
24522         Declare this_fn_optabs.
24523         (init_optabs): Add argument to init_all_optabs() call.
24524         * function.c (invoke_set_current_function_hook): Handle per
24525         function optabs.
24526         * function.h (struct function): New field optabs.
24527         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24528         optimization_current_node has changed.
24529         * target-globals.h (save_target_globals_default_opts): Protoize.
24530         * target-globals.c (save_target_globals_default_opts): New.
24532 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24534         PR target/56347
24535         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24536         registers %fr12 and %fr12R as call used.
24538         PR target/56214
24539         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24540         and HImode, require all displacements to be an integer multiple of
24541         their mode size.
24542         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24543         only allow QImode and HImode when reload is in progress and strict is
24544         true.  Likewise for symbolic addresses.  Use base14_operand to check
24545         displacements in REG+BASE addresses.
24547 2013-02-18  Richard Biener  <rguenther@suse.de>
24549         PR tree-optimization/56366
24550         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24551         handle sign-conversion of outer-loop initial induction value.
24553 2013-02-18  Richard Biener  <rguenther@suse.de>
24555         PR middle-end/56349
24556         * cfghooks.c (merge_blocks): If we merge a latch into another
24557         block adjust references to it.
24558         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24559         (verify_loop_structure): Verify that a recorded latch is in fact
24560         a latch.
24562 2013-02-18  Richard Biener  <rguenther@suse.de>
24564         PR tree-optimization/56321
24565         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24566         order SSA name release and virtual operand unlinking.
24568 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24570         * config/microblaze/microblaze.md (save_stack_block): Define.
24571         (restore_stack_block): Likewise.
24573 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24575         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24576         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24577         * config/microblaze/microblaze.c (microblaze_option_override):
24578         Bail out early for PIC modes when target does not support PIC.
24580 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24582         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24583         Replace with a microblaze version.
24584         (microblaze_trampoline_init): Adapt for microblaze.
24585         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24586         microblaze.
24588 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24589             Dodji Seketeli  <dodji@redhat.com>
24591         PR asan/56330
24592         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24593         (instrument_mem_region_access): Do not forget to always put
24594         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24595         0) statement, even for cases where either 'base' or 'base + len'
24596         are not instrumented -- because they have been previously
24597         instrumented.  Simplify the logic by putting all the statements
24598         instrument 'base + len' inside a sequence, and then insert that
24599         sequence right before the current insertion point.  Then, to
24600         instrument 'base + len', just get an iterator on that statement.
24601         And do not forget to update the pointer to iterator the function
24602         received as argument.
24604 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24606         PR rtl-optimization/56348
24607         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24609 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24611         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24612         (clean_graph_dump_file): Pass base to start_graph_dump.
24614 2013-02-14  Richard Henderson  <rth@redhat.com>
24616         PR target/55941
24617         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24619 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24621         * collect2-aix.h: Define F_LOADONLY.
24623 2013-02-14  Richard Biener  <rguenther@suse.de>
24625         PR lto/50494
24626         * varasm.c (output_constant_def_1): Get the decl representing
24627         the constant as argument.
24628         (output_constant_def): Wrap output_constant_def_1.
24629         (make_decl_rtl): Use output_constant_def_1 with the decl
24630         representing the constant.
24631         (build_constant_desc): Optionally re-use a decl already
24632         representing the constant.
24633         (tree_output_constant_def): Adjust.
24635 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24637         Fix an asan crash
24638         * asan.c (instrument_builtin_call):  Really put the length of the
24639         second source argument into src1_len.
24641 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24643         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24644         argument.  If it is false, don't create edge from then_bb to
24645         fallthru_bb.
24646         (insert_if_then_before_iter): Pass true to it.
24647         (build_check_stmt): Pass false to it.
24648         (transform_statements): Flush hash table only on extended basic
24649         block boundaries, rather than at the beginning of every bb.
24650         Don't flush hash table on nonfreeing_call_p calls.
24651         * tree-flow.h (nonfreeing_call_p): New prototype.
24652         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24654 2013-02-13  David S. Miller  <davem@davemloft.net>
24656         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24658 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24660         PR target/56184
24661         * ira.c (max_regno_before_ira): Move from ...
24662         (ira): ... here.
24663         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24664         vec_safe_length.
24666 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24668         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24670 2013-02-13  Richard Biener  <rguenther@suse.de>
24672         PR lto/56295
24673         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24674         globals in MEM_REFs.
24676 2013-02-13  Richard Biener  <rguenther@suse.de>
24678         * loop-init.c (loop_optimizer_init): Clear loop state when
24679         re-initializing preserved loops.
24680         * loop-unswitch.c (unswitch_single_loop): Return whether
24681         we unswitched the loop.  Do not verify loop state here.
24682         (unswitch_loops): When we unswitched a loop discover new loops.
24684 2013-02-13  Kostya Serebryany  <kcc@google.com>
24686         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24687         on x86_64 linux.
24688         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24690 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24692         Avoid instrumenting duplicated memory access in the same basic block
24693         * Makefile.in (asan.o): Add new dependency on hash-table.h
24694         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24695         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24696         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24697         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24698         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24699         (get_mem_ref_of_assignment): New functions.
24700         (get_mem_refs_of_builtin_call): Extract from
24701         instrument_builtin_call and tweak a little bit to make it fit with
24702         the new signature.
24703         (instrument_builtin_call): Use the new
24704         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24705         of is_gimple_builtin_call.
24706         (instrument_derefs, instrument_mem_region_access): Insert the
24707         instrumented memory reference into the hash table.
24708         (maybe_instrument_assignment): Renamed instrument_assignment into
24709         this, and change it to advance the iterator when instrumentation
24710         actually happened and return true in that case.  This makes it
24711         homogeneous with maybe_instrument_assignment, and thus give a
24712         chance to callers to be more 'regular'.
24713         (transform_statements): Clear the memory reference hash table
24714         whenever we enter a new BB, when we cross a function call, or when
24715         we are done transforming statements.  Use
24716         maybe_instrument_assignment instead of instrumentation.  No more
24717         need to special case maybe_instrument_assignment and advance the
24718         iterator after calling it; it's now handled just like
24719         maybe_instrument_call.  Update comment.
24721 2013-02-13  Richard Biener  <rguenther@suse.de>
24723         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24724         Fix loop discovery code.
24726 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24728         PR inline-asm/56148
24729         * lra-constraints.c (process_alt_operands): Match early clobber
24730         operand with itself.  Check conflicts with earlyclobber only if
24731         the operand is not reloaded.  Prefer to reload conflicting operand
24732         if earlyclobber and matching operands are the same.
24734 2013-02-12  Richard Biener  <rguenther@suse.de>
24736         PR lto/56297
24737         * lto-streamer-out.c (write_symbol): Do not output symbols
24738         for hard register variables.
24740 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24742         PR target/54222
24743         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24744         (umulsidi3_insn, mulsidi3_insn): New insns.
24746 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24748         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24749         (struct tune_params): Add vec_costs field.
24750         * config/arm/arm.c (arm_builtin_vectorization_cost)
24751         (arm_add_stmt_cost): New functions.
24752         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24753         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24754         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24755         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24756         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24757         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24758         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24760 2013-02-12  Richard Biener  <rguenther@suse.de>
24762         PR lto/56295
24763         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24764         decls again if possible.
24766 2013-02-12  Richard Biener  <rguenther@suse.de>
24768         PR middle-end/56288
24769         * tree-ssa.c (verify_ssa_name): Fix check, move
24770         SSA_NAME_IN_FREE_LIST check up.
24772 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24773             Steven Bosscher   <steven@gcc.gnu.org>
24775         PR rtl-optimization/56151
24776         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24777         equal to op0 or op1, and last_insn pattern is CODE operation
24778         with MEM dest and one of the operands matches that MEM.
24780 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24782         * doc/extend.texi: Document Function Multiversioning and "default"
24783         parameter string to target attribute.
24784         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24785         target attribute parameter is "default".
24786         (ix86_compare_version_priority): Remove checks for target attribute.
24787         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24788         Remove check for target attribute equal to NULL. Add assert.
24789         (ix86_generate_version_dispatcher_body): Change error to sorry.
24791 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24792             Jack Howarth  <howarth@bromo.med.uc.edu>
24793             Patrick Marlier  <patrick.marlier@gmail.com>
24795         PR libitm/55693
24796         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24797         define ENDFILE_SPEC as TM_DESTRUCTOR.
24798         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24800 2013-02-11  Alexander Potapenko  <glider@google.com>
24801             Jack Howarth  <howarth@bromo.med.uc.edu>
24802             Jakub Jelinek  <jakub@redhat.com>
24804         PR sanitizer/55617
24805         * config/darwin.c (cdtor_record): Rename ctor_record.
24806         (sort_cdtor_records): Rename sort_ctor_records.
24807         (finalize_dtors): New routine to sort destructors by
24808         priority before use in assemble_integer.
24809         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24811 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24813         PR rtl-optimization/56275
24814         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24815         offset is non-negative and less than cmode size before
24816         calling simplify_subreg.
24818 2013-02-11  Richard Biener  <rguenther@suse.de>
24820         PR tree-optimization/56264
24821         * cfgloop.h (fix_loop_structure): Adjust prototype.
24822         * loop-init.c (fix_loop_structure): Return the number of
24823         newly discovered loops.
24824         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24825         are discovered, do a full loop-closed SSA rewrite.
24827 2013-02-11  Richard Biener  <rguenther@suse.de>
24829         PR tree-optimization/56273
24830         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24831         first VRP run.
24832         (check_array_ref): Fix missing newline in dumps.
24833         (search_for_addr_array): Likewise.
24835 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24837         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24839 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24841         PR target/56256
24842         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24844 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24846         PR rtl-optimization/56246
24847         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24848         reload pseudo.
24849         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24850         constraints are satisfied.
24852 2013-02-08  Jeff Law  <law@redhat.com>
24854         PR debug/53948
24855         * emit-rtl.c (reg_is_parm_p): New function.
24856         * regs.h (reg_is_parm_p): New prototype.
24857         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24858         callee-clobbered registers.
24860 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24862         PR target/56043
24863         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24864         If there is no implicit builtin declaration, just return NULL.
24866 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24868         * config/i386/sse.md (FMAMODEM): New mode iterator.
24869         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24870         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24872 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24874         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24875         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24876         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24878 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24880         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24881         (microblaze*-*-elf): Likewise.
24882         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24883         LINK_SPEC.
24884         * config/microblaze/microblaze-c.c: Add builtin defines for
24885         _LITTLE_ENDIAN and _BIG_ENDIAN.
24886         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24887         add to TARGET_DEFAULT flags.
24888         Expand ASM_SPEC and LINK_SPEC.
24889         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24890         * config/microblaze/microblaze.md: Update extendsidi2 and
24891         movdi_internal instructions to use low-order / high-order reg
24892         print_operands.
24893         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24894         options and inversemask / mask of LITTLE_ENDIAN.
24895         * config/microblaze/t-microblaze: Expand multilib options to
24896         include mlittle-endian (le) and update exceptions patterns.
24898 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24900         PR rtl-optimization/56195
24901         * lra-constraints.c (get_reload_reg): Don't reuse regs
24902         if they have smaller mode than requested, if they have
24903         wider mode than requested, try to return a SUBREG.
24905         PR tree-optimization/56250
24906         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24907         if type is unsigned and code isn't MULT_EXPR.
24909 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24911         PR tree-optimization/56064
24912         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24913         bits according to mode.
24914         * fixed-value.h (fixed_from_double_int)
24915         (const_fixed_from_double_int): Adjust comments.
24917 2013-02-08  Richard Biener  <rguenther@suse.de>
24919         PR lto/56231
24920         * lto-streamer.h (struct data_in): Remove current_file, current_line
24921         and current_col members.
24922         * lto-streamer-out.c (lto_output_location): Stream changed bits
24923         en-block for efficiency.
24924         * lto-streamer-in.c (clear_line_info): Remove.
24925         (lto_input_location): Cache current file, line and column
24926         globally via local statics.  Read changed bits en-block.
24927         (input_function): Do not call clear_line_info.
24928         (lto_read_body): Likewise.
24929         (lto_input_toplevel_asms): Likewise.
24931 2013-02-08  Michael Matz  <matz@suse.de>
24933         PR tree-optimization/52448
24934         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24935         (nt_call_phase): New static.
24936         (add_or_mark_expr): Only mark accesses with newer phase than any
24937         call seen.
24938         (nonfreeing_call_p): New.
24939         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24940         (nt_fini_block): Keep blocks marked as visited.
24941         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24943 2013-02-08  Richard Biener  <rguenther@suse.de>
24945         * ira.c (ira): Free broken dominator information.
24947 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24949         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
24951 2013-02-08  Marek Polacek  <polacek@redhat.com>
24953         * cfgloop.c (verify_loop_structure): Add more checking of headers.
24955 2013-02-08  Richard Biener  <rguenther@suse.de>
24957         PR middle-end/56181
24958         * cfgloop.h (flow_loops_find): Adjust.
24959         (bb_loop_header_p): Declare.
24960         * cfgloop.c (bb_loop_header_p): New function split out from ...
24961         (flow_loops_find): ... here.  Adjust function signature,
24962         support incremental loop structure update.
24963         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
24964         * cfgloopmanip.c (fix_loop_structure): Move ...
24965         * loop-init.c (fix_loop_structure): ... here.
24966         (apply_loop_flags): Split out from ...
24967         (loop_optimizer_init): ... here.
24968         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
24969         in incremental mode, only remove dead loops here.
24971 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24973         PR target/54222
24974         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
24975         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
24976         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
24977         (*round<mode>3.libgcc): New insns for fixed-modes.
24978         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
24979         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
24980         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
24981         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
24982         implementations.  Define to __builtin_avr_absFX,
24983         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
24984         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
24985         __builtin_avr_countlsFX, respectively.
24986         * config/avr/avr-c.c (target.h): Include it.
24987         (enum avr_builtin_id): New enum.
24988         (avr_resolve_overloaded_builtin): New static function.
24989         (avr_register_target_pragmas): Use it to set
24990         targetm.resolve_overloaded_builtin.
24991         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
24992         tree nodes used by DEF_BUILTIN.
24993         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
24994         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
24995         <AVR_BUILTIN_xxBITS>: Same.
24997 2013-02-08  Richard Biener  <rguenther@suse.de>
24999         * cfgloop.c (verify_loop_structure): Properly handle
25000         a loop exiting to another loop header.
25001         * ira-int.h (ira_loops): Remove.
25002         * ira.c (ira_loops): Remove.
25003         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25004         (do_reload): Use loop_optimizer_finalize.
25005         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25006         number_of_loops to access the loop tree.
25007         (more_one_region_p): Likewise.
25008         (finish_loop_tree_nodes): Likewise.
25009         (rebuild_regno_allocno_maps): Likewise.
25010         (mark_loops_for_removal): Likewise.
25011         (mark_all_loops_for_removal): Likewise.
25012         (remove_unnecessary_regions): Likewise.
25013         (ira_build): Likewise.
25014         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25016 2013-02-08  Richard Biener  <rguenther@suse.de>
25018         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25019         * ipa-pure-const.c (analyze_function): Avoid calling
25020         mark_irreducible_loops twice.
25021         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25023 2013-02-07  David S. Miller  <davem@davemloft.net>
25025         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25026         on 'reg'.
25027         * var-tracking.c (vt_add_function_parameter): Test the presence of
25028         HAVE_window_save properly and do not remap argument registers when
25029         we have a leaf function.
25031 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25033         PR bootstrap/56227
25034         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25035         instead of "ll".
25036         * config/i386/i386.c (ix86_print_operand): Ditto.
25038 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25040         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25042 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25044         PR rtl-optimization/56225
25045         * lra-constraints.c (process_alt_operands): Check that reload hard
25046         reg can hold value for strict_low_part.
25048 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25050         PR debug/56154
25051         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25052         dwarf2out_end_function.
25053         (in_first_function_p, maybe_at_text_label_p,
25054         first_loclabel_num_not_at_text_label): New variables.
25055         (dwarf2out_var_location): In the first function find out
25056         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25057         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25058         functions.
25060 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25062         PR rtl-optimization/56178
25063         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25064         SUBREG of a register.  Tidy up related block of code.
25065         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25066         note if the source is a register or a SUBREG of a register.
25068 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25070         PR target/56228
25071         * config/rs6000/rs6000.md (ptrm): New mode attr.
25072         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25073         call_value_indirect_aix<pttrsize>,
25074         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25075         m in constraints.
25077 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25079         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25080         if -bnortl. Convert to strcmp and strncmp.
25082 2013-02-07  Alan Modra  <amodra@gmail.com>
25084         PR target/54009
25085         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25086         addresses won't wrap when offsetting.
25087         (rs6000_secondary_reload): Provide secondary reloads needed for
25088         wrapping LO_SUM addresses.
25090 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25092         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25093         MACH, just __MACH__.
25095 2013-02-06  Richard Biener  <rguenther@suse.de>
25097         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25098         instead of calling fix_loop_structure.
25100 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25102         PR middle-end/56217
25103         * omp-low.c (use_pointer_for_field): Return false if
25104         lower_send_shared_vars doesn't generate any copy-out code.
25106 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25108         PR rtl-optimization/56131
25109         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25110         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25111         of the label is NULL.  Add comment.
25113 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25115         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25117         PR sanitizer/55374
25118         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25119         (STATIC_LIBTSAN_LIBS): Likewise.
25120         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25121         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25122         is defined, don't add anything else beyond that.
25123         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25124         (LINK_COMMAND_SPEC): Use them.
25126         PR tree-optimization/56205
25127         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25128         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25129         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25131 2013-02-05  Richard Biener  <rguenther@suse.de>
25133         PR tree-optimization/53342
25134         PR tree-optimization/53185
25135         * tree-vectorizer.h (vect_check_strided_load): Remove.
25136         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25137         not disallow peeling for vectorized strided loads.
25138         (vect_check_strided_load): Make static and simplify.
25139         (vect_analyze_data_refs): Adjust.
25140         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25141         correctly when vectorizing strided loads.
25143 2013-02-05  Richard Biener  <rguenther@suse.de>
25145         * doc/install.texi: Refer to ISL, not PPL.
25147 2013-02-05  Jan Hubicka  <jh@suse.cz>
25149         PR tree-optimization/55789
25150         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25152 2013-02-05  Jan Hubicka  <jh@suse.cz>
25154         PR tree-optimization/55789
25155         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25156         the dead call anyway.
25158 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25160         PR sanitizer/55374
25161         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25163 2013-02-04  Alexander Potapenko  <glider@google.com>
25164             Jack Howarth  <howarth@bromo.med.uc.edu>
25165             Jakub Jelinek  <jakub@redhat.com>
25167         PR sanitizer/55617
25168         * config/darwin.c (sort_ctor_records): Stabilized qsort
25169         on constructor priority by using original position.
25170         (finalize_ctors): New routine to sort constructors by
25171         priority before use in assemble_integer.
25172         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25174 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25176         PR libstdc++/54314
25177         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25178         about visibility on artificial decls.
25179         * config/sol2.c (solaris_assemble_visibility): Likewise.
25181 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25183         PR target/56186
25184         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25185         argument and improve checking of return-argument types for 16-byte
25186         modes.
25187         (ix86_function_value_1): Add additional valtype argument on call
25188         of function_value_64.
25189         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25190         handling infunction_value_64 function.
25192 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25194         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25196 2013-02-04  Richard Biener  <rguenther@suse.de>
25198         PR tree-optimization/56188
25199         * tree-ssa-structalias.c (label_visit): Consider case with
25200         initially non-empty points-to set.
25201         (perform_var_substitution): Dump node mapping and clean up.
25203 2013-02-04  Richard Guenther  <rguenther@suse.de>
25205         PR lto/56168
25206         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25207         node prevail as last resort.
25208         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25209         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25211 2013-02-04  Richard Biener  <rguenther@suse.de>
25213         PR tree-optimization/56113
25214         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25215         Merge into ...
25216         (equiv_class_lookup_or_add): ... this.
25217         (label_visit): Adjust and fix error in previous patch.
25218         (perform_var_substitution): Adjust.
25220 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25222         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25223         * config/sh/sh4-300.md: Likewise.
25224         * config/sh/sh4a.md: Likewise.
25225         * config/sh/constraints.md: Likewise.
25226         * config/sh/sh.md: Likewise.
25227         * config/sh/netbsd-elf.h: Likewise.
25228         * config/sh/predicates.md: Likewise.
25229         * config/sh/sh-protos.h: Likewise.
25230         * config/sh/ushmedia.h: Likewise.
25231         * config/sh/linux.h: Likewise.
25232         * config/sh/sh.c: Likewise.
25233         * config/sh/superh.h: Likewise.
25234         * config/sh/elf.h: Likewise.
25235         * config/sh/sh4.md: Likewise.
25236         * config/sh/sh.h: Likewise.
25238 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25240         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25241         constraint to match_test floating_point_store_memory_operand().
25242         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25243         (base14_operand): New.
25244         (floating_point_store_memory_operand): New.
25245         (integer_store_memory_operand): Revise to use base14_operand and
25246         reg_plus_base_memory_operand.
25247         (move_dest_operand): Allow symbolic_memory_operands.
25248         (symbolic_memory_operand): Check for LO_SOM.
25249         (symbolic_operand): Change default case to break.
25250         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25251         CONST_DOUBLE values to be reloaded by putting them into memory when
25252         the destination is a floating point register.
25253         (movdf): Remove code to handle CONST_DOUBLE.
25254         (movsf): Likewise.
25255         (reload_indf_r1): New.
25256         (reload_insf_r1): New.
25257         Consistently use "Q" and "T" constraints with integer and floating
25258         point move instructions, respectively.
25259         (movdi): Remove FAIL.
25260         Change predicate for source operand unamed DImode move from
25261         general_operand to move_src_operand.
25262         (umulsidi3): Change predicate for destination operand to
25263         register_operand.
25264         Likewise for similar unamed patterns.
25265         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25266         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25267         (hppa_legitimize_address): Simplify mask calculation.
25268         (pa_emit_move_sequence): Revised handling of secondary reloads from
25269         REG+D addresses for floating point loads and stores.  Directly handle
25270         loading CONST0_RTX (mode) to a floating point register.
25271         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25272         to floating point registers.  Don't restrict secondary reloads to
25273         floating point registers to integer modes.  Revise some comments and
25274         cleanup some code.
25275         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25276         (pa_legitimate_address_p): New.
25277         (pa_legitimize_reload_address): New.
25278         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25279         (STRICT_REG_OK_FOR_BASE_P): New.
25280         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25281         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25283 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25284             Andrew Dixie  <andrewd@gentrack.com>
25286         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25287         flag set.
25289 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25291         * expmed.c (extract_bit_field_1): Pass the full width of the
25292         structure to get_best_reg_extraction_insn.
25294 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25296         PR target/54601
25297         * configure.ac (use_cxa_atexit): Add AIX.
25298         * configure: Regenerate.
25300         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25302 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25304         PR debug/54793
25305         * final.c (need_profile_function): New variable.
25306         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25307         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25308         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25309         notes, targetm.asm_out.function_prologue doesn't emit anything,
25310         HAVE_prologue and profiler should be emitted before prologue,
25311         set need_profile_function instead of emitting it.
25312         (final_scan_insn): If need_profile_function, emit
25313         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25314         NOTE_INSN_FUNCTION_BEG note.
25316 2013-02-01  Richard Henderson  <rth@redhat.com>
25318         * config/rs6000/rs6000.md (smulditi3): New.
25319         (umulditi3): New.
25321         * config/alpha/alpha.md (umulditi3): New.
25323 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25325         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25326         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25328 2013-02-01  Richard Biener  <rguenther@suse.de>
25330         PR tree-optimization/56113
25331         * tree-ssa-structalias.c (label_visit): Reduce work for
25332         single-predecessor nodes.
25334 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25336         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25337         range isn't testing for zero.
25339 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25341         PR middle-end/56113
25342         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25344 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25345             Nick Clifton  <nickc@redhat.com>
25347         * config/v850/constraints.md (Q): Define as a memory constraint.
25348         * config/v850/predicates.md (label_ref_operand): New predicate.
25349         (e3v5_shift_operand): New predicate.
25350         (ior_operator): New predicate.
25351         * config/v850/t-v850: Add e3v5 multilib.
25352         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25353         (v850_gen_movdi): Prototype.
25354         * config/v850/v850.c: Add support for e3v5 architecture.
25355         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25356         TARGET_V850E_UP.
25357         (construct_save_jarl): Add e3v5 long JARL support.
25358         (v850_adjust_insn_length): New function.  Adjust length of call
25359         insns when using e3v5 instructions.
25360         (v850_gen_movdi): New function: Generate instructions to move a
25361         DImode value.
25362         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25363         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25364         (TARGET_USE_FPU): Enable for e3v5.
25365         (CONST_OK_FOR_W): New macro.
25366         (ADJUST_INSN_LENGTH): Define.
25367         * config/v850/v850.md (UNSPEC_LOOP): Define.
25368         (attr cpu): Add v850e3v5.
25369         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25370         (movdi): New pattern.
25371         (movdi_internal): New pattern.
25372         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25373         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25374         (cstoresf4): Likewise.
25375         (cstoredf4): Likewise.
25376         (insv): New pattern.
25377         (rotlso3_a): New pattern.
25378         (rotlsi3_b): New pattern
25379         (rotlsi3_v850e3v5): New pattern.
25380         (doloop_begin): New pattern.
25381         (fix_loop_counter): New pattern.
25382         (doloop_end): New pattern.
25383         (branch_normal): Add e3v5 long branch support.
25384         (branch_invert): Likewise.
25385         (branch_z_normal): Likewise.
25386         (branch_z_invert): Likewise.
25387         (branch_nz_normal): Likewise.
25388         (branch_nz_invert): Likewise.
25389         (call_internal_short): Add e3v5 register-indirect JARL support.
25390         (call_internal_long): Likewise.
25391         (call_value_internal_short): Likewise.
25392         (call_value_internal_long): Likewise.
25393         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25394         (mloop): New option.
25395         * config.gcc: Add support for configuring v840e3v5 target.
25396         * doc/invoke.texi: Document new v850 specific command line options.
25398 2013-01-31  Paul Koning  <ni1d@arrl.net>
25400         PR debug/55059
25401         PR debug/54508
25402         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25403         children if parent is a class.
25404         (prune_unused_types_prune): Don't add DW_AT_declaration.
25406 2013-01-31  Richard Biener  <rguenther@suse.de>
25408         PR tree-optimization/56157
25409         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25410         match up operand with SLP child.
25412 2013-01-31  Jason Merrill  <jason@redhat.com>
25414         PR debug/54410
25415         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25416         parameters the first time.
25417         (gen_scheduled_generic_parms_dies): Check completeness here.
25419 2013-01-31  Richard Biener  <rguenther@suse.de>
25421         PR middle-end/53073
25422         * common.opt (faggressive-loop-optimizations): New flag,
25423         enabled by default.
25424         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25425         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25426         infer_loop_bounds_from_undefined by it.
25428 2013-01-31  Richard Biener  <rguenther@suse.de>
25430         PR tree-optimization/56150
25431         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25432         visit virtual operands.
25433         (find_uses_to_rename_bb): Likewise.
25435 2013-01-31  Richard Biener  <rguenther@suse.de>
25437         PR tree-optimization/56150
25438         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25439         mixed store non-store stmts.
25441 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25443         PR sanitizer/55374
25444         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25445         LIBASAN_EARLY_SPEC is defined.
25446         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25447         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25448         before %o.
25449         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25451         PR c++/55742
25452         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25453         invalid args instead of ICEing on it.
25454         (ix86_valid_target_attribute_tree): Return error_mark_node if
25455         ix86_valid_target_attribute_inner_p failed.
25456         (ix86_valid_target_attribute_p): Return false only if
25457         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25458         target("default") attribute.
25459         (sorted_attr_string): Change argument from const char * to tree,
25460         merge in all target attribute arguments rather than just one.
25461         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25462         instead of free.  Avoid using strcat.
25463         (ix86_mangle_function_version_assembler_name): Mangle
25464         target("default") as if no target attribute is present.  Adjust
25465         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25466         instead of xmalloc and XDELETEVEC instead of free.
25467         (ix86_function_versions): Don't return true if one of the decls
25468         doesn't have target attribute.  If they don't and one of the decls
25469         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25470         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25471         (ix86_supports_function_versions): Remove.
25472         (make_name): Fix up formatting.
25473         (make_dispatcher_decl): Remove resolver_name and its initialization.
25474         Avoid leaking memory.
25475         (is_function_default_version): Return true if there is
25476         target("default") attribute rather than no target attribute at all.
25477         (make_resolver_func): Avoid leaking memory.
25478         (ix86_generate_version_dispatcher_body): Likewise.
25479         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25480         * target.def (supports_function_versions): Remove.
25481         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25482         * doc/tm.texi: Regenerated.
25484 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25486         PR rtl-optimization/56144
25487         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25488         for values with side effects.
25490 2013-01-30  Richard Biener  <rguenther@suse.de>
25492         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25493         (sparseset_pop): Likewise.
25494         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25495         to be able to use quick_push in the worker loop.
25497 2013-01-30  Marek Polacek  <polacek@redhat.com>
25499         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25501 2013-01-30  Richard Biener  <rguenther@suse.de>
25503         PR lto/56147
25504         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25506 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25508         PR tree-optimization/56064
25509         * fixed-value.c (fixed_from_double_int): New function.
25510         * fixed-value.h (fixed_from_double_int): New prototype.
25511         (const_fixed_from_double_int): New static inline function.
25512         * fold-const.c (native_interpret_fixed): New static function.
25513         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25514         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25515         (native_encode_fixed): New static function.
25516         (native_encode_expr) <FIXED_CST>: Use it.
25517         (native_interpret_int): Move double_int worker code to...
25518         * double-int.c (double_int::from_buffer): ...this new static method.
25519         * double-int.h (double_int::from_buffer): Prototype it.
25521 2013-01-30  Richard Biener  <rguenther@suse.de>
25523         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25524         New pointer-map and obstack.
25525         (init_alias_vars): Allocate pointer-map and obstack.
25526         (delete_points_to_sets): Free them.
25527         (find_what_var_points_to): Cache result.
25528         (find_what_p_points_to): Adjust for changed interface of
25529         find_what_var_points_to.
25530         (compute_points_to_sets): Likewise.
25531         (ipa_pta_execute): Likewise.
25533 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25535         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25536         * configure: Regenerate.
25537         * config.in: Regenerate.
25538         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25539         #nobits/#progbits if supported.
25541 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25543         PR target/56121
25544         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25545         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25546         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25548 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25550         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25551         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25553 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25555         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25556         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25558 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25560         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25561         declaration.
25562         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25563         * config/arm/cortex-a7.md: New bypasses using
25564         arm_mac_accumulator_is_result.
25566 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25568         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25569         (cortex_a7_neon_mla): Likewise.
25570         (cortex_a7_fpfmad): New reservation.
25571         (cortex_a7_fpmacs): Use ffmas and update required units.
25572         (cortex_a7_fpmuld): Update required units and latency.
25573         (cortex_a7_fpmacd): Likewise.
25574         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25575         (cortex_a7_neon). Likewise.
25576         (bypass) Update participating units.
25578 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25580         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25581         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25582         from fmac to ffma.
25583         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25584         (vfp_fmul): Use ffmad.
25585         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25586         (cortex_r4_fmacd): Use ffmad.
25587         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25588         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25589         (cortex_a9_fmacd): Use ffmad.
25590         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25591         (cortex_a8_vfp_macd): Use ffmad.
25592         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25593         (cortex_a5_fpmacd): Use ffmad.
25594         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25595         (cortex_a15_vfp_macd): Use ffmad.
25596         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25598 2013-01-29  Jason Merrill  <jason@redhat.com>
25600         PR libstdc++/54314
25601         * varasm.c (default_assemble_visibility): Don't warn about
25602         visibility on artificial decls.
25604 2013-01-29  Richard Biener  <rguenther@suse.de>
25606         PR tree-optimization/56113
25607         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25608         the bitmap leader.
25609         (label_visit): Free duplicate bitmaps and record the leader instead.
25610         (perform_var_substitution): Adjust.
25612 2013-01-29  Richard Biener  <rguenther@suse.de>
25614         PR tree-optimization/55270
25615         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25616         the CFG, schedule loops for fixup.
25618 2013-01-29  Nick Clifton  <nickc@redhat.com>
25620         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25621         SP_REG.
25623 2013-01-28  Leif Ekblad  <leif@rdos.net>
25625         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25626         * config/i386/i386.h (TARGET_RDOS): New macro.
25627         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25628         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25629         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25630         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25631         DEFAULT_LARGE_SECTION_THRESHOLD.
25632         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25633         * config/i386/rdos.h: New file.
25634         * config/i386/rdos64.h: New file.
25636 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25638         PR other/54814
25639         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25640         TEST_HARD_REG_BIT.
25642 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25644         PR rtl-optimization/56117
25645         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25646         call cselib_lookup_from_insn on the MEM before calling
25647         add_insn_mem_dependence.
25649 2013-01-28  Richard Biener  <rguenther@suse.de>
25651         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25652         to a stmt that didn't have one.
25653         (copy_phis_for_bb): Likewise for PHI arguments.
25654         (copy_debug_stmt): Likewise for debug stmts.
25656 2013-01-28  Richard Biener  <rguenther@suse.de>
25658         PR tree-optimization/56034
25659         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25660         (partition_builtin_p): Adjust.
25661         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25662         it is the last partition.
25663         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25664         up the vertex for the definition.
25665         (classify_partition): Classify whether a partition is a
25666         PKIND_REDUCTION, thus has uses outside of the loop.
25667         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25668         Merge all PKIND_REDUCTION partitions into the last partition.
25669         (tree_loop_distribution): Seed partitions from reductions as well.
25671 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25673         PR tree-optimization/56125
25674         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25675         pow(x,c) into sqrt(x) * powi(x, n/2) or
25676         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25677         optimizing for size.
25678         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25679         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25680         integer.
25682         PR tree-optimization/56094
25683         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25684         to UNKNOWN_LOCATION while gimplifying expr.
25686 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25688         PR target/56114
25689         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25690         operand 0 in movabs insn template for -masm=intel asm alternative.
25691         (*movabs<mode>_2): Ditto for operand 1.
25693 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25695         PR target/54663
25696         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25697         of microblaze-c.o
25699 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25701         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25702         tm_file.
25704 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25706         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25707         Undef to avoid warning.
25709 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25711         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25712         * configure: Regenerate.
25714 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25716         PR tree-optimization/56098
25717         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25718         for stmts with volatile ops.
25719         (cond_store_replacement): Don't optimize if assign has volatile ops.
25720         (cond_if_else_store_replacement_1): Don't optimize if either
25721         then_assign or else_assign have volatile ops.
25722         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25723         volatile ops.
25725 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25727         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25729 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25731         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25732         missing ':' in asm example.
25734 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25736         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25737         entries into lane and laneq entries.
25738         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25739         Remove AdvSIMD scalar modes.
25740         (aarch64_sq<r>dmulh_laneq<mode>): New.
25741         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25742         modes.
25743         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25744         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25745         * config/aarch64/iterators.md (VCOND): New.
25746         (VCONQ): New.
25748 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25750         PR target/54222
25751         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25752         Add NULL LIBNAME argument to existing definitions.
25753         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25754         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25755         * config/avr/avr.c (DEF_BUILTIN): Same.
25756         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25757         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25758         implementation is available (DECL_ASSEMBLER_NAME is set).
25759         (avr_fold_absfx): New static function.
25760         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25761         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25762         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25763         AVR_BUILTIN_ABSLLK.
25764         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25765         (abshk, absk, abslk, absllk): Provide as static inline functions.
25767 2013-01-25  Marek Polacek  <polacek@redhat.com>
25769         PR tree-optimization/56035
25770         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25772 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25774         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25775         (*movtf_internal_rex64): Add (!o,C) alternative
25776         (*movxf_internal_rex64): Ditto.
25777         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25779 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25781         * doc/invoke.texi: fix typo.
25782         * doc/objc.texi: fix typo.
25784 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25786         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25787         for the first two alternatives.
25789 2013-01-24  Diego Novillo  <dnovillo@google.com>
25791         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25792         (ggc-zone.o): Remove.
25793         * configure.ac: Remove option --with-gc.
25794         * configure: Re-generate.
25795         * doc/install.texi: Remove documentation for --with-gc.
25796         * gengtype.c (write_enum_defn): Remove.  Update all users.
25797         (write_Types_process_field): Remove generation of gt_e_* argument.
25798         (output_type_enum): Remove.  Update all users.
25799         (write_enum_defn): Remove.  Update all users.
25800         (enum alloc_zone): Remove.  Update all users.
25801         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25802         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25803         Update all callers.
25804         (struct ptr_data): Remove field TYPE.  Update all users.
25805         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25806         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25807         Update all users.
25808         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25809         (struct alloc_zone): Remove.
25810         (ggc_internal_alloc_zone_stat): Remove.
25811         (ggc_internal_cleared_alloc_zone_stat): Remove.
25812         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25813         (ggc_pch_count_object): Remove last argument.  Update all users.
25814         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25815         (struct alloc_zone): Remove.
25816         * ggc-zone.c: Remove.
25817         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25818         (struct alloc_zone): Remove.
25819         (ggc_alloc_typed_stat): Remove.
25820         (ggc_alloc_typed): Remove.
25821         (ggc_splay_alloc): Remove first argument.
25822         (rtl_zone): Remove.  Update all users.
25823         (tree_zone): Remove.  Update all users.
25824         (tree_id_zone): Remove.  Update all users.
25825         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25826         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25827         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25828         * tree-ssanames.c: Remove references to zone allocator in comments.
25830 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25832         * config/avr/avr.c (avr_out_fract): Make register numbers that
25833         might be outside of source operand signed.
25835 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25837         * config/i386/constraints.md (Yf): New constraint.
25838         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25839         of f constraint to conditionaly disable x87 register preferences.
25840         (*movdf_internal): Ditto.
25841         (*movsf_internal): Ditto.
25843 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25845         PR inline-asm/55934
25846         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25847         that have operands with impossible constraints.
25848         Add a FIXME for a speed-up opportunity.
25849         * lra-constraints.c (process_alt_operands): Verify that a class
25850         selected from constraints on asms is valid for the operand mode.
25851         (curr_insn_transform): Remove incorrect comment.
25853 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25855         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25856         TOC operand is a valid symbol ref in the constant pool.
25858 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25860         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25862 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25864         PR target/54222
25865         * config/avr/stdfix.h: New file.
25866         * t-avr (stdfix-gcc.h): New rule to build it.
25867         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25869 2013-01-23  Kostya Serebryany  <kcc@google.com>
25871         * config/darwin.h: remove dependency on
25872         CoreFoundation (asan on Mac OS).
25874 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25876         PR target/49069
25877         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25878         instead of cmpdi_operand for first comparison operand.
25879         Don't assert that comparison operands aren't both constants.
25881 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25883         * doc/install.texi (Downloading the Source): Update references to
25884         downloading separate components.
25886 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25888         * doc/extend.texi (__int128): Improve grammar.
25890 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25892         PR target/56028
25893         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25894         alternative to (o,r).
25895         (*movdi_internal_rex64): Remove (!o,n) alternative.
25896         (DImode immediate->memory splitter): Remove.
25897         (DImode immediate->memory peephole2): Remove.
25898         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25899         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25900         alternative to (!o,*r).
25901         (*movtf_internal_sse): New pattern.
25902         (*movxf_internal_rex64): New pattern.
25903         (*movxf_internal): Disable for TARGET_64BIT.
25904         (*movdf_internal_rex64): Remove (!o,F) alternative.
25906 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25908         PR middle-end/56074
25909         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25910         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25911         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25912         stmt locations where LOCATION_LOCUS of the stmt location is
25913         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25915         PR target/55686
25916         * config/i386/i386.md (UNSPEC_STOS): New.
25917         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25918         *strsetqi_1): Add UNSPEC_STOS.
25920 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25922         PR c++/56067
25923         * doc/invoke.texi: Remove left over -Wsynth example.
25925 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25927         PR tree-optimization/56051
25928         * fold-const.c (fold_binary_loc): Don't fold
25929         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25930         a narrowing conversion, or widening conversion from signed
25931         to unsigned.
25933 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25935         PR rtl-optimization/56023
25936         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25937         dependent on debug instruction.
25939 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25941         PR middle-end/56022
25942         * function.c (allocate_struct_function): Call
25943         invoke_set_current_function_hook earlier.
25945 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25947         * reload1.c (init_reload): Only initialize reload_obstack
25948         during the first call.
25950 2013-01-21  Marek Polacek  <polacek@redhat.com>
25952         * cfgloop.c (verify_loop_structure): Fix up grammar.
25954 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
25956         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
25957         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
25959 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
25961         PR target/56058
25962         * config/arm/marvell-pj4.md: Update copyright year.
25963         Fix up use of alu to alu_reg and simple_alu_imm.
25965 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25967         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
25969 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
25971         PR target/55433
25972         * lra-constraints.c (curr_insn_transform): Don't reuse original
25973         insn for secondary memory move when memory mode should be different.
25975 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25977         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
25978         atomic_storedi_1): New patterns.
25980 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
25982         btver2 pipeline descriptions.
25983         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
25984         descriptions.
25985         * config/i386/i386.md (btver2_decode): New type attributes.
25986         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
25987         type attributes.
25988         * config/i386/btver2.md: New file describing btver2 pipelines.
25990 2013-01-19  Andrew Pinski  <apinski@cavium.com>
25992         PR tree-optimization/52631
25993         * tree-ssa-sccvn (visit_use): Before looking up the original
25994         statement, try looking up the simplified expression.
25996 2013-01-19  Anthony Green  <green@moxielogic.com>
25998         * config/moxie/moxie.c (moxie_expand_prologue): Set
25999         current_function_static_stack_size.
26001 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26003         PR tree-optimization/56029
26004         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26005         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26007 2013-01-18  Sharad Singhai  <singhai@google.com>
26009         PR tree-optimization/55995
26010         * dumpfile.c (dump_loc): Print location only if available.
26011         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26013 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26015         PR target/55433
26016         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26017         secondary memory move.
26018         (inherit_reload_reg): Use rclass instead of cl for
26019         check_secondary_memory_needed_p.
26021 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26023         PR middle-end/56015
26024         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26025         the case where writing real complex part of target modifies op1.
26027 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26029         * config/aarch64/aarch64-simd.md
26030         (aarch64_vcond_internal<mode>): Handle unordered cases.
26031         * config/aarch64/iterators.md (v_cmp_result): New.
26033 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26034             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26036         * config/arm/marvell-pj4.md: New file.
26037         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26038         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26039         (generic_vfp): Likewise.
26040         * config/arm/arm-cores.def: Add marvell-pj4.
26041         * config/arm/arm-tune.md: Regenerate.
26042         * config/arm/arm-tables.opt: Regenerate.
26043         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26044         * doc/invoke.texi: Document marvell-pj4.
26046 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26048         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26050 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26052         PR debug/54114
26053         PR debug/54402
26054         PR debug/49888
26055         * var-tracking.c (negative_power_of_two_p): New.
26056         (global_get_addr_cache, local_get_addr_cache): New.
26057         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26058         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26059         heading comment.
26060         (vt_stack_offset_p): Remove.
26061         (vt_canon_true_dep): Always canonicalize loc's address.
26062         (clobber_overlapping_mems): Make sure we have a MEM.
26063         (local_get_addr_clear_given_value): New.
26064         (val_reset): Clear local cached entries.
26065         (compute_bb_dataflow): Create and release the local cache.
26066         Disable duplicate MEMs clobbering.
26067         (emit_notes_in_bb): Clobber MEMs likewise.
26068         (vt_emit_notes): Create and release the local cache.
26069         (vt_initialize, vt_finalize): Create and release the global
26070         cache, respectively.
26071         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26073 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26075         PR libmudflap/53359
26076         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26077         not found in the symtab.
26079 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26081         PR debug/56006
26082         PR rtl-optimization/55547
26083         PR rtl-optimization/53827
26084         PR debug/53671
26085         PR debug/49888
26086         * alias.c (offset_overlap_p): New, factored out of...
26087         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26088         the conservative special case for symbolic constants.  Don't
26089         adjust zero sizes on alignment.
26091 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26093         PR rtl-optimization/52573
26094         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26095         REG_UNUSED for the same register.
26097 2013-01-17  Richard Biener  <rguenther@suse.de>
26098             Marek Polacek  <polacek@redhat.com>
26100         PR rtl-optimization/55833
26101         * loop-unswitch.c (unswitch_loops): Move loop verification...
26102         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26103         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26104         Set it to true when we're removing a loop from hierarchy tree in
26105         an irreducible region.
26106         (fix_bb_placements): Adjust caller.
26107         (fix_loop_placements): Likewise.
26109 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26111         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26112         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26113         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26114         Remove ID.  Adjust comments.
26115         * config/avr/avr-c.c (avr_builtin_name): Remove.
26116         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26117         * config/avr/avr.c (avr_tolower): New static function.
26118         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26119         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26120         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26121         default expansion.
26123 2013-01-17  Jan Hubicka  <jh@suse.cz>
26125         PR tree-optimization/55273
26126         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26128 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26130         PR target/55981
26131         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26132         store through atomic_store<mode>_1.
26133         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26135 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26137         PR tree-optimizations/55264
26138         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26139         for virtual methods.
26140         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26141         virtual methods before inlining is over.
26142         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26143         virtual functions.
26144         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26145         non-virtual.
26147 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26149         PR rtl-optimization/56005
26150         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26151         pending reads for prefetch.
26153 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26155         * config/aarch64/aarch64.md
26156         (*cstoresi_neg_uxtw): New pattern.
26157         (*cmovsi_insn_uxtw): New pattern.
26158         (*<optab>si3_uxtw): New pattern.
26159         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26160         (*<optab>si3_insn_uxtw): New pattern.
26161         (*bswapsi2_uxtw): New pattern.
26163 2013-01-16  Richard Biener  <rguenther@suse.de>
26165         * tree-inline.c (tree_function_versioning): Remove set but
26166         never used variable.
26168 2013-01-16  Richard Biener  <rguenther@suse.de>
26170         PR tree-optimization/55964
26171         * tree-flow.h (rename_variables_in_loop): Remove.
26172         (rename_variables_in_bb): Likewise.
26173         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26174         (copy_loop_before): Adjust and delete update-ssa status.
26175         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26176         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26177         (rename_variables_in_loop): Remove.
26178         (slpeel_update_phis_for_duplicate_loop): Likewise.
26179         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26180         use available cfg machinery instead of duplicating it.
26181         Update PHI nodes and perform poor-mans SSA update here.
26182         (slpeel_tree_peel_loop_to_edge): Adjust.
26184 2013-01-16  Richard Biener  <rguenther@suse.de>
26186         PR tree-optimization/54767
26187         PR tree-optimization/53465
26188         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26189         (vrp_visit_phi_node): For PHI arguments coming via backedges
26190         drop all symbolical range information.
26191         (execute_vrp): Compute backedges.
26193 2013-01-16  Richard Biener  <rguenther@suse.de>
26195         * doc/install.texi: Update CLooG and ISL requirements to
26196         0.18.0 and 0.11.1.
26198 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26200         PR target/55301
26201         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26202         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26203         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26205 2013-01-16  DJ Delorie  <dj@redhat.com>
26207         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26208         (UNSPECV_SP_SWITCH_E): New.
26209         (sp_switch_1): Change to an unspec.
26210         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26211         replace $r15.
26213 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26215         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26216         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26217         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26218         (expand_mem_thread_fence): Ditto.
26219         (expand_mem_signal_fence): Ditto.
26220         (expand_atomic_load): Ditto.
26221         (expand_atomic_store): Ditto.
26223 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26225         PR rtl-optimization/55547
26226         PR rtl-optimization/53827
26227         PR debug/53671
26228         PR debug/49888
26229         * alias.c (memrefs_conflict_p): Set sizes to negative after
26230         AND adjustments.
26232 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26234         PR target/55940
26235         * function.c (thread_prologue_and_epilogue_insns): Always
26236         add crtl->drap_reg to set_up_by_prologue.set, even if
26237         stack_realign_drap is false.
26239 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26241         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26242         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26243         *call): Fix indention.
26245 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26247         PR target/55876
26248         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26249         Update comment.
26251 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26253         PR rtl-optimization/55153
26254         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26256 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26258         PR tree-optimization/55920
26259         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26260         accesses as grp_to_be_debug_replaced.
26262 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26264         PR tree-optimization/55920
26265         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26266         there is non-useless type conversion needed from debug rhs to lhs,
26267         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26269 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26270             Mikael Pettersson  <mikpe@it.uu.se>
26272         PR target/43961
26273         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26274         Thumb.
26275         (ASM_OUTPUT_CASE_LABEL): Remove.
26276         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26277         * final.c (shorten_branches): Update alignment of labels before
26278         jump tables if CASE_VECTOR_SHORTEN_MODE.
26280 2013-01-15  Richard Biener  <rguenther@suse.de>
26282         PR bootstrap/55961
26283         * system.h: Do not include gmp.h for building host tools.
26285 2013-01-15  Richard Biener  <rguenther@suse.de>
26287         PR middle-end/55882
26288         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26289         account for bitpos when computing alignment.
26291 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26293         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26294         (ix86_target_macros_internal): Likewise.
26296         * config/i386/i386.c (m_CORE2I7): Removed.
26297         (m_CORE_HASWELL): New macro.
26298         (m_CORE_ALL): Likewise.
26299         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26300         (initial_ix86_arch_features): Likewise.
26301         (processor_target_table): Initializations for Core avx2.
26302         (cpu_names): New names "core-avx2".
26303         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26304         PROCESSOR_CORE_HASWELL.
26305         (ix86_issue_rate): New case.
26306         (ia32_multipass_dfa_lookahead): Likewise.
26307         (ix86_sched_init_global): Likewise.
26309         * config/i386/i386.h (TARGET_HASWELL): New macro.
26310         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26311         (processor_type): New PROCESSOR_HASWELL.
26313 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26315         PR tree-optimization/55955
26316         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26317         *SHIFT_EXPR and *ROTATE_EXPR codes.
26319         PR tree-optimization/48766
26320         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26321         -ftrapv disable -fwrapv.
26323 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26325         PR target/55974
26326         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26327         etc. to 1 and not to __flash.
26328         Use LL suffix for __INT24_MAX__ with -mint8.
26329         Use ULL suffix for __UINT24_MAX__ with -mint8.
26331 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26333         * config/avr/avr-arch.h
26334         (struct base_arch_s): Use typedef avr_arch_t instead.
26335         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26336         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26337         * config/avr/avr.c: Same.
26338         * config/avr/avr-devices.c: Same.
26339         * config/avr/driver-avr.c: Same.
26340         * config/avr/gen-avr-mmcu-texi.c: Same.
26341         * config/avr/avr-mcus.def: Adjust comment.
26343 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26345         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26346         * config/aarch64/iterators.md (VALLDI): New.
26348 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26349             Andi Kleen  <ak@linux.intel.com>
26351         PR target/55948
26352         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26353         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26354         memmodel flag.
26356 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26358         * config/avr/avr-stdint.h: Remove trailing blanks.
26359         * config/avr/avr-log.h: Same.
26360         * config/avr/avr-arch.h: Same.
26361         * config/avr/avr-devices.c: Same.
26362         * config/avr/avr-dimode.md: Same.
26363         * config/avr/predicates.md: Same.
26364         * config/avr/avr-c.c: Same.  And fix typo.
26366         * config/avr/avr-protos.h: Same.  And:
26367         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26368         (init_cumulative_args): Rename to avr_init_cumulative_args.
26369         (expand_prologue): Rename to avr_expand_prologue.
26370         (expand_epilogue): Rename to avr_expand_epilogue.
26371         (adjust_insn_length): Rename to avr_adjust_insn_length.
26372         (notice_update_cc): Rename to avr_notice_update_cc.
26373         (final_prescan_insn): Rename to avr_final_prescan_insn.
26374         * config/avr/avr.c: Same.
26375         * config/avr/avr.h: Same.
26376         * config/avr/avr.md: Remove trailing blanks.
26377         (prologue): Use avr_expand_prologue.
26378         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26380 2013-01-14  Richard Biener  <rguenther@suse.de>
26382         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26383         verify_location, collect_subblocks): New functions.
26384         (verify_gimple_in_cfg): Verify that locations only reference
26385         BLOCKs in the functions BLOCK tree.
26387 2013-01-14  Richard Biener  <rguenther@suse.de>
26389         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26390         PHI argument.
26391         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26392         unshare reference.
26393         (insert_out_of_ssa_copy_on_edge): Likewise.
26394         (rewrite_close_phi_out_of_ssa): Likewise.
26395         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26396         debug expressions.
26397         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26398         propagated constants.
26399         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26400         can not be shared.
26402 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26404         * config/avr/avr-modes.def: Add GPL copyright notice.
26406 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26408         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26409         MEMMODEL_MASK to determine memory model.
26410         (atomic_store<mode>): Ditto from operands[2].
26411         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26413 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26415         PR fortran/55935
26416         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26417         (fold_gimple_assign): Don't call unshare_expr here.
26418         (fold_ctor_reference): Call unshare_expr.
26420 2013-01-13  Terry Guo  <terry.guo@arm.com>
26422         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26423         * doc/fragments.texi: Document MULTILIB_REUSE.
26424         * gcc.c (multilib_reuse): New internal spec.
26425         (set_multilib_dir): Also search multilib from multilib_reuse.
26426         * genmultilib (tmpmultilib3): Refactor code.
26427         (tmpmultilib4): Ditto.
26428         (multilib_reuse): New multilib argument.
26430 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26432         * Makefile.in: Update copyright.
26434 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26436         PR middle-end/55890
26437         * calls.c (expand_call): Check if arg_nr is valid.
26439 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26441         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26442         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26443         documentation.  Add missing '__' in front of
26444         __builtin_ia32_packssdw256.
26446 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26448         PR target/55719
26449         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26450         NO_REGS for larl operands.
26451         (s390_reload_larl_operand): Use s390_load_address instead of
26452         emit_move_insn.
26454 2013-01-11  Richard Biener  <rguenther@suse.de>
26456         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26457         (verify_node_sharing): ... here.
26458         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26460 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26462         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26463         Substitute TREECHECKING.
26464         * configure: Regenerate.
26465         * Makefile.in (TREECHECKING): New.
26467 2013-01-11  Richard Guenther  <rguenther@suse.de>
26469         PR tree-optimization/44061
26470         * tree-vrp.c (extract_range_basic): Compute zero as
26471         value-range for __builtin_constant_p of function parameters.
26473 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26475         Update copyright years.
26477 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26479         PR rtl-optimization/55672
26480         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26481         const to be eliminable.
26483 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26485         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26486         * configure: Regenerate.
26488 2013-01-10  Richard Biener  <rguenther@suse.de>
26490         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26492 2013-01-10  Richard Biener  <rguenther@suse.de>
26494         PR bootstrap/55792
26495         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26496         locations for virtual PHI arguments.
26497         (rewrite_update_phi_arguments): Likewise.
26499 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26501         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26502         on to assembler.
26504 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26506         PR tree-optimization/55921
26507         * tree-complex.c (expand_complex_asm): New function.
26508         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26510 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26512         PR target/55718
26513         * config/s390/s390.c (s390_symref_operand_p)
26514         (s390_loadrelative_operand_p): Merge the two functions.
26515         (s390_check_qrst_address, print_operand_address): Add parameters
26516         to s390_loadrelative_operand_p invokation.
26517         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26518         (s390_reload_larl_operand, s390_secondary_reload): Use
26519         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26520         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26522 2013-01-09  Mike Stump  <mikestump@comcast.net>
26524         * dse.c (record_store): Remove unnecessary assert.
26526 2013-01-09  Jan Hubicka  <jh@suse.cz>
26528         PR tree-optimization/55569
26529         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26530         * cfgloop.h (scale_loop_profile): Likewise.
26532 2013-01-09  Jan Hubicka  <jh@suse.cz>
26534         PR lto/45375
26535         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26536         functions.
26537         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26539 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26541         PR middle-end/55114
26542         * expr.h (maybe_emit_group_store): Declare.
26543         * expr.c (maybe_emit_group_store): New function.
26544         * builtins.c (expand_builtin_int_roundingfn): Call it.
26545         (expand_builtin_int_roundingfn_2): Likewise.
26547 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26549         PR rtl-optimization/55829
26550         * lra-constraints.c (match_reload): Add code for absent output.
26551         (curr_insn_transform): Add code for reloads of matched inputs
26552         without output.
26554 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26556         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26557         attribute of movddup insn to DF.
26558         (*vec_interleave_lowv2df): Ditto.
26559         (vec_dupv2df): Ditto.
26561 2013-01-09  Jan Hubicka  <jh@suse.cz>
26563         PR tree-optimiation/55875
26564         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26565         EVERY_ITERATION parameter.
26566         (number_of_iterations_exit): Check if exit is executed every iteration.
26567         (idx_infer_loop_bounds): Similarly here.
26568         (n_of_executions_at_most): Simplify
26569         to only test for cases where statement is dominated by the
26570         particular bound; handle correctly the "postdominance" test.
26571         (scev_probably_wraps_p): Use max loop iterations info
26572         as a global bound first.
26574 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26575             Nick Clifton  <nickc@redhat.com>
26577         * config/v850/v850.md (cbranchsf4): New pattern.
26578         (cstoresf4): New pattern.
26579         (cbranchdf4): New pattern.
26580         (cstoredf4): New pattern.
26581         (movsicc): Disallow floating point comparisons.
26582         (cmpsf_le_insn): Fix order of operators.
26583         (cmpsf_lt_insn): Likewise.
26584         (cmpsf_eq_insn): Likewise.
26585         (cmpdf_le_insn): Likewise.
26586         (cmpdf_lt_insn): Likewise.
26587         (cmpdf_eq_insn): Likewise.
26588         (cmpsf_ge_insn): Use LE comparison.
26589         (cmpdf_ge_insn): Likewise.
26590         (cmpsf_gt_insn): Use LT comparison.
26591         (cmpdf_gt_insn): Likewise.
26592         (cmpsf_ne_insn): Delete pattern.
26593         (cmpdf_ne_insn): Delete pattern.
26594         * config/v850/v850.c (v850_gen_float_compare): Use
26595         gen_cmpdf_eq_insn for NE comparison.
26596         (v850_float_z_comparison_operator)
26597         (v850_float_nz_comparison_operator): Move from here ...
26598         * config/v850/predicates.md: ... to here.  Move GT and GE
26599         comparisons into v850_float_z_comparison_operator.
26600         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26601         Delete prototype.
26602         (v850_float_nz_comparison_operator): Likewise.
26604 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26606         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26607         with calls to gen_insvsi/gen_insvdi.
26609 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26611         * config/i386/i386.c (initial_ix86_tune_features): Set up
26612         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26614 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26615             Jakub Jelinek  <jakub@redhat.com>
26617         PR tree-optimization/48189
26618         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26619         If nitercst is 0, don't predict the exit edge.
26621 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26623         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26624         in asm_fprintf with reg_names.
26625         (aarch64_print_operand_address): Likewise.
26626         (aarch64_return_addr): Likewise.
26627         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26629 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26631         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26632         (INT_U6_BITS): Likewise.
26633         * config/pa/predicates.md (uint6_operand): New predicate.
26634         (shift5_operand, shift6_operand): Likewise.
26635         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26636         arith32_operand.
26637         (lshrdi3): Use shift6_operand.
26638         (shrpsi4, shrpdi4): New insn patterns.
26639         (extzv): Delete expander.
26640         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26641         predicates in unamed zero extract patterns.  Tighten common constraint.
26642         (extv): Delete expander.
26643         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26644         predicates in unamed sign extract patterns.  Tighten common constraint.
26645         (insv): Delete expander.
26646         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26647         predicates in unamed insert patterns.  Tighten common constraint.
26648         Change uint32_operand predicate to uint6_operand predicate in unamed
26649         DImode pattern to insert constant values of type 1...1xxxx.
26651 2013-01-04  Jan Hubicka  <jh@suse.cz>
26653         PR tree-optimization/55823
26654         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26655         issue.
26657 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26658             Uros Bizjak  <ubizjak@gmail.com>
26660         PR rtl-optimization/55845
26661         * df-problems.c (can_move_insns_across): Stop scanning at
26662         volatile_insn_p source instruction or give up if
26663         across_from .. across_to range contains any volatile_insn_p
26664         instructions.
26666 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26668         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26669         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26670         Declare.
26671         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26672         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26674 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26676         PR fortran/55341
26677         * asan.c (asan_clear_shadow): New function.
26678         (asan_emit_stack_protection): Use it.
26680 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26682         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26683         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26684         with tab instead of space.
26686 2013-01-08  Nick Clifton  <nickc@redhat.com>
26688         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26689         register bank 0 at the start of an interrupt handler.
26690         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26691         MDBH registers.
26693 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26695         * config/aarch64/aarch64-simd.md
26696         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26697         (aarch64_simd_bsl): Likewise.
26698         (aarch64_vcond_internal<mode>): Likewise.
26699         (vcond<mode><mode>): Likewise.
26700         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26701         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26703 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26705         * config/aarch64/aarch64-builtins.c
26706         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26708 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26710         PR debug/55579
26711         * tree-sra.c (analyze_access_subtree): Return true also after
26712         potentially creating a debug-only replacement.
26714 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26716         PR middle-end/55890
26717         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26719         PR tree-optimization/54120
26720         * tree-vrp.c (range_fits_type_p): Don't allow
26721         src_precision < precision from signed vr to unsigned_p
26722         if vr->min or vr->max is negative.
26723         (simplify_float_conversion_using_ranges): Test can_float_p
26724         against CODE_FOR_nothing.
26726 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26727             Richard Biener  <rguenther@suse.de>
26729         PR middle-end/55851
26730         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26731         types instead of just INTEGER_TYPE types.
26733 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26735         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26736         TF_SIZE): Define.
26738 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26740         PR target/42661
26741         * config/mips/mips.opt: Change mad to mmad to match documentation.
26743 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26745         PR target/55897
26746         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26747         .progmemx.data now.
26749 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26751         PR target/55897
26752         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26753         (avr_addrspace_t): Add .section_name field.
26754         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26755         array size.
26756         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26757         NULL entry.  Put __memx into .progmemx.data.
26758         (progmem_section_prefix): Remove.
26759         (avr_asm_init_sections): No need to initialize progmem_section.
26760         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26761         section name prefix.
26762         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26763         retrieve the progmem section.
26764         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26765         boundary to run over avr_addrspace[].
26766         (avr_register_target_pragmas): Ditto.
26768 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26770         * varasm.c (output_constant_def_contents): For asan_protect_global
26771         protected strings, adjust DECL_ALIGN if needed, before testing for
26772         anchored symbols.
26773         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26774         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26775         normal decls.
26776         (output_object_block): For asan protected decls, emit asan padding
26777         after their contents.
26778         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26779         (asan_finish_file): Test it here instead.
26781 2013-01-07  Nick Clifton  <nickc@redhat.com>
26782             Matthias Klose  <doko@debian.org>
26783             Doug Kwan  <dougkwan@google.com>
26784             H.J. Lu  <hongjiu.lu@intel.com>
26786         PR driver/55470
26787         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26789         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26791         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26793         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26795         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26797 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26799         PR target/54461
26800         * doc/install.texi (Cross-Compiler-Specific Options): Document
26801         --with-avrlibc.
26803 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26805         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26806         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26807         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26808         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26809         vqmovun_high_s64): Fix source operand number and update copyright.
26811 2013-01-07  Richard Biener  <rguenther@suse.de>
26813         PR middle-end/55890
26814         * gimple.h (gimple_call_builtin_p): New overload.
26815         * gimple.c (validate_call): New function.
26816         (gimple_call_builtin_p): Likewise.
26817         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26818         Use gimple_call_builtin_p.
26819         (find_func_clobbers): Likewise.
26820         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26821         (strlen_optimize_stmt): Likewise.
26823 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26825         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26826         (vld1q_dup_*): Likewise.
26827         (vld1_*): Likewise.
26828         (vld1q_*): Likewise.
26829         (vld1_lane_*): Likewise.
26830         (vld1q_lane_*): Likewise.
26832 2013-01-07  Richard Biener  <rguenther@suse.de>
26834         * lto-streamer.h (LTO_minor_version): Bump to 2.
26836 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26838         * config/aarch64/aarch64-protos.h
26839         (aarch64_const_double_zero_rtx_p): Rename to...
26840         (aarch64_float_const_zero_rtx_p): ...this.
26841         (aarch64_float_const_representable_p): New.
26842         (aarch64_output_simd_mov_immediate): Likewise.
26843         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26844         move immediate case.
26845         * config/aarch64/aarch64.c
26846         (aarch64_const_double_zero_rtx_p): Rename to...
26847         (aarch64_float_const_zero_rtx_p): ...this.
26848         (aarch64_print_operand): Allow printing of new constants.
26849         (aarch64_valid_floating_const): New.
26850         (aarch64_legitimate_constant_p): Check for valid floating-point
26851         constants.
26852         (aarch64_simd_valid_immediate): Likewise.
26853         (aarch64_vect_float_const_representable_p): New.
26854         (aarch64_float_const_representable_p): Likewise.
26855         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26856         (aarch64_output_simd_mov_immediate): New.
26857         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26858         (*movdf_aarch64): Likewise.
26859         * config/aarch64/constraints.md (Ufc): New.
26860         (Y): call aarch64_float_const_zero_rtx.
26861         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26863 2013-01-07  Richard Biener  <rguenther@suse.de>
26865         PR tree-optimization/55888
26866         PR tree-optimization/55862
26867         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26868         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26869         not if it is contained therein.
26871 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26873         * config/avr/t-avr: Typo.
26875 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26877         PR55243
26878         * config/avr/t-avr: Don't automatically rebuild
26879         $(srcdir)/config/avr/t-multilib
26880         $(srcdir)/config/avr/avr-tables.opt
26881         $(srcdir)/doc/avr-mmcu.texi
26882         (avr-mcus): New phony target to build them on request.
26883         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26884         * avr/avr-mcus.def: Adjust comments.
26886 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26888         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26890 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26892         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26894 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26896         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26898 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26900         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26901         to generate profiling.
26902         * config/rs6000/aix64.h (LIB_SPEC): Same.
26904 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26906         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26907         New function.
26908         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26910 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26912         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26913         unconditionally.
26914         (ix86_expand_move): Ditto.
26915         (ix86_zero_extend_to_Pmode): Ditto.
26916         (ix86_expand_call): Ditto.
26917         (ix86_expand_special_args_builtin): Ditto.
26918         (ix86_expand_builtin): Ditto.
26920 2013-01-04  Richard Biener  <rguenther@suse.de>
26922         PR tree-optimization/55862
26923         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26924         translating them through PHI nodes.
26926 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26928         PR tree-optimization/55755
26929         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26930         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26932 2013-01-04  Marek Polacek  <polacek@redhat.com>
26934         PR middle-end/55859
26935         * opts.c (default_options_optimization): Clarify error message.
26937 2013-01-04  Richard Biener  <rguenther@suse.de>
26939         PR middle-end/55863
26940         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26941         reassociation.
26943 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26945         PR target/53789
26946         * config/pa/pa.md (movsi): Revert previous change.
26947         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
26948         references.
26950 2013-01-03  Richard Henderson  <rth@redhat.com>
26952         * config/i386/i386.c (ix86_expand_move): Always assign to op1
26953         after eliminating TLS symbols.
26955 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
26957         PR bootstrap/50167
26958         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
26959         * graphite-poly.c (debug_gmp_value): Likewise.
26961 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
26963         PR target/55712
26964         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
26965         selected code model, define __code_mode_small__, __code_model_medium__,
26966         __code_model_large__, __code_model_32__ or __code_model_kernel__.
26967         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
26968         xchg temporary register with %k.  Declare temporary register as
26969         early clobbered.
26970         [__x86_64__]: For medium and large code models, preserve %rbx register.
26972 2013-01-03  Richard Biener  <rguenther@suse.de>
26974         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
26975         (dump_subscript): Adjust.
26976         (finalize_ddr_dependent): Do not dump redundant info.
26977         (analyze_siv_subscript): Adjust.
26978         (subscript_dependence_tester): Likewise.
26979         (compute_affine_dependence): Likewise.
26981 2013-01-03  Richard Biener  <rguenther@suse.de>
26983         Revert
26984         2013-01-03  Richard Biener  <rguenther@suse.de>
26986         PR tree-optimization/55857
26987         * tree-vect-stmts.c (vectorizable_load): Do not setup
26988         re-alignment for invariant loads.
26990         2013-01-02  Richard Biener  <rguenther@suse.de>
26992         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
26993         invariant load do not generate a vector load from the scalar location.
26995 2013-01-03  Richard Biener  <rguenther@suse.de>
26997         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
26998         for not vectorizing.
26999         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27000         not build INDIRECT_REFs, call get_name once only.
27001         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27002         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27004 2013-01-03  Richard Biener  <rguenther@suse.de>
27006         PR tree-optimization/55857
27007         * tree-vect-stmts.c (vectorizable_load): Do not setup
27008         re-alignment for invariant loads.
27010 2013-01-03  Richard Biener  <rguenther@suse.de>
27012         PR lto/55848
27013         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27014         prefer a built-in decl.
27016 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27018         * gcc.c (process_command): Update copyright notice dates.
27019         * gcov.c (print_version): Likewise.
27020         * gcov-dump.c (print_version): Likewise.
27022         PR rtl-optimization/55838
27023         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27024         iv0.step, iv1.step and step.
27026 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27027             Marc Glisse  <marc.glisse@inria.fr>
27029         PR tree-optimization/55832
27030         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27031         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27032         integer_{one,zero}_node.
27034 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27036         PR debug/54402
27037         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27038         * var-tracking.c (reverse_op): Don't add reverse ops to
27039         VALUEs that have already
27040         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27042 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27044         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27046 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27048         * dumpfile.c (dump_loc): Print filename with location.
27049         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27050         new location_t parameter to emit complete unroll message with
27051         new dump framework.
27052         (canonicalize_loop_induction_variables): Compute loops location
27053         and pass to try_unroll_loop_completely.
27054         * loop-unroll.c (report_unroll_peel): New function.
27055         (peel_loops_completely): Use new dump format with location
27056         for main dumpfile message, and invoke report_unroll_peel on success.
27057         (decide_unrolling_and_peeling): Ditto.
27058         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27059         by report_unroll_peel.
27060         (decide_peel_completely): Ditto.
27061         (decide_unroll_constant_iterations): Ditto.
27062         (decide_unroll_runtime_iterations): Ditto.
27063         (decide_peel_simple): Ditto.
27064         (decide_unroll_stupid): Ditto.
27065         * cfgloop.c (get_loop_location): New function.
27066         * cfgloop.h (get_loop_location): Declare.
27068 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27070         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27071         NULL.
27073 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27075         PR middle-end/55198
27076         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27077         BLKmode objects when EXPAND_MEMORY is specified.
27079 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27081         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27082         in loop predicate.
27083         (fold_builtin_cpu): Do not share cpu model decls across statements.
27085 2013-01-02  Jason Merrill  <jason@redhat.com>
27087         PR c++/55804
27088         * tree.c (build_array_type_1): Revert earlier change.
27090 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27092         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27093         "cortex-a57".
27094         * config/aarch64/aarch64-tune.md: Re-generate.
27096 2013-01-02  Richard Biener  <rguenther@suse.de>
27098         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27099         invariant load do not generate a vector load from the scalar location.
27101 2013-01-02  Richard Biener  <rguenther@suse.de>
27103         PR bootstrap/55784
27104         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27105         * configure: Regenerate.
27107 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27109         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27110         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27111         (expand_builtin_int_roundingfn_2): Keep the original target around
27112         for the fallback case.
27114 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27116         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27117         to be clear for sign changes.
27119 2013-01-01  Jan Hubicka  <jh@suse.cz>
27121         * ipa-inline-analysis.c: Fix formatting.
27123 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27125         PR tree-optimization/55831
27126         * tree-vect-loop.c (get_initial_def_for_induction): Use
27127         gsi_after_labels instead of gsi_start_bb.
27129 Copyright (C) 2013 Free Software Foundation, Inc.
27131 Copying and distribution of this file, with or without modification,
27132 are permitted in any medium without royalty provided the copyright
27133 notice and this notice are preserved.