re PR middle-end/58809 (ICE with complex variable in OpenMP reduction clause)
[official-gcc.git] / gcc / ChangeLog
blobd00105df9fbaecbb8d76bc45841a0e52bcf3a77d
1 2013-10-21  Marek Polacek  <polacek@redhat.com>
3         PR middle-end/58809
4         * fold-const.c (fold_range_test): Return 0 if the type is not
5         an integral type.
7 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
9         * system.h: Move hwint.h include further down.
10         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
11         gcc_checking_asserts.
12         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
14 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
16         Fix volatile issues in optimize_bit_field_compare.
17         * fold-const.c (optimize_bit_field_compare): Bail out if
18         lvolatilep or rvolatilep.
20 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
22         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
23         and get_inner_reference returning different pmode for non-volatile
24         bit-field members dependent on flag_strict_volatile_bitfields.
25         * stor-layout.c (layout_decl): Remove special handling of
26         flag_strict_volatile_bitfields.
27         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
28         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
30 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
32         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
33         calculations.
35 2013-10-21  Jeff Law  <law@redhat.com>
37         * tree-ssa-threadedge.c (thread_through_normal_block): New argument VISITED.
38         Remove VISISTED as a local variable.  When we have a threadable jump, verify
39         the destination of the jump has not been visised.
40         (thread_across_edge): Allocate VISITED bitmap once at function scope and
41         use it throughout.  Make sure to set appropriate bits in VISITED for E (start
42         of jump thread path).
44         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading through
45         a joiner if any edge on the path has a recorded jump thread.
47 2013-10-21  Ian Lance Taylor  <iant@google.com>
49         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
50         don't imply that attributes can solve all problems.
51         (Directory Options): Fix typo.
53 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
55         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
56         extend and extend_add.
58 2013-10-21  Richard Biener  <rguenther@suse.de>
60         PR tree-optimization/58794
61         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
62         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
64 2013-10-21  Richard Biener  <rguenther@suse.de>
66         PR middle-end/58742
67         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
68         to (T) X for sign-changing conversions (or no conversion).
70 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
72         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
74 2013-10-20  Jan Hubicka  <jh@suse.cz>
76         * config/i386/i386-tune.def: Add comment; organize into categories
78 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
80         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
81         argument.  Update function comment.
82         (expand_set_or_movmem_via_rep): New function combining
83         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
84         (expand_movmem_via_rep_mov): Remove.
85         expand_setmem_via_rep_stos): Remove.
86         (expand_movmem_epilogue): Update calls correspondingly.
87         (expand_setmem_epilogue_via_loop): Likewise.
88         (emit_memset): New.
89         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
90         (expand_set_or_movmem_prologue): New function combining
91         expand_movmem_prologue and expand_setmem_prologue.
92         (expand_movmem_prologue): Remove.
93         (expand_setmem_prologue): Remove.
94         (expand_set_or_movmem_constant_prologue): New function combining
95         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
96         (expand_constant_movmem_prologue): Remove.
97         (expand_constant_setmem_prologue): Remove.
98         (promote_duplicated_reg): Allow vector-const0 value.
99         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
100         and ix86_expand_setmem.
101         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
102         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
104 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
106         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
107         little endian.
108         (vec_unpacku_hi_v8hi): Likewise.
109         (vec_unpacku_lo_v16qi): Likewise.
110         (vec_unpacku_lo_v8hi): Likewise.
112 2013-10-20  Jan Hubicka  <jh@suse.cz>
114         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
115         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
116         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
118 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
120         * config/mips/mips.h (ISA_HAS_WSBH): Define.
121         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
122         constants.
123         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
125 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
127         PR target/58603
128         * system.h: Undef m_slot.
130 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
132         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
133         all elements for both endian flavors.
135 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
137         PR target/58792
138         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
139         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
140         and SI_REG for 64bit SYSV ABI targets.
142 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
144         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
145         to multi_reg_return.  Clarify that we are skipping USEs of multiple
146         return registers.  Use bool type where appropriate.
148 2013-10-18  Jan Hubicka  <jh@suse.cz>
150         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
151         for cold functions.
152         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
153         (X86_TUNE_PUSH_MEMORY): Likewise.
154         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
155         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
156         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
157         New.
158         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
159         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
160         Remove.
161         (ix86_option_override_internal): Update to use tune features instead
162         of variables.
164 2013-10-18  Cong Hou  <congh@google.com>
166         PR tree-optimization/58508
167         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
168         statement that contains data refs with zero-step.
170 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
172         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
173         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
174         * gimple-low.c (gimple_check_call_arg,
175         gimple_check_call_matching_types): Move to cgraph.c.
176         * gimple-low.h: Remove prototype.
177         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
178         Relocate from gimple-low.c.
179         * cgraph.h: Add prototype.  Don't include basic-block.h.
180         * gimplify.c: Add gimple-low to include list.
181         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
182         * tree-eh.c: Add gimple-low to include list.
183         * tree-nested.c: Likewise.
184         * cfgexpand.c: Add tree-ssa-address.h to include list.
185         * expr.c: Likewise.
186         * gimple-fold.c: Likewise.
187         * gimple-ssa-strength-reduction.c: Likewise.
188         * trans-mem.c: Likewise.
189         * tree-mudflap.c: Likewise.
190         * tree-ssa-loop-ivopts.c: Likewise.
191         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
192         (degenerate_phi_result): Move to tree-phinodes.c.
193         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
194         * tree-ssa-threadedge.c: Likewise.
195         * tree-vrp.c: Likewise.
196         * tree-phinodes.c (degenerate_phi_result): Relocate here.
197         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
198         * tree-phinodes.h (degenerate_phi_result): Add prototype.
199         * tree-ssa-copy.c: Include tree-ssa-dom.h.
200         * tree-ssa-forwprop.c: Likewise.
201         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
202         pass_data_cleanup_cfg_post_optimizing,
203         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
204         * tree-optimize.c: Delete File.
205         * graphite.c: Include tree-cfgcleanup.h.
206         * passes.c: Likewise.
207         * tree-cfg.c: Likewise.
208         * tree-profile.c: Likewise.
209         * tree-ssa-dse.c: Likewise.
210         * tree-ssa-loop-ivcanon.c: Likewise.
211         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
212         * tree-outof-ssa.c: Include sbitmap.h.
213         * tree-ssa-live.c: Likewise.
214         * tree-ssa-propagate.c: Likewise.
215         * tree-ssa-structalias.c: Likewise.
216         * tree-stdarg.c: Likewise.
217         * Makefile.in (OBJS): Delete tree-optimize.o.
218         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
219         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
220         * varasm.c: Include basic-block.h.
221         * cfgloop.h: Include function.h instead of basic-block.h
222         (bb_loop_depth): Move to cfgloop.c.
223         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
225 2013-10-18  Teresa Johnson  <tejohnson@google.com>
227         * predict.c (probably_never_executed): Compare frequency-based
228         count to number of training runs.
229         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
231 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
233         * config/arm/arm.c (cortexa9_extra_costs): New table.
234         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
236 2013-10-18  Jeff Law  <law@redhat.com>
238         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
240         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
241         (register_jump_thread): Add "registered_jump_thread" debug
242         counter support.
243         * dbgcnt.def (registered_jump_thread): New debug counter.
245 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
247         * config/rs6000/rs6000.c: Include cgraph.h.
249 2013-10-18  Teresa Johnson  <tejohnson@google.com>
251         * tree-ssa-tail-merge.c (replace_block_by): Update edge
252         weights during merging.
254 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
256         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
257         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
258         * tree-ssa-operands.h: Remove prototype.
259         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
260         * gimple.c (virtual_operand_p): Relocate from gimple.c.
261         * gimple.h: Add prototype.
262         * gimple-ssa.h: Include tree-ssa-operands.h.
263         * tree-dump.c: Add tree-cfg.h to include list.
264         * tree-ssa-alias.c: Add ipa-reference.h to include list.
265         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
266         * config/i386/i386.c: Don't include tree-flow.h.
267         * config/rs6000/rs6000.c: Likewise.
269 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
271         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
273 2013-10-18  Richard Biener  <rguenther@suse.de>
275         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
276         or TYPE_UNSIGNED of integral types.
277         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
278         NULL_TREE for zero-precision integral types.
280 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
282         * config/aarch64/arm_neon.h
283         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
285 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
287         * opth-gen.awk: Define target_flags_explicit.
289 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
291         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
292         fields to the reg_addr array that describes the valid addressing
293         mode for any register, general purpose registers, floating point
294         registers, and Altivec registers.
295         (FIRST_RELOAD_REG_CLASS): Likewise.
296         (LAST_RELOAD_REG_CLASS): Likewise.
297         (struct reload_reg_map_type): Likewise.
298         (reload_reg_map_type): Likewise.
299         (RELOAD_REG_VALID): Likewise.
300         (RELOAD_REG_MULTIPLE): Likewise.
301         (RELOAD_REG_INDEXED): Likewise.
302         (RELOAD_REG_OFFSET): Likewise.
303         (RELOAD_REG_PRE_INCDEC): Likewise.
304         (RELOAD_REG_PRE_MODIFY): Likewise.
305         (reg_addr): Likewise.
306         (mode_supports_pre_incdec_p): New helper functions to say whether
307         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
308         (mode_supports_pre_modify_p): Likewise.
309         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
310         print the valid address mode bits for each mode.
311         (rs6000_debug_print_mode): Likewise.
312         (rs6000_debug_reg_global): Likewise.
313         (rs6000_setup_reg_addr_masks): New function to set up the address
314         mask bits for each type.
315         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
316         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
317         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
318         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
319         PRE_MODIFY are supported.
320         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
321         registers, instead of {src,dest}_av_p.
322         (rs6000_print_options_internal): Tweak the debug output slightly.
324 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
326         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
327         isa attribute.
329 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
331         * tree-flow.h (struct omp_region): Move to omp-low.c.
332         Remove omp_ prototypes and variables.
333         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
334         (copy_var_decl): Relocate prototype from tree-flow.h.
335         * gimple.c (copy_var_decl): Relocate from omp-low.c.
336         * tree.h: Move prototype to omp-low.h.
337         * omp-low.h: New File.  Relocate prototypes here.
338         * omp-low.c (struct omp_region): Make local here.
339         (root_omp_region): Make static.
340         (copy_var_decl) Move to gimple.c.
341         (new_omp_region): Make static.
342         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
343         * tree-cfg.c: Include omp-low.h.
344         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
345         * gimplify.c: Include omp-low.h.
346         * tree-parloops.c: Likewise.
348 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
350         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
351         operands are in memory, prefer to force non-matched operand 1 to
352         the register.
354 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
356         PR target/58673
357         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
358         restrict TImode addresses to single indirect registers if both
359         -mquad-memory and -mvsx-timode are used.
360         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
361         we should emit load/store quad.  Remove using %y for quad memory
362         addresses.
364         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
365         constraints to allow load/store quad on machines where TImode is
366         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
367         for TImode to load integer constants.
369 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
371         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
373 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
375         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
376         handling of STACK_REG.
378 2013-10-17  Richard Biener  <rguenther@suse.de>
380         PR tree-optimization/58143
381         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
382         New function.
383         (rewrite_to_defined_overflow): Likewise.
384         (move_computations_dom_walker::before_dom): Rewrite stmts
385         with undefined signed overflow that are not always executed
386         into unsigned arithmetic.
388 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
390         PR target/57756
391         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
392         explicit isa flag to be an options variable, instead of using
393         global_options_set.  Remove define from rs6000.h.
394         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
396         * config/rs6000/rs6000.c (rs6000_option_override_internal):
397         Initialize rs6000_isa_flags_explicit.
398         (rs6000_function_specific_save): Add gcc_options* parameter, so
399         that the powerpc builds after the 2013-10-15 changes.
400         (rs6000_function_specific_restore): Likewise.
402 2013-10-16  DJ Delorie  <dj@redhat.com>
404         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
405         op is a REG before checking REGNO.
406         (rl78_alloc_physical_registers): Verify pattern is a SET before
407         checking SET_SRC.
409 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
411         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
412         endianness.
413         (vec_unpacks_lo_v4sf): Likewise.
414         (vec_unpacks_float_hi_v4si): Likewise.
415         (vec_unpacks_float_lo_v4si): Likewise.
416         (vec_unpacku_float_hi_v4si): Likewise.
417         (vec_unpacku_float_lo_v4si): Likewise.
419 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
421         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
422         (vsx_concat_v2sf): Likewise.
424 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
426         * config/aarch64/aarch64.md
427         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
429 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
431         PR tree-optimization/58697
432         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
433         estimated_loop_iterations_int.
434         (max_stmt_executions_int): Call get_max_loop_iterations_int.
435         (get_max_loop_iterations_int): New.  HWINT version of
436         get_max_loop_iterations.
437         * cfgloop.h: Add prototypes.
438         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
439         * loop-unroll.c (decide_peel_once_rolling): Call
440         get_estimated_loop_iterations_int.
441         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
442         * tree-ssa-loop-niter.h: Tweak prototypes.
444 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
446         * gengtype-parse.c (struct_field_seq): Ignore access-control
447         keywords ("public:" etc).
449 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
451         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
452         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
454 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
456         * config/arm/arm.opt (mlra): New option.
457         * config/arm/arm.c (arm_lra_p): New function.
458         (TARGET_LRA_P): Define.
460 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
462         * tree-core.h (tree_code_name): Remove.
463         * tree.h (get_tree_code_name): New prototype.
464         * tree.c (tree_code_name): Make static.
465         (get_tree_code_name): New function.
466         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
467         tree_class_check_failed, tree_range_check_failed,
468         tree_not_class_check_failed, omp_clause_check_failed,
469         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
470         wrapper get_tree_code_name instead of calling tree_code_name directly.
471         * tree-vrp.c (dump_asserts_for): Likewise.
472         * tree-dump.c (dequeue_and_dump): Likewise.
473         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
474         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
475         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
476         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
477         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
478         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
479         dump_gimple_omp_for): Likewise.
480         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
481         * tree-ssa-pre.c (print_pre_expr): Likewise.
482         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
483         * print-tree.c (print_node_brief, print_node): Likewise.
484         * gimple.c (gimple_check_failed): Likewise.
485         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
486         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
487         * config/mep/mep.c (mep_validate_vliw): Likewise.
488         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
489         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
491 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
493         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
494         for AMD bdver3.
496 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
498         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
499         (MULTILIB_MATCHES): Add multilib for -march=v8.
501 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
503         PR target/57756
504         * optc-save-gen.awk: Add extra parameter to the save and restore
505         target calls.
506         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
507         * tree.c (build_optimization_node): New parameter.  Add extra parameter
508         to call to cl_optimization_save.
509         (build_target_option_node): New parameter. Add extra parameter
510         to call to cl_target_option_save.
511         * tree.h (build_optimization_node): New parameter.
512         (build_target_option_node): New parameter.
513         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
514         build_optimization_node and build_target_option_node.
515         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
516         (handle_pragma_push_options): Ditto.
517         * toplev.c (process_options): Ditto.
518         * opts.c (init_options_struct): Check for opts_set non-null.
519         * target.def (target_option.save): New parameter.
520         (target_option.restore): New parameter.
521         * tm.texi: Generate.
522         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
523         (ix86_pragma_target_parse): Ditto.
524         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
525         parameters.
526         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
527         to  build_optimization_node and build_target_option_node.
528         (rs6000_valid_attribute_p): Ditto.
529         (rs6000_pragma_target_parse): Ditto.
530         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
531         data.
532         * config/i386/i386.h:
533         TARGET_64BIT_P: New Macro
534         TARGET_MMX_P: New Macro.
535         TARGET_3DNOW_P: New Macro.
536         TARGET_3DNOW_A_P: New Macro.
537         TARGET_SSE_P: New Macro.
538         TARGET_SSE2_P: New Macro.
539         TARGET_SSE3_P: New Macro.
540         TARGET_SSSE3_P: New Macro.
541         TARGET_SSE4_1_P: New Macro.
542         TARGET_SSE4_2_P: New Macro.
543         TARGET_AVX_P: New Macro.
544         TARGET_AVX2_P: New Macro.
545         TARGET_AVX512F_P: New Macro.
546         TARGET_AVX512PF_P: New Macro.
547         TARGET_AVX512ER_P: New Macro.
548         TARGET_AVX512CD_P: New Macro.
549         TARGET_FMA_P: New Macro.
550         TARGET_SSE4A_P: New Macro.
551         TARGET_FMA4_P: New Macro.
552         TARGET_XOP_P: New Macro.
553         TARGET_LWP_P: New Macro.
554         TARGET_ABM_P: New Macro.
555         TARGET_BMI_P: New Macro.
556         TARGET_BMI2_P: New Macro.
557         TARGET_LZCNT_P: New Macro.
558         TARGET_TBM_P: New Macro.
559         TARGET_POPCNT_P: New Macro.
560         TARGET_SAHF_P: New Macro.
561         TARGET_MOVBE_P: New Macro.
562         TARGET_CRC32_P: New Macro.
563         TARGET_AES_P: New Macro.
564         TARGET_PCLMUL_P: New Macro.
565         TARGET_CMPXCHG16B_P: New Macro.
566         TARGET_FSGSBASE_P: New Macro.
567         TARGET_RDRND_P: New Macro.
568         TARGET_F16C_P: New Macro.
569         TARGET_RTM_P: New Macro.
570         TARGET_HLE_P: New Macro.
571         TARGET_RDSEED_P: New Macro.
572         TARGET_PRFCHW_P: New Macro.
573         TARGET_ADX_P: New Macro.
574         TARGET_FXSR_P: New Macro.
575         TARGET_XSAVE_P: New Macro.
576         TARGET_XSAVEOPT_P: New Macro.
577         TARGET_LP64_P: New Macro.
578         TARGET_X32_P: New Macro.
579         TARGET_FPMATH_DEFAULT_P: New Macro.
580         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
581         * config/i386/i386.c (ix86_option_override_internal): New parameters.
582         opts and opts_set.
583         Change ix86_tune_string to access opts->x_ix86_tune_string.
584         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
585         Change ix86_arch_string to access opts->x_ix86_arch_string.
586         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
587         Change ix86_pmode to access opts->x_ix86_pmode.
588         Change ix86_abi to access opts->x_ix86_abi.
589         Change ix86_cmodel to access opts->x_ix86_cmodel.
590         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
591         Change ix86_isa_flags_explicit to access
592         opts->x_ix86_isa_flags_explicit.
593         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
594         Change ix86_regparm to access opts->x_ix86_regparm.
595         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
596         Change ix86_preferred_stack_boundary_arg to access
597         opts->x_ix86_preferred_stack_boundary_arg.
598         Change ix86_force_align_arg_pointer to access
599         opts->x_ix86_force_align_arg_pointer.
600         Change ix86_incoming_stack_boundar_arg to access
601         opts->x_ix86_incoming_stack_boundar_arg.
602         Change ix86_fpmath to access opts->x_ix86_fpmath.
603         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
604         Change ix86_recip_name to access opts->x_ix86_recip_name.
605         Change ix86_stack_protector_guard to access
606         opts->x_ix86_stack_protector_guard.
607         Change ix86_tune_memcpy_strategy to access
608         opts->x_ix86_tune_memcpy_strategy.
609         Change ix86_tune_memset_strategy to access
610         opts->x_ix86_tune_memset_strategy.
611         Change global_options to access opts.
612         Change global_options_set to access opts_set.
613         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
614         Change TARGET_MMX to TARGET_MMX_P (opts->...).
615         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
616         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
617         Change TARGET_SSE to TARGET_SSE_P (opts->...).
618         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
619         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
620         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
621         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
622         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
623         Change TARGET_AVX to TARGET_AVX_P (opts->...).
624         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
625         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
626         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
627         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
628         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
629         Change TARGET_FMA to TARGET_FMA_P (opts->...).
630         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
631         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
632         Change TARGET_XOP to TARGET_XOP_P (opts->...).
633         Change TARGET_LWP to TARGET_LWP_P (opts->...).
634         Change TARGET_ABM to TARGET_ABM_P (opts->...).
635         Change TARGET_BMI to TARGET_BMI_P (opts->...).
636         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
637         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
638         Change TARGET_TBM to TARGET_TBM_P (opts->...).
639         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
640         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
641         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
642         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
643         Change TARGET_AES to TARGET_AES_P (opts->...).
644         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
645         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
646         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
647         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
648         Change TARGET_F16C to TARGET_F16C_P (opts->...).
649         Change TARGET_RTM to TARGET_RTM_P (opts->...).
650         Change TARGET_HLE to TARGET_HLE_P (opts->...).
651         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
652         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
653         Change TARGET_ADX to TARGET_ADX_P (opts->...).
654         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
655         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
656         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
657         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
658         Change TARGET_X32 to TARGET_X32_P (opts->...).
659         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
660         Change TARGET_FLOAT_RETURNS_IN_80387 to
661         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
662         (ix86_function_specific_save): New parameter. Use opts-> fields
663         to replace global fields.
664         (ix86_function_specific_restore): Ditto.
665         (ix86_valid_target_attribute_inner_p): New parameters.
666         Fix recursive call.
667         Fix call to ix86_handle_option and set_option.
668         (ix86_valid_target_attribute_tree): New parameters.
669         Change global_options to access opts.
670         Change global_options_set to access opts_set.
671         Fix call to ix86_valid_target_attribute_inner_p.
672         Change ix86_tune_string to access opts->x_ix86_tune_string.
673         Change ix86_arch_string to access opts->x_ix86_arch_string.
674         Change ix86_fpmath to access opts->x_ix86_fpmath
675         Fix call to ix86_option_override_internal.
676         Fix call to ix86_add_new_builtins.
677         Fix calls to build_optimization_node and build_target_option_node.
678         (ix86_valid_target_attribute_p): Remove access to global_options.
679         Use new gcc_options structure func_options.
680         Fix call to ix86_valid_target_attribute_tree.
681         Fix call to  build_optimization_node.
682         (get_builtin_code_for_version): Fix call to
683         ix86_valid_target_attribute_tree.
685 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
687         * Makefile.in (PICFLAG): New.
688         (enable_host_shared): New.
689         (INTERNAL_CFLAGS): Use PICFLAG.
690         (LIBIBERTY): Use pic build of libiberty.a if configured with
691         --enable-host-shared.
692         * configure.ac: Add --enable-host-shared, setting up new
693         PICFLAG variable.
694         * configure: Regenerate.
695         * doc/install.texi (--enable-shared): Add note contrasting it with ...
696         (--enable-host-shared): New option.
698 2013-10-15  Richard Biener  <rguenther@suse.de>
700         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
701         for built-in functions.
703 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
705         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
706         (optimize_range_tests_1): New function,
707         extracted from optimize_range_tests.
708         (optimize_range_tests_xor): Similarly.
709         (optimize_range_tests_diff): New function.
710         (optimize_range_tests): Use optimize_range_tests_1.
712 2013-10-15  Cong Hou  <congh@google.com>
714         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
715         requirement of the reduction pattern so that one operand of the
716         reduction operation can come from outside of the loop.
718 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
720         * config/arm/neon-schedgen.ml: Remove.
721         * config/arm/cortex-a9-neon.md: Remove comment regarding
722         neon-schedgen.ml.
724 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
726         * config/arm/types: Remove old neon types.
728 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
730         * config/arm/cortex-a7.md
731         (cortex_a7_neon_type): New.
732         (cortex_a7_neon_mul): Update for new types.
733         (cortex_a7_neon_mla): Likewise.
734         (cortex_a7_neon): Likewise.
736 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
738         * config/arm/cortex-a15-neon.md
739         (cortex_a15_neon_type): New,
741         (cortex_a15_neon_int_1): Remove.
742         (cortex_a15_neon_int_2): Likewise.
743         (cortex_a15_neon_int_3): Likewise.
744         (cortex_a15_neon_int_4): Likewise.
745         (cortex_a15_neon_int_5): Likewise.
746         (cortex_a15_neon_vqneg_vqabs): Likewise.
747         (cortex_a15_neon_vmov): Likewise.
748         (cortex_a15_neon_vaba): Likewise.
749         (cortex_a15_neon_vaba_qqq): Likewise.
750         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
751         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
752         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
753         Likewise.
754         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
755         (cortex_a15_neon_mla_qqq_8_16): Likewise.
756         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
757         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
758         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
759         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
760         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
761         (cortex_a15_neon_shift_1): Likewise.
762         (cortex_a15_neon_shift_2): Likewise.
763         (cortex_a15_neon_shift_3): Likewise.
764         (cortex_a15_neon_vshl_ddd): Likewise.
765         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
766         (cortex_a15_neon_vsra_vrsra): Likewise.
767         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
768         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
769         (cortex_a15_neon_bp_3cycle): Likewise.
770         (cortex_a15_neon_ldm_2): Likewise.
771         (cortex_a15_neon_stm_2): Likewise.
772         (cortex_a15_neon_mcr): Likewise.
773         (cortex_a15_neon_mrc): Likewise.
774         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
775         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
776         (cortex_a15_neon_fp_vmul_ddd): Likewise.
777         (cortex_a15_neon_fp_vmul_qqd): Likewise.
778         (cortex_a15_neon_fp_vmla_ddd): Likewise.
779         (cortex_a15_neon_fp_vmla_qqq): Likewise.
780         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
781         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
782         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
783         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
784         (cortex_a15_neon_bp_simple): Likewise.
785         (cortex_a15_neon_bp_2cycle): Likewise.
786         (cortex_a15_neon_bp_3cycle): Likewise.
787         (cortex_a15_neon_vld1_1_2_regs): Likewise.
788         (cortex_a15_neon_vld1_3_4_regs): Likewise.
789         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
790         (cortex_a15_neon_vld2_4_regs): Likewise.
791         (cortex_a15_neon_vld3_vld4): Likewise.
792         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
793         (cortex_a15_neon_vst1_3_4_regs): Likewise.
794         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
795         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
796         (cortex_a15_neon_vst3_vst4): Rename to...
797         (cortex_a15_neon_vst4): This, update for new attributes.
798         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
799         (cortex_a15_neon_vld3_vld4_lane): Likewise.
800         (cortex_a15_neon_vst1_vst2_lane): Likewise.
801         (cortex_a15_neon_vst3_vst4_lane): Likewise.
802         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
803         (cortex_a15_neon_ldm_2): Likewise.
804         (cortex_a15_neon_stm_2): Likewise.
805         (cortex_a15_neon_mcr): Likewise.
806         (cortex_a15_neon_mcr_2_mcrr): Likewise.
807         (cortex_a15_neon_mrc): Likewise.
808         (cortex_a15_neon_mrrc): Likewise.
810         (cortex_a15_neon_abd): New.
811         (cortex_a15_neon_abd_q): Likewise.
812         (cortex_a15_neon_aba): Likewise.
813         (cortex_a15_neon_aba_q): Likewise.
814         (cortex_a15_neon_acc): Likewise.
815         (cortex_a15_neon_acc_q): Likewise.
816         (cortex_a15_neon_arith_basic): Likewise.
817         (cortex_a15_neon_arith_complex): Likewise.
818         (cortex_a15_neon_multiply): Likewise.
819         (cortex_a15_neon_multiply_q): Likewise.
820         (cortex_a15_neon_mla): Likewise.
821         (cortex_a15_neon_mla_q): Likewise.
822         (cortex_a15_neon_sat_mla_long): Likewise.
823         (cortex_a15_neon_shift_acc): Likewise.
824         (cortex_a15_neon_shift_imm_basic): Likewise.
825         (cortex_a15_neon_shift_imm_complex): Likewise.
826         (cortex_a15_neon_shift_reg_basic): Likewise.
827         (cortex_a15_neon_shift_reg_basic_q): Likewise.
828         (cortex_a15_neon_shift_reg_complex): Likewise.
829         (cortex_a15_neon_shift_reg_complex_q): Likewise.
830         (cortex_a15_neon_fp_negabs): Likewise
831         (cortex_a15_neon_fp_arith): Likewise
832         (cortex_a15_neon_fp_arith_q): Likewise
833         (cortex_a15_neon_fp_cvt_int): Likewise
834         (cortex_a15_neon_fp_cvt_int_q): Likewise
835         (cortex_a15_neon_fp_cvt_16): Likewise
836         (cortex_a15_neon_fp_mul): Likewise
837         (cortex_a15_neon_fp_mul_q): Likewise
838         (cortex_a15_neon_fp_mla): Likewise
839         (cortex_a15_neon_fp_mla_q): Likewise
840         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
841         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
842         (cortex_a15_neon_bitops): Likewise.
843         (cortex_a15_neon_bitops_q): Likewise.
844         (cortex_a15_neon_from_gp): Likewise.
845         (cortex_a15_neon_from_gp_q): Likewise.
846         (cortex_a15_neon_tbl3_tbl4): Likewise.
847         (cortex_a15_neon_zip_q): Likewise.
848         (cortex_a15_neon_to_gp): Likewise.
849         (cortex_a15_neon_load_a): Likewise.
850         (cortex_a15_neon_load_b): Likewise.
851         (cortex_a15_neon_load_c): Likewise.
852         (cortex_a15_neon_load_d): Likewise.
853         (cortex_a15_neon_load_e): Likewise.
854         (cortex_a15_neon_load_f): Likewise.
855         (cortex_a15_neon_store_a): Likewise.
856         (cortex_a15_neon_store_b): Likewise.
857         (cortex_a15_neon_store_c): Likewise.
858         (cortex_a15_neon_store_d): Likewise.
859         (cortex_a15_neon_store_e): Likewise.
860         (cortex_a15_neon_store_f): Likewise.
861         (cortex_a15_neon_store_g): Likewise.
862         (cortex_a15_neon_store_h): Likewise.
863         (cortex_a15_vfp_to_from_gp): Likewise.
865 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
867         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
869         (cortex_a9_neon_vshl_ddd): Remove.
870         (cortex_a9_neon_vst3_vst4): Likewise.
871         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
873         (cortex_a9_neon_bit_ops_q): New.
875         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
876         (cortex_a9_neon_int_2): Likewise.
877         (cortex_a9_neon_int_3): Likewise.
878         (cortex_a9_neon_int_4): Likewise.
879         (cortex_a9_neon_int_5): Likewise.
880         (cortex_a9_neon_vqneg_vqabs): Likewise.
881         (cortex_a9_neon_vmov): Likewise.
882         (cortex_a9_neon_vaba): Likewise.
883         (cortex_a9_neon_vaba_qqq): Likewise.
884         (cortex_a9_neon_shift_1): Likewise.
885         (cortex_a9_neon_shift_2): Likewise.
886         (cortex_a9_neon_shift_3): Likewise.
887         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
888         (cortex_a9_neon_vsra_vrsra): Likewise.
889         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
890         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
891         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
892         Likewise.
893         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
894         (cortex_a9_neon_mla_qqq_8_16): Likewise.
895         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
896         Likewise.
897         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
898         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
899         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
900         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
901         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
902         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
903         (cortex_a9_neon_fp_vsum): Likewise.
904         (cortex_a9_neon_fp_vmul_ddd): Likewise.
905         (cortex_a9_neon_fp_vmul_qqd): Likewise.
906         (cortex_a9_neon_fp_vmla_ddd): Likewise.
907         (cortex_a9_neon_fp_vmla_qqq): Likewise.
908         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
909         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
910         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
911         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
912         (cortex_a9_neon_bp_simple): Likewise.
913         (cortex_a9_neon_bp_2cycle): Likewise.
914         (cortex_a9_neon_bp_3cycle): Likewise.
915         (cortex_a9_neon_ldr): Likewise.
916         (cortex_a9_neon_str): Likewise.
917         (cortex_a9_neon_vld1_1_2_regs): Likewise.
918         (cortex_a9_neon_vld1_3_4_regs): Likewise.
919         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
920         (cortex_a9_neon_vld2_4_regs): Likewise.
921         (cortex_a9_neon_vld3_vld4): Likewise.
922         (cortex_a9_neon_vld1_vld2_lane): Likewise.
923         (cortex_a9_neon_vld3_vld4_lane): Likewise.
924         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
925         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
926         (cortex_a9_neon_vst1_3_4_regs): Likewise.
927         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
928         (cortex_a9_neon_vst1_vst2_lane): Likewise.
929         (cortex_a9_neon_vst3_vst4_lane): Likewise.
930         (cortex_a9_neon_mcr): Likewise.
931         (cortex_a9_neon_mcr_2_mcrr): Likewise.
932         (cortex_a9_neon_mrc): Likewise.
933         (cortex_a9_neon_mrrc): Likewise.
935 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
937         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
939         (cortex_a8_neon_vshl_ddd): Remove.
940         (cortex_a8_neon_vst3_vst4): Likewise.
941         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
943         (cortex_a8_neon_bit_ops_q): New.
945         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
946         (cortex_a8_neon_int_2): Likewise..
947         (cortex_a8_neon_int_3): Likewise.
948         (cortex_a8_neon_int_5): Likewise.
949         (cortex_a8_neon_vqneg_vqabs): Likewise.
950         (cortex_a8_neon_int_4): Likewise.
951         (cortex_a8_neon_vaba): Likewise.
952         (cortex_a8_neon_vaba_qqq): Likewise.
953         (cortex_a8_neon_shift_1): Likewise.
954         (cortex_a8_neon_shift_2): Likewise.
955         (cortex_a8_neon_shift_3): Likewise.
956         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
957         (cortex_a8_neon_vsra_vrsra): Likewise.
958         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
959         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
960         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
961         Likewise.
962         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
963         (cortex_a8_neon_mla_qqq_8_16): Likewise.
964         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
965         Likewise.
966         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
967         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
968         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
969         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
970         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
971         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
972         (cortex_a8_neon_fp_vsum): Likewise.
973         (cortex_a8_neon_fp_vmul_ddd): Likewise.
974         (cortex_a8_neon_fp_vmul_qqd): Likewise.
975         (cortex_a8_neon_fp_vmla_ddd): Likewise.
976         (cortex_a8_neon_fp_vmla_qqq): Likewise.
977         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
978         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
979         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
980         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
981         (cortex_a8_neon_bp_simple): Likewise.
982         (cortex_a8_neon_bp_2cycle): Likewise.
983         (cortex_a8_neon_bp_3cycle): Likewise.
984         (cortex_a8_neon_ldr): Likewise.
985         (cortex_a8_neon_str): Likewise.
986         (cortex_a8_neon_vld1_1_2_regs): Likewise.
987         (cortex_a8_neon_vld1_3_4_regs): Likewise.
988         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
989         (cortex_a8_neon_vld2_4_regs): Likewise.
990         (cortex_a8_neon_vld3_vld4): Likewise.
991         (cortex_a8_neon_vld1_vld2_lane): Likewise.
992         (cortex_a8_neon_vld3_vld4_lane): Likewise.
993         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
994         (cortex_a8_neon_vst1_3_4_regs): Likewise.
995         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
996         (cortex_a8_neon_vst1_vst2_lane): Likewise.
997         (cortex_a8_neon_vst3_vst4_lane): Likewise.
998         (cortex_a8_neon_mcr): Likewise.
999         (cortex_a8_neon_mcr_2_mcrr): Likewise.
1000         (cortex_a8_neon_mrc): Likewise.
1001         (cortex_a8_neon_mrrc): Likewise.
1003 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1005         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
1006         (fp): New.
1007         * config/aarch64/aarch64-simd.md (neon_type): Remove.
1008         (aarch64_simd_dup<mode>): Add "type" attribute.
1009         (aarch64_dup_lane<mode>): Likewise.
1010         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
1011         (*aarch64_simd_mov<mode>): Likewise.
1012         (aarch64_simd_mov_from_<mode>low): Likewise.
1013         (aarch64_simd_mov_from_<mode>high): Likewise.
1014         (orn<mode>3): Likewise.
1015         (bic<mode>3): Likewise.
1016         (add<mode>3): Likewise.
1017         (sub<mode>3): Likewise.
1018         (mul<mode>3): Likewise.
1019         (*aarch64_mul3_elt<mode>): Likewise.
1020         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1021         (*aarch64_mul3_elt_to_128df): Likewise.
1022         (*aarch64_mul3_elt_to_64v2df): Likewise.
1023         (neg<mode>2): Likewise.
1024         (abs<mode>2): Likewise.
1025         (abd<mode>_3): Likewise.
1026         (aba<mode>_3): Likewise.
1027         (fabd<mode>_3): Likewise.
1028         (*fabd_scalar<mode>3): Likewise.
1029         (and<mode>3): Likewise.
1030         (ior<mode>3): Likewise.
1031         (xor<mode>3): Likewise.
1032         (one_cmpl<mode>2): Likewise.
1033         (aarch64_simd_vec_set<mode>): Likewise.
1034         (aarch64_simd_lshr<mode>): Likewise.
1035         (aarch64_simd_ashr<mode>): Likewise.
1036         (aarch64_simd_imm_shl<mode>): Likewise.
1037         (aarch64_simd_reg_sshl<mode): Likewise.
1038         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
1039         (aarch64_simd_reg_shl<mode>_signed): Likewise.
1040         (aarch64_simd_vec_setv2di): Likewise.
1041         (aarch64_simd_vec_set<mode>): Likewise.
1042         (aarch64_mla<mode>): Likewise.
1043         (*aarch64_mla_elt<mode>): Likewise.
1044         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
1045         (aarch64_mls<mode>): Likewise.
1046         (*aarch64_mls_elt<mode>): Likewise.
1047         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
1048         (<su><maxmin><mode>3): Likewise.
1049         (move_lo_quad_<mode>): Likewise.
1050         (aarch64_simd_move_hi_quad_<mode>): Likewise.
1051         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
1052         (vec_pack_trunc_<mode>): Likewise.
1053         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
1054         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
1055         (*aarch64_<su>mlal_lo<mode>): Likewise.
1056         (*aarch64_<su>mlal_hi<mode>): Likewise.
1057         (*aarch64_<su>mlsl_lo<mode>): Likewise.
1058         (*aarch64_<su>mlsl_hi<mode>): Likewise.
1059         (*aarch64_<su>mlal<mode>): Likewise.
1060         (*aarch64_<su>mlsl<mode>): Likewise.
1061         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
1062         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
1063         (add<mode>3): Likewise.
1064         (sub<mode>3): Likewise.
1065         (mul<mode>3): Likewise.
1066         (div<mode>3): Likewise.
1067         (neg<mode>2): Likewise.
1068         (abs<mode>2): Likewise.
1069         (fma<mode>4): Likewise.
1070         (*aarch64_fma4_elt<mode>): Likewise.
1071         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
1072         (*aarch64_fma4_elt_to_128df): Likewise.
1073         (*aarch64_fma4_elt_to_64v2df): Likewise.
1074         (fnma<mode>4): Likewise.
1075         (*aarch64_fnma4_elt<mode>): Likewise.
1076         (*aarch64_fnma4_elt_<vswap_width_name><mode>
1077         (*aarch64_fnma4_elt_to_128df): Likewise.
1078         (*aarch64_fnma4_elt_to_64v2df): Likewise.
1079         (<frint_pattern><mode>2): Likewise.
1080         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
1081         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
1082         (vec_unpacks_lo_v4sf): Likewise.
1083         (aarch64_float_extend_lo_v2df): Likewise.
1084         (vec_unpacks_hi_v4sf): Likewise.
1085         (aarch64_float_truncate_lo_v2sf): Likewise.
1086         (aarch64_float_truncate_hi_v4sf): Likewise.
1087         (aarch64_vmls<mode>): Likewise.
1088         (<su><maxmin><mode>3): Likewise.
1089         (<maxmin_uns><mode>3): Likewise.
1090         (reduc_<sur>plus_<mode>): Likewise.
1091         (reduc_<sur>plus_v2di): Likewise.
1092         (reduc_<sur>plus_v2si): Likewise.
1093         (reduc_<sur>plus_<mode>): Likewise.
1094         (aarch64_addpv4sf): Likewise.
1095         (clz<mode>2): Likewise.
1096         (reduc_<maxmin_uns>_<mode>): Likewise.
1097         (reduc_<maxmin_uns>_v2di): Likewise.
1098         (reduc_<maxmin_uns>_v2si): Likewise.
1099         (reduc_<maxmin_uns>_<mode>): Likewise.
1100         (reduc_<maxmin_uns>_v4sf): Likewise.
1101         (aarch64_simd_bsl<mode>_internal): Likewise.
1102         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1103         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1104         (aarch64_get_lane<mode>): Likewise.
1105         (*aarch64_combinez<mode>): Likewise.
1106         (aarch64_combine<mode>): Likewise.
1107         (aarch64_simd_combine<mode>): Likewise.
1108         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
1109         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
1110         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
1111         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
1112         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
1113         (aarch64_<sur>h<addsub><mode>): Likewise.
1114         (aarch64_<sur><addsub>hn<mode>): Likewise.
1115         (aarch64_<sur><addsub>hn2<mode>): Likewise.
1116         (aarch64_pmul<mode>): Likewise.
1117         (aarch64_<su_optab><optab><mode>): Likewise.
1118         (aarch64_<sur>qadd<mode>): Likewise.
1119         (aarch64_sqmovun<mode>): Likewise.
1120         (aarch64_<sur>qmovn<mode>): Likewise.
1121         (aarch64_s<optab><mode>): Likewise.
1122         (aarch64_sq<r>dmulh<mode>): Likewise.
1123         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1124         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
1125         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
1126         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
1127         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1128         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
1129         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
1130         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
1131         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1132         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
1133         (aarch64_sqdmull<mode>): Likewise.
1134         (aarch64_sqdmull_lane<mode>_internal): Likewise.
1135         (aarch64_sqdmull_n<mode>): Likewise.
1136         (aarch64_sqdmull2<mode>_internal): Likewise.
1137         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1138         (aarch64_sqdmull2_n<mode>_internal): Likewise.
1139         (aarch64_<sur>shl<mode>): Likewise.
1140         (aarch64_<sur>q<r>shl<mode>
1141         (aarch64_<sur>shll_n<mode>): Likewise.
1142         (aarch64_<sur>shll2_n<mode>): Likewise.
1143         (aarch64_<sur>shr_n<mode>): Likewise.
1144         (aarch64_<sur>sra_n<mode>): Likewise.
1145         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
1146         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
1147         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
1148         (aarch64_cm<optab><mode>): Likewise.
1149         (aarch64_cm<optab>di): Likewise.
1150         (aarch64_cm<optab><mode>): Likewise.
1151         (aarch64_cm<optab>di): Likewise.
1152         (aarch64_cmtst<mode>): Likewise.
1153         (aarch64_cmtstdi): Likewise.
1154         (aarch64_cm<optab><mode>): Likewise.
1155         (*aarch64_fac<optab><mode>): Likewise.
1156         (aarch64_addp<mode>): Likewise.
1157         (aarch64_addpdi): Likewise.
1158         (sqrt<mode>2): Likewise.
1159         (vec_load_lanesoi<mode>): Likewise.
1160         (vec_store_lanesoi<mode>): Likewise.
1161         (vec_load_lanesci<mode>): Likewise.
1162         (vec_store_lanesci<mode>): Likewise.
1163         (vec_load_lanesxi<mode>): Likewise.
1164         (vec_store_lanesxi<mode>): Likewise.
1165         (*aarch64_mov<mode>): Likewise.
1166         (aarch64_ld2<mode>_dreg): Likewise.
1167         (aarch64_ld2<mode>_dreg): Likewise.
1168         (aarch64_ld3<mode>_dreg): Likewise.
1169         (aarch64_ld3<mode>_dreg): Likewise.
1170         (aarch64_ld4<mode>_dreg): Likewise.
1171         (aarch64_ld4<mode>_dreg): Likewise.
1172         (aarch64_tbl1<mode>): Likewise.
1173         (aarch64_tbl2v16qi): Likewise.
1174         (aarch64_combinev16qi): Likewise.
1175         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
1176         (aarch64_st2<mode>_dreg): Likewise.
1177         (aarch64_st2<mode>_dreg): Likewise.
1178         (aarch64_st3<mode>_dreg): Likewise.
1179         (aarch64_st3<mode>_dreg): Likewise.
1180         (aarch64_st4<mode>_dreg): Likewise.
1181         (aarch64_st4<mode>_dreg): Likewise.
1182         (*aarch64_simd_ld1r<mode>): Likewise.
1183         (aarch64_frecpe<mode>): Likewise.
1184         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
1185         (aarch64_frecps<mode>): Likewise.
1187 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1189         * config/arm/iterators.md (V_elem_ch): New.
1190         (q): Likewise.
1191         (VQH_type): Likewise.
1192         * config/arm/arm.md (is_neon_type): New.
1193         (conds): Use is_neon_type.
1194         (anddi3_insn): Update type attribute.
1195         (xordi3_insn): Likewise.
1196         (one_cmpldi2): Likewise.
1197         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
1198         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
1199         (*movmisalign<mode>_neon_store): Likewise.
1200         (*movmisalign<mode>_neon_load): Likewise.
1201         (vec_set<mode>_internal): Likewise.
1202         (vec_set<mode>_internal): Likewise.
1203         (vec_setv2di_internal): Likewise.
1204         (vec_extract<mode>): Likewise.
1205         (vec_extract<mode>): Likewise.
1206         (vec_extractv2di): Likewise.
1207         (*add<mode>3_neon): Likewise.
1208         (adddi3_neon): Likewise.
1209         (*sub<mode>3_neon): Likewise.
1210         (subdi3_neon): Likewise.
1211         (fma<VCVTF:mode>4): Likewise.
1212         (fma<VCVTF:mode>4_intrinsic): Likewise.
1213         (*fmsub<VCVTF:mode>4): Likewise.
1214         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
1215         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
1216         (ior<mode>3): Likewise.
1217         (and<mode>3): Likewise.
1218         (orn<mode>3_neon): Likewise.
1219         (orndi3_neon): Likewise.
1220         (bic<mode>3_neon): Likewise.
1221         (bicdi3_neon): Likewise.
1222         (xor<mode>3): Likewise.
1223         (one_cmpl<mode>2): Likewise.
1224         (abs<mode>2): Likewise.
1225         (neg<mode>2): Likewise.
1226         (negdi2_neon): Likewise.
1227         (*umin<mode>3_neon): Likewise.
1228         (*umax<mode>3_neon): Likewise.
1229         (*smin<mode>3_neon): Likewise.
1230         (*smax<mode>3_neon): Likewise.
1231         (vashl<mode>3): Likewise.
1232         (vashr<mode>3_imm): Likewise.
1233         (vlshr<mode>3_imm): Likewise.
1234         (ashl<mode>3_signed): Likewise.
1235         (ashl<mode>3_unsigned): Likewise.
1236         (neon_load_count): Likewise.
1237         (ashldi3_neon_noclobber): Likewise.
1238         (ashldi3_neon): Likewise.
1239         (signed_shift_di3_neon): Likewise.
1240         (unsigned_shift_di3_neon): Likewise.
1241         (ashrdi3_neon_imm_noclobber): Likewise.
1242         (lshrdi3_neon_imm_noclobber): Likewise.
1243         (<shift>di3_neon): Likewise.
1244         (widen_ssum<mode>3): Likewise.
1245         (widen_usum<mode>3): Likewise.
1246         (quad_halves_<code>v4si): Likewise.
1247         (quad_halves_<code>v4sf): Likewise.
1248         (quad_halves_<code>v8hi): Likewise.
1249         (quad_halves_<code>v16qi): Likewise.
1250         (reduc_splus_v2di): Likewise.
1251         (neon_vpadd_internal<mode>): Likewise.
1252         (neon_vpsmin<mode>): Likewise.
1253         (neon_vpsmax<mode>): Likewise.
1254         (neon_vpumin<mode>): Likewise.
1255         (neon_vpumax<mode>): Likewise.
1256         (*ss_add<mode>_neon): Likewise.
1257         (*us_add<mode>_neon): Likewise.
1258         (*ss_sub<mode>_neon): Likewise.
1259         (*us_sub<mode>_neon): Likewise.
1260         (neon_vadd<mode>_unspec): Likewise.
1261         (neon_vaddl<mode>): Likewise.
1262         (neon_vaddw<mode>): Likewise.
1263         (neon_vhadd<mode>): Likewise.
1264         (neon_vqadd<mode>): Likewise.
1265         (neon_vaddhn<mode>): Likewise.
1266         (neon_vmul<mode>): Likewise.
1267         (neon_vfms<VCVTF:mode>): Likewise.
1268         (neon_vmlal<mode>): Likewise.
1269         (neon_vmls<mode>): Likewise.
1270         (neon_vmlsl<mode>): Likewise.
1271         (neon_vqdmulh<mode>): Likewise.
1272         (neon_vqdmlal<mode>): Likewise.
1273         (neon_vqdmlsl<mode>): Likewise.
1274         (neon_vmull<mode>): Likewise.
1275         (neon_vqdmull<mode>): Likewise.
1276         (neon_vsub<mode>_unspec): Likewise.
1277         (neon_vsubl<mode>): Likewise.
1278         (neon_vsubw<mode>): Likewise.
1279         (neon_vqsub<mode>): Likewise.
1280         (neon_vhsub<mode>): Likewise.
1281         (neon_vsubhn<mode>): Likewise.
1282         (neon_vceq<mode>): Likewise.
1283         (neon_vcge<mode>): Likewise.
1284         (neon_vcgeu<mode>): Likewise.
1285         (neon_vcgt<mode>): Likewise.
1286         (neon_vcgtu<mode>): Likewise.
1287         (neon_vcle<mode>): Likewise.
1288         (neon_vclt<mode>): Likewise.
1289         (neon_vcage<mode>): Likewise.
1290         (neon_vcagt<mode>): Likewise.
1291         (neon_vtst<mode>): Likewise.
1292         (neon_vabd<mode>): Likewise.
1293         (neon_vabdl<mode>): Likewise.
1294         (neon_vaba<mode>): Likewise.
1295         (neon_vabal<mode>): Likewise.
1296         (neon_vmax<mode>): Likewise.
1297         (neon_vmin<mode>): Likewise.
1298         (neon_vpaddl<mode>): Likewise.
1299         (neon_vpadal<mode>): Likewise.
1300         (neon_vpmax<mode>): Likewise.
1301         (neon_vpmin<mode>): Likewise.
1302         (neon_vrecps<mode>): Likewise.
1303         (neon_vrsqrts<mode>): Likewise.
1304         (neon_vqabs<mode>): Likewise.
1305         (neon_vqneg<mode>): Likewise.
1306         (neon_vcls<mode>): Likewise.
1307         (clz<mode>2): Likewise.
1308         (popcount<mode>2): Likewise.
1309         (neon_vrecpe<mode>): Likewise.
1310         (neon_vrsqrte<mode>): Likewise.
1311         (neon_vget_lane<mode>_sext_internal): Likewise.
1312         (neon_vget_lane<mode>_zext_internal): Likewise.
1313         (neon_vdup_n<mode>): Likewise.
1314         (neon_vdup_n<mode>): Likewise.
1315         (neon_vdup_nv2di): Likewise.
1316         (neon_vdup_lane<mode>_interal): Likewise.
1317         (*neon_vswp<mode>): Likewise.
1318         (neon_vcombine<mode>): Likewise.
1319         (float<mode><V_cvtto>2): Likewise.
1320         (floatuns<mode><V_cvtto>2): Likewise.
1321         (fix_trunc<mode><V_cvtto>2): Likewise.
1322         (fixuns_trunc<mode><V_cvtto>2
1323         (neon_vcvt<mode>): Likewise.
1324         (neon_vcvt<mode>): Likewise.
1325         (neon_vcvtv4sfv4hf): Likewise.
1326         (neon_vcvtv4hfv4sf): Likewise.
1327         (neon_vcvt_n<mode>): Likewise.
1328         (neon_vcvt_n<mode>): Likewise.
1329         (neon_vmovn<mode>): Likewise.
1330         (neon_vqmovn<mode>): Likewise.
1331         (neon_vqmovun<mode>): Likewise.
1332         (neon_vmovl<mode>): Likewise.
1333         (neon_vmul_lane<mode>): Likewise.
1334         (neon_vmul_lane<mode>): Likewise.
1335         (neon_vmull_lane<mode>): Likewise.
1336         (neon_vqdmull_lane<mode>): Likewise.
1337         (neon_vqdmulh_lane<mode>): Likewise.
1338         (neon_vqdmulh_lane<mode>): Likewise.
1339         (neon_vmla_lane<mode>): Likewise.
1340         (neon_vmla_lane<mode>): Likewise.
1341         (neon_vmlal_lane<mode>): Likewise.
1342         (neon_vqdmlal_lane<mode>): Likewise.
1343         (neon_vmls_lane<mode>): Likewise.
1344         (neon_vmls_lane<mode>): Likewise.
1345         (neon_vmlsl_lane<mode>): Likewise.
1346         (neon_vqdmlsl_lane<mode>): Likewise.
1347         (neon_vext<mode>): Likewise.
1348         (neon_vrev64<mode>): Likewise.
1349         (neon_vrev32<mode>): Likewise.
1350         (neon_vrev16<mode>): Likewise.
1351         (neon_vbsl<mode>_internal): Likewise.
1352         (neon_vshl<mode>): Likewise.
1353         (neon_vqshl<mode>): Likewise.
1354         (neon_vshr_n<mode>): Likewise.
1355         (neon_vshrn_n<mode>): Likewise.
1356         (neon_vqshrn_n<mode>): Likewise.
1357         (neon_vqshrun_n<mode>): Likewise.
1358         (neon_vshl_n<mode>): Likewise.
1359         (neon_vqshl_n<mode>): Likewise.
1360         (neon_vqshlu_n<mode>): Likewise.
1361         (neon_vshll_n<mode>): Likewise.
1362         (neon_vsra_n<mode>): Likewise.
1363         (neon_vsri_n<mode>): Likewise.
1364         (neon_vsli_n<mode>): Likewise.
1365         (neon_vtbl1v8qi): Likewise.
1366         (neon_vtbl2v8qi): Likewise.
1367         (neon_vtbl3v8qi): Likewise.
1368         (neon_vtbl4v8qi): Likewise.
1369         (neon_vtbl1v16qi): Likewise.
1370         (neon_vtbl2v16qi): Likewise.
1371         (neon_vcombinev16qi): Likewise.
1372         (neon_vtbx1v8qi): Likewise.
1373         (neon_vtbx2v8qi): Likewise.
1374         (neon_vtbx3v8qi): Likewise.
1375         (neon_vtbx4v8qi): Likewise.
1376         (*neon_vtrn<mode>_insn): Likewise.
1377         (*neon_vzip<mode>_insn): Likewise.
1378         (*neon_vuzp<mode>_insn): Likewise.
1379         (neon_vld1<mode>): Likewise.
1380         (neon_vld1_lane<mode>): Likewise.
1381         (neon_vld1_lane<mode>): Likewise.
1382         (neon_vld1_dup<mode>): Likewise.
1383         (neon_vld1_dup<mode>): Likewise.
1384         (neon_vld1_dupv2di): Likewise.
1385         (neon_vst1<mode>): Likewise.
1386         (neon_vst1_lane<mode>): Likewise.
1387         (neon_vst1_lane<mode>): Likewise.
1388         (neon_vld2<mode>): Likewise.
1389         (neon_vld2<mode>): Likewise.
1390         (neon_vld2_lane<mode>): Likewise.
1391         (neon_vld2_lane<mode>): Likewise.
1392         (neon_vld2_dup<mode>): Likewise.
1393         (neon_vst2<mode>): Likewise.
1394         (neon_vst2<mode>): Likewise.
1395         (neon_vst2_lane<mode>): Likewise.
1396         (neon_vst2_lane<mode>): Likewise.
1397         (neon_vld3<mode>): Likewise.
1398         (neon_vld3qa<mode>): Likewise.
1399         (neon_vld3qb<mode>): Likewise.
1400         (neon_vld3_lane<mode>): Likewise.
1401         (neon_vld3_lane<mode>): Likewise.
1402         (neon_vld3_dup<mode>): Likewise.
1403         (neon_vst3<mode>): Likewise.
1404         (neon_vst3qa<mode>): Likewise.
1405         (neon_vst3qb<mode>): Likewise.
1406         (neon_vst3_lane<mode>): Likewise.
1407         (neon_vst3_lane<mode>): Likewise.
1408         (neon_vld4<mode>): Likewise.
1409         (neon_vld4qa<mode>): Likewise.
1410         (neon_vld4qb<mode>): Likewise.
1411         (neon_vld4_lane<mode>): Likewise.
1412         (neon_vld4_lane<mode>): Likewise.
1413         (neon_vld4_dup<mode>): Likewise.
1414         (neon_vst4<mode>): Likewise.
1415         (neon_vst4qa<mode>): Likewise.
1416         (neon_vst4qb<mode>): Likewise.
1417         (neon_vst4_lane<mode>): Likewise.
1418         (neon_vst4_lane<mode>): Likewise.
1419         (neon_vec_unpack<US>_lo_<mode>): Likewise.
1420         (neon_vec_unpack<US>_hi_<mode>): Likewise.
1421         (neon_vec_<US>mult_lo_<mode>): Likewise.
1422         (neon_vec_<US>mult_hi_<mode>): Likewise.
1423         (neon_vec_<US>shiftl_<mode>): Likewise.
1424         (neon_unpack<US>_<mode>): Likewise.
1425         (neon_vec_<US>mult_<mode>): Likewise.
1426         (vec_pack_trunc_<mode>): Likewise.
1427         (neon_vec_pack_trunc_<mode>): Likewise.
1428         (neon_vabd<mode>_2): Likewise.
1429         (neon_vabd<mode>_3): Likewise.
1431 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1433         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
1434         (load_pair): Update type attribute.
1435         (store_pair): Update type attribute.
1436         * config/aarch64/iterators.md (q): New.
1438 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
1440         * config/arm/types.md: Add new types for Neon insns.
1442 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1443             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1444             Sergey Lega  <sergey.s.lega@intel.com>
1445             Anna Tikhonova  <anna.tikhonova@intel.com>
1446             Ilya Tocar  <ilya.tocar@intel.com>
1447             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1448             Ilya Verbin  <ilya.verbin@intel.com>
1449             Kirill Yukhin  <kirill.yukhin@intel.com>
1450             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1452         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
1453         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
1454         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
1455         (rcp14<mode>): New.
1456         (srcp14<mode>): Ditto.
1457         (rsqrt14<mode>): Ditto.
1458         (rsqrt14<mode>): Ditto.
1459         (avx512f_vmscalef<mode>): Ditto.
1460         (avx512f_scalef<mode>): Ditto.
1461         (avx512f_getexp<mode>): Ditto.
1462         (avx512f_sgetexp<mode>): Ditto.
1463         (avx512f_fixupimm<mode>): Ditto.
1464         (avx512f_sfixupimm<mode>): Ditto.
1465         (avx512f_rndscale<mode>): Ditto.
1466         (*avx512er_exp2<mode>): Ditto.
1467         (*avx512er_rcp28<mode>): Ditto.
1468         (avx512er_rsqrt28<mode>): Ditto.
1469         (avx512f_getmant<mode>): Ditto.
1470         (avx512f_getmant<mode>): Ditto.
1471         (avx512f_rndscale<mode>): Fix formatting.
1473 2013-10-15  Martin Jambor  <mjambor@suse.cz>
1475         * ipa-utils.h (ipa_edge_within_scc): Declare.
1476         * ipa-cp.c (edge_within_scc): Moved...
1477         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
1479 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1480             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1481             Sergey Lega  <sergey.s.lega@intel.com>
1482             Anna Tikhonova  <anna.tikhonova@intel.com>
1483             Ilya Tocar  <ilya.tocar@intel.com>
1484             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1485             Ilya Verbin  <ilya.verbin@intel.com>
1486             Kirill Yukhin  <kirill.yukhin@intel.com>
1487             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1489         * config/i386/predicates.md (const_8_to_15_operand): New.
1490         (const_16_to_31_operand): Ditto.
1491         * config/i386/sse.md (V8FI): New.
1492         (V16FI): Ditto.
1493         (reduc_splus_v8df): Ditto.
1494         (reduc_splus_v16sf): Ditto.
1495         (avx512f_vextract<shuffletype>32x4_1): Ditto.
1496         (vec_extract_hi_<mode>): Ditto.
1497         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
1498         (vec_set_lo_<mode>): Ditto.
1499         (vec_set_hi_<mode>): Ditto.
1500         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
1501         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
1502         (avx512f_pshufd_1): Ditto.
1503         (avx512f_broadcast<mode>): Ditto.
1504         (avx512f_broadcast<mode>): Ditto.
1505         (define_split): Split vec_extract_lo into move.
1506         (ssequartermode): Ditto.
1507         (ssedoublemode): Extened with wider modes.
1508         (vec_extract_lo_<mode>): Ditto.
1510 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1511             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1512             Sergey Lega  <sergey.s.lega@intel.com>
1513             Anna Tikhonova  <anna.tikhonova@intel.com>
1514             Ilya Tocar  <ilya.tocar@intel.com>
1515             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1516             Ilya Verbin  <ilya.verbin@intel.com>
1517             Kirill Yukhin  <kirill.yukhin@intel.com>
1518             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1520         * config/i386/predicates.md (register_or_constm1_operand): New.
1521         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
1522         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
1523         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
1524         UNSPEC_SCATTER_PREFETCH
1525         (VI48_512): New.
1526         (avx512f_ucmp<mode>3): Ditto.
1527         (avx512f_vternlog<mode>): Ditto.
1528         (avx512f_align<mode>): Ditto.
1529         (<shift_insn><mode>3): Ditto.
1530         (avx512f_<rotate>v<mode>): Ditto.
1531         (avx512f_<rotate><mode>): Ditto.
1532         (avx512f_eq<mode>3): Ditto.
1533         (avx512f_eq<mode>3_1): Ditto.
1534         (avx512f_gt<mode>3): Ditto.
1535         (avx512f_testm<mode>3): Ditto.
1536         (avx512f_testnm<mode>3): Ditto.
1537         (avx512pf_gatherpf<mode>): Ditto.
1538         (*avx512pf_gatherpf<mode>_mask): Ditto.
1539         (*avx512pf_gatherpf<mode>): Ditto.
1540         (avx512pf_scatterpf<mode>): Ditto.
1541         (*avx512pf_scatterpf<mode>_mask): Ditto.
1542         (*avx512pf_scatterpf<mode>): Ditto.
1543         (avx512f_vec_dup_gpr<mode>): Ditto.
1544         (clz<mode>2): Ditto.
1545         (conflict<mode>): Ditto.
1546         (REDUC_SMINMAX_MODE): Extened with wider modes.
1547         (reduc_<code>_<mode>): Ditto.
1548         (vlshr<mode>3): Ditto.
1549         (vashl<mode>3): Ditto.
1551 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1552             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1553             Sergey Lega  <sergey.s.lega@intel.com>
1554             Anna Tikhonova  <anna.tikhonova@intel.com>
1555             Ilya Tocar  <ilya.tocar@intel.com>
1556             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1557             Ilya Verbin  <ilya.verbin@intel.com>
1558             Kirill Yukhin  <kirill.yukhin@intel.com>
1559             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1561         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
1562         UNSPEC_SCATTER.
1563         (VI48F_512): New.
1564         (avx512fmaskmode): Ditto.
1565         (bcstscalarsuff): Ditto.
1566         (avx512f_blendm<mode>): Ditto.
1567         (cmp_imm_predicate): Ditto.
1568         (avx512f_cmp<mode>3): Ditto.
1569         (avx512f_vec_dup<mode>): Ditto.
1570         (avx512f_vec_dup_mem<mode>): Ditto.
1571         (avx512f_vpermi2var<mode>3): Ditto.
1572         (avx512f_vpermt2var<mode>3): Ditto.
1573         (vec_init<mode>): Ditto.
1574         (avx512f_gathersi<mode>): Ditto.
1575         (*avx512f_gathersi<mode>): Ditto.
1576         (*avx512f_gathersi<mode>_2): Ditto.
1577         (avx512f_gatherdi<mode>): Ditto.
1578         (*avx512f_gatherdi<mode>): Ditto.
1579         (*avx512f_gatherdi<mode>_2): Ditto.
1580         (avx512f_scattersi<mode>): Ditto.
1581         (*avx512f_scattersi<mode>): Ditto.
1582         (avx512f_scatterdi<mode>): Ditto.
1583         (*avx512f_scatterdi<mode>): Ditto.
1584         (sseintprefix): Extened with wider modes.
1585         (VEC_GATHER_IDXSI): Ditto.
1586         (VEC_GATHER_IDXDI): Ditto.
1587         (VEC_GATHER_SRCDI): Ditto.
1589 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
1590             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1592         * config/arm/t-aprofile: New file.
1593         * config.gcc: Handle --with-multilib-list option.
1595 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
1597         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
1598         the input for r2.
1600 2013-10-15  Richard Biener  <rguenther@suse.de>
1602         * tree-loop-distribution.c (build_empty_rdg): Inline into
1603         single user.
1604         (rdg_flag_vertex): Inline into single user.
1605         (rdg_flag_vertex_and_dependent): Likewise.
1606         (build_rdg_partition_for_vertex): Remove processed bitmap.
1607         (rdg_build_partitions): Simplify.
1609 2013-10-15  Richard Biener  <rguenther@suse.de>
1611         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
1612         Restructure forwarding through conversions and copies to
1613         avoid performing copy-propagation the wrong way.  Adjust
1614         recursion invocations.
1615         (forward_propagate_addr_expr): Add argument stating if we
1616         are recursing from a single-use.
1617         (ssa_forward_propagate_and_combine): Adjust.
1619 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
1621         * dumpfile.h (gcc::dump_manager): New class, to hold state
1622         relating to dumpfile management.
1623         (get_dump_file_name): Remove in favor of method of dump_manager.
1624         (dump_initialized_p): Likewise.
1625         (dump_start): Likewise.
1626         (dump_finish): Likewise.
1627         (dump_switch_p): Likewise.
1628         (dump_register): Likewise.
1629         (get_dump_file_info): Likewise.
1630         * context.c (gcc::context::context): Construct the dump_manager
1631         instance.
1632         * context.h (gcc::context::get_dumps): New.
1633         (gcc::context::m_dumps): New.
1634         * coverage.c (coverage_init): Port to dump_manager API.
1635         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
1636         (extra_dump_files_in_use): Likewise.
1637         (extra_dump_files_alloced): Likewise.
1638         (gcc::dump_manager::dump_manager): New.
1639         (dump_register): Convert to...
1640         (gcc::dump_manager::dump_register): ...method, replacing
1641         function-static next_dump with m_next_dump field.
1642         (get_dump_file_info): Convert to...
1643         (gcc::dump_manager::get_dump_file_info): ...method.
1644         (get_dump_file_name): Convert to...
1645         (gcc::dump_manager::get_dump_file_name): ...method.
1646         (dump_start): Convert to...
1647         (gcc::dump_manager::dump_start): ...method.
1648         (dump_finish): Convert to...
1649         (gcc::dump_manager::dump_finish): ...method.
1650         (dump_begin): Replace body with...
1651         (gcc::dump_manager::dump_begin): ...new method.
1652         (dump_phase_enabled_p): Convert to...
1653         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1654         (dump_phase_enabled_p): Convert to...
1655         (gcc::dump_manager::dump_phase_enabled_p): ...method.
1656         (dump_initialized_p):  Convert to...
1657         (gcc::dump_manager::dump_initialized_p): ...method.
1658         (dump_flag_name): Replace body with...
1659         (gcc::dump_manager::dump_flag_name): ...new method.
1660         (dump_enable_all): Convert to...
1661         (gcc::dump_manager::dump_enable_all): ...new method.
1662         (opt_info_enable_passes): Convert to...
1663         (gcc::dump_manager::opt_info_enable_passes): ...new method.
1664         (dump_switch_p_1): Convert to...
1665         (gcc::dump_manager::dump_switch_p_1): ...new method.
1666         (dump_switch_p):  Convert to...
1667         (gcc::dump_manager::dump_switch_p): ...new method.
1668         (opt_info_switch_p): Port to dump_manager API.
1669         (enable_rtl_dump_file): Likewise.
1670         * opts-global.c (handle_common_deferred_options): Port to new
1671         dump_manager API.
1672         * passes.c (pass_manager::finish_optimization_passes): Likewise.
1673         (pass_manager::register_one_dump_file): Likewise.
1674         (pass_manager::register_pass): Likewise.
1675         (pass_init_dump_file): Likewise.
1676         (pass_fini_dump_file): Likewise.
1677         * statistics.c (statistics_early_init): Likewise.
1679 2013-10-14  Richard Biener  <rguenther@suse.de>
1681         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
1682         iterative_hash_canonical_type, gimple_canonical_type_hash,
1683         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
1684         gimple_register_canonical_type, print_gimple_types_stats,
1685         free_gimple_type_tables): Move to lto/lto.c
1686         (gt-gimple.h): Do not include.
1687         * gimple.h (gimple_register_canonical_type,
1688         print_gimple_types_stats, free_gimple_type_tables): Remove.
1689         * Makefile.in (GTFILES): Remove gimple.c.
1691 2013-10-14  Travis Snoozy  <quandary@remstate.com>
1693         PR target/58716
1694         * config/msp430/msp430.c (msp430_option_override): Correct thinko
1695         scanning for msp430x targets.
1697 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
1699         PR bootstrap/58509
1700         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
1701         (registers_ok_for_ldd_peep): Move around.
1702         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
1703         * config/sparc/sparc.md (widening peepholes): Use it.
1705 2013-10-14  Richard Biener  <rguenther@suse.de>
1707         PR middle-end/58712
1708         PR middle-end/55358
1709         * gimple.c (iterative_hash_canonical_type): Make sure to
1710         record the hash into the correct hashtable slot.
1712 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
1714         PR rtl-optimization/58662
1715         * combine.c (try_combine): Take into account death nodes on I2 when
1716         splitting a PARALLEL of two independent SETs.  Fix dump message.
1718 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
1720         PR target/51244
1721         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
1722         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
1723         * config/sh/t-sh (sh_treg_combine.o): New entry.
1724         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
1725         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
1726         (register_sh_passes): New function.  Register sh_treg_combine pass.
1727         (sh_option_override): Invoke it.
1728         (sh_canonicalize_comparison): Handle op0_preserve_value.
1729         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
1730         opportunities.  Canonicalize branch condition.
1731         (nott): Allow only if pseudos can be created for non-SH2A.
1733 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
1735         PR target/58690
1736         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
1737         (ix86_expand_movmem): Replace copy_addr_to_reg with
1738         ix86_copy_addr_to_reg.
1739         (ix86_expand_setmem): Likewise.
1741 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
1743         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
1744         provided by ix86_fp_compare_mode instead of CCFPUmode.
1746 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
1748         * config/aarch64/arm_neon.h
1749         (vtbx<1,3>_<psu>8): Fix register constriants.
1751 2013-10-11  Jeff Law  <law@redhat.com>
1753         PR tree-optimization/58640
1754         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
1755         threading paths that cross over two loop entry points.
1757 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1759         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
1760         handle vector float as well.
1761         (*vsx_le_perm_load_v4si): Likewise.
1762         (*vsx_le_perm_store_v2di): Likewise.
1763         (*vsx_le_perm_store_v4si): Likewise.
1765 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
1767         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
1768         directly to circumvent subtract from splat{31} workaround.
1769         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
1770         prototype.
1771         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
1772         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
1773         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
1774         (altivec_vperm_<mode>): Convert to define_insn_and_split to
1775         separate big and little endian logic.
1776         (*altivec_vperm_<mode>_internal): New define_insn.
1777         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
1778         separate big and little endian logic.
1779         (*altivec_vperm_<mode>_uns_internal): New define_insn.
1780         (vec_permv16qi): Add little endian logic.
1782 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
1784         * doc/extend.texi (returns_nonnull): Remove arguments.
1786 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1787             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1788             Sergey Lega  <sergey.s.lega@intel.com>
1789             Anna Tikhonova  <anna.tikhonova@intel.com>
1790             Ilya Tocar  <ilya.tocar@intel.com>
1791             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1792             Ilya Verbin  <ilya.verbin@intel.com>
1793             Kirill Yukhin  <kirill.yukhin@intel.com>
1794             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1796         * config/i386/sse.md (VI48F_256_512): New.
1797         (avx2_permvar<mode>): Change to ...
1798         (<avx2_avx512f>_permvar<mode>): This.
1800 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1801             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1802             Sergey Lega  <sergey.s.lega@intel.com>
1803             Anna Tikhonova  <anna.tikhonova@intel.com>
1804             Ilya Tocar  <ilya.tocar@intel.com>
1805             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1806             Ilya Verbin  <ilya.verbin@intel.com>
1807             Kirill Yukhin  <kirill.yukhin@intel.com>
1808             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1810         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
1811         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
1812         * config/i386/sse.md (VI4_AVX): New.
1813         (sf2simodelower): Ditto.
1814         (sse2_cvtps2dq): Change to ...
1815         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
1817 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1818             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1819             Sergey Lega  <sergey.s.lega@intel.com>
1820             Anna Tikhonova  <anna.tikhonova@intel.com>
1821             Ilya Tocar  <ilya.tocar@intel.com>
1822             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1823             Ilya Verbin  <ilya.verbin@intel.com>
1824             Kirill Yukhin  <kirill.yukhin@intel.com>
1825             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1827         * config/i386/sse.md (V_512): New.
1828         (VI_512): Ditto.
1829         (vcond<V_512:mode><VF_512:mode>): Ditto.
1830         (vcond<V_512:mode><VI_512:mode>): Ditto.
1831         (vcondu<V_512:mode><VI_512:mode>): Ditto.
1833 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1834             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1835             Sergey Lega  <sergey.s.lega@intel.com>
1836             Anna Tikhonova  <anna.tikhonova@intel.com>
1837             Ilya Tocar  <ilya.tocar@intel.com>
1838             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1839             Ilya Verbin  <ilya.verbin@intel.com>
1840             Kirill Yukhin  <kirill.yukhin@intel.com>
1841             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1843         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
1844         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
1845         (FMAMODE): Ditto.
1846         (fma<mode>4): Removed condition.
1847         (fms<mode>4): Ditto.
1848         (fnma<mode>4): Ditto.
1849         (fnms<mode>4): Ditto.
1850         (fma4i_fmadd_<mode>): Ditto.
1851         (*fma_fmadd_<mode>): Ditto.
1852         (*fma_fmsub_<mode>): Ditto.
1853         (*fma_fnmadd_<mode>): Ditto.
1854         (*fma_fnmsub_<mode>): Ditto.
1855         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
1856         (*fma_fmaddsub_<mode>): Ditto.
1857         (*fma_fmsubadd_<mode>): Ditto.
1858         (*fmai_fmadd_<mode>): Ditto.
1859         (*fmai_fmsub_<mode>): Ditto.
1860         (*fmai_fnmadd_<mode>): Ditto.
1861         (*fmai_fnmsub_<mode>): Ditto.
1863 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1864             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1865             Sergey Lega  <sergey.s.lega@intel.com>
1866             Anna Tikhonova  <anna.tikhonova@intel.com>
1867             Ilya Tocar  <ilya.tocar@intel.com>
1868             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1869             Ilya Verbin  <ilya.verbin@intel.com>
1870             Kirill Yukhin  <kirill.yukhin@intel.com>
1871             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1873         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
1874         (VI124_256): Changed to ...
1875         (VI124_256_48_512): This.
1876         (ssepackmode): Extended with wider modes.
1877         (<code><mode>3): Changed iterator.
1878         (*avx2_<code><mode>3): Ditto.
1879         (vec_pack_trunc_<mode>): Ditto.
1881 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1882             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1883             Sergey Lega  <sergey.s.lega@intel.com>
1884             Anna Tikhonova  <anna.tikhonova@intel.com>
1885             Ilya Tocar  <ilya.tocar@intel.com>
1886             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1887             Ilya Verbin  <ilya.verbin@intel.com>
1888             Kirill Yukhin  <kirill.yukhin@intel.com>
1889             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1891         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
1892         (VI8F_256_512): Ditto.
1893         (abs<mode>2): Changed iterator.
1894         (avx2_perm<mode>): Changed to ...
1895         (<avx2_avx512f>_perm<mode>): This.
1896         (avx2_perm<mode>_1): Changed to ...
1897         (<avx2_avx512f>_perm<mode>_1): This.
1899 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1900             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1901             Sergey Lega  <sergey.s.lega@intel.com>
1902             Anna Tikhonova  <anna.tikhonova@intel.com>
1903             Ilya Tocar  <ilya.tocar@intel.com>
1904             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1905             Ilya Verbin  <ilya.verbin@intel.com>
1906             Kirill Yukhin  <kirill.yukhin@intel.com>
1907             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1909         * config/i386/sse.md (VI48_AVX512F): New.
1910         (VI48_AVX2): Changed to ...
1911         (VI48_AVX2_48_AVX512F): This.
1912         (avx2_ashrv<mode>): Changed to ...
1913         (<avx2_avx512f>_ashrv<mode>): This.
1914         (avx2_<shift_insn>v<mode>): Changed to ...
1915         (<avx2_avx512f>_<shift_insn>v<mode>): This.
1917 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1918             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1919             Sergey Lega  <sergey.s.lega@intel.com>
1920             Anna Tikhonova  <anna.tikhonova@intel.com>
1921             Ilya Tocar  <ilya.tocar@intel.com>
1922             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1923             Ilya Verbin  <ilya.verbin@intel.com>
1924             Kirill Yukhin  <kirill.yukhin@intel.com>
1925             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1927         * config/i386/sse.md (VI4_AVX512F): New.
1928         (VI8_AVX2_AVX512F): Ditto.
1929         (mul<mode>3): Extended with wider modes.
1930         (*<sse4_1_avx2>_mul<mode>3): Ditto.
1931         (mul<mode>3): Ditto.
1932         (vec_widen_<s>mult_odd_<mode>): Ditto.
1934 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1935             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1936             Sergey Lega  <sergey.s.lega@intel.com>
1937             Anna Tikhonova  <anna.tikhonova@intel.com>
1938             Ilya Tocar  <ilya.tocar@intel.com>
1939             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1940             Ilya Verbin  <ilya.verbin@intel.com>
1941             Kirill Yukhin  <kirill.yukhin@intel.com>
1942             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1944         * config/i386/sse.md (VI2_AVX512F): New.
1945         (VI124_AVX512F): Ditto.
1946         (sseunpackmode): Extended with wider modes.
1947         (sseunpackfltmode): Ditto.
1948         (vec_unpacks_float_hi_<mode>): Ditto.
1949         (vec_unpacks_float_lo_<mode>): Ditto.
1950         (vec_unpacku_float_hi_<mode>): Ditto.
1951         (vec_unpacku_float_lo_<mode>): Ditto.
1952         (vec_unpacks_lo_<mode>): Ditto.
1953         (vec_unpacks_hi_<mode>): Ditto.
1954         (vec_unpacku_lo_<mode>): Ditto.
1955         (vec_unpacku_hi_<mode>): Ditto.
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 (multdiv): New.
1968         (multdiv_mnemonic): Ditto.
1969         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
1970         (<sse>_vm<multdiv_mnemonic><mode>3): This.
1971         (<sse>_vmdiv<mode>3): Removed.
1973 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1974             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1975             Sergey Lega  <sergey.s.lega@intel.com>
1976             Anna Tikhonova  <anna.tikhonova@intel.com>
1977             Ilya Tocar  <ilya.tocar@intel.com>
1978             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1979             Ilya Verbin  <ilya.verbin@intel.com>
1980             Kirill Yukhin  <kirill.yukhin@intel.com>
1981             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1983         * config/i386/sse.md (V): Extended with wider modes.
1984         (VF2): Ditto.
1985         (ssehalfvecmode): Ditto.
1986         (i128): Ditto.
1987         (ssepackfltmode): Ditto.
1988         (avx_vec_concat<mode>): Ditto.
1989         (V_256_512): New iterator.
1990         (VF2_512_256): Ditto.
1991         (si2dfmode): New attribute.
1992         (si2dfmodelower): Ditto.
1993         (sf2dfmode): Ditto.
1994         (concat_tg_mode): Ditto.
1995         (floatv4siv4df2): Changed to ...
1996         (float<si2dfmodelower><mode>2): This.
1997         (avx_cvtps2pd256): Changed to ...
1998         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
1999         (vec_pack_trunc_v4df): Changed to ...
2000         (vec_pack_trunc_<mode>): This.
2001         (avx_vpermil<mode>): Changed to ...
2002         (<sse2_avx_avx512f>_vpermil<mode>): This.
2003         (<fixsuffix>fix_truncv8dfv8si2): New.
2004         (vec_pack_sfix_trunc_v8df): Ditto.
2005         (avx512f_rndscale<mode>): Ditto.
2006         (avx512f_roundpd512): Ditto.
2007         (vec_pack_ufix_trunc_<mode>): Updated iterator.
2009 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2010             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2011             Sergey Lega  <sergey.s.lega@intel.com>
2012             Anna Tikhonova  <anna.tikhonova@intel.com>
2013             Ilya Tocar  <ilya.tocar@intel.com>
2014             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2015             Ilya Verbin  <ilya.verbin@intel.com>
2016             Kirill Yukhin  <kirill.yukhin@intel.com>
2017             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2019         * config/i386/i386.md (any_fix): New iterator.
2020         (fixsuffix): New attribute.
2021         * config/i386/sse.md (VF1): Extened with wider modes.
2022         (VI): Ditto.
2023         (VI_AVX2): Ditto.
2024         (VI8): Ditto.
2025         (sseintvecmodelower): Ditto.
2026         (ssescalarmode): Ditto.
2027         (ssescalarnum): Ditto.
2028         (VF1_128_256): New.
2029         (ssexmmmode): Ditto.
2030         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
2031         (<sse>_rcp<mode>2): Change iterator.
2032         (rsqrt<mode>2): Ditto.
2033         (<sse>_rsqrt<mode>2): Ditto.
2034         (avx2_vec_dup<mode>): Ditto.
2035         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
2036         (round<mode>2_sfix): Ditto.
2037         (avx2_pbroadcast<mode>): Ditto.
2038         (*andnot<mode>3): Handle XI mode.
2039         (*<code><mode>3): Ditto.
2040         (AVXTOSSEMODE): Removed.
2041         (avx_vpermil<mode>): Changed to ...
2042         (<sse2_avx_avx512f>_vpermil<mode>): This.
2044 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
2045             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
2046             Sergey Lega  <sergey.s.lega@intel.com>
2047             Anna Tikhonova  <anna.tikhonova@intel.com>
2048             Ilya Tocar  <ilya.tocar@intel.com>
2049             Andrey Turetskiy  <andrey.turetskiy@intel.com>
2050             Ilya Verbin  <ilya.verbin@intel.com>
2051             Kirill Yukhin  <kirill.yukhin@intel.com>
2052             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
2054         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
2055         (<sse>_comi): Ditto.
2056         (<sse>_ucomi): Ditto.
2057         (sse_cvtss2siq_2): Ditto.
2058         (sse2_cvtsd2si): Ditto.
2059         (sse2_cvtsd2siq): Ditto.
2060         (sse2_cvttsd2si): Ditto.
2061         (sse2_cvttsd2siq): Ditto.
2062         (<shift_insn><mode>3): Ditto.
2063         (sse2_cvtsi2sdq): Update constraint and prefix.
2064         (sse_cvtsi2ss): Update prefix.
2065         (sse_cvtsi2ssq): Ditto.
2067 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2069         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
2070         ignore internal calls.
2072 2013-10-11  Richard Biener  <rguenther@suse.de>
2074         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
2075         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
2076         TREE_OVERFLOW is set.
2078 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
2080         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
2082         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
2083         * gimple.h: Likewise.
2085         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
2087         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
2088         macros.
2090         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
2092         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
2093         * configure: Regenerate.
2095 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
2097         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
2098         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
2099         formatting fixes, use pp_colon instead of pp_character (..., ':'),
2100         similarly pp_right_paren.
2101         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
2102         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
2103         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
2104         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
2105         clauses.
2106         (omp_declare_simd_clauses_equal,
2107         omp_remove_redundant_declare_simd_attrs): New functions.
2108         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
2109         (walk_tree_1): Handle new OpenMP 4.0 clauses.
2110         * tree.h (OMP_LOOP_CHECK): Define.
2111         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
2112         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
2113         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
2114         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
2115         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
2116         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
2117         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
2118         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
2119         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
2120         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
2121         OMP_CLAUSE_SIMDLEN_EXPR): Define.
2122         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
2123         (omp_remove_redundant_declare_simd_attrs): New prototype.
2124         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
2125         GIMPLE_OMP_TEAMS): New codes.
2126         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
2127         * omp-low.c (struct omp_context): Add cancel_label and cancellable
2128         fields.
2129         (target_nesting_level): New variable.
2130         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
2131         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
2132         for collapse > 1 static schedule unless ordered.
2133         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
2134         (determine_parallel_type): Adjust get_ws_args_for caller.
2135         (install_var_field): Handle mask & 4 for double indirection.
2136         (scan_sharing_clauses): Ignore shared clause on teams construct.
2137         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
2138         (create_omp_child_function): If inside target or declare target
2139         constructs, set "omp declare target" attribute on the child function.
2140         (find_combined_for): New function.
2141         (scan_omp_parallel): Handle combined loops.
2142         (scan_omp_target, scan_omp_teams): New functions.
2143         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
2144         restrictions and set ctx->cancellable for cancellable constructs.
2145         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
2146         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
2147         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
2148         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
2149         (omp_clause_aligned_alignment): New function.
2150         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
2151         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
2152         on teams constructs.  Handle user defined reductions and new
2153         OpenMP 4.0 clauses.
2154         (lower_reduction_clauses): Don't set placeholder to address of ref
2155         if it has already the right type.
2156         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
2157         (expand_parallel_call): Use the new non-_start suffixed builtins,
2158         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
2159         and GOMP_parallel_end after the call.
2160         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
2161         (expand_omp_for_init_counts): Handle combined loops.
2162         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
2163         loops.
2164         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
2165         end of cancellable loops.
2166         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
2167         Likewise.  Handle collapse > 1 loops.
2168         (expand_omp_simd): Handle combined loops.
2169         (expand_omp_for): Add inner_stmt argument, adjust callers of
2170         expand_omp_for* functions, use expand_omp_for_static*chunk even
2171         for collapse > 1 unless ordered.
2172         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
2173         of cancellable sections.
2174         (expand_omp_single): Remove need_barrier variable, just rely on
2175         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
2176         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
2177         (expand_omp_atomic_load, expand_omp_atomic_store,
2178         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
2179         (expand_omp_target): New function.
2180         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
2181         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
2182         (build_omp_regions_1): Immediately close region for
2183         GF_OMP_TARGET_KIND_UPDATE.
2184         (maybe_add_implicit_barrier_cancel): New function.
2185         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
2186         cancellation.
2187         (lower_omp_single): Likewise.  Add clobber after the barrier.
2188         (lower_omp_taskgroup): New function.
2189         (lower_omp_for): Handle combined loops.  Adjust
2190         lower_rec_input_clauses caller.  Handle cancellation.
2191         (lower_depend_clauses): New function.
2192         (lower_omp_taskreg): Lower depend clauses.  Adjust
2193         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
2194         cancellation.
2195         (lower_omp_target, lower_omp_teams): New functions.
2196         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
2197         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
2198         and GOMP_cancellation_point calls.
2199         (lower_omp): Fold stmts inside of target region.
2200         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
2201         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2202         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
2203         (BT_FN_VOID_OMPFN_PTR_UINT,
2204         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
2205         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
2206         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
2207         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
2208         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
2209         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
2210         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
2211         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
2212         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
2213         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
2214         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
2215         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
2216         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
2217         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
2218         BUILT_IN_GOMP_PARALLEL_END.
2219         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
2220         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
2221         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
2222         GF_OMP_FOR_KIND_DISTRIBUTE.
2223         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
2224         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
2225         (dump_gimple_omp_return): Print lhs if it has any.
2226         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
2227         gimple_omp_atomic_seq_cst_p.
2228         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
2229         and GIMPLE_OMP_TEAMS.
2230         * langhooks.c (lhd_omp_mappable_type): New function.
2231         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
2232         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
2233         hook.
2234         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
2235         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
2236         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
2237         (struct gimplify_omp_ctx): Add combined_loop field.
2238         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
2239         on stmts inside of target region.
2240         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
2241         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
2242         ORT_TARGET and ORT_TARGET_DATA.
2243         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
2244         Handle GOVD_MAP.
2245         (omp_notice_threadprivate_variable): Complain about threadprivate
2246         variables in target region.
2247         (omp_notice_variable): Complain about vars with non-mappable type
2248         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
2249         (omp_check_private): Ignore ORT_TARGET* regions.
2250         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
2251         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
2252         (find_combined_omp_for): New function.
2253         (gimplify_omp_for): Handle gimplification of combined loops.
2254         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
2255         OMP_TEAMS.
2256         (gimplify_omp_target_update): New function.
2257         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
2258         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
2259         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
2260         (gimplify_body): If fndecl has "omp declare target" attribute, add
2261         implicit ORT_TARGET context around it.
2262         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
2263         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
2264         * tree-nested.c (convert_nonlocal_reference_stmt,
2265         convert_local_reference_stmt, convert_gimple_call): Handle
2266         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2267         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
2268         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
2269         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
2270         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
2271         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
2272         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
2273         BUILT_IN_GOMP_LOOP_END_CANCEL,
2274         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
2275         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
2276         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
2277         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
2278         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
2279         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
2280         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
2281         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
2282         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
2283         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
2284         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
2285         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
2286         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
2287         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
2288         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
2289         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2290         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
2291         gimple_build_omp_teams): New functions.
2292         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
2293         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
2294         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
2295         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2296         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
2297         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
2298         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
2299         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
2300         GF_OMP_ATOMIC_SEQ_CST): New.
2301         (gimple_build_omp_taskgroup, gimple_build_omp_target,
2302         gimple_build_omp_teams): New prototypes.
2303         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
2304         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2305         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
2306         GIMPLE_OMP_SINGLE as end of range.
2307         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
2308         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
2309         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
2310         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
2311         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
2312         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
2313         gimple_omp_target_kind, gimple_omp_target_set_kind,
2314         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
2315         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
2316         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
2317         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
2318         gimple_omp_teams_set_clauses): New inlines.
2319         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
2320         and GIMPLE_OMP_TASKGROUP.
2321         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
2322         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
2323         enum omp_clause_proc_bind_kind): New.
2324         (union omp_clause_subcode): Add depend_kind, map_kind and
2325         proc_bind_kind fields.
2326         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
2327         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
2328         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
2329         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
2330         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
2332 2013-10-10  Teresa Johnson  <tejohnson@google.com>
2334         * predict.c (tree_estimate_probability): Add new parameter
2335         for estimate_bb_frequencies.
2336         (estimate_bb_frequencies): Add new parameter to force estimation.
2337         (rebuild_frequencies): When max frequency in function is small,
2338         recompute counts from frequencies.
2339         * predict.h (estimate_bb_frequencies): New parameter.
2341 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2343         * ipa-inline.c (ipa_inline): Fix leak of "order" when
2344         optimizations are disabled.
2346 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
2348         * coverage.c (coverage_finish): Fix leak of da_file_name.
2350 2013-10-10  Jan Hubicka  <jh@suse.cz>
2352         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
2353         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
2354         Bobcat and generic.
2356 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
2358         PR middle-end/58670
2359         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
2360         if any labels are in FALLTHRU_BB, use a special label emitted
2361         immediately after the asm goto insn rather than label_rtx
2362         of the LABEL_DECL.
2363         (expand_asm_stmt): Adjust caller.
2364         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
2365         edge if the last insn in predecessor is a jump with single successor,
2366         but it isn't simplejump_p.
2368 2013-10-10  Richard Biener  <rguenther@suse.de>
2370         PR tree-optimization/58656
2371         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
2373 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2375         * gimplify.c: Include expr.h and tm_p.h for targets with special
2376         va-arg padding requirements.
2378 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2380         * tree-flow.h: Move some prototypes to gimple.h.
2381         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
2382         * gimple.h: Relocate some prototypes from tree-flow.h
2383         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
2384         Move to gimplify.c.
2385         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
2386         (build_va_arg_indirect_ref): Relocate and make static.
2387         (std_gimplify_va_arg_expr): Relocate here.
2388         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
2389         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
2391 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2393         * doc/md.texi: Document the mnemonic attribute.
2395 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2397         PR target/57377
2398         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
2399         (set_attr_alternative x ...) when searching for user defined
2400         mnemonic attribute.
2402 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
2404         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
2406 2013-10-09  Easwaran Raman <eraman@google.com>
2408         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
2409         * cfgexpand.c (defer_stack_allocation): ...use here
2410         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
2412 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
2414         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
2415         (operand_equal_for_value_replacement): New function, extracted from
2416         value_replacement and enhanced to catch more cases.
2417         (value_replacement): Use operand_equal_for_value_replacement.
2419 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2421         * loop-doloop.c (doloop_modify, doloop_optimize): Use
2422         get_max_loop_iterations.
2424 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2426         * config/arm/aarch-common.c (arm_early_load_addr_dep):
2427         Place comment above function.
2429 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2431         * tree-flow.h: Remove all remaining prototypes, enums and structs that
2432         are not related to tree-cfg.c.
2433         * tree-ssa-address.h: New file.  Relocate prototypes.
2434         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
2435         (addr_for_mem_ref): New.  Combine call to get_address_description and
2436         return addr_for_mem_ref.
2437         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
2438         * tree-ssa-live.h: Adjust prototypes.
2439         * passes.c: Include tree-ssa-live.h.
2440         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
2441         * graphite.c (graphite_transform_loops): Make static.
2442         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2443         make_pass_graphite, pass_data_graphite_transforms,
2444         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
2445         * ipa-pure-const.c (warn_function_noreturn): Make static.
2446         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2447         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2448         Relocate from tree-cfg.c
2449         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
2450         static.
2451         (execute_warn_function_noreturn, gate_warn_function_noreturn,
2452         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
2453         Move to ipa-pure-const.c.
2454         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
2455         Relocate from tree-optimize.c.
2456         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
2457         make_pass_fixup_cfg): Move to tree-cfg.c.
2458         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
2459         Relocate some prototypes.
2460         * tree-data-ref.h (tree_check_data_deps) Add prototype.
2461         * tree-dump.c (dump_function_to_file): Remove prototype.
2462         Add tree-flow.h to the include file.
2463         * tree-dump.h: Remove prototype.
2464         * tree-parloops.h: New File.  Add prototypes.
2465         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
2466         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
2467         from tree-ssa-loop.c.
2468         * tree-predcom.c (run_tree_predictive_commoning,
2469         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
2470         Relocate here from tree-ssa-loop.c.
2471         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
2472         ssa_name_values.release ().
2473         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
2474         (ssa_name_values): Relocate from tree-flow.h.
2475         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
2476         * tree-ssa-loop.c (run_tree_predictive_commoning,
2477         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
2478         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
2479         make_pass_graphite, pass_data_graphite_transforms,
2480         make_pass_graphite_transforms, gate_tree_parallelize_loops,
2481         tree_parallelize_loops, pass_data_parallelize_loops,
2482         make_pass_parallelize_loops): Move to other files.
2483         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
2484         moved here.
2485         * tree.h: Remove prototypes from tree-address.c.
2487 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2489         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
2490         (struct int_tree_map): Move to tree-hasher.h
2491         (SCALE, LABEL, PERCENT): Move to gimple.h
2492         * tree-flow-inline.h: Delete.  Move functions to other files.
2493         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
2494         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
2495         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
2496         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
2497         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
2498         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
2499         * tree-hasher.h: Don't include tree-flow.h.
2500         (struct int_tree_map): Relocate from tree-flow.h.
2501         * tree-sra.c (contains_view_convert_expr_p): Relocate from
2502         tree-flow-inline.h and make static.
2503         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
2504         tree-flow-inline.h.
2505         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
2506         tree-flow-inline.h and make static.
2507         * tree.h (is_global_var, may_be_aliased): Relocate from
2508         tree-flow-inline.h.
2509         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
2510         * value-prof.c: No longer include tree-flow-inline.h.
2511         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
2513 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
2515         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
2516         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
2517         (enum move_pos): Move to tree-ssa-loop-im.h
2518         * cfgloop.h: Move some prototypes.
2519         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
2520         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
2521         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
2522         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
2523         (loop_containing_stmt): Relocate from tree-flow-inline.h.
2524         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
2525         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
2526         (enum move_pos): Relocate here.
2527         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
2528         tree-ssa-loop.c.
2529         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
2530         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
2531         make_pass_lim): Relocate here from tree-ssa-loop.c.
2532         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
2533         tree-ssa-loop.c.
2534         (loop_edge_to_cancel, unloop_loops): Make static.
2535         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2536         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
2537         (tree_complete_unroll, gate_tree_complete_unroll,
2538         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
2539         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2540         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
2541         * tree-ssa-loop-ivopts.c: Remove local prototypes.
2542         (stmt_invariant_in_loop_p): Remove unused function.
2543         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
2544         * tree-ssa-loop-manip.h: New file.  Add prototypes.
2545         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
2546         (gcov_type_to_double_int): Move to cfgloop.h.
2547         (double_int_cmp, bound_index,
2548         estimate_numbers_of_iterations_loop): Make static.
2549         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
2550         (max_loop_iterations): Factor out get_max_loop_iterations.
2551         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
2552         cfgloop.c.
2553         * tree-ssa-loop-niter.h: New file.  Add prototypes.
2554         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
2555         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
2556         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
2557         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
2558         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
2559         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
2560         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
2561         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
2562         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
2563         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
2564         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
2565         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
2566         pass_data_complete_unroll, make_pass_complete_unroll,
2567         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
2568         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
2569         tree-ssa-loop-ivcanon.c.
2570         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
2571         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
2572         tree-ssa-loop-prefetch.c.
2573         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
2574         tree-ssa-loop-im.c.
2575         (get_lsm_tmp_name): Relocate and add suffix parameter.
2576         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
2577         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
2578         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
2579         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
2580         (get_estimated_loop_iterations): Factor out accessor from
2581         estimated_loop_iterations in tree-ssa-loop-niter.c.
2582         (get_max_loop_iterations): Factor out accessor from
2583         _max_loop_iterations in tree-ssa-niter.c.
2584         * loop-unroll.c (decide_unroll_constant_iterations,
2585         decide_unroll_runtime_iterations, decide_peel_simple,
2586         decide_unroll_stupid): Use new get_* accessors.
2588 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
2590         PR tree-optimization/20318
2591         * doc/extend.texi (returns_nonnull): New function attribute.
2592         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
2593         attribute.
2594         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
2595         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
2597 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
2599         PR middle-end/58570
2600         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
2601         false if both components are bitfields.
2603 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2605         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
2606         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
2607         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
2608         * config/aarch64/aarch64.h
2609         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
2610         * config/aarch64/aarch64-simd-builtins.def
2611         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
2613 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2615         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
2617 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2619         * config/aarch64/arm_neon.h (vdiv_f64): Added.
2621 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
2623         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
2624         (vneg_f64): New intrinsic.
2625         (vneg_s8): Asm replaced with C.
2626         (vneg_s16): Likewise.
2627         (vneg_s32): Likewise.
2628         (vneg_s64): New intrinsic.
2629         (vnegq_f32): Asm replaced with C.
2630         (vnegq_f64): Likewise.
2631         (vnegq_s8): Likewise.
2632         (vnegq_s16): Likewise.
2633         (vnegq_s32): Likewise.
2634         (vnegq_s64): Likewise.
2636 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
2638         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
2640 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2642         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
2643         packed stack special handling.
2644         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
2645         back to fixed stack slots for FPRs saved due to stdarg.
2647 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2649         * config/s390/s390.c (s390_frame_info): Restructure function.
2651 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2653         * config/s390/s390.c (struct s390_frame_layout): New field
2654         gpr_save_slots.
2655         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
2656         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
2657         regs_ever_clobbered to char*.
2658         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
2659         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
2660         (s390_register_info): Enable FPR save slots.  Move/Copy some
2661         functionality into ...
2662         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
2663         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
2664         function.
2665         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
2666         not imply a stack frame.
2667         (s390_init_frame_layout): Remove variable clobbered_regs.
2668         (s390_update_register_info): Remove function.
2669         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
2670         cannot be used for register renaming.
2671         (s390_hard_regno_scratch_ok): New function.
2672         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
2673         (s390_initial_elimination_offset): Change offset calculation of
2674         the return address pointer.
2675         (save_gprs): Deal with only r6 being saved from the call-saved regs.
2676         (restore_gprs): Set frame related flag.
2677         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
2678         (s390_emit_prologue): Call s390_register_info instead of
2679         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
2680         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
2681         (s390_optimize_prologue): Call s390_optimize_register_info.
2682         Try to remove also FPR slot save/restore INSNs.  Remove frame
2683         related flags from restore INSNs.
2685 2013-10-08  DJ Delorie  <dj@redhat.com>
2687         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
2688         (movhi): Likewise.
2690         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
2691         avoid conflict with the MI use of %c.
2692         * config/rl78/rl78-real.md: change %c to %C throughout.
2693         * config/rl78/rl78-virt.md: Likewise.
2695 2013-10-08  Jan Hubicka  <jh@suse.cz>
2697         * config/i386/i386.c (ix86_option_override_internal): Switch
2698         to SSE math for -ffast-math when target ISA supports SSE2.
2700 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
2702         * tree-flow.h: Remove some prototypes.
2703         * tree.h: Remove some protypes, add a couple.
2704         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
2705         using_eh_for_cleanups_p): Add interface routines for front ends.
2706         * tree-eh.h: New file.  Add protoptyes.
2707         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
2708         (add_stmt_to_eh_lp_fn): Make static.
2709         (lower_try_finally): Use new using_eh_for_cleanups_p.
2710         * emit-rtl.c: Include tree-eh.h.
2711         * gimple.h: Include tree-eh.h.
2713 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
2715         PR tree-optimization/58480
2716         * tree-vrp.c (infer_nonnull_range): New function.
2717         (infer_value_range): Call infer_nonnull_range.
2719 2013-10-08  Dehao Chen  <dehao@google.com>
2721         PR tree-optimization/58619
2722         * tree-inline.c (copy_phis_for_bb): Combine location data
2723         only if non-null.
2725 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
2727         PR target/58423
2728         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
2729         RTX_FRAME_RELATED_P on INSN.
2731 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2733         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
2734         (altivec_expand_vec_perm_const): Call it.
2736 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2738         * config/rs6000/vector.md (mov<mode>): Emit permuted move
2739         sequences for LE VSX loads and stores at expand time.
2740         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
2741         prototype.
2742         * config/rs6000/rs6000.c (rs6000_const_vec): New.
2743         (rs6000_gen_le_vsx_permute): New.
2744         (rs6000_gen_le_vsx_load): New.
2745         (rs6000_gen_le_vsx_store): New.
2746         (rs6000_gen_le_vsx_move): New.
2747         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
2748         (*vsx_le_perm_load_v4si): New.
2749         (*vsx_le_perm_load_v8hi): New.
2750         (*vsx_le_perm_load_v16qi): New.
2751         (*vsx_le_perm_store_v2di): New.
2752         (*vsx_le_perm_store_v4si): New.
2753         (*vsx_le_perm_store_v8hi): New.
2754         (*vsx_le_perm_store_v16qi): New.
2755         (*vsx_xxpermdi2_le_<mode>): New.
2756         (*vsx_xxpermdi4_le_<mode>): New.
2757         (*vsx_xxpermdi8_le_V8HI): New.
2758         (*vsx_xxpermdi16_le_V16QI): New.
2759         (*vsx_lxvd2x2_le_<mode>): New.
2760         (*vsx_lxvd2x4_le_<mode>): New.
2761         (*vsx_lxvd2x8_le_V8HI): New.
2762         (*vsx_lxvd2x16_le_V16QI): New.
2763         (*vsx_stxvd2x2_le_<mode>): New.
2764         (*vsx_stxvd2x4_le_<mode>): New.
2765         (*vsx_stxvd2x8_le_V8HI): New.
2766         (*vsx_stxvd2x16_le_V16QI): New.
2768 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
2770         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
2772 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2774         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
2775         loop to work also for 31bit ABI.
2776         Save the stack pointer for frame_size > 0.
2778 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
2780         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
2781         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
2782         constraint letters from expanders.
2783         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
2784         retry count to general_operand.
2785         ("tabort"): Give operand 0 a mode.
2786         ("tabort_1"): Add mode and constraint letter for operand 0.
2787         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
2789 2013-10-04  Jeff Law  <law@redhat.com>
2791         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
2793         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
2794         out of ...
2795         (thread_across_edge): Here.  Call it.
2797 2013-10-04  Cary Coutant  <ccoutant@google.com>
2799         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
2800         discarding a location list expression (or a piece of one).
2802 2013-10-03  Jan Hubicka  <jh@suse.cz>
2804         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
2805         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
2806         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
2808 2013-10-03  Jan Hubicka  <jh@suse.cz>
2810         * config/i386/i386.c (ix86_option_override_internal): Do not enable
2811         accumulate-outgoing-args when producing unwind info.
2813 2013-10-03  Wei Mi  <wmi@google.com>
2815         * lra-constraints.c (insert_move_for_subreg): New function
2816         extracted from simplify_operand_subreg.
2817         (simplify_operand_subreg): Add reload for paradoxical subreg.
2819 2013-10-03  Rong Xu  <xur@google.com>
2821         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
2822         the candidate of builtin_expect such that we should fix the
2823         size/time estimation.
2824         (estimate_function_body_sizes): Do the acutally size/time fix-up
2825         for builtin_expect.
2827 2013-10-03  Rong Xu  <xur@google.com>
2829         * predict.c (tree_predict_by_opcode): Get the probability
2830         for builtin_expect from param builtin_expect_probability.
2831         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
2832         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
2833         * doc/invoke.texi: Add documentation for builtin-expect-probability.
2835 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
2837         PR c++/19476
2838         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
2839         * calls.c (alloca_call_p): Use get_callee_fndecl.
2840         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
2841         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
2842         Likewise.
2843         (vrp_visit_stmt): Remove duplicated code.
2845 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
2847         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
2848         ceildf2, btruncdf2, instead of vsx_* name.
2850         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
2851         iterators to only do V2DF and V4SF here.  Move the DF code to
2852         rs6000.md where it is combined with SF mode.  Replace <VSv> with
2853         just 'v' since only vector operations are handled with these insns
2854         after moving the DF support to rs6000.md.
2855         (vsx_sub<mode>3): Likewise.
2856         (vsx_mul<mode>3): Likewise.
2857         (vsx_div<mode>3): Likewise.
2858         (vsx_fre<mode>2): Likewise.
2859         (vsx_neg<mode>2): Likewise.
2860         (vsx_abs<mode>2): Likewise.
2861         (vsx_nabs<mode>2): Likewise.
2862         (vsx_smax<mode>3): Likewise.
2863         (vsx_smin<mode>3): Likewise.
2864         (vsx_sqrt<mode>2): Likewise.
2865         (vsx_rsqrte<mode>2): Likewise.
2866         (vsx_fms<mode>4): Likewise.
2867         (vsx_nfma<mode>4): Likewise.
2868         (vsx_copysign<mode>3): Likewise.
2869         (vsx_btrunc<mode>2): Likewise.
2870         (vsx_floor<mode>2): Likewise.
2871         (vsx_ceil<mode>2): Likewise.
2872         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
2873         (vsx_sminsf3): Likewise.
2874         (vsx_fmadf4): Likewise.
2875         (vsx_fmsdf4): Likewise.
2876         (vsx_nfmadf4): Likewise.
2877         (vsx_nfmsdf4): Likewise.
2878         (vsx_cmpdf_internal1): Likewise.
2880         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
2881         simpler to select whether a target has SPE or traditional floating
2882         point support in iterators.
2883         (TARGET_DF_SPE): Likewise.
2884         (TARGET_SF_FPR): Likewise.
2885         (TARGET_DF_FPR): Likewise.
2886         (TARGET_SF_INSN): Macros to say whether floating point support
2887         exists for a given operation for expanders.
2888         (TARGET_DF_INSN): Likewise.
2890         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
2891         combining of SF/DF mode operations, using both traditional and VSX
2892         registers.
2893         (Fvsx): Likewise.
2894         (Ff): Likewise.
2895         (Fv): Likewise.
2896         (Fs): Likewise.
2897         (Ffre): Likewise.
2898         (FFRE): Likewise.
2899         (abs<mode>2): Combine SF/DF modes using traditional floating point
2900         instructions.  Add support for using the upper DF registers with
2901         VSX support, and SF registers with power8-vector support.  Update
2902         expanders for operations supported by both the SPE and traditional
2903         floating point units.
2904         (abs<mode>2_fpr): Likewise.
2905         (nabs<mode>2): Likewise.
2906         (nabs<mode>2_fpr): Likewise.
2907         (neg<mode>2): Likewise.
2908         (neg<mode>2_fpr): Likewise.
2909         (add<mode>3): Likewise.
2910         (add<mode>3_fpr): Likewise.
2911         (sub<mode>3): Likewise.
2912         (sub<mode>3_fpr): Likewise.
2913         (mul<mode>3): Likewise.
2914         (mul<mode>3_fpr): Likewise.
2915         (div<mode>3): Likewise.
2916         (div<mode>3_fpr): Likewise.
2917         (sqrt<mode>3): Likewise.
2918         (sqrt<mode>3_fpr): Likewise.
2919         (fre<Fs>): Likewise.
2920         (rsqrt<mode>2): Likewise.
2921         (cmp<mode>_fpr): Likewise.
2922         (smax<mode>3): Likewise.
2923         (smin<mode>3): Likewise.
2924         (smax<mode>3_vsx): Likewise.
2925         (smin<mode>3_vsx): Likewise.
2926         (negsf2): Delete SF operations that are merged with DF.
2927         (abssf2): Likewise.
2928         (addsf3): Likewise.
2929         (subsf3): Likewise.
2930         (mulsf3): Likewise.
2931         (divsf3): Likewise.
2932         (fres): Likewise.
2933         (fmasf4_fpr): Likewise.
2934         (fmssf4_fpr): Likewise.
2935         (nfmasf4_fpr): Likewise.
2936         (nfmssf4_fpr): Likewise.
2937         (sqrtsf2): Likewise.
2938         (rsqrtsf_internal1): Likewise.
2939         (smaxsf3): Likewise.
2940         (sminsf3): Likewise.
2941         (cmpsf_internal1): Likewise.
2942         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
2943         (negdf2): Delete DF operations that are merged with SF.
2944         (absdf2): Likewise.
2945         (nabsdf2): Likewise.
2946         (adddf3): Likewise.
2947         (subdf3): Likewise.
2948         (muldf3): Likewise.
2949         (divdf3): Likewise.
2950         (fred): Likewise.
2951         (rsqrtdf_internal1): Likewise.
2952         (fmadf4_fpr): Likewise.
2953         (fmsdf4_fpr): Likewise.
2954         (nfmadf4_fpr): Likewise.
2955         (nfmsdf4_fpr): Likewise.
2956         (sqrtdf2): Likewise.
2957         (smaxdf3): Likewise.
2958         (smindf3): Likewise.
2959         (cmpdf_internal1): Likewise.
2960         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
2961         (btrunc<mode>2): Delete separate expander, and combine with the
2962         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
2963         (btrunc<mode>2_fpr): Likewise.
2964         (ceil<mode>2): Likewise.
2965         (ceil<mode>2_fpr): Likewise.
2966         (floor<mode>2): Likewise.
2967         (floor<mode>2_fpr): Likewise.
2968         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
2969         Add support for using the upper registers with VSX and
2970         power8-vector.  Move insns to be closer to the define_expands. On
2971         VSX systems, prefer the traditional form of FMA over the VSX
2972         version, since the traditional form allows the target not to
2973         overlap with the inputs.
2974         (fms<mode>4_fpr): Likewise.
2975         (nfma<mode>4_fpr): Likewise.
2976         (nfms<mode>4_fpr): Likewise.
2978 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2979             Richard Earnshaw  <richard.earnshaw@arm.com>
2981         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
2982         (struct mult_cost_table): Likewise.
2983         (struct mem_cost_table): Likewise.
2984         (struct fp_cost_table): Likewise.
2985         (struct vector_cost_table): Likewise.
2986         (cpu_cost_table): Likewise.
2987         * config/arm/arm.opt (mold-rts-costs): New option.
2988         (mnew-generic-costs): Likewise.
2989         * config/arm/arm.c (generic_extra_costs): New table.
2990         (cortexa15_extra_costs): Likewise.
2991         (arm_slowmul_tune): Use NULL as new costs.
2992         (arm_fastmul_tune): Likewise.
2993         (arm_strongarm_tune): Likewise.
2994         (arm_xscale_tune): Likewise.
2995         (arm_9e_tune): Likewise.
2996         (arm_v6t2_tune): Likewise.
2997         (arm_cortex_a5_tune): Likewise.
2998         (arm_cortex_a9_tune): Likewise.
2999         (arm_v6m_tune): Likewise.
3000         (arm_fa726te_tune): Likewise.
3001         (arm_cortex_a15_tune): Use cortex15_extra_costs.
3002         (arm_cortex_tune): Use generict_extra_costs.
3003         (shifter_op_p): New function.
3004         (arm_unspec_cost): Likewise.
3005         (LIBCALL_COST): Define.
3006         (arm_new_rtx_costs): New function.
3007         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
3008         table is available. Use old costs otherwise unless mnew-generic-costs
3009         is specified.
3010         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
3011         (cpu_cost_table): Declare.
3013 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3015         PR target/58460
3016         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
3017         (*subs_mul_imm_<mode>)
3018         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
3019         (*sub_<shift>_<mode>)
3020         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
3021         Remove k constraint.
3023 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
3025         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
3026         code.
3027         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
3029 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3031         * predict.c (probably_never_executed): New function.
3032         (probably_never_executed_bb_p): Invoke probably_never_executed.
3033         (probably_never_executed_edge_p): Ditto.
3034         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
3035         Treat profile insanities conservatively.
3037 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
3039         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
3041 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
3043         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
3044         use it.  Use smaller increase for scratch.  Don't increase reject
3045         for early clobber scratch.
3046         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
3047         setting eliminated regs except setting fp from hfp.
3048         (lra_eliminate): Check lra_insn_recog_data on NULL.
3050 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
3052         PR target/58587
3053         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
3054         setting -mvsx-timode by default until the underlying problem is fixed.
3055         (RS6000_CPU, power7 defaults): Likewise.
3057 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3059         * config/x-linux (host-linux.o): Remove header dependencies.
3060         Use $(COMPILE) and $(POSTCOMPILE).
3061         * config/t-linux-android (linux-android.o): Ditto.
3063 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3065         * Makefile.in (expmed.o-warn): Remove.
3067 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3069         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
3070         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
3072 2013-10-02  Teresa Johnson  <tejohnson@google.com>
3074         * dojump.c (do_jump_1): Divide probability between
3075         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
3077 2013-10-02  Tom Tromey  <tromey@redhat.com>
3079         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
3081 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3083         * tree-flow.h: Remove some prototypes.
3084         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
3085         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
3086         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
3087         mark_virtual_phi_result_for_renaming): Relocate here.
3088         * tree-into-ssa.h: Add prototypes.
3089         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
3090         single_pred_before_succ_order.
3091         (blocks_in_phiopt_order): Rename and move to cfganal.c.
3092         (nonfreeing_call_p) Move to gimple.c.
3093         * cfganal.c (single_pred_before_succ_order): Move and renamed from
3094         tree-ssa-phiopt.c.
3095         * basic-block.h (single_pred_before_succ_order): Add prototype.
3096         * gimple.c (nonfreeing_call_p): Relocate here.
3097         * gimple.h: Add prototype.
3098         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
3099         * tree-ssa-dom.h: New file.  Relocate prototypes here.
3100         * tree-ssa.h: Include tree-ssa-dom.h.
3102 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
3104         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
3105         Use $(COMPILE) and $(POSTCOMPILE).
3107         * config/alpha/x-alpha (driver-alpha.o): Ditto.
3109 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3111         * tree-flow.h: Remove some prototypes.
3112         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
3113         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3114         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
3115         propagate_tree_value*): Move from here to...
3116         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
3117         propagate_tree_value*): Relocate here.
3118         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
3119         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
3120         * gimple-fold.c: Remove gimple-fold.h from include list.
3121         * tree-vrp.c: Remove gimple-fold.h from include list.
3122         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
3123         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
3124         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
3125         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
3126         * sese.c: Add tree-ssa-propagate.h to include list.
3128 2013-10-02  Richard Biener  <rguenther@suse.de>
3130         * tree-loop-distribution.c: Include tree-vectorizer.h for
3131         find_loop_location.
3132         (enum partition_kind): Remove PKIND_REDUCTION.
3133         (struct partition_s): Remove has_writes member, add reduction_p member.
3134         (partition_alloc): Adjust.
3135         (partition_builtin_p): Likewise.
3136         (partition_has_writes): Remove.
3137         (partition_reduction_p): New function.
3138         (partition_merge_into): Likewise.
3139         (generate_code_for_partition): Commonize builtin partition
3140         handling tail.
3141         (rdg_cannot_recompute_vertex_p): Remove.
3142         (already_processed_vertex_p): Likewise.
3143         (rdg_flag_vertex): Do not set has_writes.
3144         (classify_partition): Adjust.
3145         (rdg_build_partitions): Do not set has_writes, treat all
3146         partitions as useful.
3147         (distribute_loop): Record number of library calls generated.  Adjust.
3148         (tree_loop_distribution): Report number of loops and library
3149         calls generated as opt-info.
3151 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
3153         * tree-flow.h: Include new .h files.  Move prototypes.
3154         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
3155         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
3156         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
3157         * tree-pretty-print.h: Add prototypes from tree-flow.h.
3158         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
3159         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
3160         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
3161         * tree.h (get_ref_base_and_extent): Move prototype out.
3162         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
3163         tree-dfa.h.
3164         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
3165         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
3166         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
3167         * tree-scalar-evolution.c: Include tree.h.
3168         * sese.c: Include tree.h.
3169         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
3170         * dwarf2out.c: Include tree-dfa.h.
3171         * tree-chrec.c: Include tree.h.
3172         * tree-data-ref.c: Include tree.h.
3174 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
3176         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
3177         Fix whitespace.
3179 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
3181         * config/t-sol2 (sol2-c.o): Remove header dependencies.
3182         Use $(COMPILE) and $(POSTCOMPILE).
3183         (sol2-cxx.o): Likewise.
3184         (sol2-stubs.o): Likewise.
3185         (sol2.o): Likewise.
3186         * config/x-solaris (host-solaris.o): Likewise.
3188         * config/sparc/t-sparc (sparc.o): Remove.
3189         (sparc-c.o): Remove header dependencies.
3190         Use $(COMPILE) and $(POSTCOMPILE).
3191         * config/sparc/x-sparc: Likewise.
3193 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
3195         * config/arc/arc-opts.h: Add 2013 to Copyright years.
3196         * config/arc/arc700.md: Likewise.
3197         * config/arc/arc-modes.def: Likewise.
3198         * config/arc/arc-simd.h: Likewise.
3199         * config/arc/t-arc-uClibc: Likewise.
3200         * config/arc/t-arc-newlib: Likewise.
3202 2013-10-02  Renlin Li  <renlin.li@arm.com>
3204         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
3205         plus_constant.
3206         (aarch64_expand_epilogue): Likewise.
3208 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3209             Yufeng Zhang  <yufeng.zhang@arm.com>
3211         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
3212         declaration.
3213         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
3214         'base_in' represent a conversion and legal_cast_p_1 holds; set
3215         'base_in' with the returned value from get_unwidened.
3217 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3219         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
3220         plus_constant is not used.
3222 2013-10-01  Wei Mi  <wmi@google.com>
3224         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
3225         * config/i386/i386.md: Add define_peephole2 to
3226         break partial reg stall for cvtss2sd/cvtsd2ss.
3228 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3230         * config/arc/arc.c (pass_arc_ifcvt::clone):
3231         Update for ctxt_ -> m_ctxt change.
3233 2013-10-01  Jeff Law  <law@redhat.com>
3235         * tree-ssa-threadupdate.c (struct redirection_data): Delete
3236         outgoing_edge and intermediate_edge fields.  Instead store the path.
3237         (redirection_data::hash): Hash on the last edge's destination index.
3238         (redirection_data::equal): Check the entire thread path.
3239         (lookup_redirectio_data): Corresponding changes.
3240         (create_edge_and_update_destination_phis): Likewise.
3241         (thread_single_edge): Likewise.
3243 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3244             Diego Novillo <dnovillo@google.com>
3246         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
3247         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
3248         (vld32wh_insn, vld32wl_insn): Delete commented-out old
3249         versions of these patterns.
3251         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
3252         (__builtin_arc_aligned): Likewise.
3254         * config/arc/arc.md: Expand adc_0 comment stating the intended
3255         purpose and why it isn't ready.
3256         Replace commented out call_value_via_label_mixed with a
3257         plain comment about bl_s.
3259         * config/arc/arc.c (stdio.h): Don't include directly.
3260         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
3261         Remove if (1) condition.
3262         (arc_encode_section_info): Fix comment.
3264 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3266         * config/arc/arc.c (arc_conditional_register_usage):
3267         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
3268         Also set reg_alloc_order for DMA config regs.
3270 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3271             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3273         * doc/install.texi (--with-cpu): Mention ARC.
3274         (arc-*-elf32): New paragraph.
3275         (arc-linux-uclibc): Likewise.
3276         * doc/md.texi (Machine Constraints): Add ARC part.
3277         * doc/invoke.texi: (menu): Add ARC Options.
3278         (Machine Dependent Options) <ARC Options>: Add synopsis.
3279         (node ARC Options): Add.
3280         * doc/extend.texi (long_call / short_call attribute): Add ARC.
3281         (ARC Built-in Functions): New section defining
3282         generic ARC built-in functions.
3283         (ARC SIMD Built-in Functions): New section defining SIMD specific
3284         built-in functions.
3285         (Declaring Attributes of Functions): Extended
3286         description of short_call and long_call attributes for ARC and
3287         added index entries.
3289 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
3290             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
3291             Joern Rennecke  <joern.rennecke@embecosm.com>
3292             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
3293             Brendan Kehoe  <brendan@zen.org>
3294             Michael Eager  <eager@eagercon.com>
3295             Simon Cook  <simon.cook@embecosm.com>
3296             Jeremy Bennett  <jeremy.bennett@embecosm.com>
3298         * config/arc, common/config/arc: New directories.
3300 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
3301             Brendan Kehoe  <brendan@zen.org>
3302             Simon Cook  <simon.cook@embecosm.com>
3304         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
3306 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3308         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
3309         * tree-ssa-coalesce.h: New. Move prototype to here.
3310         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
3311         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
3312         (gimple_can_coalesce_p): Move to...
3313         * gimple.c (gimple_can_coalesce_p): Here.
3315 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
3317         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
3318         (dump_decl_set): Move to gimple.c.
3319         * gimple.h: Don't include tree-ssa-operands.h.
3320         (dump_decl_set): Add prototype.
3321         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3322         Move to gimple-ssa.h.
3323         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
3324         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
3325         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
3326         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
3327         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
3328         than PHI_ARG_DEF.
3329         (dump_decl_set): Relocate here.
3330         * gimple-ssa.h: New file.
3331         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
3332         Relocate from gimple.h.
3333         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
3334         * tree-ssa-operands.c (swap_ssa_operands): Rename from
3335         swap_tree_operands and remove non-ssa path.
3336         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
3337         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
3338         swap_ssa_operands.
3339         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
3340         vect_is_simple_reduction_1): Use swap_ssa_operands.
3341         * tree-flow.h: Move various prototypes to tree-phinodes.h.
3342         (enum need_phi_state): Move to tree-into-ssa.c.
3343         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3344         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
3345         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
3346         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
3347         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3348         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3349         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3350         num_imm_uses): Move to ssa-iterators.h.
3351         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
3352         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
3353         tree-phinodes.h.
3354         (op_iter_done, op_iter_next_def, op_iter_next_tree,
3355         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
3356         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
3357         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
3358         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
3359         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
3360         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
3361         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
3362         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
3363         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
3364         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
3365         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
3366         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
3367         (set_phi_nodes): Move to tree-phinodes.h.
3368         * tree-ssa-operands.h (enum ssa_op_iter_type,
3369         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
3370         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
3371         (dump_decl_set): Remove prototype.
3372         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
3373         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
3374         (set_phi_nodes): Relocate from tree-flow-inline.h.
3375         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
3376         tree-flow-inline.h
3377         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
3378         include list.  Temporarily add gimple.h to include list.
3379         * ssa-iterators.h: New file.
3380         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
3381         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
3382         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
3383         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
3384         Relocate from tree-ssa-operands.h.
3385         (delink_imm_use, link_imm_use_to_list, link_imm_use,
3386         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
3387         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
3388         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
3389         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
3390         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
3391         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
3392         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
3393         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
3394         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
3395         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
3396         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
3397         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
3398         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
3399         Relocate from tree-flow-inline.h.
3400         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
3402 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
3404         * aarch64-simd.md
3405         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
3406         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
3407         Insert '\t' to output template.
3408         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
3409         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
3410         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
3411         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
3413 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
3415         * doc/install.texi (Host/target specific installation notes for GCC):
3416         Put @anchor before @heading.
3417         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
3418         Use @email for email addresses.
3420 2013-10-01  Jeff Law  <law@redhat.com>
3422         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
3423         a vec.  Only delete the path if we create one without successfully
3424         registering a jump thread.
3425         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
3426         as a pointer.
3427         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
3428         (paths): New vector of jump threading paths.
3429         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
3430         (THREAD_PATH): New accessor macro for the entire thread path.
3431         (lookup_redirection_data): Get intermediate and final outgoing edge
3432         from the thread path.
3433         (create_edge_and_update_destination_phis): Copy the threading path.
3434         (ssa_fix_duplicate_block_edges): Get edges and block types from the
3435         jump threading path.
3436         (ssa_redirect_edges): Get edges and block types from the jump threading
3437         path.  Free the path vector.
3438         (thread_block): Get edges from the jump threading path.  Look at the
3439         entire path to see if we thread to a loop exit.  If we cancel a jump
3440         thread request, then free the path vector.
3441         (thread_single_edge): Get edges and block types from the jump threading
3442         path.  Free the path vector.
3443         (thread_through_loop_header): Get edges and block types from the jump
3444         threading path.  Free the path vector.
3445         (mark_threaded_blocks): Iterate over the vector of paths and store
3446         the path on the appropriate edge.  Get edges and block types from the
3447         jump threading path.
3448         (mark_threaded_blocks): Get edges and block types from the jump
3449         threading path.  Free the path vector.
3450         (thread_through_all_blocks): Use the vector of paths rather than
3451         a vector of 3-edge sets.
3452         (register_jump_thread): Accept pointer to a path vector rather
3453         than the path vector itself.  Store the path vector for later use.
3454         Simplify.
3456 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
3457             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
3459         PR target/58574
3460         * config/s390/s390.c (s390_split_branches): Modify check for table
3461         jump insns.
3462         (s390_chunkify_start): Rearrange table jump insn check in order to
3463         deal with compare and branch insns correctly.
3465 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
3467         PR target/58578
3468         Revert
3469         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
3470         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
3471         define_insn_and_split.
3472         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
3473         (shiftsi3_compare): New pattern.
3474         (rrx): New pattern.
3475         * config/arm/unspecs.md (UNSPEC_RRX): New.
3477 2013-10-01  Alan Modra  <amodra@gmail.com>
3479         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
3480         casing inout operands.
3482 2013-10-01  Richard Biener  <rguenther@suse.de>
3484         PR tree-optimization/58553
3485         * tree-loop-distribution.c (struct partition_s): Add niter member.
3486         (classify_partition): Populate niter member for the partition
3487         and properly identify whether the relevant store happens before
3488         or after the loop exit.
3489         (generate_memset_builtin): Use niter member from the partition.
3490         (generate_memcpy_builtin): Likewise.
3492 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3494         * vec.h (vec_prefix, vec): Prefix member names with "m_".
3495         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
3497 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
3499         * basic-block.h (edge_list): Prefix member names with "m_".
3500         * context.h (context): Likewise.
3501         * domwalk.h (dom_walker): Likewise.
3502         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
3503         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
3504         * hash-table.h (hash_table): Likewise.
3505         * machmode.h (bit_field_mode_iterator): Likewise.
3506         * pass_manager.h (pass_list): Likewise.
3507         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
3508         * tree-pass.h (pass_data): Likewise.
3509         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
3510         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
3511         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
3512         * asan.c (pass_data_asan): Update accordingly.
3513         * cfganal.c (control_dependences::find_control_dependence): Likewise.
3514         (control_dependences::control_dependences): Likewise.
3515         (control_dependences::~control_dependences): Likewise.
3516         (control_dependences::~control_dependences): Likewise.
3517         (control_dependences::get_edges_dependent_on): Likewise.
3518         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
3519         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
3520         * context.c (gcc::context::context): Likewise.
3521         * cprop.c (pass_rtl_cprop::clone): Likewise.
3522         * domwalk.c (dom_walker::walk): Likewise.
3523         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
3524         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
3525         * mode-switching.c (pass_mode_switching::clone): Likewise.
3526         * passes.c (opt_pass::opt_pass): Likewise.
3527         (pass_manager::pass_manager): Likewise.
3528         * predict.c (pass_strip_predict_hints::clone): Likewise.
3529         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
3530         (pass_split_all_insns::clone): Likewise.
3531         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
3532         Likewise.
3533         (bit_field_mode_iterator::next_mode): Likewise.
3534         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
3535         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
3536         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
3537         * tree-complex.c (pass_lower_complex::clone): Likewise.
3538         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
3539         * tree-object-size.c (pass_object_sizes::clone): Likewise.
3540         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
3541         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
3542         (pass_fold_builtins::clone): Likewise.
3543         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
3544         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
3545         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
3546         (pass_cd_dce::clone): Likewise.
3547         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
3548         (pass_phi_only_cprop::clone): Likewise.
3549         * tree-ssa-dse.c (pass_dse::clone): Likewise.
3550         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
3551         * tree-ssa-loop.c (pass_lim::clone): Likewise.
3552         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
3553         * tree-ssa-pre.c (pass_fre::clone): Likewise.
3554         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
3555         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
3556         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
3557         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
3558         * tree-vrp.c (pass_vrp::clone): Likewise.
3559         * tsan.c (pass_tsan::clone): Likewise.
3561 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
3563         PR middle-end/58564
3564         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
3565         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
3567         PR middle-end/58564
3568         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
3569         optimization, punt if sign_bit_p looked through any zero extension.
3571 2013-09-30  Teresa Johnson  <tejohnson@google.com>
3573         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
3574         Update redirected out edge count in joiner case.
3575         (ssa_redirect_edges): Common the joiner and non-joiner cases
3576         so that joiner case gets profile updates.
3578 2013-09-30  Richard Biener  <rguenther@suse.de>
3580         PR tree-optimization/58554
3581         * tree-loop-distribution.c (classify_partition): Require
3582         unconditionally executed stores for memcpy and memset recognition.
3583         (tree_loop_distribution): Calculate dominance info.
3585 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
3587         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
3588         (NO_PROFILE_COUNTERS): Likewise.
3589         (PROFILE_HOOK): Likewise.
3590         (FUNCTION_PROFILER): Likewise.
3591         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
3593 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
3595         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
3596         calls and defines in TARGET_MACHO conditional.
3597         (load_macho_picbase_di): Likewise.
3598         (reload_macho_picbase): Likewise.
3599         (reload_macho_picbase_si): Likewise.
3600         (reload_macho_picbase_di): Likewise.
3601         (nonlocal_goto_receiver): Likewise.
3603 2013-09-30  Nick Clifton  <nickc@redhat.com>
3605         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
3606         that use the MSP430X ISA.
3607         (msp430_option_override): Scan -mmcu command line option for any
3608         MCU name that supports the MSP430X ISA.
3609         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
3610         -mmcu options which enable the MSP430X ISA.
3612 2013-09-30  Richard Biener  <rguenther@suse.de>
3614         PR middle-end/58532
3615         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
3616         before looking for setjmp-like calls.
3618 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3620         PR target/10901
3621         * config/darwin-protos.h (machopic_get_function_picbase): New.
3622         * config/darwin.c (machopic_get_function_picbase): New.
3623         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
3624         label for a new func.  (load_macho_picbase_di): Likewise.
3625         (reload_macho_picbase): New expand.
3626         (reload_macho_picbase_si): New insn.
3627         (reload_macho_picbase_di): New insn.
3628         (nonlocal_goto_receiver): New define and split.
3629         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
3630         (unspecv enum): Add UNSPECV_NLGR.
3632 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3634         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
3635         that altivec registers are correctly sized on Darwin.
3637 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
3639         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
3640         darwin-driver.o): Use COMPILE and POSTCOMPILE.
3641         * config/x-darwin (host-darwin.o): Likewise.
3642         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
3643         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
3644         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
3646 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
3648         * doc/invoke.texi: Fix usage of @tie{} command.
3650 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
3652         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
3654 2013-09-28  Jan Hubicka  <jh@suse.cz>
3656         * config/i386/x86-tune.def: Add documentation for each of the options;
3657         add whitespace.
3659 2013-09-28  Jan Hubicka  <jh@suse.cz>
3661         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
3662         generic.
3663         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
3664         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
3665         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
3666         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
3667         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
3669 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
3671         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
3672         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
3673         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
3674         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
3675         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
3676         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
3677         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
3678         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
3679         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
3680         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
3681         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
3682         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
3683         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
3684         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
3685         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
3686         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
3687         gimple.h, godump.c, graphite-clast-to-gimple.c,
3688         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
3689         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
3690         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
3691         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
3692         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
3693         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
3694         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
3695         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
3696         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
3697         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
3698         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
3699         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
3700         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
3701         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
3702         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
3703         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
3704         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
3705         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
3706         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
3707         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
3708         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
3709         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
3710         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
3711         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
3712         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
3713         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
3714         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
3715         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
3716         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
3717         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
3718         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
3719         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
3720         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
3721         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
3722         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
3723         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
3724         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
3725         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
3726         whitespace before "(".
3728 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
3730         * expr.h (extract_bit_field): Remove packedp parameter.
3731         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
3732         from forward declaration.
3733         (store_split_bit_field): Remove packedp arg from calls to
3734         extract_fixed_bit_field.
3735         (extract_bit_field_1): Remove packedp parameter and packedp
3736         argument from recursive calls and calls to extract_fixed_bit_field.
3737         (extract_bit_field): Remove packedp parameter and corresponding
3738         arg to extract_bit_field_1.
3739         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
3740         to issue warnings.
3741         (extract_split_bit_field): Remove packedp arg from call to
3742         extract_fixed_bit_field.
3743         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
3744         (copy_blkmode_from_reg): Likewise.
3745         (copy_blkmode_to_reg): Likewise.
3746         (read_complex_part): Likewise.
3747         (store_field): Likewise.
3748         (expand_expr_real_1): Likewise.
3749         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
3750         to extract_bit_field.
3751         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
3752         call to extract_bit_field.
3753         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
3754         call to extract_bit_field.
3755         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
3756         and special packedp behavior from -fstrict-volatile-bitfields
3757         documentation.
3759 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3761         * lra-eliminations.c (init_elim_table): Guard value_p.
3763 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
3765         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
3766         DFmode, DImode, and SFmode in the upper VSX registers based on the
3767         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
3768         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
3769         of the VSX register handling.
3771         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
3772         (f32_sv): Likewise.
3773         (zero_extendsidi2_lfiwzx): Add support for loading into the
3774         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
3775         only do VSX memory options on Altivec registers.
3776         (extendsidi2_lfiwax): Likewise.
3777         (extendsfdf2_fpr): Likewise.
3778         (mov<mode>_hardfloat, SF/SD modes): Likewise.
3779         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
3780         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
3781         (movdi_internal64): Likewise.
3783 2013-09-27  Xinliang David Li  <davidxl@google.com>
3785         * opts.c (finish_options): Adjust parameters
3786         according to vect cost model.
3787         (common_handle_option): Set dynamic vect cost
3788         model for FDO.
3789         targhooks.c (default_add_stmt_cost): Compute stmt cost
3790         unconditionally.
3791         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
3792         Use helper function.
3793         * tree-vectorizer.h (unlimited_cost_model): New function.
3794         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
3795         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
3796         function.
3797         (vect_enhance_data_refs_alignment): Ditto.
3798         * flag-types.h: New enum.
3799         * common/config/i386/i386-common.c (ix86_option_init_struct):
3800         No need to initialize vect_cost_model flag.
3801         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
3802         unconditionally.
3804 2013-09-27  Diego Novillo  <dnovillo@google.com>
3806         * gimple.h (enum ssa_mode): Remove.
3808 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
3810         * cfgloop.h (number_of_loops): Fix typo in check for null.
3812 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
3814         PR middle-end/58551
3815         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
3816         are children of outermost saved_cfun's loop, and set it up to
3817         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
3818         if loop != loop0 and SESE region contains bbs that belong to loop0.
3820 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3822         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
3823         (binary_scale_code_p, get_base_term, get_index_term): New functions.
3824         (set_address_segment, set_address_base, set_address_index)
3825         (set_address_disp): Accept the argument unconditionally.
3826         (baseness): Remove must_be_base_p and must_be_index_p checks.
3827         (decompose_normal_address): Classify as much as possible in the
3828         main loop.
3830 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
3832         * cse.c (count_reg_usage): Handle INT_LIST.
3833         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
3834         * reginfo.c (reg_scan_mark_refs): Likewise.
3835         * reload1.c (eliminate_regs_1): Likewise.
3837 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
3839         PR middle-end/58547
3840         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
3841         signed.
3843 2013-09-27  Richard Biener  <rguenther@suse.de>
3845         PR tree-optimization/58459
3846         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
3847         restriction not propagating into loops.
3849 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
3851         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
3852         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
3853         * doc/tree-ssa.texi (Walking use-def chains): Delete.
3855 2013-09-26  Richard Biener  <rguenther@suse.de>
3857         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
3859 2013-09-26  Richard Biener  <rguenther@suse.de>
3861         * alias.h (component_uses_parent_alias_set): Rename to ...
3862         (component_uses_parent_alias_set_from): ... this.
3863         * alias.c (component_uses_parent_alias_set): Rename to ...
3864         (component_uses_parent_alias_set_from): ... this and return
3865         the desired parent.
3866         (reference_alias_ptr_type_1): Use the result from
3867         component_uses_parent_alias_set_from instead of stripping
3868         components one at a time.
3869         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
3871 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
3873         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
3874         Move prototypes to...
3875         * tree-ssa-ter.h: New File.  Move prototypes here.
3876         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
3877         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
3878         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
3879         from is_replaceable_p.
3880         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
3881         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
3882         (process_replaceable): Use ter_is_replaceable_p.
3883         (find_replaceable_in_bb): Use ter_is_replaceable_p.
3884         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
3885         newly refactored ssa_is_replaceable_p.
3886         * cfgexpand.c: Include tree-outof-ssa.h.
3887         * ssaexpand.h: Delete.
3889 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
3891         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
3892         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
3893         tree-ssa.c
3894         (create_gimple_tmp): Delete.
3895         (get_expr_type, build_assign, build_type_cast): Move to...
3896         * gimple-builder.c: New File.
3897         (get_expr_type): Relocate from gimple.c.
3898         (build_assign, build_type_cast): Change to only create ssanames.
3899         * gimple.h: Move prototypes to...
3900         * gimple-builder.h: New File. Here.
3901         * tree-ssa.h: And here.
3902         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
3903         count_uses_and_derefs): Relocate from gimple.c.
3904         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
3905         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
3906         * tree-ssa-math-opts (execute_cse_reciprocals): Use
3907         gimple_replace_ssa_lhs.
3908         * asan.c: Include gimple-builder.h.
3909         * Makefile.in: Add gimple-builder.o.
3911 2013-09-26  Richard Biener  <rguenther@suse.de>
3913         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
3914         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
3915         (loe_visit_block): Use gcc_checking_assert.
3916         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
3917         gimple_assign_ssa_name_copy_p.
3918         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
3919         change.
3921 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
3923         * config/rs6000/t-rs6000 (rs6000.o): Remove.
3924         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
3926 2013-09-26  Richard Biener  <rguenther@suse.de>
3928         PR tree-optimization/58539
3929         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
3930         the fact that debug statements are not taking part in loop-closed
3931         SSA construction.
3933 2013-09-26  Nick Clifton  <nickc@redhat.com>
3935         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
3936         time warning message.
3937         (msp430_print_operand_raw): Delete unused letter parameter.
3938         (TARGET_PRINT_OPERAND_ADDRESS): Define.
3939         (msp430_print_operand_address): New function.
3940         (msp430_print_operand): Move address printing code from here to
3941         new function.
3942         * config/msp430/msp430.md (movsipsi2): Add comment in generated
3943         assembler.
3944         (zero_extendpsisi2): Likewise.
3945         (extendpsisi2): New pattern.
3946         (andneghi3): New pattern.
3948 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
3950         * config/aarch64/aarch64.opt (mlra): New option.
3951         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
3952         (TARGET_LRA_P): Define.
3954 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
3956         * expr.c (expand_assignment): Remove obsolete comment.
3958 2013-09-25  Jeff Law  <law@redhat.com>
3960         * tree-flow.h (thread_through_all_blocks): Prototype moved into
3961         tree-ssa-threadupdate.h.
3962         (register_jump_thread): Similarly.
3963         * tree-ssa-threadupdate.h: New header file.
3964         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
3965         * tree-vrp.c: Likewise.
3966         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
3967         (thread_around_empty_blocks): Change type of path vector argument to
3968         an edge,type pair from just an edge.  Initialize both elements when
3969         appending to a jump threading path.  Tweak references to elements
3970         appropriately.
3971         (thread_across_edge): Similarly.  Release memory for the elements
3972         as needed.
3973         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
3974         (dump_jump_thread_path): New function broken out from
3975         register_jump_thread.
3976         (register_jump_thread): Use dump_jump_thread_path.  Change type of
3977         path vector entries.  Search the path for NULL edges and dump
3978         the path if one is found.  Tweak the conversion of path to 3-edge
3979         form to use the block copy type information embedded in the path.
3981 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3983         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
3985 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
3986             Vladimir Makarov  <vmakarov@redhat.com>
3988         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
3989         from the least significant bit.
3990         (strip_address_mutations): Add bitfield operations handling.
3991         (must_be_index_p): Add shifting and rotate operations handling.
3992         (set_address_base): Use must_be_base_p predicate.
3993         (set_address_index): Use must_be_index_p predicate.
3995 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
3996             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
3997             Sergey Lega  <sergey.s.lega@intel.com>
3998             Anna Tikhonova  <anna.tikhonova@intel.com>
3999             Ilya Tocar  <ilya.tocar@intel.com>
4000             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4001             Ilya Verbin  <ilya.verbin@intel.com>
4002             Kirill Yukhin  <kirill.yukhin@intel.com>
4003             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4005         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
4006         Use new names.
4007         (ix86_expand_vector_move_misalign): Support new unaligned load and
4008         stores and use new names.
4009         (CODE_FOR_sse2_storedqu): Rename to ...
4010         (CODE_FOR_sse2_storedquv16qi): ... this.
4011         (CODE_FOR_sse2_loaddqu): Rename to ...
4012         (CODE_FOR_sse2_loaddquv16qi): ... this.
4013         (CODE_FOR_avx_loaddqu256): Rename to ...
4014         (CODE_FOR_avx_loaddquv32qi): ... this.
4015         (CODE_FOR_avx_storedqu256): Rename to ...
4016         (CODE_FOR_avx_storedquv32qi): ... this.
4017         * config/i386/i386.md (fpint_logic): New.
4018         * config/i386/sse.md (VMOVE): Extend for AVX512.
4019         (VF): Ditto.
4020         (VF_128_256): New.
4021         (VF_512): Ditto.
4022         (VI_UNALIGNED_LOADSTORE): Ditto.
4023         (sse2_avx_avx512f): Ditto.
4024         (sse2_avx2): Extend for AVX512.
4025         (sse4_1_avx2): Ditto.
4026         (avx2_avx512f): New.
4027         (sse): Extend for AVX512.
4028         (sse2): Ditto.
4029         (sse4_1): Ditto.
4030         (avxsizesuffix): Ditto.
4031         (sseintvecmode): Ditto.
4032         (ssePSmode): Ditto.
4033         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
4034         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
4035         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4036         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
4037         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
4038         (<sse2_avx_avx512f>_storedqu<mode): ... this.
4039         (<sse>_movnt<mode>): Replace constraint "x" with "v".
4040         (STORENT_MODE): Extend for AVX512.
4041         (*absneg<mode>2): Replace constraint "x" with "v".
4042         (*mul<mode>3): Ditto.
4043         (*ieee_smin<mode>3): Ditto.
4044         (*ieee_smax<mode>3): Ditto.
4045         (avx_cmp<mode>3): Replace VF with VF_128_256.
4046         (*<sse>_maskcmp<mode>3_comm): Ditto.
4047         (<sse>_maskcmp<mode>3): Ditto.
4048         (<sse>_andnot<mode>3): Extend for AVX512.
4049         (<code><mode>3, anylogic): Replace VF with VF_128_256.
4050         (<code><mode>3, fpint_logic): New.
4051         (*<code><mode>3): Extend for AVX512.
4052         (avx512flogicsuff): New.
4053         (avx512f_<logic><mode>): Ditto.
4054         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
4055         VF_128_256.
4056         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
4057         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
4058         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
4059         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
4060         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
4061         (xop_vpermil2<mode>3): Ditto.
4062         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
4063         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
4064         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
4065         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
4067 2013-09-25  Tom Tromey  <tromey@redhat.com>
4069         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
4070         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
4071         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
4072         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
4073         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
4074         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
4075         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
4076         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
4077         (GRAPHITE_HTAB_H): Remove.
4079 2013-09-25  Tom Tromey  <tromey@redhat.com>
4081         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
4083 2013-09-25  Tom Tromey  <tromey@redhat.com>
4085         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
4087 2013-09-25  Tom Tromey  <tromey@redhat.com>
4089         * config/i386/t-i386 (i386.o): Remove.
4090         (i386-c.o): Use COMPILE and POSTCOMPILE.
4092 2013-09-25  Tom Tromey  <tromey@redhat.com>
4094         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
4096 2013-09-25  Tom Tromey  <tromey@redhat.com>
4098         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
4099         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
4100         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
4101         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
4102         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
4103         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
4104         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
4105         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
4106         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
4107         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
4108         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
4109         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
4110         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
4111         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
4112         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
4113         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
4114         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
4115         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
4116         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
4117         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
4118         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
4119         (tree-ssa-pre.o, tree-ssa-sccvn.o)
4120         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
4121         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
4122         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
4123         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
4124         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
4125         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
4126         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
4127         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
4128         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
4129         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
4130         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
4131         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
4132         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
4133         (graphite-clast-to-gimple.o, graphite-dependences.o)
4134         (graphite-interchange.o, graphite-poly.o)
4135         (graphite-scop-detection.o, graphite-sese-to-poly.o)
4136         (graphite-optimize-isl.o, tree-vect-loop.o)
4137         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
4138         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
4139         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
4140         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
4141         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
4142         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
4143         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
4144         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
4145         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
4146         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
4147         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
4148         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
4149         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
4150         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
4151         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
4152         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
4153         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
4154         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
4155         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
4156         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
4157         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
4158         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
4159         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
4160         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
4161         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
4162         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
4163         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
4164         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
4165         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
4166         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
4167         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
4168         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
4169         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
4170         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
4171         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
4172         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
4173         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
4174         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
4175         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
4176         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
4177         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
4178         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
4179         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
4180         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
4181         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
4182         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
4183         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
4184         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
4185         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
4186         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
4187         (gcov-dump.o): Remove.
4188         (default-c.o): Use COMPILE and POSTCOMPILE.
4189         (CFLAGS-gcc.o): New variable.
4190         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
4192 2013-09-25  Tom Tromey  <tromey@redhat.com>
4194         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
4195         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
4196         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
4197         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
4198         (c-family/c-pragma.o, c-family/c-pretty-print.o)
4199         (c-family/c-semantics.o, c-family/c-ada-spec.o)
4200         (c-family/array-notation-common.o, c-family/stub-objc.o)
4201         (c-family/c-ubsan.o): Remove.
4203 2013-09-25  Tom Tromey  <tromey@redhat.com>
4205         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
4207 2013-09-25  Tom Tromey  <tromey@redhat.com>
4209         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
4210         to add -DENABLE_SHARED_LIBGCC.
4211         (gcc.o): Don't use subshell.
4213 2013-09-25  Tom Tromey  <tromey@redhat.com>
4215         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
4216         * configure.ac: Don't invoke AM_PROG_CC_C_O.
4217         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
4218         * configure, config.in: Rebuild.
4220 2013-09-25  Tom Tromey  <tromey@redhat.com>
4222         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
4223         (COMPILE, POSTCOMPILE): New variables.
4224         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
4225         (DEPFILES): New variable.
4226         Include ".Po" files.
4227         * configure.ac: Add checks for dependency checking.
4228         * configure, aclocal.m4: Regenerate.
4230 2013-09-25  Tom Tromey  <tromey@redhat.com>
4232         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
4233         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
4235 2013-09-25  Tom Tromey  <tromey@redhat.com>
4237         * Makefile.in (generated_files): Add options.h,
4238         target-hooks-def.h, insn-opinit.h,
4239         common/common-target-hooks-def.h, pass-instances.def,
4240         c-family/c-target-hooks-def.h.
4242 2013-09-25  Jeff Law  <law@redhat.com>
4244         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
4245         than foo[foo.length () - 1] to access last member in a vec.
4246         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
4248 2013-09-25  Richard Biener  <rguenther@suse.de>
4250         PR middle-end/58521
4251         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
4253 2013-09-25  Jan Hubicka  <jh@suse.cz>
4255         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
4256         test.
4258 2013-09-25  Marek Polacek  <polacek@redhat.com>
4260         PR sanitizer/58420
4261         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
4262         when determining the type name.
4264 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
4266         * config/sh/sh.md: Fix formatting.
4268 2013-09-24  Xinliang David Li  <davidxl@google.com>
4270         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
4271         max peel iterations parameter.
4272         * param.def: New parameter.
4273         * doc/invoke.texi: Document New parameter.
4275 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
4277         * gimple-pretty-print.c: Various whitespace tweaks.
4278         * tree-core.h: Likewise.
4279         * tree-pretty-print.c: Likewise.
4280         * tree-ssa-alias.c: Likewise.
4281         * tree-ssa-copy.c: Likewise.
4282         * tree-ssanames.c: Likewise.
4283         * tree-ssanames.h: Likewise.
4284         * tree-vrp.c: Likewise.
4286 2013-09-24  Alan Modra  <amodra@gmail.com>
4288         PR middle-end/57134
4289         PR middle-end/57586
4290         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
4291         for output operands that disallow regs.  Don't use EXPAND_WRITE on
4292         inout operands.
4294 2013-09-24  Richard Biener  <rguenther@suse.de>
4296         PR middle-end/58513
4297         * tree.c (reference_alias_ptr_type): Move ...
4298         * alias.c (reference_alias_ptr_type): ... here and implement
4299         in terms of the new reference_alias_ptr_type_1.
4300         (ref_all_alias_ptr_type_p): New helper.
4301         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
4302         use ref_all_alias_ptr_type_p.
4303         (get_deref_alias_set): Add flag_strict_aliasing check here.
4304         (reference_alias_ptr_type_1): New function, split out from ...
4305         (get_alias_set): ... here.
4306         (alias_ptr_types_compatible_p): New function.
4307         * alias.h (reference_alias_ptr_type): Declare.
4308         (alias_ptr_types_compatible_p): Likewise.
4309         * tree.h (reference_alias_ptr_type): Remove.
4310         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
4311         to compare MEM_REF alias types.
4313 2013-09-24  Richard Biener  <rguenther@suse.de>
4315         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
4317 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
4319         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
4320         reload helper function arrays into a single array reg_addr.
4321         (reload_fpr_gpr): Likewise.
4322         (reload_gpr_vsx): Likewise.
4323         (reload_vsx_gpr): Likewise.
4324         (struct rs6000_reg_addr): Likewise.
4325         (reg_addr): Likewise.
4326         (rs6000_debug_reg_global): Change rs6000_vector_reload,
4327         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
4328         (rs6000_init_hard_regno_mode_ok): Likewise.
4329         (rs6000_secondary_reload_direct_move): Likewise.
4330         (rs6000_secondary_reload): Likewise.
4332         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
4333         constraints: wu, ww, and wy.  Repurpose wv constraint added during
4334         power8 changes.  Put wg constraint in alphabetical order.
4336         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
4337         for future work to add ISA 2.07 VSX single precision support.
4338         (-mvsx-scalar-double): Change default from -1 to 1, update
4339         documentation comment.
4340         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
4341         (-mupper-regs-df): New debug switch to control whether DF values
4342         can go in the traditional Altivec registers.
4343         (-mupper-regs-sf): New debug switch to control whether SF values
4344         can go in the traditional Altivec registers.
4346         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
4347         and wy constraints.
4348         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
4349         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
4350         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
4351         (rs6000_debug_legitimate_address_p): Print if we are running
4352         before, during, or after reload.
4353         (rs6000_secondary_reload): Add a comment.
4354         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
4356         * config/rs6000/constraints.md (wa constraint): Sort w<x>
4357         constraints.  Update documentation string.
4358         (wd constraint): Likewise.
4359         (wf constraint): Likewise.
4360         (wg constraint): Likewise.
4361         (wn constraint): Likewise.
4362         (ws constraint): Likewise.
4363         (wt constraint): Likewise.
4364         (wx constraint): Likewise.
4365         (wz constraint): Likewise.
4366         (wu constraint): New constraint for ISA 2.07 SFmode scalar
4367         instructions.
4368         (ww constraint): Likewise.
4369         (wy constraint): Likewise.
4370         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
4371         in the previous submissions.
4372         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
4374 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
4376         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
4377         (REG_BR_PROB): Say that the probability is stored in an int_list.
4378         * reg-notes.def: Update commentary to mention INT_LIST.
4379         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
4380         (INT_LIST): New rtx.
4381         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
4382         * rtlanal.c (int_reg_note_p): New function.
4383         (alloc_reg_note): Assert that the note does not have an int argument.
4384         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
4385         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
4386         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
4387         rather than an INSN_LIST.  Handle INT_LIST.
4388         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
4389         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
4390         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
4391         Manipulate them as ints rather than rtxes.
4392         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
4393         * regmove.c (copy_src_to_dest): Likewise.
4394         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
4396         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
4397         into the cases that need it.
4398         * config/arm/arm.c (arm_unwind_emit): Likewise.
4400         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
4401         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
4402         * loop-doloop.c (add_test, doloop_modify): Likewise.
4403         * loop-unswitch.c (compare_and_jump_seq): Likewise.
4404         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
4405         * predict.c (combine_predictions_for_insn): Likewise.
4406         * print-rtl.c (print_rtx): Handle INT_LIST.
4407         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
4408         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
4409         * config/arm/arm.c (emit_unlikely_jump): Likewise.
4410         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
4411         (ix86_split_fp_branch, predict_jump): Likewise.
4412         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
4413         * config/sh/sh.c (expand_cbranchsi4): Likewise.
4414         * config/spu/spu.c (ea_load_store_inline): Likewise.
4416         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
4417         value of a REG_BR_PROB note.
4418         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
4419         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
4420         * emit-rtl.c (try_split): Likewise.
4421         * predict.c (br_prob_note_reliable_p): Likewise.
4422         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
4423         * reorg.c (mostly_true_jump): Likewise.
4424         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
4425         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
4426         * config/i386/i386.c (ix86_print_operand): Likewise.
4427         * config/ia64/ia64.c (ia64_print_operand): Likewise.
4428         * config/mmix/mmix.c (mmix_print_operand): Likewise.
4429         * config/rs6000/rs6000.c (output_cbranch): Likewise.
4430         * config/s390/s390.c (s390_expand_tbegin): Likewise.
4431         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
4432         * config/sparc/sparc.c (output_cbranch): Likewise.
4433         * config/spu/spu.c (get_branch_target): Likewise.
4434         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
4435         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
4437 2013-09-23  Jan Hubicka  <jh@suse.cz>
4439         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
4440         for ipa-devirt.
4441         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
4442         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
4443         of external calls
4444         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
4445         (gimple_fold_call): Dump inconsistent devirtualizations; add
4446         sanity check for type based devirtualizations.
4447         * ipa-prop.c: Include ipa-utils.h
4448         (ipa_intraprocedural_devirtualization): Add sanity check.
4449         (try_make_edge_direct_virtual_call): Likewise.
4451 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
4453         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
4454         assignment statements.
4456 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
4458         * gimple-pretty-print.c (dump_ssaname_info): New function.
4459         (dump_gimple_phi): Call it.
4460         (pp_gimple_stmt_1): Likewise.
4461         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
4462         (range_info_def): Declare.
4463         * tree-pretty-print.c (pp_double_int): New function.
4464         (dump_generic_node): Call it.
4465         * tree-pretty-print.h (pp_double_int): Declare.
4466         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
4467         * tree-ssanames.h (range_info_def): New structure.
4468         (value_range_type): Move definition here.
4469         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
4470         Declare.
4471         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
4472         initialization.
4473         (set_range_info): New function.
4474         (get_range_info): Likewise.
4475         (duplicate_ssa_name_range_info): Likewise.
4476         (duplicate_ssa_name_fn): Check pointer type and call
4477         duplicate_ssa_name_range_info.
4478         * tree-ssa-copy.c (fini_copy_prop): Likewise.
4479         * tree-vrp.c (value_range_type): Remove definition, now in
4480         tree-ssanames.h.
4481         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
4482         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
4483         (SSA_NAME_RANGE_INFO): New macro.
4485 2013-09-23  Richard Biener  <rguenther@suse.de>
4487         PR tree-optimization/58464
4488         * tree-ssa-pre.c (phi_trans_lookup): Remove.
4489         (phi_trans_add): Change to add conditionally on being not present.
4490         (phi_translate_1): Remove recursion detection here.
4491         (phi_translate): Pre-seed the cache with NULL to catch
4492         recursion here in a more generic way.
4493         (bitmap_find_leader): Adjust comment.
4494         (get_representative_for): Dump value-numbers.
4495         (create_expression_by_pieces): Likewise.
4496         (insert_into_preds_of_block): Likewise.
4498 2013-09-23  Christian Bruel  <christian.bruel@st.com>
4500         PR target/58475
4501         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
4502         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
4504 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
4506         Revert r202780:
4507         2013-09-20  Renlin Li  <renlin.li@arm.com>
4509         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
4510         plus_constant.
4511         (aarch64_expand_epilogue): Likewise.
4512         (aarch64_legitimize_reload_address): Likewise.
4514 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
4516         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
4517         NULL_TREE before pushing them onto the vector.  Likewise for labels.
4519 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
4521         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
4522         comparison operators when -fno-trapping-math is in effect.
4523         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
4524         comparison operators in TFmode and assert that unsupported operators
4525         cannot reach here.
4526         (ia64_print_operand): Likewise.
4528 2013-09-21  Jan Hubicka  <jh@suse.cz>
4530         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
4531         (sse_typeless_stores): Enable for core
4532         (sse_load0_by_pxor): Likewise.
4533         (four_jump_limit): Disable for core.
4534         (pad_returns): Likewise.
4535         (avoid_vector_decode): Likewise.
4536         (fuse_cmp_and_branch): Enable for cores.
4537         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
4539 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4541         PR middle-end/56791
4542         * config/pa/pa.c (pa_option_override): Disable auto increment and
4543         decrement instructions until reload is completed.
4545         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
4546         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
4547         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
4549 2013-09-20  DJ Delorie  <dj@redhat.com>
4550             Nick Clifton  <nickc@redhat.com>
4552         * config/rl78/rl78.c: Various whitespace and comment tweaks.
4553         (need_to_save): Save bank 0 on interrupts.
4554         (characterize_address): Strip far address wrappers.
4555         (rl78_as_legitimate_address): Likewise.
4556         (transcode_memory_rtx): Likewise.
4557         (rl78_peep_movhi_p): Disable this peephole after devirt.
4558         (rl78_propogate_register_origins): Forget all origins when a
4559         CLOBBER is seen.
4560         * config/rl78/rl78-virt.md: Various whitespace tweaks.
4561         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
4562         peephole2's.
4563         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
4564         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
4565         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
4566         relocating.
4567         * config/rl78/constraints.md: Various whitespace and paren tweaks.
4569 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
4571         * config/pa/pa.md: In "scc" insn patterns, change output template to
4572         handle const0_rtx in reg_or_0_operand operands.
4574 2013-09-20  Martin Husemann  <martin@NetBSD.org>
4576         PR target/56875
4577         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
4578         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
4580 2013-09-20  Richard Biener  <rguenther@suse.de>
4582         PR middle-end/58484
4583         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
4584         remembering SSA name version and block index.
4585         (new_scev_info_str): Adjust.
4586         (hash_scev_info): Likewise.  Also hash the block index.
4587         (eq_scev_info): Adjust.
4588         (find_var_scev_info): Likewise.
4589         (struct instantiate_cache_entry): Remove.
4590         (struct instantiate_cache_type): Use a htab to map name, block
4591         to chrec.
4592         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4593         (get_instantiated_value_entry): Likewise.
4594         (hash_idx_scev_info, eq_idx_scev_info): New functions.
4595         (instantiate_scev_name): Adjust.
4597 2013-09-20  Jeff Law  <law@redhat.com>
4599         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
4601 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
4603         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
4604         Call aarch64_simd_expand_args to update op[argc].
4606 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4608         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
4609         plugin argument.
4611 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
4613         * gengtype.c (file_rules): Added rule for *.cc files.
4614         (get_output_file_with_visibility): Give fatal message when no
4615         rules found.
4617 2013-09-20  Renlin Li  <renlin.li@arm.com>
4619         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
4620         (aarch64_expand_epilogue): Likewise.
4621         (aarch64_legitimize_reload_address): Likewise.
4623 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4625         PR middle-end/57748
4626         * expr.c (expand_assignment): Remove misalignp code path.
4628 2013-09-20  Marek Polacek  <polacek@redhat.com>
4630         PR sanitizer/58413
4631         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
4632         TYPE_PRECISION.  Add asserts.
4634 2013-09-20  Richard Biener  <rguenther@suse.de>
4636         PR tree-optimization/58453
4637         * tree-loop-distribution.c (distribute_loop): Apply the cost
4638         model for -ftree-loop-distribute-patterns, too.
4640 2013-09-20  Richard Biener  <rguenther@suse.de>
4642         PR middle-end/58473
4643         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
4644         make type comparison less strict.
4646 2013-09-20  Alan Modra  <amodra@gmail.com>
4648         * configure: Regenerate.
4649         * aclocal.m4: Regenerate.
4651 2013-09-20  Marek Polacek  <polacek@redhat.com>
4653         PR other/58467
4654         * doc/extend.texi: Document that attribute used is meant to be used
4655         on variables with static storage duration.
4657 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4659         PR tree-optimization/58472
4660         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
4661         simd_lane_access set inv_p = false.
4662         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
4663         the simduid magic VAR_DECL.
4665 2013-09-19  Jan Hubicka  <jh@suse.cz>
4667         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
4669 2013-09-17  Jeff Law  <law@redhat.com>
4671         * tree-ssa-dom.c (record_temporary_equivalences): New function
4672         split out of dom_opt_dom_walker::after_dom_children.
4673         (dom_opt_dom_walker::thread_across_edge): Move common code
4674         in here from dom_opt_dom_walker::after_dom_children.
4675         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
4677 2013-09-19  Jan Hubicka  <jh@suse.cz>
4679         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
4680         (TARGET_GENERIC): Use PROCESOR_GENERIC
4681         (enum processor_type): Unify generic32 and 64.
4682         * i386.md (cpu): Likewise.
4683         * x86-tune.def (use_leave): Enable for generic32.
4684         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
4685         * athlon.md: Change generic64 to generic in all occurences.
4686         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
4687         (ix86_target_macros_internal): Likewise.
4688         * driver-i386.c (host_detect_local_cpu): Likewise.
4689         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
4690         to ..
4691         (generic_memcpy, generic_memset, generic_cost): This one.
4692         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
4693         (m_GENERIC32, m_GENERIC64): Remove.
4694         (m_GENERIC): Turn into one flag.
4695         (processor_target): Unify generic tunnings.
4696         (ix86_option_override_internal): Replace generic32/64 by generic.
4697         (ix86_issue_rate): Likewise.
4698         (ix86_adjust_cost): Likewise.
4700 2013-09-19  Jan Hubicka  <jh@suse.cz>
4702         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
4703         of speculative flag.
4705 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
4707         * omp-low.c (expand_omp_sections): Always pass len - 1 to
4708         GOMP_sections_start, even if !exit_reachable.
4710 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4712         * lra-constraints.c (need_for_all_save_p): Use macro
4713         HARD_REGNO_CALL_PART_CLOBBERED.
4714         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
4715         set up pseudo conflict hard regs.
4717 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
4719         PR target/58452
4720         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
4721         operands.
4723 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
4725         PR rtl-optimization/58438
4726         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
4727         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
4728         from previous subpasses.
4730 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
4732         * arm.c (arm_get_frame_offsets): Validate architecture supports
4733         LDRD/STRD before accepting the tuning preference.
4734         (arm_expand_prologue): Likewise.
4735         (arm_expand_epilogue): Likewise.
4737 2013-09-18  Richard Biener  <rguenther@suse.de>
4739         PR tree-optimization/58417
4740         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
4741         have chrecs with symbols defined in the loop as operands.
4742         (chrec_fold_multiply): Likewise.
4743         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
4744         parameters before folding binary operations.
4745         (struct instantiate_cache_entry_hasher): Remove.
4746         (struct instantiate_cache_type): Use a pointer-map.
4747         (instantiate_cache_type::instantiate_cache_type): New function.
4748         (instantiate_cache_type::get): Likewise.
4749         (instantiate_cache_type::set): Likewise.
4750         (instantiate_cache_type::~instantiate_cache_type): Adjust.
4751         (get_instantiated_value_entry): Likewise.
4752         (global_cache): New global.
4753         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
4754         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
4755         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
4756         (instantiate_scev_name): Adjust.
4757         (instantiate_scev): Construct global instead of local cache.
4758         (resolve_mixers): Likewise.
4760 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
4761             Paolo Carlini  <paolo.carlini@oracle.com>
4763         PR c++/58458
4764         * doc/implement-cxx.texi: Fix references to the C++ standards.
4766 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
4768         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
4769         * tree-vect-data-refs.c (vect_analyze_data_refs): For
4770         simd_lane_access drs, update also DR_ALIGNED_TO.
4772 2013-09-18  Marek Polacek  <polacek@redhat.com>
4774         PR sanitizer/58411
4775         * doc/extend.texi: Document no_sanitize_undefined attribute.
4776         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
4777         no_sanitize_undefined attribute.
4779 2013-09-18  Nick Clifton  <nickc@redhat.com>
4781         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
4782         (ASM_DECLARE_FUNCTION_NAME): Define.
4784 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
4786         * compare-elim.c (find_comparison_dom_walker): New class
4787         (find_comparisons_in_bb): Rename to
4788         find_comparison_dom_walker::before_dom_children
4789         (find_comparisons): Adjust
4790         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
4791         adjust.
4792         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
4793         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
4794         (init_walk_dominator_tree): Remove declaration.
4795         (fini_walk_dominator_tree): Remove declaration.
4796         * fwprop.c (single_def_use_dom_walker): New class
4797         (single_def_use_enter_block): Convert to
4798         single_def_use_dom_walker::before_dom_children.
4799         (single_def_use_leave_block): Convert to
4800         single_def_use_dom_walker::after_dom_children.
4801         (build_single_def_use_links): Adjust.
4802         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
4803         class.
4804         (find_candidates_in_block): Convert to
4805         find_candidates_dom_walker::before_dom_children.
4806         (execute_strength_reduction): Adjust.
4807         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
4808         (sese_dom_walker): New class.
4809         (sese_dom_walker::sese_dom_walker): New constructor.
4810         (sese_dom_walker::~sese_dom_walker): New destructor.
4811         (build_sese_conditions_before): Convert to
4812         sese_dom_walker::before_dom_children.
4813         (build_sese_conditions_after): Convert to
4814         sese_dom_walker::after_dom_children.
4815         (build_poly_scop): Adjust
4816         * tree-into-ssa.c (rewrite_dom_walker): New class
4817         (rewrite_enter_block): Convert to
4818         rewrite_dom_walker::before_dom_children.
4819         (rewrite_leave_block): Convert to
4820         rewrite_dom_walker::after_dom_children.
4821         (rewrite_update_dom_walker): New class.
4822         (rewrite_update_enter_block): Convert to
4823         rewrite_update_dom_walker::before_dom_children.
4824         (rewrite_update_leave_block): Convert to
4825         rewrite_update_dom_walker::after_dom_children.
4826         (rewrite_blocks, rewrite_into_ssa): Adjust.
4827         (mark_def_dom_walker): New class.
4828         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
4829         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
4830         (mark_def_sites_blocks): Convert to
4831         mark_def_dom_walker::before_dom_children.
4832         (mark_def_site_blocks): Remove.
4833         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
4834         (tree_ssa_dominator_optimize): Adjust.
4835         (dom_thread_across_edge): Convert to method
4836         dom_opt_dom_walker::thread_across_edge.
4837         (dom_opt_enter_block): Convert to member function
4838         dom_opt_dom_walker::before_dom_children.
4839         (dom_opt_leave_block): Convert to member function
4840         dom_opt_dom_walker::after_dom_children.
4841         * tree-ssa-dse.c (dse_dom_walker): New class.
4842         (dse_enter_block): Convert to member function
4843         dse_dom_walker::before_dom_children.
4844         (tree_ssa_dse): Adjust.
4845         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
4846         (determine_invariantness_stmt): Convert to method
4847         invariantness_dom_walker::before_dom_children.
4848         (determine_invariantness): Remove
4849         (move_computations_dom_walker): New class.
4850         (move_computations_stmt): Convert to method
4851         move_computations_dom_walker::before_dom_children.
4852         (move_computations, tree_ssa_lim): Adjust.
4853         * tree-ssa-phiopt.c (nontrapping_dom_walker): new class
4854         (nt_init_block): Make method
4855         notrappping_dom_walker::before_dom_children.
4856         (nt_fini_block): Make
4857         method nontrapping_dom_walker::after_dom_children.
4858         (get_non_trapping): Adjust.
4859         * tree-ssa-pre.c (eliminate_dom_walker): New class.
4860         (eliminate_bb): Make method eliminate_dom_walker::before_dom_children.
4861         (eliminate_leave_block): Make method.
4862         eliminate_dom_walker::after_dom_children.
4863         (eliminate): Adjust
4864         * tree-ssa-strlen.c (strlen_dom_walker): New class.
4865         (strlen_enter_block): Make method
4866         strlen_dom_walker::before_dom_children.
4867         (strlen_leave_block): Make
4868         method strlen_dom_walker::after_dom_children.
4869         (tree_ssa_strlen): Adjust.
4870         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
4871         (tree_ssa_uncprop): Adjust.
4872         (uncprop_leave_block): Make method
4873         uncprop_dom_walker::after_dom_children.
4874         (uncprop_leave_block): Make method
4875         uncprop_dom_walker::before_dom_children.
4877 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
4879         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
4880         branch in same basic block.  Check both src and dest of the move insn.
4882 2013-09-17  Nick Clifton  <nickc@redhat.com>
4884         * config/rl78/rl78-real.md (bf): New pattern.
4885         (bt): New pattern.
4886         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
4887         (rl78_print_operand): Do not put a # before a %B.
4888         * config/rl78/rl78.opt: Tweak doc strings.
4890 2013-09-17  DJ Delorie  <dj@redhat.com>
4892         * config/rl78/constraints.md (Wcv): Allow up to $r31.
4893         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
4894         (rl78_option_override): Likewise, if -mallregs.
4895         (is_virtual_register): Likewise.
4896         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
4897         (REGNO_OK_FOR_BASE_P): Likewise.
4898         * config/rl78/rl78.opt (-mallregs): New.
4900 2013-09-17  Nick Clifton  <nickc@redhat.com>
4902         * config/rl78/rl78.c (need_to_save): Change return type to bool.
4903         For interrupt functions: save all call clobbered registers if the
4904         interrupt handler is not a leaf function.
4905         (rl78_expand_prologue): Always recompute the frame information.
4906         For interrupt functions: only select bank 0 if one of the bank 0
4907         registers is going to be psuhed.
4909 2013-09-17  DJ Delorie  <dj@redhat.com>
4911         * config/rl78/constraints.md: For each W* constraint, rename to C*
4912         and create a W* constraint that checks for an optional ES: prefix
4913         pattern also.
4914         * config/rl78/rl78.md (UNS_ES_ADDR): New.
4915         (es_addr): New.  Used to wrap far addresses.
4916         * config/rl78/rl78-protos.h (rl78_es_addr): New.
4917         (rl78_es_base): New.
4918         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
4919         wrapped far addresses.
4920         (rl78_print_operand_1): Unwrap far addresses before processing.
4921         (rl78_lo16): Wrap far addresses in unspecs.
4922         (rl78_es_addr): New.
4923         (rl78_es_base): New.
4924         (insn_ok_now): Check for not-yet-wrapped far addresses.
4925         (transcode_memory_rtx): Properly re-wrap far addresses.
4927 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
4929         * config/sparc/t-rtems: Add leon3 multilibs.
4931 2013-09-17  Cong Hou  <congh@google.com>
4933         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
4934         when checking the dot production pattern. The type of rhs operand
4935         of multiply is now checked correctly.
4937 2013-09-17  Jeff Law  <law@redhat.com>
4939         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
4940         edge implied equivalences into successor phis.
4941         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
4942         here from tree-ssa-threadedge.c.
4943         (mark_threaded_blocks): When threading through a joiner, if both
4944         successors of the joiner's clone reach the same block, verify the
4945         PHI arguments are equal.  If not, cancel the jump threading request.
4946         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
4947         tree-ssa-threadupdate.c
4948         (thread_across_edge): Don't check PHI argument equality when
4949         threading through joiner block here.
4951 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
4953         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
4954         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
4955         (warn_uninit, warn_uninitialized_vars,
4956         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4957         Move to tree-ssa-uninit.c.
4958         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
4959         (has_undefined_value_p): New.  Pass dependant parts of
4960         ssa_undefined_value_p.
4961         (uninit_undefined_value_p): Use has_undefined_value_p.
4962         (warn_uninit, warn_uninitialized_vars,
4963         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
4964         Move from tree-ssa.c.
4965         * tree-ssa.h: Adjust prototypes.
4967 2013-09-17  Jan Hubicka  <jh@suse.cz>
4969         PR middle-end/58332
4970         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
4971         * ipa-inline.c (can_inline_edge_p): Do not downgrade
4972         FUNCTION_NOT_OPTIMIZED.
4973         * ipa-inline-analysis.c (compute_inline_parameters): Function
4974         not optimized is not inlinable unless it is alwaysinline.
4975         (inline_analyze_function): Force calls in not optimized
4976         function not inlinable.
4978 2013-09-17  Jan Hubicka  <jh@suse.cz>
4980         PR middle-end/58329
4981         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
4982         to return NULL.
4983         * ipa.c (function_and_variable_visibility): Likewise.
4984         * ipa-profile.c (ipa_profile): Likewise.
4986 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
4988         PR ipa/58398
4989         * cgraph.c (cgraph_function_body_availability): Check for ifunc
4990         attribute, and don't inline the resolver in this case.
4992 2013-09-17  Teresa Johnson  <tejohnson@google.com>
4994         * coverage.c (get_coverage_counts): Add missing newline.
4996 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
4998         PR tree-optimization/58088
4999         * fold-const.c (mask_with_trailing_zeros): New function.
5000         (fold_binary_loc): Make sure we don't recurse infinitely
5001         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
5002         Use mask_with_trailing_zeros where appropriate.
5004 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
5006         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
5007         of 'prev' var to get better distance estimation.
5009 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
5011         * tree-inline.h (struct copy_body_data): Add transform_parameter.
5012         * tree-inline.c (is_parameter_of): New predicate.
5013         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
5014         a parameter has been remapped.
5015         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
5016         (optimize_inline_calls): Initialize transform_parameter.
5017         (copy_gimple_seq_and_replace_locals): Likewise.
5018         (tree_function_versioning): Likewise.
5019         (maybe_inline_call_in_expr): Likewise.
5021 2013-09-17  Nick Clifton  <nickc@redhat.com>
5023         * config/msp430/msp430-protos.h: Add prototypes for new functions.
5024         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
5025         interrupt handlers.
5026         (is_attr_func): New function.
5027         (msp430_is_interrupt_func): New function.
5028         (is_naked_func): New function.
5029         (is_reentrant_func): New function.
5030         (is_critical_func): New function.
5031         (msp430_start_function): Add annotations for function attributes.
5032         (msp430_attr): New function.
5033         (msp430_attribute_table): New.
5034         (msp430_function_section): New function.
5035         (TARGET_ASM_FUNCTION_SECTION): Define.
5036         (msp430_builtin): New enum.
5037         (msp430_init_builtins): New function.
5038         (msp430_builtin_devl): New function.
5039         (msp430_expand_builtin): New function.
5040         (TARGET_INIT_BUILTINS): Define.
5041         (TARGET_EXPAND_BUILTINS): Define.
5042         (TARGET_BUILTIN_DECL): Define.
5043         (msp430_expand_prologue): Add support for naked, interrupt,
5044         critical and reentrant functions.
5045         (msp430_expand_epilogue): Likewise.
5046         (msp430_print_operand): Handle 'O' character.
5047         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
5048         NO_TRAMPOLINES.
5049         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
5050         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
5051         (pushm): Use a 'n' rather than an 'i' constraint.
5052         (msp_return): Add generation of the interrupt return instruction.
5053         (disable_interrupts): New pattern.
5054         (enable_interrupts): New pattern.
5055         (push_intr_state): New pattern.
5056         (pop_intr_state): New pattern.
5057         (bic_SR): New pattern.
5058         (bis_SR): New pattern.
5059         * doc/extend.texi: Document MSP430 function attributes and builtin
5060         functions.
5062 2013-09-17  Richard Biener  <rguenther@suse.de>
5064         PR tree-optimization/58432
5065         * tree-loop-distribution.c (tree_loop_distribution): Also
5066         scan PHIs for outside loop uses and seed a partition from them.
5068 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
5070         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
5071         (restructure_reference): Call backtrace_base_for_ref.
5073 2013-09-17  Alan Modra  <amodra@gmail.com>
5075         PR target/57589
5076         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
5077         patch.
5079 2013-09-16  DJ Delorie  <dj@redhat.com>
5081         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
5082         vregs location for RL78/G10.
5083         (rl78_expand_prologue): Avoid SEL on G10.
5084         (rl78_expand_epilogue): Likewise.
5085         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
5086         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
5087         __RL78_G10__ when appropriate.
5088         (ASM_SPEC): Pass -mg10 along to the assembler.
5089         * config/rl78/rl78.md (sel_rb): Disable for G10.
5090         * config/rl78/rl78.opt: Add -mg10 option.
5091         * config/rl78/t-rl78: Add -mg10 multilib.
5093 2013-09-16  Xinliang David Li  <davidxl@google.com>
5095         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
5096         * omp-low.c (omp_max_vf): Ditto.
5097         (expand_omp_simd): Ditto.
5098         * tree-vectorizer.c (vectorize_loops): Ditto.
5099         (gate_vect_slp): Ditto.
5100         (gate_increase_alignment): Ditto.
5101         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
5102         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
5103         (gate_tree_vectorize): Name change.
5104         (tree_vectorize): Ditto.
5105         (pass_vectorize::gate): Call new function.
5106         (pass_vectorize::execute): Ditto.
5107         * opts.c: O3 default setting change.
5108         (finish_options): Check new flag.
5109         * doc/invoke.texi: Document new flags.
5110         * common.opt: New flags.
5112 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
5114         * doc/tm.texi.in (Cond Exec Macros): Remove node.
5115         (Condition Code): Don't reference it.
5116         * doc/tm.texi: Regenerate.
5118 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
5120         PR middle-end/58418
5121         * lra-constraints.c (undo_optional_reloads): Consider all optional
5122         reload even if it did not get a hard reg.
5124 2013-09-16  Teresa Johnson  <tejohnson@google.com>
5126         * dumpfile.c (dump_loc): Remove newline emission.
5127         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
5128         emission to dump_printf_loc calls where missing.
5129         (vect_mark_for_runtime_alias_test): Ditto.
5130         (vect_analyze_data_ref_dependence): Ditto.
5131         (vect_analyze_data_ref_dependences): Ditto.
5132         (vect_slp_analyze_data_ref_dependence): Ditto.
5133         (vect_slp_analyze_data_ref_dependences): Ditto.
5134         (vect_compute_data_ref_alignment): Ditto.
5135         (vect_update_misalignment_for_peel): Ditto.
5136         (vect_verify_datarefs_alignment): Ditto.
5137         (vector_alignment_reachable_p): Ditto.
5138         (vect_get_data_access_cost): Ditto.
5139         (vect_enhance_data_refs_alignment): Ditto.
5140         (vect_find_same_alignment_drs): Ditto.
5141         (vect_analyze_data_refs_alignment): Ditto.
5142         (vect_analyze_group_access): Ditto.
5143         (vect_analyze_data_ref_access): Ditto.
5144         (vect_analyze_data_ref_accesses): Ditto.
5145         (vect_prune_runtime_alias_test_list): Ditto.
5146         (vect_analyze_data_refs): Ditto.
5147         (vect_create_addr_base_for_vector_ref): Ditto.
5148         (vect_create_data_ref_ptr): Ditto.
5149         (vect_grouped_store_supported): Ditto.
5150         (vect_grouped_load_supported): Ditto.
5151         * value-prof.c (check_counter): Ditto.
5152         (check_ic_target): Ditto.
5153         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
5154         (vect_recog_widen_mult_pattern): Ditto.
5155         (vect_recog_widen_sum_pattern): Ditto.
5156         (vect_recog_over_widening_pattern): Ditto.
5157         (vect_recog_widen_shift_pattern): Ditto.
5158         (vect_recog_rotate_pattern): Ditto.
5159         (vect_recog_vector_vector_shift_pattern): Ditto.
5160         (vect_recog_divmod_pattern): Ditto.
5161         (vect_recog_mixed_size_cond_pattern): Ditto.
5162         (vect_recog_bool_pattern): Ditto.
5163         (vect_pattern_recog_1): Ditto.
5164         (vect_pattern_recog): Ditto.
5165         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
5166         (vect_is_simple_iv_evolution): Ditto.
5167         (vect_analyze_scalar_cycles_1): Ditto.
5168         (vect_get_loop_niters): Ditto.
5169         (vect_analyze_loop_1): Ditto.
5170         (vect_analyze_loop_form): Ditto.
5171         (vect_analyze_loop_operations): Ditto.
5172         (vect_analyze_loop_2): Ditto.
5173         (vect_analyze_loop): Ditto.
5174         (report_vect_op): Ditto.
5175         (vect_is_slp_reduction): Ditto.
5176         (vect_is_simple_reduction_1): Ditto.
5177         (vect_get_known_peeling_cost): Ditto.
5178         (vect_estimate_min_profitable_iters): Ditto.
5179         (vect_model_reduction_cost): Ditto.
5180         (vect_model_induction_cost): Ditto.
5181         (get_initial_def_for_induction): Ditto.
5182         (vect_create_epilog_for_reduction): Ditto.
5183         (vectorizable_reduction): Ditto.
5184         (vectorizable_induction): Ditto.
5185         (vectorizable_live_operation): Ditto.
5186         (vect_loop_kill_debug_uses): Ditto.
5187         (vect_transform_loop): Ditto.
5188         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
5189         (vect_stmt_relevant_p): Ditto.
5190         (process_use): Ditto.
5191         (vect_mark_stmts_to_be_vectorized): Ditto.
5192         (vect_model_simple_cost): Ditto.
5193         (vect_model_promotion_demotion_cost): Ditto.
5194         (vect_model_store_cost): Ditto.
5195         (vect_get_store_cost): Ditto.
5196         (vect_model_load_cost): Ditto.
5197         (vect_get_load_cost): Ditto.
5198         (vect_init_vector_1): Ditto.
5199         (vect_get_vec_def_for_operand): Ditto.
5200         (vect_finish_stmt_generation): Ditto.
5201         (vectorizable_call): Ditto.
5202         (vectorizable_conversion): Ditto.
5203         (vectorizable_assignment): Ditto.
5204         (vectorizable_shift): Ditto.
5205         (vectorizable_operation): Ditto.
5206         (vectorizable_store): Ditto.
5207         (vectorizable_load): Ditto.
5208         (vectorizable_condition): Ditto.
5209         (vect_analyze_stmt): Ditto.
5210         (vect_transform_stmt): Ditto.
5211         (vect_is_simple_use): Ditto.
5212         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
5213         (vect_can_advance_ivs_p): Ditto.
5214         (vect_update_ivs_after_vectorizer): Ditto.
5215         (vect_do_peeling_for_loop_bound): Ditto.
5216         (vect_gen_niters_for_prolog_loop): Ditto.
5217         (vect_update_inits_of_drs): Ditto.
5218         (vect_create_cond_for_alias_checks): Ditto.
5219         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
5220         (vect_build_slp_tree_1): Ditto.
5221         (vect_supported_load_permutation_p): Ditto.
5222         (vect_analyze_slp_instance): Ditto.
5223         (vect_analyze_slp): Ditto.
5224         (vect_make_slp_decision): Ditto.
5225         (vect_detect_hybrid_slp): Ditto.
5226         (vect_bb_vectorization_profitable_p): Ditto.
5227         (vect_slp_analyze_bb_1): Ditto.
5228         (vect_update_slp_costs_according_to_vf): Ditto.
5229         (vect_get_mask_element): Ditto.
5230         (vect_transform_slp_perm_load): Ditto.
5231         (vect_schedule_slp_instance): Ditto.
5232         (vect_schedule_slp): Ditto.
5233         (vect_slp_transform_bb): Ditto.
5234         * profile.c (read_profile_edge_counts): Ditto.
5235         (compute_branch_probabilities): Ditto.
5236         * coverage.c (get_coverage_counts): Ditto.
5238 2013-09-16  Diego Novillo  <dnovillo@google.com>
5240         * tree-core.h: Add missing comment lines from refactoring of tree.h.
5242 2013-09-16  Jan Hubicka  <jh@suse.cz>
5244         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
5245         abstract functions; for static functions check the presence of body.
5247 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5249         * config/aarch64/aarch64-simd-builtins.def (fma): New.
5250         * config/aarch64/aarch64-simd.md
5251         (aarch64_mla_elt<mode>): New.
5252         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
5253         (aarch64_mls_elt<mode>): Likewise.
5254         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
5255         (aarch64_fma4_elt<mode>): Likewise.
5256         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
5257         (aarch64_fma4_elt_to_128v2df): Likewise.
5258         (aarch64_fma4_elt_to_64df): Likewise.
5259         (fnma<mode>4): Likewise.
5260         (aarch64_fnma4_elt<mode>): Likewise.
5261         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
5262         (aarch64_fnma4_elt_to_128v2df): Likewise.
5263         (aarch64_fnma4_elt_to_64df): Likewise.
5264         * config/aarch64/iterators.md (VDQSF): New.
5265         * config/aarch64/arm_neon.h
5266         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
5267         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
5269 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5271         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
5272         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
5273         (aarch64_mul3_elt_to_128df): Likewise.
5274         (aarch64_mul3_elt_to_64v2df): Likewise.
5275         * config/aarch64/iterators.md (VEL): Also handle DFmode.
5276         (VMUL): New.
5277         (VMUL_CHANGE_NLANES) Likewise.
5278         (h_con): Likewise.
5279         (f): Likewise.
5280         * config/aarch64/arm_neon.h
5281         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
5283 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
5285         * config/aarch64/arm_neon.h
5286         (vcvtx_high_f32_f64): Fix parameters.
5288 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5289             Uros Bizjak  <ubizjak@gmail.com>
5291         * config/alpha.c: Include tree-ssanames.h.
5293 2013-09-16  Richard Biener  <rguenther@suse.de>
5295         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
5296         (dot_rdg_1): Handle control_dd.
5297         (create_edge_for_control_dependence): New function.
5298         (create_rdg_edges): Add control dependences if asked for.
5299         (build_rdg): Likewise.
5300         (generate_loops_for_partition): If there are not necessary
5301         control stmts remove all their dependencies.
5302         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
5303         (distribute_loop): Pass on control dependences.
5304         (tree_loop_distribution): Compute control dependences and remove
5305         restriction on number of loop nodes.
5307 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
5309         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
5310         for internal calls.
5312 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
5314         * cse.c (try_const_anchors): Punt on CC modes.
5316 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
5318         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
5320 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
5322         PR target/58382
5323         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
5324         calls to word_mode.
5326 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5328         PR target/48094
5329         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
5330         seen.
5331         (darwin_objc1_section): Likewise.
5332         (darwin_file_end): Emit Image Info section when required.
5334 2013-09-14  Jan Hubicka  <jh@suse.cz>
5336         * tree-into-ssa.c (gate_into_ssa): New.
5337         (pass_data_build_ssa): Use it.
5338         * cgraph.h (expand_thunk): Update prototype.
5339         * cgraphunit.c (analyze_function): Expand thunks early.
5340         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
5341         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
5342         set lowered flag; do not add new function.
5343         (assemble_thunks_and_aliases): Update.
5344         * tree-ssa.c (gate_init_datastructures): New gate.
5345         (pass_data_init_datastructures): Use it.
5347 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
5349         PR target/58269
5350         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
5351         xmm register set described in the psABI.
5353 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
5355         * dwarf2out.c (should_emit_struct_debug): Add check
5356         for type_decl variable is not NULL.
5358 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
5360         * config.gcc: Use new winnt-c.c target hooks
5361         * config/t-winnt: New file
5362         * config/winnt-c.c: New file
5363         * doc/tm.texi.in: Document new hook
5364         * doc/tm.texi: Regenerated
5366 2013-09-13  Jan Hubicka  <jh@suse.cz>
5368         PR middle-end/58094
5369         * ipa-inline.c (check_callers): New function.
5370         (check_caller_edge): Remove.
5371         (want_inline_function_to_all_callers_p): Also permit alises that are
5372         called dirrectly.
5373         (inline_to_all_callers): Terminate the walk when devirtualization
5374         introduce new calls.
5376 2013-09-13  Jan Hubicka  <jh@suse.cz>
5378         * ipa-inline-analysis.c (struct growth_data): Add node.
5379         (do_estimate_growth_1): Fix detection of recursion.
5381 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
5383         PR tree-optimization/58392
5384         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
5385         to avoid shadowing of outer loop variable.  If
5386         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
5387         replace_by_duplicate_decl simduid of loops that have it set and
5388         set dest_cfun->has_simduid_loops and/or
5389         dest_cfun->has_force_vect_loops.
5390         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
5391         instead of maybe_lookup_decl.
5392         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
5393         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
5394         call.  Copy over force_vect and copy and remap simduid.  Set
5395         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
5396         (copy_cfg_body): Remove blocks_to_copy argument.  Use
5397         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
5398         caller.  Don't set cfun->has_simduid_loops and/or
5399         cfun->has_force_vect_loops here.
5400         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
5401         caller.
5402         (expand_call_inline, tree_function_versioning): Adjust copy_body
5403         callers.
5405 2013-09-13  Martin Jambor  <mjambor@suse.cz>
5407         PR bootstrap/58388
5408         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
5409         the assert if the edge was a speculative one.
5411 2013-09-13  Richard Biener  <rguenther@suse.de>
5413         * tree-data-ref.h (known_dependences_p): Move here ...
5414         * tree-loop-distribution.c (known_dependences_p): ... from here.
5415         (dump_rdg_component, debug_rdg_component): Remove.
5416         (dump_rdg): Adjust.
5417         (generate_loops_for_partition): Use gimple_uid instead of
5418         relying on matching stmt visit order.
5419         (rdg_build_partitions): Take starting stmt vector.
5420         (ldist_gen): Merge into ...
5421         (distribute_loop): ... this function.  Do not compute starting
5422         vertices vector.
5423         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
5425 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5427         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
5428         Set type attribute correctly. Set predicable_short_it attribute.
5429         (cmpsi_shiftsi): Remove %? from output template.
5431 2013-09-13  Richard Biener  <rguenther@suse.de>
5433         * tree-loop-distribution.c (struct rdg_component,
5434         rdg_defs_used_in_other_loops_p, free_rdg_components,
5435         rdg_build_components): Remove.
5436         (stmts_from_loop): Do not record virtual PHIs.
5437         (generate_loops_for_partition): Skip virtual PHIs.
5438         (build_rdg_partition_for_component): Rename to ...
5439         (build_rdg_partition_for_vertex): ... this and adjust.
5440         (rdg_build_partitions): Take a vector of starting vertices
5441         instead of components.  Remove unnecessary leftover handling.
5442         (ldist_gen): Do not build components or record other stores.
5443         (distribute_loop): Do not distribute loops containing stmts
5444         with side-effects.
5446 2013-09-13  Christian Bruel  <christian.bruel@st.com>
5448         PR target/58314
5449         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
5451 2013-09-13  Kai Tietz  <ktietz@redhat.com>
5453         * config.gcc: Separate cases for mingw and cygwin targets,
5454         and add 64-bit cygwin target case.
5456         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
5457         dll-export inline-functions.
5458         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
5461 2013-09-13  Jeff Law  <law@redhat.com>
5463         PR middle-end/58387
5464         Revert:
5465         2013-09-06  Jeff Law  <law@redhat.com>
5467         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
5468         edge implied equivalences into successor phis.
5470 2013-09-12  DJ Delorie  <dj@redhat.com>
5472         * config/rl78/rl78-virt.md: Change from | to \; for asm line
5473         separators.
5475 2013-09-12  Brooks Moses  <bmoses@google.com>
5477         PR driver/42955
5478         * Makefile.in: Do not install driver binaries in $(target)/bin.
5480 2013-09-12  DJ Delorie  <dj@redhat.com>
5482         * config/rl78/rl78.opt (mrelax): New.
5483         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
5484         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
5486         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
5487         between SP and FP.
5488         (rl78_expand_epilogue): Likewise.
5490 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
5492         PR middle-end/58335
5493         * lra-eliminations.c (remove_reg_equal_offset_note): New.
5494         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
5495         pointer elimination with using remove_reg_equal_offset_note.
5497 2013-09-12  DJ Delorie  <dj@redhat.com>
5499         * config/msp430/: New port.
5500         * config.gcc (msp430): Added.
5501         * doc/invoke.texi: Document MSP430 options.
5502         * doc/install.texi: Document msp430-elf
5503         * doc/md.texi: Document msp430-elf
5504         * doc/contrib.texi: Document msp430-elf
5506         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
5507         PSImode.
5509 2013-09-12  Martin Jambor  <mjambor@suse.cz>
5511         PR ipa/58389
5512         * ipa-prop.c (remove_described_reference): Give up if the edge in the
5513         reference descriptor is NULL.
5514         (ipa_edge_removal_hook): If owning a reference descriptor, set its
5515         edge to NULL.
5517 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5519         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
5520         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
5521         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
5522         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
5523         make_temp_ssa_name): move to tree-ssanames.h
5524         * tree-ssa-alias.h: Move prototype.
5525         * tree-ssa.h: Include tree-ssanames.h.
5526         * tree-ssanames.c (FREE_SSANAMES): Move to here.
5527         * tree-ssanames.h: New.  Move items from tree-flow*.h
5528         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
5530 2013-09-12  Richard Biener  <rguenther@suse.de>
5532         PR tree-optimization/58404
5533         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
5534         propagate non-invariant addresses into dereferences wrapped
5535         in component references.
5537 2013-09-12  Richard Biener  <rguenther@suse.de>
5539         PR tree-optimization/58402
5540         * passes.def: Move pass_late_warn_uninitialized later.
5542 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
5544         * tree-ssa.h: New.  Move content from tree-flow.h and
5545         tree-flow-inline.h.
5546         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
5547         Move prototypes belonging to tree-ssa.c.
5548         * tree-flow-inline.h (redirect_edge_var_map_def,
5549         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
5550         tree-ssa.h.
5551         * gimple.h: Adjust prototypes.
5552         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
5553         to...
5554         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
5555         * tree.h: Move prototype to tree-ssa.h.
5556         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
5557         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
5558         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
5559         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
5560         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
5561         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
5562         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
5563         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
5564         graphite-blocking.c, graphite-clast-to-gimple.c,
5565         graphite-dependences.c, graphite-interchange.c,
5566         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
5567         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
5568         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
5569         ipa-reference.c, ipa-split.c, ipa-utils.c,
5570         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
5571         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
5572         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
5573         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
5574         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
5575         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
5576         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
5577         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
5578         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
5579         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
5580         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
5581         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
5582         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
5583         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
5584         value-prof.c, var-tracking.c,
5585         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
5587 2013-09-12  Richard Biener  <rguenther@suse.de>
5589         PR tree-optimization/58396
5590         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
5591         (build_rdg): Take a loop-nest parameter, fix memleaks.
5592         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
5594 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
5596         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
5597         for SLM.
5599 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
5601         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
5602         The FMA instruction names should have a 'v' prefix.
5604 2013-09-12  Richard Biener  <rguenther@suse.de>
5606         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
5607         (dot_rdg): Use popen instead of system in optional code.
5608         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
5609         (already_processed_vertex_p): Adjust.
5610         (has_anti_or_output_dependence, predecessor_has_mem_write,
5611         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
5612         rdg_flag_uses): Remove.
5613         (rdg_flag_vertex): Simplify.
5614         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
5615         remove recursion.
5616         (build_rdg_partition_for_component): Process the first vertex
5617         of a component only.
5618         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
5620 2013-09-12  Alan Modra  <amodra@gmail.com>
5622         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
5624 2013-09-11  DJ Delorie  <dj@redhat.com>
5625             Nick Clifton  <nickc@redhat.com>
5627         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
5628         (rl78_stack_based_mem): New.
5629         * config/rl78/constraints.md (Iv08): New.
5630         (Iv16): New.
5631         (Iv24): New.
5632         (Is09): New.
5633         (Is17): New.
5634         (Is25): New.
5635         (ISsi): New.
5636         (IShi): New.
5637         (ISqi): New.
5638         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
5639         (movhi): Likewise.
5640         (movsi): Change from expand to insn-and-split.
5641         (ashrsi3): Clobber AX.
5642         (lshrsi3): New.
5643         (ashlsi3): New.
5644         (cbranchsi4): New.
5645         * config/rl78/rl78.md (CC_REG): Fix.
5646         (addsi3): Allow memory and immediate operands.
5647         (addsi3_internal): Split into...
5648         (addsi3_internal_virt): ...new, and ...
5649         (addsi3_internal_real): ...new.
5650         (subsi): New.
5651         (subsi3_internal_virt): New.
5652         (subsi3_internal_real): New.
5653         (mulsi3): Add memory operand.
5654         (mulsi3_rl78): Likewise.
5655         (mulsi3_g13): Likewise.
5656         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
5657         (cbranchqi4_real): Add more constraint options.
5658         (cbranchhi4_real): Expand pattern.
5659         (cbranchhi4_real_signed): New.
5660         (cbranchhi4_real_inverted): New.
5661         (cbranchsi4_real_lt): New.
5662         (cbranchsi4_real_ge): New.
5663         (cbranchsi4_real_signed): New.
5664         (cbranchsi4_real): New.
5665         (peephole2): New.
5666         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
5667         constant shifts.
5668         (lshrsi3_virt): Likewise.
5669         (ashlsi3_virt): Likewise.
5670         (cbranchqi4_virt_signed): New.
5671         (cbranchhi4_virt_signed): New.
5672         (cbranchsi4_virt): New.
5673         * config/rl78/rl78.c: Whitespace fixes throughout.
5674         (move_elim_pass): New.
5675         (pass_data_rl78_move_elim): New.
5676         (pass_rl78_move_elim): New.
5677         (make_pass_rl78_move_elim): New.
5678         (rl78_devirt_info): Run devirt earlier.
5679         (rl78_move_elim_info): New.
5680         (rl78_asm_file_start): Register it.
5681         (rl78_split_movsi): New.
5682         (rl78_as_legitimate_address): Allow virtual base registers when
5683         appropriate.
5684         (rl78_addr_space_convert): Remove spurious debug stuff.
5685         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
5686         (rl78_print_operand): More cases for not printing '#'.
5687         (rl78_expand_compare): Remove most of the logic.
5688         (content_memory): New.
5689         (clear_content_memory): New.
5690         (get_content_index): New.
5691         (get_content_name): New.
5692         (display_content_memory): New.
5693         (update_content): New.
5694         (record_content): New.
5695         (already_contains): New.
5696         (insn_ok_now): Re-recog insns with virtual registers.
5697         (add_postponed_content_update): New.
5698         (process_postponed_content_update): New.
5699         (gen_and_emit_move): New.
5700         (transcode_memory_rtx): Record new location content.
5701         Use gen_and_emit_move.
5702         (force_into_acc): New.
5703         (move_to_acc): Use gen_and_emit_move.
5704         (move_from_acc): Likewise.
5705         (move_acc_to_reg): Likewise.
5706         (move_to_x): Likewise.
5707         (move_to_hl): Likewise.
5708         (move_to_de): Likewise.
5709         (rl78_alloc_physical_registers_op1): Record location content.
5710         (has_constraint): New.
5711         (rl78_alloc_physical_registers_op2): Record location content.
5712         Optimize use of HL.
5713         (rl78_alloc_physical_registers_ro1): Likewise.
5714         (rl78_alloc_physical_registers_cmp): Likewise.
5715         (rl78_alloc_physical_registers_umul): Likewise.
5716         (rl78_alloc_address_registers_macax): New.
5717         (rl78_alloc_physical_registers): Initialize and set location
5718         content memory as needed.
5719         (rl78_reorg): Make sure split2 is called.
5720         (rl78_rtx_costs): New.
5722 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
5724         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
5725         for (not (neg ...)) and (neg (not ...)) cases.
5727 2013-09-11  Richard Biener  <rguenther@suse.de>
5729         PR middle-end/58377
5730         * passes.def: Split critical edges before late uninit warning passes.
5731         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
5733 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
5735         PR tree-optimization/58385
5736         * fold-const.c (build_range_check): If both low and high are NULL,
5737         use omit_one_operand_loc to preserve exp side-effects.
5739 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5741         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
5743 2013-09-11  Richard Biener  <rguenther@suse.de>
5745         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
5746         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
5747         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
5748         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
5749         stmts_from_loop, known_dependences_p, build_empty_rdg,
5750         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
5751         * tree-loop-distribution.c: ... here.
5752         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
5753         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
5754         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
5755         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
5756         * tree-loop-distribution.c: ... here.
5757         * tree-loop-distribution.c: Include gimple-pretty-print.h.
5758         (struct partition_s): Add loops member.
5759         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
5760         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
5761         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
5763 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5764             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
5765             Sergey Lega  <sergey.s.lega@intel.com>
5766             Anna Tikhonova  <anna.tikhonova@intel.com>
5767             Ilya Tocar  <ilya.tocar@intel.com>
5768             Andrey Turetskiy  <andrey.turetskiy@intel.com>
5769             Ilya Verbin  <ilya.verbin@intel.com>
5770             Kirill Yukhin  <kirill.yukhin@intel.com>
5771             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
5773         * config/i386/constraints.md (k): New.
5774         (Yk): Ditto.
5775         * config/i386/i386.c (const regclass_map): Add new mask registers.
5776         (dbx_register_map): Ditto.
5777         (dbx64_register_map): Ditto.
5778         (svr4_dbx_register_map): Ditto.
5779         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
5780         disabled.
5781         (ix86_preferred_reload_class): Disable constants for mask registers.
5782         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
5783         (ix86_hard_regno_mode_ok): Support new mask registers.
5784         (x86_order_regs_for_local_alloc): Ditto.
5785         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
5786         (FIXED_REGISTERS): Add new mask registers.
5787         (CALL_USED_REGISTERS): Ditto.
5788         (REG_ALLOC_ORDER): Ditto.
5789         (VALID_MASK_REG_MODE): New.
5790         (FIRST_MASK_REG): Ditto.
5791         (LAST_MASK_REG): Ditto.
5792         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
5793         (MAYBE_MASK_CLASS_P): New.
5794         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
5795         (REG_CLASS_CONTENTS): Ditto.
5796         (MASK_REGNO_P): New.
5797         (ANY_MASK_REG_P): Ditto.
5798         (HI_REGISTER_NAMES): Add new mask registers.
5799         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
5800         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
5801         mask registers.
5802         (attribute "type"): Add mskmov, msklog.
5803         (attribute "length_immediate"): Support them.
5804         (attribute "memory"): Ditto.
5805         (attribute "prefix_0f"): Ditto.
5806         (*movhi_internal): Support new mask registers.
5807         (*movqi_internal): Ditto.
5808         (define_split): Split out clobber pattern is a logic
5809         insn on mask registers.
5810         (*k<logic><mode>): New.
5811         (*andhi_1): Extend to support mask regs.
5812         (*andqi_1): Extend to support mask regs.
5813         (kandn<mode>): New.
5814         (define_split): Split and-not to and and not if operands
5815         are not mask regs.
5816         (*<code><mode>_1): Separate HI mode to new pattern...
5817         (*<code>hi_1): This.
5818         (*<code>qi_1): Extend to support mask regs.
5819         (kxnor<mode>): New.
5820         (kortestzhi): Ditto.
5821         (kortestchi): Ditto.
5822         (kunpckhi): Ditto.
5823         (*one_cmpl<mode>2_1): Remove HImode and handle it...
5824         (*one_cmplhi2_1): ...Here, now with mask registers support.
5825         (*one_cmplqi2_1): Support new mask registers.
5826         (HI/QImode arithmetics splitter): Don't split if mask registers
5827         are used.
5828         (HI/QImode not splitter): Ditto.
5829         * config/i386/predicated.md (mask_reg_operand): New.
5830         (general_reg_operand): Ditto.
5832 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
5834         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
5835         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
5837 2013-09-10  Jeff Law  <law@redhat.com>
5839         PR tree-optimization/58380
5840         * tree-ssa-threadupdate.c (thread_block): Recognize another case
5841         of threading through a buried loop header.
5843         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
5844         return value for single successor case.
5846 2013-09-10  Jan Hubicka  <jh@suse.cz>
5848         * ipa-devirt.c (ipa_devirt): Enable with LTO.
5850 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
5852         PR target/58361
5853         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
5854         support conditional execution.
5855         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
5857 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
5859         * lra.c (lra): Clear lra_optional_reload_pseudos before every
5860         constraint pass.
5861         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
5862         Check destination too to check move insn.
5863         (undo_optional_reloads): Add check that the original peudo did not
5864         changed its allocation and the optional reload was inherited on last
5865         inheritance pass.  Break loop after deciding to keep optional reload.
5866         (lra_undo_inheritance): Add check that inherited pseudo still in
5867         memory.
5869 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
5871         * config/aarch64/aarch64.md (generic_sched): New.
5872         * config/aarch64/aarch64-generic.md (load): Make conditional
5873         on generic_sched attribute.
5874         (nonload): Likewise.
5876 2013-09-10  Jan Hubicka  <jh@suse.cz>
5878         * lto-cgraph.c: Include ipa-utils.h.
5879         (compute_ltrans_boundary): Also add possible targets into the boundary.
5881 2013-09-10  Jan Hubicka  <jh@suse.cz>
5883         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
5884         VAR_DECL of vtable rather than full expression.
5886 2013-09-10  Jan Hubicka  <jh@suse.cz>
5887             Paolo Carlini  <paolo.carlini@oracle.com>
5889         * cgraphunit.c (analyze_functions): Save input_location, set it
5890         to UNKNOWN_LOCATION and restore it at the end.
5892 2013-09-10  Martin Jambor  <mjambor@suse.cz>
5894         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
5895         represented by a thunk.
5897 2013-09-10  Jeff Law  <law@redhat.com>
5899         PR tree-optimization/58343
5900         * tree-ssa-threadupdate.c (thread_block): Identify and disable
5901         jump threading requests through loop headers buried in the middle
5902         of a jump threading path.
5904         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
5905         in return value/type.
5907 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
5909         PR rtl-optimization/58365
5910         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
5911         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
5912         it differs.
5914 2013-09-10  Richard Biener  <rguenther@suse.de>
5916         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
5917         * tree-data-ref.c (create_rdg_vertices): Collect all data
5918         references, signal failure to the caller, use data-ref API.
5919         (build_rdg): Compute data references only once.  Maintain lifetime
5920         of data references and data dependences from within RDG.
5921         (free_rdg): Free dependence relations.
5922         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
5923         inventing extra dependences.
5924         (distribute_loop): Update for RDG API changes.
5926 2013-09-10  Kai Tietz  <ktietz@redhat.com>
5928         * doc/invoke.texi (fms-extensions): Document changed
5929         behavior for ms-abi targets.
5930         * config/i386/i386.c (ix86_option_override_internal):
5931         Set default value of option -fms-extension for ms-abi targets.
5933 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
5935         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
5937 2013-09-10  Alan Modra  <amodra@gmail.com>
5939         PR target/58330
5940         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
5942 2013-09-10  Alan Modra  <amodra@gmail.com>
5944         * config/rs6000/predicates.md (add_cint_operand): New.
5945         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
5946         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
5947         using add_cint_operand.
5948         (largetoc_high_plus_aix): Likewise.
5950 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
5952         PR tree-optimization/58364
5953         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
5954         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
5955         the current range can't be an unconditional true or false.
5957 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
5959         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
5961 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
5963         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
5965 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
5967         PR c++/43452
5968         * doc/invoke.texi (-Wdelete-incomplete): Document it.
5970 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
5972         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
5973         NO_REGS for immediate that can't be moved directly into FP_REGS.
5975 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5977         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5978         comparison with negated operand.
5979         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
5980         RTL form.
5982 2013-09-09  Richard Biener  <rguenther@suse.de>
5984         PR middle-end/58326
5985         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
5986         of a subloop record all its block as affecting loop-closed SSA form.
5988 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5990         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
5991         of an rtx/bitpos pair.
5992         (store_fixed_bit_field): Update accordingly.
5994 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
5996         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
5997         GEN_INT.
5998         * builtins.c (expand_errno_check): Likewise.
5999         * dwarf2cfi.c (init_return_column_size): Likewise.
6000         * except.c (sjlj_mark_call_sites): Likewise.
6001         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
6002         * lra-constraints.c (emit_inc): Likewise.
6003         * ree.c (combine_set_extension): Likewise.
6004         * regmove.c (fixup_match_2): Likewise.
6005         * reload1.c (inc_for_reload): Likewise.
6007 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6009         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
6010         (force_to_mode, simplify_shift_const_1, simplify_comparison):
6011         Use gen_int_mode with the mode of the associated simplify_* call.
6012         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
6013         * expmed.c (expand_shift_1): Likewise.
6014         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6015         * loop-iv.c (iv_number_of_iterations): Likewise.
6016         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6017         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
6019 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6021         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
6022         of the associated expand_* call.
6023         (asan_emit_stack_protection): Likewise.
6024         * builtins.c (round_trampoline_addr): Likewise.
6025         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
6026         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
6027         (emit_store_flag): Likewise.
6028         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
6029         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
6030         Likewise.
6031         * function.c (instantiate_virtual_regs_in_insn): Likewise.
6032         * ifcvt.c (noce_try_store_flag_constants): Likewise.
6033         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
6034         * modulo-sched.c (generate_prolog_epilog): Likewise.
6035         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
6036         (expand_ctz, expand_ffs, expand_unop): Likewise.
6038 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6040         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
6041         of the associated gen_rtx_* call.
6042         * caller-save.c (init_caller_save): Likewise.
6043         * combine.c (find_split_point, make_extraction): Likewise.
6044         (make_compound_operation): Likewise.
6045         * dwarf2out.c (mem_loc_descriptor): Likewise.
6046         * explow.c (plus_constant, probe_stack_range): Likewise.
6047         * expmed.c (expand_mult_const): Likewise.
6048         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
6049         * reload1.c (init_reload): Likewise.
6050         * valtrack.c (cleanup_auto_inc_dec): Likewise.
6051         * var-tracking.c (adjust_mems): Likewise.
6052         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
6053         rather than gen_rtx_fmt_ee.
6055 2013-09-09  Jan Hubicka  <jh@suse.cz>
6057         PR middle-end/58294
6058         * value-prof.c (gimple_ic): Copy also abnormal edges.
6060 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
6062         * asan.c (asan_shadow_cst): Use gen_int_mode.
6064 2013-09-08  Jan Hubicka  <jh@suse.cz>
6066         * ipa-profile.c: Add toplevel comment.
6067         (ipa_propagate_frequency_1): Be more conservative when profile is read.
6068         (contains_hot_call_p): New function.
6069         (ipa_propagate_frequency): Set frequencies based on counts when
6070         profile is read.
6071         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
6072         profile; do not tamper with profile after inlining if it is read.
6074 2013-09-08  Jan Hubicka  <jh@suse.cz>
6076         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
6077         speculative edges.
6079 2013-09-08  Jan Hubicka  <jh@suse.cz>
6081         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
6082         summary generation.
6084 2013-09-08  Jeff Law  <law@redhat.com>
6086         PR bootstrap/58340
6087         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
6088         of 'found'.
6090 2013-09-08  Andi Kleen  <ak@linux.intel.com>
6092         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
6094 2013-09-08  Jan Hubicka  <jh@suse.cz>
6096         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
6098 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
6100         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
6101         for non-debug insns.
6102         * lra.c (new_insn_reg): Take the containing insn as a parameter.
6103         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
6104         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
6105         accordingly.
6107 2013-09-08  Jan Hubicka  <jh@suse.cz>
6109         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
6110         targets and devirtualize to BUILT_IN_UNREACHABLE.
6111         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
6112         * ipa.c (walk_polymorphic_call_targets): New function.
6113         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
6114         functions; use the new timevar.
6115         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
6116         was removed from the program.
6117         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
6118         not consider it in the walk when its vtable is dead.
6119         (possible_polymorphic_call_targets_1): Pass anonymous flag to
6120         record_binfo.
6121         (devirt_variable_node_removal_hook): New function.
6122         (possible_polymorphic_call_targets): Also register
6123         devirt_variable_node_removal_hook.
6124         (ipa_devirt): Do not do non-speculative devirtualization.
6125         (gate_ipa_devirt): One execute if devirtualizing speculatively.
6127 2013-09-08  Jan Hubicka  <jh@suse.cz>
6129         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
6130         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
6131         varpool_remove_variable_insertion_hook): Declare.
6132         * varpool.c (varpool_node_hook_list): New structure.
6133         (first_varpool_node_removal_hook,
6134         first_varpool_variable_insertion_hook): New variables.
6135         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
6136         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
6137         varpool_remove_variable_insertion_hook,
6138         varpool_call_variable_insertion_hooks): New functions.
6139         (varpool_remove_node): Use it.
6141 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
6143         PR c++/54941
6144         * diagnostic.c (diagnostic_build_prefix): When s.file is
6145         "<built-in>" don't output line and column numbers.
6147 2013-09-06  Jan Hubicka  <jh@suse.cz>
6149         * cgraphunit.c (expand_thunk): Get body before touching arguments.
6150         * lto-streamer-out.c: Stream thunks, too.
6151         * lto-streamer-in.c (input_function): Pop cfun here
6152         (lto_read_body): Instead of here.
6154 2013-09-06  Caroline Tice  <cmtice@google.com>
6156         * doc/install.texi: Add documentation for the --enable-vtable-verify
6157         and the --disable-libvtv configure options.
6159 2013-09-06  Jeff Law  <law@redhat.com>
6161         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
6162         edge implied equivalences into successor phis.
6164 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
6166         * resource.c (mark_referenced_resources): Handle COND_EXEC.
6168 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
6170         * resource.c (mark_target_live_regs): Compute resources taking
6171         into account if a call is predicated or not.
6173 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6175         * toplev.c (output_stack_usage): Be prepared for suffixes created by
6176         the compiler in the function names.
6178 2013-09-06  Jan Hubicka  <jh@suse.cz>
6180         PR middle-end/58094
6181         * ipa-inline.c (has_caller_p): New function.
6182         (want_inline_function_to_all_callers_p): Use it.
6183         (sum_callers, inline_to_all_callers): Break out from ...
6184         (ipa_inline): ... here.
6186 2013-09-06  Jan Hubicka  <jh@suse.cz>
6188         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
6189         only when AVX is enabled.
6191 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6193         * config/aarch64/aarch64.md
6194         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
6195         is fpsimd_<load/store>2.
6196         (load_pair<mode>): Likewise.
6197         (store_pair<mode>): Likewise.
6199 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6201         * config/arm/types.md (type): Add "mrs" type.
6202         * config/aarch64/aarch64.md
6203         (aarch64_load_tp_hard): Make type "mrs".
6204         * config/arm/arm.md
6205         (load_tp_hard): Make type "mrs".
6206         * config/arm/cortex-a15.md: Update with new attributes.
6207         * config/arm/cortex-a5.md: Update with new attributes.
6208         * config/arm/cortex-a53.md: Update with new attributes.
6209         * config/arm/cortex-a7.md: Update with new attributes.
6210         * config/arm/cortex-a8.md: Update with new attributes.
6211         * config/arm/cortex-a9.md: Update with new attributes.
6212         * config/arm/cortex-m4.md: Update with new attributes.
6213         * config/arm/cortex-r4.md: Update with new attributes.
6214         * config/arm/fa526.md: Update with new attributes.
6215         * config/arm/fa606te.md: Update with new attributes.
6216         * config/arm/fa626te.md: Update with new attributes.
6217         * config/arm/fa726te.md: Update with new attributes.
6219 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6221         * config/aarch64/aarch64.md
6222         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
6223         (*movtf_aarch64): Likewise.
6224         * config/arm/arm.md
6225         (thumb1_movdi_insn): Use "multiple" for type where more than one
6226         instruction is used for a move.
6227         (*arm32_movhf): Likewise.
6228         (*thumb_movdf_insn): Likewise.
6230 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6232         * config/arm/types.md (type): Rename fcpys to fmov.
6233         * config/arm/vfp.md
6234         (*arm_movsi_vfp): Rename type fcpys as fmov.
6235         (*thumb2_movsi_vfp): Likewise
6236         (*movhf_vfp_neon): Likewise
6237         (*movhf_vfp): Likewise
6238         (*movsf_vfp): Likewise
6239         (*thumb2_movsf_vfp): Likewise
6240         (*movsfcc_vfp): Likewise
6241         (*thumb2_movsfcc_vfp): Likewise
6242         * config/aarch64/aarch64-simd.md
6243         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
6244         * config/aarch64/aarch64.md
6245         (*movsi_aarch64): Replace type mov_reg with fmovs.
6246         (*movdi_aarch64): Likewise
6247         (*movsf_aarch64): Likewise
6248         (*movdf_aarch64): Likewise
6249         * config/arm/arm.c
6250         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
6251         * config/arm/iwmmxt.md
6252         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
6253         * config/arm/arm1020e.md: Update with new attributes.
6254         * config/arm/cortex-a15-neon.md: Update with new attributes.
6255         * config/arm/cortex-a5.md: Update with new attributes.
6256         * config/arm/cortex-a53.md: Update with new attributes.
6257         * config/arm/cortex-a7.md: Update with new attributes.
6258         * config/arm/cortex-a8-neon.md: Update with new attributes.
6259         * config/arm/cortex-a9.md: Update with new attributes.
6260         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6261         * config/arm/cortex-r4f.md: Update with new attributes.
6262         * config/arm/marvell-pj4.md: Update with new attributes.
6263         * config/arm/vfp11.md: Update with new attributes.
6265 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6267         * config/aarch64/aarch64.md
6268         (*madd<mode>): Fix type attribute.
6269         (*maddsi_uxtw): Likewise.
6270         (*msub<mode>): Likewise.
6271         (*msubsi_uxtw): Likewise.
6272         (<su_optab>maddsidi4): Likewise.
6273         (<su_optab>msubsidi4): Likewise.
6275 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6277         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
6278         * config/arm/arm.md (core_cycles): Remove fdiv.
6279         * config/arm/vfp.md:
6280         (*sqrtsf2_vfp): Update for attribute changes.
6281         (*sqrtdf2_vfp): Likewise.
6282         * config/aarch64/aarch64.md:
6283         (sqrt<mode>2): Update for attribute changes.
6284         * config/arm/arm1020e.md: Update with new attributes.
6285         * config/arm/cortex-a15-neon.md: Update with new attributes.
6286         * config/arm/cortex-a5.md: Update with new attributes.
6287         * config/arm/cortex-a53.md: Update with new attributes.
6288         * config/arm/cortex-a7.md: Update with new attributes.
6289         * config/arm/cortex-a8-neon.md: Update with new attributes.
6290         * config/arm/cortex-a9.md: Update with new attributes.
6291         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6292         * config/arm/cortex-r4f.md: Update with new attributes.
6293         * config/arm/marvell-pj4.md: Update with new attributes.
6294         * config/arm/vfp11.md: Update with new attributes.
6296 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6298         * config/arm/types.md
6299         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
6300         * config/aarch64/aarch64.md
6301         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
6302         new attributes.
6303         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
6304         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
6305         (float<GPI:mode><GPF:mode>2): Likewise.
6306         * config/arm/vfp.md
6307         (*truncsisf2_vfp): Update with new attributes.
6308         (*truncsidf2_vfp): Likewise.
6309         (fixuns_truncsfsi2): Likewise.
6310         (fixuns_truncdfsi2): Likewise.
6311         (*floatsisf2_vfp): Likewise.
6312         (*floatsidf2_vfp): Likewise.
6313         (floatunssisf2): Likewise.
6314         (floatunssidf2): Likewise.
6315         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
6316         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
6317         * config/arm/arm1020e.md: Update with new attributes.
6318         * config/arm/cortex-a15-neon.md: Update with new attributes.
6319         * config/arm/cortex-a5.md: Update with new attributes.
6320         * config/arm/cortex-a53.md: Update with new attributes.
6321         * config/arm/cortex-a7.md: Update with new attributes.
6322         * config/arm/cortex-a8-neon.md: Update with new attributes.
6323         * config/arm/cortex-a9.md: Update with new attributes.
6324         * config/arm/cortex-m4-fpu.md: Update with new attributes.
6325         * config/arm/cortex-r4f.md: Update with new attributes.
6326         * config/arm/marvell-pj4.md: Update with new attributes.
6327         * config/arm/vfp11.md: Update with new attributes.
6329 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6331         * config/aarch64/arm_neon.h
6332         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
6333         (vqtbx<1,2,3,4><q>_s8): Likewise.
6335 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6337         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
6338         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
6339         (add<mode>3): Add type attribute.
6340         (add<mode>3): Likewise.
6341         (usadd<mode>3): Likewise.
6342         (ssadd<mode>3): Likewise.
6343         (sub<mode>3): Likewise.
6344         (sub<mode>3): Likewise.
6345         (ussub<mode>3): Likewise.
6346         (sssub<mode>3): Likewise.
6347         (ssmulsa3): Likewise.
6348         (usmulusa3): Likewise.
6349         (arm_usatsihi): Likewise.
6350         * config/arm/vfp.md
6351         (*movdi_vfp): Add types for all instructions.
6352         (*movdi_vfp_cortexa8): Likewise.
6353         (*movhf_vfp_neon): Likewise.
6354         (*movhf_vfp): Likewise.
6355         (*movdf_vfp): Likewise.
6356         (*thumb2_movdf_vfp): Likewise.
6357         (*thumb2_movdfcc_vfp): Likewise.
6358         * config/arm/arm.md: Add type attribute to all insn patterns.
6359         (*thumb1_adddi3): Add type attribute.
6360         (*arm_adddi3): Likewise.
6361         (*adddi_sesidi_di): Likewise.
6362         (*adddi_zesidi_di): Likewise.
6363         (*thumb1_addsi3): Likewise.
6364         (addsi3_compare0): Likewise.
6365         (*addsi3_compare0_scratch): Likewise.
6366         (*compare_negsi_si): Likewise.
6367         (cmpsi2_addneg): Likewise.
6368         (*addsi3_carryin_<optab>): Likewise.
6369         (*addsi3_carryin_alt2_<optab>): Likewise.
6370         (*addsi3_carryin_clobercc_<optab>): Likewise.
6371         (*subsi3_carryin): Likewise.
6372         (*subsi3_carryin_const): Likewise.
6373         (*subsi3_carryin_compare): Likewise.
6374         (*subsi3_carryin_compare_const): Likewise.
6375         (*arm_subdi3): Likewise.
6376         (*thumb_subdi3): Likewise.
6377         (*subdi_di_zesidi): Likewise.
6378         (*subdi_di_sesidi): Likewise.
6379         (*subdi_zesidi_di): Likewise.
6380         (*subdi_sesidi_di): Likewise.
6381         (*subdi_zesidi_ze): Likewise.
6382         (thumb1_subsi3_insn): Likewise.
6383         (*arm_subsi3_insn): Likewise.
6384         (*anddi3_insn): Likewise.
6385         (*anddi_zesidi_di): Likewise.
6386         (*anddi_sesdi_di): Likewise.
6387         (*ne_zeroextracts): Likewise.
6388         (*ne_zeroextracts): Likewise.
6389         (*ite_ne_zeroextr): Likewise.
6390         (*ite_ne_zeroextr): Likewise.
6391         (*anddi_notdi_di): Likewise.
6392         (*anddi_notzesidi): Likewise.
6393         (*anddi_notsesidi): Likewise.
6394         (andsi_notsi_si): Likewise.
6395         (thumb1_bicsi3): Likewise.
6396         (*iordi3_insn): Likewise.
6397         (*iordi_zesidi_di): Likewise.
6398         (*iordi_sesidi_di): Likewise.
6399         (*thumb1_iorsi3_insn): Likewise.
6400         (*xordi3_insn): Likewise.
6401         (*xordi_zesidi_di): Likewise.
6402         (*xordi_sesidi_di): Likewise.
6403         (*arm_xorsi3): Likewise.
6404         (*andsi_iorsi3_no): Likewise.
6405         (*smax_0): Likewise.
6406         (*smax_m1): Likewise.
6407         (*arm_smax_insn): Likewise.
6408         (*smin_0): Likewise.
6409         (*arm_smin_insn): Likewise.
6410         (*arm_umaxsi3): Likewise.
6411         (*arm_uminsi3): Likewise.
6412         (*minmax_arithsi): Likewise.
6413         (*minmax_arithsi_): Likewise.
6414         (*satsi_<SAT:code>): Likewise.
6415         (arm_ashldi3_1bit): Likewise.
6416         (arm_ashrdi3_1bit): Likewise.
6417         (arm_lshrdi3_1bit): Likewise.
6418         (*arm_negdi2): Likewise.
6419         (*thumb1_negdi2): Likewise.
6420         (*arm_negsi2): Likewise.
6421         (*thumb1_negsi2): Likewise.
6422         (*negdi_extendsid): Likewise.
6423         (*negdi_zero_extend): Likewise.
6424         (*arm_abssi2): Likewise.
6425         (*thumb1_abssi2): Likewise.
6426         (*arm_neg_abssi2): Likewise.
6427         (*thumb1_neg_abss): Likewise.
6428         (one_cmpldi2): Likewise.
6429         (extend<mode>di2): Likewise.
6430         (*compareqi_eq0): Likewise.
6431         (*arm_extendhisi2addsi): Likewise.
6432         (*arm_movdi): Likewise.
6433         (*thumb1_movdi_insn): Likewise.
6434         (*arm_movt): Likewise.
6435         (*thumb1_movsi_insn): Likewise.
6436         (pic_add_dot_plus_four): Likewise.
6437         (pic_add_dot_plus_eight): Likewise.
6438         (tls_load_dot_plus_eight): Likewise.
6439         (*thumb1_movhi_insn): Likewise.
6440         (*thumb1_movsf_insn): Likewise.
6441         (*movdf_soft_insn): Likewise.
6442         (*thumb_movdf_insn): Likewise.
6443         (cbranchsi4_insn): Likewise.
6444         (cbranchsi4_scratch): Likewise.
6445         (*negated_cbranchsi4): Likewise.
6446         (*tbit_cbranch): Likewise.
6447         (*tlobits_cbranch): Likewise.
6448         (*tstsi3_cbranch): Likewise.
6449         (*cbranchne_decr1): Likewise.
6450         (*addsi3_cbranch): Likewise.
6451         (*addsi3_cbranch_scratch): Likewise.
6452         (*arm_cmpdi_insn): Likewise.
6453         (*arm_cmpdi_unsig): Likewise.
6454         (*arm_cmpdi_zero): Likewise.
6455         (*thumb_cmpdi_zero): Likewise.
6456         (*deleted_compare): Likewise.
6457         (*mov_scc): Likewise.
6458         (*mov_negscc): Likewise.
6459         (*mov_notscc): Likewise.
6460         (*cstoresi_eq0_thumb1_insn): Likewise.
6461         (cstoresi_nltu_thumb1): Likewise.
6462         (cstoresi_ltu_thu): Likewise.
6463         (thumb1_addsi3_addgeu): Likewise.
6464         (*arm_jump): Likewise.
6465         (*thumb_jump): Likewise.
6466         (*check_arch2): Likewise.
6467         (arm_casesi_internal): Likewise.
6468         (thumb1_casesi_dispatch): Likewise.
6469         (*arm_indirect_jump): Likewise.
6470         (*thumb1_indirect_jump): Likewise.
6471         (nop): Likewise.
6472         (*and_scc): Likewise.
6473         (*ior_scc): Likewise.
6474         (*compare_scc): Likewise.
6475         (*cond_move): Likewise.
6476         (*cond_arith): Likewise.
6477         (*cond_sub): Likewise.
6478         (*cmp_ite0): Likewise.
6479         (*cmp_ite1): Likewise.
6480         (*cmp_and): Likewise.
6481         (*cmp_ior): Likewise.
6482         (*ior_scc_scc): Likewise.
6483         (*ior_scc_scc_cmp): Likewise.
6484         (*and_scc_scc): Likewise.
6485         (*and_scc_scc_cmp): Likewise.
6486         (*and_scc_scc_nod): Likewise.
6487         (*negscc): Likewise.
6488         (movcond_addsi): Likewise.
6489         (movcond): Likewise.
6490         (*ifcompare_plus_move): Likewise.
6491         (*if_plus_move): Likewise.
6492         (*ifcompare_move_plus): Likewise.
6493         (*if_move_plus): Likewise.
6494         (*ifcompare_arith_arith): Likewise.
6495         (*if_arith_arith): Likewise.
6496         (*ifcompare_arith_move): Likewise.
6497         (*if_arith_move): Likewise.
6498         (*ifcompare_move_arith): Likewise.
6499         (*if_move_arith): Likewise.
6500         (*ifcompare_move_not): Likewise.
6501         (*if_move_not): Likewise.
6502         (*ifcompare_not_move): Likewise.
6503         (*if_not_move): Likewise.
6504         (*ifcompare_shift_move): Likewise.
6505         (*if_shift_move): Likewise.
6506         (*ifcompare_move_shift): Likewise.
6507         (*if_move_shift): Likewise.
6508         (*ifcompare_shift_shift): Likewise.
6509         (*ifcompare_not_arith): Likewise.
6510         (*ifcompare_arith_not): Likewise.
6511         (*if_arith_not): Likewise.
6512         (*ifcompare_neg_move): Likewise.
6513         (*if_neg_move): Likewise.
6514         (*ifcompare_move_neg): Likewise.
6515         (*if_move_neg): Likewise.
6516         (prologue_thumb1_interwork): Likewise.
6517         (*cond_move_not): Likewise.
6518         (*sign_extract_onebit): Likewise.
6519         (*not_signextract_onebit): Likewise.
6520         (stack_tie): Likewise.
6521         (align_4): Likewise.
6522         (align_8): Likewise.
6523         (consttable_end): Likewise.
6524         (consttable_1): Likewise.
6525         (consttable_2): Likewise.
6526         (consttable_4): Likewise.
6527         (consttable_8): Likewise.
6528         (consttable_16): Likewise.
6529         (*thumb1_tablejump): Likewise.
6530         (prefetch): Likewise.
6531         (force_register_use): Likewise.
6532         (thumb_eh_return): Likewise.
6533         (load_tp_hard): Likewise.
6534         (load_tp_soft): Likewise.
6535         (tlscall): Likewise.
6536         (*arm_movtas_ze): Likewise.
6537         (*arm_rev): Likewise.
6538         (*arm_revsh): Likewise.
6539         (*arm_rev16): Likewise.
6540         * config/arm/thumb2.md
6541         (*thumb2_smaxsi3): Likewise.
6542         (*thumb2_sminsi3): Likewise.
6543         (*thumb32_umaxsi3): Likewise.
6544         (*thumb2_uminsi3): Likewise.
6545         (*thumb2_negdi2): Likewise.
6546         (*thumb2_abssi2): Likewise.
6547         (*thumb2_neg_abss): Likewise.
6548         (*thumb2_movsi_insn): Likewise.
6549         (tls_load_dot_plus_four): Likewise.
6550         (*thumb2_movhi_insn): Likewise.
6551         (*thumb2_mov_scc): Likewise.
6552         (*thumb2_mov_negs): Likewise.
6553         (*thumb2_mov_negs): Likewise.
6554         (*thumb2_mov_nots): Likewise.
6555         (*thumb2_mov_nots): Likewise.
6556         (*thumb2_movsicc_): Likewise.
6557         (*thumb2_movsfcc_soft_insn): Likewise.
6558         (*thumb2_indirect_jump): Likewise.
6559         (*thumb2_and_scc): Likewise.
6560         (*thumb2_ior_scc): Likewise.
6561         (*thumb2_ior_scc_strict_it): Likewise.
6562         (*thumb2_cond_move): Likewise.
6563         (*thumb2_cond_arith): Likewise.
6564         (*thumb2_cond_ari): Likewise.
6565         (*thumb2_cond_sub): Likewise.
6566         (*thumb2_negscc): Likewise.
6567         (*thumb2_movcond): Likewise.
6568         (thumb2_casesi_internal): Likewise.
6569         (thumb2_casesi_internal_pic): Likewise.
6570         (*thumb2_alusi3_short): Likewise.
6571         (*thumb2_mov<mode>_shortim): Likewise.
6572         (*thumb2_addsi_short): Likewise.
6573         (*thumb2_subsi_short): Likewise.
6574         (thumb2_addsi3_compare0): Likewise.
6575         (*thumb2_cbz): Likewise.
6576         (*thumb2_cbnz): Likewise.
6577         (*thumb2_one_cmplsi2_short): Likewise.
6578         (*thumb2_negsi2_short): Likewise.
6579         (*orsi_notsi_si): Likewise.
6580         * config/arm/arm1020e.md: Update with new attributes.
6581         * config/arm/arm1026ejs.md: Update with new attributes.
6582         * config/arm/arm1136jfs.md: Update with new attributes.
6583         * config/arm/arm926ejs.md: Update with new attributes.
6584         * config/arm/cortex-a15.md: Update with new attributes.
6585         * config/arm/cortex-a5.md: Update with new attributes.
6586         * config/arm/cortex-a53.md: Update with new attributes.
6587         * config/arm/cortex-a7.md: Update with new attributes.
6588         * config/arm/cortex-a8.md: Update with new attributes.
6589         * config/arm/cortex-a9.md: Update with new attributes.
6590         * config/arm/cortex-m4.md: Update with new attributes.
6591         * config/arm/cortex-r4.md: Update with new attributes.
6592         * config/arm/fa526.md: Update with new attributes.
6593         * config/arm/fa606te.md: Update with new attributes.
6594         * config/arm/fa626te.md: Update with new attributes.
6595         * config/arm/fa726te.md: Update with new attributes.
6597 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
6599         * config/aarch64/aarch64-simd.md
6600         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
6601         <vwx> iterator to ensure correct register choice.
6602         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
6603         (aarch64_sqdmull_n<mode>): Likewise.
6604         (aarch64_sqdmull2_n<mode>_internal): Likewise.
6605         * config/aarch64/arm_neon.h
6606         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
6607         (vml<as><q>_n_<su>16): Likewise.
6608         (vml<as>l_high_lane<q>_<su>16): Likewise.
6609         (vml<as>l_high_n_<su>16): Likewise.
6610         (vml<as>l_lane<q>_<su>16): Likewise.
6611         (vml<as>l_n_<su>16): Likewise.
6612         (vmul<q>_lane<q>_<su>16): Likewise.
6613         (vmul<q>_n_<su>16): Likewise.
6614         (vmull_lane<q>_<su>16): Likewise.
6615         (vmull_n_<su>16): Likewise.
6616         (vmull_high_lane<q>_<su>16): Likewise.
6617         (vmull_high_n_<su>16): Likewise.
6618         (vqrdmulh<q>_n_s16): Likewise.
6620 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
6622         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
6623         have the correct lane parameter.
6625 2013-09-06  Richard Biener <rguenther@suse.de>
6627         * cfganal.c (control_dependences::~control_dependences):
6628         Properly free all of the vector.
6630 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
6632         PR target/58269
6633         * config/i386/i386.c (ix86_conditional_register_usage):
6634         Proper initialize extended SSE registers.
6636 2013-09-06  Jan Hubicka  <jh@suse.cz>
6638         PR tree-optimization/58311
6639         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
6641 2013-09-06  Jan Hubicka  <jh@suse.cz>
6643         * Makefile.in (tree-sra.o): Update dependencies.
6644         * tree-sra.c: Include ipa-utils.h
6645         (scan_function): Use recursive_call_p.
6646         (has_caller_p): New function.
6647         (cgraph_for_node_and_aliases): Count also callers of aliases.
6649 2013-09-06  Jan Hubicka  <jh@suse.cz>
6651         PR middle-end/58094
6652         * cgraph.h (symtab_semantically_equivalent_p): Declare.
6653         * tree-tailcall.c: Include ipa-utils.h.
6654         (find_tail_calls): Use it.
6655         * ipa-pure-const.c (check_call): Likewise.
6656         * ipa-utils.c (recursive_call_p): New function.
6657         * ipa-utils.h (recursive_call_p): Dclare.
6658         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
6659         (symtab_semantically_equivalent_p): New function.
6660         * Makefile.in (tree-tailcall.o): Update dependencies.
6662 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
6664         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
6665         non-inlinable part.
6667 2013-09-06  Richard Biener  <rguenther@suse.de>
6669         * lto-streamer.h (lto_global_var_decls): Remove.
6670         * Makefile.in (OBJS): Remove lto-symtab.o.
6671         (lto-symtab.o): Remove.
6672         (GTFILES): Remove lto-symtab.c
6673         * lto-symtab.c: Move to lto/
6675 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6677         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
6678         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
6679         (UNSPEC_FPINT_RINT): New constant definitions.
6680         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
6681         definition with 2 attributes.
6682         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
6683         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
6684         definitions.
6686 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6688         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
6689         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
6690         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
6691         attribute to "z196_alone".
6692         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
6693         "zEC12_simple".
6695 2013-09-06  Richard Biener  <rguenther@suse.de>
6697         * basic-block.h (class control_dependences): New.
6698         * tree-ssa-dce.c (control_dependence_map): Remove.
6699         (cd): New global.
6700         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
6701         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
6702         find_pdom, find_control_dependence, find_all_control_dependences):
6703         Move to cfganal.c.
6704         (mark_control_dependent_edges_necessary,
6705         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
6706         tree_dce_done, perform_tree_ssa_dce): Adjust.
6707         * cfganal.c (set_control_dependence_map_bit,
6708         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
6709         find_all_control_dependences): Move from tree-ssa-dce.c and
6710         implement as methods of control_dependences class.
6711         (control_dependences::control_dependences): New.
6712         (control_dependences::~control_dependences): Likewise.
6713         (control_dependences::get_edges_dependent_on): Likewise.
6714         (control_dependences::get_edge): Likewise.
6716 2013-09-04  Jan Hubicka  <jh@suse.cz>
6718         * tree.c (types_same_for_odr): Drop overactive check.
6719         * ipa-devirt.c (hash_type_name): Likewise.
6721 2013-09-04  Jan Hubicka  <jh@suse.cz>
6723         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
6724         (analyze_functions): ... here.
6726 2013-09-04  Jan Hubicka  <jh@suse.cz>
6728         PR middle-end/58201
6729         * cgraphunit.c (analyze_functions): Clear AUX fields
6730         after processing; initialize assembler name has.
6732 2013-09-05  Jeff Law  <law@redhat.com>
6734         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
6735         from thread_around_empty_block.  Record threading path into PATH.
6736         Recurse if threading through the initial block is successful.
6737         (thread_across_edge): Corresponding changes to slightly simplify.
6739 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6741         * config/aarch64/aarch64.md
6742         (type): Remove frecpe, frecps, frecpx.
6743         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
6744         fix to be a TARGET_SIMD instruction.
6745         (aarch64_frecps): Remove.
6746         * config/aarch64/aarch64-simd.md
6747         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
6748         (aarch64_frecps<mode>): Handle all float/vector of float modes.
6750 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6751             Sofiane Naci  <sofiane.naci@arm.com>
6753         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
6754         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
6755         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
6756         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
6757         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
6758         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
6759         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
6760         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
6761         into "clz, "rbit".  Rename "shift" to "shift_imm".
6762         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
6763         changes.  Update for attribute changes all occurrences of arlo_* and
6764         shift* types.
6765         * config/arm/arm-fixed.md: Update for attribute changes
6766         all occurrences of arlo_* types.
6767         * config/arm/thumb2.md: Update for attribute changes all occurrences
6768         of arlo_* types.
6769         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
6770         (cortexa7_older_only): Likewise.
6771         (cortexa7_younger):  Likewise.
6772         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
6773         (1020alu_shift_op): Likewise.
6774         (1020alu_shift_reg_op): Likewise.
6775         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
6776         (alu_shift_op): Likewise.
6777         (alu_shift_reg_op): Likewise.
6778         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
6779         (11_alu_shift_op): Likewise.
6780         (11_alu_shift_reg_op): Likewise.
6781         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
6782         (9_alu_shift_reg_op): Likewise.
6783         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
6784         attribute changes.
6785         (cortex_a15_alu_shift): Likewise.
6786         (cortex_a15_alu_shift_reg): Likewise.
6787         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
6788         attribute changes.
6789         (cortex_a5_alu_shift): Likewise.
6790         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
6791         attribute changes.
6792         (cortex_a53_alu_shift): Likewise.
6793         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
6794         attribute changes.
6795         (cortex_a7_alu_reg): Likewise.
6796         (cortex_a7_alu_shift): Likewise.
6797         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
6798         attribute changes.
6799         (cortex_a8_alu_shift): Likewise.
6800         (cortex_a8_alu_shift_reg): Likewise.
6801         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
6802         (cortex_a9_dp_shift): Likewise.
6803         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
6804         attribute changes.
6805         * config/arm/cortex-r4.md
6806         (cortex_r4_alu): Update for attribute changes.
6807         (cortex_r4_mov): Likewise.
6808         (cortex_r4_alu_shift_reg): Likewise.
6809         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
6810         (526_alu_shift_op): Likewise.
6811         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
6812         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
6813         (626te_alu_shift_op): Likewise.
6814         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
6815         (726te_alu_shift_op): Likewise.
6816         (726te_alu_shift_reg_op): Likewise.
6817         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
6818         (mp626_alu_shift_op): Likewise.
6819         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
6820         (pj4_alu_conds): Likewise.
6821         (pj4_shift): Likewise.
6822         (pj4_shift_conds): Likewise.
6823         (pj4_alu_shift): Likewise.
6824         (pj4_alu_shift_conds): Likewise.
6825         * config/aarch64/aarch64.md: Update for attribute change
6826         all occurrences of arlo_* and shift* types.
6828 2013-09-05  Mike Stump  <mikestump@comcast.net>
6830         * tree.h: Move documentation for tree_function_decl to tree-core.h
6831         with the declaration.
6833 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
6835         PR target/58139
6836         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
6837         looking for widest mode.
6839 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
6841         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
6843 2013-09-05  Richard Biener  <rguenther@suse.de>
6845         PR tree-optimization/58137
6846         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
6847         Do not create vectors of pointers.
6848         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
6849         types for the components of the vector initializer.
6850         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
6851         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
6853 2013-09-05  Martin Jambor  <mjambor@suse.cz>
6855         * ipa-prop.c (remove_described_reference): Accept missing references,
6856         return false if that hppens, otherwise return true.
6857         (cgraph_node_for_jfunc): New function.
6858         (try_decrement_rdesc_refcount): Likewise.
6859         (try_make_edge_direct_simple_call): Use them.
6860         (ipa_edge_removal_hook): Remove references from rdescs.
6861         (ipa_edge_duplication_hook): Clone rdescs and their references
6862         when the new edge has the same caller as the old one.
6863         * cgraph.c (cgraph_resolve_speculation): Remove speculative
6864         reference before removing any edges.
6866 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
6868         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
6869         initial store.
6870         * thumb2.md (thumb2_storewb_parisi): New pattern.
6872 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
6874         * config/aarch64/aarch64-option-extensions.def: Add
6875         AARCH64_OPT_EXTENSION of 'crc'.
6876         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
6877         (AARCH64_ISA_CRC): Ditto.
6878         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
6879         description of the CRC extension.
6881 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6883         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
6884         * config/rs6000/linux.h: Ditto.
6885         * alpha/linux.h: Ditto.
6886         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
6887         no_c99_libc_has_function.
6888         * config/c6x/uclinux-elf.h: Ditto.
6889         * config/lm32/uclinux-elf.h: Ditto.
6890         * config/m68k/uclinux.h: Ditto.
6891         * config/moxie/uclinux.h: Ditto.
6892         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
6893         (crisv32-*-linux*, cris-*-linux*): Ditto.
6894         * config/bfin/bfin.c: Include "tm_p.h".
6896 2013-09-05  Richard Biener  <rguenther@suse.de>
6898         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
6899         check for a definition without a basic-block.
6901 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
6902             Sofiane Naci <sofiane.naci@arm.com>
6904         * config/aarch64/aarch64.md
6905         (*movti_aarch64): Rename r_2_f and f_2_r.
6906         (*movsf_aarch64): Likewise.
6907         (*movdf_aarch64): Likewise.
6908         (*movtf_aarch64): Likewise.
6909         (aarch64_movdi_<mode>low): Likewise.
6910         (aarch64_movdi_<mode>high): Likewise.
6911         (aarch64_mov<mode>high_di): Likewise.
6912         (aarch64_mov<mode>low_di): Likewise.
6913         (aarch64_movtilow_tilow): Likewise.
6914         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
6915         values to config/arm/types.md
6916         (attribute "conds"): Update for attribute change.
6917         (anddi3_insn): Likewise.
6918         (iordi3_insn): Likewise.
6919         (xordi3_insn): Likewise.
6920         (one_cmpldi2): Likewise.
6921         * config/arm/types.md (type): Add Neon types.
6922         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
6923         use "type" attribute.
6924         (movmisalign<mode>_neon_store): Likewise.
6925         (movmisalign<mode>_neon_load): Likewise.
6926         (vec_set<mode>_internal): Likewise.
6927         (vec_setv2di_internal): Likewise.
6928         (vec_extract<mode>): Likewise.
6929         (vec_extractv2di): Likewise.
6930         (add<mode>3_neon): Likewise.
6931         (adddi3_neon): Likewise.
6932         (sub<mode>3_neon): Likewise.
6933         (subdi3_neon): Likewise.
6934         (mul<mode>3_neon): Likewise.
6935         (mul<mode>3add<mode>_neon): Likewise.
6936         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
6937         (fma<VCVTF:mode>4)): Likewise.
6938         (fma<VCVTF:mode>4_intrinsic): Likewise.
6939         (fmsub<VCVTF:mode>4)): Likewise.
6940         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
6941         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
6942         (ior<mode>3): Likewise.
6943         (and<mode>3): Likewise.
6944         (anddi3_neon): Likewise.
6945         (orn<mode>3_neon): Likewise.
6946         (orndi3_neon): Likewise.
6947         (bic<mode>3_neon): Likewise.
6948         (bicdi3_neon): Likewise.
6949         (xor<mode>3): Likewise.
6950         (one_cmpl<mode>2): Likewise.
6951         (abs<mode>2): Likewise.
6952         (neg<mode>2): Likewise.
6953         (umin<mode>3_neon): Likewise.
6954         (umax<mode>3_neon): Likewise.
6955         (smin<mode>3_neon): Likewise.
6956         (smax<mode>3_neon): Likewise.
6957         (vashl<mode>3): Likewise.
6958         (vashr<mode>3_imm): Likewise.
6959         (vlshr<mode>3_imm): Likewise.
6960         (ashl<mode>3_signed): Likewise.
6961         (ashl<mode>3_unsigned): Likewise.
6962         (neon_load_count): Likewise.
6963         (ashldi3_neon_noclobber): Likewise.
6964         (signed_shift_di3_neon): Likewise.
6965         (unsigned_shift_di3_neon): Likewise.
6966         (ashrdi3_neon_imm_noclobber): Likewise.
6967         (lshrdi3_neon_imm_noclobber): Likewise.
6968         (widen_ssum<mode>3): Likewise.
6969         (widen_usum<mode>3): Likewise.
6970         (quad_halves_<code>v4si): Likewise.
6971         (quad_halves_<code>v4sf): Likewise.
6972         (quad_halves_<code>v8hi): Likewise.
6973         (quad_halves_<code>v16qi): Likewise.
6974         (reduc_splus_v2di): Likewise.
6975         (neon_vpadd_internal<mode>): Likewise.
6976         (neon_vpsmin<mode>): Likewise.
6977         (neon_vpsmax<mode>): Likewise.
6978         (neon_vpumin<mode>): Likewise.
6979         (neon_vpumax<mode>): Likewise.
6980         (ss_add<mode>_neon): Likewise.
6981         (us_add<mode>_neon): Likewise.
6982         (ss_sub<mode>_neon): Likewise.
6983         (us_sub<mode>_neon): Likewise.
6984         (neon_vadd<mode>_unspec): Likewise.
6985         (neon_vaddl<mode>): Likewise.
6986         (neon_vaddw<mode>): Likewise.
6987         (neon_vhadd<mode>): Likewise.
6988         (neon_vqadd<mode>): Likewise.
6989         (neon_vaddhn<mode>): Likewise.
6990         (neon_vmul<mode>): Likewise.
6991         (neon_vmla<mode>): Likewise.
6992         (neon_vmlal<mode>): Likewise.
6993         (neon_vmls<mode>): Likewise.
6994         (neon_vmlsl<mode>): Likewise.
6995         (neon_vqdmulh<mode>): Likewise.
6996         (neon_vqdmlal<mode>): Likewise.
6997         (neon_vqdmlsl<mode>): Likewise.
6998         (neon_vmull<mode>): Likewise.
6999         (neon_vqdmull<mode>): Likewise.
7000         (neon_vsub<mode>_unspec): Likewise.
7001         (neon_vsubl<mode>): Likewise.
7002         (neon_vsubw<mode>): Likewise.
7003         (neon_vqsub<mode>): Likewise.
7004         (neon_vhsub<mode>): Likewise.
7005         (neon_vsubhn<mode>): Likewise.
7006         (neon_vceq<mode>): Likewise.
7007         (neon_vcge<mode>): Likewise.
7008         (neon_vcgeu<mode>): Likewise.
7009         (neon_vcgt<mode>): Likewise.
7010         (neon_vcgtu<mode>): Likewise.
7011         (neon_vcle<mode>): Likewise.
7012         (neon_vclt<mode>): Likewise.
7013         (neon_vcage<mode>): Likewise.
7014         (neon_vcagt<mode>): Likewise.
7015         (neon_vtst<mode>): Likewise.
7016         (neon_vabd<mode>): Likewise.
7017         (neon_vabdl<mode>): Likewise.
7018         (neon_vaba<mode>): Likewise.
7019         (neon_vabal<mode>): Likewise.
7020         (neon_vmax<mode>): Likewise.
7021         (neon_vmin<mode>): Likewise.
7022         (neon_vpaddl<mode>): Likewise.
7023         (neon_vpadal<mode>): Likewise.
7024         (neon_vpmax<mode>): Likewise.
7025         (neon_vpmin<mode>): Likewise.
7026         (neon_vrecps<mode>): Likewise.
7027         (neon_vrsqrts<mode>): Likewise.
7028         (neon_vqabs<mode>): Likewise.
7029         (neon_vqneg<mode>): Likewise.
7030         (neon_vcls<mode>): Likewise.
7031         (clz<mode>2): Likewise.
7032         (popcount<mode>2): Likewise.
7033         (neon_vrecpe): Likewise.
7034         (neon_vrsqrte): Likewise.
7035         (neon_vget_lane<mode>_sext_internal): Likewise.
7036         (neon_vget_lane<mode>_zext_internal): Likewise.
7037         (neon_vdup_n<mode>): Likewise.
7038         (neon_vdup_nv2di): Likewise.
7039         (neon_vdpu_lane<mode>_internal): Likewise.
7040         (neon_vswp<mode>): Likewise.
7041         (float<mode><V_cvtto>2): Likewise.
7042         (floatuns<mode><V_cvtto>2): Likewise.
7043         (fix_trunc<mode><V_cvtto>)2): Likewise
7044         (fixuns_trunc<mode><V_cvtto)2): Likewise.
7045         (neon_vcvt<mode>): Likewise.
7046         (neon_vcvtv4sfv4hf): Likewise.
7047         (neon_vcvtv4hfv4sf): Likewise.
7048         (neon_vcvt_n<mode>): Likewise.
7049         (neon_vmovn<mode>): Likewise.
7050         (neon_vqmovn<mode>): Likewise.
7051         (neon_vqmovun<mode>): Likewise.
7052         (neon_vmovl<mode>): Likewise.
7053         (neon_vmul_lane<mode>): Likewise.
7054         (neon_vmull_lane<mode>): Likewise.
7055         (neon_vqdmull_lane<mode>): Likewise.
7056         (neon_vqdmulh_lane<mode>): Likewise.
7057         (neon_vmla_lane<mode>): Likewise.
7058         (neon_vmlal_lane<mode>): Likewise.
7059         (neon_vqdmlal_lane<mode>): Likewise.
7060         (neon_vmls_lane<mode>): Likewise.
7061         (neon_vmlsl_lane<mode>): Likewise.
7062         (neon_vqdmlsl_lane<mode>): Likewise.
7063         (neon_vext<mode>): Likewise.
7064         (neon_vrev64<mode>): Likewise.
7065         (neon_vrev32<mode>): Likewise.
7066         (neon_vrev16<mode>): Likewise.
7067         (neon_vbsl<mode>_internal): Likewise.
7068         (neon_vshl<mode>): Likewise.
7069         (neon_vqshl<mode>): Likewise.
7070         (neon_vshr_n<mode>): Likewise.
7071         (neon_vshrn_n<mode>): Likewise.
7072         (neon_vqshrn_n<mode>): Likewise.
7073         (neon_vqshrun_n<mode>): Likewise.
7074         (neon_vshl_n<mode>): Likewise.
7075         (neon_vqshl_n<mode>): Likewise.
7076         (neon_vqshlu_n<mode>): Likewise.
7077         (neon_vshll_n<mode>): Likewise.
7078         (neon_vsra_n<mode>): Likewise.
7079         (neon_vsri_n<mode>): Likewise.
7080         (neon_vsli_n<mode>): Likewise.
7081         (neon_vtbl1v8qi): Likewise.
7082         (neon_vtbl2v8qi): Likewise.
7083         (neon_vtbl3v8qi): Likewise.
7084         (neon_vtbl4v8qi): Likewise.
7085         (neon_vtbx1v8qi): Likewise.
7086         (neon_vtbx2v8qi): Likewise.
7087         (neon_vtbx3v8qi): Likewise.
7088         (neon_vtbx4v8qi): Likewise.
7089         (neon_vtrn<mode>_internal): Likewise.
7090         (neon_vzip<mode>_internal): Likewise.
7091         (neon_vuzp<mode>_internal): Likewise.
7092         (neon_vld1<mode>): Likewise.
7093         (neon_vld1_lane<mode>): Likewise.
7094         (neon_vld1_dup<mode>): Likewise.
7095         (neon_vld1_dupv2di): Likewise.
7096         (neon_vst1<mode>): Likewise.
7097         (neon_vst1_lane<mode>): Likewise.
7098         (neon_vld2<mode>): Likewise.
7099         (neon_vld2_lane<mode>): Likewise.
7100         (neon_vld2_dup<mode>): Likewise.
7101         (neon_vst2<mode>): Likewise.
7102         (neon_vst2_lane<mode>): Likewise.
7103         (neon_vld3<mode>): Likewise.
7104         (neon_vld3qa<mode>): Likewise.
7105         (neon_vld3qb<mode>): Likewise.
7106         (neon_vld3_lane<mode>): Likewise.
7107         (neon_vld3_dup<mode>): Likewise.
7108         (neon_vst3<mode>): Likewise.
7109         (neon_vst3qa<mode>): Likewise.
7110         (neon_vst3qb<mode>): Likewise.
7111         (neon_vst3_lane<mode>): Likewise.
7112         (neon_vld4<mode>): Likewise.
7113         (neon_vld4qa<mode>): Likewise.
7114         (neon_vld4qb<mode>): Likewise.
7115         (neon_vld4_lane<mode>): Likewise.
7116         (neon_vld4_dup<mode>): Likewise.
7117         (neon_vst4<mode>): Likewise.
7118         (neon_vst4qa<mode>): Likewise.
7119         (neon_vst4qb<mode>): Likewise.
7120         (neon_vst4_lane<mode>): Likewise.
7121         (neon_vec_unpack<US>_lo_<mode>): Likewise.
7122         (neon_vec_unpack<US>_hi_<mode>): Likewise.
7123         (neon_vec_<US>mult_lo_<mode>): Likewise.
7124         (neon_vec_<US>mult_hi_<mode>): Likewise.
7125         (neon_vec_<US>shiftl_<mode>): Likewise.
7126         (neon_unpack<US>_<mode>): Likewise.
7127         (neon_vec_<US>mult_<mode>): Likewise.
7128         (vec_pack_trunc_<mode>): Likewise.
7129         (neon_vec_pack_trunk_<mode>): Likewise.
7130         (neon_vabd<mode>_2): Likewise.
7131         (neon_vabd<mode>_3): Likewise.
7132         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
7133         (thumb2_movsi_vfp): Likewise.
7134         (movdi_vfp): Likewise.
7135         (movdi_vfp_cortexa8): Likewise.
7136         (movhf_vfp_neon): Likewise.
7137         (movhf_vfp): Likewiwse.
7138         (movsf_vfp): Likewiwse.
7139         (thumb2_movsf_vfp): Likewiwse.
7140         (movdf_vfp): Likewise.
7141         (thumb2_movdf_vfp): Likewise.
7142         (movsfcc_vfp): Likewise.
7143         (thumb2_movsfcc_vfp): Likewise.
7144         (movdfcc_vfp): Likewise.
7145         (thumb2_movdfcc_vfp): Likewise.
7146         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
7147         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
7148         (v10_v2c): Likewise.
7149         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
7150         attribute change.
7151         (cortex_a15_neon_int_2): Likewise.
7152         (cortex_a15_neon_int_3): Likewise.
7153         (cortex_a15_neon_int_4): Likewise.
7154         (cortex_a15_neon_int_5): Likewise.
7155         (cortex_a15_neon_vqneg_vqabs): Likewise.
7156         (cortex_a15_neon_vmov): Likewise.
7157         (cortex_a15_neon_vaba): Likewise.
7158         (cortex_a15_neon_vaba_qqq): Likewise.
7159         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7160         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7161         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
7162         scalar_64_32_long_scalar): Likewise.
7163         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7164         (cortex_a15_neon_mla_qqq_8_16): Likewise.
7165         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7166         lotype_qdd_64_32_long): Likewise.
7167         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7168         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7169         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
7170         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7171         (cortex_a15_neon_shift_1): Likewise.
7172         (cortex_a15_neon_shift_2): Likewise.
7173         (cortex_a15_neon_shift_3): Likewise.
7174         (cortex_a15_neon_vshl_ddd): Likewise.
7175         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7176         (cortex_a15_neon_vsra_vrsra): Likewise.
7177         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
7178         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
7179         (cortex_a15_neon_fp_vmul_ddd): Likewise.
7180         (cortex_a15_neon_fp_vmul_qqd): Likewise.
7181         (cortex_a15_neon_fp_vmla_ddd): Likewise.
7182         (cortex_a15_neon_fp_vmla_qqq): Likewise.
7183         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
7184         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
7185         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7186         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7187         (cortex_a15_neon_bp_simple): Likewise.
7188         (cortex_a15_neon_bp_2cycle): Likewise.
7189         (cortex_a15_neon_bp_3cycle): Likewise.
7190         (cortex_a15_neon_vld1_1_2_regs): Likewise.
7191         (cortex_a15_neon_vld1_3_4_regs): Likewise.
7192         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7193         (cortex_a15_neon_vld2_4_regs): Likewise.
7194         (cortex_a15_neon_vld3_vld4): Likewise.
7195         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7196         (cortex_a15_neon_vst1_3_4_regs): Likewise.
7197         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
7198         (cortex_a15_neon_vst3_vst4): Likewise.
7199         (cortex_a15_neon_vld1_vld2_lane): Likewise.
7200         (cortex_a15_neon_vld3_vld4_lane" 10
7201         (cortex_a15_neon_vst1_vst2_lane): Likewise.
7202         (cortex_a15_neon_vst3_vst4_lane): Likewise.
7203         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
7204         (cortex_a15_neon_ldm_2): Likewise.0
7205         (cortex_a15_neon_stm_2): Likewise.
7206         (cortex_a15_neon_mcr): Likewise.
7207         (cortex_a15_neon_mcr_2_mcrr): Likewise.
7208         (cortex_a15_neon_mrc): Likewise.
7209         (cortex_a15_neon_mrrc): Likewise.
7210         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
7211         change.
7212         (cortex_a15_alu_shift): Likewise.
7213         (cortex_a15_alu_shift_reg): Likewise.
7214         (cortex_a15_mult32): Likewise.
7215         (cortex_a15_mult64): Likewise.
7216         (cortex_a15_block): Likewise.
7217         (cortex_a15_branch): Likewise.
7218         (cortex_a15_load1): Likewise.
7219         (cortex_a15_load3): Likewise.
7220         (cortex_a15_store1): Likewise.
7221         (cortex_a15_store3): Likewise.
7222         (cortex_a15_call): Likewise.
7223         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
7224         (cortex_a5_f2r): Likewise.
7225         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
7226         change.
7227         (cortex_a53_f2r): Likewise.
7228         * config/arm/cortex-a7.md
7229         (cortex_a7_branch): Update for attribute change.
7230         (cortex_a7_call): Likewise.
7231         (cortex_a7_alu_imm): Likewise.
7232         (cortex_a7_alu_reg): Likewise.
7233         (cortex_a7_alu_shift): Likewise.
7234         (cortex_a7_mul): Likewise.
7235         (cortex_a7_load1): Likewise.
7236         (cortex_a7_store1): Likewise.
7237         (cortex_a7_load2): Likewise.
7238         (cortex_a7_store2): Likewise.
7239         (cortex_a7_load3): Likewise.
7240         (cortex_a7_store3): Likewise.
7241         (cortex_a7_load4): Likewise.
7242         (cortex_a7_store4): Likewise.
7243         (cortex_a7_fpalu): Likewise.
7244         (cortex_a7_fconst): Likewise.
7245         (cortex_a7_fpmuls): Likewise.
7246         (cortex_a7_neon_mul): Likewise.
7247         (cortex_a7_fpmacs): Likewise.
7248         (cortex_a7_neon_mla: Likewise.
7249         (cortex_a7_fpmuld: Likewise.
7250         (cortex_a7_fpmacd: Likewise.
7251         (cortex_a7_fpfmad: Likewise.
7252         (cortex_a7_fdivs: Likewise.
7253         (cortex_a7_fdivd: Likewise.
7254         (cortex_a7_r2f: Likewise.
7255         (cortex_a7_f2r: Likewise.
7256         (cortex_a7_f_flags: Likewise.
7257         (cortex_a7_f_loads: Likewise.
7258         (cortex_a7_f_loadd: Likewise.
7259         (cortex_a7_f_stores: Likewise.
7260         (cortex_a7_f_stored: Likewise.
7261         (cortex_a7_neon): Likewise.
7262         * config/arm/cortex-a8-neon.md
7263         (cortex_a8_neon_mrc): Update for attribute change.
7264         (cortex_a8_neon_mrrc): Likewise.
7265         (cortex_a8_neon_int_1): Likewise.
7266         (cortex_a8_neon_int_2): Likewise.
7267         (cortex_a8_neon_int_3): Likewise.
7268         (cortex_a8_neon_int_4): Likewise.
7269         (cortex_a8_neon_int_5): Likewise.
7270         (cortex_a8_neon_vqneg_vqabs): Likewise.
7271         (cortex_a8_neon_vmov): Likewise.
7272         (cortex_a8_neon_vaba): Likewise.
7273         (cortex_a8_neon_vaba_qqq): Likewise.
7274         (cortex_a8_neon_vsma): Likewise.
7275         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7276         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7277         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7278         long_scalar): Likewise.
7279         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7280         (cortex_a8_neon_mla_qqq_8_16): Likewise.
7281         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7282         long_scalar_qdd_64_32_long): Likewise.
7283         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7284         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7285         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
7286         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7287         (cortex_a8_neon_shift_1): Likewise.
7288         (cortex_a8_neon_shift_2): Likewise.
7289         (cortex_a8_neon_shift_3): Likewise.
7290         (cortex_a8_neon_vshl_ddd): Likewise.
7291         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7292         (cortex_a8_neon_vsra_vrsra): Likewise.
7293         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
7294         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
7295         (cortex_a8_neon_fp_vsum): Likewise.
7296         (cortex_a8_neon_fp_vmul_ddd): Likewise.
7297         (cortex_a8_neon_fp_vmul_qqd): Likewise.
7298         (cortex_a8_neon_fp_vmla_ddd): Likewise.
7299         (cortex_a8_neon_fp_vmla_qqq): Likewise.
7300         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
7301         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
7302         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7303         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7304         (cortex_a8_neon_bp_simple): Likewise.
7305         (cortex_a8_neon_bp_2cycle): Likewise.
7306         (cortex_a8_neon_bp_3cycle): Likewise.
7307         (cortex_a8_neon_ldr): Likewise.
7308         (cortex_a8_neon_str): Likewise.
7309         (cortex_a8_neon_vld1_1_2_regs): Likewise.
7310         (cortex_a8_neon_vld1_3_4_regs): Likewise.
7311         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7312         (cortex_a8_neon_vld2_4_regs): Likewise.
7313         (cortex_a8_neon_vld3_vld4): Likewise.
7314         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7315         (cortex_a8_neon_vst1_3_4_regs): Likewise.
7316         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
7317         (cortex_a8_neon_vst3_vst4): Likewise.
7318         (cortex_a8_neon_vld1_vld2_lane): Likewise.
7319         (cortex_a8_neon_vld3_vld4_lane): Likewise.
7320         (cortex_a8_neon_vst1_vst2_lane): Likewise.
7321         (cortex_a8_neon_vst3_vst4_lane): Likewise.
7322         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
7323         (cortex_a8_neon_mcr): Likewise.
7324         (cortex_a8_neon_mcr_2_mcrr): Likewise.
7325         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
7326         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
7327         change.
7328         (ca9_neon_mrrc): Likewise.
7329         (cortex_a9_neon_int_1): Likewise.
7330         (cortex_a9_neon_int_2): Likewise.
7331         (cortex_a9_neon_int_3): Likewise.
7332         (cortex_a9_neon_int_4): Likewise.
7333         (cortex_a9_neon_int_5): Likewise.
7334         (cortex_a9_neon_vqneg_vqabs): Likewise.
7335         (cortex_a9_neon_vmov): Likewise.
7336         (cortex_a9_neon_vaba): Likewise.
7337         (cortex_a9_neon_vaba_qqq): Likewise.
7338         (cortex_a9_neon_vsma): Likewise.
7339         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7340         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
7341         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
7342         long_scalar): Likewise.
7343         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
7344         (cortex_a9_neon_mla_qqq_8_16): Likewise.
7345         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
7346         long_scalar_qdd_64_32_long): Likewise.
7347         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
7348         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
7349         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
7350         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
7351         (cortex_a9_neon_shift_1): Likewise.
7352         (cortex_a9_neon_shift_2): Likewise.
7353         (cortex_a9_neon_shift_3): Likewise.
7354         (cortex_a9_neon_vshl_ddd): Likewise.
7355         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
7356         (cortex_a9_neon_vsra_vrsra): Likewise.
7357         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
7358         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
7359         (cortex_a9_neon_fp_vsum): Likewise.
7360         (cortex_a9_neon_fp_vmul_ddd): Likewise.
7361         (cortex_a9_neon_fp_vmul_qqd): Likewise.
7362         (cortex_a9_neon_fp_vmla_ddd): Likewise.
7363         (cortex_a9_neon_fp_vmla_qqq): Likewise.
7364         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
7365         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
7366         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
7367         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
7368         (cortex_a9_neon_bp_simple): Likewise.
7369         (cortex_a9_neon_bp_2cycle): Likewise.
7370         (cortex_a9_neon_bp_3cycle): Likewise.
7371         (cortex_a9_neon_ldr): Likewise.
7372         (cortex_a9_neon_str): Likewise.
7373         (cortex_a9_neon_vld1_1_2_regs): Likewise.
7374         (cortex_a9_neon_vld1_3_4_regs): Likewise.
7375         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
7376         (cortex_a9_neon_vld2_4_regs): Likewise.
7377         (cortex_a9_neon_vld3_vld4): Likewise.
7378         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
7379         (cortex_a9_neon_vst1_3_4_regs): Likewise.
7380         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
7381         (cortex_a9_neon_vst3_vst4): Likewise.
7382         (cortex_a9_neon_vld1_vld2_lane): Likewise.
7383         (cortex_a9_neon_vld3_vld4_lane): Likewise.
7384         (cortex_a9_neon_vst1_vst2_lane): Likewise.
7385         (cortex_a9_neon_vst3_vst4_lane): Likewise.
7386         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
7387         (cortex_a9_neon_mcr): Likewise.
7388         (cortex_a9_neon_mcr_2_mcrr): Likewise.
7389         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
7390         (cortex_a9_fps): Likewise.
7391         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
7392         change.
7393         (cortex_m4_fmuls): Likewise.
7394         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
7395         change.
7396         (cortex_r4_mrc): Likewise.
7397         * config/arm/iterators.md: Update comment referring to neon_type.
7398         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
7399         (iwmmxt_movsi_insn): Likewise.
7400         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
7401         attribute change.
7402         (pj4_core_to_vfp): Likewise.
7403         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
7404         attribute change.
7405         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
7406         (vfp_fstore): Likewise.
7407         * doc/md.texi: Change references to neon_type to refer to type.
7409 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7411         * tree.h (DECL_BUILT_IN): Fix typo in comment.
7413 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
7415         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
7416         lglobl if not weak.
7418 2013-09-04  Easwaran Raman  <eraman@google.com>
7420         PR middle-end/57370
7421         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
7422         (build_and_add_sum): Use it.
7423         (appears_later_in_bb): Simplify code.
7425 2013-09-04  Teresa Johnson  <tejohnson@google.com>
7427         * dumpfile.c (dump_finish): Don't close stderr/stdout.
7429 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
7431         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
7433 2013-09-04  Jan Hubicka  <jh@suse.cz>
7435         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
7436         * ipa-devirt.c: Include diganostic.h
7437         (odr_type_d): Add types and types_set.
7438         (hash_type_name): Work for types with vtables during LTO.
7439         (odr_hasher::remove): Fix comment; destroy types_set.
7440         (add_type_duplicate): New function,
7441         (get_odr_type): Use it.
7442         (dump_type_inheritance_graph): Dump type duplicates.
7443         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
7444         graph.
7445         * tree.c (types_same_for_odr): Give exact answers on types with
7446         virtual tables.
7448 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
7450         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
7451         explaining their differences.
7453 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
7455         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
7457 2013-09-03  Jeff Law  <law@redhat.com>
7459         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
7460         when not threading through a joiner block.  Pass joiner/no joiner
7461         state to register_jump_thread.
7462         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
7463         state from argument rather than implying on path length.
7464         Dump the entire jump thread path into debugging dump.
7465         * tree-flow.h (register_jump_thread): Update prototype.
7467 2013-08-29  Xinliang David Li  <davidxl@google.com>
7469         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7470         Remove a trivial gcc_assert.
7472 2013-08-29  Xinliang David Li  <davidxl@google.com>
7474         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
7475         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
7476         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
7477         Delay base decl alignment adjustment.
7478         * tree-vectorizer.c (vect_destroy_datarefs): New function.
7479         * tree-vectorizer.h: New data structure.
7480         (set_dr_misalignment): New function.
7481         (dr_misalignment): Ditto.
7482         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
7483         (vectorizable_load): Ditto.
7484         (ensure_base_align): New function.
7485         (vectorize_loops): Add dbg_cnt support.
7486         (execute_vect_slp): Ditto.
7487         * dbgcnt.def: New debug counter.
7488         * Makefile: New dependency.
7490 2013-09-03  Meador Inge  <meadori@codesourcery.com>
7492         Revert:
7494         2013-08-30  Meador Inge  <meadori@codesourcery.com>
7496         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7498 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
7500         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
7501         function descriptor.
7503 2013-09-03  Richard Biener  <rguenther@suse.de>
7505         * tree-affine.c (add_elt_to_tree): Fix association issue,
7506         avoid useless converts and make sure to always return a
7507         properly typed result.
7509 2013-09-03  Richard Biener  <rguenther@suse.de>
7511         PR middle-end/57656
7512         * fold-const.c (negate_expr_p): Fix division case.
7513         (negate_expr): Likewise.
7515 2013-09-03  Richard Biener  <rguenther@suse.de>
7517         PR lto/58285
7518         * tree-streamer-out.c: Include tm.h.
7519         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
7521 2013-09-03  Jan Hubicka  <jh@suse.cz>
7523         * tree-profile.c (tree_profiling): Cleanup CFG when done.
7525 2013-09-03  Alan Modra  <amodra@gmail.com>
7527         * config.gcc (powerpc*-*-linux*): Add support for little-endian
7528         multilibs to big-endian target and vice versa.
7529         * config/rs6000/t-linux64: Use := assignment on all vars.
7530         (MULTILIB_EXTRA_OPTS): Remove fPIC.
7531         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
7532         * config/rs6000/t-linux64le: New file.
7533         * config/rs6000/t-linux64bele: New file.
7534         * config/rs6000/t-linux64lebe: New file.
7536 2013-09-02  Jan Hubicka  <jh@suse.cz>
7538         * ipa-inline-transform.c (inline_transform): Do not
7539         optimize_inline_calls when not optimizing.
7541 2013-09-02  Jan Hubicka  <jh@suse.cz>
7543         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
7544         duplicated nodes for assembler names.
7545         * symtab.c (symtab_unregister_node): Do not attempt to unlink
7546         hard registers from assembler name hash.
7548 2013-09-02  Jan Hubicka  <jh@suse.cz>
7550         * ipa-split.c (execute_split_functions): Split externally visible
7551         functions called once.
7553 2013-09-02  Martin Jambor  <mjambor@suse.cz>
7555         PR ipa/58106
7556         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
7557         linked list.  When finding the correct duplicate, also consider also
7558         the caller in additon to its inlined_to node.
7560 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
7562         * config/aarch64/aarch64-simd-builtins.def
7563         (dup_lane_scalar): Remove.
7564         * config/aarch64/aarch64-simd.md
7565         (aarch64_simd_dup): Add 'w->w' alternative.
7566         (aarch64_dup_lane<mode>): Allow for VALL.
7567         (aarch64_dup_lane_scalar<mode>): Remove.
7568         (aarch64_dup_lane_<vswap_width_name><mode>): New.
7569         (aarch64_get_lane_signed<mode>): Add w->w altenative.
7570         (aarch64_get_lane_unsigned<mode>): Likewise.
7571         (aarch64_get_lane<mode>): Likewise.
7572         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
7573         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
7574         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
7575         (VCON): Change container of V2SF.
7576         (vswap_width_name): Likewise.
7577         * config/aarch64/arm_neon.h
7578         (__aarch64_vdup_lane_any): New.
7579         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7580         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
7581         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
7583 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
7585         PR middle-end/56382
7586         * expr.c (emit_move_complex): Do not move complex FP values as parts if
7587         the source or the destination is a single hard register.
7589 2013-09-02  Richard Biener  <rguenther@suse.de>
7591         PR middle-end/57511
7592         * tree-scalar-evolution.c (instantiate_scev_name): Allow
7593         non-linear SCEVs.
7595 2013-09-02  Richard Biener  <rguenther@suse.de>
7597         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
7598         arithmetic to sizetype.
7600 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
7602         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
7603         Find auto-increment use both before and after candidate.
7605 2013-09-02  Marek Polacek  <polacek@redhat.com>
7607         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
7609 2013-09-01  Jan Hubicka  <jh@suse.cz>
7611         * Makefile.in: Add ipa-profile.o
7612         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
7613         * cgraph.c (struct cgraph_propagate_frequency_data,
7614         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
7615         ipa-profile.c; replace cgraph_ by ipa_ prefix.
7616         * cgraph.h (cgraph_propagate_frequency): Remove.
7617         * ipa-inline-analysis.c: Include ipa-utils.h;
7618         drop duplicated cfgloop.h.
7619         (inline_update_callee_summaries): Update.
7620         * ipa-profile.c: New file.
7621         * ipa-utils.h (ipa_propagate_frequency): Declare.
7622         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
7623         data-streamer.h, value-prof.h.
7624         (symtab_remove_unreachable_nodes): Update profile.
7625         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
7626         account_time_size, cmp_counts, dump_histogram,
7627         ipa_profile_generate_summary, ipa_profile_write_summary,
7628         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
7629         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
7630         Move to ipa-profile.c.
7632 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
7634         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
7636 2013-09-01  Jan Hubicka  <jh@suse.cz>
7638         * common.opt (fdevirtualize-speculatively): New function.
7639         * invoke.texi (fdevirtualize-speculatively): Document.
7640         * ipa-devirt.c: Include ipa-inline.h
7641         (likely_target_p): New function.
7642         (ipa_devirt): New function.
7643         (gate_ipa_devirt): New function.
7644         (pass_data_ipa_devirt): New static var.
7645         (pass_ipa_devirt): Likewise.
7646         (make_pass_ipa_devirt): New function.
7647         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
7648         (common_handle_option): Disable devirtualization when
7649         value range profiling is available.
7650         * passes.def (pass_ipa_devirt): Add.
7651         * timever.def (TV_IPA_DEVIRT): New timevar.
7652         * tree-pass.h (make_pass_ipa_devirt):
7654 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
7656         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
7657         include sanitize(undefined).
7659 2013-08-31  Diego Novillo  <dnovillo@google.com>
7661         * Makefile.in (TREE_CORE_H): Define.
7662         (TREE_H): Use.
7663         (GTFILES): Add tree-core.h.
7664         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
7665         size the array.
7666         * tree-core.h: New file.
7667         Move all data structures, enum, typedefs, global
7668         declarations and constants from ...
7669         * tree.h: ... here.
7671 2013-08-31  Jan Hubicka  <jh@suse.cz>
7673         * bulitins.c (expand_builtin): Do not early exit for gcov
7674         instrumented functions.
7676 2013-08-31  Marek Polacek  <polacek@redhat.com>
7678         * ubsan.c: Include tm_p.h.
7680 2013-08-31  Jan Hubicka  <jh@suse.cz>
7682         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
7683         warning.
7685         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
7686         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
7687         * tree-cfg.c (verify_gimple_label): ... here.
7688         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
7689         (ipa_merge_profiles): New function.
7690         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
7691         (lto_input_function_body): Likewise.
7692         * ipa-utils.h (ipa_merge_profiles): Declare.
7693         * lto-streamer.h (lto_input_function_body): Update prototype.
7694         (emit_label_in_global_context_p): Remove.
7695         * lto-symtab.c: Include ipa-utils.h
7696         (lto_cgraph_replace_node): Use ipa_merge_profiles.
7698 2013-08-31  Jan Hubicka  <jh@suse.cz>
7700         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
7702 2013-08-31  Jan Hubicka  <jh@suse.cz>
7704         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
7706 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
7708         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
7709         "cmp" RTX before signed_comparison_operator check to account
7710         for "code" changes.
7712 2013-08-30  Jan Hubicka  <jh@suse.cz>
7714         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
7715         (detect_type_change_1): Rename to ...
7716         (detect_type_change): ... this one; early return on non-polymorphic
7717         types.
7718         (detect_type_change_ssa): Add comp_type parameter; update
7719         use of detect_type_change.
7720         (compute_complex_assign_jump_func): Add param_type parameter;
7721         update use of detect_type_change_ssa.
7722         (compute_complex_ancestor_jump_func): Likewise.
7723         (ipa_get_callee_param_type): New function.
7724         (ipa_compute_jump_functions_for_edge): Compute parameter type;
7725         update calls to the jump function computation functions.
7727 2013-08-30  Teresa Johnson  <tejohnson@google.com>
7728             Steven Bosscher  <steven@gcc.gnu.org>
7730         * cfgrtl.c (fixup_new_cold_bb): New routine.
7731         (commit_edge_insertions): Invoke fixup_partitions.
7732         (find_partition_fixes): New routine.
7733         (fixup_partitions): Ditto.
7734         (verify_hot_cold_block_grouping): Update comments.
7735         (rtl_verify_edges): Invoke find_partition_fixes.
7736         (rtl_verify_bb_pointers): Update comments.
7737         (rtl_verify_bb_layout): Ditto.
7738         * basic-block.h (probably_never_executed_edge_p): Declare.
7739         (fixup_partitions): Ditto.
7740         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
7741         * bb-reorder.c (sanitize_hot_paths): New function.
7742         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
7743         sanitize_hot_paths.
7744         * predict.c (probably_never_executed_edge_p): New routine.
7745         * cfg.c (check_bb_profile): Add partition insanity warnings.
7747 2013-08-30  Meador Inge  <meadori@codesourcery.com>
7749         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
7751 2013-08-30  Marek Polacek  <polacek@redhat.com>
7753         * Makefile.in (ubsan.o): Add.
7754         (c-family/c-ubsan.o): Add.
7755         (builtins.o): Add ubsan.h dependency.
7756         * ubsan.h: New file.
7757         * ubsan.c: New file.
7758         * common.opt: Add -fsanitize=undefined option.
7759         (flag_sanitize): Add variable.
7760         (fsanitize=): Add option.  Add Driver.
7761         (fsanitize=thread): Remove option.
7762         (fsanitize=address): Likewise.
7763         (static-libubsan): New option.
7764         * doc/invoke.texi: Document the new flag and -static-libubsan.
7765         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
7766         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
7767         * builtin-attrs.def (ATTR_COLD): Define.
7768         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7769         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
7770         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
7771         * flag-types.h (sanitize_code): New enum.
7772         * opts.c (common_handle_option): Parse command line arguments
7773         of -fsanitize=.  Add -fsanitize=unreachable option.
7774         * varasm.c (get_variable_section): Adjust.
7775         (assemble_noswitch_variable): Likewise.
7776         (assemble_variable): Likewise.
7777         (output_constant_def_contents): Likewise.
7778         (categorize_decl_for_section): Likewise.
7779         (place_block_symbol): Likewise.
7780         (output_object_block): Likewise.
7781         * builtins.def: Likewise.
7782         * toplev.c (compile_file): Likewise.
7783         (process_options): Likewise.
7784         * cppbuiltin.c: Likewise.
7785         * tsan.c (tsan_pass): Likewise.
7786         (tsan_gate): Likewise.
7787         (tsan_gate_O0): Likewise.
7788         * cfgexpand.c (partition_stack_vars): Likewise.
7789         (expand_stack_vars): Likewise.
7790         (defer_stack_allocation): Likewise.
7791         (expand_used_vars): Likewise.
7792         * cfgcleanup.c (old_insns_match_p): Likewise.
7793         * asan.c (asan_finish_file): Likewise.
7794         (asan_instrument): Likewise.
7795         (gate_asan): Likewise.
7796         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
7797         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
7798         (asan_global_struct): Use pointer_sized_int_node instead
7799         calling build_nonstandard_integer_type.
7800         (initialize_sanitizer_builtins): Likewise.
7801         (asan_finish_file): Likewise.
7802         * gcc.c: Document %{%:function(args):X}.
7803         (static_spec_functions): Add sanitize.
7804         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
7805         store funcval != NULL there.
7806         (do_spec_1): Adjust handle_spec_function caller.
7807         (handle_braces): Allow %:function(args) as condition.
7808         (sanitize_spec_function): New function.
7809         (ADD_STATIC_LIBUBSAN_LIBS): Define.
7810         (LIBUBSAN_SPEC): Likewise.
7811         (LIBUBSAN_EARLY_SPEC): Likewise.
7812         (SANITIZER_SPEC): Handle libubsan.
7813         (SANITIZER_EARLY_SPEC): Likewise.
7814         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
7815         instead of fsanitize=address.
7816         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
7817         instead of fsanitize=address*.
7818         * builtins.c: Include ubsan.h.
7819         (fold_builtin_0): Instrument __builtin_unreachable.
7820         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
7821         instead of flag_asan.
7822         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
7823         (pointer_sized_int_node): Define.
7824         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
7826 2013-08-30  Mike Stump  <mikestump@comcast.net>
7828         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
7829         with RE patterns.
7831 2013-08-29  Jan Hubicka  <jh@suse.cz>
7833         * cgraph.c (cgraph_function_body_availability): Handle weakref
7834         correctly.
7835         * passes.def: Remove pass_fixup_cfg.
7836         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
7837         track when we need to remove functions.
7838         (gate_ipa_inline): Execute inlining always; add comment why.
7839         (pass_data_ipa_inline): Remove TODO_remove_functions.
7840         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
7841         do not produce summaries.
7842         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
7843         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
7844         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
7845         constant pool and vtable.
7847 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
7849         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
7850         New arm_neon.h's internal macros to specify 64-bit constants.
7851         Avoid using stdint.h's macros.
7853 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
7855         * recog.c (verify_changes): Verify that changes[i].old is non-zero
7856         before applying REG_P.
7858 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
7860         PR tree-optimization/58277
7861         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
7862         after seeing too many stmts with vdef in between dombb and current
7863         bb, invalidate everything.
7865 2013-08-30  Richard Biener  <rguenther@suse.de>
7867         * fold-const.c (fold_single_bit_test): Fix overflow test.
7869 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
7871         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
7872         and which can live in a register, always retrieve the value on entry.
7873         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
7874         passed by invisible reference specially.
7875         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
7876         (vt_add_function_parameter): Correctly deal with a parameter passed by
7877         invisible reference.
7879 2013-08-30  Jan Hubicka  <jh@suse.cz>
7881         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
7883 2013-08-30  Richard Biener  <rguenther@suse.de>
7885         PR tree-optimization/58228
7886         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
7887         allow invariant loads in nested loop vectorization.
7889 2013-08-30  Richard Biener  <rguenther@suse.de>
7891         PR tree-optimization/58223
7892         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
7893         (has_anti_or_output_dependence): ... this and adjust to also
7894         look for output dependences.
7895         (mark_nodes_having_upstream_mem_writes): Adjust.
7896         (rdg_flag_uses): Likewise.
7898 2013-08-30  Richard Biener  <rguenther@suse.de>
7900         PR tree-optimization/58010
7901         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
7902         assert that we have a loop-closed PHI.
7904 2013-08-29  Jan Hubicka  <jh@suse.cz>
7906         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
7907         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
7908         * lto-section-in.c (lto_free_function_in_decl_state): New function.
7909         (lto_free_function_in_decl_state_for_node): New function.
7911 2013-08-29  Xinliang David Li  <davidxl@google.com>
7913         * loop-unroll.c (report_unroll_peel): Minor message change.
7914         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
7915         Emit alignment peeling message with default -fopt-info.
7916         (vect_loop_versioning): Emit loop version info message.
7917         * tree-vectorizer.c (vectorize_loops): Minor message change.
7918         (execute_vect_slp): Ditto.
7920 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
7922         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
7923         of the clone from the DECL_NAME of the original function.
7925 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
7927         * passes.c (register_pass): Add overload.
7928         * tree-pass.h (register_pass): Forward declare it.  Add comment.
7930 2013-08-29  Jan Hubicka  <jh@suse.cz>
7932         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
7933         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
7934         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
7935         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7936         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
7937         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
7938         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
7939         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
7941 2013-08-29  Teresa Johnson  <tejohnson@google.com>
7943         * dumpfile.c (dump_loc): Output column number.
7944         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
7945         * doc/invoke.texi: Document optall -fopt-info flag.
7946         * profile.c (read_profile_edge_counts): Use new dump framework.
7947         (compute_branch_probabilities): Ditto.
7948         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
7949         when pass not in any opt group.
7950         * pass_manager.h (pass_manager::get_pass_profile): New method.
7951         * value-prof.c (check_counter): Use new dump framework.
7952         (check_ic_target): Ditto.
7953         * coverage.c (get_coverage_counts): Ditto.
7954         (coverage_init): Setup new dump framework.
7956 2013-08-29  Richard Biener  <rguenther@suse.de>
7958         PR tree-optimization/58246
7959         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
7960         handle the dominance check inside a basic-block.
7962 2013-08-29  Richard Biener  <rguenther@suse.de>
7964         PR middle-end/57287
7965         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
7966         of default defs that appear in abnormal PHI nodes.
7968 2013-08-29  Richard Biener  <rguenther@suse.de>
7970         PR tree-optimization/57685
7971         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
7972         single-use operands to avoid exponential complexity.
7974 2013-08-28  Dehao Chen  <dehao@google.com>
7976         * ipa-inline.c (edge_badness): Fix integer underflow.
7978 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
7980         * gtm-builtins.def (_ITM_free): Declare leaf.
7982 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
7984         PR target/58067
7985         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
7986         (*tls_local_dynamic_base_64_largepic): Likewise.
7987         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
7988         Remove predicate from call operand.
7989         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
7990         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
7992 2013-08-28  Jeff Law  <law@redhat.com>
7994         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
7995         checks for the number of predecessors and successors allowed.
7996         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
7997         which require copying a joiner block if there is a request which
7998         is a subpath that requires no joiner block copying.
8000 2013-08-28  Jan Hubicka  <jh@suse.cz>
8002         * lto-streamer-out.c (DFS_write_tree_body): Drop
8003         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
8004         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
8005         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
8006         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
8007         Do not read DECL_ERROR_ISSUED.
8008         (unpack_ts_decl_with_vis_value_fields): Do not read
8009         DECL_DEFER_OUTPUT.
8010         (lto_input_ts_binfo_tree_pointers): Do not read
8011         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
8012         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
8013         write DECL_ERROR_ISSUED..
8014         (pack_ts_decl_with_vis_value_fields): Do not write
8015         DECL_DEFER_OUTPUT.
8016         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
8017         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
8018         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
8019         * tree.h (tree_decl_common): Update comment.
8020         (DECL_ERROR_ISSUED): Remove.
8022 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
8024         PR middle-end/58257
8025         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
8027 2013-08-28  Jan Hubicka  <jh@suse.cz>
8029         * builtins.def (free): Declare leaf.
8031 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
8033         * gdbhooks.py: New.
8034         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
8035         * configure: Regenerate.
8037 2013-08-27  Martin Jambor  <mjambor@suse.cz>
8039         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
8040         (ipa_ancestor_jf_data): Likewise.
8041         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
8042         (ipa_get_jf_pass_through_type_preserved): New function.
8043         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
8044         (ipa_get_jf_ancestor_type_preserved): New function.
8045         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
8046         (ipa_get_jf_ancestor_result): Likewise.
8047         (propagate_vals_accross_pass_through): Use
8048         ipa_get_jf_pass_through_result to do all the value mappings.
8049         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
8050         type_preserved flag.
8051         (ipa_set_jf_cst_copy): New function.
8052         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
8053         (ipa_set_jf_arith_pass_through): Likewise.
8054         (ipa_set_ancestor_jf): Likewise.
8055         (compute_complex_assign_jump_func): Set type_preserved instead of
8056         punting.
8057         (ipa_compute_jump_functions_for_edge): Likewise.
8058         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
8059         (update_jump_functions_after_inlining): Update type_preserved.
8060         Explicitely create jump functions when combining one with pass_through.
8061         (ipa_write_jump_function): Stream the type_preserved flags.
8062         (ipa_read_jump_function): Likewise.
8064 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
8065             Aldy Hernandez  <aldyh@redhat.com>
8067         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
8068         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
8069         * function.h (struct function): Add has_force_vect_loops and
8070         has_simduid_loops.
8071         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
8072         * gimple.c (gimple_build_omp_critical): Add KIND argument and
8073         handle it.
8074         * gimple.def: Update CLAUSES comments.
8075         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
8076         (gimple_build_omp_for): Add argument to prototype.
8077         (gimple_omp_for_kind): New.
8078         (gimple_omp_for_set_kind): New.
8079         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
8080         GOVD_DATA_SHARE_CLASS.
8081         (enum omp_region_type): Add ORT_SIMD.
8082         (gimple_add_tmp_var): Handle ORT_SIMD.
8083         (gimplify_var_or_parm_decl): Same.
8084         (is_gimple_stmt): Same.
8085         (omp_firstprivatize_variable): Same.
8086         (omp_add_variable): Only use splay_tree_insert if lookup failed.
8087         (omp_notice_variable): Handle ORT_SIMD.
8088         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
8089         (omp_check_private): Handle ORT_SIMD.
8090         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8091         OMP_CLAUSE_SAFELEN.
8092         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
8093         Handle OMP_CLAUSE_LASTPRIVATE.
8094         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
8095         OMP_CLAUSE_SAFELEN.
8096         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
8097         (gimplify_expr): Handle OMP_SIMD.
8098         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
8099         (expand_GOMP_SIMD_VF): New.
8100         (expand_GOMP_SIMD_LAST_LANE): New.
8101         * internal-fn.def (GOMP_SIMD_LANE): New.
8102         (GOMP_SIMD_VF): New.
8103         (GOMP_SIMD_LAST_LANE): New.
8104         * omp-low.c: Include target.h.
8105         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
8106         OMP_CLAUSE_SAFELEN.
8107         (check_omp_nesting_restrictions): Same.
8108         (omp_max_vf): New.
8109         (lower_rec_simd_input_clauses): New.
8110         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
8111         OMP_CLAUSE_LINEAR.
8112         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
8113         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
8114         (expand_omp_build_assign): New.
8115         (expand_omp_for_init_counts): New.
8116         (expand_omp_for_init_vars): New.
8117         (extract_omp_for_update_vars): New.
8118         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
8119         and rewrite accordingly.
8120         (expand_omp_simd): New.
8121         (expand_omp_for): Use expand_omp_simd.
8122         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
8123         (lower_omp_for): Do not lower the body.
8124         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
8125         in their own loops.
8126         * tree-flow.h (find_omp_clause): Remove prototype.
8127         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
8128         forcing vectorization of the loop, or if flag_tree_vectorize.
8129         (gate_tree_if_conversion): Similarly.
8130         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
8131         gimple_build_omp_for.
8132         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
8133         * tree-parloops (create_parallel_loop): Pass kind argument to
8134         gimple_build_omp_for.
8135         * tree-pretty-print.c (dump_omp_clause): Add cases for
8136         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
8137         OMP_CLAUSE__SIMDUID_.
8138         (dump_generic_node): Handle OMP_SIMD.
8139         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
8140         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
8141         unroll OMP_SIMD loops here.
8142         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
8143         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
8144         loop->safelen.
8145         (vect_analyze_data_refs): Handle simd loops.
8146         * tree-vect-loop.c (vectorizable_live_operation): Handle
8147         IFN_GOMP_SIMD*.
8148         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
8149         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
8150         (vectorizable_load): Same.
8151         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
8152         (struct simduid_to_vf): New.
8153         (simduid_to_vf::hash): New.
8154         (simduid_to-vf::equal): New.
8155         (struct simd_array_to_simduid): New.
8156         (simd_array_to_simduid::hash): New.
8157         (simd_array_to_simduid::equal): New.
8158         (adjust_simduid_builtins): New.
8159         (struct note_simd_array_uses_struct): New.
8160         (note_simd_array_uses_cb): New.
8161         (note_simd_array_uses): New.
8162         (vectorize_loops): Handle simd hints and adjust simd builtins
8163         accordingly.
8164         * tree-vectorizer.h (struct _stmt_vec_info): Add
8165         simd_lane_access_p field.
8166         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
8167         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
8168         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
8169         (omp_clause_code_name): Same.
8170         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
8171         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
8172         * tree.def (OMP_SIMD): New entry.
8173         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
8174         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
8175         (OMP_CLAUSE_DECL): Adjust range for new clauses.
8176         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
8177         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
8178         (OMP_CLAUSE_LINEAR_STEP): New.
8179         (OMP_CLAUSE_SAFELEN_EXPR): New.
8180         (OMP_CLAUSE__SIMDUID__DECL): New.
8181         (find_omp_clause): New prototype.
8183 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
8185         * config/i386/driver-i386.c (host_detect_local_cpu): Update
8186         Haswell processor detection.
8188 2013-08-27  Christian Widmer  <shadow@umbrox.de>
8190         PR target/57927
8191         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
8192         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
8193         AVX2 capable processors.
8195 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
8197         * config/aarch64/arm_neon.h: Replace all inline asm implementations
8198         of vget_low_* with implementations in terms of other intrinsics.
8200 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
8202         PR middle-end/57219
8203         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
8204         values to -1, 0 and 1.
8206 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
8208         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
8209         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
8210         (<optab><mode>3_insn): Remove.
8211         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
8212         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8213         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
8214         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
8215         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
8216         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
8217         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
8218         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
8219         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
8220         (ror<mode>3_insn): Likewise.
8221         * config/aarch64/predicates.md (aarch64_simd_register): New.
8223 2013-08-27  Richard Biener  <rguenther@suse.de>
8225         PR tree-optimization/57521
8226         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
8227         one edge is non-critical.
8228         (find_phi_replacement_condition): Make sure to use a non-critical
8229         edge.  Cleanup and remove old bug workarounds.
8230         (bb_postdominates_preds): Remove.
8231         (if_convertible_loop_p_1): Do not compute post-dominators.
8232         (combine_blocks): Do not free post-dominators.
8233         (main_tree_if_conversion): Likewise.
8234         (pass_data_if_conversion): Add TODO_verify_ssa.
8236 2013-08-27  DJ Delorie  <dj@redhat.com>
8238         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
8240 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
8242         * function.c (assign_parm_find_data_types): Set passed_mode and
8243         nominal_mode to the TYPE_MODE of nominal_type for the built
8244         pointer type in case of the struct-pass-by-reference.
8246 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8248         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
8249         (UINT16_TYPE): Change default to "unsigned int".
8251         * config/avr/avr.opt (mfract-convert-truncate): New option.
8252         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
8253         is set, round negative fractional integers according to n1169
8254         when converting to integer types.
8256 2013-08-26  Jan Hubicka  <jh@suse.cz>
8258         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
8259         methods can not be called indirectly when their address is not taken.
8261 2013-08-26  Jan Hubicka  <jh@suse.cz>
8263         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
8264         ctor_for_folding.
8266 2013-08-26  Jan Hubicka  <jh@suse.cz>
8268         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
8269         can be unshared.
8271 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8273         * reload.c (find_valid_class): Allow classes that do not include
8274         FIRST_PSEUDO_REGISTER - 1.
8276 2013-08-26  Jan Hubicka  <jh@suse.cz>
8278         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
8279         fix edge count/frequency when speculation failed; fix type check
8280         for the direct call.
8282 2013-08-26  Jan Hubicka  <jh@suse.cz>
8284         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
8286 2013-08-26  Jan Hubicka  <jh@suse.cz>
8288         * ipa-inline-transform.c (inline_transform): Be ready for basic block
8289         to be changed by edge redirection.
8291 2013-08-26  Jan Hubicka  <jh@suse.cz>
8293         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
8294         formating; add sanity check.
8295         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
8296         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
8297         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
8298         (resolve_noninline_speculation): Update callee keys, too.
8300 2013-08-26  Jan Hubicka  <jh@suse.cz>
8302         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
8303         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
8305 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
8307         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
8308         into proper place.
8310 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
8312         * config/i386/i386.c (ix86_debug_options): Remove prototype.
8313         (x86_64_elf_select_section): Ditto.
8314         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
8315         arguments.
8316         (ix86_pass_by_reference): Ditto.
8317         (output_set_got): Ditto.
8318         (ix86_unary_operator_ok): Ditto.
8319         (ix86_expand_builtin): Ditto.
8321 2013-08-23  Jan Hubicka  <jh@suse.cz>
8323         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
8325 2013-08-23  Jan Hubicka  <jh@suse.cz>
8327         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
8328         (tree_decl_with_vis): Add FINAL field.
8330 2013-08-23  Jeff Law  <law@redhat.com>
8332         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
8333         the debugging dump when the expression is fully redundant.
8335 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8337         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
8338         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
8339         * pretty-print.c (pp_formatted_text_data): Likewise.
8340         (pp_write_text_to_stream): Likewise.
8341         (pp_write_text_as_dot_label_to_stream): Likewise.
8342         (pp_append_r): Likewise.
8343         (pp_format): Likewise.
8344         (pp_flush): Likewise.
8345         (pp_clear_output_area): Likewise.
8346         (pp_append_text): Likewise.
8347         (pp_formatted_text): Likewise.
8348         (pp_remaining_character_count_for_line): Likewise.
8349         (pp_newline): Likewise.
8350         (pp_character): Likewise.
8351         (output_buffer::~output_buffer): Define.
8352         (pretty_printer::~pretty_printer): Destruct output buffer.
8353         * pretty-print.h (output_buffer::~output_buffer): Declare.
8354         (pretty_printer::~pretty_printer): Declare virtual.
8356 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
8358         PR other/57324
8359         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
8360         HOST_WIDE_INT_M1U): New macros.
8361         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
8362         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
8363         unsigned -1 for lshift.
8364         * cse.c (cse_insn): Likewise.
8365         * double-int.c (rshift_double, lshift_double): Likewise.
8366         * builtins.c (fold_builtin_bitop): Likewise.
8367         * combine.c (force_to_mode): Likewise.
8368         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
8369         * simplify-rtx.c (simplify_const_unary_operation,
8370         simplify_const_binary_operation): Likewise.
8371         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
8372         check_va_list_escapes): Likewise.
8373         * rtlanal.c (nonzero_bits1): Likewise.
8374         * expmed.c (expand_smod_pow2): Likewise.
8375         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
8377 2013-08-23  Jan Hubicka  <jh@suse.cz>
8379         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
8380         as having address taken.
8382 2013-08-23  Jan Hubicka  <jh@suse.cz>
8384         * ipa-utils.h (method_class_type): Declare.
8385         * ipa-devirt.c (method_class_type): Export.
8387         * cgraphunit.c (analyze_functions): Do basic devirtualization;
8388         do not walk base classes of anonymous types.
8390 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
8392         PR rtl-optimization/58220
8393         PR regression/58221
8394         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
8395         handle SEQUENCE insns properly.
8397 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8399         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
8400         definition.
8401         (pp_newline_and_indent): Likewise.
8402         (pp_separate_with): Likewise.
8403         * pretty-print.c (pp_newline_and_flush): Define.
8404         (pp_newline_and_indent): Likewise.
8405         (pp_separate_with): Likewise.
8407 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8409         PR target/58218
8410         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
8411         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
8413 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
8415         * config/i386/predicates.md (ext_sse_reg_operand): New.
8416         * config/i386/i386.md (*movti_internal): Use
8417         predicate to determine if EVEX is needed.
8418         (*movsi_internal): Ditto.
8419         (*movdf_internal): Ditto.
8420         (*movsf_internal): Ditto.
8421         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8423 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
8425         PR tree-optimization/58209
8426         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
8427         (find_tail_calls): Give up for pointer result types if m is non-NULL.
8428         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
8429         emit POINTER_PLUS_EXPR.
8430         (create_tailcall_accumulator): For pointer result type accumulate in
8431         sizetype type.
8433 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
8435         * configure.ac: Add backslashes missing from the last change.
8436         * configure: Regenerate.
8438 2013-08-22  Jan Hubicka  <jh@susue.cz>
8440         * ipa.c (function_and_variable_visibility): First remember function
8441         was global and then make it local.
8443 2013-08-22  Julian Brown  <julian@codesourcery.com>
8445         * configure.ac: Add aarch64 to list of arches which use "nop" in
8446         debug_line test.
8447         * configure: Regenerate.
8449 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
8451         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
8452         gnu_libc_has_function.
8453         * config/s390/tpf.h: Likewise.
8455 2013-08-22  Jan Hubicka  <jh@susue.cz>
8457         * timevar.c (validate_phases): Add cast.
8459 2013-08-22  Jan Hubicka  <jh@susue.cz>
8461         * timevar.c (validate_phases): Use size_t for memory.
8462         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
8464 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
8466         * pretty-print.h (output_buffer::output_buffer): Declare.
8467         (pretty_printer::pretty_printer): Likewise.
8468         (pp_construct): Remove.
8469         * pretty-print.c (output_buffer::output_buffer): Define.
8470         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
8471         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
8472         (print_gimple_expr): Likewise.
8473         (print_gimple_seq): Likewise.
8474         (gimple_dump_bb): Likewise.
8475         * sched-vis.c (dump_value_slim): Likewise.
8476         (dump_insn_slim): Likewise.
8477         (dump_rtl_slim): Likewise.
8478         (str_pattern_slim): Likewise.
8479         * tree-mudflap.c (mf_varname_tree): Likewise.
8480         * graph.c (print_graph_cfg): Likewise.
8481         (start_graph_dump): Likewise.
8482         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
8483         placement-new.
8484         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
8485         pretty printer initialization.
8486         * coretypes.h (diagnostic_context): Remove superflous type alias
8487         declaration.
8488         (pretty_printer): Likewise.  Declare directly as a class.
8489         (pretty_print_info): Remove declaration as class.
8490         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
8491         and pp_clear_output_area.
8492         (asan_add_global): Likewise.
8494 2013-08-22  Jan Hubicka  <jh@suse.cz>
8496         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
8497         * ipa-utils.h (update_type_inheritance_graph): Declare.
8498         (possible_polymorphic_call_target_p): Declare.
8499         (possible_polymorphic_call_target_p): New.
8500         * ipa-devirt.c: Update toplevel comments.
8501         (cached_polymorphic_call_targets): Move up.
8502         (odr_type_d): Move ID down.
8503         (polymorphic_type_binfo_p): Update comment.
8504         (odr_hasher::remove): Likewise;
8505         (get_odr_type): Set anonymous_namespace.
8506         (dump_odr_type): Dump it.
8507         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
8508         (maybe_record_node): Record node in cached_polymorphic_call_targets.
8509         (record_binfo): Add comment.
8510         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
8511         built.
8512         (devirt_node_removal_hook): Do not iCE when cache is freed.
8513         (possible_polymorphic_call_target_p): New predicate.
8514         (update_type_inheritance_graph): New function.
8516 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8517             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8518             Sergey Lega  <sergey.s.lega@intel.com>
8519             Anna Tikhonova  <anna.tikhonova@intel.com>
8520             Ilya Tocar  <ilya.tocar@intel.com>
8521             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8522             Ilya Verbin  <ilya.verbin@intel.com>
8523             Kirill Yukhin  <kirill.yukhin@intel.com>
8524             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8526         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
8527         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
8528         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
8529         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
8530         (OPTION_MASK_ISA_AVX2_UNSET): Update.
8531         (OPTION_MASK_ISA_AVX512F_UNSET): New.
8532         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
8533         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
8534         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
8535         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
8536         OPT_mavx512pf, OPT_mavx512er cases.
8537         * config/i386/constraints.md (v): New constraint.
8538         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
8539         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
8540         (bit_AVX512CD): New.
8541         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8542         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
8543         * config/i386/i386-c.c (ix86_target_macros_internal):
8544         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
8545         __AVX512PF__.
8546         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
8547         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
8548         * config/i386/i386.c (regclass_map, dbx_register_map)
8549         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
8550         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
8551         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
8552         -mavx512pf options.
8553         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
8554         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
8555         -mavx512pf options.  Fix formatting.
8556         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
8557         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
8558         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
8559         -mavx512cd, -mavx512pf options.
8560         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
8561         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
8562         (ix86_preferred_output_reload_class): Replace SSE_REGS with
8563         ALL_SSE_REGS.
8564         (ix86_hard_regno_mode_ok): Support 512-bit registers.
8565         (ix86_set_reg_reg_cost): Ditto.
8566         (x86_order_regs_for_local_alloc): Ditto.
8567         (MAX_VECT_LEN): Extend to 64-byte.
8568         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
8569         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
8570         (TARGET_AVX512ER, TARGET_AVX512CD): New.
8571         (BIGGEST_ALIGNMENT): Extend to 512-bits.
8572         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
8573         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
8574         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
8575         (SSE_REG_MODE_P): Support new modes.
8576         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
8577         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
8578         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
8579         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
8580         (REG_CLASS_CONTENTS): Add new registers.
8581         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
8582         (EXT_REX_SSE_REGNO_P): New.
8583         (HI_REGISTER_NAMES): Add new registers.
8584         * config/i386/i386.md: Define constants for new registers.
8585         (mode): Add new 512-bit modes.
8586         (prefix): Support evex prefix.
8587         (isa): Support avx512f, noavx512f, fma_avx512f.
8588         (ssemodesuffix): Add new 512-bit modes.
8589         (movxi): New.
8590         (*movxi_internal_avx512f): Ditto.
8591         (*movdi_internal): Replace constraint "x" with the new constraint "v".
8592         Support MODE_XI.
8593         (*movsi_internal): Likewise.
8594         (*movdf_internal): Likewise.
8595         (*movsf_internal): Likewise.
8596         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
8597         (<code><mode>3): Likewise.
8598         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
8599         New.
8600         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
8601         with the new constraint "v".
8602         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
8603         modes.
8604         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
8605         with the new constraint "v".
8606         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
8607         (<sse2>_storedqu<avxsizesuffix>): Likewise.
8608         (*<plusminus_insn><mode>3): Likewise.
8609         (<sse>_vm<plusminus_insn><mode>3): Likewise.
8610         (*mul<mode>3): Likewise.
8611         (<sse>_vmmul<mode>3): Likewise.
8612         (<sse>_div<mode>3): Likewise.
8613         (<sse>_vmdiv<mode>3): Likewise.
8614         (<sse>_sqrt<mode>2): Likewise.
8615         (<sse>_vmsqrt<mode>2): Likewise.
8616         (*<code><mode>3_finite): Likewise.
8617         (*<code><mode>3) <smaxmin>: Likewise.
8618         (<sse>_vm<code><mode>3): Likewise.
8619         (*<code><mode>3) <any_logic>: Likewise.
8620         (*fma_fmadd_<mode>): Likewise.
8621         (*fma_fmsub_<mode>): Likewise.
8622         (*fma_fnmadd_<mode>): Likewise.
8623         (*fma_fnmsub_<mode>): Likewise.
8624         (*fma_fmaddsub_<mode>): Likewise.
8625         (*fma_fmsubadd_<mode>): Likewise.
8626         (*fmai_fmadd_<mode>): Likewise.
8627         (*fmai_fmsub_<mode>): Likewise.
8628         (*fmai_fnmadd_<mode>): Likewise.
8629         (*fmai_fnmsub_<mode>): Likewise.
8630         (sse_cvtsi2ss): Likewise.
8631         (sse_cvtsi2ssq): Likewise.
8632         (sse_cvtss2si): Likewise.
8633         (sse_cvtss2si_2): Likewise.
8634         (sse_cvtss2siq): Likewise.
8635         (sse_cvtss2siq_2): Likewise.
8636         (sse_cvttss2si): Likewise.
8637         (sse_cvtss2siq_2): Likewise.
8638         (float<sseintvecmodelower><mode>2): Likewise.
8639         (sse2_cvtsd2si_2): Likewise.
8640         (sse2_cvtsd2siq_2): Likewise.
8641         (*<plusminus_insn><mode>3): Likewise.
8642         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
8643         (*<sse4_1_avx2>_mul<mode>3): Likewise.
8644         (ashr<mode>3): Likewise.
8645         (<shift_insn><mode>3): Likewise.
8646         (avx2_<code><mode>3): Likewise.
8647         (*avx2_<code><mode>3): Likewise.
8648         (*andnot<mode>3): Likewise.
8649         (*<code><mode>3) <any_logic>: Likewise.
8650         (abs<mode>2): Likewise.
8651         (avx2_permvar<mode>): Likewise.
8652         (avx2_perm<mode>_1): Likewise.
8653         (*avx_vpermilp<mode>): Likewise.
8654         (avx_vpermilvar<mode>3): Likewise.
8655         (avx2_ashrv<mode>): Likewise.
8656         (avx2_<shift_insn>v<mode>): Likewise.
8657         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
8658         -mavx512cd.
8659         * doc/rtl.texi: Document XImode.
8661 2013-08-21  Jeff Law  <law@redhat.com>
8663         * tree-flow.h (register_jump_thread): Pass vector of edges
8664         instead of each important edge.
8665         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
8666         thread path into a vector and pass that to register_jump_thread.
8667         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
8668         passed in edge vector to the current 3-edge form.
8670         Revert:
8671         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8673         * dce.c (fini_dce): Call df_analyze again just in case
8674         delete_unmarked_insns removed anything.
8676 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
8678         * reload.h (struct reg_equivs): Rename to ..
8679         (struct reg_equivs_s): .. this.
8681 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
8683         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
8685 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
8687         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
8689 2013-08-21  Jeff Law  <law@redhat.com>
8691         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
8692         simplify assignments too.  If the RHS collapses to a singleton
8693         range, then return the value for the range.
8695 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
8697         * config/i386/sse.md (V16): Rename to...
8698         (VMOVE): this.
8699         (mov<mode>): Update iterator name.
8700         (*mov<mode>_internal): Ditto.
8701         (push<mode>1): Ditto.
8702         (movmisalign<mode>): Ditto.
8704 2013-08-20  Jan Hubicka  <jh@suse.cz>
8706         PR bootstrap/58186
8707         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
8708         entry for direct edges.
8709         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
8711 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8713         Revert my last two changes, r201865 and r201864:
8715         Revert r201865:
8716         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8718         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8719         instances can own GC refs.
8721         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8722         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8723         (gcc::context::gt_pch_nx): Likewise.
8724         (gcc::context::gt_pch_nx):  Likewise.
8725         * ggc.h (gt_ggc_mx <T>): New.
8726         (gt_pch_nx_with_op <T>): New.
8727         (gt_pch_nx <T>): New.
8728         * passes.c (opt_pass::gt_ggc_mx): New.
8729         (opt_pass::gt_pch_nx): New.
8730         (opt_pass::gt_pch_nx_with_op): New.
8731         (pass_manager::gt_ggc_mx): New.
8732         (pass_manager::gt_pch_nx): New.
8733         (pass_manager::gt_pch_nx_with_op): New.
8734         (pass_manager::operator new): Use
8735         ggc_internal_cleared_alloc_stat rather than xcalloc.
8736         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8737         (pass_manager::gt_ggc_mx): New.
8738         (pass_manager::gt_pch_nx): New.
8739         (pass_manager::gt_pch_nx_with_op): New.
8740         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8741         (opt_pass::operator new): New.
8742         (opt_pass::gt_ggc_mx): New.
8743         (opt_pass::gt_pch_nx): New.
8744         (opt_pass::gt_pch_nx_with_op): New.
8746         Revert r201864:
8747         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8749         * Makefile.in (GTFILES): Add context.h.
8750         * context.c (gcc::context::operator new): New.
8751         (gcc::context::gt_ggc_mx): New.
8752         (gcc::context::gt_pch_nx): New.
8753         (gcc::context::gt_pch_nx): New.
8754         * context.h (gcc::context): Add GTY((user)) marking.
8755         (gcc::context::operator new): New.
8756         (gcc::context::gt_ggc_mx): New.
8757         (gcc::context::gt_pch_nx): New.
8758         (gcc::context::gt_pch_nx): New.
8759         (g): Add GTY marking.
8760         (gt_ggc_mx (gcc::context *)): New.
8761         (gt_pch_nx (gcc::context *)): New.
8762         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8763         void *cookie)): New.
8764         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8766 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
8768         * dce.c (fini_dce): Call df_analyze again just in case
8769         delete_unmarked_insns removed anything.
8771 2013-08-20  Teresa Johnson  <tejohnson@google.com>
8773         PR rtl-optimizations/57451
8774         * final.c (reemit_insn_block_notes): Prevent lexical blocks
8775         from crossing split section boundaries.
8777 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
8779         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
8780         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
8781         with MULTLIB_DEFAULTS.
8783 2013-08-20  Nick Clifton  <nickc@redhat.com>
8785         * target.def (narrow_volatile_bitfield): Note that the default
8786         value is false, not !TARGET_STRICT_ALIGN.
8787         * doc/tm.texi: Regenerate.
8789 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
8791         Fix LIB_SPEC for systems without libpthread.
8793         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
8794         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
8795         for Android.
8796         * config/i386/linux-common.h: Likewise.
8797         * config/mips/linux-common.h: Likewise.
8799 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
8801         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
8802         checks.
8804 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8806         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
8807         instances can own GC refs.
8809         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
8810         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
8811         (gcc::context::gt_pch_nx): Likewise.
8812         (gcc::context::gt_pch_nx):  Likewise.
8813         * ggc.h (gt_ggc_mx <T>): New.
8814         (gt_pch_nx_with_op <T>): New.
8815         (gt_pch_nx <T>): New.
8816         * passes.c (opt_pass::gt_ggc_mx): New.
8817         (opt_pass::gt_pch_nx): New.
8818         (opt_pass::gt_pch_nx_with_op): New.
8819         (pass_manager::gt_ggc_mx): New.
8820         (pass_manager::gt_pch_nx): New.
8821         (pass_manager::gt_pch_nx_with_op): New.
8822         (pass_manager::operator new): Use
8823         ggc_internal_cleared_alloc_stat rather than xcalloc.
8824         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
8825         (pass_manager::gt_ggc_mx): New.
8826         (pass_manager::gt_pch_nx): New.
8827         (pass_manager::gt_pch_nx_with_op): New.
8828         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
8829         (opt_pass::operator new): New.
8830         (opt_pass::gt_ggc_mx): New.
8831         (opt_pass::gt_pch_nx): New.
8832         (opt_pass::gt_pch_nx_with_op): New.
8834 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
8836         * Makefile.in (GTFILES): Add context.h.
8837         * context.c (gcc::context::operator new): New.
8838         (gcc::context::gt_ggc_mx): New.
8839         (gcc::context::gt_pch_nx): New.
8840         (gcc::context::gt_pch_nx): New.
8841         * context.h (gcc::context): Add GTY((user)) marking.
8842         (gcc::context::operator new): New.
8843         (gcc::context::gt_ggc_mx): New.
8844         (gcc::context::gt_pch_nx): New.
8845         (gcc::context::gt_pch_nx): New.
8846         (g): Add GTY marking.
8847         (gt_ggc_mx (gcc::context *)): New.
8848         (gt_pch_nx (gcc::context *)): New.
8849         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
8850         void *cookie)): New.
8851         * gengtype.c (open_base_files) <ifiles>: Add context.h.
8853 2013-08-20  Alan Modra  <amodra@gmail.com>
8855         PR target/57865
8856         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
8857         (rs6000_emit_epilogue): Likewise.
8859 2013-08-19  Dehao Chen  <dehao@google.com>
8861         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
8863 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
8864             Jakub Jelinek  <jakub@redhat.com>
8866         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
8867         (BUILT_IN_FABSD64): Likewise.
8868         (BUILT_IN_FABSD128): Likewise.
8869         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
8870         (fold_builtin_1): Likewise.
8871         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
8872         destination and source operands.
8873         (*abstd2_fpr): Likewise.
8874         (*nabstd2_fpr): Likewise.
8876 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
8878         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
8879         JUMP_P and INSN_P.
8881 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
8883         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
8884         incomplete.
8886 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8888         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
8889         * builtins.c (default_libc_has_function): New.
8890         (gnu_libc_has_function): Ditto.
8891         (no_c99_libc_has_function): Ditto.
8892         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
8893         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
8894         (fold_builtin_sincos): Likewise.
8895         (fold_builtin_cexp): Likewise.
8896         * builtins.def (DEF_C94_BUILTIN): Likewise.
8897         (DEF_C99_BUILTIN): Likewise.
8898         (DEF_C99_C90RES_BUILTIN): Likewise.
8899         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
8900         definitions to using this define.
8901         * config/darwin-protos.h (darwin_libc_has_function): New.
8902         * config/darwin.c (darwin_libc_has_function): Ditto.
8903         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
8904         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
8905         * config/darwin.h: Ditto.
8906         * config/elfos.h: Ditto.
8907         * config/freebsd.h: Ditto.
8908         * config/i386/cygming.h: Ditto.
8909         * config/i386/djgpp.h: Ditto.
8910         * config/i386/i386-interix.h: Ditto.
8911         * config/microblaze/microblaze.h: Ditto.
8912         * config/mmix/mmix.h: Ditto.
8913         * config/gnu-user.h: Ditto.
8914         * config/ia64/hpux.h: Ditto.
8915         * config/pa/pa-hpux.h: Ditto.
8916         * config/pdp11/pdp11.h: Ditto.
8917         * config/picochip/picochip.h: Ditto.
8918         * config/linux.h: Ditto.
8919         * config/netbsd.h: Ditto.
8920         * config/openbsd.h: Ditto.
8921         * config/rs6000/aix43.h: Ditto.
8922         * config/rs6000/aix51.h: Ditto.
8923         * config/rs6000/aix52.h: Ditto.
8924         * config/rs6000/aix53.h: Ditto.
8925         * config/rs6000/aix61.h: Ditto.
8926         * config/rs6000/darwin.h: Ditto.
8927         * config/rs6000/linux.h: Ditto.
8928         * config/rs6000/linux64.h: Ditto.
8929         * config/s390/tpf.h: Ditto.
8930         * config/sol2-10.h: Ditto.
8931         * config/sol2.h: Ditto.
8932         * config/vms/vms.h: Ditto.
8933         * config/vxworks.h: Ditto.
8934         * config/linux-android.c (linux_android_libc_has_function):
8935         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
8936         * config/linux-protos.h (linux_android_libc_has_function):
8937         New declaration.
8938         * config/i386/i386.c (ix86_libc_has_function): New.
8939         * config/i386/i386-protos.h
8940         (ix86_libc_has_function): New declaration.
8941         * config/i386/i386.md
8942         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
8943         ("isinf<mode>2): Likewise.
8944         * convert.c (convert_to_integer): Using new target hook
8945         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
8946         TARGET_C99_FUNCTIONS.
8947         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
8948         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
8949         * coretypes.h (function_class): New enum for different
8950         classes of functions.
8951         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
8952         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
8953         (TARGET_HAS_SINCOS): Likewise.
8954         (TARGET_LIBC_HAS_FUNCTION): New.
8955         * doc/tm.texi: Regenerated.
8956         * targhooks.h (default_libc_has_function): New declaration.
8957         (no_c99_libc_has_function): Ditto.
8958         (gnu_libc_has_function): Ditto.
8959         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
8960         and TARGET_HAS_SINCOS.
8962 2013-08-18  Jan Hubicka  <jh@suse.cz>
8964         * Makeifle-in (ipa-devirt.o): New.
8965         (GTFILES): Add ipa-utils.h and ipa-devirt.c
8966         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
8967         (analyze_functions): Look into possible targets of polymorphic call.
8968         * dumpfile.c (dump_files): Add type-inheritance dump.
8969         * dumpfile.h (TDI_inheritance): New.
8970         * ipa-devirt.c: New file.
8971         * ipa-utils.h (odr_type_d): Forward declare.
8972         (odr_type): New type.
8973         (build_type_inheritance_graph): Declare.
8974         (possible_polymorphic_call_targets): Declare and introduce inline
8975         variant when only edge is pased.
8976         (dump_possible_polymorphic_call_targets): Likewise.
8977         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
8978         * tree.c (type_in_anonymous_namespace_p): Break out from ...
8979         (types_same_for_odr): ... here.
8980         * tree.h (type_in_anonymous_namespace_p): Declare.
8982 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
8984         PR tree-optimization/58006
8985         * tree-parloops.c (take_address_of): Don't ICE if get_name
8986         returns NULL.
8987         (eliminate_local_variables_stmt): Remove clobber stmts.
8989 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
8991         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
8992         error message is issued for an alias to undefined symbol.
8994 2013-08-18  Jan Hubicka  <jh@suse.cz>
8996         * cgraph.c (cgraph_create_indirect_edge): Discover
8997         polymorphic calls and record basic info into indirect_info.
8998         * gimple-fold.c (gimple_fold_call): When doing BINFO based
8999         devirtualization, ignore objc function calls.
9000         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
9001         call with no parm index info.
9002         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
9003         * tree.c (virtual_method_call_p): New function.
9004         * tree.h (virtual_method_call_p): Declare.
9006 2013-08-16  Jan Hubicka  <jh@suse.cz>
9008         PR middle-end/58179
9009         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
9011 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
9013         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
9014         attribute.
9016 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9018         * gengtype.c (type_for_name): Add special-case support for
9019         locating types within the "gcc::" namespace.
9020         (open_base_files): Emit a "using namespace gcc" directive.
9022 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
9024         PR target/58160
9025         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
9026         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
9028         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
9029         array instead of each individual operand as a separate argument.
9030         (emit_fusion_gpr_load): Likewise.
9031         (expand_fusion_gpr_load): Add new function declaration.
9033         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
9034         signature to have the operands passed as an array, instead of as
9035         separate arguments.  Allow ZERO_EXTEND to be in the memory
9036         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
9037         depend on the register live/dead flags when peepholes are run.
9038         (expand_fusion_gpr_load): New function to be called from the
9039         peephole2 pass, to change the register that addis sets to be the
9040         target register.
9041         (emit_fusion_gpr_load): Change the calling signature to have the
9042         operands passed as an array, instead of as separate arguments.
9043         Allow ZERO_EXTEND to be in the memory address, and also
9044         SIGN_EXTEND if -mpower8-fusion-sign.
9046         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
9047         unspec enumeration.
9048         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
9049         adjust the register addis loads up in the peephole2 pass.  Do not
9050         depend on the register live/dead state when the peephole pass is done.
9052 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
9054         * gengtype.c (create_user_defined_type): Ensure that the kind
9055         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
9056         declaration is seen before the GTY((user)) marking.
9058 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9060         PR target/58105
9061         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
9063 2013-08-16  Jan Hubicka  <jh@suse.cz>
9065         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
9066         arugment expected_type.
9067         (gimple_fold_call): Use it.
9068         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
9069         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
9070         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
9071         (try_make_edge_direct_virtual_call): Likewise.
9072         * tree.c (obj_type_ref_class): New.
9073         * tree.h (obj_type_ref_class): Use it.
9075 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
9077         * sched-vis.c (rtl_slim_pp_initialized): Remove.
9078         (rtl_slim_pp): Likewise.
9079         (init_rtl_slim_pretty_print): Likewise.
9080         (dump_value_slim):  Don't call it.  Use local pretty printer.
9081         (dump_insn_slim): Likewise.
9082         (dump_rtl_slim): Likewise.
9083         (str_pattern_slim): Likewise.
9084         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
9085         Simplify.
9087 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
9089         PR tree-optimization/58164
9090         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
9091         walk gimple_goto_dest of GIMPLE_GOTO.
9093         PR tree-optimization/58165
9094         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
9095         bi_call must be the last stmt in a bb, don't split_block, instead
9096         use fallthru edge from it and give up if there is none.
9097         Release conds vector when returning early.
9099 2013-08-14  Xinliang David Li  <davidxl@google.com>
9101         * config/i386/i386.c (ix86_option_override_internal):
9102         Remove unused variable and field.
9104 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
9106         PR target/57949
9107         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
9108         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
9109         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
9110         and Linux, correct BLKmode alignment when 128-bit alignment is
9111         required and compatibility flag is not set.
9112         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
9113         for zero-size arguments when compatibility flag is not set.
9115 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9117         PR tree-optimization/58145
9118         * tree-sra.c (build_ref_for_offset): If prev_base has
9119         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
9121 2013-08-14  Xinliang David Li  <davidxl@google.com>
9123         * config/i386/i386.c (ix86_option_override_internal):
9124         Fix uninitialized variable error.
9126 2013-08-14  Xinliang David Li  <davidxl@google.com>
9128         * config/i386/i386.opt: Define two new options.
9129         * config/i386/x86-tune.def: Add arch selector field in macros.
9130         * config/i386/i386.h: Adjust macro definition.
9131         * config/i386/i386.c (ix86_option_override_internal):
9132         Refactor the code.
9133         (parse_mtune_ctrl_str): New function.
9134         (set_ix86_tune_features): New function.
9135         (ix86_function_specific_restore): Call the new helper function.
9137 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
9139         PR rtl-optimization/57662
9140         * sel-sched.c (code_motion_process_successors): When the current insn
9141         is removed after the recursive traversal, break from the loop.
9142         Add comments and debug printouts.
9144 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
9145             Alexandre Oliva  <aoliva@redhat.com>
9147         PR target/58067
9148         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
9149         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
9150         there also UNSPEC_PLTOFF.
9152 2013-08-14  Marek Polacek  <polacek@redhat.com>
9154         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
9155         NUM_CONDITIONS bit positions.
9157 2013-08-13  Cary Coutant  <ccoutant@google.com>
9159         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
9160         (attr_checksum): Hash vector contents instead of pointer.
9161         (attr_checksum_ordered): Likewise.
9163 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
9165         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
9166         when Pmode != word_mode.  Add length_address attribute.
9167         (sse3_monitor_<mode>): Merge from sse3_monitor and
9168         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
9169         Pmode != word_mode.  Update insn length attribute.
9170         * config/i386/i386.c (ix86_option_override_internal): Update
9171         ix86_gen_monitor selection for merged sse3_monitor insn.
9173 2013-08-13  Julian Brown  <julian@codesourcery.com>
9175         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
9176         perform invalid legitimization on greater-than-word-size modes for
9177         TARGET_E500_DOUBLE.
9179 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
9181         * ira.c (setup_class_translate_array): Use aclass instead of cl
9182         for classes not fully covered by allocno classes.
9184 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
9186         PR tree-optimization/57661
9187         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
9188         * tree-inline.c (tree_function_versioning): Initialize it.
9189         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
9190         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
9191         that is not being copied.
9193         PR sanitizer/56417
9194         * asan.c (instrument_strlen_call): Fix typo in comment.
9195         Use char * type even for the lhs of POINTER_PLUS_EXPR.
9197 2013-08-13  Steve Ellcey  <sellcey@mips.com>
9199         * config/mips/mips.md (prefetch): Use lw instead of ld on
9200         loongson in 32bit mode.
9202 2013-08-13  Nick Clifton  <nickc@redhat.com>
9204         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
9206 2013-08-13  Jan Hubicka  <jh@suse.cz>
9208         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
9209         introduced edge; fix typo in sanity check.
9210         (cgraph_resolve_speculation): Export; improve diagnostic.
9211         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
9212         speculation at type mismatch.
9213         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
9214         (cgraph_resolve_speculation): Declare.
9215         (symtab_can_be_discarded): New function.
9216         * value-prof.c (gimple_ic_transform): Remove actual transform code.
9217         * ipa-inline-transform.c (speculation_removed): New global var.
9218         (clone_inlined_nodes): See if speculation can be removed.
9219         (inline_call): If speculations was removed, we growths may not match.
9220         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
9221         (speculation_useful_p): New function.
9222         (resolve_noninline_speculation): New function.
9223         (inline_small_functions): Resolve useless speculations.
9224         * ipa-inline.h (speculation_useful_p): Declare
9225         * ipa.c (can_replace_by_local_alias): Simplify.
9226         (ipa_profile): Produce speculative calls in non-lto, too;
9227         add simple cost model; produce local aliases.
9229 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
9231         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
9232         PASS_MANAGER_H.
9234 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
9236         * config/i386/i386.c (ix86_function_versions): Use error + inform.
9238 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
9240         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
9241         iterator instead of X87MODEF.
9243 2013-08-12  Perez Read  <netfirewall@gmail.com>
9245         PR target/58132
9246         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
9247         operand 0 for intel asm alternative.
9248         (*movabs<mode>_2): Ditto for operand 1.
9250 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
9252         * config/aarch64/arm_none.h
9253         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
9255 2013-08-12  Nick Clifton  <nickc@redhat.com>
9257         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
9259 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
9261         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
9262         expand for QI/HImode operand to produce more effictive code for
9263         unsigned char(short) --> float(double) conversion.
9265 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
9267         * doc/invoke.texi: Mention that -ftls-model does not force the final
9268         model.
9270 2013-08-12  Marek Polacek  <polacek@redhat.com>
9271             Marc Glisse  <marc.glisse@inria.fr>
9273         PR tree-optimization/57980
9274         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
9275         when creating -1 constant.
9277 2013-08-10  Jan Hubicka  <jh@suse.cz>
9279         Workaround binutils PR14342.
9280         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
9281         (gimple_init_edge_profiler): Likewise.
9282         (gimple_gen_ic_func_profiler): Likewise.
9284 2013-08-09  Jan Hubicka  <jh@suse.cz>
9286         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
9288 2013-08-09  Xinliang David Li  <davidxl@google.com>
9290         * config/i386/stringop.def: New file.
9291         * config/i386/stringop.opt: New file.
9292         * config/i386/i386-opts.h: Include stringopt.def.
9293         * config/i386/i386.opt: Include stringopt.opt.
9294         * config/i386/i386.c (ix86_option_override_internal):
9295         Override default size based stringop inline strategies with options.
9296         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
9297         New function.
9299 2013-08-09  Jan Hubicka  <jh@suse.cz>
9301         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
9303 2013-08-09  Jan Hubicka  <jh@suse.cz>
9305         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
9306         CGRAPH_FREQ_MAX.
9307         (dump_cgraph_node): Dump profile-id.
9308         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
9309         and common_target_probability.
9310         * lto-cgraph.c (lto_output_edge): Stream common targets.
9311         (lto_output_node): Stream profile ids.
9312         (input_node): Stream profile ids.
9313         (input_edge): Stream common targets.
9314         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
9315         * ipa.c: Include value-prof.h
9316         (ipa_profile_generate_summary): Turn indirect call statement histograms
9317         into common targets.
9318         (ipa_profile): Turn common targets into speculative edges.
9320 2013-08-09  Jan Hubicka  <jh@suse.cz>
9322         * cgraph.h (cgraph_node): Add profile_id.
9323         * value-prof.c (cgraph_node_map): Turn into pointer_map.
9324         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
9325         (del_node_map): Update.
9326         (find_func_by_funcdef_no): Replace by ...
9327         (find_func_by_profile_id): ... this one.
9328         (gimple_ic_transform): Do not remove useful histograms when
9329         speculation is not done; dump info when indirect call removal
9330         can happen at LTO.
9331         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
9332         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
9333         (__gcov_indirect_call_profiler_v2): .. this one.
9334         * profile.h (init_node_map): Update.
9335         * coverage.c (coverage_compute_profile_id): New function.
9336         * coverage.h (coverage_compute_profile_id): Declare.
9337         * tree-profile.c (init_ic_make_global_vars): Make
9338         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
9339         (gimple_init_edge_profiler): Update prototype of
9340         __gcov_indirect_call_profiler.
9341         (gimple_gen_ic_func_profiler): Simplify.
9342         (tree_profiling): Use init_node_map
9344 2013-08-09  Jan Hubicka  <jh@suse.cz>
9346         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
9347         non-speculative refs.
9348         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
9349         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
9350         (cgraph_set_call_stmt): Likewise.
9351         (cgraph_create_edge_1): Fix release checking compilatoin;
9352         clear lto_stmt_uid.
9353         (cgraph_free_edge): Free indirect info.
9354         (cgraph_turn_edge_to_speculative): New function.
9355         (cgraph_speculative_call_info): New function.
9356         (cgraph_make_edge_direct): Return direct edge; handle speculation.
9357         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
9358         (dump_cgraph_node): Dump speculation.
9359         (verify_edge_count_and_frequency): Accept speculative edges.
9360         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
9361         (verify_cgraph_node): Handle speculation.
9362         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
9363         (cgraph_set_call_stmt): Update prototype.
9364         (cgraph_make_edge_direct): Update prototype.
9365         (cgraph_speculative_call_info): Declare.
9366         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
9367         to change; update call of ipa_find_references.
9368         * ipa-ref.c (ipa_record_reference): Fix return value; clear
9369         lto_stmt_uid and speculative flags.
9370         (ipa_dump_references): Dump speculation.
9371         (ipa_clone_references): Clone speculative flag.
9372         (ipa_clone_referring): Likewise.
9373         (ipa_clone_ref): New function.
9374         (ipa_find_reference): Look into lto_stmt_uids
9375         (ipa_clear_stmts_in_references): Do not clear speculative calls.
9376         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
9377         (ipa_find_reference): Update declaration.
9378         (ipa_clone_ref): Declare.
9379         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
9380         stream speculative flag.
9381         (lto_output_ref): Stream statements uids and speculation.
9382         (input_ref): Likewise.
9383         (input_edge): Stream speuclation.
9384         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
9385         (cgraph_set_call_stmt_including_clones): Handle speculation.
9386         * ipa-inline.c (heap_edge_removal_hook): New function.
9387         (inline_small_functions): Register it.
9388         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
9389         also initialize refs.
9390         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
9391         edge to change.
9392         (try_make_edge_direct_simple_call): Likewise.
9393         (try_make_edge_direct_simple_call): Likewise.
9394         (update_indirect_edges_after_inlining): Likewise.
9395         (remove_described_reference): Look proper lto_stmt_uid.
9396         (propagate_controlled_uses): Likewise.
9397         (propagate_controlled_uses): Liekwise.
9398         * tree-inline.c (copy_bb): Copy speculative edges.
9399         (redirect_all_calls): New function.
9400         (copy_cfg_body): Do redirection after loop info is updated.
9401         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
9403 2013-08-09  Jan Hubicka  <jh@suse.cz>
9405         * lto-streamer-out.c (output_function): Renumber PHIs.
9406         * lto-streamer-in.c (input_function): Likewise.
9408 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
9410         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
9411         (get_lane_unsigned): Likewise.
9412         (dup_lane_scalar): Likewise.
9413         (get_lane): enable for VALL.
9414         * config/aarch64/aarch64-simd.md
9415         (aarch64_dup_lane_scalar<mode>): Remove.
9416         (aarch64_get_lane_signed<mode>): Likewise.
9417         (aarch64_get_lane_unsigned<mode>): Likewise.
9418         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
9419         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
9420         (aarch64_get_lane<mode>): Enable for all vector modes.
9421         (aarch64_get_lanedi): Remove misleading constraints.
9422         * config/aarch64/arm_neon.h
9423         (__aarch64_vget_lane_any): Define.
9424         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
9425         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
9426         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
9427         * config/aarch64/iterators.md (VDQQH): New.
9428         (VDQQHS): Likewise.
9429         (vwcore): Likewise.
9431 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
9433         * configure.ac: Add GAS check for LEON instructions on SPARC.
9434         * configure: Regenerate.
9435         * config.in: Likewise.
9436         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
9437         sparc*-*-* block.
9438         * config/sparc/sparc.opt (LEON, LEON3): New masks.
9439         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
9440         for LEON or LEON3.
9441         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
9442         (AS_LEON_FLAG): New macro.
9443         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
9444         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
9445         Deal with LEON and LEON3 for the memory model.
9446         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
9447         (atomic_compare_and_swap<mode>_1): Likewise.
9448         (*atomic_compare_and_swap<mode>_1): Likewise.
9450 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
9452         * config/arm/neon.md (vcond): Fix floating-point vector
9453         comparisons against 0.
9455 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
9457         * lra-constraints.c (emit_spill_move): Remove assert.
9458         (process_alt_operands): Add more debugging
9459         output.  Increase reject for spilling into memory.  Decrease
9460         reject for reloading scratch.
9461         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
9463 2013-08-08  Steve Ellcey  <sellcey@mips.com>
9465         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
9466         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
9467         micromips incompatible.  Add nan2008.
9468         (MULTILIB_DIRNAMES): Add nan2008.
9469         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9470         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
9471         and micromips incompatible.  Add nan2008.
9472         (MULTILIB_DIRNAMES): Add nan2008.
9473         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
9475 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
9477         PR rtl-optimization/58079
9478         * combine.c (combine_simplify_rtx): Avoid using SUBST if
9479         simplify_comparison has widened a comparison with an integer.
9481 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
9483         * config/arm/neon.md (movmisalign<mode>): Disable when we
9484         don't allow unaligned accesses.
9485         (*movmisalign<mode>_neon_store): Likewise.
9486         (*movmisalign<mode>_neon_load): Likewise.
9487         (*movmisalign<mode>_neon_store): Likewise.
9488         (*movmisalign<mode>_neon_load): Likewise.
9490 2013-08-08  Jan Hubicka  <jh@suse.cz>
9492         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
9493         (make_pass_rebuild_cgraph_edges): Also clear references.
9494         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
9495         * ipa-inline-transform.c (inline_transform): Remove all references
9496         after inlining.
9497         * cgraphunit.c (expand_function): Remove all references after
9498         expansion.
9499         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
9500         (ipa_find_reference): Rewrite to iterator.
9501         (remove_stmt_references): Likewise.
9502         (ipa_clear_stmts_in_references): New function.
9503         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
9504         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
9505         clear references.
9506         * ipa-split.c (split_function): Remove references in split function.
9508 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9510         PR target/57431
9511         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
9512         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
9514 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
9516         PR target/56979
9517         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
9518         suggested mode for the assignment isn't compatible with the
9519         registers required.
9521 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
9523         PR target/58065
9524         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
9526 2013-08-07  Xinliang David Li  <davidxl@google.com>
9528         * config/i386/i386.opt: New option -mtune-ctrl=.
9529         * config/i386/x86-tune.def: New file.
9530         * config/i386/i386.h: include x86-tune.def.
9531         * config/i386/i386.c (ix86_option_override_internal):
9532         Parsing -mtune-ctrl= option and set tune features.
9534 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
9536         PR other/12081
9537         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
9538         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
9539         to gen_2arg_fn_t.
9541 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9543         * rtl.h (update_alignments): Declare.
9544         * final.c (grow_label_align): New function extracted from...
9545         (shorten_branches): ...here.  Call it.
9546         (update_alignments): New function.
9547         * reorg.c (sibling_labels): New variable.
9548         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
9549         the new label along with it onto the sibling_labels vector.
9550         (fill_simple_delay_slots): Adjust call to get_label_before.
9551         (fill_slots_from_thread): Likewise.
9552         (relax_delay_slots): Likewise.
9553         (make_return_insns): Likewise.
9554         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
9556 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
9558         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
9559         document its semantics.
9560         (diagnostic_report_diagnostic): Adjust accordingly.
9562 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9564         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
9565         (sparc_option_override): ...and port to new C++ pass API.
9566         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
9568 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
9570         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
9572 2013-08-06  Caroline Tice  <cmtice@google.com>
9574         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
9575         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
9576         * tree-pass.h: Add pass_vtable_verify.
9577         * varasm.c (assemble_variable): Add code to properly set the comdat
9578         section and name for the .vtable_map_vars section.
9579         (assemble_vtyv_preinit_initializer): New function.
9580         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
9581         LINK_ONCE flag.
9582         * output.h: Add function decl for assemble_vtv_preinit_initializer.
9583         * vtable-verify.c: New file.
9584         * vtable-verify.h: New file.
9585         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
9586         initialiation levels.
9587         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
9588         * passes.def: Insert pass_vtable_verify.
9589         * aclocal.m4: Reorder includes.
9590         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
9591         -fvtv-counts options.
9592         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
9593         as appropriate, if -fvtable-verify=... is used.
9594         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
9595         -fvtable-verify=... is used.
9596         * Makefile.in (OBJS):  Add vtable-verify.o to list.
9597         (vtable-verify.o): Add new build rule.
9598         (GTFILES): Add vtable-verify.c to list.
9599         * common.opt (fvtable-verify=): New flag.
9600         (vtv_priority): Values for fvtable-verify= flag.
9601         (fvtv-counts): New flag.
9602         (fvtv-debug): New flag.
9603         * tree.h (save_vtable_map_decl): New extern function decl.
9605 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9607         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
9608         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
9609         (pass_data_rl78_devirt): ...new pass_data instance and...
9610         (make_pass_rl78_devirt): ...new function.
9611         (rl78_asm_file_start): Port pass registration to new C++ API.
9613 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
9615         * coretypes.h (rtl_opt_pass): Add.
9616         (gcc::context): Add.
9617         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
9618         (epiphany_init): Port to new C++ pass API.
9619         (epiphany_optimize_mode_switching): Likewise.
9620         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
9621         (pass_manager::get_pass_mode_switching): New.
9622         (pass_manager::get_pass_peephole2): New.
9623         * mode-switching.c (pass_mode_switching): Add clone method.
9624         * recog.c (pass_peephole2): Add clone method.
9625         (pass_split_all_insns): Add clone method.
9627 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
9629         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
9630         (mips_option_override): ...here, porting to new C++ API for passes.
9632 2013-08-06  Jan Hubicka  <jh@suse.cz>
9634         * cgraph.c (cgraph_get_body): New function based on lto.c
9635         implementation.
9636         * cgraph.h (cgraph_get_body): Declare.
9637         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
9638         LTO paths.
9639         * cgraphunit.c (expand_function): Get body prior expanding.
9640         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
9641         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
9642         really need.
9643         * passes.c (do_per_function_toporder): Get body.
9644         * tree-inline.c (expand_call_inline): Get body prior inlining it.
9645         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
9647 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9649         PR fortran/57987
9650         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
9651         is not re-finalized.  Rename second parameter to no_collect.
9653 2013-08-06  Martin Jambor  <mjambor@suse.cz>
9655         PR middle-end/58041
9656         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
9657         MEM_REF has proper alignment information.
9659 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
9661         PR other/12081
9662         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
9663         class insn_gen_fn.
9664         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
9665         rtx (*) (rtx, ...) with insn_gen_fn.
9666         * genoutput.c (output_insn_data): Cast gen_? function pointers to
9667         insn_gen_fn::stored_funcptr.  Add initializer braces.
9669 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9671         Rewrite how instances of passes are cloned to remove assumptions
9672         about their sizes (thus allowing pass subclasses to have
9673         additional data fields, albeit non-GC-managed ones at this point).
9675         * passes.c (make_pass_instance): Now that passes have clone
9676         methods, rewrite this function to eliminate XNEW and memcpy
9677         calls that used hardcoded sizes.  Since this function no longer
9678         creates pass instances, rename it to...
9679         (add_pass_instance): ...this.  Document the old way that passes were
9680         numbered and flagged, and rework this function to continue using it.
9681         (next_pass_1): Add an initial_pass argument for use by
9682         add_pass_instance.
9683         (position_pass): When adding multiple instances of a pass, use
9684         the pass's clone method, rather than relying on the XNEW/memcpy
9685         within the former make_pass_instance (now add_pass_instance).
9686         (pass_manager::pass_manager): When invoking next_pass_1, also supply
9687         the initial instance of the current pass within the pass manager.
9689 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
9691         This is the automated part of the conversion of passes from C
9692         structs to C++ classes.
9694         Patch autogenerated by refactor_passes.py from
9695         https://github.com/davidmalcolm/gcc-refactoring-scripts
9696         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
9698         * asan.c (pass_asan): Convert from a global struct to a subclass of
9699         gimple_opt_pass along with...
9700         (pass_data_asan): ...new pass_data instance and...
9701         (make_pass_asan): ...new function.
9702         (pass_asan_O0): Convert from a global struct to a subclass of
9703         gimple_opt_pass along with...
9704         (pass_data_asan_O0): ...new pass_data instance and...
9705         (make_pass_asan_O0): ...new function.
9706         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
9707         subclass of rtl_opt_pass along with...
9708         (pass_data_inc_dec): ...new pass_data instance and...
9709         (make_pass_inc_dec): ...new function.
9710         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
9711         a subclass of rtl_opt_pass along with...
9712         (pass_data_reorder_blocks): ...new pass_data instance and...
9713         (make_pass_reorder_blocks): ...new function.
9714         (pass_duplicate_computed_gotos): Convert from a global struct to a
9715         subclass of rtl_opt_pass along with...
9716         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
9717         (make_pass_duplicate_computed_gotos): ...new function.
9718         (pass_partition_blocks): Convert from a global struct to a subclass of
9719         rtl_opt_pass along with...
9720         (pass_data_partition_blocks): ...new pass_data instance and...
9721         (make_pass_partition_blocks): ...new function.
9722         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
9723         struct to a subclass of rtl_opt_pass along with...
9724         (pass_data_branch_target_load_optimize1): ...new pass_data instance
9725         and...
9726         (make_pass_branch_target_load_optimize1): ...new function.
9727         (pass_branch_target_load_optimize2): Convert from a global struct to a
9728         subclass of rtl_opt_pass along with...
9729         (pass_data_branch_target_load_optimize2): ...new pass_data instance
9730         and...
9731         (make_pass_branch_target_load_optimize2): ...new function.
9732         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
9733         of rtl_opt_pass along with...
9734         (pass_data_jump): ...new pass_data instance and...
9735         (make_pass_jump): ...new function.
9736         (pass_jump2): Convert from a global struct to a subclass of
9737         rtl_opt_pass along with...
9738         (pass_data_jump2): ...new pass_data instance and...
9739         (make_pass_jump2): ...new function.
9740         * cfgexpand.c (pass_expand): Convert from a global struct to a
9741         subclass of rtl_opt_pass along with...
9742         (pass_data_expand): ...new pass_data instance and...
9743         (make_pass_expand): ...new function.
9744         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
9745         of rtl_opt_pass along with...
9746         (pass_data_free_cfg): ...new pass_data instance and...
9747         (make_pass_free_cfg): ...new function.
9748         (pass_into_cfg_layout_mode): Convert from a global struct to a
9749         subclass of rtl_opt_pass along with...
9750         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
9751         (make_pass_into_cfg_layout_mode): ...new function.
9752         (pass_outof_cfg_layout_mode): Convert from a global struct to a
9753         subclass of rtl_opt_pass along with...
9754         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
9755         (make_pass_outof_cfg_layout_mode): ...new function.
9756         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
9757         struct to a subclass of gimple_opt_pass along with...
9758         (pass_data_build_cgraph_edges): ...new pass_data instance and...
9759         (make_pass_build_cgraph_edges): ...new function.
9760         (pass_rebuild_cgraph_edges): Convert from a global struct to a
9761         subclass of gimple_opt_pass along with...
9762         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
9763         (make_pass_rebuild_cgraph_edges): ...new function.
9764         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
9765         subclass of gimple_opt_pass along with...
9766         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
9767         and...
9768         (make_pass_remove_cgraph_callee_edges): ...new function.
9769         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
9770         struct to a subclass of rtl_opt_pass along with...
9771         (pass_data_stack_adjustments): ...new pass_data instance and...
9772         (make_pass_stack_adjustments): ...new function.
9773         * combine.c (pass_combine): Convert from a global struct to a subclass
9774         of rtl_opt_pass along with...
9775         (pass_data_combine): ...new pass_data instance and...
9776         (make_pass_combine): ...new function.
9777         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
9778         global struct to a subclass of rtl_opt_pass along with...
9779         (pass_data_compare_elim_after_reload): ...new pass_data instance
9780         and...
9781         (make_pass_compare_elim_after_reload): ...new function.
9782         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
9783         of rtl_opt_pass along with...
9784         (pass_data_rtl_cprop): ...new pass_data instance and...
9785         (make_pass_rtl_cprop): ...new function.
9786         * cse.c (pass_cse): Convert from a global struct to a subclass of
9787         rtl_opt_pass along with...
9788         (pass_data_cse): ...new pass_data instance and...
9789         (make_pass_cse): ...new function.
9790         (pass_cse2): Convert from a global struct to a subclass of
9791         rtl_opt_pass along with...
9792         (pass_data_cse2): ...new pass_data instance and...
9793         (make_pass_cse2): ...new function.
9794         (pass_cse_after_global_opts): Convert from a global struct to a
9795         subclass of rtl_opt_pass along with...
9796         (pass_data_cse_after_global_opts): ...new pass_data instance and...
9797         (make_pass_cse_after_global_opts): ...new function.
9798         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
9799         of rtl_opt_pass along with...
9800         (pass_data_ud_rtl_dce): ...new pass_data instance and...
9801         (make_pass_ud_rtl_dce): ...new function.
9802         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
9803         rtl_opt_pass along with...
9804         (pass_data_fast_rtl_dce): ...new pass_data instance and...
9805         (make_pass_fast_rtl_dce): ...new function.
9806         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
9807         a subclass of rtl_opt_pass along with...
9808         (pass_data_df_initialize_opt): ...new pass_data instance and...
9809         (make_pass_df_initialize_opt): ...new function.
9810         (pass_df_initialize_no_opt): Convert from a global struct to a
9811         subclass of rtl_opt_pass along with...
9812         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
9813         (make_pass_df_initialize_no_opt): ...new function.
9814         (pass_df_finish): Convert from a global struct to a subclass of
9815         rtl_opt_pass along with...
9816         (pass_data_df_finish): ...new pass_data instance and...
9817         (make_pass_df_finish): ...new function.
9818         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
9819         rtl_opt_pass along with...
9820         (pass_data_rtl_dse1): ...new pass_data instance and...
9821         (make_pass_rtl_dse1): ...new function.
9822         (pass_rtl_dse2): Convert from a global struct to a subclass of
9823         rtl_opt_pass along with...
9824         (pass_data_rtl_dse2): ...new pass_data instance and...
9825         (make_pass_rtl_dse2): ...new function.
9826         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
9827         subclass of rtl_opt_pass along with...
9828         (pass_data_dwarf2_frame): ...new pass_data instance and...
9829         (make_pass_dwarf2_frame): ...new function.
9830         * except.c (pass_set_nothrow_function_flags): Convert from a global
9831         struct to a subclass of rtl_opt_pass along with...
9832         (pass_data_set_nothrow_function_flags): ...new pass_data instance
9833         and...
9834         (make_pass_set_nothrow_function_flags): ...new function.
9835         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
9836         subclass of rtl_opt_pass along with...
9837         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
9838         and...
9839         (make_pass_convert_to_eh_region_ranges): ...new function.
9840         * final.c (pass_compute_alignments): Convert from a global struct to a
9841         subclass of rtl_opt_pass along with...
9842         (pass_data_compute_alignments): ...new pass_data instance and...
9843         (make_pass_compute_alignments): ...new function.
9844         (pass_final): Convert from a global struct to a subclass of
9845         rtl_opt_pass along with...
9846         (pass_data_final): ...new pass_data instance and...
9847         (make_pass_final): ...new function.
9848         (pass_shorten_branches): Convert from a global struct to a subclass of
9849         rtl_opt_pass along with...
9850         (pass_data_shorten_branches): ...new pass_data instance and...
9851         (make_pass_shorten_branches): ...new function.
9852         (pass_clean_state): Convert from a global struct to a subclass of
9853         rtl_opt_pass along with...
9854         (pass_data_clean_state): ...new pass_data instance and...
9855         (make_pass_clean_state): ...new function.
9856         * function.c (pass_instantiate_virtual_regs): Convert from a global
9857         struct to a subclass of rtl_opt_pass along with...
9858         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
9859         (make_pass_instantiate_virtual_regs): ...new function.
9860         (pass_leaf_regs): Convert from a global struct to a subclass of
9861         rtl_opt_pass along with...
9862         (pass_data_leaf_regs): ...new pass_data instance and...
9863         (make_pass_leaf_regs): ...new function.
9864         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
9865         subclass of rtl_opt_pass along with...
9866         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
9867         and...
9868         (make_pass_thread_prologue_and_epilogue): ...new function.
9869         (pass_match_asm_constraints): Convert from a global struct to a
9870         subclass of rtl_opt_pass along with...
9871         (pass_data_match_asm_constraints): ...new pass_data instance and...
9872         (make_pass_match_asm_constraints): ...new function.
9873         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
9874         subclass of rtl_opt_pass along with...
9875         (pass_data_rtl_fwprop): ...new pass_data instance and...
9876         (make_pass_rtl_fwprop): ...new function.
9877         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
9878         rtl_opt_pass along with...
9879         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
9880         (make_pass_rtl_fwprop_addr): ...new function.
9881         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
9882         rtl_opt_pass along with...
9883         (pass_data_rtl_pre): ...new pass_data instance and...
9884         (make_pass_rtl_pre): ...new function.
9885         (pass_rtl_hoist): Convert from a global struct to a subclass of
9886         rtl_opt_pass along with...
9887         (pass_data_rtl_hoist): ...new pass_data instance and...
9888         (make_pass_rtl_hoist): ...new function.
9889         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
9890         subclass of gimple_opt_pass along with...
9891         (pass_data_lower_cf): ...new pass_data instance and...
9892         (make_pass_lower_cf): ...new function.
9893         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
9894         from a global struct to a subclass of gimple_opt_pass along with...
9895         (pass_data_strength_reduction): ...new pass_data instance and...
9896         (make_pass_strength_reduction): ...new function.
9897         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
9898         of rtl_opt_pass along with...
9899         (pass_data_rtl_ifcvt): ...new pass_data instance and...
9900         (make_pass_rtl_ifcvt): ...new function.
9901         (pass_if_after_combine): Convert from a global struct to a subclass of
9902         rtl_opt_pass along with...
9903         (pass_data_if_after_combine): ...new pass_data instance and...
9904         (make_pass_if_after_combine): ...new function.
9905         (pass_if_after_reload): Convert from a global struct to a subclass of
9906         rtl_opt_pass along with...
9907         (pass_data_if_after_reload): ...new pass_data instance and...
9908         (make_pass_if_after_reload): ...new function.
9909         * init-regs.c (pass_initialize_regs): Convert from a global struct to
9910         a subclass of rtl_opt_pass along with...
9911         (pass_data_initialize_regs): ...new pass_data instance and...
9912         (make_pass_initialize_regs): ...new function.
9913         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
9914         of ipa_opt_pass_d along with...
9915         (pass_data_ipa_cp): ...new pass_data instance and...
9916         (make_pass_ipa_cp): ...new function.
9917         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
9918         global struct to a subclass of gimple_opt_pass along with...
9919         (pass_data_inline_parameters): ...new pass_data instance and...
9920         (make_pass_inline_parameters): ...new function.
9921         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
9922         subclass of gimple_opt_pass along with...
9923         (pass_data_early_inline): ...new pass_data instance and...
9924         (make_pass_early_inline): ...new function.
9925         (pass_ipa_inline): Convert from a global struct to a subclass of
9926         ipa_opt_pass_d along with...
9927         (pass_data_ipa_inline): ...new pass_data instance and...
9928         (make_pass_ipa_inline): ...new function.
9929         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
9930         struct to a subclass of gimple_opt_pass along with...
9931         (pass_data_local_pure_const): ...new pass_data instance and...
9932         (make_pass_local_pure_const): ...new function.
9933         (pass_ipa_pure_const): Convert from a global struct to a subclass of
9934         ipa_opt_pass_d along with...
9935         (pass_data_ipa_pure_const): ...new pass_data instance and...
9936         (make_pass_ipa_pure_const): ...new function.
9937         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
9938         to a subclass of ipa_opt_pass_d along with...
9939         (pass_data_ipa_reference): ...new pass_data instance and...
9940         (make_pass_ipa_reference): ...new function.
9941         * ipa-split.c (pass_split_functions): Convert from a global struct to
9942         a subclass of gimple_opt_pass along with...
9943         (pass_data_split_functions): ...new pass_data instance and...
9944         (make_pass_split_functions): ...new function.
9945         (pass_feedback_split_functions): Convert from a global struct to a
9946         subclass of gimple_opt_pass along with...
9947         (pass_data_feedback_split_functions): ...new pass_data instance and...
9948         (make_pass_feedback_split_functions): ...new function.
9949         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
9950         global struct to a subclass of simple_ipa_opt_pass along with...
9951         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
9952         instance and...
9953         (make_pass_ipa_function_and_variable_visibility): ...new function.
9954         (pass_ipa_free_inline_summary): Convert from a global struct to a
9955         subclass of simple_ipa_opt_pass along with...
9956         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
9957         (make_pass_ipa_free_inline_summary): ...new function.
9958         (pass_ipa_whole_program_visibility): Convert from a global struct to a
9959         subclass of ipa_opt_pass_d along with...
9960         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
9961         and...
9962         (make_pass_ipa_whole_program_visibility): ...new function.
9963         (pass_ipa_profile): Convert from a global struct to a subclass of
9964         ipa_opt_pass_d along with...
9965         (pass_data_ipa_profile): ...new pass_data instance and...
9966         (make_pass_ipa_profile): ...new function.
9967         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
9968         ipa_opt_pass_d along with...
9969         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
9970         (make_pass_ipa_cdtor_merge): ...new function.
9971         * ira.c (pass_ira): Convert from a global struct to a subclass of
9972         rtl_opt_pass along with...
9973         (pass_data_ira): ...new pass_data instance and...
9974         (make_pass_ira): ...new function.
9975         (pass_reload): Convert from a global struct to a subclass of
9976         rtl_opt_pass along with...
9977         (pass_data_reload): ...new pass_data instance and...
9978         (make_pass_reload): ...new function.
9979         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
9980         subclass of rtl_opt_pass along with...
9981         (pass_data_cleanup_barriers): ...new pass_data instance and...
9982         (make_pass_cleanup_barriers): ...new function.
9983         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
9984         of rtl_opt_pass along with...
9985         (pass_data_loop2): ...new pass_data instance and...
9986         (make_pass_loop2): ...new function.
9987         (pass_rtl_loop_init): Convert from a global struct to a subclass of
9988         rtl_opt_pass along with...
9989         (pass_data_rtl_loop_init): ...new pass_data instance and...
9990         (make_pass_rtl_loop_init): ...new function.
9991         (pass_rtl_loop_done): Convert from a global struct to a subclass of
9992         rtl_opt_pass along with...
9993         (pass_data_rtl_loop_done): ...new pass_data instance and...
9994         (make_pass_rtl_loop_done): ...new function.
9995         (pass_rtl_move_loop_invariants): Convert from a global struct to a
9996         subclass of rtl_opt_pass along with...
9997         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
9998         (make_pass_rtl_move_loop_invariants): ...new function.
9999         (pass_rtl_unswitch): Convert from a global struct to a subclass of
10000         rtl_opt_pass along with...
10001         (pass_data_rtl_unswitch): ...new pass_data instance and...
10002         (make_pass_rtl_unswitch): ...new function.
10003         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
10004         subclass of rtl_opt_pass along with...
10005         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
10006         and...
10007         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10008         (pass_rtl_doloop): Convert from a global struct to a subclass of
10009         rtl_opt_pass along with...
10010         (pass_data_rtl_doloop): ...new pass_data instance and...
10011         (make_pass_rtl_doloop): ...new function.
10012         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
10013         a subclass of rtl_opt_pass along with...
10014         (pass_data_lower_subreg): ...new pass_data instance and...
10015         (make_pass_lower_subreg): ...new function.
10016         (pass_lower_subreg2): Convert from a global struct to a subclass of
10017         rtl_opt_pass along with...
10018         (pass_data_lower_subreg2): ...new pass_data instance and...
10019         (make_pass_lower_subreg2): ...new function.
10020         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
10021         struct to a subclass of ipa_opt_pass_d along with...
10022         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
10023         (make_pass_ipa_lto_gimple_out): ...new function.
10024         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
10025         of ipa_opt_pass_d along with...
10026         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
10027         (make_pass_ipa_lto_finish_out): ...new function.
10028         * mode-switching.c (pass_mode_switching): Convert from a global struct
10029         to a subclass of rtl_opt_pass along with...
10030         (pass_data_mode_switching): ...new pass_data instance and...
10031         (make_pass_mode_switching): ...new function.
10032         * modulo-sched.c (pass_sms): Convert from a global struct to a
10033         subclass of rtl_opt_pass along with...
10034         (pass_data_sms): ...new pass_data instance and...
10035         (make_pass_sms): ...new function.
10036         * omp-low.c (pass_expand_omp): Convert from a global struct to a
10037         subclass of gimple_opt_pass along with...
10038         (pass_data_expand_omp): ...new pass_data instance and...
10039         (make_pass_expand_omp): ...new function.
10040         (pass_lower_omp): Convert from a global struct to a subclass of
10041         gimple_opt_pass along with...
10042         (pass_data_lower_omp): ...new pass_data instance and...
10043         (make_pass_lower_omp): ...new function.
10044         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
10045         of gimple_opt_pass along with...
10046         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
10047         (make_pass_diagnose_omp_blocks): ...new function.
10048         * passes.c (pass_early_local_passes): Convert from a global struct to
10049         a subclass of simple_ipa_opt_pass along with...
10050         (pass_data_early_local_passes): ...new pass_data instance and...
10051         (make_pass_early_local_passes): ...new function.
10052         (pass_all_early_optimizations): Convert from a global struct to a
10053         subclass of gimple_opt_pass along with...
10054         (pass_data_all_early_optimizations): ...new pass_data instance and...
10055         (make_pass_all_early_optimizations): ...new function.
10056         (pass_all_optimizations): Convert from a global struct to a subclass
10057         of gimple_opt_pass along with...
10058         (pass_data_all_optimizations): ...new pass_data instance and...
10059         (make_pass_all_optimizations): ...new function.
10060         (pass_all_optimizations_g): Convert from a global struct to a subclass
10061         of gimple_opt_pass along with...
10062         (pass_data_all_optimizations_g): ...new pass_data instance and...
10063         (make_pass_all_optimizations_g): ...new function.
10064         (pass_rest_of_compilation): Convert from a global struct to a subclass
10065         of rtl_opt_pass along with...
10066         (pass_data_rest_of_compilation): ...new pass_data instance and...
10067         (make_pass_rest_of_compilation): ...new function.
10068         (pass_postreload): Convert from a global struct to a subclass of
10069         rtl_opt_pass along with...
10070         (pass_data_postreload): ...new pass_data instance and...
10071         (make_pass_postreload): ...new function.
10072         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
10073         subclass of rtl_opt_pass along with...
10074         (pass_data_gcse2): ...new pass_data instance and...
10075         (make_pass_gcse2): ...new function.
10076         * postreload.c (pass_postreload_cse): Convert from a global struct to
10077         a subclass of rtl_opt_pass along with...
10078         (pass_data_postreload_cse): ...new pass_data instance and...
10079         (make_pass_postreload_cse): ...new function.
10080         * predict.c (pass_profile): Convert from a global struct to a subclass
10081         of gimple_opt_pass along with...
10082         (pass_data_profile): ...new pass_data instance and...
10083         (make_pass_profile): ...new function.
10084         (pass_strip_predict_hints): Convert from a global struct to a subclass
10085         of gimple_opt_pass along with...
10086         (pass_data_strip_predict_hints): ...new pass_data instance and...
10087         (make_pass_strip_predict_hints): ...new function.
10088         * recog.c (pass_peephole2): Convert from a global struct to a subclass
10089         of rtl_opt_pass along with...
10090         (pass_data_peephole2): ...new pass_data instance and...
10091         (make_pass_peephole2): ...new function.
10092         (pass_split_all_insns): Convert from a global struct to a subclass of
10093         rtl_opt_pass along with...
10094         (pass_data_split_all_insns): ...new pass_data instance and...
10095         (make_pass_split_all_insns): ...new function.
10096         (pass_split_after_reload): Convert from a global struct to a subclass
10097         of rtl_opt_pass along with...
10098         (pass_data_split_after_reload): ...new pass_data instance and...
10099         (make_pass_split_after_reload): ...new function.
10100         (pass_split_before_regstack): Convert from a global struct to a
10101         subclass of rtl_opt_pass along with...
10102         (pass_data_split_before_regstack): ...new pass_data instance and...
10103         (make_pass_split_before_regstack): ...new function.
10104         (pass_split_before_sched2): Convert from a global struct to a subclass
10105         of rtl_opt_pass along with...
10106         (pass_data_split_before_sched2): ...new pass_data instance and...
10107         (make_pass_split_before_sched2): ...new function.
10108         (pass_split_for_shorten_branches): Convert from a global struct to a
10109         subclass of rtl_opt_pass along with...
10110         (pass_data_split_for_shorten_branches): ...new pass_data instance
10111         and...
10112         (make_pass_split_for_shorten_branches): ...new function.
10113         * ree.c (pass_ree): Convert from a global struct to a subclass of
10114         rtl_opt_pass along with...
10115         (pass_data_ree): ...new pass_data instance and...
10116         (make_pass_ree): ...new function.
10117         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
10118         subclass of rtl_opt_pass along with...
10119         (pass_data_stack_regs): ...new pass_data instance and...
10120         (make_pass_stack_regs): ...new function.
10121         (pass_stack_regs_run): Convert from a global struct to a subclass of
10122         rtl_opt_pass along with...
10123         (pass_data_stack_regs_run): ...new pass_data instance and...
10124         (make_pass_stack_regs_run): ...new function.
10125         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
10126         subclass of rtl_opt_pass along with...
10127         (pass_data_cprop_hardreg): ...new pass_data instance and...
10128         (make_pass_cprop_hardreg): ...new function.
10129         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
10130         subclass of rtl_opt_pass along with...
10131         (pass_data_reginfo_init): ...new pass_data instance and...
10132         (make_pass_reginfo_init): ...new function.
10133         * regmove.c (pass_regmove): Convert from a global struct to a subclass
10134         of rtl_opt_pass along with...
10135         (pass_data_regmove): ...new pass_data instance and...
10136         (make_pass_regmove): ...new function.
10137         * regrename.c (pass_regrename): Convert from a global struct to a
10138         subclass of rtl_opt_pass along with...
10139         (pass_data_regrename): ...new pass_data instance and...
10140         (make_pass_regrename): ...new function.
10141         * reorg.c (pass_delay_slots): Convert from a global struct to a
10142         subclass of rtl_opt_pass along with...
10143         (pass_data_delay_slots): ...new pass_data instance and...
10144         (make_pass_delay_slots): ...new function.
10145         (pass_machine_reorg): Convert from a global struct to a subclass of
10146         rtl_opt_pass along with...
10147         (pass_data_machine_reorg): ...new pass_data instance and...
10148         (make_pass_machine_reorg): ...new function.
10149         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
10150         of rtl_opt_pass along with...
10151         (pass_data_sched): ...new pass_data instance and...
10152         (make_pass_sched): ...new function.
10153         (pass_sched2): Convert from a global struct to a subclass of
10154         rtl_opt_pass along with...
10155         (pass_data_sched2): ...new pass_data instance and...
10156         (make_pass_sched2): ...new function.
10157         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
10158         to a subclass of rtl_opt_pass along with...
10159         (pass_data_stack_ptr_mod): ...new pass_data instance and...
10160         (make_pass_stack_ptr_mod): ...new function.
10161         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
10162         to a subclass of rtl_opt_pass along with...
10163         (pass_data_rtl_store_motion): ...new pass_data instance and...
10164         (make_pass_rtl_store_motion): ...new function.
10165         * tracer.c (pass_tracer): Convert from a global struct to a subclass
10166         of gimple_opt_pass along with...
10167         (pass_data_tracer): ...new pass_data instance and...
10168         (make_pass_tracer): ...new function.
10169         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
10170         to a subclass of gimple_opt_pass along with...
10171         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
10172         (make_pass_diagnose_tm_blocks): ...new function.
10173         (pass_lower_tm): Convert from a global struct to a subclass of
10174         gimple_opt_pass along with...
10175         (pass_data_lower_tm): ...new pass_data instance and...
10176         (make_pass_lower_tm): ...new function.
10177         (pass_tm_init): Convert from a global struct to a subclass of
10178         gimple_opt_pass along with...
10179         (pass_data_tm_init): ...new pass_data instance and...
10180         (make_pass_tm_init): ...new function.
10181         (pass_tm_mark): Convert from a global struct to a subclass of
10182         gimple_opt_pass along with...
10183         (pass_data_tm_mark): ...new pass_data instance and...
10184         (make_pass_tm_mark): ...new function.
10185         (pass_tm_edges): Convert from a global struct to a subclass of
10186         gimple_opt_pass along with...
10187         (pass_data_tm_edges): ...new pass_data instance and...
10188         (make_pass_tm_edges): ...new function.
10189         (pass_tm_memopt): Convert from a global struct to a subclass of
10190         gimple_opt_pass along with...
10191         (pass_data_tm_memopt): ...new pass_data instance and...
10192         (make_pass_tm_memopt): ...new function.
10193         (pass_ipa_tm): Convert from a global struct to a subclass of
10194         simple_ipa_opt_pass along with...
10195         (pass_data_ipa_tm): ...new pass_data instance and...
10196         (make_pass_ipa_tm): ...new function.
10197         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
10198         subclass of gimple_opt_pass along with...
10199         (pass_data_call_cdce): ...new pass_data instance and...
10200         (make_pass_call_cdce): ...new function.
10201         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
10202         subclass of gimple_opt_pass along with...
10203         (pass_data_build_cfg): ...new pass_data instance and...
10204         (make_pass_build_cfg): ...new function.
10205         (pass_split_crit_edges): Convert from a global struct to a subclass of
10206         gimple_opt_pass along with...
10207         (pass_data_split_crit_edges): ...new pass_data instance and...
10208         (make_pass_split_crit_edges): ...new function.
10209         (pass_warn_function_return): Convert from a global struct to a
10210         subclass of gimple_opt_pass along with...
10211         (pass_data_warn_function_return): ...new pass_data instance and...
10212         (make_pass_warn_function_return): ...new function.
10213         (pass_warn_function_noreturn): Convert from a global struct to a
10214         subclass of gimple_opt_pass along with...
10215         (pass_data_warn_function_noreturn): ...new pass_data instance and...
10216         (make_pass_warn_function_noreturn): ...new function.
10217         (pass_warn_unused_result): Convert from a global struct to a subclass
10218         of gimple_opt_pass along with...
10219         (pass_data_warn_unused_result): ...new pass_data instance and...
10220         (make_pass_warn_unused_result): ...new function.
10221         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
10222         a subclass of gimple_opt_pass along with...
10223         (pass_data_merge_phi): ...new pass_data instance and...
10224         (make_pass_merge_phi): ...new function.
10225         * tree-complex.c (pass_lower_complex): Convert from a global struct to
10226         a subclass of gimple_opt_pass along with...
10227         (pass_data_lower_complex): ...new pass_data instance and...
10228         (make_pass_lower_complex): ...new function.
10229         (pass_lower_complex_O0): Convert from a global struct to a subclass of
10230         gimple_opt_pass along with...
10231         (pass_data_lower_complex_O0): ...new pass_data instance and...
10232         (make_pass_lower_complex_O0): ...new function.
10233         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
10234         subclass of gimple_opt_pass along with...
10235         (pass_data_lower_eh): ...new pass_data instance and...
10236         (make_pass_lower_eh): ...new function.
10237         (pass_refactor_eh): Convert from a global struct to a subclass of
10238         gimple_opt_pass along with...
10239         (pass_data_refactor_eh): ...new pass_data instance and...
10240         (make_pass_refactor_eh): ...new function.
10241         (pass_lower_resx): Convert from a global struct to a subclass of
10242         gimple_opt_pass along with...
10243         (pass_data_lower_resx): ...new pass_data instance and...
10244         (make_pass_lower_resx): ...new function.
10245         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
10246         of gimple_opt_pass along with...
10247         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
10248         (make_pass_lower_eh_dispatch): ...new function.
10249         (pass_cleanup_eh): Convert from a global struct to a subclass of
10250         gimple_opt_pass along with...
10251         (pass_data_cleanup_eh): ...new pass_data instance and...
10252         (make_pass_cleanup_eh): ...new function.
10253         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
10254         to a subclass of simple_ipa_opt_pass along with...
10255         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
10256         (make_pass_ipa_lower_emutls): ...new function.
10257         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
10258         a subclass of gimple_opt_pass along with...
10259         (pass_data_if_conversion): ...new pass_data instance and...
10260         (make_pass_if_conversion): ...new function.
10261         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
10262         subclass of gimple_opt_pass along with...
10263         (pass_data_build_ssa): ...new pass_data instance and...
10264         (make_pass_build_ssa): ...new function.
10265         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
10266         global struct to a subclass of gimple_opt_pass along with...
10267         (pass_data_loop_distribution): ...new pass_data instance and...
10268         (make_pass_loop_distribution): ...new function.
10269         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
10270         subclass of gimple_opt_pass along with...
10271         (pass_data_mudflap_1): ...new pass_data instance and...
10272         (make_pass_mudflap_1): ...new function.
10273         (pass_mudflap_2): Convert from a global struct to a subclass of
10274         gimple_opt_pass along with...
10275         (pass_data_mudflap_2): ...new pass_data instance and...
10276         (make_pass_mudflap_2): ...new function.
10277         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
10278         subclass of gimple_opt_pass along with...
10279         (pass_data_mudflap_1): ...new pass_data instance and...
10280         (make_pass_mudflap_1): ...new function.
10281         (pass_mudflap_2): Convert from a global struct to a subclass of
10282         gimple_opt_pass along with...
10283         (pass_data_mudflap_2): ...new pass_data instance and...
10284         (make_pass_mudflap_2): ...new function.
10285         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
10286         gimple_opt_pass along with...
10287         (pass_data_nrv): ...new pass_data instance and...
10288         (make_pass_nrv): ...new function.
10289         (pass_return_slot): Convert from a global struct to a subclass of
10290         gimple_opt_pass along with...
10291         (pass_data_return_slot): ...new pass_data instance and...
10292         (make_pass_return_slot): ...new function.
10293         * tree-object-size.c (pass_object_sizes): Convert from a global struct
10294         to a subclass of gimple_opt_pass along with...
10295         (pass_data_object_sizes): ...new pass_data instance and...
10296         (make_pass_object_sizes): ...new function.
10297         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
10298         global struct to a subclass of gimple_opt_pass along with...
10299         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
10300         and...
10301         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10302         (pass_fixup_cfg): Convert from a global struct to a subclass of
10303         gimple_opt_pass along with...
10304         (pass_data_fixup_cfg): ...new pass_data instance and...
10305         (make_pass_fixup_cfg): ...new function.
10306         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
10307         (make_pass_mudflap_1): ...new function.
10308         (pass_mudflap_2): Replace declaration with that of...
10309         (make_pass_mudflap_2): ...new function.
10310         (pass_asan): Replace declaration with that of...
10311         (make_pass_asan): ...new function.
10312         (pass_asan_O0): Replace declaration with that of...
10313         (make_pass_asan_O0): ...new function.
10314         (pass_tsan): Replace declaration with that of...
10315         (make_pass_tsan): ...new function.
10316         (pass_tsan_O0): Replace declaration with that of...
10317         (make_pass_tsan_O0): ...new function.
10318         (pass_lower_cf): Replace declaration with that of...
10319         (make_pass_lower_cf): ...new function.
10320         (pass_refactor_eh): Replace declaration with that of...
10321         (make_pass_refactor_eh): ...new function.
10322         (pass_lower_eh): Replace declaration with that of...
10323         (make_pass_lower_eh): ...new function.
10324         (pass_lower_eh_dispatch): Replace declaration with that of...
10325         (make_pass_lower_eh_dispatch): ...new function.
10326         (pass_lower_resx): Replace declaration with that of...
10327         (make_pass_lower_resx): ...new function.
10328         (pass_build_cfg): Replace declaration with that of...
10329         (make_pass_build_cfg): ...new function.
10330         (pass_early_tree_profile): Replace declaration with that of...
10331         (make_pass_early_tree_profile): ...new function.
10332         (pass_cleanup_eh): Replace declaration with that of...
10333         (make_pass_cleanup_eh): ...new function.
10334         (pass_sra): Replace declaration with that of...
10335         (make_pass_sra): ...new function.
10336         (pass_sra_early): Replace declaration with that of...
10337         (make_pass_sra_early): ...new function.
10338         (pass_early_ipa_sra): Replace declaration with that of...
10339         (make_pass_early_ipa_sra): ...new function.
10340         (pass_tail_recursion): Replace declaration with that of...
10341         (make_pass_tail_recursion): ...new function.
10342         (pass_tail_calls): Replace declaration with that of...
10343         (make_pass_tail_calls): ...new function.
10344         (pass_tree_loop): Replace declaration with that of...
10345         (make_pass_tree_loop): ...new function.
10346         (pass_tree_loop_init): Replace declaration with that of...
10347         (make_pass_tree_loop_init): ...new function.
10348         (pass_lim): Replace declaration with that of...
10349         (make_pass_lim): ...new function.
10350         (pass_tree_unswitch): Replace declaration with that of...
10351         (make_pass_tree_unswitch): ...new function.
10352         (pass_predcom): Replace declaration with that of...
10353         (make_pass_predcom): ...new function.
10354         (pass_iv_canon): Replace declaration with that of...
10355         (make_pass_iv_canon): ...new function.
10356         (pass_scev_cprop): Replace declaration with that of...
10357         (make_pass_scev_cprop): ...new function.
10358         (pass_empty_loop): Replace declaration with that of...
10359         (make_pass_empty_loop): ...new function.
10360         (pass_record_bounds): Replace declaration with that of...
10361         (make_pass_record_bounds): ...new function.
10362         (pass_graphite): Replace declaration with that of...
10363         (make_pass_graphite): ...new function.
10364         (pass_graphite_transforms): Replace declaration with that of...
10365         (make_pass_graphite_transforms): ...new function.
10366         (pass_if_conversion): Replace declaration with that of...
10367         (make_pass_if_conversion): ...new function.
10368         (pass_loop_distribution): Replace declaration with that of...
10369         (make_pass_loop_distribution): ...new function.
10370         (pass_vectorize): Replace declaration with that of...
10371         (make_pass_vectorize): ...new function.
10372         (pass_slp_vectorize): Replace declaration with that of...
10373         (make_pass_slp_vectorize): ...new function.
10374         (pass_complete_unroll): Replace declaration with that of...
10375         (make_pass_complete_unroll): ...new function.
10376         (pass_complete_unrolli): Replace declaration with that of...
10377         (make_pass_complete_unrolli): ...new function.
10378         (pass_parallelize_loops): Replace declaration with that of...
10379         (make_pass_parallelize_loops): ...new function.
10380         (pass_loop_prefetch): Replace declaration with that of...
10381         (make_pass_loop_prefetch): ...new function.
10382         (pass_iv_optimize): Replace declaration with that of...
10383         (make_pass_iv_optimize): ...new function.
10384         (pass_tree_loop_done): Replace declaration with that of...
10385         (make_pass_tree_loop_done): ...new function.
10386         (pass_ch): Replace declaration with that of...
10387         (make_pass_ch): ...new function.
10388         (pass_ccp): Replace declaration with that of...
10389         (make_pass_ccp): ...new function.
10390         (pass_phi_only_cprop): Replace declaration with that of...
10391         (make_pass_phi_only_cprop): ...new function.
10392         (pass_build_ssa): Replace declaration with that of...
10393         (make_pass_build_ssa): ...new function.
10394         (pass_build_alias): Replace declaration with that of...
10395         (make_pass_build_alias): ...new function.
10396         (pass_build_ealias): Replace declaration with that of...
10397         (make_pass_build_ealias): ...new function.
10398         (pass_dominator): Replace declaration with that of...
10399         (make_pass_dominator): ...new function.
10400         (pass_dce): Replace declaration with that of...
10401         (make_pass_dce): ...new function.
10402         (pass_dce_loop): Replace declaration with that of...
10403         (make_pass_dce_loop): ...new function.
10404         (pass_cd_dce): Replace declaration with that of...
10405         (make_pass_cd_dce): ...new function.
10406         (pass_call_cdce): Replace declaration with that of...
10407         (make_pass_call_cdce): ...new function.
10408         (pass_merge_phi): Replace declaration with that of...
10409         (make_pass_merge_phi): ...new function.
10410         (pass_split_crit_edges): Replace declaration with that of...
10411         (make_pass_split_crit_edges): ...new function.
10412         (pass_pre): Replace declaration with that of...
10413         (make_pass_pre): ...new function.
10414         (pass_profile): Replace declaration with that of...
10415         (make_pass_profile): ...new function.
10416         (pass_strip_predict_hints): Replace declaration with that of...
10417         (make_pass_strip_predict_hints): ...new function.
10418         (pass_lower_complex_O0): Replace declaration with that of...
10419         (make_pass_lower_complex_O0): ...new function.
10420         (pass_lower_complex): Replace declaration with that of...
10421         (make_pass_lower_complex): ...new function.
10422         (pass_lower_vector): Replace declaration with that of...
10423         (make_pass_lower_vector): ...new function.
10424         (pass_lower_vector_ssa): Replace declaration with that of...
10425         (make_pass_lower_vector_ssa): ...new function.
10426         (pass_lower_omp): Replace declaration with that of...
10427         (make_pass_lower_omp): ...new function.
10428         (pass_diagnose_omp_blocks): Replace declaration with that of...
10429         (make_pass_diagnose_omp_blocks): ...new function.
10430         (pass_expand_omp): Replace declaration with that of...
10431         (make_pass_expand_omp): ...new function.
10432         (pass_expand_omp_ssa): Replace declaration with that of...
10433         (make_pass_expand_omp_ssa): ...new function.
10434         (pass_object_sizes): Replace declaration with that of...
10435         (make_pass_object_sizes): ...new function.
10436         (pass_strlen): Replace declaration with that of...
10437         (make_pass_strlen): ...new function.
10438         (pass_fold_builtins): Replace declaration with that of...
10439         (make_pass_fold_builtins): ...new function.
10440         (pass_stdarg): Replace declaration with that of...
10441         (make_pass_stdarg): ...new function.
10442         (pass_early_warn_uninitialized): Replace declaration with that of...
10443         (make_pass_early_warn_uninitialized): ...new function.
10444         (pass_late_warn_uninitialized): Replace declaration with that of...
10445         (make_pass_late_warn_uninitialized): ...new function.
10446         (pass_cse_reciprocals): Replace declaration with that of...
10447         (make_pass_cse_reciprocals): ...new function.
10448         (pass_cse_sincos): Replace declaration with that of...
10449         (make_pass_cse_sincos): ...new function.
10450         (pass_optimize_bswap): Replace declaration with that of...
10451         (make_pass_optimize_bswap): ...new function.
10452         (pass_optimize_widening_mul): Replace declaration with that of...
10453         (make_pass_optimize_widening_mul): ...new function.
10454         (pass_warn_function_return): Replace declaration with that of...
10455         (make_pass_warn_function_return): ...new function.
10456         (pass_warn_function_noreturn): Replace declaration with that of...
10457         (make_pass_warn_function_noreturn): ...new function.
10458         (pass_cselim): Replace declaration with that of...
10459         (make_pass_cselim): ...new function.
10460         (pass_phiopt): Replace declaration with that of...
10461         (make_pass_phiopt): ...new function.
10462         (pass_forwprop): Replace declaration with that of...
10463         (make_pass_forwprop): ...new function.
10464         (pass_phiprop): Replace declaration with that of...
10465         (make_pass_phiprop): ...new function.
10466         (pass_tree_ifcombine): Replace declaration with that of...
10467         (make_pass_tree_ifcombine): ...new function.
10468         (pass_dse): Replace declaration with that of...
10469         (make_pass_dse): ...new function.
10470         (pass_nrv): Replace declaration with that of...
10471         (make_pass_nrv): ...new function.
10472         (pass_rename_ssa_copies): Replace declaration with that of...
10473         (make_pass_rename_ssa_copies): ...new function.
10474         (pass_sink_code): Replace declaration with that of...
10475         (make_pass_sink_code): ...new function.
10476         (pass_fre): Replace declaration with that of...
10477         (make_pass_fre): ...new function.
10478         (pass_check_data_deps): Replace declaration with that of...
10479         (make_pass_check_data_deps): ...new function.
10480         (pass_copy_prop): Replace declaration with that of...
10481         (make_pass_copy_prop): ...new function.
10482         (pass_vrp): Replace declaration with that of...
10483         (make_pass_vrp): ...new function.
10484         (pass_uncprop): Replace declaration with that of...
10485         (make_pass_uncprop): ...new function.
10486         (pass_return_slot): Replace declaration with that of...
10487         (make_pass_return_slot): ...new function.
10488         (pass_reassoc): Replace declaration with that of...
10489         (make_pass_reassoc): ...new function.
10490         (pass_rebuild_cgraph_edges): Replace declaration with that of...
10491         (make_pass_rebuild_cgraph_edges): ...new function.
10492         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
10493         (make_pass_remove_cgraph_callee_edges): ...new function.
10494         (pass_build_cgraph_edges): Replace declaration with that of...
10495         (make_pass_build_cgraph_edges): ...new function.
10496         (pass_local_pure_const): Replace declaration with that of...
10497         (make_pass_local_pure_const): ...new function.
10498         (pass_tracer): Replace declaration with that of...
10499         (make_pass_tracer): ...new function.
10500         (pass_warn_unused_result): Replace declaration with that of...
10501         (make_pass_warn_unused_result): ...new function.
10502         (pass_diagnose_tm_blocks): Replace declaration with that of...
10503         (make_pass_diagnose_tm_blocks): ...new function.
10504         (pass_lower_tm): Replace declaration with that of...
10505         (make_pass_lower_tm): ...new function.
10506         (pass_tm_init): Replace declaration with that of...
10507         (make_pass_tm_init): ...new function.
10508         (pass_tm_mark): Replace declaration with that of...
10509         (make_pass_tm_mark): ...new function.
10510         (pass_tm_memopt): Replace declaration with that of...
10511         (make_pass_tm_memopt): ...new function.
10512         (pass_tm_edges): Replace declaration with that of...
10513         (make_pass_tm_edges): ...new function.
10514         (pass_split_functions): Replace declaration with that of...
10515         (make_pass_split_functions): ...new function.
10516         (pass_feedback_split_functions): Replace declaration with that of...
10517         (make_pass_feedback_split_functions): ...new function.
10518         (pass_strength_reduction): Replace declaration with that of...
10519         (make_pass_strength_reduction): ...new function.
10520         (pass_ipa_lower_emutls): Replace declaration with that of...
10521         (make_pass_ipa_lower_emutls): ...new function.
10522         (pass_ipa_function_and_variable_visibility): Replace declaration with
10523         that of...
10524         (make_pass_ipa_function_and_variable_visibility): ...new function.
10525         (pass_ipa_tree_profile): Replace declaration with that of...
10526         (make_pass_ipa_tree_profile): ...new function.
10527         (pass_early_local_passes): Replace declaration with that of...
10528         (make_pass_early_local_passes): ...new function.
10529         (pass_ipa_whole_program_visibility): Replace declaration with that
10530         of...
10531         (make_pass_ipa_whole_program_visibility): ...new function.
10532         (pass_ipa_lto_gimple_out): Replace declaration with that of...
10533         (make_pass_ipa_lto_gimple_out): ...new function.
10534         (pass_ipa_increase_alignment): Replace declaration with that of...
10535         (make_pass_ipa_increase_alignment): ...new function.
10536         (pass_ipa_inline): Replace declaration with that of...
10537         (make_pass_ipa_inline): ...new function.
10538         (pass_ipa_free_lang_data): Replace declaration with that of...
10539         (make_pass_ipa_free_lang_data): ...new function.
10540         (pass_ipa_free_inline_summary): Replace declaration with that of...
10541         (make_pass_ipa_free_inline_summary): ...new function.
10542         (pass_ipa_cp): Replace declaration with that of...
10543         (make_pass_ipa_cp): ...new function.
10544         (pass_ipa_reference): Replace declaration with that of...
10545         (make_pass_ipa_reference): ...new function.
10546         (pass_ipa_pure_const): Replace declaration with that of...
10547         (make_pass_ipa_pure_const): ...new function.
10548         (pass_ipa_pta): Replace declaration with that of...
10549         (make_pass_ipa_pta): ...new function.
10550         (pass_ipa_lto_finish_out): Replace declaration with that of...
10551         (make_pass_ipa_lto_finish_out): ...new function.
10552         (pass_ipa_tm): Replace declaration with that of...
10553         (make_pass_ipa_tm): ...new function.
10554         (pass_ipa_profile): Replace declaration with that of...
10555         (make_pass_ipa_profile): ...new function.
10556         (pass_ipa_cdtor_merge): Replace declaration with that of...
10557         (make_pass_ipa_cdtor_merge): ...new function.
10558         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
10559         of...
10560         (make_pass_cleanup_cfg_post_optimizing): ...new function.
10561         (pass_init_datastructures): Replace declaration with that of...
10562         (make_pass_init_datastructures): ...new function.
10563         (pass_fixup_cfg): Replace declaration with that of...
10564         (make_pass_fixup_cfg): ...new function.
10565         (pass_expand): Replace declaration with that of...
10566         (make_pass_expand): ...new function.
10567         (pass_instantiate_virtual_regs): Replace declaration with that of...
10568         (make_pass_instantiate_virtual_regs): ...new function.
10569         (pass_rtl_fwprop): Replace declaration with that of...
10570         (make_pass_rtl_fwprop): ...new function.
10571         (pass_rtl_fwprop_addr): Replace declaration with that of...
10572         (make_pass_rtl_fwprop_addr): ...new function.
10573         (pass_jump): Replace declaration with that of...
10574         (make_pass_jump): ...new function.
10575         (pass_jump2): Replace declaration with that of...
10576         (make_pass_jump2): ...new function.
10577         (pass_lower_subreg): Replace declaration with that of...
10578         (make_pass_lower_subreg): ...new function.
10579         (pass_cse): Replace declaration with that of...
10580         (make_pass_cse): ...new function.
10581         (pass_fast_rtl_dce): Replace declaration with that of...
10582         (make_pass_fast_rtl_dce): ...new function.
10583         (pass_ud_rtl_dce): Replace declaration with that of...
10584         (make_pass_ud_rtl_dce): ...new function.
10585         (pass_rtl_dce): Replace declaration with that of...
10586         (make_pass_rtl_dce): ...new function.
10587         (pass_rtl_dse1): Replace declaration with that of...
10588         (make_pass_rtl_dse1): ...new function.
10589         (pass_rtl_dse2): Replace declaration with that of...
10590         (make_pass_rtl_dse2): ...new function.
10591         (pass_rtl_dse3): Replace declaration with that of...
10592         (make_pass_rtl_dse3): ...new function.
10593         (pass_rtl_cprop): Replace declaration with that of...
10594         (make_pass_rtl_cprop): ...new function.
10595         (pass_rtl_pre): Replace declaration with that of...
10596         (make_pass_rtl_pre): ...new function.
10597         (pass_rtl_hoist): Replace declaration with that of...
10598         (make_pass_rtl_hoist): ...new function.
10599         (pass_rtl_store_motion): Replace declaration with that of...
10600         (make_pass_rtl_store_motion): ...new function.
10601         (pass_cse_after_global_opts): Replace declaration with that of...
10602         (make_pass_cse_after_global_opts): ...new function.
10603         (pass_rtl_ifcvt): Replace declaration with that of...
10604         (make_pass_rtl_ifcvt): ...new function.
10605         (pass_into_cfg_layout_mode): Replace declaration with that of...
10606         (make_pass_into_cfg_layout_mode): ...new function.
10607         (pass_outof_cfg_layout_mode): Replace declaration with that of...
10608         (make_pass_outof_cfg_layout_mode): ...new function.
10609         (pass_loop2): Replace declaration with that of...
10610         (make_pass_loop2): ...new function.
10611         (pass_rtl_loop_init): Replace declaration with that of...
10612         (make_pass_rtl_loop_init): ...new function.
10613         (pass_rtl_move_loop_invariants): Replace declaration with that of...
10614         (make_pass_rtl_move_loop_invariants): ...new function.
10615         (pass_rtl_unswitch): Replace declaration with that of...
10616         (make_pass_rtl_unswitch): ...new function.
10617         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
10618         (make_pass_rtl_unroll_and_peel_loops): ...new function.
10619         (pass_rtl_doloop): Replace declaration with that of...
10620         (make_pass_rtl_doloop): ...new function.
10621         (pass_rtl_loop_done): Replace declaration with that of...
10622         (make_pass_rtl_loop_done): ...new function.
10623         (pass_web): Replace declaration with that of...
10624         (make_pass_web): ...new function.
10625         (pass_cse2): Replace declaration with that of...
10626         (make_pass_cse2): ...new function.
10627         (pass_df_initialize_opt): Replace declaration with that of...
10628         (make_pass_df_initialize_opt): ...new function.
10629         (pass_df_initialize_no_opt): Replace declaration with that of...
10630         (make_pass_df_initialize_no_opt): ...new function.
10631         (pass_reginfo_init): Replace declaration with that of...
10632         (make_pass_reginfo_init): ...new function.
10633         (pass_inc_dec): Replace declaration with that of...
10634         (make_pass_inc_dec): ...new function.
10635         (pass_stack_ptr_mod): Replace declaration with that of...
10636         (make_pass_stack_ptr_mod): ...new function.
10637         (pass_initialize_regs): Replace declaration with that of...
10638         (make_pass_initialize_regs): ...new function.
10639         (pass_combine): Replace declaration with that of...
10640         (make_pass_combine): ...new function.
10641         (pass_if_after_combine): Replace declaration with that of...
10642         (make_pass_if_after_combine): ...new function.
10643         (pass_ree): Replace declaration with that of...
10644         (make_pass_ree): ...new function.
10645         (pass_partition_blocks): Replace declaration with that of...
10646         (make_pass_partition_blocks): ...new function.
10647         (pass_match_asm_constraints): Replace declaration with that of...
10648         (make_pass_match_asm_constraints): ...new function.
10649         (pass_regmove): Replace declaration with that of...
10650         (make_pass_regmove): ...new function.
10651         (pass_split_all_insns): Replace declaration with that of...
10652         (make_pass_split_all_insns): ...new function.
10653         (pass_fast_rtl_byte_dce): Replace declaration with that of...
10654         (make_pass_fast_rtl_byte_dce): ...new function.
10655         (pass_lower_subreg2): Replace declaration with that of...
10656         (make_pass_lower_subreg2): ...new function.
10657         (pass_mode_switching): Replace declaration with that of...
10658         (make_pass_mode_switching): ...new function.
10659         (pass_sms): Replace declaration with that of...
10660         (make_pass_sms): ...new function.
10661         (pass_sched): Replace declaration with that of...
10662         (make_pass_sched): ...new function.
10663         (pass_ira): Replace declaration with that of...
10664         (make_pass_ira): ...new function.
10665         (pass_reload): Replace declaration with that of...
10666         (make_pass_reload): ...new function.
10667         (pass_clean_state): Replace declaration with that of...
10668         (make_pass_clean_state): ...new function.
10669         (pass_branch_prob): Replace declaration with that of...
10670         (make_pass_branch_prob): ...new function.
10671         (pass_value_profile_transformations): Replace declaration with that
10672         of...
10673         (make_pass_value_profile_transformations): ...new function.
10674         (pass_postreload_cse): Replace declaration with that of...
10675         (make_pass_postreload_cse): ...new function.
10676         (pass_gcse2): Replace declaration with that of...
10677         (make_pass_gcse2): ...new function.
10678         (pass_split_after_reload): Replace declaration with that of...
10679         (make_pass_split_after_reload): ...new function.
10680         (pass_branch_target_load_optimize1): Replace declaration with that
10681         of...
10682         (make_pass_branch_target_load_optimize1): ...new function.
10683         (pass_thread_prologue_and_epilogue): Replace declaration with that
10684         of...
10685         (make_pass_thread_prologue_and_epilogue): ...new function.
10686         (pass_stack_adjustments): Replace declaration with that of...
10687         (make_pass_stack_adjustments): ...new function.
10688         (pass_peephole2): Replace declaration with that of...
10689         (make_pass_peephole2): ...new function.
10690         (pass_if_after_reload): Replace declaration with that of...
10691         (make_pass_if_after_reload): ...new function.
10692         (pass_regrename): Replace declaration with that of...
10693         (make_pass_regrename): ...new function.
10694         (pass_cprop_hardreg): Replace declaration with that of...
10695         (make_pass_cprop_hardreg): ...new function.
10696         (pass_reorder_blocks): Replace declaration with that of...
10697         (make_pass_reorder_blocks): ...new function.
10698         (pass_branch_target_load_optimize2): Replace declaration with that
10699         of...
10700         (make_pass_branch_target_load_optimize2): ...new function.
10701         (pass_leaf_regs): Replace declaration with that of...
10702         (make_pass_leaf_regs): ...new function.
10703         (pass_split_before_sched2): Replace declaration with that of...
10704         (make_pass_split_before_sched2): ...new function.
10705         (pass_compare_elim_after_reload): Replace declaration with that of...
10706         (make_pass_compare_elim_after_reload): ...new function.
10707         (pass_sched2): Replace declaration with that of...
10708         (make_pass_sched2): ...new function.
10709         (pass_stack_regs): Replace declaration with that of...
10710         (make_pass_stack_regs): ...new function.
10711         (pass_stack_regs_run): Replace declaration with that of...
10712         (make_pass_stack_regs_run): ...new function.
10713         (pass_df_finish): Replace declaration with that of...
10714         (make_pass_df_finish): ...new function.
10715         (pass_compute_alignments): Replace declaration with that of...
10716         (make_pass_compute_alignments): ...new function.
10717         (pass_duplicate_computed_gotos): Replace declaration with that of...
10718         (make_pass_duplicate_computed_gotos): ...new function.
10719         (pass_variable_tracking): Replace declaration with that of...
10720         (make_pass_variable_tracking): ...new function.
10721         (pass_free_cfg): Replace declaration with that of...
10722         (make_pass_free_cfg): ...new function.
10723         (pass_machine_reorg): Replace declaration with that of...
10724         (make_pass_machine_reorg): ...new function.
10725         (pass_cleanup_barriers): Replace declaration with that of...
10726         (make_pass_cleanup_barriers): ...new function.
10727         (pass_delay_slots): Replace declaration with that of...
10728         (make_pass_delay_slots): ...new function.
10729         (pass_split_for_shorten_branches): Replace declaration with that of...
10730         (make_pass_split_for_shorten_branches): ...new function.
10731         (pass_split_before_regstack): Replace declaration with that of...
10732         (make_pass_split_before_regstack): ...new function.
10733         (pass_convert_to_eh_region_ranges): Replace declaration with that
10734         of...
10735         (make_pass_convert_to_eh_region_ranges): ...new function.
10736         (pass_shorten_branches): Replace declaration with that of...
10737         (make_pass_shorten_branches): ...new function.
10738         (pass_set_nothrow_function_flags): Replace declaration with that of...
10739         (make_pass_set_nothrow_function_flags): ...new function.
10740         (pass_dwarf2_frame): Replace declaration with that of...
10741         (make_pass_dwarf2_frame): ...new function.
10742         (pass_final): Replace declaration with that of...
10743         (make_pass_final): ...new function.
10744         (pass_rtl_seqabstr): Replace declaration with that of...
10745         (make_pass_rtl_seqabstr): ...new function.
10746         (pass_release_ssa_names): Replace declaration with that of...
10747         (make_pass_release_ssa_names): ...new function.
10748         (pass_early_inline): Replace declaration with that of...
10749         (make_pass_early_inline): ...new function.
10750         (pass_inline_parameters): Replace declaration with that of...
10751         (make_pass_inline_parameters): ...new function.
10752         (pass_update_address_taken): Replace declaration with that of...
10753         (make_pass_update_address_taken): ...new function.
10754         (pass_convert_switch): Replace declaration with that of...
10755         (make_pass_convert_switch): ...new function.
10756         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
10757         to a subclass of simple_ipa_opt_pass along with...
10758         (pass_data_ipa_tree_profile): ...new pass_data instance and...
10759         (make_pass_ipa_tree_profile): ...new function.
10760         * tree-sra.c (pass_sra_early): Convert from a global struct to a
10761         subclass of gimple_opt_pass along with...
10762         (pass_data_sra_early): ...new pass_data instance and...
10763         (make_pass_sra_early): ...new function.
10764         (pass_sra): Convert from a global struct to a subclass of
10765         gimple_opt_pass along with...
10766         (pass_data_sra): ...new pass_data instance and...
10767         (make_pass_sra): ...new function.
10768         (pass_early_ipa_sra): Convert from a global struct to a subclass of
10769         gimple_opt_pass along with...
10770         (pass_data_early_ipa_sra): ...new pass_data instance and...
10771         (make_pass_early_ipa_sra): ...new function.
10772         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
10773         subclass of gimple_opt_pass along with...
10774         (pass_data_ccp): ...new pass_data instance and...
10775         (make_pass_ccp): ...new function.
10776         (pass_fold_builtins): Convert from a global struct to a subclass of
10777         gimple_opt_pass along with...
10778         (pass_data_fold_builtins): ...new pass_data instance and...
10779         (make_pass_fold_builtins): ...new function.
10780         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
10781         subclass of gimple_opt_pass along with...
10782         (pass_data_copy_prop): ...new pass_data instance and...
10783         (make_pass_copy_prop): ...new function.
10784         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
10785         global struct to a subclass of gimple_opt_pass along with...
10786         (pass_data_rename_ssa_copies): ...new pass_data instance and...
10787         (make_pass_rename_ssa_copies): ...new function.
10788         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
10789         subclass of gimple_opt_pass along with...
10790         (pass_data_dce): ...new pass_data instance and...
10791         (make_pass_dce): ...new function.
10792         (pass_dce_loop): Convert from a global struct to a subclass of
10793         gimple_opt_pass along with...
10794         (pass_data_dce_loop): ...new pass_data instance and...
10795         (make_pass_dce_loop): ...new function.
10796         (pass_cd_dce): Convert from a global struct to a subclass of
10797         gimple_opt_pass along with...
10798         (pass_data_cd_dce): ...new pass_data instance and...
10799         (make_pass_cd_dce): ...new function.
10800         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
10801         subclass of gimple_opt_pass along with...
10802         (pass_data_dominator): ...new pass_data instance and...
10803         (make_pass_dominator): ...new function.
10804         (pass_phi_only_cprop): Convert from a global struct to a subclass of
10805         gimple_opt_pass along with...
10806         (pass_data_phi_only_cprop): ...new pass_data instance and...
10807         (make_pass_phi_only_cprop): ...new function.
10808         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
10809         subclass of gimple_opt_pass along with...
10810         (pass_data_dse): ...new pass_data instance and...
10811         (make_pass_dse): ...new function.
10812         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
10813         a subclass of gimple_opt_pass along with...
10814         (pass_data_forwprop): ...new pass_data instance and...
10815         (make_pass_forwprop): ...new function.
10816         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
10817         struct to a subclass of gimple_opt_pass along with...
10818         (pass_data_tree_ifcombine): ...new pass_data instance and...
10819         (make_pass_tree_ifcombine): ...new function.
10820         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
10821         subclass of gimple_opt_pass along with...
10822         (pass_data_ch): ...new pass_data instance and...
10823         (make_pass_ch): ...new function.
10824         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
10825         subclass of gimple_opt_pass along with...
10826         (pass_data_tree_loop): ...new pass_data instance and...
10827         (make_pass_tree_loop): ...new function.
10828         (pass_tree_loop_init): Convert from a global struct to a subclass of
10829         gimple_opt_pass along with...
10830         (pass_data_tree_loop_init): ...new pass_data instance and...
10831         (make_pass_tree_loop_init): ...new function.
10832         (pass_lim): Convert from a global struct to a subclass of
10833         gimple_opt_pass along with...
10834         (pass_data_lim): ...new pass_data instance and...
10835         (make_pass_lim): ...new function.
10836         (pass_tree_unswitch): Convert from a global struct to a subclass of
10837         gimple_opt_pass along with...
10838         (pass_data_tree_unswitch): ...new pass_data instance and...
10839         (make_pass_tree_unswitch): ...new function.
10840         (pass_predcom): Convert from a global struct to a subclass of
10841         gimple_opt_pass along with...
10842         (pass_data_predcom): ...new pass_data instance and...
10843         (make_pass_predcom): ...new function.
10844         (pass_vectorize): Convert from a global struct to a subclass of
10845         gimple_opt_pass along with...
10846         (pass_data_vectorize): ...new pass_data instance and...
10847         (make_pass_vectorize): ...new function.
10848         (pass_graphite): Convert from a global struct to a subclass of
10849         gimple_opt_pass along with...
10850         (pass_data_graphite): ...new pass_data instance and...
10851         (make_pass_graphite): ...new function.
10852         (pass_graphite_transforms): Convert from a global struct to a subclass
10853         of gimple_opt_pass along with...
10854         (pass_data_graphite_transforms): ...new pass_data instance and...
10855         (make_pass_graphite_transforms): ...new function.
10856         (pass_check_data_deps): Convert from a global struct to a subclass of
10857         gimple_opt_pass along with...
10858         (pass_data_check_data_deps): ...new pass_data instance and...
10859         (make_pass_check_data_deps): ...new function.
10860         (pass_iv_canon): Convert from a global struct to a subclass of
10861         gimple_opt_pass along with...
10862         (pass_data_iv_canon): ...new pass_data instance and...
10863         (make_pass_iv_canon): ...new function.
10864         (pass_scev_cprop): Convert from a global struct to a subclass of
10865         gimple_opt_pass along with...
10866         (pass_data_scev_cprop): ...new pass_data instance and...
10867         (make_pass_scev_cprop): ...new function.
10868         (pass_record_bounds): Convert from a global struct to a subclass of
10869         gimple_opt_pass along with...
10870         (pass_data_record_bounds): ...new pass_data instance and...
10871         (make_pass_record_bounds): ...new function.
10872         (pass_complete_unroll): Convert from a global struct to a subclass of
10873         gimple_opt_pass along with...
10874         (pass_data_complete_unroll): ...new pass_data instance and...
10875         (make_pass_complete_unroll): ...new function.
10876         (pass_complete_unrolli): Convert from a global struct to a subclass of
10877         gimple_opt_pass along with...
10878         (pass_data_complete_unrolli): ...new pass_data instance and...
10879         (make_pass_complete_unrolli): ...new function.
10880         (pass_parallelize_loops): Convert from a global struct to a subclass
10881         of gimple_opt_pass along with...
10882         (pass_data_parallelize_loops): ...new pass_data instance and...
10883         (make_pass_parallelize_loops): ...new function.
10884         (pass_loop_prefetch): Convert from a global struct to a subclass of
10885         gimple_opt_pass along with...
10886         (pass_data_loop_prefetch): ...new pass_data instance and...
10887         (make_pass_loop_prefetch): ...new function.
10888         (pass_iv_optimize): Convert from a global struct to a subclass of
10889         gimple_opt_pass along with...
10890         (pass_data_iv_optimize): ...new pass_data instance and...
10891         (make_pass_iv_optimize): ...new function.
10892         (pass_tree_loop_done): Convert from a global struct to a subclass of
10893         gimple_opt_pass along with...
10894         (pass_data_tree_loop_done): ...new pass_data instance and...
10895         (make_pass_tree_loop_done): ...new function.
10896         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
10897         struct to a subclass of gimple_opt_pass along with...
10898         (pass_data_cse_reciprocals): ...new pass_data instance and...
10899         (make_pass_cse_reciprocals): ...new function.
10900         (pass_cse_sincos): Convert from a global struct to a subclass of
10901         gimple_opt_pass along with...
10902         (pass_data_cse_sincos): ...new pass_data instance and...
10903         (make_pass_cse_sincos): ...new function.
10904         (pass_optimize_bswap): Convert from a global struct to a subclass of
10905         gimple_opt_pass along with...
10906         (pass_data_optimize_bswap): ...new pass_data instance and...
10907         (make_pass_optimize_bswap): ...new function.
10908         (pass_optimize_widening_mul): Convert from a global struct to a
10909         subclass of gimple_opt_pass along with...
10910         (pass_data_optimize_widening_mul): ...new pass_data instance and...
10911         (make_pass_optimize_widening_mul): ...new function.
10912         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
10913         subclass of gimple_opt_pass along with...
10914         (pass_data_phiopt): ...new pass_data instance and...
10915         (make_pass_phiopt): ...new function.
10916         (pass_cselim): Convert from a global struct to a subclass of
10917         gimple_opt_pass along with...
10918         (pass_data_cselim): ...new pass_data instance and...
10919         (make_pass_cselim): ...new function.
10920         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
10921         subclass of gimple_opt_pass along with...
10922         (pass_data_phiprop): ...new pass_data instance and...
10923         (make_pass_phiprop): ...new function.
10924         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
10925         subclass of gimple_opt_pass along with...
10926         (pass_data_pre): ...new pass_data instance and...
10927         (make_pass_pre): ...new function.
10928         (pass_fre): Convert from a global struct to a subclass of
10929         gimple_opt_pass along with...
10930         (pass_data_fre): ...new pass_data instance and...
10931         (make_pass_fre): ...new function.
10932         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
10933         subclass of gimple_opt_pass along with...
10934         (pass_data_reassoc): ...new pass_data instance and...
10935         (make_pass_reassoc): ...new function.
10936         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
10937         subclass of gimple_opt_pass along with...
10938         (pass_data_sink_code): ...new pass_data instance and...
10939         (make_pass_sink_code): ...new function.
10940         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
10941         subclass of gimple_opt_pass along with...
10942         (pass_data_strlen): ...new pass_data instance and...
10943         (make_pass_strlen): ...new function.
10944         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
10945         struct to a subclass of gimple_opt_pass along with...
10946         (pass_data_build_alias): ...new pass_data instance and...
10947         (make_pass_build_alias): ...new function.
10948         (pass_build_ealias): Convert from a global struct to a subclass of
10949         gimple_opt_pass along with...
10950         (pass_data_build_ealias): ...new pass_data instance and...
10951         (make_pass_build_ealias): ...new function.
10952         (pass_ipa_pta): Convert from a global struct to a subclass of
10953         simple_ipa_opt_pass along with...
10954         (pass_data_ipa_pta): ...new pass_data instance and...
10955         (make_pass_ipa_pta): ...new function.
10956         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
10957         subclass of gimple_opt_pass along with...
10958         (pass_data_uncprop): ...new pass_data instance and...
10959         (make_pass_uncprop): ...new function.
10960         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
10961         global struct to a subclass of gimple_opt_pass along with...
10962         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
10963         (make_pass_late_warn_uninitialized): ...new function.
10964         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
10965         to a subclass of gimple_opt_pass along with...
10966         (pass_data_init_datastructures): ...new pass_data instance and...
10967         (make_pass_init_datastructures): ...new function.
10968         (pass_early_warn_uninitialized): Convert from a global struct to a
10969         subclass of gimple_opt_pass along with...
10970         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
10971         (make_pass_early_warn_uninitialized): ...new function.
10972         (pass_update_address_taken): Convert from a global struct to a
10973         subclass of gimple_opt_pass along with...
10974         (pass_data_update_address_taken): ...new pass_data instance and...
10975         (make_pass_update_address_taken): ...new function.
10976         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
10977         struct to a subclass of gimple_opt_pass along with...
10978         (pass_data_release_ssa_names): ...new pass_data instance and...
10979         (make_pass_release_ssa_names): ...new function.
10980         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
10981         subclass of gimple_opt_pass along with...
10982         (pass_data_stdarg): ...new pass_data instance and...
10983         (make_pass_stdarg): ...new function.
10984         * tree-switch-conversion.c (pass_convert_switch): Convert from a
10985         global struct to a subclass of gimple_opt_pass along with...
10986         (pass_data_convert_switch): ...new pass_data instance and...
10987         (make_pass_convert_switch): ...new function.
10988         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
10989         to a subclass of gimple_opt_pass along with...
10990         (pass_data_tail_recursion): ...new pass_data instance and...
10991         (make_pass_tail_recursion): ...new function.
10992         (pass_tail_calls): Convert from a global struct to a subclass of
10993         gimple_opt_pass along with...
10994         (pass_data_tail_calls): ...new pass_data instance and...
10995         (make_pass_tail_calls): ...new function.
10996         * tree-vect-generic.c (pass_lower_vector): Convert from a global
10997         struct to a subclass of gimple_opt_pass along with...
10998         (pass_data_lower_vector): ...new pass_data instance and...
10999         (make_pass_lower_vector): ...new function.
11000         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
11001         gimple_opt_pass along with...
11002         (pass_data_lower_vector_ssa): ...new pass_data instance and...
11003         (make_pass_lower_vector_ssa): ...new function.
11004         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
11005         to a subclass of gimple_opt_pass along with...
11006         (pass_data_slp_vectorize): ...new pass_data instance and...
11007         (make_pass_slp_vectorize): ...new function.
11008         (pass_ipa_increase_alignment): Convert from a global struct to a
11009         subclass of simple_ipa_opt_pass along with...
11010         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
11011         (make_pass_ipa_increase_alignment): ...new function.
11012         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
11013         gimple_opt_pass along with...
11014         (pass_data_vrp): ...new pass_data instance and...
11015         (make_pass_vrp): ...new function.
11016         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
11017         subclass of simple_ipa_opt_pass along with...
11018         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
11019         (make_pass_ipa_free_lang_data): ...new function.
11020         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
11021         gimple_opt_pass along with...
11022         (pass_data_tsan): ...new pass_data instance and...
11023         (make_pass_tsan): ...new function.
11024         (pass_tsan_O0): Convert from a global struct to a subclass of
11025         gimple_opt_pass along with...
11026         (pass_data_tsan_O0): ...new pass_data instance and...
11027         (make_pass_tsan_O0): ...new function.
11028         * var-tracking.c (pass_variable_tracking): Convert from a global
11029         struct to a subclass of rtl_opt_pass along with...
11030         (pass_data_variable_tracking): ...new pass_data instance and...
11031         (make_pass_variable_tracking): ...new function.
11032         * web.c (pass_web): Convert from a global struct to a subclass of
11033         rtl_opt_pass along with...
11034         (pass_data_web): ...new pass_data instance and...
11035         (make_pass_web): ...new function.
11036         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
11037         declaration with that of...
11038         (make_pass_mode_switch_use): ...new function.
11039         (pass_resolve_sw_modes): Replace declaration with that of...
11040         (make_pass_resolve_sw_modes): ...new function.
11041         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
11042         from a global struct to a subclass of rtl_opt_pass along with...
11043         (pass_data_mode_switch_use): ...new pass_data instance and...
11044         (make_pass_mode_switch_use): ...new function.
11045         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
11046         from a global struct to a subclass of rtl_opt_pass along with...
11047         (pass_data_resolve_sw_modes): ...new pass_data instance and...
11048         (make_pass_resolve_sw_modes): ...new function.
11049         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
11050         struct to a subclass of rtl_opt_pass along with...
11051         (pass_data_insert_vzeroupper): ...new pass_data instance and...
11052         (make_pass_insert_vzeroupper): ...new function.
11053         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
11054         global struct to a subclass of rtl_opt_pass along with...
11055         (pass_data_work_around_errata): ...new pass_data instance and...
11056         (make_pass_work_around_errata): ...new function.
11057         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
11058         struct to a subclass of rtl_opt_pass along with...
11059         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
11060         (make_pass_mips_machine_reorg2): ...new function.
11062 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11064         * passes.c (pass_manager::operator new): New.
11066 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
11068         Handwritten part of conversion of passes to C++ classes.
11070         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
11071         (toplev.o): Add dep on PASS_MANAGER_H.
11072         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
11073         of early local pases to reflect this moving from a global to a
11074         member of gcc::pass_manager.
11075         (cgraph_add_new_function): Likewise.
11076         * lto-cgraph.c (lto_output_node): Update for conversion of
11077         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
11078         * passes.c (opt_pass::clone): New.
11079         (opt_pass::gate): New.
11080         (opt_pass::execute): New.
11081         (opt_pass::opt_pass): New.
11082         (pass_manager::execute_early_local_passes): New.
11083         (pass_manager::execute_pass_mode_switching): new.
11084         (finish_optimization_passes): Convert to...
11085         (pass_manager::finish_optimization_passes): ...this.
11086         (finish_optimization_passes): Update for conversion of passes to
11087         C++ classes.
11088         (register_dump_files_1): Use has_gate since we cannot portably
11089         check a vtable entry against NULL.
11090         (dump_one_pass): Likewise.
11091         (ipa_write_summaries_2): Likewise.
11092         (ipa_write_optimization_summaries_1): Likewise.
11093         (ipa_read_summaries_1): Likewise.
11094         (ipa_read_optimization_summaries_1): Likewise.
11095         (execute_ipa_stmt_fixups): Likewise.
11096         (pass_manager::pass_manager): Rewrite pass-creation, invoking
11097         pass-creation functions rather than wiring up globals, and
11098         storing the results in fields of pass_manager generated using
11099         pass-instances.def.
11100         (pass_manager::dump_profile_report): Update for conversion of
11101         passes to C++ classes.
11102         (pass_manager::execute_ipa_summary_passes): Likewise.
11103         (execute_one_ipa_transform_pass): Likewise.
11104         (execute_one_pass): Use has_gate and has_execute since we cannot
11105         portably check a vtable entry against NULL.
11106         * pass_manager.h (pass_manager::finish_optimization_passes): New.
11107         (pass_manager): Use pass-instances.def to add fields for the
11108         various pass instances.
11109         * toplev.c (finalize): Update for move of
11110         finish_optimization_passes to a method of gcc::pass_manager.
11111         * toplev.h (finish_optimization_passes): Move to method of class
11112         pass_manager.
11113         * tree-pass.h (struct pass_data): New.
11114         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
11115         (opt_pass::gate): Convert to virtual function.
11116         (opt_pass::~opt_pass): New.
11117         (opt_pass::clone): New.
11118         (opt_pass::execute): Convert to virtual function.
11119         (opt_pass::opt_pass): New.
11120         (opt_pass::ctxt_): new.
11121         (gimple_opt_pass): Convert to subclass of opt_pass.
11122         (gimple_opt_pass::gimple_opt_pass): New.
11123         (rtl_opt_pass): Convert to subclass of opt_pass.
11124         (rtl_opt_pass::rtl_opt_pass): New.
11125         (ipa_opt_pass_d): Convert to subclass of opt_pass.
11126         (ipa_opt_pass_d::ipa_opt_pass_d): New.
11127         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
11128         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
11129         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
11130         invocation of pass_mode_switching to reflect this moving from a
11131         global to a member of gcc::pass_manager.
11132         (ix86_option_override): Rework how pass_insert_vzeroupper is
11133         added to the pass_manager to reflect autogenerated changes.
11134         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
11136 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
11138         PR rtl-optimization/57708
11139         * recog.c (peep2_find_free_register): Validate all regs in a
11140         multi-reg mode.
11142 2013-08-05  Jan Hubicka  <jh@suse.cz>
11144         PR lto/57602
11145         * cgraph.c (verify_cgraph_node): Accept local flags from other
11146         partitions.
11147         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
11148         (function_and_variable_visibility): Likewise.
11149         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
11151 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11153         * graph.c (init_graph_slim_pretty_print): Remove.
11154         (print_graph_cfg): Do not call it.  Use local pretty printer.
11155         (start_graph_dump): Likewise.
11157 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11159         * gimple-pretty-print.c (buffer): Remove.
11160         (initialized): Likewise.
11161         (maybe_init_pretty_print): Likewise.
11162         (print_gimple_stmt): Do not call it.  Use non-static local
11163         pretty_printer variable.
11164         (print_gimple_expr): Likewise.
11165         (print_gimple_seq): Likewise.
11166         (gimple_dump_bb): Likewise.
11168 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11170         * asan.c (asan_pp): Remove.
11171         (asan_pp_initialized): Likewise.
11172         (asan_pp_initialize): Likewise.
11173         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
11174         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
11175         (asan_add_global): Likewise.
11177 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11179         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
11180         * diagnostic.h (diagnostic_flush_buffer): Adjust.
11181         * pretty-print.c (pp_formatted_text_data): Likewise.
11182         (pp_indent): Rename from pp_base_indent.
11183         (pp_format): Rename from pp_base_format.
11184         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
11185         (pp_format_verbatim): Rename from pp_base_format_verbatim.
11186         (pp_flush): Rename from pp_base_flush.
11187         (pp_set_line_maximum_length): Rename from
11188         pp_base_set_line_maximum_length.
11189         (pp_clear_output_area): Rename from pp_base_clear_output_area.
11190         (pp_set_prefix): Rename from pp_base_set_prefix.
11191         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
11192         (pp_emit_prefix): Rename from pp_base_emit_prefix.
11193         (pp_append_text): Rename from pp_base_append_text.
11194         (pp_formatted_text): Rename from pp_base_formatted_text.
11195         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
11196         (pp_remaining_character_count_for_line): Rename from
11197         pp_base_remaining_character_count_for_line.
11198         (pp_newline): Rename from pp_base_newline.
11199         (pp_character): Rename from pp_base_character.
11200         (pp_string): Rename from pp_base_string.
11201         (pp_maybe_space): Rename from pp_base_maybe_space.
11202         * asan.c (asan_pp_string): Adjust.
11203         (asan_emit_stack_protection): Likewise.
11204         (asan_add_global): Likewise.
11205         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
11206         * tree-mudflap.c (mf_varname_tree): Likewise.
11207         * tree-pretty-print.c (pp_tree_identifier): Rename from
11208         pp_base_tree_identifier.
11209         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
11210         Declare as function.
11212 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11214         * pretty-print.h (pp_bar_bar): New.
11215         (pp_ampersand_ampersand): Likewise.
11216         (pp_less_equal): Likewise.
11217         (pp_greater_equal): Likewise.
11218         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
11219         printer functions instead of pp_string or operators and punctuators.
11220         (dump_gimple_call): Likewise.
11221         (dump_gimple_omp_for): Likewise.
11222         (dump_gimple_transaction): Likewise.
11223         (dump_gimple_phi): Likewise.
11224         (pp_gimple_stmt_1): Likewise.
11225         * sched-vis.c (print_insn): Likewise.
11226         * tree-mudflap.c (mf_varname_tree): Likewise.
11227         * tree-pretty-print.c (dump_block_node): Likewise.
11228         (dump_generic_node): Likewise.
11230 2013-08-02  Jan Hubicka  <jh@suse.cz>
11232         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
11233         boundaries.
11234         * lto-streamer-out.c (tree_is_indexable): Results decls and
11235         parm decls are not indexable.
11236         (DFS_write_tree_body): Do not follow args and results.
11237         (hash_tree): Likewise.
11238         (output_functions): Rearrange so struct function is needed
11239         only when real body is output; be able to also ouptut abstract
11240         functions; output DECL_ARGUMENTS and DECL_RESULT.
11241         (lto_output): When not in WPA, ale store abstract functions.
11242         (write_symbol): Do not care about RESULT_DECL.
11243         (output_symbol_p): Handle correctly sbtract decls.
11244         * lto-streamer-in.c (input_function): Rearrange so struct
11245         function can be NULL at entry; allow streaming of
11246         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
11247         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
11248         sanity check during LTO.
11249         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
11250         RESULT_DECl and DECL_ARGUMENTS.
11251         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
11252         Likewise.
11254 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11256         * pretty-print.h (pp_underscore): New.
11257         (pp_comma): Tidy.
11258         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
11259         printer functions instead of pp_character.
11260         (dump_binary_rhs): Likewise.
11261         (dump_ternary_rhs): Likewise.
11262         (dump_gimple_call_args): Likewise.
11263         (pp_points_to_solution): Likewise.
11264         (dump_gimple_call): Likewise.
11265         (dump_gimple_switch): Likewise.
11266         (dump_gimple_cond): Likewise.
11267         (dump_gimple_bind): Likewise.
11268         (dump_gimple_try): Likewise.
11269         (dump_gimple_omp_for): Likewise.
11270         (dump_gimple_omp_continue): Likewise.
11271         (dump_gimple_omp_single): Likewise.
11272         (dump_gimple_omp_sections): Likewise.
11273         (dump_gimple_omp_block): Likewise.
11274         (dump_gimple_omp_critical): Likewise.
11275         (dump_gimple_transaction): Likewise.
11276         (dump_gimple_asm): Likewise.
11277         (dump_gimple_phi): Likewise.
11278         (dump_gimple_omp_parallel): Likewise.
11279         (dump_gimple_omp_task): Likewise.
11280         (dump_gimple_omp_atomic_load): Likewise.
11281         (dump_gimple_omp_atomic_store): Likewise.
11282         (dump_gimple_mem_ops): Likewise.
11283         (pp_gimple_stmt_1): Likewise.
11284         (pp_cfg_jump): Likewise.
11285         (dump_implicit_edges): Likewise.
11286         (gimple_dump_bb_for_graph): Likewise.
11287         * graph.c (draw_cfg_node): Likewise.
11288         * langhooks.c (lhd_print_error_function): Likewise.
11289         * sched-vis.c (print_exp): Likewise.
11290         (print_value): Likewise.
11291         (print_pattern): Likewise.
11292         (print_insn): Likewise.
11293         (rtl_dump_bb_for_graph): Likewise.
11294         * tree-pretty-print.c (dump_function_declaration): Likewise.
11295         (dump_array_domain): Likewise.
11296         (dump_omp_clause): Likewise.
11297         (dump_location): Likewise.
11298         (dump_generic_node): Likewise.
11299         (print_struct_decl): Likewise.
11300         * diagnostic.c (diagnostic_show_locus): Use pp_space.
11302 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11304         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
11305         candidate table when replacing a candidate statement.
11306         (replace_rhs_if_not_dup): Likewise.
11307         (replace_one_candidate): Likewise.
11309 2013-08-02  Jan Hubicka  <jh@suse.cz>
11310             Martin Liska  <marxin.liska@gmail.com>
11312         * cgraphunit.c (add_new_function): Fix logic when adding from
11313         late IPA pass.
11314         (assemble_thunk): Rename to ...
11315         (expand_thunk); .. this one; export; get it working with
11316         general functions; make produced gimple valid.
11317         * cgraph.h (expand_thunk): Declare.
11319 2013-08-02  Jan Hubicka  <jh@suse.cz>
11321         * ipa-cp.c (gather_context_independent_values): Use
11322         ipa_get_param_move_cost.
11323         (get_replacement_map): Remove PARAM; move parameter folding
11324         into tree-inline.c
11325         (create_specialized_node): Update.
11326         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
11327         assert that we have gimple body; update move_cost.
11328         (count_formal_params): Assert that we have gimple body.
11329         (ipa_dump_param): New function.
11330         (ipa_alloc_node_params): Break out from ...
11331         (ipa_initialize_node_params): ... here.
11332         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
11333         (ipa_write_node_info): Stream move costs.
11334         (ipa_read_node_info): Read move costs.
11335         (ipa_update_after_lto_read): Do not recompute node params.
11336         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
11337         (ipa_get_param): Check we are not in WPA.
11338         (ipa_get_param_move_cost): New.
11339         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
11340         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
11341         parm numbers to be present.
11343 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11345         PR rtl-optimization/58048
11346         * lra-constraints.c (process_alt_operands): Don't check asm
11347         operand on register.
11349 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
11351         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
11352         the implied StoreLoad barrier for atomic operations if before.
11354 2013-08-02  Jan Hubicka  <jh@suse.cz>
11355             Martin Liska  <marxin.liska@gmail.com>
11357         * cgraph.c (cgraph_function_body_availability): Do not check
11358         cgraph flags.
11359         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
11360         symtab_node_availability): Declare.
11361         * ipa.c (can_replace_by_local_alias): New.
11362         (function_and_variable_visibility): Use it.
11363         * symtab.c (symtab_for_node_and_aliases,
11364         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
11366 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
11368         PR rtl-optimization/57963
11369         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
11370         (lra_constraints): Use them.
11372 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
11374         * config/arm/types.md (define_attr "type"): Add "load_acq"
11375         and "store_rel".
11376         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
11377         changes.
11378         (cortex_a53_store1): Likewise.
11380 2013-08-01  Jan Hubicka  <jh@suse.cz>
11382         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
11383         partitions are not needed.
11385 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
11387         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
11388         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
11389         MAYBE_NON_Q_CLASS_P where appropriate.
11391 2013-08-01  Jan Hubicka  <jh@suse.cz>
11393         * cgraph.h (release_function_body): Declare.
11394         * tree.c (free_lang_data_in_decl): Free, parameters and return values
11395         of unused delcarations.
11397 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11399         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
11400         RTL form when subtracting a constant.
11402 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11404         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
11405         Generate canonical plus rtx with negated immediate instead of minus
11406         where appropriate.
11407         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
11409 2013-08-01  Jan Hubicka  <jh@suse.cz>
11411         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
11412         (cgraph_release_function_body): Likewise.
11413         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
11414         * cgraph.h (cgrpah_node): Rename abstract_and_needed
11415         to used_as_abstract_origin.
11416         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
11417         symbols used as abstract origins.
11418         * cgraphunit.c (analyze_functions): Update.
11419         * ipa.c (symtab_remove_unreachable_nodes): Recompute
11420         used_as_abstract_origin.
11421         * tree-inline.c (tree_function_versioning): Update
11422         used_as_abstract_origin; be ready for DECL_RESULT and
11423         DECL_ARGUMENTS to be NULL.
11425         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
11426         for abstract functions.
11427         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
11428         real symbols.
11430 2013-08-01  Jan Hubicka  <jh@suse.cz>
11432         * profile.c (compute_value_histograms): Fix thinko.
11434 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11436         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
11437         aarch-common-protos.h to extra_headers.
11438         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11439         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
11440         * config/aarch64/t-aarch64 (aarch-common.o): Define.
11442 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
11444         * config/aarch64/aarch64.md (define_attr "type"): Delete.
11445         Include "../arm/types.md".  Define "type" attribute for all patterns.
11446         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
11447         attribute changes.
11449 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
11451         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
11452         to support power8 load fusion.
11453         (fusion_gpr_mem_load): Likewise.
11455         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
11457         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
11458         declarations for power8 load fusion.
11459         (emit_fusion_gpr_load): Likewise.
11461         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
11462         tuning for power8, turn on fusion mode by default.  Turn on sign
11463         extending fusion mode if normal fusion mode is on, and we are at
11464         -O2 or -O3.
11465         (fusion_gpr_load_p): New function, return true if we can fuse an
11466         addis instruction with a dependent load to a GPR.
11467         (emit_fusion_gpr_load): Emit the instructions for power8 load
11468         fusion to GPRs.
11470         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
11471         (VSX load fusion peepholes): New peepholes to fuse together an
11472         addi instruction with a VSX load instruction.
11474         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
11475         peepholes to fuse an addis instruction with a load to a GPR base
11476         register.  If we are supporting sign extending fusions, convert
11477         sign extending loads to zero extending loads and add an explicit
11478         sign extension.
11480 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11482         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
11483         aarch-common-protos.h to extra_headers.
11484         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
11485         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
11486         (arm_early_store_addr_dep): Likewise.
11487         (arm_no_early_alu_shift_dep): Likewise.
11488         (arm_no_early_alu_shift_value_dep): Likewise.
11489         (arm_no_early_mul_dep): Likewise.
11490         (arm_no_early_store_addr_dep): Likewise.
11491         (arm_mac_accumulator_is_mul_result): Likewise.
11492         (arm_mac_accumulator_is_result): Likewise.
11493         * config/arm/aarch-common.c: ... here.  New file.
11494         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
11495         here to ...
11496         (arm_early_store_addr_dep): Likewise.
11497         (arm_no_early_alu_shift_dep): Likewise.
11498         (arm_no_early_alu_shift_value_dep): Likewise.
11499         (arm_no_early_mul_dep): Likewise.
11500         (arm_no_early_store_addr_dep): Likewise.
11501         (arm_mac_accumulator_is_mul_result): Likewise.
11502         (arm_mac_accumulator_is_result): Likewise.
11503         * config/arm/aarch-common-protos.h: ... here.  New file.
11504         * config/arm/t-arm (aarch-common.o): Define.
11506 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
11508         * config/arm/arm.md: Include new file "types.md".
11509         (define_attr "type"): Move from here to ...
11510         (define_attr "mul32"): Likewise.
11511         (define_attr "mul64"): Likewise.
11512         * config/arm/types.md: ... here.  New file.
11514 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
11516         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
11517         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
11519 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
11521         * gen-pass-instances.awk: Fix offset of substr().
11523 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
11525         * Makefile.in (pass-instances.def): New.
11526         (passes.o): Replace dependency on passes.def with one on
11527         pass-instances.def
11529         * gen-pass-instances.awk: New.
11531         * passes.c (pass_manager::pass_manager): Use pass-instances.def
11532         rather than passes.def, updating local definition of NEXT_PASS
11533         macro to add an extra NUM parameter (currently unused).
11535 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
11537         * Makefile.in (PASS_MANAGER_H): New.
11538         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
11539         (passes.o): Likewise.
11540         (statistics.o): Likewise.
11541         (cgraphunit.o): Likewise.
11542         (context.o): Depend on PASS_MANAGER_H.
11544         * pass_manager.h: New.
11546         * cgraphunit.c (cgraph_add_new_function): Update for moves
11547         of globals to fields of pass_manager.
11548         (analyze_function): Likewise.
11549         (expand_function): Likewise.
11550         (ipa_passes): Likewise.
11551         (compile): Likewise.
11553         * context.c (context::context): New.
11554         * context.h  (context::context): New.
11555         (context::get_passes): New.
11556         (context::passes_): New.
11558         * lto-cgraph.c (input_node): Update for moves of globals to
11559         fields of pass_manager.
11561         * passes.c (all_passes): Remove, in favor of a field of the
11562         same name within the new class pass_manager.
11563         (all_small_ipa_passes): Likewise.
11564         (all_lowering_passes): Likewise.
11565         (all_regular_ipa_passes): Likewise.
11566         (all_late_ipa_passes): Likewise.
11567         (all_lto_gen_passes): Likewise.
11568         (passes_by_id): Likewise.
11569         (passes_by_id_size): Likewise.
11570         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11571         the new class pass_manager.
11572         (set_pass_for_id): Convert to...
11573         (pass_manager::set_pass_for_id): ...method.
11574         (get_pass_for_id): Convert to...
11575         (pass_manager::get_pass_for_id): ...method.
11576         (register_one_dump_file): Move body of implementation into...
11577         (pass_manager::register_one_dump_file): ...here.
11578         (register_dump_files_1): Convert to...
11579         (pass_manager::register_dump_files_1): ...method.
11580         (register_dump_files): Convert to...
11581         (pass_manager::register_dump_files): ...method.
11582         (create_pass_tab): Update for moves of globals to fields of
11583         pass_manager.
11584         (dump_passes): Move body of implementation into...
11585         (pass_manager::dump_passes): ...here.
11586         (register_pass): Move body of implementation into...
11587         (pass_manager::register_pass): ...here.
11588         (init_optimization_passes): Convert into...
11589         (pass_manager::pass_manager): ...constructor for new
11590         pass_manager class, and initialize the pass_lists array.
11591         (check_profile_consistency): Update for moves of globals to
11592         fields of pass_manager.
11593         (dump_profile_report): Move body of implementation into...
11594         (pass_manager::dump_profile_report): ...here.
11595         (ipa_write_summaries_1): Update for moves of pass lists from
11596         being globals to fields of pass_manager.
11597         (ipa_write_optimization_summaries): Likewise.
11598         (ipa_read_summaries):  Likewise.
11599         (ipa_read_optimization_summaries): Likewise.
11600         (execute_all_ipa_stmt_fixups): Likewise.
11602         * statistics.c (statistics_fini): Update for moves of globals to
11603         fields of pass_manager.
11605         * toplev.c (general_init): Replace call to
11606         init_optimization_passes with construction of the pass_manager
11607         instance.
11609         * tree-pass.h (all_passes): Remove, in favor of a field of the
11610         same name within the new class pass_manager.
11611         (all_small_ipa_passes): Likewise.
11612         (all_lowering_passes): Likewise.
11613         (all_regular_ipa_passes): Likewise.
11614         (all_lto_gen_passes): Likewise.
11615         (all_late_ipa_passes): Likewise.
11616         (passes_by_id): Likewise.
11617         (passes_by_id_size): Likewise.
11618         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
11619         the new class pass_manager.
11620         (get_pass_for_id): Remove.
11622 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11624         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
11625         configs.
11627 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
11629         * arm.md (mulhi3): New expand pattern.
11631 2013-07-30  Jan Hubicka  <jh@suse.cz>
11632             Martin Liska  <marxin.liska@gmail.com>
11634         * profile.c (compute_value_histograms): Do not ICE when
11635         there is mismatch only on some counters.
11637 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
11639         PR rtl-optimization/57637
11640         * function.c (move_insn_for_shrink_wrap): Also check the
11641         GEN set of the LIVE problem for the liveness analysis
11642         if it exists, otherwise give up.
11644 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11646         PR tree-optimization/57993
11647         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
11648         replaced statement in the candidate table.
11649         (phi_add_costs): Return infinite cost when the hidden basis does
11650         not dominate all phis on which the candidate is dependent.
11651         (replace_one_candidate): Record replaced statement in the
11652         candidate table.
11654 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
11656         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
11657         (ashlv2si3): New expander.
11658         (*ashlv2si3_i): New define_insn_and_split.
11659         * predicates.md (float_operation): Allow patterns with three
11660         basic sub-patterns.
11662         PR rtl-optimization/58021
11663         * mode-switching.c (create_pre_exit): Always split off preceding
11664         insns if we are not at the basic block head.
11666 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
11668         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
11669         (UCLIBC_DYNAMIC_LINKER): New macro.
11670         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
11671         `-mnan=2008'.
11672         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
11673         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
11674         `-mnan=2008'.
11675         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
11676         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
11677         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
11678         for SF and DF modes.  Use ieee_quad_format for TF mode.
11679         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
11680         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
11681         (mips_option_override): Handle `-mnan=legacy'.
11682         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
11683         `-mabs=2008' and `-mnan=2008'.
11684         (OPTION_DEFAULT_SPECS): Add "nan" default.
11685         (ASM_SPEC): Handle `-mnan='.
11686         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
11687         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
11688         comment accordingly.
11689         (neg<mode>2): Likewise.
11690         * config/mips/mips.opt (mabs, mnan): New options.
11691         * doc/install.texi (Configuration): Document `--with-nan=' option.
11692         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
11693         `-mnan=' options.
11694         (MIPS Options): Document them.
11695         * config.gcc <mips*-*-*>: Handle `--with-nan='.
11696         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
11697         * configure: Regenerate.
11698         * config.in: Regenerate.
11700 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11702         * config/i386/i386.md (float post-reload splitters): Do not check
11703         for subregs of SSE registers.
11705 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
11706             H.J. Lu  <hongjiu.lu@intel.com>
11708         PR target/57954
11709         PR target/57988
11710         * config/i386/i386.md (post-reload splitter
11711         to avoid partial SSE reg dependency stalls): New pattern.
11713 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
11715         * config/s390/s390.md ("movcc"): Swap load and store instructions.
11717 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
11719         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
11720         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
11722 2013-07-26  Cary Coutant  <ccoutant@google.com>
11724         * dwarf2out.c (die_checksum_ordered): Don't include template
11725         instantiations in signature.
11726         (is_template_parameter): New function.
11727         (is_template_instantiation): New function.
11728         (generate_skeleton_bottom_up): Don't include template instantiations
11729         in type unit DIE.
11730         (generate_skeleton): Likewise.
11731         (break_out_comdat_types): Move recursive call to break out nested
11732         types earlier.
11733         (prune_unused_types_mark_generic_parms_dies): Call
11734         is_template_parameter.
11736 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
11738         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
11739         uses vector registers.
11740         * config/aarch64/iterators.md: Add attributes rtn and vas.
11742 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11743             Richard Earnshaw  <richard.earnshaw@arm.com>
11745         * combine.c (simplify_comparison): Re-canonicalize operands
11746         where appropriate.
11747         * config/arm/arm.md (movcond_addsi): New splitter.
11749 2013-07-25  Sterling Augustine  <saugustine@google.com>
11751         * dwarf2out.c (size_of_pubnames): Move code to...
11752         (include_pubname_in_output): ...here.  New.
11753         (want_pubnames): Rearrange.
11754         (output_pubnames): Call include_pubname_in_output.  Move assertion.
11756 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11758         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
11760 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
11762         PR target/38836
11763         * doc/extend.texi: Remove obsolete builtins. Fix
11764         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
11766 2013-07-25  Jan Hubicka  <jh@suse.cz>
11768         * cgraph.c (release_function_body): Break out from ...
11769         (cgraph_release_function_body): ... this one; also release DECL_RESULT
11770         and DECL_ARGUMENTS.
11771         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
11772         old_tree in the map.
11773         (create_specialized_node): Update.
11774         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
11775         into index.
11776         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
11777         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
11778         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
11779         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
11780         DECL_RESULT.
11782 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11784         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
11785         addsi3_carryin_alt2_<optab>): Correct output template.
11787 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
11789         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
11790         Adjust for arm_restrict_it.
11791         Remove trailing whitespace.
11793 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
11795         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
11796         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
11798         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
11800 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
11802         PR rtl-optimization/57960
11803         * lra-constraints.c (process_alt_operands): Use the right mode
11804         when checking strict_low.
11806 2013-07-25  Jan Hubicka  <jh@suse.cz>
11808         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
11809         * cgraph.c (cgraph_remove_node): Do not release function body
11810         when in cgraph streaming.
11811         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
11812         in other partitions are not considered reachable; fix handling of
11813         clones.
11815 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11817         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
11819 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
11821         PR target/19599
11822         PR target/57731
11823         PR target/57837
11824         * config/arm/arm.md ("*sibcall_insn): Replace use of
11825         Ss with US. Adjust output for v5 and v4t.
11826         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
11828         * config/arm/constraints.md ("Ss"): Rename to US.
11830 2013-07-25  Terry Guo  <terry.guo@arm.com>
11832         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
11833         shift_add/shift_sub0/shift_sub1 RTXs.
11835 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
11836             Anton Blanchard  <anton@au1.ibm.com>
11838         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
11839         (altivec_vpks<VI_char>ss): Likewise.
11840         (altivec_vpks<VI_char>us): Likewise.
11841         (altivec_vpku<VI_char>us): Likewise.
11842         (altivec_vpku<VI_char>um): Likewise.
11844 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
11846         Introduce context class.
11848         * Makefile.in (CONTEXT_H): New.
11849         (OBJS): Add context.o.
11850         (toplev.o): Add CONTEXT_H to dependencies.
11851         (context.o): New.
11853         * toplev.c (general_init): Create the singleton gcc::context instance.
11855         * context.c: New.
11857         * context.h: New.
11859 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
11861         PR rtl-optimization/57968
11862         * mode-switching.c (create_pre_exit): Allow instructions that
11863         don't set a return register to need a non-exit mode.
11865 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
11866             Anton Blanchard  <anton@au1.ibm.com>
11868         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
11869         operands to vperm for little endian.
11870         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
11871         of lvsl to create the control mask for a vperm for little endian.
11873 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11874             Anton Blanchard  <anton@au1.ibm.com>
11876         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
11877         two operands for little-endian.
11879 2013-07-23  Steve Ellcey  <sellcey@mips.com>
11881         * config/mips/mips.c (mips_case_values_threshold): New.
11882         (TARGET_CASE_VALUES_THRESHOLD): Define.
11884 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11885             Anton Blanchard  <anton@au1.ibm.com>
11887         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
11888         selection of field for vector splat in little endian mode.
11890 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
11892         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
11893         expanders to rs6000.md.
11894         (ior<mode>3): Likewise.
11895         (and<mode>3): Likewise.
11896         (one_cmpl<mode>2): Likewise.
11897         (nor<mode>3): Likewise.
11898         (andc<mode>3): Likewise.
11899         (eqv<mode>3): Likewise.
11900         (nand<mode>3): Likewise.
11901         (orc<mode>3): Likewise.
11903         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
11904         declaration.
11906         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
11907         to split multi-word logical operations.
11908         (rs6000_split_logical_di): Likewise.
11909         (rs6000_split_logical): Likewise.
11911         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
11912         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
11913         and allow TImode operations in 32-bit.
11914         (vsx_and<mode>3_64bit): Likewise.
11915         (vsx_ior<mode>3_32bit): Likewise.
11916         (vsx_ior<mode>3_64bit): Likewise.
11917         (vsx_xor<mode>3_32bit): Likewise.
11918         (vsx_xor<mode>3_64bit): Likewise.
11919         (vsx_one_cmpl<mode>2_32bit): Likewise.
11920         (vsx_one_cmpl<mode>2_64bit): Likewise.
11921         (vsx_nor<mode>3_32bit): Likewise.
11922         (vsx_nor<mode>3_64bit): Likewise.
11923         (vsx_andc<mode>3_32bit): Likewise.
11924         (vsx_andc<mode>3_64bit): Likewise.
11925         (vsx_eqv<mode>3_32bit): Likewise.
11926         (vsx_eqv<mode>3_64bit): Likewise.
11927         (vsx_nand<mode>3_32bit): Likewise.
11928         (vsx_nand<mode>3_64bit): Likewise.
11929         (vsx_orc<mode>3_32bit): Likewise.
11930         (vsx_orc<mode>3_64bit): Likewise.
11932         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
11933         logical types in GPRs.
11935         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
11936         logical insns to rs6000.md, and allow TImode operations in
11937         32-bit.
11938         (altivec_ior<mode>3): Likewise.
11939         (altivec_xor<mode>3): Likewise.
11940         (altivec_one_cmpl<mode>2): Likewise.
11941         (altivec_nor<mode>3): Likewise.
11942         (altivec_andc<mode>3): Likewise.
11944         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
11945         attributes for moving the 128-bit logical operations into
11946         rs6000.md.
11947         (BOOL_REGS_OUTPUT): Likewise.
11948         (BOOL_REGS_OP1): Likewise.
11949         (BOOL_REGS_OP2): Likewise.
11950         (BOOL_REGS_UNARY): Likewise.
11951         (BOOL_REGS_AND_CR0): Likewise.
11952         (one_cmpl<mode>2): Add support for DI logical operations on
11953         32-bit, splitting the operations to 32-bit.
11954         (anddi3): Likewise.
11955         (iordi3): Likewise.
11956         (xordi3): Likewise.
11957         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
11958         changes to combine the 32/64-bit code, allow logical operations on
11959         TI mode in 32-bit, and to use similar match_operator patterns like
11960         scalar mode uses.  Combine the Altivec and VSX code for logical
11961         operations, and move it here.
11962         (ior<mode>3, 128-bit types): Likewise.
11963         (xor<mode>3, 128-bit types): Likewise.
11964         (one_cmpl<mode>3, 128-bit types): Likewise.
11965         (nor<mode>3, 128-bit types): Likewise.
11966         (andc<mode>3, 128-bit types): Likewise.
11967         (eqv<mode>3, 128-bit types): Likewise.
11968         (nand<mode>3, 128-bit types): Likewise.
11969         (orc<mode>3, 128-bit types): Likewise.
11970         (and<mode>3_internal): Likewise.
11971         (bool<mode>3_internal): Likewise.
11972         (boolc<mode>3_internal1): Likewise.
11973         (boolc<mode>3_internal2): Likewise.
11974         (boolcc<mode>3_internal1): Likewise.
11975         (boolcc<mode>3_internal2): Likewise.
11976         (eqv<mode>3_internal1): Likewise.
11977         (eqv<mode>3_internal2): Likewise.
11978         (one_cmpl1<mode>3_internal): Likewise.
11980 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11982         * config/microblaze/microblaze.c (microblaze_expand_prologue):
11983         Rename flag_stack_usage to flag_stack_usage_info.
11985 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
11987         * config/microblaze/sync.md: New file.
11988         * config/microblaze/microblaze.md: Include sync.md
11989         * config/microblaze/microblaze.c: Add print_operand 'y'.
11990         * config/microblaze/constraints.md: Add memory_contraint
11991         'Q' which is a single register.
11993 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
11995         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
11997 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
11999         * reload.c (find_reloads): Exit loop once we find this operand
12000         cannot be reloaded somehow for this alternative.
12002         * reload.c (find_reloads): Exit loop once we find a hard register.
12004         * rtlanal.c (computed_jump_p): Exit loop once we find label
12005         reference is used.
12007         * i386.c (ix86_pad_returns): Exit loop after setting replace.
12009         * cfgloopmanip.c (remove_path): Exit loop after setting
12010         irred_invalidated.
12012         * gensupport.c (subst_dup): Avoid loop if code is not
12013         MATCH_DUP nor MATCH_OP_DUP.
12015 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
12017         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
12019 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12021         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
12022         true for SP_REGNUM if mode == ptr_mode.
12023         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
12024         with value R0_REGNUM + 31.
12026 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12028         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
12029         pad pointer-typed argument downward.
12031 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12033         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
12034         and __ILP32__ when the ILP32 model is in use.
12036 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12038         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
12039         (aarch64_load_symref_appropriately): In the case of
12040         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
12041         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
12042         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
12043         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
12044         if 'mode' doesn't equal to 'ptr_mode'.
12045         (aarch64_output_mi_thunk): Add an assertion on the alignment of
12046         'vcall_offset'; change to call aarch64_emit_move differently depending
12047         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
12048         to calculate the upper bound of 'vcall_offset'.
12049         (aarch64_cannot_force_const_mem): Change to also return true if
12050         mode != ptr_mode.
12051         (aarch64_legitimize_reload_address): In the case of large
12052         displacements, add new local variable 'xmode' and an assertion
12053         based on it; change to use 'xmode' to generate the new rtx and
12054         reload.
12055         (aarch64_asm_trampoline_template): Change to generate the template
12056         differently depending on TARGET_ILP32 or not; change to use
12057         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
12058         (aarch64_trampoline_size): Removed.
12059         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
12060         and replace immediate literals with it.  Change to use 'ptr_mode'
12061         instead of 'DImode' and call convert_memory_address if the mode
12062         of 'fnaddr' doesn't equal to 'ptr_mode'.
12063         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
12064         to output symbol.
12065         (aarch64_elf_asm_destructor): Likewise.
12066         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
12067         on TARGET_ILP32 instead of aarch64_trampoline_size.
12068         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
12069         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
12070         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
12071         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
12072         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
12073         (storewb_pair<GPI:mode>_<P:mode>): ... this.
12074         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
12075         depending on the value of 'mode'.
12076         (add_losym_<mode>): New.
12077         (ldr_got_small_<mode>): New, based on ldr_got_small.
12078         (ldr_got_small): Remove.
12079         (ldr_got_small_sidi): New.
12080         * config/aarch64/iterators.md (P): New.
12081         (PTR): Change to 'ptr_mode' in the condition.
12083 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
12085         * config.gcc (aarch64*-*-*): Support --with-abi.
12086         (aarch64*-*-elf): Support --with-multilib-list.
12087         (aarch64*-*-linux*): Likewise.
12088         (supported_defaults): Add abi to aarch64*-*-*.
12089         * configure.ac: Mention AArch64 for --with-multilib-list.
12090         * configure: Re-generated.
12091         * config/aarch64/biarchilp32.h: New file.
12092         * config/aarch64/biarchlp64.h: New file.
12093         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
12094         (ABI_SPEC): Ditto.
12095         (MULTILIB_DEFAULTS): Ditto.
12096         (DRIVER_SELF_SPECS): Ditto.
12097         (ASM_SPEC): Update to also substitute -mabi.
12098         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
12099         file whose name depends on -mabi= and -mbig-endian.
12100         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
12101         TARGET_ILP32.
12102         (POINTER_SIZE): New define.
12103         (POINTERS_EXTEND_UNSIGNED): Ditto.
12104         (enum aarch64_abi_type): New enumeration tag.
12105         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
12106         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
12107         (TARGET_ILP32): New define.
12108         * config/aarch64/aarch64.opt (mabi): New.
12109         (aarch64_abi): New.
12110         (ilp32, lp64): New values for -mabi.
12111         * config/aarch64/t-aarch64 (comma): New define.
12112         (MULTILIB_OPTIONS): Ditto.
12113         (MULTILIB_DIRNAMES): Ditto.
12114         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
12115         * doc/invoke.texi: Document -mabi for AArch64.
12117 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
12119         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
12121 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
12122             Anton Blanchard  <anton@au1.ibm.com>
12124         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
12125         endianness when selecting field to splat.
12127 2013-07-22  Eric Christopher  <echristo@gmail.com>
12129         * dwarf2out.c (die_odr_checksum): New function to use
12130         CHECKSUM_ macros and ULEB128 for DIE tag.
12131         (generate_type_signature): Use.
12133 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
12135         * config.gcc (sparc*-*-*): Accept leon3 processor.
12136         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
12137         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
12138         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
12139         * config/sparc/sparc.opt (enum processor_type): Add leon3.
12140         (mfix-ut699): Adjust comment.
12141         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
12142         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
12143         (CPP_CPU_SPEC): Likewise.
12144         (ASM_CPU_SPEC): Likewise.
12145         * config/sparc/sparc.c (leon3_cost): New constant.
12146         (sparc_option_override): Add leon3 support.
12147         (mem_ref): New function.
12148         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
12149         (sparc_do_work_around_errata): Look into the instruction in the delay
12150         slot and adjust accordingly.  Add fix for the data cache nullify issues
12151         of the UT699.  Change insertion position for the NOP.
12152         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
12153         (leon3_load): New reservation.
12154         (leon_store): Bump latency to 2.
12155         (grfpu): New automaton.
12156         (grfpu_alu): New unit.
12157         (grfpu_ds): Likewise.
12158         (leon_fp_alu): Adjust.
12159         (leon_fp_mult): Delete.
12160         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
12161         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
12162         * config/sparc/sparc.md (cpu): Add leon3.
12163         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
12164         (swapsi): Likewise.
12165         (atomic_test_and_set): Likewise.
12166         (ldstub): Likewise.
12168 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
12170         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
12171         default for R5900 targets.
12172         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
12173         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
12174         * config/mips/mips.c (mips_option_override): Report an error for
12175         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
12176         for -march=r5900 -mhard-float.
12178 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
12180         * df-problems.c (can_move_insns_across): Exit loop once we
12181         find a non-fixed, non-global register.
12183         * ipa-pure-const.c (propagate_nothrow): Exit loop after
12184         setting can_throw.
12186         * omega.c (omega_eliminate_red): Break after setting red_found.
12187         (omega_problem_has_red_equations): Similarly after setting found.
12188         (omega_query_variable): Similarly after setting coupled.
12190 2013-07-22  Marek Polacek  <polacek@redhat.com>
12192         * gimplify.c: Don't include gimple.h twice.
12194 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12196         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
12197         instead of TARGET_THUMB1.
12198         (Pz): New constraint.
12199         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
12200         encodings.
12201         (compare_negsi_si): Likewise.
12202         (compare_addsi2_op0): Likewise.
12203         (compare_addsi2_op1): Likewise.
12204         (addsi3_carryin_<optab>): Likewise.
12205         (addsi3_carryin_alt2_<optab>): Likewise.
12206         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
12207         for arm_restrict_it.
12208         (subsi3_carryin): Likewise.
12209         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
12210         (minmax_arithsi): Disable for arm_restrict_it.
12211         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
12212         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
12213         (satsi_<SAT:code>_shift): Likewise.
12214         (arm_shiftsi3): Add alternative for 16-bit encoding.
12215         (arm32_movhf): Disable for arm_restrict_it.
12216         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
12217         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
12219 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
12221         * config/arm/arm.md (attribute "insn"): Delete.
12222         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
12223         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
12224         (not_shiftsi): Update for attribute change.
12225         (not_shiftsi_compare0): Likewise.
12226         (not_shiftsi_compare0_scratch): Likewise.
12227         (arm_one_cmplsi2): Likewise.
12228         (thumb1_one_cmplsi2): Likewise.
12229         (notsi_compare0): Likewise.
12230         (notsi_compare0_scratch): Likewise.
12231         (thumb1_movdi_insn): Likewise.
12232         (arm_movsi_insn): Likewise.
12233         (movhi_insn_arch4): Likewise.
12234         (movhi_bytes): Likewise.
12235         (arm_movqi_insn): Likewise.
12236         (thumb1_movqi_insn): Likewise.
12237         (arm32_movhf): Likewise.
12238         (thumb1_movhf): Likewise.
12239         (arm_movsf_soft_insn): Likewise.
12240         (thumb1_movsf_insn): Likewise.
12241         (thumb_movdf_insn): Likewise.
12242         (movsicc_insn): Likewise.
12243         (movsfcc_soft_insn): Likewise.
12244         (and_scc): Likewise.
12245         (cond_move): Likewise.
12246         (if_move_not): Likewise.
12247         (if_not_move): Likewise.
12248         (if_shift_move): Likewise.
12249         (if_move_shift): Likewise.
12250         (if_shift_shift): Likewise.
12251         (if_not_arith): Likewise.
12252         (if_arith_not): Likewise.
12253         (cond_move_not): Likewise.
12254         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
12255         (neon_mov<mode>): Likewise.
12256         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
12257         (thumb2_movsi_vfp): Likewise.
12258         (movsf_vfp): Likewise.
12259         (thumb2_movsf_vfp): Likewise.
12260         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12261         change.
12262         (cortexa7_older_only): Likewise.
12263         (cortexa7_younger): Likewise.
12264         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
12265         (1020alu_shift_op): Likewise.
12266         (1020alu_shift_reg_op): Likewise.
12267         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
12268         (alu_shift_op): Likewise.
12269         (alu_shift_reg_op): Likewise.
12270         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
12271         (11_alu_shift_op): Likewise.
12272         (11_alu_shift_reg_op): Likewise.
12273         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
12274         (9_alu_shift_reg_op): Likewise.
12275         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12276         change.
12277         (cortex_a15_alu_shift): Likewise.
12278         (cortex_a15_alu_shift_reg): Likewise.
12279         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
12280         (cortex_a5_alu_shift): Likewise.
12281         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
12282         change.
12283         (cortex_a53_alu_shift): Likewise.
12284         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12285         change.
12286         (cortex_a7_alu_reg): Likewise.
12287         (cortex_a7_alu_shift): Likewise.
12288         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12289         (cortex_a8_alu_shift): Likewise.
12290         (cortex_a8_alu_shift_reg): Likewise.
12291         (cortex_a8_mov): Likewise.
12292         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
12293         (cortex_a9_dp_shift): Likewise.
12294         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
12295         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
12296         (cortex_r4_mov): Likewise.
12297         (cortex_r4_alu_shift): Likewise.
12298         (cortex_r4_alu_shift_reg): Likewise.
12299         * config/arm/fa526.md (526_alu_op): Update for attribute change.
12300         (526_alu_shift_op): Likewise.
12301         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
12302         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
12303         (626te_alu_shift_op): Likewise.
12304         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
12305         (726te_alu_op): Likewise.
12306         (726te_alu_shift_op): Likewise.
12307         (726te_alu_shift_reg_op): Likewise.
12308         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
12309         (mp626_alu_shift_op): Likewise.
12310         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
12311         (pj4_alu_e1_conds): Likewise.
12312         (pj4_alu): Likewise.
12313         (pj4_alu_conds): Likewise.
12314         (pj4_shift): Likewise.
12315         (pj4_shift_conds): Likewise.
12316         (pj4_alu_shift): Likewise.
12317         (pj4_alu_shift_conds): Likewise.
12319 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12321         * config/arm/predicates.md (shiftable_operator_strict_it):
12322         New predicate.
12323         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
12324         Disable cond_exec version for arm_restrict_it.
12325         (thumb2_smaxsi3): Convert to generate cond_exec.
12326         (thumb2_sminsi3): Likewise.
12327         (thumb32_umaxsi3): Likewise.
12328         (thumb2_uminsi3): Likewise.
12329         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
12330         (thumb2_neg_abssi2): Likewise.
12331         (thumb2_mov_scc): Add alternative for 16-bit encoding.
12332         (thumb2_movsicc_insn): Adjust alternatives.
12333         (thumb2_mov_negscc): Disable for arm_restrict_it.
12334         (thumb2_mov_negscc_strict_it): New pattern.
12335         (thumb2_mov_notscc_strict_it): New pattern.
12336         (thumb2_mov_notscc): Disable for arm_restrict_it.
12337         (thumb2_ior_scc): Likewise.
12338         (thumb2_ior_scc_strict_it): New pattern.
12339         (thumb2_cond_move): Adjust for arm_restrict_it.
12340         (thumb2_cond_arith): Disable for arm_restrict_it.
12341         (thumb2_cond_arith_strict_it): New pattern.
12342         (thumb2_cond_sub): Adjust for arm_restrict_it.
12343         (thumb2_movcond): Likewise.
12344         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
12345         (thumb2_zero_extendhisi2_v6): Likewise.
12346         (thumb2_zero_extendqisi2_v6): Likewise.
12347         (orsi_notsi_si): Likewise.
12348         (orsi_not_shiftsi_si): Likewise.
12350 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
12352         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
12353         instruction sequence is 1 byte shorter.
12355 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
12357         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
12358         it is not needed after split.
12360 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
12362         PR target/51784
12363         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
12364         second label for nonlocal goto receivers. Don't output pic base labels
12365         unless we're producing PIC; mark that action unreachable().
12366         (ix86_save_reg): If the function contains a nonlocal label, save the
12367         PIC base reg.
12368         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
12369         * config/darwin.c (emitted_pic_label_num): New GTY.
12370         (update_pic_label_number_if_needed): New.
12371         (machopic_output_function_base_name): Adjust for nonlocal receiver
12372         case.
12373         (machopic_should_output_picbase_label): New.
12374         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
12375         (nonlocal_goto_receiver): New insn and split.
12377 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
12379         * config/aarch64/aarch64-builtins.c
12380         (aarch64_fold_builtin): Fold abs in all modes.
12381         * config/aarch64/aarch64-simd-builtins.def
12382         (abs): Enable for all modes.
12383         * config/aarch64/arm_neon.h
12384         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
12385         (vabs_f64): Add missing intrinsic.
12387 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
12389         * config/aarch64/arm_neon.h (vabs_s64): New function
12391 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
12393         PR target/57516
12394         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
12395         * config/avr/avr.md (adjust_len): Add `round'.
12396         * config/avr/avr-protos.h (avr_out_round): New prototype.
12397         (avr_out_plus): Add `out_label' argument.
12398         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
12399         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
12400         Handle the case where `insn' is just a pattern.
12401         (avr_out_bitop): Handle the case where `insn' is just a pattern.
12402         (avr_out_round): New function.
12403         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
12405 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
12407         * config/microblaze/microblaze.c (microblaze_expand_prologue):
12408         Add check for flag_stack_usage to handle -fstack-usage support
12410 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
12412         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
12413         interaction for new Power8 flags and VSX.
12415 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
12417         PR middle-end/57698
12418         * tree-inline.c (expand_call_inline): Emit errors during
12419         early_inlining only if optimization is not turned on.
12421 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12423         * passes.def: New.
12425         * passes.c (init_optimization_passes): Move the construction of
12426         the pass hierarchy into a new passes.def file.
12428         * Makefile.in (passes.o): Add dependency on passes.def.
12430 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12432         * passes.c (init_optimization_passes): Introduce macros for
12433         constructing the tree of passes (INSERT_PASSES_AFTER,
12434         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
12435         TERMINATE_PASS_LIST).
12437 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
12438             Wei Mi  <wmi@google.com>
12440         PR rtl-optimization/57878
12441         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
12442         top.
12443         (reload_pseudo_compare_func): Check nregs first for reload
12444         pseudos.
12446 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12448         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
12450 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
12452         * read-rtl.c (validate_const_int): Once an invalid character is
12453         seen, quit the loop.
12455         * gengtype.c (write_roots): Similarly once we find the "deletable"
12456         or "if_marked" option.
12458 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12460         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
12461         "xtab" and "sat".  Move value "clz" from here to ...
12462         (attriubte "type"): ... here.
12463         (satsi_<SAT:code>): Delete "insn" attribute.
12464         (satsi_<SAT:code>_shift): Likewise.
12465         (arm_zero_extendqisi2addsi): Likewise.
12466         (arm_extendqisi2addsi): Likewise.
12467         (clzsi2): Update for attribute changes.
12468         (rbitsi2): Likewise.
12469         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
12470         attribute.
12471         (arm_usatsihi): Likewise.
12472         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
12474 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
12476         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
12477         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
12478         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
12479         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
12480         in alphabetical order.
12481         (attribute "core_cycles"): Update for attribute changes.
12482         (arm_addsi3): Likewise.
12483         (addsi3_compare0): Likewise.
12484         (addsi3_compare0_scratch): Likewise.
12485         (addsi3_compare_op1): Likewise.
12486         (addsi3_compare_op2): Likewise.
12487         (compare_addsi2_op0): Likewise.
12488         (compare_addsi2_op1): Likewise.
12489         (addsi3_carryin_shift_<optab>): Likewise.
12490         (subsi3_carryin_shift): Likewise.
12491         (rsbsi3_carryin_shift): Likewise.
12492         (arm_subsi3_insn): Likewise.
12493         (subsi3_compare0): Likewise.
12494         (subsi3_compare): Likewise.
12495         (arm_andsi3_insn): Likewise.
12496         (thumb1_andsi3_insn): Likewise.
12497         (andsi3_compare0): Likewise.
12498         (andsi3_compare0_scratch): Likewise.
12499         (zeroextractsi_compare0_scratch
12500         (andsi_not_shiftsi_si): Likewise.
12501         (iorsi3_insn): Likewise.
12502         (iorsi3_compare0): Likewise.
12503         (iorsi3_compare0_scratch): Likewise.
12504         (arm_xorsi3): Likewise.
12505         (thumb1_xorsi3_insn): Likewise.
12506         (xorsi3_compare0): Likewise.
12507         (xorsi3_compare0_scratch): Likewise.
12508         (satsi_<SAT:code>_shift): Likewise.
12509         (rrx): Likewise.
12510         (arm_shiftsi3): Likewise.
12511         (shiftsi3_compare0): Likewise.
12512         (not_shiftsi): Likewise.
12513         (not_shiftsi_compare0): Likewise.
12514         (not_shiftsi_compare0_scratch): Likewise.
12515         (arm_one_cmplsi2): Likewise.
12516         (thumb_one_complsi2): Likewise.
12517         (notsi_compare0): Likewise.
12518         (notsi_compare0_scratch): Likewise.
12519         (thumb1_zero_extendhisi2): Likewise.
12520         (arm_zero_extendhisi2): Likewise.
12521         (arm_zero_extendhisi2_v6): Likewise.
12522         (arm_zero_extendhisi2addsi): Likewise.
12523         (thumb1_zero_extendqisi2): Likewise.
12524         (thumb1_zero_extendqisi2_v6): Likewise.
12525         (arm_zero_extendqisi2): Likewise.
12526         (arm_zero_extendqisi2_v6): Likewise.
12527         (arm_zero_extendqisi2addsi): Likewise.
12528         (thumb1_extendhisi2): Likewise.
12529         (arm_extendhisi2): Likewise.
12530         (arm_extendhisi2_v6): Likewise.
12531         (arm_extendqisi): Likewise.
12532         (arm_extendqisi_v6): Likewise.
12533         (arm_extendqisi2addsi): Likewise.
12534         (thumb1_extendqisi2): Likewise.
12535         (thumb1_movdi_insn): Likewise.
12536         (arm_movsi_insn): Likewise.
12537         (movsi_compare0): Likewise.
12538         (movhi_insn_arch4): Likewise.
12539         (movhi_bytes): Likewise.
12540         (arm_movqi_insn): Likewise.
12541         (thumb1_movqi_insn): Likewise.
12542         (arm32_movhf): Likewise.
12543         (thumb1_movhf): Likewise.
12544         (arm_movsf_soft_insn): Likewise.
12545         (thumb1_movsf_insn): Likewise.
12546         (movdf_soft_insn): Likewise.
12547         (thumb_movdf_insn): Likewise.
12548         (arm_cmpsi_insn): Likewise.
12549         (cmpsi_shiftsi): Likewise.
12550         (cmpsi_shiftsi_swp): Likewise.
12551         (arm_cmpsi_negshiftsi_si): Likewise.
12552         (movsicc_insn): Likewise.
12553         (movsfcc_soft_insn): Likewise.
12554         (arith_shiftsi): Likewise.
12555         (arith_shiftsi_compare0
12556         (arith_shiftsi_compare0_scratch
12557         (sub_shiftsi): Likewise.
12558         (sub_shiftsi_compare0
12559         (sub_shiftsi_compare0_scratch
12560         (and_scc): Likewise.
12561         (cond_move): Likewise.
12562         (if_plus_move): Likewise.
12563         (if_move_plus): Likewise.
12564         (if_move_not): Likewise.
12565         (if_not_move): Likewise.
12566         (if_shift_move): Likewise.
12567         (if_move_shift): Likewise.
12568         (if_shift_shift): Likewise.
12569         (if_not_arith): Likewise.
12570         (if_arith_not): Likewise.
12571         (cond_move_not): Likewise.
12572         (thumb1_ashlsi3): Set type attribute.
12573         (thumb1_ashrsi3): Likewise.
12574         (thumb1_lshrsi3): Likewise.
12575         (thumb1_rotrsi3): Likewise.
12576         (shiftsi3_compare0_scratch): Likewise.
12577         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
12578         (neon_mov<mode>): Likewise.
12579         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
12580         attribute changes.
12581         (thumb2_movsi_insn): Likewise.
12582         (thumb2_cmpsi_neg_shiftsi): Likewise.
12583         (thumb2_extendqisi_v6): Likewise.
12584         (thumb2_zero_extendhisi2_v6): Likewise.
12585         (thumb2_zero_extendqisi2_v6): Likewise.
12586         (thumb2_shiftsi3_short): Likewise.
12587         (thumb2_addsi3_compare0_scratch): Likewise.
12588         (orsi_not_shiftsi_si): Likewise.
12589         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
12590         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
12591         changes.
12592         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
12593         (1020alu_shift_op): Likewise.
12594         (1020alu_shift_reg_op): Likewise.
12595         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
12596         (alu_shift_op): Likewise.
12597         (alu_shift_reg_op): Likewise.
12598         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
12599         (11_alu_shift_op): Likewise.
12600         (11_alu_shift_reg_op): Likewise.
12601         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
12602         (9_alu_shift_reg_op): Likewise.
12603         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
12604         changes.
12605         (cortex_a15_alu_shift): Likewise.
12606         (cortex_a15_alu_shift_reg): Likewise.
12607         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
12608         changes.
12609         (cortex_a5_alu_shift): Likewise.
12610         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
12611         changes.
12612         (cortex_a53_alu_shift): Likewise.
12613         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
12614         changes.
12615         (cortex_a7_alu_reg): Likewise.
12616         (cortex_a7_alu_shift): Likewise.
12617         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
12618         changes.
12619         (cortex_a8_alu_shift): Likewise.
12620         (cortex_a8_alu_shift_reg): Likewise.
12621         (cortex_a8_mov): Likewise.
12622         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
12623         (cortex_a9_dp_shift): Likewise.
12624         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
12625         changes.
12626         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
12627         changes.
12628         (cortex_r4_mov): Likewise.
12629         (cortex_r4_alu_shift): Likewise.
12630         (cortex_r4_alu_shift_reg): Likewise.
12631         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
12632         (526_alu_shift_op): Likewise.
12633         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
12634         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
12635         (626te_alu_shift_op): Likewise.
12636         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
12637         (726te_alu_op): Likewise.
12638         (726te_alu_shift_op): Likewise.
12639         (726te_alu_shift_reg_op): Likewise.
12640         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
12641         (mp626_alu_shift_op): Likewise.
12642         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
12643         (pj4_alu_e1_conds): Likewise.
12644         (pj4_alu): Likewise.
12645         (pj4_alu_conds): Likewise.
12646         (pj4_shift): Likewise.
12647         (pj4_shift_conds): Likewise.
12648         (pj4_alu_shift): Likewise.
12649         (pj4_alu_shift_conds): Likewise.
12650         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
12651         changes.
12652         (cortexa7_older_only): Likewise.
12653         (cortexa7_younger): Likewise.
12655 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
12657         * ipa-pure-const.c (generate_summary): Rename to...
12658         (pure_const_generate_summary): ... this.
12660 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
12662         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
12664 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
12666         PR target/57909
12667         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
12668         usage in HI mode.
12670 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12672         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
12673         enabled without -march=zEC12.
12674         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
12675         flags to be set.
12677 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12679         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
12680         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
12681         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
12682         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
12683         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
12684         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
12685         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
12686         ISA_HAS_FP4.
12687         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
12688         and ISA_HAS_NMADD3_NMSUB3.
12689         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
12690         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
12691         (nmsub4<mode>, nmsub3<mode>): Likewise.
12692         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
12694 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12696         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
12697         TARGET_MIPS5400 checking.
12699 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
12700             Peter Bergner  <bergner@vnet.ibm.com>
12702         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
12703         registers in the comment.
12704         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
12705         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
12706         rather than FIRST_PSEUDO_REGISTERS.
12708 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
12710         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
12711         enable extra ISA flags with TARGET_HTM.
12713 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
12715         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
12716         Fix comment typos.
12718 2013-07-15  Cong Hou  <congh@google.com>
12720         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
12721         in compare function for sorting.
12723 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
12725         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
12726         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
12727         * config/rs6000/rs6000.opt: Add -mhtm option.
12728         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
12729         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
12730         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
12731         __HTM__ if the HTM instructions are available.
12732         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
12733         htm_spr_reg_operand): New define_predicates.
12734         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
12735         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
12736         Include htm.md.
12737         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
12738         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
12739         HTM builtin functions.
12740         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
12741         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
12742         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
12743         (rs6000_builtin_mask_calculate): Likewise.
12744         (rs6000_option_override_internal): Likewise.
12745         (bdesc_htm): Add new HTM builtin support.
12746         (htm_spr_num): New function.
12747         (htm_spr_regno): Likewise.
12748         (rs6000_htm_spr_icode): Likewise.
12749         (htm_expand_builtin): Likewise.
12750         (htm_init_builtins): Likewise.
12751         (rs6000_expand_builtin): Add support for HTM builtin functions.
12752         (rs6000_init_builtins): Likewise.
12753         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
12754         option.
12755         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
12756         (TARGET_HTM, MASK_HTM): Define macros.
12757         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
12758         (FIXED_REGISTERS): Likewise.
12759         (CALL_USED_REGISTERS): Likewise.
12760         (CALL_REALLY_USED_REGISTERS): Likewise.
12761         (REG_ALLOC_ORDER): Likewise.
12762         (enum reg_class): Likewise.
12763         (REG_CLASS_NAMES): Likewise.
12764         (REG_CLASS_CONTENTS): Likewise.
12765         (REGISTER_NAMES): Likewise.
12766         (ADDITIONAL_REGISTER_NAMES): Likewise.
12767         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
12768         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
12769         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
12770         * config/rs6000/htm.md: New file.
12771         * config/rs6000/htmintrin.h: New file.
12772         * config/rs6000/htmxlintrin.h: New file.
12774 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
12776         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
12777         Define SYMBOL_TINY_GOT, update comment.
12778         * config/aarch64/aarch64.c
12779         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
12780         (aarch64_expand_mov_immediate): Likewise.
12781         (aarch64_print_operand): Likewise.
12782         (aarch64_classify_symbol): Likewise.
12783         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
12784         (ldr_got_tiny): Define.
12786 2013-07-13  Tobias Grosser  <tobias@grosser.es>
12788         PR tree-optimization/54094
12789         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
12790           scheduling dimension for the parallelism check from the polyhedral
12791           information in the AST.
12792         * graphite-dependences.c (carries_deps): Do not assume the schedule is
12793           in 2D + 1 form.
12795 2013-07-13  Jason Merrill  <jason@redhat.com>
12797         * print-tree.c (debug_vec_tree): Use debug_raw.
12798         (debug_raw (vec<tree, va_gc> &)): New.
12799         (debug_raw (vec<tree, va_gc> *)): New.
12800         * tree.h: Declare them.
12802 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
12804         * ifcvt.c (ifcvt_after_combine): New static variable.
12805         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
12806         for size.
12807         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
12808         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
12809         rest_of_handle_if_after_reload): Pass new argument for if_convert.
12811 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
12813         * config/mips/mips.c (mips_expand_call): Remove empty statement.
12815 2013-07-12  Michael Matz  <matz@suse.de>
12817         PR middle-end/55771
12818         * convert.c (convert_to_real): Reject non-float inner types.
12820 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
12822         * config/aarch64/aarch64-protos.h
12823         (aarch64_simd_immediate_valid_for_move): Remove.
12824         * config/aarch64/aarch64.c (simd_immediate_info): New member.
12825         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
12826         cases.
12827         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
12829 2013-07-11  Steve Ellcey  <sellcey@mips.com>
12831         * config/mips/mips.c (mips_conditional_register_usage): Do not
12832         use t[0-7] registers in MIPS16 mode when optimizing for size.
12834 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
12836         * config/i386/i386.c (dispatch_function_versions): Fix array
12837         indexing of function_version_info to match actual_versions.
12839 2013-07-11  Teresa Johnson  <tejohnson@google.com>
12841         * vec.h (struct va_gc): Move release out-of-line.
12842         (va_gc::release): Call ggc_free on released vec.
12844 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
12846         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
12847         Require GOT register as additional operand in UNSPEC.
12848         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
12849         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12850         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
12851         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
12852         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
12853         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
12854         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
12856 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
12858         PR target/57631
12859         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
12860         name seen by assembler/linker rather if available.
12862 2013-07-11  Andreas Schwab  <schwab@suse.de>
12864         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
12866 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
12868         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
12870 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
12872         * doc/tm.texi.in: Move hook documentation to ....
12873         * target.def: ... here.
12875         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
12876         text on @hook line.
12877         * doc/tm.texi: Regenerate.
12879 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
12881         PR c++/57869
12882         * doc/invoke.texi: Document -Wconditionally-supported.
12884 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12886         PR target/57844
12887         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
12888         of my_fp.
12890 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12892         PR target/57506
12893         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
12894         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
12895         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
12896         Remove duplicate devices.
12897         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
12898         * config/avr/t-multilib: Regenerate.
12899         * config/avr/avr-tables.opt: Regenerate.
12900         * doc/avr-mmcu.texi: Regenerate.
12902 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
12904         PR target/56987
12905         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
12907 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
12909         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
12910         the cost of MULT when optimizing for size.
12912 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
12914         * config/cr16/cr16-protos.h: Don't include target.h.
12916 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
12918         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
12919         adjust register size for TDmode and TFmode for VSX registers.
12921 2013-07-08  Kai Tietz  <ktietz@redhat.com>
12923         PR target/56892
12924         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
12925         hook_bool_const_tree_true.
12927 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12929         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
12930         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
12931         * config/s390/s390.md: Define FPR*_REGNUM constants.
12932         Fix FPR2_REGNUM constant (18 -> 17).
12933         ("*trunc<BFP:mode><DFP_ALL:mode>2")
12934         ("*trunc<DFP_ALL:mode><BFP:mode>2")
12935         ("trunc<BFP:mode><DFP_ALL:mode>2")
12936         ("trunc<DFP_ALL:mode><BFP:mode>2")
12937         ("*extend<BFP:mode><DFP_ALL:mode>2")
12938         ("*extend<DFP_ALL:mode><BFP:mode>2")
12939         ("extend<BFP:mode><DFP_ALL:mode>2")
12940         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
12941         FPR4_REGNUM.
12943 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
12945         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
12947 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
12949         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
12950         and cfun_fpr_bit_p to cfun_fpr_save_p.
12951         (s390_frame_area, s390_register_info, s390_frame_info)
12952         (s390_emit_prologue, s390_emit_epilogue)
12953         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
12954         register numbers.
12955         * config/s390/s390.h: Define *_REGNUM macros for floating point
12956         register numbers.
12958 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
12960         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
12962 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
12964         PR rtl-optimization/57786
12965         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
12966         and break out of the loop when it is set to false.
12968 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
12970         PR target/57819
12971         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
12972         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
12973         (const_int 63)) 0)).
12974         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
12975         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
12976         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
12978         PR rtl-optimization/57829
12979         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
12980         mask bits outside of mode are just sign-extension from mode to HWI.
12982 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
12984         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
12985         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
12986         adjust_address instead of change_address to keep info about alignment.
12987         (emit_strmov): Remove.
12988         (emit_memmov): New function.
12989         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
12990         (expand_movmem_epilogue): Likewise and return updated rtx for
12991         destination.
12992         (expand_constant_movmem_prologue): Likewise and return updated rtx for
12993         destination and source.
12994         (decide_alignment): Refactor, handle vector_loop.
12995         (ix86_expand_movmem): Likewise.
12996         (ix86_expand_setmem): Likewise.
12997         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
12999 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
13001         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
13002         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
13004 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13006         * config/i386/sse.md (sse_movlhps): Change alternative 3
13007         of operand 2 to "m".
13009 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
13011         PR target/57807
13012         * config/i386/sse.md (iptr): New mode attribute.
13013         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
13014         (<sse>_vm<plusminus_insn><mode>3): Ditto.
13015         (<sse>_vmmul<mode>3): Ditto.
13016         (<sse>_vmdiv<mode>3): Ditto.
13017         (sse_vmrcpv4sf2): Ditto.
13018         (<sse>_vmsqrt<mode>2): Ditto.
13019         (sse_vmrsqrtv4sf2): Ditto.
13020         (<sse>_vm<code><mode>3): Ditto.
13021         (avx_vmcmp<mode>3): Ditto.
13022         (<sse>_vmmaskcmp<mode>3): Ditto.
13023         (<sse>_comi): Ditto.
13024         (<sse>_ucomi): Ditto.
13025         (*xop_vmfrcz_<mode>): Ditto.
13026         (*fmai_fmadd_<mode>): Ditto.
13027         (*fmai_fmsub_<mode>): Ditto.
13028         (*fmai_fnmadd_<mode>): Ditto.
13029         (*fmai_fnmsub_<mode>): Ditto.
13030         (*fma4i_vmfmadd_<mode>): Ditto.
13031         (*fma4i_vmfmsub_<mode>): Ditto.
13032         (*fma4i_vmfnmadd_<mode>): Ditto.
13033         (*fma4i_vmfnmsub_<mode>): Ditto.
13034         (*xop_vmfrcz_<mode>): Ditto.
13035         (sse_cvtps2pi): Ditto.
13036         (sse_cvttps2pi): Ditto.
13037         (sse_cvtss2si): Ditto.
13038         (sse_cvtss2si_2): Ditto.
13039         (sse_cvtss2siq_2): Ditto.
13040         (sse_cvttss2si): Ditto.
13041         (sse_cvttss2siq): Ditto.
13042         (sse_cvtsd2si): Ditto.
13043         (sse_cvtsd2si_2): Ditto.
13044         (sse_cvtsd2siq_2): Ditto.
13045         (sse_cvttsd2si): Ditto.
13046         (sse_cvttsd2siq): Ditto.
13047         (sse_cvtsd2ss): Ditto.
13048         (sse_cvtss2sd): Ditto.
13049         (avx2_pbroadcast<mode>): Ditto.
13050         (avx2_pbroadcast<mode>_1): Ditto.
13051         (*avx_vperm_broadcast_v4sf): Ditto.
13053         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
13054         (sse_movlhps): Ditto.
13055         (sse_storehps): Ditto.
13056         (sse_loadhps): Ditto.
13057         (sse_storelps): Ditto.
13058         (sse_loadlps): Ditto.
13059         (*vec_concatv4sf): Ditto.
13060         (*vec_interleave_highv2df): Ditto.
13061         (*vec_interleave_lowv2df): Ditto.
13062         (*vec_extractv2df_1_sse): Ditto.
13063         (*vec_extractv2df_0_sse): Ditto.
13064         (sse2_storelpd): Ditto.
13065         (sse2_loadlpd): Ditto.
13066         (sse2_movsd): Ditto.
13067         (*vec_concatv4si): Ditto.
13068         (vec_concatv2di): Ditto.
13070         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
13071         for Intel asm dialect.
13072         (mmx_punpcklwd): Ditto.
13073         (mmx_punpckldq): Ditto.
13075         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
13076         for intel assembler dialect.
13078 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
13080         PR target/29776
13081         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
13082         for BUILT_IN_C{LZ,LRSB}*.
13083         * tree.h (CASE_INT_FN): Add FN##IMAX case.
13084         * tree-vrp.c (extract_range_basic): Handle
13085         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
13086         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
13087         fall thru to code calling set_value*.
13088         * builtins.c (expand_builtin): Remove *IMAX cases.
13089         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
13090         if width is bigger than 2*HWI.
13092 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
13094         PR rtl-optimization/55342
13095         * lra-int.h (lra_subreg_reload_pseudos): New.
13096         * lra.c: Add undoing optional reloads to the block diagram.
13097         (lra_subreg_reload_pseudos): New.
13098         (lra_optional_reload_pseudos): Change comments.
13099         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
13100         lra_optional_reload_pseudos after undo transformations.
13101         * lra-assigns.c (pseudo_prefix_title): New.
13102         (lra_setup_reg_renumber): Use it.
13103         (spill_for): Ditto.  Check subreg reload pseudos too.
13104         (assign_by_spills): Consider subreg reload pseudos too.
13105         * lra-constraints.c (simplify_operand_subreg): Use
13106         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
13107         (curr_insn_transform): Recognize and do optional reloads.
13108         (undo_optional_reloads): New.
13109         (lra_undo_inheritance): Call undo_optional_reloads.
13111 2013-07-05  Thomas Quinot  <quinot@adacore.com>
13113         * tree-complex.c (expand_complex_operations_1): Fix typo.
13115 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
13117         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
13118         (tune_params): New member 'const vec_costs'.
13119         * config/aarch64/aarch64.c (generic_vector_cost): New.
13120         (generic_tunings): New member 'generic_vector_cost'.
13121         (aarch64_builtin_vectorization_cost): New.
13122         (aarch64_add_stmt_cost): New.
13123         (TARGET_VECTORIZE_ADD_STMT_COST): New.
13124         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
13126 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
13128         PR target/57777
13129         * config/i386/predicates.md (vsib_address_operand): Disallow
13130         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
13132 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
13134         PR middle-end/55030
13135         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
13136         expand_builtin_setjmp_receiver.
13137         (expand_label): Adjust, call expand_builtin_setjmp_receiver
13138         with NULL for the label parameter.
13139         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
13140         the frame-pointer.  Adjust comments.
13141         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
13142         only if LABEL is non-NULL.
13144 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
13146         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
13147         (ARM_ABI_AAPCS64): Ditto.
13148         (arm_abi): Ditto.
13149         (ARM_DEFAULT_ABI): Ditto.
13151 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
13153         * config/aarch64/aarch64-builtins.c
13154         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
13155         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
13156         (st1): Likewise.
13157         * config/aarch64/aarch64-simd.md
13158         (aarch64_ld1<VALL:mode>): New.
13159         (aarch64_st1<VALL:mode>): Likewise.
13160         * config/aarch64/arm_neon.h
13161         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
13163 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
13165         * config/i386/i386.c (gate_insert_vzeroupper): Check if
13166         target ISA is AVX.
13167         (ix86_option_override_internal):Turn on all -mavx target flags by
13168         default as they are dependent on AVX anyway.
13170 2013-07-02  Cary Coutant  <ccoutant@google.com>
13172         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
13173         deterministic hash.
13174         (loc_checksum_ordered): Likewise.
13175         (hash_loc_operands): Remove inline keyword.
13177 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
13179         PR tree-optimization/57741
13180         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
13181         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
13182         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
13183         Allow REAL_CST step_exprs if flag_associative_math.
13184         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
13186 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13188         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
13190 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
13192         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
13194 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13196         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
13197         encoding.
13198         (iorsi3_insn): Likewise.
13199         (arm_xorsi3): Likewise.
13201 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
13203         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
13204         to ...
13205         (attribute "type"): ... here, and prefix with "wmmx_".
13206         (attribute "core_cycles"): Update for attribute changes.
13207         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
13208         (tbcstv4hi): Likewise.
13209         (tbcstv2si): Likewise.
13210         (iwmmxt_iordi3): Likewise.
13211         (iwmmxt_xordi3): Likewise.
13212         (iwmmxt_anddi3): Likewise.
13213         (iwmmxt_nanddi3): Likewise.
13214         (iwmmxt_arm_movdi): Likewise.
13215         (iwmmxt_movsi_insn): Likewise.
13216         (mov<mode>_internal): Likewise.
13217         (and<mode>3_iwmmxt): Likewise.
13218         (ior<mode>3_iwmmxt): Likewise.
13219         (xor<mode>3_iwmmxt): Likewise.
13220         (add<mode>3_iwmmxt): Likewise.
13221         (ssaddv8qi3): Likewise.
13222         (ssaddv4hi3): Likewise.
13223         (ssaddv2si3): Likewise.
13224         (usaddv8qi3): Likewise.
13225         (usaddv4hi3): Likewise.
13226         (usaddv2si3): Likewise.
13227         (sub<mode>3_iwmmxt): Likewise.
13228         (sssubv8qi3): Likewise.
13229         (sssubv4hi3): Likewise.
13230         (sssubv2si3): Likewise.
13231         (ussubv8qi3): Likewise.
13232         (ussubv4hi3): Likewise.
13233         (ussubv2si3): Likewise.
13234         (mulv4hi3_iwmmxt): Likewise.
13235         (smulv4hi3_highpart): Likewise.
13236         (umulv4hi3_highpart): Likewise.
13237         (iwmmxt_wmacs): Likewise.
13238         (iwmmxt_wmacsz): Likewise.
13239         (iwmmxt_wmacu): Likewise.
13240         (iwmmxt_wmacuz): Likewise.
13241         (iwmmxt_clrdi): Likewise.
13242         (iwmmxt_clrv8qi): Likewise.
13243         (iwmmxt_clr4hi): Likewise.
13244         (iwmmxt_clr2si): Likewise.
13245         (iwmmxt_uavgrndv8qi3): Likewise.
13246         (iwmmxt_uavgrndv4hi3): Likewise.
13247         (iwmmxt_uavgv8qi3): Likewise.
13248         (iwmmxt_uavgv4hi3): Likewise.
13249         (iwmmxt_tinsrb): Likewise.
13250         (iwmmxt_tinsrh): Likewise.
13251         (iwmmxt_tinsrw): Likewise.
13252         (iwmmxt_textrmub): Likewise.
13253         (iwmmxt_textrmsb): Likewise.
13254         (iwmmxt_textrmuh): Likewise.
13255         (iwmmxt_textrmsh): Likewise.
13256         (iwmmxt_textrmw): Likewise.
13257         (iwmxxt_wshufh): Likewise.
13258         (eqv8qi3): Likewise.
13259         (eqv4hi3): Likewise.
13260         (eqv2si3): Likewise.
13261         (gtuv8qi3): Likewise.
13262         (gtuv4hi3): Likewise.
13263         (gtuv2si3): Likewise.
13264         (gtv8qi3): Likewise.
13265         (gtv4hi3): Likewise.
13266         (gtv2si3): Likewise.
13267         (smax<mode>3_iwmmxt): Likewise.
13268         (umax<mode>3_iwmmxt): Likewise.
13269         (smin<mode>3_iwmmxt): Likewise.
13270         (umin<mode>3_iwmmxt): Likewise.
13271         (iwmmxt_wpackhss): Likewise.
13272         (iwmmxt_wpackwss): Likewise.
13273         (iwmmxt_wpackdss): Likewise.
13274         (iwmmxt_wpackhus): Likewise.
13275         (iwmmxt_wpackwus): Likewise.
13276         (iwmmxt_wpackdus): Likewise.
13277         (iwmmxt_wunpckihb): Likewise.
13278         (iwmmxt_wunpckihh): Likewise.
13279         (iwmmxt_wunpckihw): Likewise.
13280         (iwmmxt_wunpckilb): Likewise.
13281         (iwmmxt_wunpckilh): Likewise.
13282         (iwmmxt_wunpckilw): Likewise.
13283         (iwmmxt_wunpckehub): Likewise.
13284         (iwmmxt_wunpckehuh): Likewise.
13285         (iwmmxt_wunpckehuw): Likewise.
13286         (iwmmxt_wunpckehsb): Likewise.
13287         (iwmmxt_wunpckehsh): Likewise.
13288         (iwmmxt_wunpckehsw): Likewise.
13289         (iwmmxt_wunpckelub): Likewise.
13290         (iwmmxt_wunpckeluh): Likewise.
13291         (iwmmxt_wunpckeluw): Likewise.
13292         (iwmmxt_wunpckelsb): Likewise.
13293         (iwmmxt_wunpckelsh): Likewise.
13294         (iwmmxt_wunpckelsw): Likewise.
13295         (ror<mode>3): Likewise.
13296         (ashr<mode>3_iwmmxt): Likewise.
13297         (lshr<mode>3_iwmmxt): Likewise.
13298         (ashl<mode>3_iwmmxt): Likewise.
13299         (ror<mode>3_di): Likewise.
13300         (ashr<mode>3_di): Likewise.
13301         (lshr<mode>3_di): Likewise.
13302         (ashl<mode>3_di): Likewise.
13303         (iwmmxt_wmadds): Likewise.
13304         (iwmmxt_wmaddu): Likewise.
13305         (iwmmxt_tmia): Likewise.
13306         (iwmmxt_tmiaph): Likewise.
13307         (iwmmxt_tmiabb): Likewise.
13308         (iwmmxt_tmiatb): Likewise.
13309         (iwmmxt_tmiabt): Likewise.
13310         (iwmmxt_tmiatt): Likewise.
13311         (iwmmxt_tmovmskb): Likewise.
13312         (iwmmxt_tmovmskh): Likewise.
13313         (iwmmxt_tmovmskw): Likewise.
13314         (iwmmxt_waccb): Likewise.
13315         (iwmmxt_wacch): Likewise.
13316         (iwmmxt_waccw): Likewise.
13317         (iwmmxt_waligni): Likewise.
13318         (iwmmxt_walignr): Likewise.
13319         (iwmmxt_walignr0): Likewise.
13320         (iwmmxt_walignr1): Likewise.
13321         (iwmmxt_walignr2): Likewise.
13322         (iwmmxt_walignr3): Likewise.
13323         (iwmmxt_wsadb): Likewise.
13324         (iwmmxt_wsadh): Likewise.
13325         (iwmmxt_wsadbz): Likewise.
13326         (iwmmxt_wsadhz): Likewise.
13327         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
13328         (iwmmxt_wabsdiffb): Likewise.
13329         (iwmmxt_wabsdiffh): Likewise.
13330         (iwmmxt_wabsdiffw): Likewise.
13331         (iwmmxt_waddsubhx): Likewise
13332         (iwmmxt_wsubaddhx): Likewise.
13333         (addc<mode>3): Likewise.
13334         (iwmmxt_avg4): Likewise.
13335         (iwmmxt_avg4r): Likewise.
13336         (iwmmxt_wmaddsx): Likewise.
13337         (iwmmxt_wmaddux): Likewise.
13338         (iwmmxt_wmaddsn): Likewise.
13339         (iwmmxt_wmaddun): Likewise.
13340         (iwmmxt_wmulwsm): Likewise.
13341         (iwmmxt_wmulwum): Likewise.
13342         (iwmmxt_wmulsmr): Likewise.
13343         (iwmmxt_wmulumr): Likewise.
13344         (iwmmxt_wmulwsmr): Likewise.
13345         (iwmmxt_wmulwumr): Likewise.
13346         (iwmmxt_wmulwl): Likewise.
13347         (iwmmxt_wqmulm): Likewise.
13348         (iwmmxt_wqmulwm): Likewise.
13349         (iwmmxt_wqmulmr): Likewise.
13350         (iwmmxt_wqmulwmr): Likewise.
13351         (iwmmxt_waddbhusm): Likewise.
13352         (iwmmxt_waddbhusl): Likewise.
13353         (iwmmxt_wqmiabb): Likewise.
13354         (iwmmxt_wqmiabt): Likewise.
13355         (iwmmxt_wqmiatb): Likewise.
13356         (iwmmxt_wqmiatt): Likewise.
13357         (iwmmxt_wqmiabbn): Likewise.
13358         (iwmmxt_wqmiabtn): Likewise.
13359         (iwmmxt_wqmiatbn): Likewise.
13360         (iwmmxt_wqmiattn): Likewise.
13361         (iwmmxt_wmiabb): Likewise.
13362         (iwmmxt_wmiabt): Likewise.
13363         (iwmmxt_wmiatb): Likewise.
13364         (iwmmxt_wmiatt): Likewise.
13365         (iwmmxt_wmiabbn): Likewise.
13366         (iwmmxt_wmiabtn): Likewise.
13367         (iwmmxt_wmiatbn): Likewise.
13368         (iwmmxt_wmiattn): Likewise.
13369         (iwmmxt_wmiawbb): Likewise.
13370         (iwmmxt_wmiawbt): Likewise.
13371         (iwmmxt_wmiawtb): Likewise.
13372         (iwmmxt_wmiawtt): Likewise.
13373         (iwmmxt_wmiawbbn): Likewise.
13374         (iwmmxt_wmiawbtn): Likewise.
13375         (iwmmxt_wmiawtbn): Likewise.
13376         (iwmmxt_wmiawttn): Likewise.
13377         (iwmmxt_wmerge): Likewise.
13378         (iwmmxt_tandc<mode>3): Likewise.
13379         (iwmmxt_torc<mode>3): Likewise.
13380         (iwmmxt_torvsc<mode>3): Likewise.
13381         (iwmmxt_textrc<mode>3): Likewise.
13382         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
13383         (wmmxt_pack): Likewise.
13384         (wmmxt_mult_c1): Likewise.
13385         (wmmxt_mult_c2): Likewise.
13386         (wmmxt_alu_c1): Likewise.
13387         (wmmxt_alu_c2): Likewise.
13388         (wmmxt_alu_c3): Likewise.
13389         (wmmxt_transfer_c1): Likewise.
13390         (wmmxt_transfer_c2): Likewise.
13391         (wmmxt_transfer_c3): Likewise.
13392         (marvell_f_iwmmxt_wstr): Likewise.
13393         (marvell_f_iwmmxt_wldr): Likewise.
13395 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
13397         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
13399 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13401         Revert:
13402         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13403         * lra-constraints.c (need_for_split_p): Check call used hard regs
13404         living through calls.
13406         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
13407         call used regs for call insn.
13409 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
13411         PR target/57736
13412         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
13413         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
13415 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
13417         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
13419 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
13421         * lra-constraints.c (need_for_split_p): Check call used hard regs
13422         living through calls.
13424 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
13426         PR target/57744
13427         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
13428         to tie with any other modes.  Eliminate Altivec vector mode tests,
13429         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
13430         code, to return 0 if testing MODE2 for a condition, if we've
13431         already tested MODE1 for the same condition.
13433 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13435         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
13436         layout.
13438 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13440         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
13441         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
13443 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
13445         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
13446         Define.
13447         (aarch64_symbolic_constant_p): Remove.
13448         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
13449         static.  Fix line length and white space.
13450         (aarch64_symbolic_constant_p): Remove.
13451         * config/aarch64/predicates.md (aarch64_valid_symref):
13452         Use aarch64_classify_symbol_expression.
13454 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13456         * config/arm/constraints.md (Ts): New constraint.
13457         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
13458         16-bit encodings.
13459         (compare_scc): Use "Ts" constraint for operand 0.
13460         (ior_scc_scc): Likewise.
13461         (and_scc_scc): Likewise.
13462         (and_scc_scc_nodom): Likewise.
13463         (ior_scc_scc_cmp): Likewise for operand 7.
13464         (and_scc_scc_cmp): Likewise.
13465         * config/arm/thumb2.md (thumb2_movsi_insn):
13466         Add alternatives for 16-bit encodings.
13467         (thumb2_movhi_insn): Likewise.
13468         (thumb2_movsicc_insn): Likewise.
13469         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
13470         (thumb2_negscc): Use "Ts" constraint.
13471         Move mvn instruction outside cond_exec block.
13472         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
13473         for 16-bit encodings.
13475 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13477         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
13478         encoding.
13479         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
13480         (mulsi3subsi): Likewise.
13481         (mulsidi3adddi): Likewise.
13482         (mulsidi3_v6): Likewise.
13483         (umulsidi3_v6): Likewise.
13484         (umulsidi3adddi_v6): Likewise.
13485         (smulsi3_highpart_v6): Likewise.
13486         (umulsi3_highpart_v6): Likewise.
13487         (mulhisi3tb): Likewise.
13488         (mulhisi3bt): Likewise.
13489         (mulhisi3tt): Likewise.
13490         (maddhisi4): Likewise.
13491         (maddhisi4tb): Likewise.
13492         (maddhisi4tt): Likewise.
13493         (maddhidi4): Likewise.
13494         (maddhidi4tb): Likewise.
13495         (maddhidi4tt): Likewise.
13496         (zeroextractsi_compare0_scratch): Likewise.
13497         (insv_zero): Likewise.
13498         (insv_t2): Likewise.
13499         (anddi_notzesidi_di): Likewise.
13500         (anddi_notsesidi_di): Likewise.
13501         (andsi_notsi_si): Likewise.
13502         (iordi_zesidi_di): Likewise.
13503         (xordi_zesidi_di): Likewise.
13504         (andsi_iorsi3_notsi): Likewise.
13505         (smax_0): Likewise.
13506         (smax_m1): Likewise.
13507         (smin_0): Likewise.
13508         (not_shiftsi): Likewise.
13509         (unaligned_loadsi): Likewise.
13510         (unaligned_loadhis): Likewise.
13511         (unaligned_loadhiu): Likewise.
13512         (unaligned_storesi): Likewise.
13513         (unaligned_storehi): Likewise.
13514         (extv_reg): Likewise.
13515         (extzv_t2): Likewise.
13516         (divsi3): Likewise.
13517         (udivsi3): Likewise.
13518         (arm_zero_extendhisi2addsi): Likewise.
13519         (arm_zero_extendqisi2addsi): Likewise.
13520         (compareqi_eq0): Likewise.
13521         (arm_extendhisi2_v6): Likewise.
13522         (arm_extendqisi2addsi): Likewise.
13523         (arm_movt): Likewise.
13524         (thumb2_ldrd): Likewise.
13525         (thumb2_ldrd_base): Likewise.
13526         (thumb2_ldrd_base_neg): Likewise.
13527         (thumb2_strd): Likewise.
13528         (thumb2_strd_base): Likewise.
13529         (thumb2_strd_base_neg): Likewise.
13530         (arm_negsi2): Add alternative for 16-bit encoding.
13531         (arm_one_cmplsi2): Likewise.
13533 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
13535         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
13536         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
13537         (movdfcc): Likewise.
13538         * config/arm/vfp.md (*thumb2_movsf_vfp):
13539         Disable predication for arm_restrict_it.
13540         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
13541         (*thumb2_movdfcc_vfp): Likewise.
13542         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
13543         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
13544         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
13545         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
13546         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
13547         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
13548         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
13549         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
13550         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
13551         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
13552         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
13553         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
13554         Disable predication for arm_restrict_it.
13556 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
13558         * config/i386/bmiintrin.h (_bextr_u32): New.
13559         (_bextr_u64): Ditto.
13561 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
13563         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
13564         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
13565         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
13566         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
13567         * config/mips/n32-elf.h: ...this new file.
13569 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
13571         PR target/57224
13572         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
13573         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
13575 2013-06-27  Catherine Moore  <clm@codesourcery.com>
13577         * config/mips/mips-tables.opt: Regenerate.
13578         * config/mips/mips-cpus.def: Add m14ke and m14kec.
13579         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
13580         * doc/invoke.texi: Add -m14kc.
13582 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13584         PR target/57623
13585         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
13586         constraints of operand 1 and 2.
13588         PR target/57623
13589         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
13590         to match RTL canonicalization.  Swap predicates and
13591         constraints of operand 1 and 2.
13593 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
13595         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
13596         Process OP_INOUT regs for splitting too.
13598 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
13600         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
13601         decl before the loop, initialize to NULL.
13602         (vectorizable_load): Initialize ptr_incr to NULL.
13604 2013-06-27  Martin Jambor  <mjambor@suse.cz>
13606         PR lto/57208
13607         * ipa-ref.h (ipa_maybe_record_reference): Declare.
13608         * ipa-ref.c (ipa_maybe_record_reference): New function.
13609         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
13610         * ipa-cp.c (create_specialized_node): Record potential references from
13611         aggvals.
13612         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
13614 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13616         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
13617         parameter 'mode' of type 'enum machine_mode mode'; change to pass
13618         'mode' to force_reg.
13619         (aarch64_add_offset): Update calls to aarch64_force_temporary.
13620         (aarch64_expand_mov_immediate): Likewise.
13622 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
13624         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
13625         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
13627 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
13629         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
13630         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13631         (struct machine_function): Add tbegin_p.
13632         (s390_canonicalize_comparison): Fold CC mode compares to
13633         conditional jump if possible.
13634         (s390_emit_jump): Return the emitted jump.
13635         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
13636         Handle CCRAWmode compares.
13637         (s390_option_override): Default to -mhtm if available.
13638         (s390_reg_clobbered_rtx): Handle floating point regs as well.
13639         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
13640         FPRs instead of df_regs_ever_live_p.
13641         (s390_optimize_nonescaping_tx): New function.
13642         (s390_init_frame_layout): Extend clobbered_regs array to cover
13643         FPRs as well.
13644         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
13645         (s390_expand_tbegin): New function.
13646         (enum s390_builtin): New enum definition.
13647         (code_for_builtin): New array definition.
13648         (s390_init_builtins): New function.
13649         (s390_expand_builtin): New function.
13650         (TARGET_INIT_BUILTINS): Define.
13651         (TARGET_EXPAND_BUILTIN): Define.
13652         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
13653         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
13654         (s390_alc_comparison): Likewise.
13655         * config/s390/s390-modes.def: Add CCRAWmode.
13656         * config/s390/s390.h (processor_flags): Add PF_TX.
13657         (TARGET_CPU_HTM): Define macro.
13658         (TARGET_HTM): Define macro.
13659         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
13660         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
13661         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
13662         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
13663         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
13664         (TBEGIN_MASK, TBEGINC_MASK): New constants.
13665         ("*cc_to_int"): Move up.
13666         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
13667         constants other than 0.
13668         ("*ccraw_to_int"): New insn and splitter definition.
13669         ("tbegin", "tbegin_nofloat", "tbegin_retry")
13670         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
13671         ("tx_assist"): New expander.
13672         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
13673         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
13674         * config/s390/s390.opt: Add -mhtm option.
13675         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
13676         * config/s390/htmxlintrin.h: New file.
13677         * config/s390/htmintrin.h: New file.
13678         * config/s390/s390intrin.h: New file.
13679         * doc/extend.texi: Document htm builtins.
13680         * config.gcc: Add the new header files to extra_headers.
13682 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
13684         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
13685         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
13687 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
13688             Pat Haugen  <pthaugen@us.ibm.com>
13689             Peter Bergner  <bergner@vnet.ibm.com>
13691         * config/rs6000/power8.md: New.
13692         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
13693         setting for power8 entry.
13694         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
13695         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
13696         test for Power4/Power5 only.
13697         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
13698         support.
13699         (force_new_group): Adjust comment.
13700         * config/rs6000/rs6000.md: Include power8.md.
13702 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
13704         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
13705         * config/arm/arm-protos.h (arm_max_conditional_execute): New
13706         declaration.
13707         (tune_params): Update comment.
13708         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
13709         (arm_max_conditional_execute): New function.
13710         (thumb2_final_prescan_insn): Use max_insn_skipped and
13711         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
13713 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
13715         PR tree-optimization/57705
13716         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
13717         SSA_NAME step, provided that it is not defined inside the loop.
13718         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
13719         (get_initial_def_for_induction): Handle SSA_NAME IV step.
13721 2013-06-25  Martin Jambor  <mjambor@suse.cz>
13723         PR middle-end/57670
13724         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
13725         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
13726         calls in the dump.
13727         (ipa_note_param_call): Initialize member_ptr flag.
13728         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
13729         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
13730         (ipa_write_indirect_edge_info): Stream member_ptr flag.
13731         (ipa_read_indirect_edge_info): Likewise.
13733 2013-06-25  Richard Biener  <rguenther@suse.de>
13735         PR middle-end/56977
13736         * passes.c (init_optimization_passes): Move pass_fold_builtins
13737         and pass_dce earlier with -Og.
13739 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
13741         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
13742         <BIT_FIELD_REF>: Remove trailing TAB.
13743         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
13744         remove blank line.
13746 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13748         PR tree-optimization/57358
13749         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
13750         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
13751         (ipa_analyze_params_uses): Generate pessimistic info when true.
13753 2013-06-24  Martin Jambor  <mjambor@suse.cz>
13755         PR tree-optimization/57539
13756         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
13757         global.inlined_to of the new node to it.  All callers changed.
13758         * ipa-inline-transform.c (clone_inlined_nodes): New variable
13759         inlining_into, pass it to cgraph_clone_node.
13760         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
13761         ipa_free_edge_args_substructures.
13762         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
13763         rdesc linked list.  Do not assert rdesc edges have inlined caller.
13764         Assert we have found an rdesc in the rdesc list.
13766 2013-06-24  Richard Biener  <rguenther@suse.de>
13768         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
13769         (pointer_set_lookup): Declare.
13770         (class pointer_map): New template class implementing a
13771         generic pointer to T map.
13772         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
13773         pointer_map<T>::contains, pointer_map<T>::insert,
13774         pointer_map<T>::traverse): New functions.
13775         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
13776         (pointer_set_lookup): New function.
13777         (pointer_set_contains): Use pointer_set_lookup.
13778         (pointer_set_insert): Likewise.
13779         (insert_aux): Remove.
13780         (struct pointer_map_t): Embed a pointer_set_t.
13781         (pointer_map_create): Adjust.
13782         (pointer_map_destroy): Likewise.
13783         (pointer_map_contains): Likewise.
13784         (pointer_map_insert): Likewise.
13785         (pointer_map_traverse): Likewise.
13786         * tree-streamer.h (struct streamer_tree_cache_d): Use a
13787         pointer_map<unsigned> instead of a pointer_map_t.
13788         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
13789         (streamer_tree_cache_lookup): Likewise.
13790         (streamer_tree_cache_create): Likewise.
13791         (streamer_tree_cache_delete): Likewise.
13792         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
13793         pointer_map<unsigned> instead of a pointer_map_t.
13794         (lto_init_tree_ref_encoder): Adjust.
13795         (lto_destroy_tree_ref_encoder): Likewise.
13796         * lto-section-out.c (lto_output_decl_index): Likewise.
13797         (lto_record_function_out_decl_state): Likewise.
13798         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
13800 2013-06-24  Richard Biener  <rguenther@suse.de>
13802         PR tree-optimization/57488
13803         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
13805 2013-06-24  Alan Modra  <amodra@gmail.com>
13807         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
13808         (gen_easy_altivec_constant): Likewise.
13809         * config/rs6000/predicates.md (easy_vector_constant_add_self,
13810         easy_vector_constant_msb): Likewise.
13812 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
13814         PR target/57688
13815         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
13816         add missing return true.
13818 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
13820         PR target/52483
13821         * config/sh/predicates.md (general_extend_operand): Invoke
13822         general_movsrc_operand for memory operands.
13823         (general_movsrc_operand): Allow reg+reg addressing, do not use
13824         general_operand for memory operands.
13826 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
13828         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
13829         when current target options does not apply.
13830         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
13831         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
13832         * config/i386/bmiintrin.h: Pass appropriate target
13833         attributes to header.
13834         * config/i386/mmintrin.h: Ditto.
13835         * config/i386/nmmintrin.h: Ditto.
13836         * config/i386/avx2intrin.h: Ditto.
13837         * config/i386/fxsrintrin.h: Ditto.
13838         * config/i386/tbmintrin.h: Ditto.
13839         * config/i386/xsaveintrin.h: Ditto.
13840         * config/i386/f16cintrin.h: Ditto.
13841         * config/i386/xtestintrin.h: Ditto.
13842         * config/i386/xsaveoptintrin.h: Ditto.
13843         * config/i386/bmi2intrin.h: Ditto.
13844         * config/i386/lzcntintrin.h: Ditto.
13845         * config/i386/smmintrin.h: Ditto.
13846         * config/i386/wmmintrin.h: Ditto.
13847         * config/i386/x86intrin.h: Remove all header include guards.
13848         * config/i386/prfchwintrin.h: Ditto.
13849         * config/i386/pmmintrin.h: Ditto.
13850         * config/i386/tmmintrin.h: Ditto.
13851         * config/i386/xmmintrin.h: Ditto.
13852         * config/i386/popcntintrin.h: Ditto.
13853         * config/i386/rdseedintrin.h: Ditto.
13854         * config/i386/ammintrin.h: Ditto.
13855         * config/i386/emmintrin.h: Ditto.
13856         * config/i386/immintrin.h: Remove all header include guards.
13857         * config/i386/fma4intrin.h: Ditto.
13858         * config/i386/lwpintrin.h: Ditto.
13859         * config/i386/xopintrin.h: Ditto.
13860         * config/i386/ia32intrin.h: Ditto.
13861         * config/i386/avxintrin.h: Ditto.
13862         * config/i386/rtmintrin.h: Ditto.
13863         * config/i386/fmaintrin.h: Ditto.
13864         * config/i386/mm3dnow.h: Ditto.
13866 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
13868         * common/config/i386/i386-common.c: Handle LZCNT.
13870 2013-06-22  Andi Kleen  <ak@linux.intel.com>
13872         * doc/extend.texi: Use __atomic_store_n instead of
13873         __atomic_store in HLE example.
13875 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
13877         * config/sh/sh.c: Remove <cstdlib> workaround.
13879 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13881         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
13883 2013-06-21  Andi Kleen  <ak@linux.intel.com>
13885         * doc/extend.texi: Document that __atomic_clear and
13886           __atomic_test_and_set should only be used with bool.
13888 2013-06-20  Jan Hubicka  <jh@suse.cz>
13890         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
13891         types_same_for_odr.
13892         * tree.c (decls_same_for_odr): New function.
13893         (same_for_edr): New function.
13894         (types_same_for_odr): New function.
13895         (get_binfo_at_offset): Use it.
13896         * tree.h (types_same_for_odr): Declare.
13898 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
13899             Jason Merrill  <jason@redhat.com>
13901         * system.h: Include <cstdlib> as well as <stdlib.h>.
13903 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
13905         PR target/57655
13906         * config/i386/i386.c (construct_container): Report error if
13907         long double is used with disabled x87 float returns.
13909 2013-06-20  Jan Hubicka  <jh@suse.cz>
13911         * lto-cgraph.c (input_symtab): Do not set cgraph state.
13913 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
13915         PR rtl-optimization/57425
13916         PR rtl-optimization/57569
13917         * alias.c (write_dependence_p): Remove parameters mem_mode and
13918         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
13919         Changed all callers.
13920         (canon_anti_dependence): Get comments and semantics in sync.
13921         Add parameter mem_canonicalized.  Changed all callers.
13922         * rtl.h (canon_anti_dependence): Update prototype.
13924 2013-06-20  Richard Biener  <rguenther@suse.de>
13926         * data-streamer-in.c (streamer_read_uhwi): Optimize single
13927         byte case, inline streamer_read_uchar and defer section
13928         overrun check.
13930 2013-06-20  Richard Biener  <rguenther@suse.de>
13932         PR tree-optimization/57584
13933         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
13934         SSA names into the expanded expression that take part in
13935         abnormal coalescing.
13937 2013-06-19  Sharad Singhai  <singhai@google.com>
13939         * gcov.c (print_usage): Handle new option.
13940         (process_args): Ditto.
13941         (get_gcov_intermediate_filename): New function.
13942         (output_intermediate_file): New function.
13943         (output_gcov_file): New function
13944         (generate_results): Handle new option.
13945         (release_function): Relase demangled name.
13946         (read_graph_file): Handle demangled name.
13947         (output_lines): Ditto.
13948         * doc/gcov.texi: Document gcov intermediate format.
13950 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
13952         PR bootstrap/57604
13953         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
13954         (lra_emit_add): Use the functions.  Add comment about Y as an
13955         address segment.
13957 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
13959         PR driver/57652
13960         * collect2.c (collect_atexit): New.
13961         (collect_exit): Delete.
13962         (main): Register collect_atexit with atexit.
13963         (collect_wait): Change collect_exit to exit.
13964         (do_wait): Same.
13965         * collect2.h (collect_exit): Delete.
13966         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
13968 2013-06-19  Wei Mi  <wmi@google.com>
13970         PR rtl-optimization/57518
13971         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
13972         if regno is used in paradoxical subreg.
13973         (update_equiv_regs): Check pdx_subregs[regno] before
13974         set a reg to be equivalent with a mem.
13976 2013-06-19  Matthias Klose  <doko@ubuntu.com>
13978         PR driver/57651
13979         * file-find.h (find_a_file): Add a mode parameter.
13980         * file-find.c (find_a_file): Likewise.
13981         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
13982         with X_OK for the executables.
13983         * collect2.c (main): Call find_a_file with X_OK.
13985 2013-06-19  Steve Ellcey  <sellcey@mips.com>
13987         PR target/56942
13988         * config/mips/mips.md (casesi_internal_mips16_<mode>):
13989         Use NEXT_INSN instead of next_real_insn.
13991 2013-06-19  Jan Hubicka  <jh@suse.cz>
13993         * cgraph.h (const_value_known_p): Replace by ...
13994         (ctor_for_folding): .. this one.
13995         * cgraphunit.c (process_function_and_variable_attributes): Use it.
13996         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
13997         * expr.c (expand_expr_real_1): Likewise.
13998         (string_constant): Likewise.
13999         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
14000         * ipa.c (process_references): Likewise.
14001         (symtab_remove_unreachable_nodes): Likewise.
14002         * ipa-inline-analysis.c (param_change_prob): Likewise.
14003         * gimple-fold.c (canonicalize_constructor_val): Likewise.
14004         (get_base_constructor): Likwise.
14005         * varpool.c (varpool_remove_node): Likewise.
14006         (varpool_remove_initializer): LIkewise.
14007         (dump_varpool_node): LIkwise.
14008         (const_value_known_p): Rewrite to ...
14009         (ctor_for_folding): ... this one.
14011 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
14013         PR driver/57651
14014         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
14015         PERSONALITY in $PATH derived prefixes.
14017 2013-06-19  Jeff Law  <law@redhat.com>
14019         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
14020         in comment.
14022         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
14023         (simplify_bitwise_binary): Use it to simpify certain binary ops on
14024         booleans.
14026 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
14028         * config/arm/vfp.md: Move VFP instruction classification documentation
14029         to ...
14030         * config/arm/arm.md: ... here.  Update instruction classification
14031         documentation.
14033 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
14035         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
14036         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
14037         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
14039 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
14041         PR target/57609
14042         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14043         with NEXT_INSN.  Use tablejump_p to check for jump table data
14044         insns.
14046 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
14048         PR c++/56544
14049         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
14050         that now in C++ the value is correct per the C++ standards.
14052 2013-06-19  Richard Biener  <rguenther@suse.de>
14054         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
14055         for global context.
14057 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14059         Revert:
14060         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14062         PR target/57609
14063         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14064         with next_active_insn.
14066 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
14068         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
14069         functions are inlined during failures to flag an error.
14070         * tree-inline.c (expand_call_inline): Allow the error to be flagged
14071         in early inline pass.
14073 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
14075         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
14076         in comments.
14078 2013-06-18  Julian Brown  <julian@codesourcery.com>
14080         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
14081         Permit virtual register pre-reload if !strict.
14082         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
14083         change.
14084         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
14085         prototype.
14086         * config/arm/neon.md (movmisalign<mode>): Use
14087         neon_perm_struct_or_reg_operand instead of
14088         neon_struct_or_register_operand.
14089         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
14090         neon_permissive_struct_operand instead of neon_struct_operand.
14091         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
14092         neon_vector_mem_operand.
14093         * config/arm/predicates.md (neon_struct_operand): Adjust call to
14094         neon_vector_mem_operand.
14095         (neon_permissive_struct_operand): New.
14096         (neon_struct_or_register_operand): Rename to...
14097         (neon_perm_struct_or_reg_operand): This. Adjust call to
14098         neon_vector_mem_operand.
14100 2013-06-18  Richard Biener  <rguenther@suse.de>
14102         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
14103         * lto-streamer.h: Include pointer-set.h.
14104         (struct lto_decl_slot): Remove.
14105         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
14106         Remove next_index entry.
14107         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14108         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14109         (lto_init_tree_ref_encoder): Adjust.
14110         (lto_destroy_tree_ref_encoder): Likewise.
14111         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
14112         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
14113         (lto_output_decl_index): Adjust.
14114         (lto_new_out_decl_state): Likewise.
14115         (lto_record_function_out_decl_state): Likewise.
14116         * lto-streamer-out.c (copy_function): Likewise.
14118 2013-06-18  Richard Biener  <rguenther@suse.de>
14120         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
14121         * cgraphunit.c: Include cfgloop.h.
14122         (init_lowered_empty_function): Initialize the loop tree.
14123         (assemble_thunk): Insert new BBs into loops.
14125 2013-06-18  Richard Biener  <rguenther@suse.de>
14127         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
14128         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
14129         the map from cache entry to cache index optional.
14130         (streamer_tree_cache_replace_tree): Adjust accordingly.
14131         (streamer_tree_cache_append): Likewise.
14132         (streamer_tree_cache_delete): Likewise.
14133         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
14134         streamer cache map from cache entry to cache index.
14135         * lto-streamer-out.c (create_output_block): Adjust.
14137 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
14139         * config/arm/arm.md (attribute "insn"): Move multiplication and
14140         division attributes to...
14141         (attribute "type"): ... here.  Remove mult.
14142         (attribute "mul32"): New attribute.
14143         (attribute "mul64"): Add umaal.
14144         (*arm_mulsi3): Update attributes.
14145         (*arm_mulsi3_v6): Likewise.
14146         (*thumb_mulsi3): Likewise.
14147         (*thumb_mulsi3_v6): Likewise.
14148         (*mulsi3_compare0): Likewise.
14149         (*mulsi3_compare0_v6): Likewise.
14150         (*mulsi_compare0_scratch): Likewise.
14151         (*mulsi_compare0_scratch_v6): Likewise.
14152         (*mulsi3addsi): Likewise.
14153         (*mulsi3addsi_v6): Likewise.
14154         (*mulsi3addsi_compare0): Likewise.
14155         (*mulsi3addsi_compare0_v6): Likewise.
14156         (*mulsi3addsi_compare0_scratch): Likewise.
14157         (*mulsi3addsi_compare0_scratch_v6): Likewise.
14158         (*mulsi3subsi): Likewise.
14159         (*mulsidi3adddi): Likewise.
14160         (*mulsi3addsi_v6): Likewise.
14161         (*mulsidi3adddi_v6): Likewise.
14162         (*mulsidi3_nov6): Likewise.
14163         (*mulsidi3_v6): Likewise.
14164         (*umulsidi3_nov6): Likewise.
14165         (*umulsidi3_v6): Likewise.
14166         (*umulsidi3adddi): Likewise.
14167         (*umulsidi3adddi_v6): Likewise.
14168         (*smulsi3_highpart_nov6): Likewise.
14169         (*smulsi3_highpart_v6): Likewise.
14170         (*umulsi3_highpart_nov6): Likewise.
14171         (*umulsi3_highpart_v6): Likewise.
14172         (mulhisi3): Likewise.
14173         (*mulhisi3tb): Likewise.
14174         (*mulhisi3bt): Likewise.
14175         (*mulhisi3tt): Likewise.
14176         (maddhisi4): Likewise.
14177         (*maddhisi4tb): Likewise.
14178         (*maddhisi4tt): Likewise.
14179         (maddhidi4): Likewise.
14180         (*maddhidi4tb): Likewise.
14181         (*maddhidi4tt): Likewise.
14182         (divsi3): Likewise.
14183         (udivsi3): Likewise.
14184         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
14185         (thumb2_mulsi_short_compare0): Likewise.
14186         (thumb2_mulsi_short_compare0_scratch): Likewise.
14187         * config/arm/arm1020e.md (1020mult1): Update attribute change.
14188         (1020mult2): Likewise.
14189         (1020mult3): Likewise.
14190         (1020mult4): Likewise.
14191         (1020mult5): Likewise.
14192         (1020mult6): Likewise.
14193         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
14194         change.
14195         (cortex_a15_mult64): Likewise.
14196         (cortex_a15_sdiv): Likewise.
14197         (cortex_a15_udiv): Likewise.
14198         * config/arm/arm1026ejs.md (mult1): Update attribute change.
14199         (mult2): Likewise.
14200         (mult3): Likewise.
14201         (mult4): Likewise.
14202         (mult5): Likewise.
14203         (mult6): Likewise.
14204         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
14205         (pj4_ir_div): Likewise.
14206         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
14207         (11_mult2): Likewise.
14208         (11_mult3): Likewise.
14209         (11_mult4): Likewise.
14210         (11_mult5): Likewise.
14211         (11_mult6): Likewise.
14212         (11_mult7): Likewise.
14213         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
14214         (cortex_a8_mla): Likewise.
14215         (cortex_a8_mull): Likewise.
14216         (cortex_a8_smulwy): Likewise.
14217         (cortex_a8_smlald): Likewise.
14218         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
14219         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
14220         (cortex_r4_mul_3): Likewise.
14221         (cortex_r4_mla_4): Likewise.
14222         (cortex_r4_mla_3): Likewise.
14223         (cortex_r4_smlald): Likewise.
14224         (cortex_r4_mull): Likewise.
14225         (cortex_r4_sdiv): Likewise.
14226         (cortex_r4_udiv): Likewise.
14227         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
14228         (cortex_a7_idiv): Likewise.
14229         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
14230         (9_mult2): Likewise.
14231         (9_mult3): Likewise.
14232         (9_mult4): Likewise.
14233         (9_mult5): Likewise.
14234         (9_mult6): Likewise.
14235         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
14236         (cortex_a53_sdiv): Likewise.
14237         (cortex_a53_udiv): Likewise.
14238         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
14239         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
14240         (mp626_mult2): Likewise.
14241         (mp626_mult3): Likewise.
14242         (mp626_mult4): Likewise.
14243         * config/arm/fa526.md (526_mult1): Update attribute change.
14244         (526_mult2): Likewise.
14245         * config/arm/arm-generic.md (mult): Update attribute change.
14246         (mult_ldsched_strongarm): Likewise.
14247         (mult_ldsched): Likewise.
14248         (multi_cycle): Likewise.
14249         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
14250         * config/arm/fa606te.md (606te_mult1): Update attribute change.
14251         (606te_mult2): Likewise.
14252         (606te_mult3): Likewise.
14253         (606te_mult4): Likewise.
14254         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
14255         (cortex_a9_mac16): Likewise.
14256         (cortex_a9_multiply): Likewise.
14257         (cortex_a9_mac): Likewise.
14258         (cortex_a9_multiply_long): Likewise.
14259         * config/arm/fa626te.md (626te_mult1): Update attribute change.
14260         (626te_mult2): Likewise.
14261         (626te_mult3): Likewise.
14262         (626te_mult4): Likewise.
14264 2013-06-18  Richard Biener  <rguenther@suse.de>
14266         PR lto/57334
14267         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
14269 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14271         PR target/57609
14272         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
14273         with next_active_insn.
14275 2013-06-18  Alan Modra  <amodra@gmail.com>
14277         * config/rs6000/rs6000.h (enum data_align): New.
14278         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
14279         (DATA_ABI_ALIGNMENT): Define.
14280         (CONSTANT_ALIGNMENT): Correct comment.
14281         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
14282         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
14284 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
14286         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
14287         ATTRIBUTE_UNUSED marking.
14289 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
14291         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
14292         alternative and update.
14293         (aarch64_dup_lanedi): Delete.
14294         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
14295         * config/aarch64/aarch64-simd-builtins.def: Update.
14297 2013-06-17  Richard Biener  <rguenther@suse.de>
14299         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
14300         (lto_input_scc): Declare.
14301         (lto_input_tree_1): Likewise.
14302         (struct lto_stats_d): Add num_tree_bodies_output and
14303         num_pickle_refs_output.
14304         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
14305         (lto_read_tree_1): Split out from ...
14306         (lto_read_tree): ... this.
14307         (lto_input_scc): New function.
14308         (lto_input_tree_1): Split out from ...
14309         (lto_input_tree): ... this.  Handle LTO_tree_scc.
14310         (lto_data_in_create): Create the streamer cache without hashes.
14311         * lto-streamer-out.c (create_output_block): Create the streamer
14312         cache with hashes when not doing WPA.
14313         (lto_write_tree_1): Split out from ...
14314         (lto_write_tree): ... this.
14315         (get_symbol_initial_value): New function.
14316         (lto_output_tree_1): Split out from ...
14317         (lto_output_tree): ... this.  Write trees as series of SCCs
14318         using a DFS walk via DFS_write_tree.
14319         (struct sccs, struct scc_entry): New types.
14320         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
14321         (DFS_write_tree_body): New function.
14322         (DFS_write_tree): Likewise.
14323         (hash_tree): Likewise.
14324         (scc_entry_compare): Likewise.
14325         (hash_scc): Likewise.
14326         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
14327         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
14328         TREE_CHAIN as regular reference.
14329         (streamer_read_integer_cst): Remove.
14330         (streamer_get_pickled_tree): Adjust.
14331         * tree-streamer-out.c (streamer_write_chain): Disable streaming
14332         of DECL_EXTERNALs in BLOCK_VARS for now.
14333         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
14334         reference.
14335         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
14336         Add hash value argument and record that if hashes are recorded
14337         in the cache.
14338         (streamer_tree_cache_insert_1): Adjust.
14339         (streamer_tree_cache_insert): Likewise.
14340         (streamer_tree_cache_insert_at): Rename to ...
14341         (streamer_tree_cache_replace_tree): ... this and adjust.
14342         (streamer_tree_cache_append): Adjust.
14343         (record_common_node): Likewise.
14344         (streamer_tree_cache_create): Add argument whether to
14345         record hash values together with trees.
14346         (streamer_tree_cache_delete): Adjust.
14347         * tree-streamer.h (struct streamer_tree_cache_d): Add
14348         vector of hashes.
14349         (streamer_read_integer_cst): Remove.
14350         (streamer_tree_cache_insert): Adjust.
14351         (streamer_tree_cache_append): Likewise.
14352         (streamer_tree_cache_insert_at): Rename to ...
14353         (streamer_tree_cache_replace_tree): ... this and adjust.
14354         (streamer_tree_cache_create): Add argument whether to record hashes.
14355         (streamer_tree_cache_get): Rename to ...
14356         (streamer_tree_cache_get_tree): ... this.
14357         (streamer_tree_cache_get_hash): New function.
14358         * tree.c (cache_integer_cst): New function.
14359         * tree.h (cache_integer_cst): Declare.
14360         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
14361         * lto-symtab.c (lto_varpool_replace_node): Only release
14362         DECL_INITIAL of non-prevailing decls.
14363         * varpool.c (varpool_remove_initializer): Do not release
14364         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
14366 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14368         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
14369         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
14370         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
14371         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
14372         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
14373         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
14374         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
14375         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
14376         instead of TARGET_64BIT.
14377         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
14378         Require ISA_HAS_<D>DIV.
14380 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
14382         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
14383         (mips*-*-linux*): Move default with_llsc setting to where other
14384         defaults are set.
14385         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
14386         with_arch block.
14387         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
14388         Likewise.  Remove default with_tune setting.  Move default float
14389         setting to its own block.  Handle with_llsc in the same block as above.
14391 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
14393         PR rtl-optimization/57425
14394         PR rtl-optimization/57569
14395         * alias.c (write_dependence_p): Add new parameters mem_mode,
14396         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
14397         Changed all callers.
14398         (canon_anti_dependence): New function.
14399         * cse.c (check_dependence): Use canon_anti_dependence.
14400         * cselib.c (cselib_invalidate_mem): Likewise.
14401         * rtl.h (canon_anti_dependence): Declare.
14403 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
14405         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
14406         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
14407         ".set mips3" for 64-bit targets.
14409 2013-06-15  Dehao Chen  <dehao@google.com>
14411         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
14412         * gimple-low.c (gimple_check_call_matching_types): Likewise.
14413         (gimple_check_call_args): Likewise.
14414         * value-prof.c (check_ic_target): Likewise.
14415         * ipa-inline.c (early_inliner): Likewise.
14416         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
14417         * cgraph.c (cgraph_create_edge_1): Likewise.
14418         (cgraph_make_edge_direct): Likewise.
14420 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
14422         PR target/57615
14423         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
14424         rs6000_output_move_128bit to handle emitting quad memory
14425         operations.  Set attribute length to 8 bytes.
14427 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
14429         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
14430         New pattern.
14431         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
14432         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
14433         (aarch64_<su>mlsl<mode>): Likewise.
14435 2013-06-14  Mike Stump  <mikestump@comcast.net>
14437         * Makefile.in (TARGET_H): Add insn-codes.h.
14439 2013-06-14  Alan Modra  <amodra@gmail.com>
14441         PR middle-end/57134
14442         PR middle-end/57586
14443         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
14444         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
14445         bitfield expansion when EXPAND_MEMORY.
14446         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
14448 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
14450         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
14451         test for clearing quad memory on 32-bit later.
14453 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14455         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
14456         (fold_negate_expr): Likewise.
14457         (fold_real_zero_addition_p): Handle vectors.
14458         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
14460 2013-06-14  Alan Modra  <amodra@gmail.com>
14462         * varasm.c (force_const_mem): Revert 2013-06-07 change.
14464 2013-06-13  Jan Hubicka  <jh@suse.cz>
14466         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
14467         Local comdats are not externally visible.
14468         * symtab.c (dump_symtab_base): Dump externally visible.
14469         (verify_symtab_base): Verify back links in the symtab hash.
14471 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14473         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
14474         CONVERT_EXPR as equal nodes.
14476 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
14478         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
14480 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
14482         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
14483         Generalize to complex and vector.
14484         * tree.c (build_all_ones_cst): New function.
14485         * tree.h (build_all_ones_cst): Declare it.
14487 2013-06-13  Alan Modra  <amodra@gmail.com>
14489         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
14490         * config/rs6000/rs6000.md (signbittf2): New insn.
14491         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
14492         (abstf2_internal, cmptf_internal2): Likewise.
14493         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
14495 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
14496             Pat Haugen  <pthaugen@us.ibm.com>
14497             Peter Bergner  <bergner@vnet.ibm.com>
14499         * config/rs6000/rs6000.c (emit_load_locked): Add support for
14500         power8 byte, half-word, and quad-word atomic instructions.
14501         (emit_store_conditional): Likewise.
14502         (rs6000_expand_atomic_compare_and_swap): Likewise.
14503         (rs6000_expand_atomic_op): Likewise.
14505         * config/rs6000/sync.md (larx): Add new modes for power8.
14506         (stcx): Likewise.
14507         (AINT): New mode iterator to include TImode as well as normal
14508         integer modes on power8.
14509         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
14510         that VSX registers are not considered.  Use AINT mode iterator
14511         instead of INT1 to allow inclusion of quad word atomic operations
14512         on power8.
14513         (load_locked<mode>): Likewise.
14514         (store_conditional<mode>): Likewise.
14515         (atomic_compare_and_swap<mode>): Likewise.
14516         (atomic_exchange<mode>): Likewise.
14517         (atomic_nand<mode>): Likewise.
14518         (atomic_fetch_<fetchop_name><mode>): Likewise.
14519         (atomic_nand_fetch<mode>): Likewise.
14520         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
14521         each type.
14522         (ATOMIC): On power8, add QImode, HImode modes.
14523         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
14524         modes that promote to SImode.
14525         (load_lockedti): Convert TImode arguments to PTImode, so that we
14526         get a guaranteed even/odd register pair.
14527         (load_lockedpti): Likewise.
14528         (store_conditionalti): Likewise.
14529         (store_conditionalpti): Likewise.
14531         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
14532         atomic load/store instructions.
14533         (HSI): Likewise.
14535 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
14537         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
14538         loads.
14539         (insn_count): New attribute, with most cases extracted from...
14540         (length): ...here.  Redefine most cases in terms of insn_count.
14541         (single_insn): Delete.
14542         (can_delay): Use insn_count to check for single instructions.
14543         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
14544         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
14545         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
14546         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
14547         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
14548         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
14549         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
14550         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
14551         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
14552         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
14553         rather than "length".
14554         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
14555         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
14556         Use "insn_count" rather than "length".
14557         * config/mips/mips-dsp.md
14558         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
14559         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
14560         length attributes.
14562 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
14564         PR tree-optimization/57361
14565         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
14567 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
14569         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
14570         to split.
14571         (aarch64_simd_combine<mode>): New instruction expansion.
14572         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
14573         function prototype.
14574         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
14575         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
14577 2013-06-12  Jan Hubicka  <jh@suse.cz>
14579         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
14580         decl has when in streaming stage.
14581         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
14582         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
14584 2013-06-12  Roland Stigge  <stigge@antcom.de>
14586         PR target/57578
14587         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
14589 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14591         PR tree-optimization/57537
14592         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
14593         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
14595 2013-06-12  Richard Biener  <rguenther@suse.de>
14597         * data-streamer.h (streamer_write_char_stream): CSE
14598         obs->current_pointer.
14599         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
14600         streamer_write_char_stream manually and optimize the resulting loop.
14601         (streamer_write_hwi_stream): Likewise.
14603 2013-06-12  Jan Hubicka  <jh@suse.cz>
14605         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
14606         * cgraph.h (varpool_create_empty_node): Declare.
14607         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
14608         duplicated nodes.
14609         * symtab.c (symtab_unregister_node): Be lax about missin entries
14610         in node hash.
14611         (symtab_get_node): Update comment.
14612         * varpool.c (varpool_create_empty_node): Break out from ...
14613         (varpool_node_for_decl): ... here.
14614         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
14616 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
14618         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
14619         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
14620         part.  Use straight-line flow at the end.
14621         <COMPONENT_REF>: Remove superfluous else.
14622         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
14624 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
14626         PR target/56564
14627         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
14628         target hook even for !TREE_PUBLIC decls.  If no resolution info
14629         is available, return false for common and external decls.
14631 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
14633         * config/rl78/constraints.md (U): New constraint.
14634         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
14635         valloc attribute.
14637 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
14639         PR target/57589
14640         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
14641         to allow returning address to AT_PLATFORM name.
14643 2013-06-11  Jan Hubicka  <jh@suse.cz>
14645         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
14646         * cgraph.h (symtab_node_base): Add weakref flag.
14647         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
14648         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
14649         (output_weakrefs): Use weakref flag.
14650         * fold-const.c (simple_operand_p): Handle WEAK.
14651         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
14652         * ipa.c (varpool_externally_visible_p): Drop weakref.
14653         (function_and_variable_visibility): Update comment; fix weakref
14654         sanity checks; do not clear DECL_WEAK on them.
14655         * lto-cgraph.c (lto_output_node): update.
14656         (lto_output_varpool_node): Update.
14657         (input_overwrite_node): Update.
14658         (input_node): Update.
14659         (input_varpool_node): Update.
14660         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
14661         (lto_symtab_merge_symbols): Add sanity check.
14662         (lto_symtab_prevailing_decl): Do not special case weakrefs.
14663         * passes.c (rest_of_decl_compilation): Set static flag, too.
14664         * symtab.c (dump_symtab_base): Dump weakref.
14665         (verify_symtab_base): Sanity check weakrefs.
14666         (symtab_make_decl_local): Remove duplicated code.
14667         (symtab_alias_ultimate_target): Simplify.
14668         * varpool.c (varpool_create_variable_alias): Set weakref flag.
14670 2013-06-11  Tom de Vries  <tom@codesourcery.com>
14672         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
14673         sequence_vect == NULL.
14675 2013-06-11  DJ Delorie  <dj@redhat.com>
14677         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
14678         (rl78_unwind_word_mode): New.
14680 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14682         * final.c (debug_prefix_maps): Make static.
14684 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
14686         * function.c (initial_trampoline): Remove stray copy.
14688 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
14690         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
14692 2013-06-11  Martin Jambor  <mjambor@suse.cz>
14694         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
14695         within bounds at the beginning of the function.
14697 2013-06-11  Alan Modra  <amodra@gmail.com>
14699         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
14700         reporting.
14701         (get_named_section): Don't NULL !DECL_P decl.
14703 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
14705         * doc/invoke.texi (core-avx2): Document.
14706         (slm): Likewise.
14707         (atom): Updated with MOVBE.
14709 2013-06-11  Richard Biener  <rguenther@suse.de>
14711         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
14713 2013-06-11  Anton Blanchard  <anton@samba.org>
14715         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
14716         correct shift value in little-endian mode.
14718 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
14720         PR target/56564
14721         * varasm.c (get_variable_align): Move #endif to the right place.
14723 2013-06-10  Cary Coutant  <ccoutant@google.com>
14725         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
14726         for hash so that hash table traversal order is deterministic.
14728 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
14729             Pat Haugen  <pthaugen@us.ibm.com>
14730             Peter Bergner  <bergner@vnet.ibm.com>
14732         * config/rs6000/vector.md (GPR move splitter): Do not split moves
14733         of vectors in GPRS if they are direct moves or quad word load or
14734         store moves.
14736         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
14737         declaration.
14738         (direct_move_p): Likewise.
14739         (quad_load_store_p): Likewise.
14741         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
14742         classes into bins based on the physical register type.
14743         (reg_class_to_reg_type): Likewise.
14744         (IS_STD_REG_TYPE): Likewise.
14745         (IS_FP_VECT_REG_TYPE): Likewise.
14746         (reload_fpr_gpr): Arrays to determine what insn to use if we can
14747         use direct move instructions.
14748         (reload_gpr_vsx): Likewise.
14749         (reload_vsx_gpr): Likewise.
14750         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
14751         information that is a simplification of register classes.  Also
14752         precalculate direct move reload helpers.
14753         (direct_move_p): New function to return true if the operation can
14754         be done as a direct move instruciton.
14755         (quad_load_store_p): New function to return true if the operation
14756         is a quad memory operation.
14757         (rs6000_legitimize_address): If quad memory, only allow register
14758         indirect for TImode addresses.
14759         (rs6000_legitimate_address_p): Likewise.
14760         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
14761         (rs6000_reload_register_type): Likewise.
14762         (register_to_reg_type): Return register type.
14763         (rs6000_secondary_reload_simple_move): New helper function for
14764         secondary reload and secondary memory needed to identify anything
14765         that is a simple move, and does not need reloading.
14766         (rs6000_secondary_reload_direct_move): New helper function for
14767         secondary reload to identify cases that can be done with several
14768         instructions via the direct move instructions.
14769         (rs6000_secondary_reload_move): New helper function for secondary
14770         reload to identify moves between register types that can be done.
14771         (rs6000_secondary_reload): Add support for quad memory operations
14772         and for direct move.
14773         (rs6000_secondary_memory_needed): Likewise.
14774         (rs6000_debug_secondary_memory_needed): Change argument names.
14775         (rs6000_output_move_128bit): New function to return the move to
14776         use for 128-bit moves, including knowing about the various
14777         limitations of quad memory operations.
14779         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
14780         memory operations.  call rs6000_output_move_128bit for the actual
14781         instruciton(s) to generate.
14782         (vsx_movti_64bit): Likewise.
14784         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
14785         (UNSPEC_P8V_MTVSRWZ): Likewise.
14786         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
14787         (UNSPEC_P8V_MTVSRD): Likewise.
14788         (UNSPEC_P8V_XXPERMDI): Likewise.
14789         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
14790         (UNSPEC_FUSION_GPR): Likewise.
14791         (FMOVE128_GPR): New iterator for direct move.
14792         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
14793         (f32_sv): Likewise.
14794         (f32_dm): Likewise.
14795         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
14796         loads and direct move instructions.
14797         (zero_extendsidi2_lfiwzx): Likewise.
14798         (extendsidi2_lfiwax): Likewise.
14799         (extendsidi2_nocell): Likewise.
14800         (floatsi<mode>2_lfiwax): Likewise.
14801         (lfiwax): Likewise.
14802         (floatunssi<mode>2_lfiwzx): Likewise.
14803         (lfiwzx): Likewise.
14804         (fix_trunc<mode>_stfiwx): Likewise.
14805         (fixuns_trunc<mode>_stfiwx): Likewise.
14806         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
14807         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
14808         (parity<mode>2_cmpb): Set length/type attr.
14809         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
14810         for 'mr.' to fast_compare.
14811         (bpermd_<mode>): Change type attr to popcnt.
14812         (p8_fmrgow_<mode>): New insns for power8 direct move support.
14813         (p8_mtvsrwz_1): Likewise.
14814         (p8_mtvsrwz_2): Likewise.
14815         (reload_fpr_from_gpr<mode>): Likewise.
14816         (p8_mtvsrd_1): Likewise.
14817         (p8_mtvsrd_2): Likewise.
14818         (p8_xxpermdi_<mode>): Likewise.
14819         (reload_vsx_from_gpr<mode>): Likewise.
14820         (reload_vsx_from_gprsf): Likewise.
14821         (p8_mfvsrd_3_<mode>): LIkewise.
14822         (reload_gpr_from_vsx<mode>): Likewise.
14823         (reload_gpr_from_vsxsf): Likewise.
14824         (p8_mfvsrd_4_disf): Likewise.
14825         (multi-word GPR splits): Do not split direct moves or quad memory
14826         operations.
14828 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
14830         * tree-into-ssa.c (interesting_blocks): Make static.
14832 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
14834         PR target/56564
14835         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
14836         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
14837         Use DATA_ABI_ALIGNMENT for that case instead if defined.
14838         (get_variable_align): New function.
14839         (get_variable_section, emit_bss, emit_common,
14840         assemble_variable_contents, place_block_symbol): Use
14841         get_variable_align instead of DECL_ALIGN.
14842         (assemble_noswitch_variable): Add align argument, use it
14843         instead of DECL_ALIGN.
14844         (assemble_variable): Adjust caller.  Use get_variable_align
14845         instead of DECL_ALIGN.
14846         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
14847         caller.
14848         (DATA_ABI_ALIGNMENT): Define.
14849         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
14850         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
14851         opt is false, only return the psABI mandated alignment increase.
14852         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
14853         (DATA_ABI_ALIGNMENT): ... this.
14854         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
14855         (DATA_ABI_ALIGNMENT): ... this.
14856         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
14857         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
14858         (DATA_ABI_ALIGNMENT): ... this.
14859         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
14860         * doc/tm.texi: Regenerated.
14862 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
14864         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
14865         cmp_code to construct REG_EQUAL note.
14867 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
14869         PR target/57568
14870         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
14871         that operands[2] doesn't overlap with operands[0].
14873 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
14874             Jan Hubicka  <jh@suse.cz>
14876         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
14877         hack to mark symbols as used.
14879 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
14881         PR rtl-optimization/57559
14882         * lra-constraints.c (process_alt_operands): Don't discourage
14883         memory with known offset for offsetable memory constraint.
14884         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
14886 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
14888         * varasm.c (struct oc_local_state): Reorder fields.
14889         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
14890         and adjust accordingly.
14891         (output_constructor): Reorder initialization code and adjust call to
14892         output_constructor_bitfield.
14894 2013-06-07  Jan Hubicka  <jh@suse.cz>
14896         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
14898 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
14900         * tree-object-size.c (unknown): Make const.
14902 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
14904         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
14905         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
14906         for last alternative in the cpu_facility attribute.
14908 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14910         PR target/56315
14911         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
14912         (xordi3): Change operand 2 constraint to arm_xordi_operand.
14913         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
14914         * config/arm/constraints.md (Dg): New constraint.
14915         * config/arm/neon.md (xordi3_neon): Remove.
14916         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
14917         * config/arm/predicates.md (arm_xordi_operand): New predicate.
14919 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14921         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
14922         Clean up alternatives.
14924 2013-06-07  Alan Modra  <amodra@gmail.com>
14926         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
14927         va_list_gpr_size.
14929 2013-06-07  Alan Modra  <amodra@gmail.com>
14931         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
14933 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14935         * config/arm/constraints.md (Df): New constraint.
14936         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
14937         Correct length attribute for last two alternatives.
14939 2013-06-07  Alan Modra  <amodra@gmail.com>
14941         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
14942         override user -mfp-in-toc.
14943         (offsettable_ok_by_alignment): Consider just the current access
14944         rather than the whole object, unless BLKmode.  Handle
14945         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
14946         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
14947         for -mcmodel=medium.
14948         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
14949         override user -mfp-in-toc or -msum-in-toc.  Default to
14950         -mno-fp-in-toc for -mcmodel=medium.
14952 2013-06-06  DJ Delorie  <dj@redhat.com>
14954         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
14955         TARGET_VALID_POINTER_MODE.
14957 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
14958             Pat Haugen  <pthaugen@us.ibm.com>
14959             Peter Bergner  <bergner@vnet.ibm.com>
14961         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
14962         Document new power8 builtins.
14964         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
14965         condition code register, to allow 128-bit logical operations to be
14966         done in the VSX or GPR registers.
14967         (nor<mode>3): Use the canonical form for nor.
14968         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
14969         vclz*, and vpopcnt* vector instructions.
14970         (nand<mode>3): Likewise.
14971         (orc<mode>3): Likewise.
14972         (clz<mode>2): LIkewise.
14973         (popcount<mode>2): Likewise.
14975         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
14976         that only the GPRs are recognized.
14978         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14979         support for new power8 builtins.
14981         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
14982         builtin functions.
14983         (xscvdpspn): Likewise.
14984         (vclz): Likewise.
14985         (vclzb): Likewise.
14986         (vclzh): Likewise.
14987         (vclzw): Likewise.
14988         (vclzd): Likewise.
14989         (vpopcnt): Likewise.
14990         (vpopcntb): Likewise.
14991         (vpopcnth): Likewise.
14992         (vpopcntw): Likewise.
14993         (vpopcntd): Likewise.
14994         (vgbbd): Likewise.
14995         (vmrgew): Likewise.
14996         (vmrgow): Likewise.
14997         (eqv): Likewise.
14998         (eqv_v16qi3): Likewise.
14999         (eqv_v8hi3): Likewise.
15000         (eqv_v4si3): Likewise.
15001         (eqv_v2di3): Likewise.
15002         (eqv_v4sf3): Likewise.
15003         (eqv_v2df3): Likewise.
15004         (nand): Likewise.
15005         (nand_v16qi3): Likewise.
15006         (nand_v8hi3): Likewise.
15007         (nand_v4si3): Likewise.
15008         (nand_v2di3): Likewise.
15009         (nand_v4sf3): Likewise.
15010         (nand_v2df3): Likewise.
15011         (orc): Likewise.
15012         (orc_v16qi3): Likewise.
15013         (orc_v8hi3): Likewise.
15014         (orc_v4si3): Likewise.
15015         (orc_v2di3): Likewise.
15016         (orc_v4sf3): Likewise.
15017         (orc_v2df3): Likewise.
15019         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
15020         allow power8 quad mode in 64-bit.
15021         (rs6000_builtin_vectorized_function): Add support to vectorize
15022         ISA 2.07 count leading zeros, population count builtins.
15023         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
15024         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
15025         (builtin_function_type): Add vgbbd builtin function which takes an
15026         unsigned argument.
15027         (altivec_expand_vec_perm_const): Add support for new power8 merge
15028         instructions.
15030         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
15031         that does not include TImdoe for use with 32-bit.
15032         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
15033         instructions.
15034         (UNSPEC_VSX_CVDPSPN): Likewise.
15035         (vsx_xscvdpspn): Likewise.
15036         (vsx_xscvspdpn): Likewise.
15037         (vsx_xscvdpspn_scalar): Likewise.
15038         (vsx_xscvspdpn_directmove): Likewise.
15039         (vsx_and<mode>3): Split logical operations into 32-bit and
15040         64-bit. Add support to do logical operations on TImode as well as
15041         VSX vector types.  Allow logical operations to be done in either
15042         VSX registers or in general purpose registers in 64-bit mode.  Add
15043         splitters if GPRs were used. For AND, add clobber of CCmode to
15044         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
15045         encoding.
15046         (vsx_and<mode>3_32bit): Likewise.
15047         (vsx_and<mode>3_64bit): Likewise.
15048         (vsx_ior<mode>3): Likewise.
15049         (vsx_ior<mode>3_32bit): Likewise.
15050         (vsx_ior<mode>3_64bit): Likewise.
15051         (vsx_xor<mode>3): Likewise.
15052         (vsx_xor<mode>3_32bit): Likewise.
15053         (vsx_xor<mode>3_64bit): Likewise.
15054         (vsx_one_cmpl<mode>2): Likewise.
15055         (vsx_one_cmpl<mode>2_32bit): Likewise.
15056         (vsx_one_cmpl<mode>2_64bit): Likewise.
15057         (vsx_nor<mode>3): Likewise.
15058         (vsx_nor<mode>3_32bit): Likewise.
15059         (vsx_nor<mode>3_64bit): Likewise.
15060         (vsx_andc<mode>3): Likewise.
15061         (vsx_andc<mode>3_32bit): Likewise.
15062         (vsx_andc<mode>3_64bit): Likewise.
15063         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
15064         and xxlorc instructions.
15065         (vsx_eqv<mode>3_64bit): Likewise.
15066         (vsx_nand<mode>3_32bit): Likewise.
15067         (vsx_nand<mode>3_64bit): Likewise.
15068         (vsx_orc<mode>3_32bit): Likewise.
15069         (vsx_orc<mode>3_64bit): Likewise.
15071         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
15073         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
15074         instruction.
15075         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
15076         (p8_vmrgow): Likewise.
15077         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
15078         GPRs to be split under VSX.
15079         (p8v_clz<mode>2): Add power8 count leading zero support.
15080         (p8v_popcount<mode>2): Add power8 population count support.
15081         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
15082         support.
15084         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
15085         instruction.
15087         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
15088         builtin functions.
15089         (vec_nand): Likewise.
15090         (vec_vclz): Likewise.
15091         (vec_vclzb): Likewise.
15092         (vec_vclzd): Likewise.
15093         (vec_vclzh): Likewise.
15094         (vec_vclzw): Likewise.
15095         (vec_vgbbd): Likewise.
15096         (vec_vmrgew): Likewise.
15097         (vec_vmrgow): Likewise.
15098         (vec_vpopcnt): Likewise.
15099         (vec_vpopcntb): Likewise.
15100         (vec_vpopcntd): Likewise.
15101         (vec_vpopcnth): Likewise.
15102         (vec_vpopcntw): Likewise.
15104 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15106         PR rtl-optimization/57468
15107         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
15108         spilled pseudos.
15110 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15112         PR rtl-optimization/57459
15113         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
15114         type when setting live regs.
15116 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
15118         * config/s390/s390.opt (mlra): New option.
15119         * config/s390/s390.c (s390_decompose_address): Check displacement
15120         for all registers for LRA.
15121         (s390_secondary_reload): Don't used secondary reloads for LRA.
15122         (s390_lra_p): New function.
15123         (TARGET_LRA_P): Define.
15124         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
15125         of attribute cpu_facility to zarch for the last alternative.
15126         (*cmpmem_short): Ditto.
15128 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
15130         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
15131         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
15132         (arm_expand_prologue): Likewise.
15134 2013-06-06  Teresa Johnson  <tejohnson@google.com>
15136         PR c++/53743
15137         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
15138         as this is now done by redirect_edge_and_branch_force.
15139         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
15140         barriers, and fix interaction with splitting.
15141         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
15142         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
15143         reflect changes made in the routine.
15144         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
15145         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
15146         since this is called in cfglayout mode, and replace partition fixup
15147         with assert as that is now done by force_nonfallthru_and_redirect.
15148         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
15149         already be marked with region crossing note.
15150         (insert_section_boundary_note): Make non-static, gate on flag
15151         has_bb_partition, rewrite to also check for multiple partitions.
15152         (rest_of_handle_reorder_blocks): Remove call to
15153         insert_section_boundary_note, now done later during free_cfg.
15154         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
15155         * bb-reorder.h (insert_section_boundary_note): Declare.
15156         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
15157         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
15158         invoke insert_section_boundary_note.
15159         (try_redirect_by_replacing_jump): Remove unnecessary
15160         check for region crossing note.
15161         (fixup_partition_crossing): New function.
15162         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
15163         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
15164         in non-cfglayout mode.
15165         (force_nonfallthru_and_redirect): Fixup partition boundaries,
15166         remove old code that tried to do this. Emit barrier correctly
15167         when we are in cfglayout mode.
15168         (last_bb_in_partition): New function.
15169         (rtl_split_edge): Correctly fixup partition boundaries.
15170         (commit_one_edge_insertion): Remove old code that tried to
15171         fixup region crossing edge since this is now handled in
15172         split_block, and set up insertion point correctly since
15173         block may now end in a jump.
15174         (verify_hot_cold_block_grouping): Guard against checking when not in
15175         linearized RTL mode.
15176         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
15177         notes.
15178         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
15179         rtl_verify_flow_info, so not called in cfglayout mode.
15180         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
15181         (fixup_reorder_chain): Remove old code that attempted to fixup region
15182         crossing note as this is now handled in force_nonfallthru_and_redirect.
15183         (duplicate_insn_chain): Don't duplicate switch section notes.
15184         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
15185         note.
15186         * basic-block.h (emit_barrier_after_bb): Declare.
15188 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15190         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
15191         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
15192         arm_usatsihi): Adjust alternatives for arm_restrict_it.
15194 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15196         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
15197         where appropriate.
15198         * config/arm/ldmstm.md: Regenerate.
15200 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15202         * config/arm/sync.md (atomic_loaddi_1):
15203         Disable predication for arm_restrict_it.
15204         (arm_load_exclusive<mode>): Likewise.
15205         (arm_load_exclusivesi): Likewise.
15206         (arm_load_exclusivedi): Likewise.
15207         (arm_load_acquire_exclusive<mode>): Likewise.
15208         (arm_load_acquire_exclusivesi): Likewise.
15209         (arm_load_acquire_exclusivedi): Likewise.
15210         (arm_store_exclusive<mode>): Likewise.
15211         (arm_store_exclusive<mode>): Likewise.
15212         (arm_store_release_exclusivedi): Likewise.
15213         (arm_store_release_exclusive<mode>): Likewise.
15215 2013-06-06  Richard Biener  <rguenther@suse.de>
15217         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
15218         after LTO_null.
15219         (lto_tag_is_tree_code_p): Adjust.
15220         (lto_tag_is_gimple_code_p): Likewise.
15221         (lto_gimple_code_to_tag): Likewise.
15222         (lto_tag_to_gimple_code): Likewise.
15223         (lto_tree_code_to_tag): Likewise.
15224         (lto_tag_to_tree_code): Likewise.
15225         * data-streamer.h (streamer_write_hwi_in_range): Use
15226         uhwi streaming to stream the normalized range.
15227         (streamer_read_hwi_in_range): Likewise.
15229 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15231         * config/arm/arm.md (enabled_for_depr_it): New attribute.
15232         (predicable_short_it): Likewise.
15233         (predicated): Likewise.
15234         (enabled): Handle above.
15235         (define_cond_exec): Set predicated attribute to yes.
15237 2013-06-05  Mike Stump  <mikestump@comcast.net>
15239         * gdbinit.in (__FUNCTION__): Add.
15241 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
15243         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
15244         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
15246 2013-06-05  Jan Hubicka  <jh@suse.cz>
15248         * varasm.c (mark_decl_referenced): Revert the removal until targets
15249         are fixed.
15251 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
15253         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
15254         instead of mark_decl_referenced.
15256 2013-06-05  Jan Hubicka  <jh@suse.cz>
15258         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
15259         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
15260         and symtab_used_from_object_file_p.
15261         (cgraph_make_node_local_1): Clear forced_by_abi.
15262         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
15263         * cgraph.h (symtab_node_base): Add forced_by_abi.
15264         (decide_is_variable_needed): Remove.
15265         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
15266         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
15267         (decide_is_symbol_needed): ... this one; handle symbols in general;
15268         always analyze virtuals; honnor forced_by_abi.
15269         (cgraph_finalize_function): Update.
15270         (varpool_finalize_decl): Update.
15271         (symbol_defined_and_needed): Remove.
15272         (analyze_functions): Update.
15273         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
15274         output_refs, input_overwrite_node): Handle forced_by_abi.
15275         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
15276         (address_taken_from_non_vtable_p): ... this one.
15277         (comdat_can_be_unshared_p_1): New function.
15278         (cgraph_comdat_can_be_unshared_p): Rename to ...
15279         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
15280         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
15281         (function_and_variable_visibility): Clear forced_by_abi as needed.
15282         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
15283         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
15284         * symtab.c (dump_symtab_base): Dump forced_by_abi.
15285         * varpool.c (decide_is_variable_needed): Remove.
15287 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15289         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
15290         (arm_option_override): Override arm_restrict_it where appropriate.
15291         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
15292         * config/arm/arm.opt (mrestrict-it): New command-line option.
15293         * doc/invoke.texi: Document -mrestrict-it.
15295 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15297         * tsan.c (tsan_atomic_table): Make const.
15299 2013-06-05  Richard Biener  <rguenther@suse.de>
15301         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
15302         index associated with the tree we are supposed to replace.
15303         * tree-streamer-out.c (pack_ts_base_value_fields): Output
15304         TREE_ASM_WRITTEN as zero for everything but SSA names.
15306 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
15308         * tree-ssa-structalias.c (call_stmt_vars): Make static.
15310 2013-06-04  Jan Hubicka  <jh@suse.cz>
15312         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
15313         (input_node, input_varpool_node): Handle correctly external same
15314         body aliases.
15315         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
15316         nodes at ltrans stage.
15318 2013-06-04  Jan Hubicka  <jh@suse.cz>
15320         * ipa-inline.c (update_caller_keys): Fix availability test.
15321         (update_callee_keys): Likewise.
15322         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
15323         to follow ELF standard.
15325 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
15327         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
15328         (mips64r5900el-*-elf*): New configurations.
15329         * config/mips/mips-cpus.def (r5900): New processor.
15330         * config/mips/mips-tables.opt: Regenerate.
15331         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
15332         (mips_issue_rate): Handle PROCESSOR_R5900.
15333         (mips_reorg_process_insns): Force reorder mode for the R5900.
15334         * config/mips/mips.h (TARGET_MIPS5900): Define.
15335         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
15336         TARGET_MIPS5900.
15337         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
15338         TARGET_MIPS5900.
15339         * config/mips/mips.md (processor): Add r5900.
15340         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
15342 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15344         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
15345         into function to generate MOVI instruction.
15346         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
15347         (aarch64_preferred_simd_mode): Turn into wrapper.
15348         (aarch64_output_scalar_simd_mov_immediate): New function.
15349         * config/aarch64/aarch64-protos.h: Add prototype for above.
15351 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15353         * config/aarch64/aarch64.c (simd_immediate_info): Remove
15354         element_char member.
15355         (sizetochar): Return signed char.
15356         (aarch64_simd_valid_immediate): Remove elchar and other
15357         unnecessary variables.
15358         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
15359         Calculate element_char as required.
15360         * config/aarch64/aarch64-protos.h: Update and move prototype
15361         for aarch64_output_simd_mov_immediate.
15362         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15363         Update arguments.
15365 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15367         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
15368         information completed by aarch64_simd_valid_immediate.
15369         (aarch64_legitimate_constant_p): Update arguments.
15370         (aarch64_simd_valid_immediate): Work with struct rather than many
15371         pointers.
15372         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
15373         (aarch64_simd_make_constant): Update arguments.
15374         (aarch64_output_simd_mov_immediate): Work with struct rather than
15375         many pointers.  Output immediate directly rather than as operand.
15376         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15377         Update prototype.
15378         * config/aarch64/constraints.md (Dn): Update arguments.
15380 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15382         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
15383         longer static.
15384         (aarch64_simd_immediate_valid_for_move): Remove.
15385         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
15386         (aarch64_simd_make_constant): Update call.
15387         (aarch64_output_simd_mov_immediate): Update call.
15388         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
15389         Add prototype.
15390         * config/aarch64/constraints.md (Dn): Update call.
15392 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
15394         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
15395         return type to bool for prototype.
15396         (aarch64_legitimate_constant_p): Check for true instead of not -1.
15397         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
15398         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
15400 2013-06-04  Catherine Moore  <clm@codesourcery.com>
15402         * config/mips/mips.opt (meva): New.
15403         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
15404         (ASM_SPEC): Handle -meva.
15405         * doc/invoke.texi (meva):  Document.
15407 2013-06-04  Alan Modra  <amodra@gmail.com>
15409         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
15410         constant output.
15412 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15414         * rtl.def: Add extra fourth optional field to define_cond_exec.
15415         * gensupport.c (process_one_cond_exec): Process attributes from
15416         define_cond_exec.
15417         * doc/md.texi: Document fourth field in define_cond_exec.
15419 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
15421         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
15422         out the processing order as in store_bit_field_1.
15424 2013-06-04  Jan Hubicka  <jh@suse.cz>
15426         PR middle-end/57500
15427         * cgraphunit.c (cgraph_process_same_body_aliases): Create
15428         non-VAR_DECL node if it does not exist yet.
15430 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
15432         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
15433         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
15434         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
15435         target_cpu_default setting.
15437 2013-06-03  Teresa Johnson  <tejohnson@google.com>
15439         * dumpfile.c (opt_info_switch_p): Change -fopt-info
15440         default to -fopt-info=optimized instead of all.
15441         * doc/invoke.texi: Ditto.
15442         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
15443         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
15444         (execute_vect_slp): Emit BB vectorization success under
15445         MSG_OPTIMIZED_LOCATIONS.
15446         * tree-vect-slp.c (vect_slp_transform_bb): Change
15447         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
15448         * tree-vect-loop.c (vect_transform_loop): Ditto.
15450 2013-06-03  Jason Merrill  <jason@redhat.com>
15452         PR c++/57415
15453         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15454         Use TARGET_EXPR for C++.
15456 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
15458         PR rtl-optimization/57268
15459         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
15460         if DEBUG_INSN_P (insn).
15462         Reapply
15463         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15465         PR rtl-optimization/57268
15466         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15467         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15469 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15471         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
15472         (ix86_avoid_lea_for_addr): Likewise.
15473         (exact_dependency_1): Likewise.
15474         (ix86_adjust_cost): Likewise.
15475         (swap_top_of_ready_list): Fix formatting and !reload_completed check
15476         removed.
15477         (do_reorder_for_imul): Fix typo, formatting and
15478         !reload_completed check removed.
15479         (ix86_sched_reorder): Fix typo and formatting.
15480         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
15481         list.
15483 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
15485         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
15487 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
15489         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
15490         <STRING_CST>: Likewise.
15491         <VECTOR_CST>: Likewise.
15493 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
15494             Mikael Morin  <mikael@gcc.gnu.org>
15496         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
15497         * config.in: Regenerated.
15498         * configure: Regenerated.
15500 2013-06-01  Jan Hubicka  <jh@suse.cz>
15502         PR middle-end/57366
15503         * cgraphunit.c (compile): When weakref is not supported,
15504         set up transparent aliases before final output pass.
15505         * varasm.c (assemble_alias): Do not try to do it here.
15507 2013-06-01  Jan Hubicka  <jh@suse.cz>
15509         PR middle-end/57467
15510         * passes.c (for_per_function): Skip unanalyzed functions.
15512 2013-06-01  Jan Hubicka  <jh@suse.cz>
15514         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
15515         (lto_symtab_merge_symbols_1): ... this one.
15516         (lto_symtab_merge_cgraph_nodes): Rename to ...
15517         (lto_symtab_merge_symbols): ... this one; simplify.
15518         * cgraph.c (same_body_aliases_done): Rename to ...
15519         (cpp_implicit_aliases_done): ... this one.
15520         (cgraph_create_function_alias): Update.
15521         (cgraph_same_body_alias): Update.
15522         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
15523         (verify_edge_corresponds_to_fndecl): Simplify.
15524         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
15525         (cgraph_node): Remove same_body_alias.
15526         (varpool_node): Remove alias_of and extra_name_alias.
15527         (same_body_aliases_done): Rename to ..
15528         (cpp_implicit_aliases_done): ... this one.
15529         (symtab_alias_ultimate_target): Add default parameter.
15530         (symtab_resolve_alias): New function.
15531         (fixup_same_cpp_alias_visibility): Declare.
15532         (cgraph_function_node): Add default parameter.
15533         (cgraph_node_asm_name): Likewise.
15534         (cgraph_function_or_thunk_node): Add default parameter; do
15535         not ICE when it is NULL.
15536         (varpool_variable_node): Likewise.
15537         * tree-emutls.c (create_emultls_var): Update.
15538         (ipa_lower_emutls): Update.
15539         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
15540         (cgraph_reset_node): Reset alias info.
15541         (cgraph_finalize_function): Update.
15542         (fixup_same_cpp_alias_visibility): Move to symtab.c.
15543         (analyze_function): Simplify.
15544         (cgraph_process_same_body_aliases): Simplify.
15545         (analyze_functions): Fixup same body aliases.
15546         (handle_alias_pairs): Simplify.
15547         (assemble_thunk): Update.
15548         (assemble_thunks_and_aliases): Update.
15549         (output_weakrefs): Rewrite.
15550         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
15551         (lto_output_varpool_node): Likewise.
15552         (compute_ltrans_boundary): Remve assert.
15553         (get_alias_symbol): New functoin.
15554         (input_node): Rewrite alias handling.
15555         (input_varpool_node): Likewise.
15556         * ipa-pure-const.c (propagate_pure_const): Fix formating.
15557         * ipa.c (process_references): Handle weakrefs correctly.
15558         (symtab_remove_unreachable_nodes): Likewise.
15559         * trans-mem.c (get_cg_data): Update.
15560         (ipa_tm_create_version_alias): Update.
15561         (ipa_tm_execute): Update.
15562         * symtab.c (dump_symtab_base): Dump aliases.
15563         (verify_symtab_base): Verify aliases.
15564         (symtab_node_availability): New function.
15565         (symtab_alias_ultimate_target): Simplify.
15566         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
15567         handle all the fixup cases.
15568         (symtab_resolve_alias): New function.
15569         * passes.c (ipa_write_summaries): Handle weakrefs.
15570         * varpool.c (varpool_analyze_node): Simplify.
15571         (assemble_aliases): Update.
15572         (varpool_create_variable_alias): Simplify.
15573         (varpool_extra_name_alias): Simplify.
15574         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
15575         (lto_symtab_merge_symbols): ... this one.
15577 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
15579         Revert
15580         PR rtl-optimization/57268
15581         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15582         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15584 2013-06-01  Tobias Burnus  <burnus@net-b.de>
15586         Partially reverted:
15587         2013-05-31  Tobias Burnus  <burnus@net-b.de>
15589         PR middle-end/57073
15590         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
15591         further up.
15593 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
15595         PR rtl-optimization/57268
15596         * sched-deps.c (sched_analyze_2): Flush dependence lists if
15597         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
15599 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15601         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
15602         unordered comparison operators when -fno-trapping-math is in effect
15603         on the e500.
15604         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
15605         and implement unordered comparison operators properly on the e500.
15607 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
15609         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
15610         for constant scalar integers.
15611         (simplify_relational_operation_1): Likewise.
15613 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
15615         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
15616         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
15617         Fix comment.
15619 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15620             Igor Zamyatin  <igor.zamyatin@intel.com>
15622         Silvermont (SLM) architecture performance tuning.
15623         * config/i386/i386.h (enum ix86_tune_indices): Add
15624         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
15625         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
15627         * config/i386/i386.c (initial_ix86_tune_features)
15628         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
15629         (ix86_lea_outperforms): Handle Silvermont tuning.
15630         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
15631         call.
15632         (ix86_use_lea_for_mov): Likewise.
15633         (ix86_avoid_lea_for_addr): Likewise.
15634         (ix86_lea_for_add_ok): Likewise.
15635         (exact_dependency_1): New function.
15636         (exact_store_load_dependency): Likewise.
15637         (ix86_adjust_cost): Handle Silvermont tuning.
15638         (do_reoder_for_imul): Likewise.
15639         (swap_top_of_ready_list): New function.
15640         (ix86_sched_reorder): Changed to handle Silvermont tuning.
15642         * config/i386/i386.md (peepholes that split memory operand in fp
15643         converts): New.
15645 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15647         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
15648         Remove un-necessary braces.
15650 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15652         * config/aarch64/aarch64.c (aarch64_classify_symbol):
15653         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
15655 2013-05-31  Tobias Burnus  <burnus@net-b.de>
15657         PR middle-end/57073
15658         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
15660 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15662         PR target/56315
15663         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
15664         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
15665         * config/arm/neon.md (iordi3_neon): Remove.
15666         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
15667         * config/arm/predicates.md (imm_for_neon_logic_operand):
15668         Move to earlier in the file.
15669         (neon_logic_op2): Likewise.
15670         (arm_iordi_operand_neon): New predicate.
15672 2013-05-31  Richard Biener  <rguenther@suse.de>
15674         PR tree-optimization/57478
15675         PR tree-optimization/57453
15676         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
15677         are life as well.
15679 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
15681         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
15682         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
15684 2013-05-30  Tobias Burnus  <burnus@net-b.de>
15685             Thomas Koenig  <tkoenig@gcc.gnu.org>
15687         PR middle-end/57073
15688         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
15689         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
15691 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
15693         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
15695 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
15697         * target.def (register_usage_leveling_p): New hook.
15698         * targhooks.c (default_register_usage_leveling_p): New.
15699         * targhooks.h (default_register_usage_leveling_p): New prototype.
15700         * lra-assigns.c (register_usage_leveling_p): Use the hook.
15701         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
15702         * doc/tm.texi: Update.
15703         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
15705 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
15707         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
15708         (*insv_reg<mode>): New define_insn.
15710 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
15712         PR rtl-optimization/57439
15713         * postreload.c (move2add_valid_value_p): Check that we have
15714         a zero subreg_regno_offset when accessing the register in
15715         the requested mode.
15717 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
15718             Igor Zamyatin  <igor.zamyatin@intel.com>
15720         Silvermont (SLM) architecture pipeline model, tuning and
15721         insn selection.
15722         * config.gcc: Add slm config options and target.
15724         * config/i386/slm.md: New.
15726         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
15728         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
15729         PROCESSOR_SLM.
15730         (ix86_target_macros_internal): Likewise.
15732         * config/i386/i386.c (slm_cost): New cost.
15733         (m_SLM): New macro flag.
15734         (initial_ix86_tune_features): Set m_SLM.
15735         (x86_accumulate_outgoing_args): Likewise.
15736         (x86_arch_always_fancy_math_387): Likewise.
15737         (processor_target_table): Add slm cost.
15738         (cpu_names): Add slm cpu name.
15739         (x86_option_override_internal): Set SLM ISA.
15740         (ix86_issue_rate): New case PROCESSOR_SLM.
15741         (ia32_multipass_dfa_lookahead): Likewise.
15742         (fold_builtin_cpu): Add slm.
15744         * config/i386/i386.h (TARGET_SLM): New target macro.
15745         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
15746         (processor_type): Add PROCESSOR_SLM.
15748         * config/i386/i386.md (cpu): Add new value "slm".
15749         (slm.md): Include slm.md.
15751 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
15752             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15754         * config/arm/arm-protos.h: Add and update function protos.
15755         * config/arm/arm.c (use_simple_return_p): New added.
15756         (thumb2_expand_return): Check simple_return flag.
15757         * config/arm/arm.md: Add simple_return and conditional simple_return.
15758         * config/arm/iterators.md: Add iterator for return and simple_return.
15760 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
15762         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
15763         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
15764         (arm_emit_vfp_multi_reg_pop): Likewise.
15765         (thumb2_emit_ldrd_pop): Likewise.
15766         (arm_expand_epilogue): Add misc REG_CFA notes.
15767         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
15769 2013-05-29  Lawrence Crowl  <crowl@google.com>
15771         * config/arm/t-arm: Update for below.
15773         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
15774         Change type to hash_table.  Update dependent calls and types.
15776         * config/i386/t-cygming: Update for below.
15778         * config/i386/t-interix: Update for below.
15780         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
15781         Change type to hash_table.  Update dependent calls and types.
15782         (i386_find_on_wrapper_list::wrappers): Likewise.
15784         * config/ia64/t-ia64: Update for below.
15786         * config/ia64/ia64.c (bundle_state_table):
15787         Change type to hash_table.  Update dependent calls and types.
15789         * config/mips/mips.c (mips_reorg_process_insns::htab):
15790         Change type to hash_table.  Update dependent calls and types.
15792         * config/sol2.c (solaris_comdat_htab):
15793         Change type to hash_table.  Update dependent calls and types.
15795         * config/t-sol2: Update for above.
15797 2013-05-29  Teresa Johnson  <tejohnson@google.com>
15799         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
15800         functions are not yet marked as defined.
15802 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
15803             Pat Haugen  <pthaugen@us.ibm.com>
15804             Peter Bergner  <bergner@vnet.ibm.com>
15806         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
15807         instructions.
15808         (VEC_A): Likewise.
15809         (VEC_C): Likewise.
15810         (vrotl<mode>3): Likewise.
15811         (vashl<mode>3): Likewise.
15812         (vlshr<mode>3): Likewise.
15813         (vashr<mode>3): Likewise.
15815         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15816         support for power8 V2DI builtins.
15818         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
15819         power8 V2DI builtins.
15820         (vupkhsw): Likewise.
15821         (vupklsw): Likewise.
15822         (vaddudm): Likewise.
15823         (vminsd): Likewise.
15824         (vmaxsd): Likewise.
15825         (vminud): Likewise.
15826         (vmaxud): Likewise.
15827         (vpkudum): Likewise.
15828         (vpksdss): Likewise.
15829         (vpkudus): Likewise.
15830         (vpksdus): Likewise.
15831         (vrld): Likewise.
15832         (vsld): Likewise.
15833         (vsrd): Likewise.
15834         (vsrad): Likewise.
15835         (vsubudm): Likewise.
15836         (vcmpequd): Likewise.
15837         (vcmpgtsd): Likewise.
15838         (vcmpgtud): Likewise.
15839         (vcmpequd_p): Likewise.
15840         (vcmpgtsd_p): Likewise.
15841         (vcmpgtud_p): Likewise.
15842         (vupkhsw): Likewise.
15843         (vupklsw): Likewise.
15844         (vaddudm): Likewise.
15845         (vmaxsd): Likewise.
15846         (vmaxud): Likewise.
15847         (vminsd): Likewise.
15848         (vminud): Likewise.
15849         (vpksdss): Likewise.
15850         (vpksdus): Likewise.
15851         (vpkudum): Likewise.
15852         (vpkudus): Likewise.
15853         (vrld): Likewise.
15854         (vsld): Likewise.
15855         (vsrad): Likewise.
15856         (vsrd): Likewise.
15857         (vsubudm): Likewise.
15859         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
15860         support for power8 V2DI instructions.
15862         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
15863         power8 V2DI instructions.  Combine pack and unpack insns to use an
15864         iterator for each mode.  Check whether a particular mode supports
15865         Altivec instructions instead of just checking TARGET_ALTIVEC.
15866         (UNSPEC_VPKUWUM): Likewise.
15867         (UNSPEC_VPKSHSS): Likewise.
15868         (UNSPEC_VPKSWSS): Likewise.
15869         (UNSPEC_VPKUHUS): Likewise.
15870         (UNSPEC_VPKSHUS): Likewise.
15871         (UNSPEC_VPKUWUS): Likewise.
15872         (UNSPEC_VPKSWUS): Likewise.
15873         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
15874         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
15875         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
15876         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
15877         (UNSPEC_VUPKHSB): Likewise.
15878         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
15879         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
15880         (UNSPEC_VUPKHSH): Likewise.
15881         (UNSPEC_VUPKLSB): Likewise.
15882         (UNSPEC_VUPKLSH): Likewise.
15883         (VI2): Likewise.
15884         (VI_char): Likewise.
15885         (VI_scalar): Likewise.
15886         (VI_unit): Likewise.
15887         (VP): Likewise.
15888         (VP_small): Likewise.
15889         (VP_small_lc): Likewise.
15890         (VU_char): Likewise.
15891         (add<mode>3): Likewise.
15892         (altivec_vaddcuw): Likewise.
15893         (altivec_vaddu<VI_char>s): Likewise.
15894         (altivec_vadds<VI_char>s): Likewise.
15895         (sub<mode>3): Likewise.
15896         (altivec_vsubcuw): Likewise.
15897         (altivec_vsubu<VI_char>s): Likewise.
15898         (altivec_vsubs<VI_char>s): Likewise.
15899         (altivec_vavgs<VI_char>): Likewise.
15900         (altivec_vcmpbfp): Likewise.
15901         (altivec_eq<mode>): Likewise.
15902         (altivec_gt<mode>): Likewise.
15903         (altivec_gtu<mode>): Likewise.
15904         (umax<mode>3): Likewise.
15905         (smax<mode>3): Likewise.
15906         (umin<mode>3): Likewise.
15907         (smin<mode>3): Likewise.
15908         (altivec_vpkuhum): Likewise.
15909         (altivec_vpkuwum): Likewise.
15910         (altivec_vpkshss): Likewise.
15911         (altivec_vpkswss): Likewise.
15912         (altivec_vpkuhus): Likewise.
15913         (altivec_vpkshus): Likewise.
15914         (altivec_vpkuwus): Likewise.
15915         (altivec_vpkswus): Likewise.
15916         (altivec_vpks<VI_char>ss): Likewise.
15917         (altivec_vpks<VI_char>us): Likewise.
15918         (altivec_vpku<VI_char>us): Likewise.
15919         (altivec_vpku<VI_char>um): Likewise.
15920         (altivec_vrl<VI_char>): Likewise.
15921         (altivec_vsl<VI_char>): Likewise.
15922         (altivec_vsr<VI_char>): Likewise.
15923         (altivec_vsra<VI_char>): Likewise.
15924         (altivec_vsldoi_<mode>): Likewise.
15925         (altivec_vupkhsb): Likewise.
15926         (altivec_vupkhs<VU_char>): Likewise.
15927         (altivec_vupkls<VU_char>): Likewise.
15928         (altivec_vupkhsh): Likewise.
15929         (altivec_vupklsb): Likewise.
15930         (altivec_vupklsh): Likewise.
15931         (altivec_vcmpequ<VI_char>_p): Likewise.
15932         (altivec_vcmpgts<VI_char>_p): Likewise.
15933         (altivec_vcmpgtu<VI_char>_p): Likewise.
15934         (abs<mode>2): Likewise.
15935         (vec_unpacks_hi_v16qi): Likewise.
15936         (vec_unpacks_hi_v8hi): Likewise.
15937         (vec_unpacks_lo_v16qi): Likewise.
15938         (vec_unpacks_hi_<VP_small_lc>): Likewise.
15939         (vec_unpacks_lo_v8hi): Likewise.
15940         (vec_unpacks_lo_<VP_small_lc>): Likewise.
15941         (vec_pack_trunc_v8h): Likewise.
15942         (vec_pack_trunc_v4si): Likewise.
15943         (vec_pack_trunc_<mode>): Likewise.
15945         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
15946         V2DI builtins.
15947         (vec_vmaxsd): Likewise.
15948         (vec_vmaxud): Likewise.
15949         (vec_vminsd): Likewise.
15950         (vec_vminud): Likewise.
15951         (vec_vpksdss): Likewise.
15952         (vec_vpksdus): Likewise.
15953         (vec_vpkudum): Likewise.
15954         (vec_vpkudus): Likewise.
15955         (vec_vrld): Likewise.
15956         (vec_vsld): Likewise.
15957         (vec_vsrad): Likewise.
15958         (vec_vsrd): Likewise.
15959         (vec_vsubudm): Likewise.
15960         (vec_vupkhsw): Likewise.
15961         (vec_vupklsw): Likewise.
15963 2013-05-29  Jan Hubicka  <jh@suse.cz>
15965         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
15966         flags; reorder rest of fields in more consistent way.
15967         (varpool_node): Remove analyzed, finalized and alias.
15968         (cgraph_ndoe): Likewise.
15969         (symtab_alias_ultimate_target): New function.
15970         (cgraph_function_node): Move offline.
15971         (cgraph_reset_node): Declare.
15972         (cgraph_comdat_can_be_unshared_p): Remove.
15973         (varpool_remove_initializer): Declare.
15974         (varpool_first_defined_variable, varpool_next_defined_variable
15975         cgraph_first_defined_function, cgraph_next_defined_function): Update.
15976         (cgraph_function_with_gimple_body_p): Update.
15977         (varpool_all_refs_explicit_p): Update.
15978         (symtab_alias_target): New function.
15979         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
15980         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
15981         (cgraph_function_or_thunk_node): Simplify using
15982         symtab_alias_ultimate_target.
15983         (varpool_variable_node): Likewise.
15984         * cgraph.c (cgraph_create_function_alias): Update.
15985         (cgraph_add_thunk): Update.
15986         (cgraph_remove_node): Update.
15987         (dump_cgraph_node): Do not dump removed flags.
15988         (cgraph_function_body_availability): Update.
15989         (cgraph_propagate_frequency): Update.
15990         (verify_cgraph_node): Check sanity of local flag.
15991         (cgraph_function_node): Move here from cgraph.h; revamp for
15992         cgraph_function_or_thunk_node.
15993         * lto-symtab.c (lto_varpool_replace_node): Update.
15994         (lto_symtab_resolve_can_prevail_p): Update.
15995         (lto_symtab_merge_cgraph_nodes): Update.
15996         * ipa-cp.c (determine_versionability, initialize_node_lattices,
15997         propagate_constants_accross_call, devirtualization_time_bonus,
15998         ipcp_propagate_stage): Update.
15999         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
16000         * ipa-inline-transform.c (clone_inlined_nodes,
16001         preserve_function_body_p): Update.
16002         * ipa-reference.c (propagate): Update.
16003         (write_node_summary_p): Update.
16004         * toplev.c (wrapup_global_declaration_2): Update.
16005         * cgraphunit.c (cgraph_analyze_function): Rename to ...
16006         (analyze_function) ... this one.
16007         (cgraph_process_new_functions): Update.
16008         (cgraph_reset_node): Export.
16009         (cgraph_finalize_function): Update.
16010         (cgraph_add_new_function): Update.
16011         (process_function_and_variable_attributes): Update.
16012         (varpool_finalize_decl): Update.
16013         (symbol_finalized): Remove.
16014         (symbol_finalized_and_needed): Rename to ...
16015         (symbol_defined_and_needed): ... update.
16016         (cgraph_analyze_functions): Update.
16017         (handle_alias_pairs): Update.
16018         (mark_functions_to_output): Update.
16019         (assemble_thunk): Update.
16020         (output_in_order): Update.
16021         (output_weakrefs): Update.
16022         (finalize_compilation_unit): Update.
16023         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
16024         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
16025         input_node, input_varpool_node): Update.
16026         * dbxout.c (dbxout_expand_expr): Update.
16027         * cgraphclones.c (cgraph_clone_node): Update.
16028         (cgraph_copy_node_for_versioning): Update.
16029         (cgraph_materialize_clone): Update.
16030         (cgraph_materialize_all_clones): Update.
16031         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
16032         propagate_pure_const, propagate_nothrow): Update.
16033         * lto-streamer-out.c (lto_output, write_symbol): Update.
16034         * ipa-utils.c (ipa_reverse_postorder): Update.
16035         * ipa-inline.c (can_inline_edge_p): Update.
16036         (update_caller_keys, ipa_inline): Update.
16037         * dwarf2out.c (reference_to_unused,
16038         premark_types_used_by_global_vars_helper): Update.
16039         * tree-eh.c (tree_could_trap_p): Update.
16040         * ipa-split.c (consider_split, execute_split_functions): Update.
16041         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
16042         has_addr_references_p): Update; move ahead in file for better
16043         readability.
16044         (process_references): Simplify.
16045         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
16046         bodies are removed.
16047         (cgraph_comdat_can_be_unshared_p): Make static.
16048         (cgraph_externally_visible_p): Update.
16049         (varpool_externally_visible_p): Update.
16050         (function_and_variable_visibility): Update.
16051         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
16052         ipa_tm_mark_force_output_node): Update.
16053         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
16054         estimate_edge_devirt_benefit, inline_generate_summary,
16055         inline_write_summary): Update.
16056         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
16057         * ipa-prop.c (ipa_compute_jump_functions): Update.
16058         (ipa_print_node_params, ipa_prop_read_section,
16059         ipa_update_after_lto_read, read_replacements_section): Update.
16060         * varasm.c (mark_decl_referenced): Update.
16061         (assemble_alias, dump_tm_clone_pairs): Update.
16062         * tree-inline.c (copy_bb): Update.
16063         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
16064         Update.
16065         * symtab.c (dump_symtab_base): Print new flags.
16066         (verify_symtab_base): Verify new flags.
16067         (symtab_alias_ultimate_target): New function.
16068         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
16069         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
16070         Update.
16071         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
16072         Update.
16073         * i386.c (ix86_get_function_versions_dispatcher,
16074         ix86_generate_version_dispatcher_body): Update.
16075         (fold_builtin_cpu): Use varpool_add_new_variable.
16076         * varpool.c (varpool_remove_initializer): Break out from ...
16077         (varpool_remove_node): ... this one.
16078         (dump_varpool_node, varpool_node_for_asm,
16079         cgraph_variable_initializer_availability, varpool_analyze_node,
16080         varpool_assemble_decl, varpool_remove_unreferenced_decls,
16081         varpool_finalize_named_section_flags, varpool_create_variable_alias):
16082         Update.
16084 2013-05-29  Jan Hubicka  <jh@suse.cz>
16086         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
16088 2013-05-29  Easwaran Raman  <eraman@google.com>
16090         PR tree-optimization/57442
16091         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
16092         when control exits the main loop.
16094 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
16096         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
16097         and RX600.
16098         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
16099         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
16100         * rx/t-rx: Add rx100 under multi library matches option for nofpu
16101         option.
16103 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16105         PR tree-optimization/57441
16106         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
16107         Don't limit size of incr_vec to number of candidates.
16109 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
16111         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
16112         and mips16 directories.
16113         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
16114         (MULTILIB_DIRNAMES): Ditto.
16115         (MULTILIB_EXCEPTIONS): Add new exceptions.
16116         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
16117         (MULTILIB_DIRNAMES): Ditto.
16118         (MULTILIB_EXCEPTIONS): Add new exceptions.
16120 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16121             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16123         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
16124         SYMBOL_TINY_ABSOLUTE.
16125         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
16126         SYMBOL_TINY_ABSOLUTE.
16127         (aarch64_expand_mov_immediate): Likewise.
16128         (aarch64_classify_symbol): Likewise.
16129         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
16130         Permit SYMBOL_TINY_ABSOLUTE.
16131         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
16133 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16134             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16136         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
16137         Refactor if/switch.  Replace gcc_assert with if.
16139 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
16141         * config/i386/i386.c (initial_ix86_tune_features): Enable
16142         FP Reassociation for AMD bdver1 and bdver2.
16144 2013-05-29  Martin Jambor  <mjambor@suse.cz>
16146         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
16147         and IMAGPART_EXPR do not occur within other handled_components.
16149 2013-05-29  Richard Biener  <rguenther@suse.de>
16151         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
16152         access on whether the use is in the BB we currently try to
16153         vectorize.
16154         (vect_bb_vectorization_profitable_p): Pass the BB we currently
16155         vectorize to vect_bb_slp_scalar_cost.
16157 2013-05-29  Richard Biener  <rguenther@suse.de>
16159         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
16160         computing scalar cost offsetted by stmts that are kept live
16161         by scalar uses.
16162         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
16163         for computation of scalar cost.
16165 2013-05-28  Steve Ellcey  <sellcey@mips.com>
16167         * config/mips/mips-cpus.def (mips32r2): Change processor type.
16169 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16171         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
16172         array notation built-in reduction functions.
16173         * doc/passes.texi (Passes): Added documentation about changes done
16174         for Cilk Plus.
16175         * doc/invoke.texi (C Dialect Options): Added documentation about
16176         the -fcilkplus flag.
16177         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
16178         (BUILTINS_DEF): Depend on cilkplus.def.
16179         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
16180         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
16181         * cilkplus.def: New file.
16183 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
16185         PR rtl-optimization/57439
16186         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
16188 2013-05-28  Easwaran Raman  <eraman@google.com>
16190         PR tree-optimization/57337
16191         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
16192         (find_insert_point): Correctly identify the insertion point
16193         when two statements with the same UID is compared.
16195 2013-05-28  Richard Biener  <rguenther@suse.de>
16197         PR tree-optimization/56787
16198         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
16199         from the list of data references.
16200         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
16201         clobbers.
16202         (vect_analyze_loop_operations): Likewise.
16203         (vect_transform_loop): Remove clobbers.
16205 2013-05-28  Martin Jambor  <mjambor@suse.cz>
16207         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
16208         and REALPART_EXPRs have scalar type.
16210 2013-05-28  Richard Biener  <rguenther@suse.de>
16212         PR tree-optimization/57411
16213         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
16214         virtual operands.
16215         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
16216         virtual operand propagation.
16218 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16220         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
16221         destination register for bmasksi_vis.
16222         (vector_init_bshuffle): Likewise.
16223         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
16225 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
16227         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
16228         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
16229         mode if the instruction isn't available in the original mode.
16230         * config/sparc/sparc.opt (mfix-ut699): New option.
16231         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
16232         (divdf3): Turn into expander.
16233         (divdf3_nofix): New insn.
16234         (divdf3_fix): Likewise.
16235         (divsf3): Disable if -mfix-ut699.
16236         (sqrtdf2): Turn into expander.
16237         (sqrtdf2_nofix): New insn.
16238         (sqrtdf2_fix): Likewise.
16239         (sqrtsf2): Disable if -mfix-ut699.
16241 2013-05-27  Richard Biener  <rguenther@suse.de>
16243         PR middle-end/57412
16244         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
16245         block for the new loop.
16247 2013-05-27  Richard Biener  <rguenther@suse.de>
16249         PR tree-optimization/57343
16250         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
16251         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
16252         (number_of_iterations_cond): Do not build the folded tree.
16254 2013-05-27  Richard Biener  <rguenther@suse.de>
16256         Revert
16257         PR middle-end/57381
16258         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16259         OEP_CONSTANT_ADDRESS_OF retained.
16261         PR tree-optimization/57417
16262         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
16263         for unchanged base.
16264         (set_ssa_val_to): Compare addresses using
16265         get_addr_base_and_unit_offset.
16267 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
16269         PR rtl-optimization/56833
16270         * postreload.c (move2add_record_mode): New function.
16271         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
16272         (move2add_use_add2_insn): Use move2add_record_sym_value.
16273         (move2add_use_add3_insn): Likewise.
16274         (reload_cse_move2add): Use move2add_valid_value_p and
16275         move2add_record_mode.  Invalidate call-clobbered and REG_INC
16276         affected regs by setting reg_mode to VOIDmode.
16277         (move2add_note_store): Don't pretend the inside of a SUBREG is
16278         the actual destination.  Invalidate single/leading registers by
16279         setting reg_mode to VOIDmode.
16280         Use move2add_record_sym_value, move2add_valid_value_p and
16281         move2add_record_mode.
16283 2013-05-27  Richard Biener  <rguenther@suse.de>
16285         PR tree-optimization/57396
16286         * tree-affine.c (double_int_constant_multiple_p): Properly
16287         return false for val == 0 and div != 0.
16289 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16291         * config/mips/mips.h: Use #elif in preprocessor conditions.
16293 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16295         PR target/53916
16296         * config/mips/constraints.md (kl): New constraint.
16297         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
16298         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
16299         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
16300         from LO for MIPS16.
16301         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
16303 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
16305         PR target/55777
16306         * config/mips/mips.c (mips_can_inline_p): New function.
16307         (TARGET_CAN_INLINE_P): Define.
16309 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
16311         * sched-int.h (ds_t, dw_t): Make unsigned int.
16312         Fix documentation that describes how all the ds_t bits are used.
16313         Reserve the last bit for delayed-branch scheduling.
16314         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
16315         (BITS_PER_DEP_WEAK): Fix definition and documentation.
16316         (gen_dep_weak_1): Remove prototype.
16317         * sched-deps.c (get_dep_weak_1): Make static.
16318         * target.def (speculate_insn, needs_block_p, gen_spec_check,
16319         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
16320         * doc/tm.texi: Regenerate.
16321         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
16323 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
16325         PR debug/56950
16326         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
16328 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
16329             Sandra Loosemore  <sandra@codesourcery.com>
16331         * config.gcc (powerpc-*): Allow native for with-cpu.
16333 2013-05-24  Jeff Law  <law@redhat.com>
16335         PR tree-optimization/57124
16336         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
16337         conversion feeding a condition if the range has an overflow
16338         if -fstrict-overflow.  Add warnings for when we do make the
16339         transformation.
16341 2013-05-24  Dehao Chen  <dehao@google.com>
16343         * tree-cfg.c (locus_discrim_map): Fix the typo.
16344         (locus_discrim_hasher): Likewise.
16345         (locus_discrim_hasher::hash): Likewise.
16346         (locus_discrim_hasher::equal): Likewise.
16348 2013-05-24  Martin Jambor  <mjambor@suse.cz>
16350         PR tree-optimization/57294
16351         * cgraph.h (ipa_record_stmt_references): Declare.
16352         * cgraphbuild.c (ipa_record_stmt_references): New function.
16353         (build_cgraph_edges): Use ipa_record_stmt_references.
16354         (rebuild_cgraph_edges): Likewise.
16355         (cgraph_rebuild_references): Likewise.
16356         * ipa-prop.c (ipa_modify_call_arguments): Discard references
16357         associated with the old statement and build references from the
16358         newly built statements.
16359         * ipa-ref.c (ipa_remove_stmt_references): New function.
16360         * ipa-ref.h (ipa_remove_stmt_references): Declare.
16362 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
16364         * lra-constraints.c (emit_spill_move): Use smaller mode for
16365         mem-mem moves.
16366         (check_and_process_move): Consider mem-reg moves for secondary
16367         too.
16368         (curr_insn_transform): Don't lose insns emitted before for
16369         secondary memory moves.
16370         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
16371         reg set up in the current insn.
16373 2013-05-24  Dehao Chen  <dehao@google.com>
16375         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
16376         hash function.
16377         (locus_descrim_hasher::equal): Likewise.
16378         (build_gimple_cfg): New discriminator assignment algorithm.
16379         (make_edges): Likewise.
16380         (next_discriminator_for_locus): Likewise.
16381         (same_line_p): Likewise.
16382         (assign_discriminators): Likewise.
16383         (make_cond_expr_edges): Likewise.
16384         (make_gimple_switch_edges): Likewise.
16385         (make_goto_expr_edges): Likewise.
16386         (make_gimple_asm_edges): Likewise.
16388 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
16390         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
16391         X format specifier to only display bottom 16 bits.
16392         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
16393         immediate to match for operand 2, since it will be masked.
16395 2013-05-24  Richard Biener  <rguenther@suse.de>
16397         PR tree-optimization/57287
16398         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
16399         all SSA names that occur in abnormal PHIs.
16401 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
16403         PR tree-ssa/57385
16404         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
16405         that index is not negative.
16407 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
16409         PR rtl-optimization/55177
16410         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
16411         (simplify_byte_swapping_operation): New.
16412         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
16413         (simplify_relational_operation_1): Deal with BSWAP.
16415 2013-05-23  Richard Henderson  <rth@redhat.com>
16417         PR target/56742
16418         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
16419         (ix86_reorg): Call it.
16421 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
16423         PR target/57379
16424         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
16425         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
16426         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
16428 2013-05-23  Christian Bruel  <christian.bruel@st.com>
16430         PR debug/57351
16431         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
16433 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16434             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16436         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
16437         * config/aarch64/constraints.md (Usa): Remove.
16438         * doc/md.texi (AArch64 Usa): Remove.
16440 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
16441             Marcus Shawcroft  <marcus.shawcroft@arm.com>
16443         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
16444         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
16445         * config/aarch64/predicates.md (aarch64_const_address): Remove.
16446         (aarch64_mov_operand): Use aarch64_mov_operand_p.
16448 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
16450         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
16451         instruction (AdvSIMD).
16452         * config/aarch64/aarch64-builtins.c
16453         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
16454         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
16456 2013-05-23  Martin Jambor  <mjambor@suse.cz>
16458         PR middle-end/57347
16459         * tree.h (contains_bitfld_component_ref_p): Declare.
16460         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
16461         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
16462         caller.
16463         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
16464         not access a bit-field.  Assert all final offsets are byte-aligned.
16466 2013-05-23  Richard Biener  <rguenther@suse.de>
16468         PR tree-optimization/57380
16469         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
16470         least one invariant or re-used load.
16471         * passes.c (init_optimization_passes): Move pass_phiprop before
16472         pass_forwprop.
16474 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
16476         * config/aarch64/aarch64-simd.md
16477         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
16479 2013-05-23  Richard Biener  <rguenther@suse.de>
16481         PR middle-end/57381
16482         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
16483         OEP_CONSTANT_ADDRESS_OF retained.
16485 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
16487         PR middle-end/57344
16488         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
16489         don't lower unit.  Handle unit not being always BITS_PER_WORD.
16491 2013-05-23  Richard Biener  <rguenther@suse.de>
16493         PR rtl-optimization/57341
16494         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
16495         instead of true_dependence.
16497 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
16499         * bb-reorder.c (branch_threshold): Make const.
16500         (exec_threshold): Ditto.
16502 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16503             Pat Haugen  <pthaugen@us.ibm.com>
16504             Peter Bergner  <bergner@vnet.ibm.com>
16506         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
16507         documentation for the power8 crypto builtins.
16509         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
16511         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
16512         macros for defining power8 builtin functions.
16513         (BU_P8V_AV_2): Likewise.
16514         (BU_P8V_AV_P): Likewise.
16515         (BU_P8V_VSX_1): Likewise.
16516         (BU_P8V_OVERLOAD_1): Likewise.
16517         (BU_P8V_OVERLOAD_2): Likewise.
16518         (BU_CRYPTO_1): Likewise.
16519         (BU_CRYPTO_2): Likewise.
16520         (BU_CRYPTO_3): Likewise.
16521         (BU_CRYPTO_OVERLOAD_1): Likewise.
16522         (BU_CRYPTO_OVERLOAD_2): Likewise.
16523         (XSCVSPDP): Fix typo, point to the correct instruction.
16524         (VCIPHER): Add power8 crypto builtins.
16525         (VCIPHERLAST): Likewise.
16526         (VNCIPHER): Likewise.
16527         (VNCIPHERLAST): Likewise.
16528         (VPMSUMB): Likewise.
16529         (VPMSUMH): Likewise.
16530         (VPMSUMW): Likewise.
16531         (VPERMXOR_V2DI): Likewise.
16532         (VPERMXOR_V4SI: Likewise.
16533         (VPERMXOR_V8HI: Likewise.
16534         (VPERMXOR_V16QI: Likewise.
16535         (VSHASIGMAW): Likewise.
16536         (VSHASIGMAD): Likewise.
16537         (VPMSUM): Likewise.
16538         (VPERMXOR): Likewise.
16539         (VSHASIGMA): Likewise.
16541         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
16542         __CRYPTO__ if the crypto instructions are available.
16543         (altivec_overloaded_builtins): Add support for overloaded power8
16544         builtins.
16546         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
16547         support for power8 crypto builtins.
16548         (builtin_function_type): Likewise.
16549         (altivec_init_builtins): Add support for builtins that take vector
16550         long long (V2DI) arguments.
16552         * config/rs6000/crypto.md: New file, define power8 crypto
16553         instructions.
16555 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
16556             Pat Haugen  <pthaugen@us.ibm.com>
16557             Peter Bergner  <bergner@vnet.ibm.com>
16559         * doc/invoke.texi (Option Summary): Add power8 options.
16560         (RS/6000 and PowerPC Options): Likewise.
16562         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
16563         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
16564         wm, wn, wr documentation.
16566         * config/rs6000/constraints.md (wm): New constraint for VSX
16567         registers if direct move instructions are enabled.
16568         (wn): New constraint for no registers.
16569         (wq): New constraint for quad word even GPR registers.
16570         (wr): New constraint if 64-bit instructions are enabled.
16571         (wv): New constraint if power8 vector instructions are enabled.
16572         (wQ): New constraint for quad word memory locations.
16574         * config/rs6000/predicates.md (const_0_to_15_operand): New
16575         constraint for 0..15 for crypto instructions.
16576         (gpc_reg_operand): If VSX allow registers in VSX registers as well
16577         as GPR and floating point registers.
16578         (int_reg_operand): New predicate to match only GPR registers.
16579         (base_reg_operand): New predicate to match base registers.
16580         (quad_int_reg_operand): New predicate to match even GPR registers
16581         for quad memory operations.
16582         (vsx_reg_or_cint_operand): New predicate to allow vector logical
16583         operations in both GPR and VSX registers.
16584         (quad_memory_operand): New predicate for quad memory operations.
16585         (reg_or_indexed_operand): New predicate for direct move support.
16587         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
16588         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
16589         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
16590         (POWERPC_MASKS): Add power8 options.
16591         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
16592         various options.
16594         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
16595         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
16597         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
16598         (-mpower8-fusion): New power8 options.
16599         (-mpower8-fusion-sign): Likewise.
16600         (-mpower8-vector): Likewise.
16601         (-mcrypto): Likewise.
16602         (-mdirect-move): Likewise.
16603         (-mquad-memory): Likewise.
16605         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
16606         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
16607         registers.
16608         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
16609         (rs6000_debug_vector_unit): Add p8_vector.
16610         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
16611         definitions.  Also print fusion state.
16612         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
16613         (rs6000_builtin_mask_calculate): Add power8 builtin support.
16614         (rs6000_option_override_internal): Add support for power8.
16615         (rs6000_common_init_builtins): Add debugging for skipped builtins
16616         if -mdebug=builtin.
16617         (rs6000_adjust_cost): Add power8 support.
16618         (rs6000_issue_rate): Likewise.
16619         (insn_must_be_first_in_group): Likewise.
16620         (insn_must_be_last_in_group): Likewise.
16621         (force_new_group): Likewise.
16622         (rs6000_register_move_cost): Likewise.
16623         (rs6000_opt_masks): Likewise.
16625         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
16626         power8 capable assembler, default to power7 options.
16627         (TARGET_DIRECT_MOVE): Likewise.
16628         (TARGET_CRYPTO): Likewise.
16629         (TARGET_P8_VECTOR): Likewise.
16630         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
16631         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
16632         (VECTOR_MEM_P8_VECTOR_P): Likewise.
16633         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
16634         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
16635         (TARGET_XSCVDPSPN): Likewise.
16636         (TARGET_XSCVSPDPN): Likewsie.
16637         (TARGET_SYNC_HI_QI): Likewise.
16638         (TARGET_SYNC_TI): Likewise.
16639         (MASK_CRYPTO): Likewise.
16640         (MASK_DIRECT_MOVE): Likewise.
16641         (MASK_P8_FUSION): Likewise.
16642         (MASK_P8_VECTOR): Likewise.
16643         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
16644         temporary used by some of the direct move instructions to get two FP
16645         temporary registers does not force creation of a stack frame.
16646         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
16647         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
16648         that any VSX registers are tieable, even if they are also an
16649         Altivec vector mode.
16650         (r6000_reg_class_enum): Add wm, wr, wv constraints.
16651         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
16652         (RS6000_BTM_CRYPTO): Likewise.
16653         (RS6000_BTM_COMMON): Likewise.
16655         * config/rs6000/rs6000.md (cpu attribute): Add power8.
16656         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
16657         (enum rs6000_vector): Add power8 vector support.
16659 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
16661         PR target/19599
16662         PR target/57340
16663         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
16664         (any_sibcall_could_use_r3): this and handle indirect calls.
16665         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
16667 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
16669         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
16671 2013-05-22  Richard Biener  <rguenther@suse.de>
16673         PR middle-end/57349
16674         * profile.c (branch_prob): Do not split blocks that are
16675         abnormally receiving from ECF_RETURNS_TWICE functions.
16677 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
16679         * recog.c (offsettable_address_addr_space_p): Fix calculation of
16680         address mode.  Move pointer mode initialization to the same place.
16682 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
16684         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
16685         while it has any effect.
16687 2013-05-21  Easwaran Raman  <eraman@google.com>
16689         PR tree-optimization/57322
16690         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
16691         UID of the statement added to the BB to be 1.
16693 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
16695         PR tree-optimization/57331
16696         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
16697         of conversion from pointer type to integral type with integer.
16699 2013-05-21  Martin Jambor  <mjambor@suse.cz>
16701         PR lto/57289
16702         * ipa-prop.c (ipa_read_node_info): Process param_used and
16703         controlled_uses in the same order as when writing.
16705 2013-05-21  Magnus Granberg  <baldrick@free.fr>
16707         PR plugins/56754
16708         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
16710 2013-05-21  Richard Biener  <rguenther@suse.de>
16712         PR tree-optimization/57318
16713         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
16714         estimate stmts with side-effects as likely eliminated.
16716 2013-05-21  Richard Biener  <rguenther@suse.de>
16718         PR tree-optimization/57330
16719         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
16720         preserve the call stmts fntype.
16722 2013-05-21  Richard Biener  <rguenther@suse.de>
16724         PR tree-optimization/57303
16725         * tree-ssa-sink.c (statement_sink_location): Improve killing
16726         stmt detection and properly handle self-assignments.
16728 2013-05-21  Christian Bruel  <christian.bruel@st.com>
16730         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
16731         spanning registers. LEAF_REG_REMAP is supported only for contiguous
16732         registers. Set register size out of the PARALLEL loop.
16734 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
16736         PR target/56547
16737         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
16738         (*fmasf4, *fmasf4_media): New insns.
16740 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16742         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
16743         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
16744         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
16745         (mips_idiv_insns): Update the comments to say that the returned
16746         instruction counts are in units of BASE_INSN_LENGTH.
16747         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
16748         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
16749         using 2 rather than 4 as the length of indirect MIPS16 and
16750         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
16751         length of a NOP.  Don't divide MIPS16 lengths by 2.
16752         (mips16_split_long_branches): Assume a branch is long if the
16753         length is greater than 4 rather than 8.
16754         * config/mips/mips.md (length): Give MIPS16 lengths directly,
16755         rather than multiplying them by 2.  Multiply instruction counts
16756         by BASE_INSN_LENGTH rather than 4.
16757         (*jump_mips16, tls_get_tp_mips16_<mode>)
16758         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
16760 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16762         * config/mips/mips.md (extended_mips16): Remove branch case.
16763         (length): Remove duplicated extended_mips16 test.
16765 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
16767         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
16769 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
16771         * recog.h (Recog_data): Rename to...
16772         (recog_data_d): ...this.
16773         (recog_data): Update accordingly.
16774         * recog.c (recog_data): Likewise.
16775         * reload.c (save_recog_data): Likewise.
16776         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
16777         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
16779 2013-05-17  Julian Brown  <julian@codesourcery.com>
16781         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
16782         found in a REG_EQUAL note, invalidate it.
16784 2013-05-17  Easwaran Raman  <eraman@google.com>
16786         * tree-ssa-reassoc.c (find_insert_point): New function.
16787         (insert_stmt_after): Likewise.
16788         (get_def_stmt): Likewise.
16789         (ensure_ops_are_available): Likewise.
16790         (not_dominated_by): Likewise.
16791         (rewrite_expr_tree): Do not move statements beyond what is
16792         necessary. Remove call to swap_ops_for_binary_stmt...
16793         (reassociate_bb): ... and move it here.
16794         (build_and_add_sum): Assign UIDs for new statements.
16795         (linearize_expr): Likewise.
16796         (do_reassoc): Renumber gimple statement UIDs.
16798 2013-05-17  Jan Hubicka  <jh@suse.cz>
16800         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
16801         weakrefs.
16802         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
16803         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
16804         weakrefs.
16805         (output_weakrefs): Update.
16807 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
16808             Martin Jambor  <mjambor@suse.cz>
16810         PR middle-end/57276
16811         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
16812         value that corresponds to the given aggval is found in values vector.
16814 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
16816         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
16817         sse, sse2, sse3, ssse3 and sse4a flags to options.
16819 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
16821         * gengtype-state.c: (s_expr_writer): New class, to handle
16822         prettifying of output layout of s-expressions.
16823         (state_writer): New class, to write out gtype.state.
16824         (state_written_type_count): Move this variable into member data of
16825         state_writer.
16826         (s_expr_writer::s_expr_writer): New code: constructor for new class
16827         (state_writer::state_writer(): ditto
16828         (s_expr_writer::write_new_line): New function
16829         (s_expr_writer::write_any_indent): ditto
16830         (s_expr_writer::begin_s_expr): ditto
16831         (s_expr_writer::end_s_expr): ditto
16832         (write_state_fileloc): convert to method of state_writer...
16833         (state_writer:: write_state_fileloc): ...and use methods of
16834         s_expr_writer to write indentation into the gtype.state output file
16835         to visually represent the hierarchical structure of the list
16836         structures
16837         (write_state_fields): ditto, renaming to...
16838         (state_writer::write_state_fields)
16839         (write_state_a_string): ditto, renaming to...
16840         (state_writer::write_state_a_string)
16841         (write_state_string_option): ditto, renaming to...
16842         (state_writer::write_state_string_option)
16843         (write_state_type_option): ditto, renaming to...
16844         (state_writer::write_state_type_option)
16845         (write_state_nested_option): ditto, renaming to...
16846         (state_writer::write_state_nested_option)
16847         (write_state_option): ditto, renaming to...
16848         (state_writer::write_state_option)
16849         (write_state_options): ditto, renaming to...
16850         (state_writer::write_state_options)
16851         (write_state_lang_bitmap): ditto, renaming to...
16852         (state_writer::write_state_lang_bitmap)
16853         (write_state_version): ditto, renaming to...
16854         (state_writer::write_state_version)
16855         (write_state_scalar_type): ditto, renaming to...
16856         (state_writer::write_state_scalar_type)
16857         (write_state_string_type): ditto, renaming to...
16858         (state_writer::write_state_string_type)
16859         (write_state_undefined_type): ditto, renaming to...
16860         (state_writer::write_state_undefined_type)
16861         (write_state_struct_union_type): ditto, renaming to...
16862         (state_writer::write_state_struct_union_type)
16863         (write_state_struct_type): ditto, renaming to...
16864         (state_writer::write_state_struct_type)
16865         (write_state_user_struct_type): ditto, renaming to...
16866         (state_writer::write_state_user_struct_type)
16867         (write_state_lang_struct_type): ditto, renaming to...
16868         (state_writer::write_state_lang_struct_type)
16869         (write_state_param_struct_type): ditto, renaming to...
16870         (state_writer::write_state_param_struct_type)
16871         (write_state_pointer_type): ditto, renaming to...
16872         (state_writer::write_state_pointer_type)
16873         (write_state_array_type): ditto, renaming to...
16874         (state_writer::write_state_array_type)
16875         (write_state_gc_used): ditto, renaming to...
16876         (state_writer::write_state_gc_used)
16877         (write_state_common_type_content): ditto, renaming to...
16878         (state_writer::write_state_common_type_content)
16879         (write_state_type): ditto, renaming to...
16880         (state_writer::write_state_type)
16881         (write_state_pair_list): ditto, renaming to...
16882         (state_writer::write_state_pair_list)
16883         (write_state_pair): ditto, renaming to...
16884         (state_writer::write_state_pair)
16885         (write_state_typedefs): ditto, renaming to...
16886         (state_writer::write_state_typedefs)
16887         (write_state_structures): ditto, renaming to...
16888         (state_writer::write_state_structures)
16889         (write_state_param_structs): ditto, renaming to...
16890         (state_writer::write_state_param_structs)
16891         (write_state_variables): ditto, renaming to...
16892         (state_writer::write_state_variables)
16893         (write_state_srcdir): ditto, renaming to...
16894         (state_writer::write_state_srcdir)
16895         (write_state_files_list): ditto, renaming to...
16896         (state_writer::write_state_files_list)
16897         (write_state_languages): ditto, renaming to...
16898         (state_writer::write_state_languages)
16899         (write_state): create a state_writer instance and use it when
16900         writing out the state file
16902 2013-05-17  Mike Stump  <mikestump@comcast.net>
16904         PR rtl-optimization/57304
16905         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
16906         accessing DF_REF_REAL_LOC.
16908 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16910         PR rtl-optimization/57281
16911         PR rtl-optimization/57300
16912         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
16913         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
16914         what the other splitter did if the registers are dead.
16916 2013-05-17  Richard Biener  <rguenther@suse.de>
16918         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
16919         MEM_REF offsets.
16921 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16923         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
16924         linking.
16926 2013-05-17  Marek Polacek  <polacek@redhat.com>
16928         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
16929         length when doing non-zero store of storing '\0' to '\0'.
16931 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
16933         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
16934         vect_external_def oprnd1 with loop_vinfo, try to emit
16935         optional cast, negation and and stmts on the loop preheader
16936         edge instead of into the pattern def seq.
16938         PR tree-optimization/57051
16939         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
16940         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
16942 2013-05-16  Nick Clifton  <nickc@redhat.com>
16944         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
16945         (rl78_is_naked_func): New function.
16946         (rl78_expand_prologue): Skip prologue generation for naked functions.
16947         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
16948         * doc/extend.texi (naked): Add RL78 to the list of processors
16949         that supports this attribute.
16951 2013-05-16  Jeff Law  <law@redhat.com>
16953         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
16955 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16957         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
16958         cache parameters using detect_caches_amd also for CYRIX,
16959         NSC and TM2 signatures.
16961 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
16962             Dzianis Kahanovich  <mahatma@eu.by>
16964         PR target/45359
16965         PR target/46396
16966         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
16967         VIA/Centaur processors and determine their cache parameters
16968         using detect_caches_amd.
16970 2013-05-16  Teresa Johnson  <tejohnson@google.com>
16972         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
16973         (rtl_verify_edges): New function.
16974         (rtl_verify_bb_insns): Ditto.
16975         (rtl_verify_bb_pointers): Ditto.
16976         (rtl_verify_bb_insn_chain): Ditto.
16977         (rtl_verify_fallthru): Ditto.
16978         (rtl_verify_bb_layout): Ditto.
16979         (rtl_verify_flow_info_1): Outline checks into new functions.
16980         (rtl_verify_flow_info): Ditto.
16982 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
16984         * cfghooks.c (copy_bbs): Add update_dominance argument.
16985         * cfghooks.h (copy_bbs): Update prototype.
16986         * tree-cfg.c (gimple_duplicate_sese_region):
16987         Add update_dominance argument.
16988         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
16989         * tree-ssa-loop-ch.c (copy_loop_headers): Update
16990         gimple_duplicate_sese_region call.
16991         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
16992         Update copy_bbs call.
16993         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
16994         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
16996 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
16998         * tree-vectorizer.h (NUM_PATTERNS): Increment.
16999         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
17000         vect_recog_rotate_pattern.
17001         (vect_recog_rotate_pattern): New function.
17003 2013-05-16  Jason Merrill  <jason@redhat.com>
17005         * Makefile.in (LLINKER): New variable.
17006         (mostlyclean): Remove link mutex.
17007         * configure.ac: Handle --enable-link-mutex.
17008         * lock-and-run.sh: New script.
17010 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17012         PR target/19599
17013         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
17014         for NULL decl.
17016 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17018         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
17020 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
17022         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
17023         * config/arm/arm.c (next_consecutive_mem): New function.
17024         (gen_movmem_ldrd_strd): Likewise.
17025         * config/arm/arm.md (movmemqi): Update condition and code.
17026         (unaligned_loaddi, unaligned_storedi): New patterns.
17028 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17030         * config.gcc: Obsolete *-*-solaris2.9*.
17031         * doc/install.texi (Specific, *-*-solaris2*): Document it.
17033 2013-05-16  Richard Biener  <rguenther@suse.de>
17035         * passes.c (init_optimization_passes): Move pass_parallelize_loops
17036         earlier, after GRAPHITE transforms and IV canonicalization.
17038 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
17040         * omp-low.c (extract_omp_for_data): For collapsed loops,
17041         if at least one of the loops is known at compile time to
17042         iterate zero times, set count to 0.
17043         (expand_omp_regimplify_p): New function.
17044         (expand_omp_for_generic): For collapsed loops, if at least
17045         one of the loops isn't known to iterate at least once,
17046         add runtime check with setting count to 0.
17047         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
17048         For unsigned types if it isn't known at compile time that
17049         the loop will iterate at least once, add runtime check to bypass
17050         the whole loop if initial condition isn't true.
17052 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
17054         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
17056 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
17058         PR middle-end/57286
17059         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
17060         transformations to avoid an infinite loop.
17062 2013-05-16  Marek Polacek  <polacek@redhat.com>
17064         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
17066 2013-05-15  Leif Ekblad  <leif@rdos.net>
17068         * config/i386/i386.c (ix86_decompose_address): Use
17069         DEFAULT_TLS_SEG_REG to access TLS segment register.
17070         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
17071         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
17072         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
17074 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
17076         PR target/57260
17077         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
17078         sibling calls to functions that would normally be lazily bound,
17079         unless $gp is call-clobbered.
17081 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
17083         * config/i386/i386.c (ix86_option_override_internal): Update
17084         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
17085         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
17086         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
17087         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
17088         of TARGET_3DNOW.
17089         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
17091 2013-05-15  Andreas Schwab  <schwab@suse.de>
17093         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
17094         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
17095         third operand.
17097 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17099         * loop-unroll.c (report_unroll_peel): Check decision before
17100         emitting unroll/peel message.
17102 2013-05-15  Teresa Johnson  <tejohnson@google.com>
17104         * function.h (has_bb_partition): New rtl_data flag.
17105         (bb_reorder_complete): Ditto.
17106         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
17107         instead of flag_reorder_blocks_and_partition.
17108         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
17109         with some enhancements.
17110         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
17111         * bb-reorder.c (connect_traces): Check for has_bb_partition
17112         instead of flag_reorder_blocks_and_partition.
17113         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
17114         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
17115         verify_hot_cold_block_grouping.
17116         (partition_hot_cold_basic_blocks): Set has_bb_partition.
17118 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
17120         PR target/19599
17121         * config/arm/predicates.md (call_insn_operand): New predicate.
17122         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
17123         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
17124         if insn is not a tail call.
17125         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
17126         registers.
17127         * config/arm/arm.h (enum reg_class): New caller save register class.
17128         (REG_CLASS_NAMES): Likewise.
17129         (REG_CLASS_CONTENTS): Likewise.
17130         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
17131         without decls.
17133 2013-05-15  Richard Biener  <rguenther@suse.de>
17135         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
17136         of MSG_OPTIMIZED_LOCATIONS.
17137         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
17138         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
17139         message.
17140         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
17141         of MSG_OPTIMIZED_LOCATIONS.
17142         (execute_vect_slp): Likewise.
17143         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
17144         (vect_create_cond_for_alias_checks): Likewise.
17145         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17146         (vect_recog_widen_mult_pattern): Likewise.
17147         (vect_recog_widen_sum_pattern): Likewise.
17148         (vect_recog_over_widening_pattern): Likewise.
17149         (vect_recog_widen_shift_pattern): Likewise.
17150         (vect_recog_vector_vector_shift_pattern): Likewise.
17151         (vect_recog_divmod_pattern): Likewise.
17152         (vect_recog_mixed_size_cond_pattern): Likewise.
17153         (vect_recog_bool_pattern): Likewise.
17154         (vect_pattern_recog_1): Likewise.
17156 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17158         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
17159         non-functions to builtin_unreachable.
17160         * ipa-inline-transform.c (inline_call): Do not assert estimates were
17161         correct when new direct edges were discovered.
17163 2013-05-15  Martin Jambor  <mjambor@suse.cz>
17165         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
17166         header, print symbol order instead of node uid, print more information
17167         about indirect edge targets.
17168         (ipa_make_edge_direct_to_target): Print symbol order instead of node
17169         uids.
17170         (ipa_make_edge_direct_to_target): Likewise.
17171         (remove_described_reference): Likewise.
17172         (propagate_controlled_uses): Likewise.
17173         (ipa_print_node_params): Also print symbol order.
17174         (ipcp_transform_function): Print symbol order instead of node uids.
17175         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
17176         (cgraph_get_create_real_symbol_node): Likewise.
17177         * ipa-cp.c (print_lattice): Likewise.
17178         (print_all_lattices): Likewise.
17179         (determine_versionability): Likewise.
17180         (initialize_node_lattices): Likewise.
17181         (estimate_local_effects): Likewise.
17182         (update_profiling_info): Likewise.
17183         (create_specialized_node): Likewise.
17184         (perhaps_add_new_callers): Likewise.
17185         (decide_about_value): Likewise.
17186         (decide_whether_version_node): Likewise.
17187         (identify_dead_nodes): Likewise.
17188         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
17189         (dump_inline_summary): Likewise.
17190         (estimate_node_size_and_time): Likewise.
17191         (inline_analyze_function): Likewise.
17192         * ipa-inline.c (report_inline_failed_reason): Likewise.
17193         (want_early_inline_function_p): Likewise.
17194         (edge_badness): Likewise.
17195         (update_edge_key): Likewise.
17196         (inline_small_functions): Likewise.  Add dumping of order to two other
17197         dumps.
17198         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
17199         instead of node uids.
17200         (propagate_pure_const): Likewise.
17201         (propagate_pure_const): Likewise.
17202         * ipa-utils.c (dump_cgraph_node_set): Likewise.
17203         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
17204         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
17205         of node uids.
17206         * tree-pretty-print.c (dump_function_header): Likewise.
17207         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
17208         Print symbol order instead of node uids.
17210 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17212         * config/s390/s390.c (s390_register_move_cost): Don't impose the
17213         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
17215 2013-05-15  Richard Biener  <rguenther@suse.de>
17217         PR tree-optimization/57275
17218         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
17219         return value for fail to do runtime alias checks for gather loads.
17221 2013-05-15  Jan Hubicka  <jh@suse.cz>
17223         PR lto/57038
17224         PR lto/47375
17225         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
17226         weakrefs are not external.
17227         (lto_symtab_merge_decls): Fix thinko when dealing with
17228         non-lto_symtab decls.
17229         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
17230         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
17231         * varpool.c (dump_varpool_node): Dump more flags.
17233 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17235         * config/i386/i386.c (processor_alias_table): Add instruction
17236         FSGSBASE for AMD bdver3 architecture.
17238 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17240         * tree.c (warn_deprecated_use): Print file:line using locus color.
17241         * diagnostic.c (diagnostic_report_current_module): Print file:line
17242         and file:line:column using locus color.
17244 2013-05-14  Mike Stump  <mikestump@comcast.net>
17246         * gdbinit.in: Add __null.
17248 2013-05-14  Mike Stump  <mikestump@comcast.net>
17250         * recog.h: Rename struct recog_data to Recog_data.
17251         * recog.c: Likewise.
17252         * reload.c (can_reload_into): Likewise.
17253         * config/picochip/picochip.c: Likewise.
17255 2013-05-14  Mike Stump  <mikestump@comcast.net>
17257         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
17259 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
17261         * resource.h (struct resources): Remove unch_memory member.
17262         (CLEAR_RESOURCE): Don't clear unch_memory.
17263         * resource.c (mark_referenced_resources): Don't set it.
17264         (mark_set_resources): Likewise.
17265         (mark_target_live_regs): Don't clear it.
17266         (init_resource_info): Likewise.
17267         * reorg.c (resource_conflicts_p): Don't compare it.
17268         (redundant_insn): Don't set it.
17270         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
17271         Remove prototypes.
17272         * emit-rtl.c (next_label): Remove unused function.
17273         (skip_consecutive_labels, link_cc0_insns): Move to ...
17274         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
17275         only place where these functions are used, and make them static.
17277 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17279         * fold-const.c (fold_negate_expr): Handle vectors.
17280         (fold_truth_not_expr): Make it static.
17281         (fold_invert_truthvalue): New static function.
17282         (invert_truthvalue_loc): Handle vectors. Do not call
17283         fold_truth_not_expr directly.
17284         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
17285         <TRUTH_NOT_EXPR>: Do not cast to boolean.
17286         (fold_comparison): Handle vector constants.
17287         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
17288         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
17289         * tree.h (fold_truth_not_expr): Remove declaration.
17291 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
17293         * config/aarch64/aarch64-simd.md
17294         (aarch64_vcond_internal<mode>): Rename to...
17295         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
17296         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
17297         float modes. Clarify all iterator modes.
17298         (vcond<mode><mode>): Use new name for vcond expanders.
17299         (vcond<v_cmp_result><mode>): Likewise.
17300         (vcondu<mode><mode>: Likewise.
17301         * config/aarch64/iterators.md (VDQF_COND): New.
17303 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
17305         PR bootstrap/57266
17306         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
17307         variable for the shift amount. Check that we shift by non-negative
17308         amounts.
17310 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
17312         PR target/42017
17313         * config/arm/arm.h (EPILOGUE_USES): Only return true
17314         for LR_REGNUM after epilogue_completed.
17316 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
17318         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
17319         is error_mark_node.
17321 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
17323         PR target/57261
17324         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
17325         and Solaris 11+/x86 with gld.
17326         * configure: Regenerate.
17328 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17330         * expmed.c (expand_shift_1): Canonicalize rotates by
17331         constant bitsize / 2 to bitsize - 1.
17332         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
17333         case ROTATERT>: Likewise.
17335         Revert:
17336         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17338         * config/i386/i386.md (rotateinv): New code attr.
17339         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17340         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17341         roll $31, %eax, etc.
17343 2013-05-14  Richard Biener  <rguenther@suse.de>
17345         PR middle-end/57235
17346         * tree-eh.c (sink_clobbers): Give up for successors with
17347         multiple predecessors and no virtual uses.
17349 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
17351         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
17352         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
17354 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
17356         PR middle-end/57251
17357         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
17358         the case when both op0 and op1 have VOIDmode.
17360 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17362         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
17363         in multiply-accumulate mode.
17365 2013-05-13  Guozhi Wei  <carrot@google.com>
17367         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
17369 2013-05-13  Kai Tietz  <ktietz@redhat.com>
17371         PR target/56975
17372         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
17373         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
17374         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
17375         * config/i386/i386.c (ix86_option_override_internal): Likewise.
17376         (ix86_expand_prologue): Likewise.
17377         (ix86_expand_split_stack_prologue): Likewise.
17378         (legitimate_pic_address_disp_p): Likewise.
17379         (legitimize_pic_address): Likewise.
17380         (legitimize_tls_address): Likewise.
17381         (legitimize_pe_coff_symbol): Likewise.
17382         (output_pic_addr_const): Likewise.
17383         (construct_plt_address): Likewise.
17384         (ix86_expand_call): Likewise.
17385         (x86_output_mi_thunk): Likewise.
17386         (x86_function_profiler): Likewise.
17388 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
17390         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
17391         similar switch cases.
17392         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
17393         (aarch64_simd_mov_to_<mode>low): Delete.
17394         (aarch64_simd_mov_to_<mode>high): Delete.
17395         (move_lo_quad_<mode>): Add w<-r alternative.
17396         (aarch64_simd_move_hi_quad_<mode>): Likewise.
17397         (aarch64_simd_mov_from_*): Update type attribute.
17398         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
17399         statement.
17401 2013-05-13  Jan Hubicka  <jh@suse.cz>
17403         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
17404         * config/i386/i386.c (ix86_compute_frame_layout,
17405         ix86_expand_epilogue, emit_i387_cw_initialization,
17406         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
17407         ix86_local_alignment): Fix use of size/speed predicates.
17409 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17411         PR tree-optimization/45216
17412         PR tree-optimization/57157
17413         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
17414         the (-Y) & (B - 1) variant if OP is |.
17415         * expmed.c (expand_shift_1): For rotations by const0_rtx just
17416         return shifted.  Use (-op1) & (prec - 1) as other_amount
17417         instead of prec - op1.
17419 2013-05-13  Martin Jambor  <mjambor@suse.cz>
17421         PR middle-end/42371
17422         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
17423         (ipa_constant_data): New type.
17424         (ipa_jump_func): Use ipa_constant_data to hold information about
17425         constant jump functions.
17426         (ipa_get_jf_constant): Adjust to jump function type changes.
17427         (ipa_get_jf_constant_rdesc): New function.
17428         (ipa_param_descriptor): New field controlled_uses.
17429         (ipa_get_controlled_uses): New function.
17430         (ipa_set_controlled_uses): Likewise.
17431         * ipa-ref.h (ipa_find_reference): Declare.
17432         * ipa-prop.c (ipa_cst_ref_desc): New type.
17433         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
17434         changes.
17435         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
17436         New parameter cs.  Adjust all callers.
17437         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
17438         (remove_described_reference): New function.
17439         (jfunc_rdesc_usable): Likewise.
17440         (try_make_edge_direct_simple_call): Decrement controlled use count,
17441         attempt to remove reference if it hits zero.
17442         (combine_controlled_uses_counters): New function.
17443         (propagate_controlled_uses): Likewise.
17444         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
17445         (ipa_edge_duplication_hook): Duplicate reference descriptions.
17446         (ipa_print_node_params): Print described use counter.
17447         (ipa_write_jump_function): Adjust to jump function type changes.
17448         (ipa_read_jump_function): New parameter CS, pass it to
17449         ipa_set_jf_constant.  Adjust caller.
17450         (ipa_write_node_info): Stream controlled use count
17451         (ipa_read_node_info): Likewise.
17452         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
17453         asserting.
17454         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
17455         count.  Remove cloning-added reference if it reaches zero.
17456         * ipa-ref.c (ipa_find_reference): New function.
17458 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
17460         * config/i386/i386.c (processor_target_table): Modified default
17461         alignment values for AMD BD and BT architectures.
17463 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
17465         * tree-vect-generic.c (uniform_vector_p): Move ...
17466         * tree.c (uniform_vector_p): ... here.
17467         * tree.h (uniform_vector_p): Declare it.
17468         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
17469         into a scalar.
17471 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
17473         PR tree-optimization/57230
17474         * tree-ssa-strlen.c (handle_char_store): Record length for
17475         array store from STRING_CST.
17477         PR tree-optimization/57230
17478         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
17479         check.
17481 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
17483         * config/epiphany/epiphany.c (epiphany_init): Check size of
17484         NUM_MODES_FOR_MODE_SWITCHING.
17485         (epiphany_expand_prologue):
17486         Remove CONFIG_REGNUM initial value handling code.
17487         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
17488         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
17489         (emit_set_fp_mode, epiphany_mode_after): Likewise.
17490         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
17491         Don't return 1 for FP_MODE_NONE.
17492         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
17493         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
17494         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
17495         * config/epiphany/epiphany.md (save_config): New pattern.
17497 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
17499         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
17501 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
17503         * config/i386/i386.md (memory): Handle sseishft1.
17504         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
17505         (*vec_extractv2di_1): Ditto.
17507 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
17509         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
17510         saved registers.
17512 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
17514         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
17515         Add mthumb/march=armv7-a multilib.
17516         Add mthumb/march=armv7-r multilib.
17517         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
17519 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
17521         * config/v850/t-rtems: Add more multilibs.
17523 2013-05-10  Richard Biener  <rguenther@suse.de>
17525         PR tree-optimization/57214
17526         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
17527         not propagate from SSA names that occur in abnormal PHI nodes.
17529 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
17531         * stor-layout.c (element_precision): New function.
17532         * machmode.h (element_precision): Declare it.
17533         * tree.c (build_minus_one_cst): New function.
17534         (element_precision): Likewise.
17535         * tree.h (build_minus_one_cst): Declare new function.
17536         (element_precision): Likewise.
17537         * fold-const.c (operand_equal_p): Use element_precision.
17538         (fold_binary_loc): Handle vector types.
17539         * convert.c (convert_to_integer): Use element_precision.
17540         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
17541         separately.
17543 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
17545         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
17546         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
17547         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
17548         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17549         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
17550         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
17551         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
17552         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
17553         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
17554         (Uuw8): New constraints.
17555         (Usb4): Move into alphabetical order.
17556         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
17557         (sd8_operand, ub8_operand, uw8_operand): New predicates.
17558         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
17559         previously unnamed patterns.
17560         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
17561         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
17562         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
17563         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
17564         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
17565         of specific lengths.
17567 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
17569         * config/i386/i386.md (rotateinv): New code attr.
17570         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
17571         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
17572         roll $31, %eax, etc.
17574         PR tree-optimization/45216
17575         PR tree-optimization/57157
17576         * tree-ssa-forwprop.c (simplify_rotate): New function.
17577         (ssa_forward_propagate_and_combine): Call it.
17579 2013-05-10  Richard Biener  <rguenther@suse.de>
17581         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
17582         disable peeling when we version for aliasing.
17583         (vector_alignment_reachable_p): Honor explicit user alignment.
17584         (vect_supportable_dr_alignment): Likewise.
17585         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
17586         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
17587         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
17588         then peeling to arrange for the cost-model check to come first.
17590 2013-05-10  Alan Modra  <amodra@gmail.com>
17592         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
17593         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
17594         * configure: Regenerate.
17596 2013-05-10  Alan Modra  <amodra@gmail.com>
17598         PR target/55033
17599         * varasm.c (default_elf_select_section): Move !DECL_P check..
17600         (get_named_section): ..to here before calling get_section_name.
17601         Adjust assertion.
17602         (default_section_type_flags): Add DECL_P check.
17603         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
17604         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
17606 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
17608         * config/epiphany/epiphany.c (epiphany_expand_prologue):
17609         When using gen_stack_adjust_str with a register offset, add a
17610         REG_FRAME_RELATED_EXPR note.
17612 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
17614         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
17615         (*vec_extractv4si_zext_mem): Ditto.
17616         (*vec_extractv2di): Add 0->x and x->x alternatives.
17617         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
17618         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
17620 2013-05-09  Jason Merrill  <jason@redhat.com>
17622         N3639 C++1y VLA support
17623         * gimplify.c (gimplify_vla_decl): Don't touch an existing
17624         DECL_VALUE_EXPR.
17626         * tree.c (build_constructor_va): New.
17627         * tree.h: Declare it.
17629 2013-05-09  Martin Jambor  <mjambor@suse.cz>
17631         PR lto/57084
17632         * gimple-fold.c (canonicalize_constructor_val): Call
17633         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
17635 2013-05-09  Jan Hubicka  <jh@suse.cz>
17636             Richard Biener  <rguenther@suse.de>
17638         PR lto/54095
17639         * symtab.c (symtab_make_decl_local): Do not add private names.
17641 2013-05-09  Jan Hubicka  <jh@suse.cz>
17643         PR lto/54095
17644         * symtab.c (insert_to_assembler_name_hash): Handle clones.
17645         (unlink_from_assembler_name_hash): Likewise.
17646         (symtab_prevail_in_asm_name_hash, symtab_register_node,
17647         symtab_unregister_node, symtab_initialize_asm_name_hash,
17648         change_decl_assembler_name): Update.
17650 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
17652         * config/aarch64/aarch64.md: New movtf split.
17653         (*movtf_aarch64): Update.
17654         (aarch64_movdi_tilow): Handle TF modes and rename to
17655         aarch64_movdi_<mode>low.
17656         (aarch64_movdi_tihigh): Handle TF modes and rename to
17657         aarch64_movdi_<mode>high
17658         (aarch64_movtihigh_di): Handle TF modes and rename to
17659         aarch64_mov<mode>high_di
17660         (aarch64_movtilow_di): Handle TF modes and rename to
17661         aarch64_mov<mode>low_di
17662         (aarch64_movtilow_tilow): Remove spurious whitespace.
17663         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
17664         splits.
17665         (aarch64_print_operand): Update.
17667 2013-05-09  Alan Modra  <amodra@gmail.com>
17669         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
17670         powerpc64le.
17671         * configure: Regenerate.
17673 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17675         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
17676         splitter preparation statements.
17677         * config/i386/sse.md (*vec_extract* splitters): Ditto.
17678         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
17679         adjust_address_nv.
17681 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17683         * gimple-ssa-strength-reduction.c (count_candidates): Change
17684         return value to int.
17685         (analyze_candidates_and_replace): Change type of length to int.
17687 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
17689         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
17690         (*vec_extract<mode>): Use VI12_128 mode iterator.
17691         (*vec_extract<mode>_mem): Ditto.
17692         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
17693         attribute.
17695 2013-05-08  Diego Novillo  <dnovillo@google.com>
17697         PR bootstrap/54659
17699         Revert:
17700         2012-08-17  Diego Novillo  <dnovillo@google.com>
17702         PR bootstrap/54281
17703         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
17704         * config.in: Regenerate.
17705         * configure: Regenerate.
17706         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
17708 2013-05-08  Jan Hubicka  <jh@suse.cz>
17710         PR lto/54095
17711         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
17712         * cgraph.h (symtab_node_base): Add unique_name.
17713         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
17714         input_overwrite_node, input_varpool_node): Stream unique_name.
17715         * cgraphclones.c (cgraph_create_virtual_clone,
17716         cgraph_function_versioning): Set unique_name.
17717         * ipa.c (function_and_variable_visibility): Set unique_name.
17719 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17721         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
17722         (alloc_cand_and_find_basis): Restrict conditional candidate
17723         processing to CAND_MULTs.
17725 2013-05-08  Jan Hubicka  <jh@suse.cz>
17727         PR lto/54095
17728         lto-symtab.c (lto_symtab_symbol_p): New function.
17729         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
17730         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
17731         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
17732         Skip static symbols.
17734 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
17736         PR tree-optimization/57200
17737         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
17738         Only call inform if the preceding warning_at returns true.
17740 2013-05-07  Han Shen  <shenhan@google.com>
17742         * cfgexpand.c (record_or_union_type_has_array_p): New function.
17743         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
17744         * common.opt (fstack-protector-strong): New option.
17745         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
17746         * doc/invoke.texi (Optimization Options): Document
17747         "-fstack-protector-strong".
17748         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
17750 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17752         * config/mips/mips.c (mips_machine_reorg2): Return 0.
17754 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
17756         * ira.c (update_equiv_regs): Add insn having equiv memory even if
17757         it is not lhs of the insn.
17758         (setup_reg_equiv): Remove insn having equiv memory which it is not
17759         lhs of the insn.
17760         * lra-constraints.c (process_address): Try to improve generation
17761         code for address base + disp.
17762         (lra_constraints): Make correct the code for checking insn setting
17763         up backward equivalence.  Remove insn only if it is in the init
17764         insn list.
17765         * lra-eliminations.c (update_reg_eliminate): Change return value.
17766         (lra_eliminate): Use the result.
17768 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
17770         * config/i386/sse.md (ssescalarnummask): New mode attribute.
17771         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
17772         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
17773         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
17774         register target operands.
17775         (*vec_extractv8hi_sse2): New pattern.
17776         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
17777         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
17778         (*vec_extract<mode>_mem): New insn and split pattern.
17780 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
17782         * config/arm/arm.c (arm_asan_shadow_offset): New function.
17783         (TARGET_ASAN_SHADOW_OFFSET): Define.
17784         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
17785         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
17787 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17789         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
17790         (incr_vec_index): Return -1 if increment not found.
17791         (create_add_on_incoming_edge): Assert if increment not found.
17792         (record_increment): Limit number of increments recorded.
17793         (all_phi_incrs_profitable): Return false if an increment not found.
17794         (replace_profitable_candidates): Don't process increments that were
17795         not recorded.
17796         (analyze_candidates_and_replace): Limit size of incr_vec.
17798 2013-05-07  Richard Biener  <rguenther@suse.de>
17800         * calls.c (special_function_p): setjmp-like functions are leaf.
17801         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
17802         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
17804 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
17806         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
17807         (aarch64_simd_mov<mode>): New expander.
17808         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
17809         (aarch64_simd_mov_to_<mode>high): Likewise.
17810         (aarch64_simd_mov_from_<mode>low): Likewise.
17811         (aarch64_simd_mov_from_<mode>high): Likewise.
17812         (aarch64_dup_lane<mode>): Update.
17813         (aarch64_dup_lanedi): New instruction pattern.
17814         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
17815         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
17817 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17819         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
17820         (replace_mult_candidate): Remove unnecessary argument; remove
17821         unnecessary parameter from call to introduce_cast_before_cand.
17822         (replace_unconditional_candidate): Remove unnecessary parameter
17823         from call to replace_mult_candidate.
17824         (replace_conditional_candidate): Likewise.
17825         (insert_initializers): Use make_temp_ssa_name.
17826         (introduce_cast_before_cand): Remove unnecessary argument; use
17827         make_temp_ssa_name.
17828         (replace_one_candidate): Remove unnecessary argument; remove
17829         unnecessary parameter from calls to introduce_cast_before_cand.
17830         (replace_profitable_candidates): Remove unnecessary parameters
17831         from calls to replace_one_candidate.
17833 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17835         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
17836         phi def as possibly hiding a basis for a CAND_ADD whose operands
17837         have been commuted in the analysis.
17838         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
17840 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
17842         * config/aarch64/aarch64.md
17843         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
17844         shift value between 0-4.
17846 2013-05-07  Richard Biener  <rguenther@suse.de>
17848         * double-int.h (rshift): New overload.
17849         * double-int.c (rshift): New function.
17850         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
17851         (create_reference_ops_from_ref): Remove.
17852         (vn_reference_insert): Use shared ops for constructing the
17853         reference and copy it.
17855 2013-05-07  Richard Biener  <rguenther@suse.de>
17857         PR middle-end/57190
17858         * tree-eh.c (sink_clobbers): Properly propagate
17859         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
17861 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
17863         PR tree-optimization/57149
17864         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
17865         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
17866         collect_phi_def_edges, execute_late_warn_uninitialized): Use
17867         uninit_undefined_value_p instead of ssa_undefined_value_p.
17869         PR debug/57184
17870         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
17871         for modifier == EXPAND_INITIALIZER.
17873 2013-05-07  Anton Blanchard  <anton@samba.org>
17875         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
17876         for powerpc64 little endian.
17877         * configure: Regenerate.
17879 2013-05-06  Graham Stott  <grahams@btinternet.com>
17881         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
17882         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
17883         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
17884         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
17886 2013-05-06  Graham Stott  <grahams@btinternet.com>
17888         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
17889         codes which allow non-lvalues.
17891 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
17893         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
17894         components are all 1s.
17895         (integer_minus_onep): New function.
17896         * tree.h (integer_minus_onep): Declare it.
17897         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
17898         integer_minus_onep instead of integer_all_onesp.
17900 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17902         PR target/52933
17903         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
17904         variations of these patterns.
17906 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17908         * config/i386/i386.md (isa): Add x64_sse4 member.
17909         (enabled): Handle x64_sse4.
17910         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
17911         instruction for 64bit SSE4_1 targets.  Update insn attributes.
17912         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
17913         instruction for SSE4_1 targets.  Update insn attributes.
17914         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
17915         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
17916         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
17917         const_1 selector.
17918         (*vec_extractv4si): Rename from *sse4_1_pextrd.
17919         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
17920         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
17922 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
17924         PR target/57108
17925         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
17927 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
17929         * final.c (do_assembler_dialects): Don't handle curly braces and
17930         vertical bar escaped by % as dialect delimiters.
17931         (output_asm_insn): Print curly braces and vertical bar if escaped
17932         by % and ASSEMBLER_DIALECT defined.
17933         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
17934         * doc/tm.texi: Regenerated.
17936 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
17938         * config/mips/mips.c: Include tree-pass.h.
17939         (mips_reorg): Split in pre- and post-dbr_schedule parts.
17940         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
17941         (pass_mips_machine_reorg2): New machine specific pass.
17942         (insert_pass_mips_machine_reorg2): New pass plugin definition.
17943         (mips_option_override): Register the new pass.
17944         * rtl.h (cleanup_barriers): Remove prototype.
17945         (dbr_schedule): Likewise.
17946         * jump.c (cleanup_barriers): Make static.
17947         * reorg.c (dbr_schedule): Likewise.
17949 2013-05-06  Richard Biener  <rguenther@suse.de>
17951         PR tree-optimization/57185
17952         * tree-parloops.c (add_field_for_reduction): Handle anonymous
17953         SSA names properly.
17955 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
17957         PR target/57106
17958         * config/i386/i386.c (add_parameter_dependencies): Add dependence
17959         between "first_arg" and "insn", not "last" and "insn".
17961 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
17963         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
17964         (find_candidates_in_block): Re-enable slsr_process_phi.
17965         (create_phi_basis): Fix double counting of candidate adjustment.
17967 2013-05-06  Richard Biener  <rguenther@suse.de>
17969         PR middle-end/57147
17970         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
17971         the edge is also fallthru, preserve it and just clear the
17972         abnormal flag.
17973         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
17974         also complex, preserve that and just clear the fallthru flag.
17975         * tree-inline.c (update_ssa_across_abnormal_edges): Also
17976         update virtual operands.
17978 2013-05-06  Alan Modra  <amodra@gmail.com>
17980         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
17981         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
17982         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
17983         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
17984         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17985         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
17987 2013-05-06  Alan Modra  <amodra@gmail.com>
17989         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
17990         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
17991         (DEFAULT_ASM_ENDIAN): Define.
17992         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
17993         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
17994         Update -K PIC clause from sysv4.h.
17995         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
17996         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
17998 2013-05-06  Alan Modra  <amodra@gmail.com>
18000         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
18001         twice for little-endian.
18002         (ashrdi3_no_power, ashrdi3): Support little-endian.
18004 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
18006         PR target/55303
18007         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
18008         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
18009         related expanders.
18010         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
18011         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
18012         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
18013         New predicates.
18015 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
18016             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
18018         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
18019         * config/pa/pa.opt: Make mbig-switch a no-op.
18020         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
18021         (CASE_VECTOR_MODE): Always return SImode.
18022         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
18023         for the !TARGET_BIG_SWITCH case.
18024         * config/pa/pa-linux.h: Likewise.
18025         * config/pa/pa-openbsd.h: Likewise.
18026         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
18027         * config/pa/pa.md (short_jump): Remove define_insn.
18028         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
18029         (casesi0): Remove define_insn.
18030         (type): Remove btable_branch.
18031         (pa_combine_type): Likewise.
18032         (in_nullified_branch_delay): Likewise.
18033         (in_call_delay): Likewise.
18034         (define_delay): Likewise.
18035         (define_insn_reservation "Z3"): Likewise.
18036         (define_insn_reservation "Z4"): Likewise.
18037         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
18038         (pa_adjust_insn_length): Remove adjustment for btable branches.
18039         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
18040         and mno-big-switch
18042 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
18044         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
18045         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
18046         Add m->r,x alternatives.
18047         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
18048         splitters using SWI48x mode iterator.
18049         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
18050         TARGET_64BIT.  Add m->x alternative.
18051         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
18052         Add o->x alternative.  Enable for TARGET_SSE.
18053         (sse_storeq): Remove expander.
18054         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
18055         with memory input operand.
18056         (*vec_extractv2di_1 splitter): New.
18057         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
18058         * config/i386/i386.md (ssevecmodelower): New mode attribute.
18060 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18062         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
18063         (INT_LOWPART): Delete.
18064         (extract_MB): Adjust.
18065         (extract_ME): Adjust.
18066         (print_operand): Adjust.
18068 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
18070         * config/rs6000/predicates.md (reg_or_add_cint_operand,
18071         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18072         (reg_or_logical_cint_operand, easy_fp_constant,
18073         logical_const_operand): Delete "CONST_DOUBLE" case.
18074         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
18075         "HOST_BITS_PER_WIDE_INT == 64" test.
18076         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
18077         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
18078         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
18079         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
18080         test.
18081         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
18082         CONST_DOUBLE DImode/VOIDmode case.
18083         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
18084         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18085         CONST_DOUBLE VOIDmode case.
18086         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
18087         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
18088         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
18089         Delete CONST_DOUBLE case.
18090         (splitters for mov FMOVE64 const_double): Delete
18091         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
18092         "HOST_BITS_PER_WIDE_INT >= 64" test.
18093         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
18094         case.
18095         (mov DI const_double): Delete.
18097 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
18099         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
18100         on op shows all bits zero in mode of a lowpart subreg, return zero.
18102 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
18104         PR target/57150
18105         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
18106         to save TFmode registers and DImode to save TImode registers for
18107         caller save operations.
18108         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
18109         mark being partially clobbered since they only use the first
18110         double word.
18112         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
18113         and TDmode only use the upper 64-bits of each VSX register.
18115 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18117         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
18118         (find_candidates_in_block): Disable slsr_process_phi.
18120 2013-05-03  Guozhi Wei  <carrot@google.com>
18122         * coverage.c (coverage_obj_init): Move the construction of gcov
18123         constructor to ...
18124         (build_init_ctor): ... here.
18126 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18128         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
18129         (slsr_cand_d): Redefine def_phi.
18130         (stride_status, phi_adjust_status, count_phis_status): New enums.
18131         (find_phi_def): New.
18132         (find_basis_for_base_expr): New.
18133         (find_basis_for_candidate): Handle hidden bases.
18134         (alloc_cand_and_find_basis): Handle phi candidates.
18135         (slsr_process_phi): New.
18136         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
18137         (create_mul_imm_cand): Likewise.
18138         (create_add_ssa_cand): Exclude phi base candidates.
18139         (create_add_imm_cand): Likewise.
18140         (slsr_process_cast): Likewise.
18141         (slsr_process_copy): Likewise.
18142         (find_candidates_in_block): Handle phi candidates.
18143         (dump_candidate): Likewise.
18144         (unconditional_cands): Delete.
18145         (unconditional_cands_with_known_stride_p): Delete.
18146         (phi_dependent_cand_p): New.
18147         (cand_increment): Handle phi-dependent candidates.
18148         (replace_dependent): Delete.
18149         (replace_mult_candidate): New.
18150         (replace_unconditional_candidate): New.
18151         (incr_vec_index): Move to avoid forward reference.
18152         (create_add_on_incoming_edge): New.
18153         (create_phi_basis): New.
18154         (replace_dependents): Delete.
18155         (replace_conditional_candidate): New.
18156         (phi_add_costs): New.
18157         (replace_uncond_cands_and_profitable_phis): New.
18158         (record_increment): Handle phi adjustments.
18159         (record_phi_increments): New.
18160         (record_increments): Handle phi adjustments.
18161         (phi_incr_cost): New.
18162         (lowest_cost_path): Handle phis.
18163         (total_savings): Likewise.
18164         (analyze_increments): Likewise.
18165         (ncd_with_phi): New.
18166         (ncd_of_cand_and_phis): New.
18167         (nearest_common_dominator_for_cands): Handle phi increments.
18168         (all_phi_incrs_profitable): New.
18169         (replace_profitable_candidates): Handle phi-dependent candidates.
18170         (analyze_candidates_and_replace): Likewise.
18172 2013-05-03  Teresa Johnson  <tejohnson@google.com>
18174         PR bootstrap/57154
18175         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
18176         do not exceed REG_BR_PROB_BASE.
18178 2013-05-03  Jeff Law  <law@redhat.com>
18180         PR tree-optimization/57144
18181         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
18182         operand of the condition will bit into the new type when eliminating
18183         a cast feeding a condition.
18185 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
18187         PR rtl-optimization/57130
18188         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
18189         of COMPARE as in_code to the recursive call if needed.
18191 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
18193         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
18194         (enabled): Handle new members.
18195         * config/i386/sse.md (*vec_concatv2si): Merge from
18196         *vec_concatv2si_sse2 and vec_concatv2si_sse.
18197         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
18199 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
18201         PR tree-optimization/57027
18202         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
18203         for fnms opportunity, check we got the prerequisite kind
18204         of tree / gimple before using accessor functions.
18206 2013-05-03  Richard Biener  <rguenther@suse.de>
18208         * double-int.h (lshift): New overload without precision
18209         and arith argument.
18210         (operator *=, operator +=, operator -=): Move ...
18211         * double-int.c (operator *=, operator +=, operator -=): ... here
18212         and implement more efficiently.
18213         (mul_double_with_sign): Remove.
18214         (lshift_double): Adjust to take unsinged shift argument, push
18215         dispatching code to callers.
18216         (mul_double_wide_with_sign): Add early out for callers that
18217         are not interested in high parts or overflow.
18218         (lshift): New function.
18219         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
18220         dispatch code here.
18221         (lrotate, rrotate): Use logical shifts.
18222         * expr.c (get_inner_reference): Use lshift.
18223         * fixed-value.c (do_fixed_divide): Likewise.
18224         * tree-dfa.c (get_ref_base_and_extent): Likewise.
18225         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
18226         (indirect_refs_may_alias_p): Likewise.
18227         (stmt_kills_ref_p_1): Likewise.
18229 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18231         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
18233 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
18235         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
18236         scalar form of FABD instruction.
18238 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
18240         * lra-constraints.c (process_alt_operands): Add checking alt
18241         number to choose the best alternative.
18243 2013-05-02  Richard Biener  <rguenther@suse.de>
18245         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
18246         bitmap and its handling.
18247         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
18249 2013-05-02  Richard Biener  <rguenther@suse.de>
18251         PR middle-end/57140
18252         * tree-inline.c (copy_loops): Properly handle removed loops.
18253         (copy_cfg_body): Mark destination loops for fixup if source
18254         loops needed fixup.
18256 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
18258         PR target/56732
18259         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
18260         generating simple_return for naked functions.
18262 2013-05-02  Martin Jambor  <mjambor@suse.cz>
18264         PR middle-end/56988
18265         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
18266         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
18267         flags match.
18268         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
18269         ipa_agg_replacement_value structures.
18270         (known_aggs_to_agg_replacement_list): Likewise.
18271         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
18272         (read_agg_replacement_chain): Likewise.
18273         (ipcp_transform_function): Also check that by_ref flags match.
18275 2013-05-02  Richard Biener  <rguenther@suse.de>
18277         * graphds.h (struct graph): Add obstack member.
18278         * graphds.c (new_graph): Initialize obstack and allocate
18279         vertices from it.
18280         (add_edge): Allocate edge from the obstack.
18281         (free_graph): Free the obstack instead of all edges and vertices.
18283 2013-05-02  Teresa Johnson  <tejohnson@google.com>
18285         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
18286         divides.
18287         * cfg.c (update_bb_profile_for_threading): Ditto.
18288         * tree-inline.c (copy_bb): Ditto.
18289         (copy_edges_for_bb): Ditto.
18290         (initialize_cfun): Ditto.
18291         (copy_cfg_body): Ditto.
18292         (expand_call_inline): Ditto.
18293         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
18294         (estimate_node_size_and_time): Ditto.
18295         (inline_merge_summary): Ditto.
18296         * cgraphclones.c (cgraph_clone_edge): Ditto.
18297         (cgraph_clone_node): Ditto.
18298         * sched-rgn.c (compute_dom_prob_ps): Ditto.
18299         (compute_trg_info): Ditto.
18301 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18303         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
18304         S reg when fp attribute set.
18305         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
18307 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
18309         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
18310         New pattern.
18311         (*and_one_cmplsi3_compare0_uxtw): Likewise.
18312         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
18313         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
18315 2013-05-02  Richard Biener  <rguenther@suse.de>
18317         * tree-scalar-evolution.c (scev_info_hasher): Remove.
18318         (struct instantiate_cache_entry): New type.
18319         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
18320         (struct instantiate_cache_type): New type.
18321         (set_instantiated_value, get_instantiated_value): Remove.
18322         (get_instantiated_value_entry): New function.
18323         (instantiate_scev_name): Use the new cache and adjust.
18324         (instantiate_scev_poly): Adjust.
18325         (instantiate_scev_binary): Likewise.
18326         (instantiate_array_ref): Likewise.
18327         (instantiate_scev_convert): Likewise.
18328         (instantiate_scev_not): Likewise.
18329         (instantiate_scev_3): Likewise.
18330         (instantiate_scev_2): Likewise.
18331         (instantiate_scev_r): Likewise.
18332         (instantiate_scev): Likewise.
18333         (resolve_mixers): Likewise.
18335 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
18337         PR target/57091
18338         * lra-constraints.c (best_small_class_operands_num): Remove.
18339         (process_alt_operands): Remove small_class_operands_num.  Take
18340         small classes operands into losers and only if the operand is not
18341         matched.  Modify debugging output.
18342         (curr_insn_transform): Remove best_small_class_operands_num.
18343         Print insn name.
18345 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18347         * config/aarch64/aarch64-builtins.c
18348         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
18349         * config/aarch64/aarch64-simd-builtins.def
18350         (reduc_splus_): Add new modes.
18351         (reduc_uplus_): New.
18352         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
18353         (reduc_uplus_v4sf): Likewise.
18354         (reduc_splus_v4sf): Likewise.
18355         (aarch64_addv<mode>): Likewise.
18356         (reduc_uplus_<mode>): Likewise.
18357         (reduc_splus_<mode>): Likewise.
18358         (aarch64_addvv2di): Likewise.
18359         (reduc_uplus_v2di): Likewise.
18360         (reduc_splus_v2di): Likewise.
18361         (aarch64_addvv2si): Likewise.
18362         (reduc_uplus_v2si): Likewise.
18363         (reduc_splus_v2si): Likewise.
18364         (reduc_<sur>plus_<mode>): New.
18365         (reduc_<sur>plus_v2di): Likewise.
18366         (reduc_<sur>plus_v2si): Likewise.
18367         (reduc_<sur>plus_v4sf): Likewise.
18368         (aarch64_addpv4sf): Likewise.
18369         * config/aarch64/arm_neon.h
18370         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
18371         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
18372         add UNSPEC_SADDV, UNSPEC_UADDV.
18373         (SUADDV): New.
18374         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
18376 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18378         * config/aarch64/arm_neon.h
18379         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
18381 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18383         * config/aarch64/aarch64-builtins
18384         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
18386 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18388         * config/aarch64/aarch64-simd-builtins.def
18389         (reduc_smax_): New.
18390         (reduc_smin_): Likewise.
18391         (reduc_umax_): Likewise.
18392         (reduc_umin_): Likewise.
18393         (reduc_smax_nan_): Likewise.
18394         (reduc_smin_nan_): Likewise.
18395         (fmax): Remove.
18396         (fmin): Likewise.
18397         (smax): Update for V2SF, V4SF and V2DF modes.
18398         (smin): Likewise.
18399         (smax_nan): New.
18400         (smin_nan): Likewise.
18401         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
18402         (<su><maxmin><mode>3): ...This, refactor.
18403         (s<maxmin><mode>3): New.
18404         (<maxmin_uns><mode>3): Likewise.
18405         (reduc_<maxmin_uns>_<mode>): Refactor.
18406         (reduc_<maxmin_uns>_v4sf): Likewise.
18407         (reduc_<maxmin_uns>_v2si): Likewise.
18408         (aarch64_<fmaxmin><mode>: Remove.
18409         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
18410         new builtin names.
18411         (vmin<q>_f<32,64>): Likewise.
18412         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18413         (FMAXMIN): New.
18414         (su): Add mappings for smax, smin, umax, umin.
18415         (maxmin): New.
18416         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
18417         (FMAXMIN): Rename as...
18418         (FMAXMIN_UNS): ...This.
18419         (maxminv): Remove.
18420         (fmaxminv): Likewise.
18421         (fmaxmin): Likewise.
18422         (maxmin_uns): New.
18423         (maxmin_uns_op): Likewise.
18425 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18427         * config/aarch64/arm_neon.h
18428         (vac<ge, gt><sd>_f<32, 64>): Rename to...
18429         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
18430         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
18432 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18434         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
18435         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
18437 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18439         * config/aarch64/aarch64-simd.md
18440         (vcond<mode>_internal): Handle special cases for constant masks.
18441         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
18442         (vcondu<mode><mode>): Likewise.
18443         (vcond<v_cmp_result><mode>): New.
18445 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18447         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
18448         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
18449         * config/aarch64/aarch64-simd-builtins.def
18450         (cmeq): Update to BUILTIN_VALLDI.
18451         (cmgt): Likewise.
18452         (cmge): Likewise.
18453         (cmle): Likewise.
18454         (cmlt): Likewise.
18455         * config/aarch64/arm_neon.h
18456         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
18457         to builtins or C as appropriate.
18459 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
18461         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
18462         (cmgeu): ...This.
18463         (cmhi): Rename to...
18464         (cmgtu): ...This.
18465         * config/aarch64/aarch64-simd.md
18466         (simd_mode): Add SF.
18467         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
18468         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
18469         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
18470         (cstore<mode>_neg): ...This.
18471         * config/aarch64/iterators.md
18472         (VALLF): new.
18473         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
18474         (COMPARISONS): New.
18475         (UCOMPARISONS): Likewise.
18476         (optab): Add missing comparisons.
18477         (n_optab): New.
18478         (cmp_1): Likewise.
18479         (cmp_2): Likewise.
18480         (CMP): Likewise.
18481         (cmp): Remove.
18482         (VCMP_S): Likewise.
18483         (VCMP_U): Likewise.
18484         (V_cmp_result): Add DF, SF modes.
18485         (v_cmp_result): Likewise.
18486         (v): Likewise.
18487         (vmtype): Likewise.
18488         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
18490 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
18492         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
18493         define_insn to define_insn_and_split.
18494         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
18495         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
18496         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
18497         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
18498         (thumb2_negscc): Likewise.
18500 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18502         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
18504 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
18506         * config/arm/thumb2.md: Remove trailing whitespaces.
18508 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
18510         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
18511         Use gen_int_mode rather than GEN_INT.
18513 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
18515         * value-prof.c (stream_in_histogram_value): Remove the strayed
18516         debug_gimple_stmt.
18518 2013-04-30  Richard Biener  <rguenther@suse.de>
18520         PR middle-end/57122
18521         * cfghooks.c (split_edge): Properly check for the loop latch edge.
18523 2013-04-30  Richard Biener  <rguenther@suse.de>
18525         PR middle-end/57107
18526         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
18528 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
18530         PR rtl-optimization/56957
18531         PR rtl-optimization/57105
18532         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
18533         variable.  Use just INSN_UID for determining whether an insn
18534         should be only disconnected from the insn stream.
18535         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
18537 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
18539         PR tree-optimization/57104
18540         * tsan.c (instrument_expr): Don't instrument accesses to
18541         DECL_HARD_REGISTER VAR_DECLs.
18543 2013-04-30  Richard Biener  <rguenther@suse.de>
18545         * function.h (loops_for_fn): New inline function.
18546         (set_loops_for_fn): Likewise.
18547         * cfgloop.h (place_new_loop): Add struct function parameter.
18548         (get_loop): Likewise.
18549         (get_loops): Likewise.
18550         (number_of_loops): Likewise.
18551         (fel_next): Adjust.
18552         (fel_init): Likewise.
18553         * cfg.c (get_loop_copy): Adjust.
18554         * cfgloop.c (flow_loops_dump): Likewise.
18555         (record_loop_exits): Likewise.
18556         (verify_loop_structure): Likewise.
18557         * cfgloopanal.c (mark_irreducible_loops): Likewise.
18558         (estimate_reg_pressure_cost): Likewise.
18559         (mark_loop_exit_edges): Likewise.
18560         * cfgloopmanip.c (place_new_loop): Likewise.
18561         (add_loop): Likewise.
18562         (duplicate_loop): Likewise.
18563         * graph.c (draw_cfg_nodes): Likewise.
18564         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
18565         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
18566         (extract_affine_chrec): Likewise.
18567         (build_scop_iteration_domain): Likewise.
18568         * graphite.c (graphite_initialize): Likewise.
18569         * ira-build.c (create_loop_tree_nodes): Likewise.
18570         (more_one_region_p): Likewise.
18571         (rebuild_regno_allocno_maps): Likewise.
18572         (mark_loops_for_removal): Likewise.
18573         (mark_all_loops_for_removal): Likewise.
18574         (remove_unnecessary_regions): Likewise.
18575         (ira_build): Likewise.
18576         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
18577         * loop-init.c (fix_loop_structure): Likewise.
18578         (gate_rtl_move_loop_invariants): Likewise.
18579         (gate_rtl_unswitch): Likewise.
18580         (gate_rtl_unroll_and_peel_loops): Likewise.
18581         (rtl_doloop): Likewise.
18582         * lto-streamer-in.c (input_cfg): Likewise.
18583         * lto-streamer-out.c (output_cfg): Likewise.
18584         * modulo-sched.c (sms_schedule): Likewise.
18585         * predict.c (tree_estimate_probability): Likewise.
18586         (tree_estimate_probability_driver): Likewise.
18587         (estimate_loops): Likewise.
18588         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
18589         (move_sese_region_to_fn): Likewise.
18590         (debug_loop_num): Likewise.
18591         * tree-chrec.c (chrec_evaluate): Likewise.
18592         (hide_evolution_in_other_loops_than_loop): Likewise.
18593         (chrec_component_in_loop_num): Likewise.
18594         (reset_evolution_in_loop): Likewise.
18595         (evolution_function_is_invariant_rec_p): Likewise.
18596         * tree-if-conv.c (main_tree_if_conversion): Likewise.
18597         * tree-inline.c (copy_loops): Likewise.
18598         (copy_cfg_body): Likewise.
18599         (tree_function_versioning): Likewise.
18600         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
18601         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
18602         Likewise.
18603         (add_to_evolution_1): Likewise.
18604         (scev_const_prop): Likewise.
18605         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
18606         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
18607         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
18608         (tree_ssa_lim_initialize): Likewise.
18609         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
18610         (verify_loop_closed_ssa): Likewise.
18611         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
18612         (tree_ssa_loop_im): Likewise.
18613         (tree_ssa_loop_unswitch): Likewise.
18614         (tree_vectorize): Likewise.
18615         (check_data_deps): Likewise.
18616         (tree_ssa_loop_ivcanon): Likewise.
18617         (tree_ssa_loop_bounds): Likewise.
18618         (tree_complete_unroll): Likewise.
18619         (tree_complete_unroll_inner): Likewise.
18620         (tree_parallelize_loops): Likewise.
18621         (tree_ssa_loop_prefetch): Likewise.
18622         (tree_ssa_loop_ivopts): Likewise.
18623         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
18624         * tree-vectorizer.c (vectorize_loops): Likewise.
18626 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
18628         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
18629         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
18630         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
18631         with EABI_LINK_SPEC.
18633 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18635         PR target/44578
18636         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
18637         alternative.
18639 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
18641         PR target/57097
18642         * lra-constraints.c (process_alt_operands): Discourage a bit more
18643         using memory for pseudos.  Print cost dump for alternatives.
18644         Modify cost values for conflicts with early clobbers.
18645         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
18647 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18649         PR target/57098
18650         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
18652 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18654         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
18655         from/to S register.
18656         (movdi_aarch64): Support LDR/STR from/to D register.
18658 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
18660         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
18661         or higher by default.
18663 2013-04-29  Richard Biener  <rguenther@suse.de>
18665         PR middle-end/57075
18666         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
18667         even if not adding abnormal edges for calls that can make
18668         abnormal gotos.
18670 2013-04-29  Richard Biener  <rguenther@suse.de>
18672         PR middle-end/57103
18673         * tree-cfg.c (move_stmt_op): Fix condition under which to update
18674         TREE_BLOCK.
18675         (move_stmt_r): Remove redundant checking.
18677 2013-04-29  Teresa Johnson  <tejohnson@google.com>
18679         PR bootstrap/57077
18680         * basic-block.h (apply_scale): New function.
18681         (apply_probability): Use apply_scale.
18682         * gimple-streamer-in.c (input_bb): Ditto.
18683         * lto-streamer-in.c (input_cfg): Ditto.
18684         * lto-cgraph.c (merge_profile_summaries): Ditto.
18685         * tree-optimize.c (execute_fixup_cfg): Ditto.
18686         * tree-inline.c (copy_bb): Update comment to use apply_scale.
18687         (copy_edges_for_bb): Ditto.
18688         (copy_cfg_body): Ditto.
18690 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18692         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
18693         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
18694         (tail_merge_optimize): Handle current_loops == NULL.
18696 2013-04-26  Jeff Law  <law@redhat.com>
18698         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
18699         (simplify_cond_using_ranges): Generalize code to simplify
18700         COND_EXPRs where one argument is a constant and the other
18701         is an SSA_NAME created by an integral type conversion.
18703 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18705         * config/arm/arm.md (store_minmaxsi): Use only when
18706         optimize_insn_for_size_p.
18708 2013-04-29  Christian Bruel  <christian.bruel@st.com>
18710         PR target/57108
18711         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
18713 2013-04-29  Richard Biener  <rguenther@suse.de>
18715         PR middle-end/57089
18716         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
18717         loop tree make sure to schedule a fixup for the child as well.
18718         (expand_omp_for_generic): Properly add loops.
18719         (expand_omp_for_static_nochunk): Likewise.
18720         (expand_omp_for_static_chunk): Likewise.
18721         (expand_omp_for): For the degenerate case fixup loops.
18722         (expand_omp_sections): Fix default bb placement in loops.
18723         (expand_omp_atomic_pipeline): Properly add loops.
18725 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18727         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
18729 2013-04-29  Tom de Vries  <tom@codesourcery.com>
18731         * tree-ssa-tail-merge.c: Update header comment.
18733 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18735         * config/aarch64/arm_neon.h
18736         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
18737         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
18738         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
18739         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
18740         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
18741         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
18742         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
18743         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
18745 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18747         * config/aarch64/aarch64-simd.md
18748         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
18749         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
18750         fix_trunc, fixuns_trunc.
18751         (ftrunc<VDQF:mode>2): New.
18752         * config/aarch64/iterators.md (optab): Add fix, fixuns.
18753         (fix_trunc_optab): New.
18755 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18757         * config/aarch64/aarch64-builtins.c
18758         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
18759         iceilf, lround, iroundf.
18761 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
18763         PR target/54349
18764         * config/i386/i386.h (enum ix86_tune_indices)
18765         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
18766         New, split from X86_TUNE_INTER_UNIT_MOVES.
18767         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
18768         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
18769         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
18770         (TARGET_INTER_UNIT_MOVES): Remove.
18771         * config/i386/i386.c (initial_ix86_tune_features): Update.
18772         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
18773         (ix86_expand_convert_uns_didf_sse): Use
18774         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18775         (ix86_expand_vector_init_one_nonzero): Ditto.
18776         (ix86_expand_vector_init_interleave): Ditto.
18777         (inline_secondary_memory_needed): Return true for moves from SSE class
18778         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
18779         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18780         * config/i386/constraints.md (Yi, Ym): Depend on
18781         TARGET_INTER_UNIT_MOVES_TO_VEC.
18782         (Yj, Yn): New constraints.
18783         * config/i386/i386.md (*movdi_internal): Change constraints of
18784         operand 1 from Yi to Yj and from Ym to Yn.
18785         (*movsi_internal): Ditto.
18786         (*movdf_internal): Ditto.
18787         (*movsf_internal): Ditto.
18788         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
18789         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
18790         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
18791         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
18792         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
18793         * config/i386/sse.md (movdi_to_sse): Ditto.
18794         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
18795         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
18796         TARGET_INTER_UNIT_MOVES.
18797         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
18798         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
18799         instead of TARGET_INTER_UNIT_MOVES.
18800         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
18801         operand 1 from Yi to Yj and from Ym to Yn.
18803 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18805         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
18806         (float_truncate_hi_): Likewise.
18807         (float_extend_lo_): Likewise.
18808         (float_truncate_lo_): Likewise.
18809         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
18810         (aarch64_float_extend_lo_v2df): Likewise.
18811         (vec_unpacks_hi_v4sf): Likewise.
18812         (aarch64_float_truncate_lo_v2sf): Likewise.
18813         (aarch64_float_truncate_hi_v4sf): Likewise.
18814         (vec_pack_trunc_v2df): Likewise.
18815         (vec_pack_trunc_df): Likewise.
18817 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18819         * config/aarch64/aarch64-builtins.c
18820         (aarch64_fold_builtin): Fold float conversions.
18821         * config/aarch64/aarch64-simd-builtins.def
18822         (floatv2si, floatv4si, floatv2di): New.
18823         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
18824         * config/aarch64/aarch64-simd.md
18825         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
18826         * config/aarch64/iterators.md (FLOATUORS): New.
18827         (optab): Add float, floatuns.
18828         (su_optab): Likewise.
18830 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18832         * config/aarch64/aarch64-builtins.c
18833         (aarch64_builtin_vectorized_function): Use new names for
18834         fcvt builtins.
18835         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
18836         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
18837         (fcvtzu): Split as...
18838         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
18839         (fcvtas): Split as...
18840         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
18841         (fcvtau): Split as...
18842         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
18843         (fcvtps): Split as...
18844         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
18845         (fcvtpu): Split as...
18846         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
18847         (fcvtms): Split as...
18848         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
18849         (fcvtmu): Split as...
18850         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
18851         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
18852         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
18853         (lfrintnusf, lfrintnudf): Likewise.
18854         * config/aarch64/aarch64-simd.md
18855         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
18856         define_insn.
18857         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
18858         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
18859         (fcvt_pattern): Likewise.
18861 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18863         * config/aarch64/aarch64-simd.md
18864         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
18865         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
18867 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18869         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
18870         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
18871         (vrnd<a,m,n,p>_f32): Implement using builtins.
18872         (vrnd<i,x><q>_f<32, 64>): New.
18874 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
18876         * config/aarch64/aarch64-builtins.c
18877         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
18878         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
18879         (frintz): Rename to...
18880         (btrunc): ...this.
18881         (frintp): Rename to...
18882         (ceil): ...this.
18883         (frintm): Rename to...
18884         (floor): ...this.
18885         (frinti): Rename to...
18886         (nearbyint): ...this.
18887         (frintx): Rename to...
18888         (rint): ...this.
18889         (frinta): Rename to...
18890         (round): ...this.
18891         * config/aarch64/aarch64-simd.md
18892         (aarch64_frint<frint_suffix><mode>): Delete.
18893         (<frint_pattern><mode>2): Convert to insn.
18894         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
18895         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
18896         (frint_pattern): Likewise.
18897         (frint_suffix): Likewise.
18899 2013-04-29  Richard Biener  <rguenther@suse.de>
18901         PR tree-optimization/57081
18902         * loop-init.c: Include tree-flow.h.
18903         (loop_optimizer_finalize): Free number of iteration estimates.
18904         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
18906 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
18908         PR tree-optimization/57083
18909         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
18910         non-singleton shift count range, zero extend low_bound for uns case.
18912         * config/i386/predicates.md (general_vector_operand): New predicate.
18913         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
18914         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
18915         if they aren't nonimmediate operands.  If their original values
18916         satisfy const_vector_equal_evenodd_p, don't shift them.
18917         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
18918         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
18919         if not nonimmediate_operand.
18920         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
18921         instead of register_operand.
18922         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
18924 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
18926         * stor-layout.c (finalize_size_functions): Allocate a structure and
18927         reset cfun before dumping the functions.
18929 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
18931         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
18933         PR target/56866
18934         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
18935         use xop_pmacsdqh if uns_p.
18936         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
18937         the immediate rotate count.
18939 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
18941         * rtl.h (struct rtx_def): Add comment for field jump.
18942         (LRA_SUBREG_P): New macro.
18943         * recog.c (register_operand): Check LRA_SUBREG_P.
18944         * lra.c (lra): Add note at the end of RTL code. Align non-empty
18945         stack frame.
18946         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
18947         (lra_final_code_change): Skip subreg change for operators.
18948         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
18949         if there are no operand changes.
18950         * lra-constraints.c (curr_insn_set): New.
18951         (match_reload): Set LRA_SUBREG_P.
18952         (emit_spill_move): Ditto.
18953         (check_and_process_move): Use curr_insn_set. Process only single
18954         set insns.  Don't initialize sec_mem_p and change_p.
18955         (simplify_operand_subreg): Use LRA_SUBREG_P.
18956         (reg_in_class_p): New function.
18957         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
18958         of #ifdef.  Add code to remove cycling.
18959         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
18960         non-null disp.  Reload inner instead of disp when base and index
18961         are null.  Try to put lo_sum into register.
18962         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
18963         (check_and_process_move): Move code for move cost check to
18964         simple_move_p.  Remove equiv_substitution.
18965         (simple_move_p): New function.
18966         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
18967         curr_insn_set.  Call check_and_process_move only for single set
18968         insns.  Use the new function.  Move call of check_and_process_move
18969         after operand equiv substitution and address process.
18971 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18973         PR go/57045
18974         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
18975         with nonlocal goto receivers or returns twice calls, ignore
18976         unininitialized values from abnormal edges to nl goto receiver
18977         or returns twice call.
18979 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
18981         PR tree-optimization/57051
18982         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
18983         and VEC_RSHIFT_EXPR if shift count is a multiple of element
18984         bitsize.
18986 2013-04-26  Richard Biener  <rguenther@suse.de>
18988         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
18989         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
18990         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
18991         (fixup_loop_arrays_after_move): New function.
18992         (move_sese_region_to_fn): Properly outline the loop tree parts
18993         of the SESE region.
18995 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
18997         * config/i386/i386.md (type, unit): Fix long lines.
18999 2013-04-26  Richard Biener  <rguenther@suse.de>
19001         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
19002         (lto-streamer-out.o): Likewise.
19003         * cfgloop.c (init_loops_structure): Export, add struct function
19004         argument and adjust.
19005         (flow_loops_find): Adjust.
19006         * cfgloop.h (enum loop_estimation): Add EST_LAST.
19007         (init_loops_structure): Declare.
19008         * lto-streamer-in.c: Include cfgloop.h.
19009         (input_cfg): Input the loop tree.
19010         * lto-streamer-out.c: Include cfgloop.h.
19011         (output_cfg): Output the loop tree.
19012         (output_struct_function_base): Do not drop PROP_loops.
19014 2013-03-26  Richard Biener  <rguenther@suse.de>
19016         * tree-cfg.c (execute_build_cfg): Build the loop tree.
19017         (pass_build_cfg): Provide PROP_loops.
19018         (move_sese_region_to_fn): Remove loops that are outlined into fn
19019         for now.
19020         * tree-inline.c: Include cfgloop.h.
19021         (initialize_cfun): Do not drop PROP_loops.
19022         (copy_loops): New function.
19023         (copy_cfg_body): Copy loop structure.
19024         (tree_function_versioning): Initialize destination loop tree.
19025         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
19026         (pass_parallelize_loops): Do IL verification.
19027         * loop-init.c (loop_optimizer_init): Fixup loops if required.
19028         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
19029         the CFG make sure we fixup loops as well.
19030         * tree-ssa-tail-merge.c: Include cfgloop.h.
19031         (replace_block_by): When merging loop latches mark loops for fixup.
19032         * lto-streamer-out.c (output_struct_function_base): Drop
19033         PROP_loops for now.
19034         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
19035         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
19036         * ipa-split.c: Include cfgloop.h.
19037         (split_function): Add the new return block to the loop tree root.
19038         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
19039         whether we have removed the forwarder block.
19040         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
19041         * cfgloop.h (place_new_loop): Declare.
19042         * cfgloopmanip.c (place_new_loop): Export.
19043         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
19044         (tree-switch-conversion.o): Likewise.
19045         (tree-complex.o): Likewise.
19046         (tree-inline.o): Likewise.
19047         (tree-ssa-tailmerge.o): Likewise.
19048         (ipa-split.o): Likewise.
19049         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
19050         (tree-ssa-copy.o): Likewise.
19051         * tree-switch-conversion.c: Include cfgloop.h
19052         (process_switch): If we emit a bit-test cascade, schedule loops
19053         for fixup.
19054         * tree-complex.c: Include cfgloop.h.
19055         (expand_complex_div_wide): Properly add new basic-blocks to loops.
19056         * asan.c: Include cfgloop.h.
19057         (create_cond_insert_point): Properly add new basic-blocks to
19058         loops, schedule loop fixup.
19059         * cfgloop.c (verify_loop_structure): Check that looks are not
19060         marked for fixup.
19061         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
19062         to loops.
19063         (expand_omp_for_generic): Likewise.
19064         (expand_omp_sections): Likewise.
19065         (expand_omp_atomic_pipeline): Schedule loops for fixup.
19066         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
19067         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
19068         is initialized, not when loops are present.
19069         * tree-parloops.c (parallelize_loops): Remove checking here.
19070         * passes.c (init_optimization_passes): Schedule a copy-propagation
19071         pass before complete unrolling of inner loops.
19073 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
19075         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
19076         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
19077         (colorize_init): Add argument to _WIN32 version.
19078         * toplev.c: Include diagnostic-color.h.
19079         (process_options): Default to -fdiagnostics-color=auto if
19080         GCC_COLORS env var is in the environment.
19081         * common.opt (fdiagnostics-color=): Add Var and Init.
19082         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
19083         env var is in the environment, the default is auto rather than never.
19085         * diagnostic.h (file_name_as_prefix): Add context argument.
19086         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
19087         the string as locus.
19088         * langhooks.c (lhd_print_error_function): Adjust caller.
19090 2013-04-25  Lawrence Crowl  <crowl@google.com>
19092         * var-tracking.c (shared_hash_def::htab):
19093         Change type to hash_table.  Update dependent calls and types.
19095 2013-04-25  Lawrence Crowl  <crowl@google.com>
19097         * Makefile.in: Update as needed below.
19099         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
19100         Move declaration to after the type's method definitons.
19102         * attribs.c (htab_t scoped_attributes::attribute_hash):
19103         Change type to hash_table.  Update dependent calls and types.
19105         * bitmap.c (htab_t bitmap_desc_hash):
19106         Change type to hash_table.  Update dependent calls and types.
19108         * cselib.c (htab_t cselib_hash_table):
19109         Change type to hash_table.  Update dependent calls and types.
19111         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
19112         (hash_string_slot_node): Move implementation into lto-streamer.h
19113         struct string_slot_hasher.
19114         (eq_string_slot_node): Likewise.
19116         * data-streamer-out.c: Update output_block::string_hash_table
19117         dependent calls and types.
19119         * dwarf2cfi.c (htab_t trace_index):
19120         Change type to hash_table.  Update dependent calls and types.
19122         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
19123         Change type to hash_table.  Update dependent calls and types.
19124         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
19125         (htab_t optimize_external_refs::map): Likewise.
19126         (htab_t output_comp_unit::extern_map): Likewise.
19127         (htab_t output_comdat_type_unit::extern_map): Likewise.
19128         (htab_t output_macinfo::macinfo_htab): Likewise.
19129         (htab_t optimize_location_lists::htab): Likewise.
19130         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
19132         * except.c (htab_t ehspec_hash_type):
19133         Change type to hash_table.  Update dependent calls and types.
19134         (assign_filter_values::ttypes): Likewise.
19135         (assign_filter_values::ehspec): Likewise.
19136         (sjlj_assign_call_site_values::ar_hash): Likewise.
19137         (convert_to_eh_region_ranges::ar_hash): Likewise.
19139         * gcse.c (htab_t pre_ldst_table):
19140         Change type to hash_table.  Update dependent calls and types.
19142         * ggc-common.c (htab_t saving_htab):
19143         Change type to hash_table.  Update dependent calls and types.
19144         (htab_t loc_hash): Likewise.
19145         (htab_t ptr_hash): Likewise.
19146         (call_count): Rename ggc_call_count.
19147         (call_alloc): Rename ggc_call_alloc.
19148         (loc_descriptor): Rename make_loc_descriptor.
19149         (add_statistics): Rename ggc_add_statistics.
19151         * ggc-common.c (saving_htab):
19152         Change type to hash_table.  Update dependent calls and types.
19154         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
19155         (push_gimplify_context): Likewise.
19156         (pop_gimplify_context): Likewise.
19157         (struct gimple_temp_hash_elt): Added.
19158         (struct gimplify_hasher): Likewise.
19159         (struct gimplify_ctx.temp_htab):
19160         Change type to hash_table.  Update dependent calls and types.
19162         * gimple-fold.c: Include gimplify-ctx.h.
19164         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
19165         Change type to hash_table.  Update dependent calls and types.
19166         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
19167         avoid potential global name collision.
19169         * gimplify.c: Include gimplify-ctx.h.
19170         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
19171         (htab_t gimplify_ctx::temp_htab):
19172         Update dependent calls and types for new type hash_table.
19173         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
19174         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
19176         * gimplify-ctx.h: New.
19177         (struct gimple_temp_hash_elt): Move from gimplify.c.
19178         (class gimplify_hasher): New.
19179         (struct gimplify_ctx): Move from gimple.h.
19180         (htab_t gimplify_ctx::temp_htab):
19181         Change type to hash_table.  Update dependent calls and types.
19183         * graphite-clast-to-gimple.c: Include graphite-htab.h.
19184         (htab_t ivs_params::newivs_index):
19185         Change type to hash_table.  Update dependent calls and types.
19186         (htab_t ivs_params::params_index): Likewise.
19187         (htab_t print_generated_program::params_index): Likewise.
19188         (htab_t gloog::newivs_index): Likewise.
19189         (htab_t gloog::params_index): Likewise.
19191         * graphite.c: Include graphite-htab.h.
19192         4htab_t graphite_transform_loops::bb_pbb_mapping):
19193         Change type to hash_table.  Update dependent calls and types.
19195         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
19196         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
19197         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
19199         * graphite-dependences.c: Include graphite-htab.h.
19200         (loop_is_parallel_p): Change hash table type of parameter.
19202         * graphite-htab.h: New.
19203         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
19204         (extern find_pbb_via_hash): Move from graphite-poly.h.
19205         (extern loop_is_parallel_p): Move from graphite-poly.h.
19206         (extern get_loop_body_pbbs): Move from graphite-poly.h.
19208         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
19209         (extern loop_is_parallel_p): Move to graphite-htab.h.
19210         (extern get_loop_body_pbbs): Move to graphite-htab.h.
19212         * haifa-sched.c (htab_t delay_htab):
19213         Change type to hash_table.  Update dependent calls and types.
19214         (htab_t delay_htab_i2): Likewise.
19216         * ira-color.c (htab_t allocno_hard_regs_htab):
19217         Change type to hash_table.  Update dependent calls and types.
19219         * ira-costs.c (htab_t cost_classes_htab):
19220         Change type to hash_table.  Update dependent calls and types.
19222         * loop-invariant.c (htab_t merge_identical_invariants::eq):
19223         Change type to hash_table.  Update dependent calls and types.
19225         * loop-iv.c (htab_t bivs):
19226         Change type to hash_table.  Update dependent calls and types.
19228         * loop-unroll.c (htab_t opt_info::insns_to_split):
19229         Change type to hash_table.  Update dependent calls and types.
19230         (htab_t opt_info::insns_with_var_to_expand): Likewise.
19232         * lto-streamer.h (struct string_slot): Move from data-streamer.h
19233         (struct string_slot_hasher): New.
19234         (htab_t output_block::string_hash_table):
19235         Change type to hash_table.  Update dependent calls and types.
19237         * lto-streamer-in.c (freeing_string_slot_hasher): New.
19238         (htab_t file_name_hash_table):
19239         Change type to hash_table.  Update dependent calls and types.
19241         * lto-streamer-out.c: Update output_block::string_hash_table dependent
19242         calls and types.
19244         * lto-streamer.c (htab_t tree_htab):
19245         Change type to hash_table.  Update dependent calls and types.
19247         * omp-low.c: Include gimplify-ctx.h.
19249         * passes.c (htab_t name_to_pass_map):
19250         Change type to hash_table.  Update dependent calls and types.
19251         (pass_traverse): Rename to passes_pass_traverse.
19253         * plugin.c (htab_t event_tab):
19254         Change type to hash_table.  Update dependent calls and types.
19256         * postreload-gcse.c (htab_t expr_table):
19257         Change type to hash_table.  Update dependent calls and types.
19258         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
19260         * sese.c (debug_rename_map_1): Make extern.
19261         (htab_t copy_bb_and_scalar_dependences::rename_map):
19262         Change type to hash_table.  Update dependent calls and types.
19264         * sese.h (extern debug_rename_map): Move to .c file.
19266         * store-motion.c (htab_t store_motion_mems_table):
19267         Change type to hash_table.  Update dependent calls and types.
19269         * trans-mem.c (htab_t tm_new_mem_hash):
19270         Change type to hash_table.  Update dependent calls and types.
19272         * tree-browser.c (htab_t TB_up_ht):
19273         Change type to hash_table.  Update dependent calls and types.
19275         * tree-cfg.c (htab_t discriminator_per_locus):
19276         Change type to hash_table.  Update dependent calls and types.
19278         * tree-complex.c: Include tree-hasher.h
19279         (htab_t complex_variable_components):
19280         Change type to hash_table.  Update dependent calls and types.
19282         * tree-eh.c (htab_t finally_tree):
19283         Change type to hash_table.  Update dependent calls and types.
19285         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
19286         struct int_tree_hasher.
19287         (extern int_tree_map_eq): Likewise.
19288         (uid_decl_map_hash): Removed.
19289         (extern decl_tree_map_eq): Likewise.
19291         * tree-hasher.h: New.
19292         (struct int_tree_hasher): New.
19293         (typedef int_tree_htab_type): New.
19295         * tree-inline.c: Include gimplify-ctx.h.
19297         * tree-mudflap.c: Include gimplify-ctx.h.
19299         * tree-parloops.c: Include tree-hasher.h.
19300         (htab_t eliminate_local_variables_stmt::decl_address):
19301         Change type to hash_table.  Update dependent calls and types.
19302         (htab_t separate_decls_in_region::decl_copies): Likewise.
19304         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
19305         Change type to hash_table.  Update dependent calls and types.
19307         * tree-sra.c (candidates):
19308         Change type to hash_table.  Update dependent calls and types.
19310         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
19311         in tree-flow.h.
19312         (int_tree_map_hash): Likewise.
19314         * tree-ssa-dom.c (htab_t avail_exprs):
19315         Change type to hash_table.  Update dependent calls and types.
19317         * tree-ssa-live.c (var_map_base_init::tree_to_index):
19318         Change type to hash_table.  Update dependent calls and types.
19320         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
19321         Change type to hash_table.  Update dependent calls and types.
19323         * tree-ssa-phiopt.c (seen_ssa_names):
19324         Change type to hash_table.  Update dependent calls and types.
19326         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
19327         Change type to hash_table.  Update dependent calls and types.
19329         * tree-ssa-uncprop.c (equiv):
19330         Change type to hash_table.  Update dependent calls and types.
19332 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
19334         PR rtl-optimization/57003
19335         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
19336         call note_stores with kill_clobbered_value callback again after
19337         killing regs_invalidated_by_call.
19339 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19341         * config/aarch64/aarch64-simd.md
19342         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
19343         (aarch64_simd_bsl<mode>): Likewise.
19344         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
19346 2013-04-25  Marek Polacek  <polacek@redhat.com>
19348         PR tree-optimization/57066
19349         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
19351 2013-04-25  James Greenhalgh  <jame.greenhalgh@arm.com>
19353         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
19355 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19357         * config/aarch64/aarch64-builtins.c
19358         (aarch64_fold_builtin): New.
19359         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
19360         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
19361         * config/aarch64/aarch64-simd-builtins.def (abs): New.
19362         * config/aarch64/arm_neon.h
19363         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
19365 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19366             Tejas Belagod  <tejas.belagod@arm.com>
19368         * config/aarch64/aarch64-builtins.c
19369         (aarch64_gimple_fold_builtin): New.
19370         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
19371         * config/aarch64/aarch64-simd-builtins.def (addv): New.
19372         * config/aarch64/aarch64-simd.md (addpv4sf): New.
19373         (addvv4sf): Update.
19374         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
19376 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19378         * config/aarch64/aarch64.md
19379         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
19381 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
19383         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
19384         (*ngcsi_uxtw): New pattern.
19386 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
19387             Julian Brown  <julian@codesourcery.com>
19389         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
19390         (TB_DREG): Add T_V4HF.
19391         (v4hf_UP): New macro.
19392         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19393         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19394         Handle initialisation of V4HF. Adjust initialisation of reinterpret
19395         built-ins.
19396         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
19397         (arm_vector_mode_supported_p): Handle V4HF.
19398         (arm_mangle_map): Handle V4HFmode.
19399         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
19400         * config/arm/arm_neon_builtins.def: Add entries for
19401         vcvtv4hfv4sf, vcvtv4sfv4hf.
19402         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
19403         (neon_vcvtv4hfv4sf): Likewise.
19404         * config/arm/neon-gen.ml: Handle half-precision floating point
19405         features.
19406         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
19407         * config/arm/arm_neon.h: Regenerate.
19408         * config/arm/neon.ml (type elts): Add F16.
19409         (type vectype): Add T_float16x4, T_floatHF.
19410         (type vecmode): Add V4HF.
19411         (type features): Add Requires_FP_bit feature.
19412         (elt_width): Handle F16.
19413         (elt_class): Likewise.
19414         (elt_of_class_width): Likewise.
19415         (mode_of_elt): Refactor.
19416         (type_for_elt): Handle F16, fix error messages.
19417         (vectype_size): Handle T_float16x4.
19418         (vcvt_sh): New function.
19419         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
19420         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
19421         (string_of_mode): Handle V4HF.
19422         * doc/arm-neon-intrinsics.texi: Regenerate.
19424 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
19426         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
19427         format specifier in 'X' case.
19429 2013-04-25  Alan Modra  <amodra@gmail.com>
19431         PR target/57052
19432         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
19433         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
19434         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
19435         Repeat for many other rotate/shift and mask patterns using subregs.
19436         Name lshiftrt insns.
19437         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
19438         on WORDS_BIG_ENDIAN.
19440 2013-04-25  Alan Modra  <amodra@gmail.com>
19442         * config.gcc: Support little-endian powerpc-linux targets.
19443         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
19444         (LINK_OS_LINUX_SPEC): Define.
19445         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
19446         Preserve MASK_LITTLE_ENDIAN.
19447         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
19448         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
19449         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
19450         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
19451         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
19452         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
19453         Correct fp word order for little-endian.  Don't shift toc entries
19454         smaller than a word for little-endian.
19455         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
19456         (bswapdi2 splits): Correct low-part subreg for little-endian.
19457         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
19458         low/high where such is correct only for be.
19459         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
19460         little-endian for -mcall-aixdesc.
19462 2013-04-25  Alan Modra  <amodra@gmail.com>
19464         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
19465         replace_equiv_address_nv.
19467 2013-04-25  Alan Modra  <amodra@gmail.com>
19469         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
19471 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19473         Revert:
19474         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19475         * rtl.h (struct rtx_def): ...
19477 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19479         PR rtl-optimizations/57046
19480         * lra-constraints (split_reg): Set up lra_risky_transformations_p
19481         for multi-reg splits.
19483 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
19485         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
19487 2013-04-24  Sterling Augustine  <saugustine@google.com>
19489         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
19490         (comp_dir_string, debug_str_dwo_section): New.
19491         (DEBUG_STR_DWO_SECTION): Rename to ...
19492         (DEBUG_DWO_STR_SECTION): ... this.
19493         (DEBUG_NORM_STR_SECTION): Delete.
19494         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
19495         (DEBUG_STR_DWO_SECTION_FLAGS): New.
19496         (find_AT_string): Move most logic to ...
19497         (find_AT_string_in_table): ... here.  New.
19498         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
19499         add_skeleton_AT_string.  Delete logic.
19500         (output_skeleton_debug_sections): Remove call to
19501         add_top_level_skeleton_die_attrs.
19502         (add_comp_dir_attribute): Move logic to comp_dir_string.
19503         (dwarf2out_init): Initialize debug_str_dwo_section.
19504         (output_indirect_string): Call find_string_form.
19505         (output_indirect_strings): Rewrite.
19506         (prune_unused_types): Empty skeleton_debug_str_hash.
19507         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
19508         (dwarf2out_finish):  Call output_indirect_strings.
19510 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19512         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
19514 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
19516         * rtl.h (struct rtx_def): Add comment for field jump.
19517         (LRA_SUBREG_P): New macro.
19518         * recog.c (register_operand): Check LRA_SUBREG_P.
19519         * lra.c (lra): Add note at the end of RTL code. Align non-empty
19520         stack frame.
19521         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
19522         (lra_final_code_change): Skip subreg change for operators.
19523         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
19524         if there are no operand changes.
19525         * lra-constraints.c (curr_insn_set): New.
19526         (match_reload): Set LRA_SUBREG_P.
19527         (emit_spill_move): Ditto.
19528         (check_and_process_move): Use curr_insn_set. Process only single
19529         set insns.  Don't initialize sec_mem_p and change_p.
19530         (simplify_operand_subreg): Use LRA_SUBREG_P.
19531         (reg_in_class_p): New function.
19532         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
19533         of #ifdef.  Add code to remove cycling.
19534         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
19535         non-null disp.  Reload inner instead of disp when base and index
19536         are null.  Try to put lo_sum into register.
19537         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
19538         (check_and_process_move): Move code for move cost check to
19539         simple_move_p.  Remove equiv_substitution.
19540         (simple_move_p): New function.
19541         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
19542         curr_insn_set.  Call check_and_process_move only for single set
19543         insns.  Use the new function.  Move call of check_and_process_move
19544         after operand equiv substitution and address process.
19546 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
19548         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
19549         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
19550         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
19552 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
19554         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
19556 2013-04-24  Marek Polacek  <polacek@redhat.com>
19558         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
19559         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
19560         (select_loops_exit_conditions): Likewise.
19561         (number_of_iterations_for_all_loops): Likewise.
19562         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
19563         (scev_analysis): Likewise.
19565 2013-04-02  Catherine Moore  <clm@codesourcery.com>
19566             Chao-ying Fu  <fu@mips.com>
19568         * config/mips/micromips.md (jraddiusp): New pattern.
19569         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
19570         instruction if possible.
19572 2013-04-24  Alan Modra  <amodra@gmail.com>
19574         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
19576 2013-04-24  Julian Brown  <julian@codesourcery.com>
19577             Chung-Lin Tang  <cltang@codesourcery.com>
19579         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
19580         dependency behavior in enumeration type DIE generation. Add TODO note
19581         to comments about future DW_FORM_sdata/udata re-work of related code.
19583 2013-04-23  Lawrence Crowl  <crowl@google.com>
19585         * Makefile.in: Update as needed below.
19587         * hash-table.h (class hash_table):
19588         Correct many methods with parameter types compare_type to the correct
19589         value_type.  (Correct code was unlikely to notice the change.)
19590         (hash_table::elements_with_deleted) New.
19591         (class hashtable::iterator): New.
19592         (hashtable::begin()): New.
19593         (hashtable::end()): New.
19594         (FOR_EACH_HASH_TABLE_ELEMENT): New.
19596         * statistics.c (statistics_hashes):
19597         Change type to hash_table.  Update dependent calls and types.
19599         * tree-into-ssa.c (var_infos):
19600         Change type to hash_table.  Update dependent calls and types.
19602         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
19603         Change type to hash_table.  Update dependent calls and types.
19605         * tree-ssa-loop-im.c (struct mem_ref.refs):
19606         Change type to hash_table.  Update dependent calls and types.
19608         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
19609         Change type to hash_table.  Update dependent calls and types.
19611         * tree-ssa-sccvn.c (vn_tables_s::nary):
19612         Change type to hash_table.  Update dependent calls and types.
19613         (vn_tables_s::phis): Likewise.
19614         (vn_tables_s::references): Likewise.
19616         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
19617         (vn_reference_eq): Update parameter and return types.
19619         * tree-ssa-structalias.c (pointer_equiv_class_table):
19620         Change type to hash_table.  Update dependent calls and types.
19621         (location_equiv_class_table): Likewise.
19623         * tree-vect-data-refs.c: Consequential changes for making
19624         peeling a hash_table.
19626         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
19627         (destroy_loop_vec_info): Dependent hash_table update.
19629         * tree-vectorizer.h (peeling_htab):
19630         Change type to hash_table.  Update dependent calls and types.
19632 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
19634         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
19635         to check the register content is equal or not.
19636         * lra-constraints.c (match_reload): Use lra_assign_reg_val
19637         to assign register content record.
19638         * lra-eliminations.c (update_reg_eliminate): Use
19639         lra_update_reg_val_offset to update register content offset.
19640         * lra-int.h (struct lra_reg): Add offset member.
19641         (lra_reg_val_equal_p): New static inline function.
19642         (lra_update_reg_val_offset): New static inline function.
19643         (lra_assign_reg_val): New static inline function.
19644         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
19645         to assign register content record.
19646         (initialize_lra_reg_info_element): Initial offset to zero.
19648 2013-04-23  Catherine Moore  <clm@codesourcery.com>
19650         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
19651         operands.  Record compression.
19653 2013-04-23  Xinliang David Li  <davidxl@google.com>
19655         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
19657 2013-04-23  Richard Biener  <rguenther@suse.de>
19659         PR middle-end/57036
19660         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
19661         parameter, only add abnormal goto edges from the copied body
19662         if the call could perform abnormal gotos.
19663         (copy_cfg_body): Adjust.
19665 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
19667         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
19669 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
19671         * coretypes.h (gimple_stmt_iterator): Add struct to make
19672         compatible with C.
19674 2013-04-23  Richard Biener  <rguenther@suse.de>
19676         PR tree-optimization/57026
19677         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
19678         from SSA names occuring in abnormal PHI nodes.
19680 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19682         * lto/lto.c (print_lto_report_1): Fix LTO report names.
19684 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19686         * lto/lto.c (print_lto_report_1): Declare early.
19687         (read_cgraph_and_symbols): Call print_lto_report_1 early.
19689 2013-04-22  Andi Kleen  <ak@linux.intel.com>
19691         * common.opt (-flto-report-wpa): Add.
19692         * doc/invoke.texi (-flto-report-wpa): Add.
19693         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
19694         (lto_main): dito.
19696 2013-04-22  Xinliang David Li  <davidxl@google.com>
19698         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
19699         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
19700         * Makefile.in: New dependency
19702             David Daney <ddaney.cavm@gmail.com>
19704         * configure.ac (gcc_cv_as_micromips_support):  Use the
19705         --fatal-warnings option.
19706         * configure: Regenerate.
19708 2013-04-22  Marek Polacek  <polacek@redhat.com>
19710         PR sanitizer/56990
19711         * tsan.c (instrument_expr): Don't instrument expression
19712         in case its size is zero.
19714 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
19716         PR target/57032
19717         Revert:
19718         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
19720         * config/alpha/alpha.c (TARGET_LRA_P): New define.
19722 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19724         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
19725         (gimple_stmt_iterator): New typedef.
19726         * gimple.h (gimple_stmt_iterator): Rename to...
19727         (gimple_stmt_iterator_d): ... This.
19728         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
19729         trees be valid for GIMPLE and GENERIC.
19730         (TARGET_GIMPLE_FOLD_BUILTIN): New.
19731         * gimple-fold.c (gimple_fold_call): Call target hook
19732         gimple_fold_builtin.
19733         * hooks.c (hook_bool_gsiptr_false): New.
19734         * hooks.h (hook_bool_gsiptr_false): New.
19735         * target.def (fold_stmt): New.
19736         * doc/tm.texi: Regenerate.
19738 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
19740         PR target/57018
19741         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
19742         a set sp if no stack realignment.
19744 2013-04-22  Nick Clifton  <nickc@redhat.com>
19746         * config.gcc (tilegx-linux): Extend extra_objs rather than
19747         overwriting it.
19748         (tilepro-linux): Likewise.
19750 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19752         * config/aarch64/aarch64-builtins.c
19753         (CF): Remove.
19754         (CF0, CF1, CF2, CF3, CF4, CF10): New.
19755         (VAR<1-12>): Add MAP parameter.
19756         (BUILTIN_*): Likewise.
19757         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
19758         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
19759         (aarch64_ushl_n<mode>): Likewise.
19760         (aarch64_sshr_n<mode>): Likewise.
19761         (aarch64_ushr_n<mode>): Likewise.
19762         (aarch64_<maxmin><mode>): Likewise.
19763         (aarch64_sqrt<mode>): Likewise.
19764         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
19765         (vshr<q>_n_*): Likewise.
19767 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
19769         * config/aarch64/aarch64-builtins.c
19770         (aarch64_simd_builtin_type_mode): Handle SF types.
19771         (sf_UP): Define.
19772         (BUILTIN_GPF): Define.
19773         (aarch64_init_simd_builtins): Handle SF types.
19774         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
19775         (frecps): Likewise.
19776         (frecpx): Likewise.
19777         * config/aarch64/aarch64-simd.md
19778         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
19779         (aarch64_frecpe<mode>): New.
19780         (aarch64_frecps<mode>): Likewise.
19781         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
19782         (v8type): Add frecp<esx>.
19783         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
19784         (aarch64_frecps<mode>): Likewise.
19785         * config/aarch64/iterators.md (FRECP): New.
19786         (frecp_suffix): Likewise.
19787         * config/aarch64/arm_neon.h
19788         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
19790 2013-04-22  Christian Bruel  <christian.bruel@st.com>
19792         PR target/56995
19793         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
19794         (REG_CLASS_NAMES): Idem.
19795         (REG_CLASS_CONTENTS): Idem.
19796         (REGCLASS_HAS_FP_REG): Idem.
19797         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
19798         (sh_conditional_register_usage): Idem.
19800 2013-04-21  Jeff Law  <law@redhat.com>
19802         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
19803         (ssa_forward_propagate_and_combine): Use it.
19805 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19807         * lra.c: Update the flow chart diagram.
19809 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
19811         PR rtl-optimization/56847
19812         * lra-constraints.c (process_alt_operands): Discourage alternative
19813         with non-matche doffsettable memory constraint fro memory with
19814         known offset.
19816 2013-04-19  Richard Biener  <rguenther@suse.de>
19818         PR tree-optimization/56982
19819         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
19820         function.
19821         * gimplify.c (gimplify_call_expr): Notice special calls.
19822         (gimplify_modify_expr): Likewise.
19823         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
19824         abnormal control flow receivers.
19825         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
19826         in the same way as cfun->has_nonlocal_labels.
19827         (gimple_purge_dead_abnormal_call_edges): Likewise.
19828         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
19829         receivers start a basic-block.
19831 2013-04-19  Richard Biener  <rguenther@suse.de>
19833         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
19834         member ...
19835         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
19836         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
19837         (SLP_TREE_LOAD_PERMUTATION): Add.
19838         (vect_transform_slp_perm_load): Adjust prototype.
19839         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
19840         (vect_free_slp_instance): Likewise.
19841         (vect_create_new_slp_node): Likewise.
19842         (vect_supported_slp_permutation_p): Remove.
19843         (vect_slp_rearrange_stmts): Adjust.
19844         (vect_supported_load_permutation_p): Likewise.  Inline
19845         vect_supported_slp_permutation_p here.
19846         (vect_analyze_slp_instance): Compute load permutations per
19847         slp node instead of per instance.
19848         (vect_get_slp_defs): Adjust.
19849         (vect_transform_slp_perm_load): Likewise.
19850         (vect_schedule_slp_instance): Remove redundant code.
19851         (vect_schedule_slp): Remove hack for PR56270, add it ...
19852         * tree-vect-stmts.c (vectorizable_load): ... here, do not
19853         CSE loads for SLP.  Adjust.
19855 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19857         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
19858         spelling in two comments.
19860 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
19862         PR target/56797
19863         * config/arm/arm.c (load_multiple_sequence): Require SP
19864         as base register for loads if SP is in the register list.
19866 2013-04-19  Martin Jambor  <mjambor@suse.cz>
19868         PR tree-optimization/56718
19869         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
19870         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
19871         and made public.  Adjusted all callers.
19872         (ipa_intraprocedural_devirtualization): New function.
19873         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
19874         (ipa_intraprocedural_devirtualization): Likewise.
19875         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
19877 2013-04-19  Richard Biener  <rguenther@suse.de>
19879         PR tree-optimization/57000
19880         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
19882 2013-04-19  Terry Guo  <terry.guo@arm.com>
19884         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
19885         Replace with ...
19886         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
19887         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
19888         (cortex_m4_fmacs): Use new reservations.
19889         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
19891 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
19893         PR rtl-optimization/56999
19894         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
19895         related code.
19896         (lra_coalesce): Remove split_origin_bitmap and related code.
19897         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
19898         ranges if necessary.
19900 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
19902         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
19903         New array.
19904         (ix86_expand_call): Remove clobbered_registers array and use
19905         x86_64_ms_sysv_extra_clobbered_registers instead.
19906         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
19907         Declare here.
19908         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
19909         predicate.
19910         * config/i386/i386.md (*call_rex64_ms_sysv): Use
19911         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
19912         (*call_value_rex64_ms_sysv): Ditto.
19914 2013-04-18  Cary Coutant  <ccoutant@google.com>
19916         * dwarf2out.c (output_pubnames): Check die_perennial_p of
19917         parent instead of die_mark.
19919 2013-04-18  Diego Novillo  <dnovillo@google.com>
19921         * gimple.c (create_gimple_tmp): New.
19922         (get_expr_type): New.
19923         (build_assign): New.
19924         (build_type_cast): New.
19925         * gimple.h (enum ssa_mode): Define.
19926         (gimple_seq_set_location): New.
19927         * asan.c (build_check_stmt): Change some gimple_build_* calls
19928         to use build_assign and build_type_cast.
19930 2013-04-18  Richard Biener  <rguenther@suse.de>
19932         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
19933         handle negative step.  Remove redundant checks.
19934         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
19935         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
19936         for negative step and grouped loads fail to vectorize.
19938 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19940         * emit-rtl.c (reset_insn_used_flags): New function.
19941         (reset_all_used_flags): Use it.
19942         (verify_insn_sharing): New function.
19943         (verify_rtl_sharing): Fix verification for SEQUENCEs.
19945 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
19947         PR tree-optimization/56984
19948         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
19949         and (x >> M) >= N don't register any assertion if N << M is the
19950         minimum value.
19952 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
19954         * lower-subreg.c (resolve_simple_move): If called self-recursive,
19955         do not delete_insn insns that have not yet been emitted, only
19956         unlink them with remove_insn.
19957         * df-scan.c (df_insn_delete): Revert r197492.
19959 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
19961         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
19962         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
19964 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19966         * config/arm/arm.md (movsicc_insn): Convert define_insn into
19967         define_insn_and_split.
19968         (and_scc,ior_scc,negscc): Likewise.
19969         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
19971 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
19973         * config/arm/arm.c (use_return_insn): Return 0 for targets that
19974         can benefit from using a sequence of LDRD instructions in epilogue
19975         instead of a single LDM instruction.
19977 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
19979         PR 45688
19980         * doc/extend.texi: Fix typo.
19982 2013-04-17  Richard Biener  <rguenther@suse.de>
19984         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
19985         (vect_build_slp_tree): ... here.
19986         (vect_build_slp_tree_1): Compute which stmts of the SLP group
19987         match.  Remove special-casing of mismatched complex loads.
19988         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
19989         re-try the match with swapped commutative operands.
19990         (vect_supported_load_permutation_p): Remove special-casing of
19991         mismatched complex loads.
19992         (vect_analyze_slp_instance): Adjust.
19994 2013-04-17  Richard Biener  <rguenther@suse.de>
19996         PR rtl-optimization/56921
19997         * cfgloop.h (struct loop): Add simple_loop_desc member.
19998         (struct niter_desc): Mark with GTY(()).
19999         (simple_loop_desc): Do not use aux field but simple_loop_desc.
20000         * loop-iv.c (get_simple_loop_desc): Likewise.
20001         (free_simple_loop_desc): Likewise.
20003         Revert
20004         2013-04-16  Richard Biener  <rguenther@suse.de>
20006         PR rtl-optimization/56921
20007         * loop-init.c (pass_rtl_move_loop_invariants): Add
20008         TODO_do_not_ggc_collect to todo_flags_finish.
20009         (pass_rtl_unswitch): Same.
20010         (pass_rtl_unroll_and_peel_loops): Same.
20011         (pass_rtl_doloop): Same.
20013 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
20015         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
20016         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
20017         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
20018         references.
20019         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
20020         * tree-streamer.c (record_common_node): Adjust reference in comment.
20022 2013-04-17  Terry Guo  <terry.guo@arm.com>
20024         * config/arm/cortex-m4.md: Add a new bypass.
20026 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20028         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
20029         New pattern.
20030         (*subs_<optab><mode>_multp2): New pattern.
20031         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20032         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
20034 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20036         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
20037         (*subs_mul_imm_<mode>): New pattern.
20039 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
20041         PR target/56948
20042         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
20043         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
20044         (vsx_movti_32bit): Change j->wa to O->wa.
20046 2013-04-16  Richard Biener  <rguenther@suse.de>
20048         PR rtl-optimization/56921
20049         * loop-init.c (pass_rtl_move_loop_invariants): Add
20050         TODO_do_not_ggc_collect to todo_flags_finish.
20051         (pass_rtl_unswitch): Same.
20052         (pass_rtl_unroll_and_peel_loops): Same.
20053         (pass_rtl_doloop): Same.
20055 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
20057         * config/arm/arm.c (emit_multi_reg_push): New declaration
20058         for an existing function.
20059         (arm_emit_strd_push): New function.
20060         (arm_expand_prologue): Used here.
20061         (arm_emit_ldrd_pop): New function.
20062         (arm_expand_epilogue): Used here.
20063         (arm_get_frame_offsets): Update condition.
20064         (arm_emit_multi_reg_pop): Add a special case for load of a single
20065         register with writeback.
20067 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
20069         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
20070         description.
20072 2013-04-16  Richard Biener  <rguenther@suse.de>
20074         PR tree-optimization/56756
20075         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
20076         (first_mem_ref_loc): New.
20077         (execute_sm): Place the load temporarily before a previous
20078         access instead of in the latch edge to ensure its SSA dependencies
20079         are defined at points dominating the load.
20081 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
20083         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
20084         correct fix by moving header and footer insn to the footer of
20085         the merged basic block.  Clear BB_END of the merged-away block.
20087         PR middle-end/43631
20088         * emit-rtl.c (make_note_raw): New function.
20089         (link_insn_into_chain): New static inline function.
20090         (add_insn): Use it.
20091         (add_insn_before, add_insn_after): Factor insn chain linking code...
20092         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
20093         using link_insn_into_chain.
20094         (note_outside_basic_block_p): New helper function for emit_note_after
20095         and emit_note_before.
20096         (emit_note_after): Use nobb variant of add_insn_after if the note
20097         should not be contained in a basic block.
20098         (emit_note_before): Use nobb variant of add_insn_before if the note
20099         should not be contained in a basic block.
20100         (emit_note_copy): Use make_note_raw.
20101         (emit_note): Likewise.
20102         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
20103         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
20104         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
20105         the moved barrier the tail of the basic block it follows.
20106         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
20108 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
20110         PR tree-optimization/56962
20111         * gimple-ssa-strength-reduction.c (record_increment): Only set
20112         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
20113         either rhs1 or rhs2 is equal to c->base_expr.
20115 2013-04-15  Richard Biener  <rguenther@suse.de>
20117         PR tree-optimization/56933
20118         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
20119         member.
20120         (GROUP_READ_WRITE_DEPENDENCE): Remove.
20121         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
20122         * tree-vect-data-refs.c (vect_analyze_group_access): Move
20123         dependence check ...
20124         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
20125         ... here.
20126         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
20127         GROUP_READ_WRITE_DEPENDENCE.
20129 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20131         * emit-rtl.c (reset_all_used_flags): New function.
20132         (verify_rtl_sharing): Call reset_all_used_flags before and after
20133         performing the checks.
20135 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20137         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
20138         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
20139         * config/arm/constraints.md (De): New constraint.
20140         * config/arm/neon.md (anddi3_neon): Delete.
20141         (neon_vand<mode>): Expand to standard anddi3 pattern.
20142         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
20143         Move earlier in the file.
20144         (neon_inv_logic_op2): Likewise.
20145         (arm_anddi_operand_neon): New predicate.
20147 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20149         * configure.ac (gcc_cv_ld_as_needed): Set
20150         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
20151         Use -z ignore, -z record on *-*-solaris2*.
20152         (HAVE_LD_AS_NEEDED): Update comment.
20153         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
20154         * configure: Regenerate.
20155         * config.in: Regenerate.
20156         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
20157         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
20158         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
20159         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
20160         equivalents.  Fix markup.
20161         * doc/tm.texi: Regenerate.
20163 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
20165         * config/i386/i386.opt: New option mstack-protector-guard=.
20166         * config/i386/i386-opts.h: Add enum stack_protector_guard.
20167         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
20168         TARGET_SSP_TLS_GUARD.
20169         * config/i386/i386.c (ix86_option_override_internal): Set
20170         ix86_stack_protector_guard.
20171         * config/i386/i386.md (stack_protect_set): Enable for
20172         TARGET_SSP_TLS_GUARD only.
20173         (stack_protect_set_<mode>): Ditto.
20174         (stack_protect_test): Ditto.
20175         (stack_protect_test_<mode>): Ditto.
20176         * doc/invoke.texi (i386 Option): Document.
20178 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
20180         PR target/56890
20181         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
20182         (S_MODES): Set H_MODE bit.
20183         (SF_MODES): Set only S_MODE and SF_MODE bits.
20184         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
20185         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
20186         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
20187         <MODE_FLOAT>: Likewise.
20189 2013-04-15  Joey Ye  <joey.ye@arm.com>
20191         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
20193 2013-04-15  Joey Ye  <joey.ye@arm.com>
20195         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
20196         for real far jump.
20197         (thumb_far_jump_used_p): Count instruction size and set
20198         far_jump_used.
20200 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
20202         * reorg.c (fill_simple_delay_slots): Reindent block of code.
20203         * resource.c (mark_target_live_regs): Reformat conditional block.
20205 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20207         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
20208         notes, they are emitted only just before final.
20209         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
20211 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
20213         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
20214         * cfgrtl.c (delete_insn): Call it here instead.
20215         * lra-spills.c (lra_final_code_change): Use delete_insn.
20216         * haifa-sched.c (sched_remove_insn): Likewise.
20217         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
20218         returning to the nop pool.
20219         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
20220         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
20222 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20224         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
20225         * doc/tm.texi: Regenerated.
20227 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
20229         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
20230         QImode checks.
20232 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
20234         * df-core.c (df_find_def): Compare register numbers.
20235         (df_find_use): Likewise.
20237 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
20239         PR target/56903
20240         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
20241         lra_in_progress for return.
20243 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20245         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
20246         define_insn into define_insn_and_split and emit movsicc patterns.
20248 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
20250         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
20252 2013-04-12  Richard Biener  <rguenther@suse.de>
20254         * tree-pass.h (TODO_do_not_ggc_collect): New.
20255         * passes.c (execute_one_ipa_transform_pass): Honor
20256         TODO_do_not_ggc_collect.
20257         (execute_one_pass): Likewise.
20259         Revert
20260         2013-04-10  Richard Biener  <rguenther@suse.de>
20262         * passes.c (init_optimization_passes): Remove reload pass.
20263         * ira.c (do_reload): Merge into ...
20264         (ira): ... this.
20265         (rest_of_handle_reload): Remove.
20266         (pass_reload): Likewise.
20267         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20268         of reload for vzeroupper pass placement.
20270 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
20272         PR tree-optimization/56918
20273         PR tree-optimization/56920
20274         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
20275         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
20276         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
20277         use wide_mul_with_sign method.
20279 2013-04-12  Richard Biener  <rguenther@suse.de>
20281         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
20282         not be considered a gimple constant.
20284 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
20286         * fold-const.c (const_binop): Handle vector shifts by a scalar.
20287         (fold_binary_loc): Call const_binop also for mixed vector-scalar
20288         operations.
20290 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
20291             Jakub Jelinek  <jakub@redhat.com>
20293         * opts.c: Include diagnostic-color.h.
20294         (common_handle_option): Handle OPT_fdiagnostics_color_.
20295         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
20296         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
20297         (diagnostic-color.o): New.
20298         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
20299         (diagnostic_color_rule): New enum.
20300         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
20301         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
20302         the location string.
20303         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
20304         either NULL, or color kind.
20305         * diagnostic-color.c: New file.
20306         * diagnostic-color.h: New file.
20307         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
20308         arguments.
20309         * doc/invoke.texi (-fdiagnostics-color): Document.
20310         * pretty-print.h (pp_show_color): Define.
20311         (struct pretty_print_info): Add show_color field.
20312         * diagnostic.c: Include diagnostic-color.h.
20313         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
20314         macros.  Colorize error:, warning: etc. strings and also the location
20315         string.
20316         (diagnostic_show_locus): Colorize the caret line.
20317         * pretty-print.c: Include diagnostic-color.h.
20318         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
20319         inside of %< %> quotes or quoted through q format modifier.
20321 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20323         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
20325 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20327         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
20328         code in CC_NZ mode.
20329         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
20330         pattern.
20332 2013-04-11  Marek Polacek  <polacek@redhat.com>
20334         PR tree-optimization/48184
20335         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
20337 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
20339         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
20340         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
20341         (skip_simple_arithmetic): Tidy up.
20342         * tree.h (skip_simple_constant_arithmetic): Declare.
20344 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20346         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
20348 2013-04-11  Richard Biener  <rguenther@suse.de>
20350         * tree-vect-loop.c (get_initial_def_for_induction): Properly
20351         generate vector constants.
20353 2013-04-11  Richard Biener  <rguenther@suse.de>
20355         PR tree-optimization/56878
20356         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
20357         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
20358         New function.
20359         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20360         Prefer to align the DR with the most invariant base address.
20362 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20364         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
20365         comment.
20367 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
20369         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
20370         floating-point vector comparisons against 0.
20372 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
20374         PR tree-optimization/56899
20375         * fold-const.c (extract_muldiv_1): Apply distributive law
20376         only if TYPE_OVERFLOW_WRAPS (ctype).
20378 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
20380         PR target/56124
20381         * ira-costs.c (scan_one_insn): Check whether the source rtx of
20382         loading has side effect.
20384 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
20386         * config/sparc/sparc.c: Include tree-pass.h.
20387         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
20388         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
20389         head of file.  Change return type.  Split off gate function.
20390         (sparc_gate_work_around_errata): New function.
20391         (pass_work_around_errata): New pass definition.
20392         (insert_pass_work_around_errata) New pass insert definition to
20393         insert pass_work_around_errata just after delayed-branch scheduling.
20394         (sparc_option_override): Insert the pass.
20395         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
20397 2013-04-10  David S. Miller  <davem@davemloft.net>
20399         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
20400         or -mcpu=hypersparc.
20402         * target.def (cstore_mode): New hook.
20403         * target.h: Include insn-codes.h
20404         * targhooks.c: Likewise.
20405         (default_cstore_mode): New function.
20406         * targhooks.h: Declare it.
20407         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
20408         * doc/tm.texi: Rebuild.
20409         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
20410         target hook, rather than inspecting the insn_data.
20411         * config/sparc/sparc.c (sparc_cstore_mode): New function.
20412         (TARGET_CSTORE_MODE): Redefine.
20413         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
20414         result patterns.
20415         * config/sparc/predicates.md (cstore_result_operand): New special
20416         predicate.
20417         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
20418         Use it for operand 0.
20419         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
20420         (*snesi_special): Likewise.
20421         (*snesi_zero): Likewise.
20422         (*seqsi_zero): Likewise.
20423         (*sltu_insn): Likewise.
20424         (*sgeu_insn): Likewise.
20425         (*seqdi_special): Make operand 0 and comparison operation be of
20426         DImode.
20427         (*snedi_special): Likewise.
20428         (*snedi_special_vis3): Likewise.
20429         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
20430         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
20431         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
20432         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
20433         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
20434         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
20435         (*sltu_extend_sp64): Likewise.
20436         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
20437         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
20438         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
20439         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
20440         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
20442 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
20444         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
20445         (aarch64_start_file): Use the new function.
20447 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
20448             Jason Merrill  <jason@redhat.com>
20450         * common.opt: Add -gdwarf.
20451         * opts.c (common_handle_option): Handle it.
20452         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
20454 2013-04-10  Richard Biener  <rguenther@suse.de>
20456         * passes.c (execute_todo): Do not call ggc_collect conditional here.
20457         (execute_one_ipa_transform_pass): But unconditionally here.
20458         (execute_one_pass): And here.
20459         (init_optimization_passes): Remove reload pass.
20460         * tree-pass.h (TODO_ggc_collect): Remove.
20461         (pass_reload): Likewise.
20462         * ira.c (do_reload): Merge into ...
20463         (ira): ... this.
20464         (rest_of_handle_reload): Remove.
20465         (pass_reload): Likewise.
20466         * config/i386/i386.c (ix86_option_override): Refer to ira instead
20467         of reload for vzeroupper pass placement.
20468         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
20469         and todo_flags_finish of all passes.
20471 2013-04-10  Richard Biener  <rguenther@suse.de>
20473         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
20474         first_const_oprnd field, rename first_def_type to first_op_type.
20475         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
20476         (vect_get_and_check_slp_defs): Always use the type of the
20477         operand.  Allow mixed vect_external_def, vect_constant_def types.
20478         (vect_get_constant_vectors): Handle mixed vect_external_def,
20479         vect_constant_def types.
20481 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
20483         PR tree-optimization/55524
20484         * tree-ssa-math-opts.c
20485         (convert_mult_to_fma): Don't use an fms construct
20486         when we don't have an fms operation, but fnma, and it looks
20487         likely that we'll be able to use the latter.
20489 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
20491         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
20492         function.
20493         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
20494         inline fail caused by overwritable functions.
20496 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
20498         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
20499         unnecessary bits in the constant power of two case.
20501 2013-04-10  Richard Biener  <rguenther@suse.de>
20503         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
20504         broken code swapping operands.
20505         (vect_build_slp_tree): Do not compute load permutations here.
20506         (vect_analyze_slp_instance): Compute load permutations here,
20507         after building the SLP tree.
20509 2013-04-09  Christian Bruel  <christian.bruel@st.com>
20511         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
20512         of next/prev_real_insn.
20514 2013-04-09  Jan Hubicka  <jh@suse.cz>
20516         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
20517         Drop aliased parameter.
20518         (function_and_variable_visibility): Do not handle alias pairs.
20519         * cgraph.c (varpool_externally_visible_p): Update prototype.
20520         * varpool.c (varpool_add_new_variable): Update.
20522 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20524         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
20526 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
20528         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
20530         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
20532 2013-04-09  Marek Polacek  <polacek@redhat.com>
20534         PR tree-optimization/48762
20535         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
20537 2013-04-09  Richard Biener  <rguenther@suse.de>
20539         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
20540         dealing with cost.
20541         (vect_build_slp_tree): Likewise.
20542         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
20543         calculating the cost of a SLP instance.
20544         (vect_analyze_slp_instance): Use it from here, after building
20545         the SLP tree.
20547 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20549         PR middle-end/56883
20550         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
20551         expand_omp_for_static_chunk): Use simple_p = true in
20552         force_gimple_operand_gsi calls when assigning to addressable decls.
20554 2013-04-09  Jeff Law  <law@redhat.com>
20556         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
20557         when the boolean was created by converting a wider object which
20558         had a boolean range.
20560 2013-04-09  Richard Biener  <rguenther@suse.de>
20562         * tree-vectorizer.h (slp_void_p): Remove.
20563         (slp_tree): Typedef before _slp_tree declaration.
20564         (struct _slp_tree): Use a vector of slp_tree as children.
20565         (vect_get_place_in_interleaving_chain): Remove.
20566         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
20567         Move ...
20568         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
20569         and make static.
20570         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
20571         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
20572         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
20573         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
20574         Use slp_node instead of slp_void_p and adjust.
20576 2013-04-09  Richard Biener  <rguenther@suse.de>
20578         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
20579         work that is not necessary.
20581 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
20583         PR tree-optimization/56854
20584         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
20585         forward into clobber stmts if it would change MEM_REF lhs into
20586         non-MEM_REF.
20588 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
20590         * tree.c (type_hash_lookup, type_hash_add): Make static.
20591         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
20593 2013-04-09  Richard Biener  <rguenther@suse.de>
20595         * tree.h (unsave_expr_now): Remove.
20596         * tree-inline.c (mark_local_for_remap_r): Remove.
20597         (unsave_expr_1): Likewise.
20598         (unsave_r): Likewise.
20599         (unsave_expr_now): Likewise.
20600         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
20601         (propagate_tree_value): Likewise.
20603 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
20605         * doc/rtl.texi (sequence): Rewrite documentation to match the
20606         current use of SEQUENCE rtl objects.
20607         * rtl.def (SEQUENCE): Likewise.
20609         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
20610         Update documentation.
20611         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
20612         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
20614         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
20616 2013-04-08  Teresa Johnson  <tejohnson@google.com>
20618         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
20619         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
20620         methods.
20621         (estimate_edge_size_and_time): Add comment to suggest using rounding
20622         methods.
20623         (estimate_node_size_and_time): Ditto.
20624         (remap_edge_change_prob): Use helper rounding divide methods.
20625         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
20626         (gimple_mod_pow2_value_transform): Ditto.
20627         (gimple_mod_subtract_transform): Ditto.
20628         (gimple_ic_transform): Ditto.
20629         (gimple_stringops_transform): Ditto.
20630         * stmt.c (conditional_probability): Ditto.
20631         (emit_case_dispatch_table): Ditto.
20632         * lto-cgraph.c (merge_profile_summaries): Ditto.
20633         * tree-optimize.c (execute_fixup_cfg): Ditto.
20634         * cfgcleanup.c (try_forward_edges): Ditto.
20635         * cfgloopmanip.c (scale_loop_profile): Ditto.
20636         (loopify): Ditto.
20637         (duplicate_loop_to_header_edge): Ditto.
20638         (lv_adjust_loop_entry_edge): Ditto.
20639         * tree-vect-loop.c (vect_transform_loop): Ditto.
20640         * profile.c (compute_branch_probabilities): Ditto.
20641         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
20642         * lto-streamer-in.c (input_cfg): Ditto.
20643         * gimple-streamer-in.c (input_bb): Ditto.
20644         * ipa-cp.c (update_profiling_info): Ditto.
20645         (update_specialized_profile): Ditto.
20646         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
20647         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
20648         rounding methods.
20649         * sched-rgn.c (compute_dom_prob_ps): Ditto.
20650         (compute_trg_info): Ditto.
20651         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
20652         (purge_dead_edges): Ditto.
20653         * loop-unswitch.c (unswitch_loop): Ditto.
20654         * cgraphclones.c (cgraph_clone_edge): Ditto.
20655         (cgraph_clone_node): Ditto.
20656         * tree-inline.c (copy_bb): Ditto.
20657         (copy_edges_for_bb): Ditto.
20658         (initialize_cfun): Ditto.
20659         (copy_cfg_body): Ditto.
20660         (expand_call_inline): Ditto.
20662 2013-04-08  Kai Tietz  <ktietz@redhat.com>
20664         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
20665         TARGET_CYGWIN64 by TARGET_64BIT.
20667 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20669         * config/epiphany/epiphany.md (GPR_1): New constant.
20670         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
20671         * config/epiphany/epiphany.c (gen_compare_reg):
20672         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
20673         is already in place.
20674         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
20675         Don't require being called during rtl expansion; If y operlaps r0,
20676         return 0.
20677         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
20678         (epiphany_expand_epilogue): Likewise.
20680         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
20681         Don't use CC_FPmode for ORDERED / UNORDERED.
20682         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
20684         * config/epiphany/constraints.md (CnL): New constraint.
20685         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
20686         * config/epiphany/predicates.md (add_operand): Allow 1024.
20688         * config/epiphany/epiphany.md (logical_op): New code iterator.
20689         (op_mnc): New code attribute.
20690         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
20691         (mov_f+1, mov_f+2): New peephole2 patterns.
20693         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
20694         (cstoresi4): Also allow re-use of zero result when doing a NE
20695         comparison to a non-zero operand.
20696         Use (clobber (scratch)) for first insn if the gpr output is not needed.
20698         * config/epiphany/epiphany.md (<insn_opname>v2si3):
20699         Use gen_addsi3_i / gen_subsi3_i.
20701 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20703         PR c++/34949
20704         PR c++/50243
20705         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
20706         contain anything but clobbers, at most one __builtin_stack_restore,
20707         optionally debug stmts and final resx, and if it has at least one
20708         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
20709         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
20710         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
20711         which isn't defaut definition, remove them.
20712         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
20713         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
20714         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
20715         with MEM_REF LHS with SSA_NAME address.
20717 2013-04-08  Jeff Law  <law@redhat.com>
20719         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
20721 2013-04-08  Richard Biener  <rguenther@suse.de>
20723         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
20724         extra newline.
20725         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
20726         determined vector type.
20727         (vect_analyze_data_refs): Likewise.
20728         (vect_get_new_vect_var): Adjust.
20729         (vect_create_destination_var): Preserve SSA name versions.
20730         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
20731         not dump anything here.
20733 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
20735         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
20736         Add member lr_slot_known.
20737         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
20738         if necessary.
20739         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
20740         Remove code that sets lr_slot_offset according to what a previous
20741         version of epiphany_emit_save_restore used to do.
20742         (epiphany_emit_save_restore): When doing an lr save or restore,
20743         set/verify lr_slot_known and lr_slot_offset.
20745 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
20747         PR target/54338
20748         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
20749         in ALL_REGS.
20751 2013-04-08  Richard Biener  <rguenther@suse.de>
20753         * alias.c (find_base_term): Fix thinko in previous change.
20755 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
20757         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
20758         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
20759         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
20760         if possible to compute val.
20761         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
20762         For QImode integers don't require anything about precision.  Use
20763         const_with_all_bytes_same to find out if the constant doesn't have
20764         repeated bytes in it.
20766 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20768         * config/s390/s390.c (s390_expand_insv): Only accept insertions
20769         within mode size.
20771 2013-04-08  Marek Polacek  <polacek@redhat.com>
20773         PR rtl-optimization/48182
20774         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
20775         value to 1.
20777 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20779         PR target/55487
20780         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
20781         nuses, make sure we have a label.
20783 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20785         PR target/56843
20786         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
20787         (rs6000_emit_swdiv_low_precision): Remove.
20788         (rs6000_emit_swdiv): Rewrite to handle between one and four
20789         iterations of Newton-Raphson generally; modify required number of
20790         iterations for some cases.
20791         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
20793 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
20795         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
20796         set-but-unused variable.
20798         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
20799         basic blocks of released function bodies garbage-collectable.
20801         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
20802         (struct rtl_opt_pass): Add TODO_df_finish.
20804         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
20806 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20808         * config/arm/constraints.md (q): New constraint.
20809         * config/arm/ldrdstrd.md: New file.
20810         * config/arm/arm.md (ldrdstrd.md) New include.
20811         (arm_movdi): Use "q" instead of "r" constraint
20812         for double-word memory access.
20813         (movdf_soft_insn): Likewise.
20814         * config/arm/vfp.md (movdi_vfp): Likewise.
20815         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
20816         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
20817         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
20818         (mem_ok_for_ldrd_strd): Likewise.
20819         (output_move_double): Update assertion.
20821 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20823         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
20825 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20827         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
20828         define_insn_and_split.
20829         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
20831 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20833         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
20834         define_insn_and_split.
20835         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
20836         (shiftsi3_compare): New pattern.
20837         (rrx): New pattern.
20838         * config/arm/unspecs.md (UNSPEC_RRX): New.
20840 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20842         * config/arm/arm.md (negdi_extendsidi): New pattern.
20843         (negdi_zero_extendsidi): Likewise.
20845 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20847         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
20848         define_insn_and_split.
20849         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
20850         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
20852 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20854         * config/arm/arm.md (arm_subdi3): Convert define_insn into
20855         define_insn_and_split.
20856         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
20857         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
20859 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20861         * config/arm/arm.md (subsi3_carryin): New pattern.
20862         (subsi3_carryin_const): Likewise.
20863         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
20864         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
20866 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20868         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
20870 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
20872         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
20873         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
20875 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20877         * config/arm/arm.c (arm_expand_builtin): Change fcode
20878         type to unsigned int.
20880 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
20882         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
20884 2013-04-04  Ian Lance Taylor  <iant@google.com>
20886         * doc/standards.texi (Standards): The Go frontend supports the Go 1
20887         language standard.
20889 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
20891         PR middle-end/56729
20892         * df-scan.c (df_insn_delete): Disable failing assert.
20894 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20896         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
20897         New function prototype.
20898         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
20899         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
20900         (arm_builtin_vectorized_function): New function.
20902 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20904         * config/arm/arm_neon_builtins.def: New file.
20905         * config/arm/arm.c (neon_builtin_data): Move contents to
20906         arm_neon_builtins.def.
20907         (enum arm_builtins): Include neon builtin definitions.
20908         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
20909         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
20911 2013-04-04  Marek Polacek  <polacek@redhat.com>
20913         PR tree-optimization/48186
20914         * predict.c (maybe_hot_frequency_p): Return false if
20915         HOT_BB_FREQUENCY_FRACTION is 0.
20916         (cgraph_maybe_hot_edge_p): Likewise.
20918 2013-04-04  Richard Biener  <rguenther@suse.de>
20920         PR tree-optimization/56826
20921         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
20922         more accurately.
20924 2013-04-04  Richard Biener  <rguenther@suse.de>
20926         PR tree-optimization/56213
20927         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
20928         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
20930 2013-04-04  Richard Biener  <rguenther@suse.de>
20932         PR tree-optimization/56837
20933         * tree-loop-distribution.c (classify_partition): For non-zero
20934         values require that the value has the same precision as its
20935         mode to be useful as memset value.
20937 2013-04-03  Nick Clifton  <nickc@redhat.com>
20939         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
20940         (fmssf4): Use fmsf.s on E3V5 architectures.
20941         (fnmasf4): Use fnmaf.s on E3V5 architectures.
20942         (fnmssf4): Use fnmsf.s on E3V5 architectures.
20944 2013-04-03  Jeff Law  <law@redhat.com>
20946         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
20947         (lra-eliminations.o): Likewise.
20949 2013-04-03  Teresa Johnson  <tejohnson@google.com>
20951         * gcov-io.c (compute_working_sets): Moved most of body of old
20952         compute_working_sets here from profile.c.
20953         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
20954         (gcov_working_set_t): Moved typedef here from basic-block.h
20955         (compute_working_set): Declare.
20956         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
20957         (get_working_sets): Renamed from compute_working_set,
20958         replace most of body with call to new compute_working_sets.
20959         (get_exec_counts): Replace call to compute_working_sets
20960         to get_working_sets.
20961         * profile.h (get_working_sets): Renamed from compute_working_set.
20962         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
20963         to get_working_sets.
20964         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
20965         * gcov-dump.c (dump_working_sets): New function.
20967 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
20969         * hwint.c (sext_hwi, zext_hwi): New functions.
20970         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
20971         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
20972         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
20973         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
20974         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
20975         (sext_hwi, zext_hwi): New functions.
20977 2013-04-03  Jeff Law  <law@redhat.com>
20979         PR tree-optimization/56799
20980         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
20981         back test for widening conversion erroneously dropped in prior change.
20983 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20985         PR target/56809
20986         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
20987         instead of next_real_insn.
20989 2013-04-03  Marek Polacek  <polacek@redhat.com>
20991         PR sanitizer/55702
20992         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
20994 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
20996         PR target/56809
20997         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
20998         next_real_insn.
20999         (thumb1_output_casesi): Likewise.
21000         (thumb2_output_casesi): Likewise.
21002 2013-04-03  Richard Biener  <rguenther@suse.de>
21004         PR tree-optimization/56817
21005         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
21006         Split out ...
21007         (tree_unroll_loops_completely_1): ... new function to manually
21008         walk the loop tree, properly defering outer loops of unrolled
21009         loops to later iterations.
21011 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21013         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
21014         (vectorizable_load): Likewise.
21015         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
21016         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
21018 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21020         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
21021         BIT_FIELD_REF.
21023 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
21025         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
21027 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
21029         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
21031 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21033         PR tree-optimization/56790
21034         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
21035         folding.
21037 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
21039         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
21040         Handle VEC_MERGE.
21041         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
21042         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
21043         equal arguments.
21045 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
21047         PR c/19449
21048         * tree.h (force_folding_builtin_constant_p): New decl.
21049         * builtins.c (force_folding_builtin_constant_p): New variable.
21050         (fold_builtin_constant_p): Fold immediately also if
21051         force_folding_builtin_constant_p.
21053 2013-04-03  Richard Biener  <rguenther@suse.de>
21055         PR tree-optimization/56812
21056         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
21057         DRs of the same interleaving chain are independent.
21059 2013-04-02  Jason Merrill  <jason@redhat.com>
21061         * gdbinit.in (pbb): Use debug fn.
21063 2013-04-02  Lawrence Crowl  <crowl@google.com>
21065         * sese.h (struct ivtype_map_elt_s): Remove unused.
21066         (extern debug_ivtype_map): Remove unused.
21067         (extern eq_ivtype_map_elts): Remove unused.
21068         * sese.c (debug_ivtype_map): Removed unused.
21069         (debug_ivtype_map_1): Removed unused.
21070         (debug_ivtype_elt): Remove unused.
21071         (eq_ivtype_map_elts): Remove unused.
21074 2013-04-02  Kai Tietz  <ktietz@redhat.com>
21076         PR target/52790
21077         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
21078         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
21079         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
21080         function.
21081         (legitimize_pe_coff_symbol): Likewise.
21082         (is_imported_p): New helper-function.
21083         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
21084         for Windows x64 targets.
21085         (ix86_expand_prologue): Optimize for pe-coff targets.
21086         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
21087         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
21088         medium/large code-model.
21089         (legitimize_pic_address): Likewise.
21090         (legitimize_tls_address): Likewise.
21091         (ix86_expand_call): Likewise.
21092         (x86_output_mi_thunk): Likewise.
21093         (get_dllimport_decl): Add new beimport argument.
21094         (construct_plt_address): Don't assert for x64 pe-coff targets.
21095         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
21096         targets.
21097         (SYMBOL_FLAG_STUBVAR): New macro.
21098         (SYMBOL_REF_STUBVAR_P): Likewise.
21099         * config/i386/winnt.c (stub_list): New structure.
21100         (stub_head): New local variable.
21101         (i386_pe_record_stub): New function.
21102         (i386_pe_file_end): Emit refptr-stubs.
21104 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
21106         PR rtl-optimization/56745
21107         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
21108         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
21110         PR c++/34949
21111         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
21112         and both of them are MEM_REFs, just compare first argument for
21113         equality and attempt to deal even with differing offsets.
21115         PR c++/34949
21116         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
21117         of gimple_clobber_p to be MEM_REF.
21118         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
21119         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
21120         after gimplification.
21121         * asan.c (get_mem_ref_of_assignment): Don't instrument
21122         gimple_clobber_p stmts.
21123         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
21124         gimple_clobber_p stmt if they have MEM_REF lhs and
21125         are dead because of another gimple_clobber_p stmt.
21126         * tree-ssa-live.c (clear_unused_block_pointer): Treat
21127         gimple_clobber_p stmts like debug stmts.
21128         (remove_unused_locals): Remove clobbers with MEM_REF lhs
21129         that refer to unused VAR_DECLs or uninitialized values.
21130         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
21131         gimple_clobber_p stmts if they refer to removed parameters.
21132         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
21133         formatting.
21135 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
21137         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
21138         using SWI48 mode attribute.
21140 2013-04-02  Wei Mi  <wmi@google.com>
21142         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
21143         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
21144         *<rotate_insn><mode>3_mask in i386.md.
21146 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21148         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
21150 2013-04-02  Richard Biener  <rguenther@suse.de>
21152         PR tree-optimization/56778
21153         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
21154         Runtime alias tests are not supported for gather loads.
21155         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
21156         stmts referenced from SSA operands before updating SSA form.
21158 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
21159             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
21161         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
21162         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
21163         * config/arm/cortex-a53.md: New file.
21164         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
21165         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
21166         * config/arm/arm.c (arm_issue_rate): Likewise.
21167         * config/arm/arm-tune.md: Regenerate
21168         * config/arm/arm-tables.opt: Regenerate.
21169         * config/arm/arm-cores.def: Add cortex-a53.
21171 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
21173         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
21174         non-static link.
21176 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21178         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
21179         scalar load/store operations using B/H registers.
21180         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
21182 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
21184         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
21185         scalar move.
21186         * config/aarch64/aarch64.c
21187         (aarch64_simd_scalar_immediate_valid_for_move): New.
21188         * config/aarch64/aarch64-protos.h
21189         (aarch64_simd_scalar_immediate_valid_for_move): New.
21190         * config/aarch64/constraints.md (Dh, Dq): New.
21191         * config/aarch64/iterators.md (hq): New.
21193 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
21195         * reorg.c (get_branch_condition): Deal with conditional returns.
21196         (fill_simple_delay_slots): Remove dead code dealing with jumps.
21198 2013-04-01  Wei Mi  <wmi@google.com>
21200         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
21201         Truncate operand 2 using %b asm operand modifier.
21202         (*<shift_insn><mode>3_mask): Ditto.
21203         (*<rotate_insn><mode>3_mask): Ditto.
21205 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
21207         PR middle-end/56798
21208         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
21210 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
21212         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
21213         of next_real_insn.
21214         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
21216 2013-03-30  Lawrence Crowl  <crowl@google.com>
21218         * dse.c (clear_alias_sets): Remove never set.
21219         (disqualified_clear_alias_sets): Remove never set.
21220         (clear_alias_mode_pool): Remove never set.
21221         (dse_step0): Remove condition that is never true.
21222         (canon_address): Remove condition that is never true.
21223         (dse_step7): Remove condition that is never true.
21224         (rest_of_handle_dse): Remove condition that is never true.
21225         (rest_of_handle_dse::did_global): Remove never read from above.
21226         (dse_step2_spill): Remove never called from above.
21227         (dse_step5_spill): Remove never called from above.
21229 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
21231         * doc/md.texi (Standard Names) <casesi>: Update documentation for
21232         JUMP_TABLE_DATA changes.
21233         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
21234         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
21235         (Insns) <jump_table_data>: New entry.
21236         * doc/tm.texi: Regenerate.
21238         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
21240         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
21241         for table jump at the end of a basic block using tablejump_p.
21242         * targhooks.c (default_invalid_within_doloop): Likewise.
21243         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
21244         target hook implementation that is identical to the default hook.
21245         (rs6000_invalid_within_doloop): Remove.
21247         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
21248         unused variable from tablejump_p call.
21250         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
21251         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
21252         (INSN_DELETED_P): Likewise.
21253         (emit_jump_table_data): New prototype.
21254         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
21255         after 4th as unused.
21256         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
21257         * sched-vis.c (print_insn): Likewise.
21258         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
21259         insn for compatibility with back ends that use next_active_insn to
21260         identify jump table data.
21261         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
21262         (remove_insn): Likewise.
21263         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
21264         to be emitted.
21265         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
21266         (emit_jump_table_data): New function.
21268         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
21269         basic block, a JUMP_TABLE_DATA never is.
21270         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
21271         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
21272         off from code handling real insns.
21273         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
21274         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
21275         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
21276         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
21277         is not a NONDEBUG_INSN_P.
21278         * ira-costs.c (scan_one_insn): Likewise.
21279         * jump.c (mark_all_labels): Likewise.
21280         (mark_jump_label_1): Likewise.
21281         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21282         * lra.c (get_insn_freq): Expect all insns reaching here to be in
21283         a basic block.
21284         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
21285         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
21286         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
21287         JUMP_TABLE_DATA_P insns.
21288         (calculate_elim_costs_all_insns): Likewise.
21289         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
21290         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
21291         (delete_output_reload): Code style fixups.
21292         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
21293         insn flags on this non-insn.
21294         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
21295         as scheduling barriers, for pre-change compatibility.
21296         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
21297         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
21299         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
21300         redundant JUMP_TABLE_DATA_P test.
21301         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
21302         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21303         (frv_for_each_packet): Likewise.
21304         * config/i386/i386.c (min_insn_size): Likewise.
21305         (ix86_avoid_jump_mispredicts): Likewise.
21306         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21307         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
21308         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21309         (mips16_insn_length): Robustify.
21310         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
21311         (mips16_split_long_branches): Likewise.
21312         * config/pa/pa.c (pa_combine_instructions): Likewise.
21313         * config/rs6000/rs6000.c (get_next_active_insn): Treat
21314         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
21315         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
21316         as contributing to pool range lengths.
21317         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
21318         Remove redundant JUMP_TABLE_DATA_P test.
21319         (sh_loop_align): Likewise.
21320         (split_branches): Likewise.
21321         (sh_insn_length_adjustment): Likewise.
21322         * config/spu/spu.c (get_branch_target): Likewise.
21324 2013-03-29  Jan Hubicka  <jh@suse.cz>
21326         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
21327         gcov streaming; stream hot bb threshold to ltrans.
21328         * predict.c (get_hot_bb_threshold): Break out from ....
21329         (maybe_hot_count_p): ... here.
21330         (set_hot_bb_threshold): New function.
21331         * lto-section-in.c (lto_section_name): Add profile.
21332         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
21333         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
21334         and data-streamer.h
21335         (histogram_entry): New structure.
21336         (histogram, histogram_pool): New global vars.
21337         (histogram_hash): New structure.
21338         (histogram_hash::hash): New method.
21339         (histogram_hash::equal): Likewise.
21340         (account_time_size): New function.
21341         (cmp_counts): New function.
21342         (dump_histogram): New function.
21343         (ipa_profile_generate_summary): New function.
21344         (ipa_profile_write_summary): New function.
21345         (ipa_profile_read_summary): New function.
21346         (ipa_profile): Decide on threshold.
21347         (pass_ipa_profile): Add ipa_profile_write_summary and
21348         ipa_profile_read_summary.
21349         * Makefile.in (ipa.o): Update dependencies.
21350         * lto-streamer.h (LTO_section_ipa_profile): New section.
21352 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21354         * tree.h (VAR_P): New.
21356 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
21358         PR lto/56777
21359         * doc/invoke.texi ([-fwhole-program]): Fix typo.
21361 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
21363         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
21364         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21365         (control_flow_insn_p): Likewise.
21366         * cfgrtl.c (duplicate_insn_chain): Likewise.
21367         * final.c (get_attr_length_1): Likewise.
21368         (shorten_branches): Likewise.
21369         (final_scan_insn): Likewise.
21370         * function.c (instantiate_virtual_regs): Likewise.
21371         * gcse.c (insert_insn_end_basic_block): Likewise.
21372         * ira-costs.c (scan_one_insn): Likewise.
21373         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
21374         * lra.c (check_rtl): Likewise.
21375         * reload1.c (elimination_costs_in_insn): Likewise.
21376         * reorg.c (follow_jumps): Likewise.
21378         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
21379         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
21380         (thumb_far_jump_used_p): Likewise.
21381         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
21382         (workaround_speculation): Likewise.
21383         (add_sched_insns_for_speculation): Likewise.
21384         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
21385         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
21386         (frv_for_each_packet): Likewise.
21387         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
21388         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
21389         (final_emit_insn_group_barriers): Likewise.
21390         * config/m32r/m32r.c (m32r_is_insn): Likewise.
21391         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
21392         (mips16_insn_length): Likewise.
21393         * config/pa/pa.c (pa_reorg): Likewise.
21394         (pa_combine_instructions): Likewise.
21395         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
21396         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
21397         (sh_reorg): Likewise.
21398         (split_branches): Likewise.
21399         * config/spu/spu.c (get_branch_target): Likewise.
21401         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
21402         JUMP_TABLE_DATA_P.
21404 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
21406         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
21407         Fix declaration name.
21409 2013-03-28  Lawrence Crowl  <crowl@google.com>
21411         * graphds.h (struct graph.indicies): Remove unused.
21412         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
21413         (SCOP_ORIGINAL_PDDRS): Remove unused.
21414         * sese.h (extern insert_loop_close_phis): Removed unused.
21415         (extern insert_guard_phis): Removed unused.
21416         (extern ivtype_map_elt_info): Removed unused.
21417         (new_ivtype_map_elt): Removed unused.
21418         * sese.c (ivtype_map_elt_info): Removed unused.
21420 2013-03-28  Lawrence Crowl  <crowl@google.com>
21422         * Makefile.in: Add several missing include dependences.
21423         (DUMPFILE_H): New.
21424         (test-dump.o): New.  This object is not added to any executable,
21425         but is present for ad-hoc testing.
21426         * bitmap.c
21427         (debug (const bitmap_head_def &)): New.
21428         (debug (const bitmap_head_def *)): New.
21429         * bitmap.h
21430         (extern debug (const bitmap_head_def &)): New.
21431         (extern debug (const bitmap_head_def *)): New.
21432         * cfg.c
21433         (debug (edge_def &)): New.
21434         (debug (edge_def *)): New.
21435         * cfghooks.c
21436         (debug (basic_block_def &)): New.
21437         (debug (basic_block_def *)): New.
21438         * dumpfile.h
21439         (dump_node (const_tree, int, FILE *)): Correct source file.
21440         * dwarf2out.c
21441         (debug (die_struct &)): New.
21442         (debug (die_struct *)): New.
21443         * dwarf2out.h
21444         (extern debug (die_struct &)): New.
21445         (extern debug (die_struct *)): New.
21446         * gimple-pretty-print.c
21447         (debug (gimple_statement_d &)): New.
21448         (debug (gimple_statement_d *)): New.
21449         * gimple-pretty-print.h
21450         (extern debug (gimple_statement_d &)): New.
21451         (extern debug (gimple_statement_d *)): New.
21452         * ira-build.c
21453         (debug (ira_allocno_copy &)): New.
21454         (debug (ira_allocno_copy *)): New.
21455         (debug (ira_allocno &)): New.
21456         (debug (ira_allocno *)): New.
21457         * ira-int.h
21458         (extern debug (ira_allocno_copy &)): New.
21459         (extern debug (ira_allocno_copy *)): New.
21460         (extern debug (ira_allocno &)): New.
21461         (extern debug (ira_allocno *)): New.
21462         * ira-lives.c
21463         (debug (live_range &)): New.
21464         (debug (live_range *)): New.
21465         * lra-int.h
21466         (debug (lra_live_range &)): New.
21467         (debug (lra_live_range *)): New.
21468         * lra-lives.c
21469         (debug (lra_live_range &)): New.
21470         (debug (lra_live_range *)): New.
21471         * omega.c
21472         (debug (omega_pb_d &)): New.
21473         (debug (omega_pb_d *)): New.
21474         * omega.h
21475         (extern debug (omega_pb_d &)): New.
21476         (extern debug (omega_pb_d *)): New.
21477         * print-rtl.c
21478         (debug (const rtx_def &)): New.
21479         (debug (const rtx_def *)): New.
21480         * print-tree.c
21481         (debug_tree (tree): Move within file.
21482         (debug_raw (const tree_node &)): New.
21483         (debug_raw (const tree_node *)): New.
21484         (dump_tree_via_hooks (const tree_node *, int)): New.
21485         (debug (const tree_node &)): New.
21486         (debug (const tree_node *)): New.
21487         (debug_verbose (const tree_node &)): New.
21488         (debug_verbose (const tree_node *)): New.
21489         (debug_head (const tree_node &)): New.
21490         (debug_head (const tree_node *)): New.
21491         (debug_body (const tree_node &)): New.
21492         (debug_body (const tree_node *)): New.
21493         (debug_vec_tree (tree): Move and reimplement in terms of dump.
21494         (debug (vec<tree, va_gc> &)): New.
21495         (debug (vec<tree, va_gc> *)): New.
21496         * rtl.h
21497         (extern debug (const rtx_def &)): New.
21498         (extern debug (const rtx_def *)): New.
21499         * sbitmap.c
21500         (debug_raw (simple_bitmap_def &)): New.
21501         (debug_raw (simple_bitmap_def *)): New.
21502         (debug (simple_bitmap_def &)): New.
21503         (debug (simple_bitmap_def *)): New.
21504         * sbitmap.h
21505         (extern debug (simple_bitmap_def &)): New.
21506         (extern debug (simple_bitmap_def *)): New.
21507         (extern debug_raw (simple_bitmap_def &)): New.
21508         (extern debug_raw (simple_bitmap_def *)): New.
21509         * sel-sched-dump.c
21510         (debug (vinsn_def &)): New.
21511         (debug (vinsn_def *)): New.
21512         (debug_verbose (vinsn_def &)): New.
21513         (debug_verbose (vinsn_def *)): New.
21514         (debug (expr_def &)): New.
21515         (debug (expr_def *)): New.
21516         (debug_verbose (expr_def &)): New.
21517         (debug_verbose (expr_def *)): New.
21518         (debug (vec<rtx> &)): New.
21519         (debug (vec<rtx> *)): New.
21520         * sel-sched-dump.h
21521         (extern debug (vinsn_def &)): New.
21522         (extern debug (vinsn_def *)): New.
21523         (extern debug_verbose (vinsn_def &)): New.
21524         (extern debug_verbose (vinsn_def *)): New.
21525         (extern debug (expr_def &)): New.
21526         (extern debug (expr_def *)): New.
21527         (extern debug_verbose (expr_def &)): New.
21528         (extern debug_verbose (expr_def *)): New.
21529         (extern debug (vec<rtx> &)): New.
21530         (extern debug (vec<rtx> *)): New.
21531         * sel-sched-ir.h
21532         (_list_iter_cond_expr): Make inline instead of static.
21533         * sreal.c
21534         (debug (sreal &)): New.
21535         (debug (sreal *)): New.
21536         * sreal.h
21537         (extern debug (sreal &)): New.
21538         (extern debug (sreal *)): New.
21539         * tree.h
21540         (extern debug_raw (const tree_node &)): New.
21541         (extern debug_raw (const tree_node *)): New.
21542         (extern debug (const tree_node &)): New.
21543         (extern debug (const tree_node *)): New.
21544         (extern debug_verbose (const tree_node &)): New.
21545         (extern debug_verbose (const tree_node *)): New.
21546         (extern debug_head (const tree_node &)): New.
21547         (extern debug_head (const tree_node *)): New.
21548         (extern debug_body (const tree_node &)): New.
21549         (extern debug_body (const tree_node *)): New.
21550         (extern debug (vec<tree, va_gc> &)): New.
21551         (extern debug (vec<tree, va_gc> *)): New.
21552         * tree-cfg.c
21553         (debug (struct loop &)): New.
21554         (debug (struct loop *)): New.
21555         (debug_verbose (struct loop &)): New.
21556         (debug_verbose (struct loop *)): New.
21557         * tree-dump.c: Add header dependence.
21558         * tree-flow.h
21559         (extern debug (struct loop &)): New.
21560         (extern debug (struct loop *)): New.
21561         (extern debug_verbose (struct loop &)): New.
21562         (extern debug_verbose (struct loop *)): New.
21563         * tree-data-ref.c
21564         (debug (data_reference &)): New.
21565         (debug (data_reference *)): New.
21566         (debug (vec<data_reference_p> &)): New.
21567         (debug (vec<data_reference_p> *)): New.
21568         (debug (vec<ddr_p> &)): New.
21569         (debug (vec<ddr_p> *)): New.
21570         * tree-data-ref.h
21571         (extern debug (data_reference &)): New.
21572         (extern debug (data_reference *)): New.
21573         (extern debug (vec<data_reference_p> &)): New.
21574         (extern debug (vec<data_reference_p> *)): New.
21575         (extern debug (vec<ddr_p> &)): New.
21576         (extern debug (vec<ddr_p> *)): New.
21577         * tree-ssa-alias.c
21578         (debug (pt_solution &)): New.
21579         (debug (pt_solution *)): New.
21580         * tree-ssa-alias.h
21581         (extern debug (pt_solution &)): New.
21582         (extern debug (pt_solution *)): New.
21583         * tree-ssa-alias.c
21584         (debug (_var_map &)): New.
21585         (debug (_var_map *)): New.
21586         (debug (tree_live_info_d &)): New.
21587         (debug (tree_live_info_d *)): New.
21588         * tree-ssa-alias.h
21589         (extern debug (_var_map &)): New.
21590         (extern debug (_var_map *)): New.
21591         (extern debug (tree_live_info_d &)): New.
21592         (extern debug (tree_live_info_d *)): New.
21594 2013-03-28  Jan Hubicka  <jh@suse.cz>
21596         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
21598 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
21600         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
21601         record only when desired or required.
21603 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
21605         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
21606         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
21608 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
21610         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
21611         (*andsi3_compare0_uxtw): New pattern.
21612         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
21613         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
21615 2013-03-28  Jan Hubicka  <jh@suse.cz>
21617         * data-streamer-in.c (streamer_read_gcov_count): New function.
21618         * gimple-streamer-out.c: Include value-prof.h.
21619         (output_gimple_stmt): Output histogram.
21620         (output_bb): Use streamer_write_gcov_count.
21621         * value-prof.c: Include data-streamer.h
21622         (dump_histogram_value): Add HIST_TYPE_MAX.
21623         (stream_out_histogram_value): New function.
21624         (stream_in_histogram_value): New function.
21625         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
21626         (stream_out_histogram_value, stream_in_histogram_value): Declare.
21627         * data-streamer-out.c (streamer_write_gcov_count): New function.
21628         (streamer_write_gcov_count_stream): New function.
21629         * lto-cgraph.c (lto_output_edge): Update counter streaming.
21630         (lto_output_node): Likewise.
21631         (input_node, input_edge): Likewise.
21632         * lto-streamer-out.c (output_cfg): Update streaming.
21633         * lto-streamer-in.c (input_cfg): Likewise.
21634         * data-streamer.h (streamer_write_gcov_count,
21635         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
21636         * gimple-streamer-in.c: Include value-prof.h
21637         (input_gimple_stmt): Input histograms.
21638         (input_bb): Update profile streaming.
21640 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21642         * genmodes.c (emit_max_int): New function.
21643         (emit_insn_modes_h): Added call to emit_max_function.
21644         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
21645         Added doc.
21646         * machmode.def: Fixed comment.
21648 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
21650         * combine.c (try_combine): Removed useless assert.
21651         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
21653 2013-03-28  Marek Polacek  <polacek@redhat.com>
21654             Richard Biener  <rguenther@suse.de>
21656         PR tree-optimization/56695
21657         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
21658         build signed result of a vector comparison.
21659         * tree-cfg.c (verify_gimple_comparison): Check that a result
21660         of a vector comparison has signed type.
21662 2013-03-28  Richard Biener  <rguenther@suse.de>
21664         PR tree-optimization/37021
21665         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
21666         do not restrict gaps between groups.
21667         * tree-vect-stmts.c (vectorizable_load): Properly account for
21668         a gap between groups.
21670 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
21672         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
21673         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
21674         is not enabled.
21676 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
21678         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
21679         * doc/extend.texi (Named Address Spaces): Ditto.
21680         (Variable Attributes): Ditto.
21682 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21684         * config.build: Add support for cygwin x64 target.
21685         * config.gcc: Likewise.
21686         * config.host: Likewise.
21687         * configure.ac: Likewise
21688         * configure: Regenerated.
21690 2013-03-27  Kai Tietz  <ktietz@redhat.com>
21692         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
21693         * config/i386/t-cygwin-w64: New file.
21694         * config/i386/cygwin-w64.h: New file.
21695         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
21696         and add support for x64-cygwin target.
21697         (CPP_SPEC): Likewise.
21698         (CXX_WRAP_SPEC_LIST): Undefine before define.
21699         (LIBGCJ_SONAME): Use 15 as version.
21701 2013-03-27  Richard Biener  <rguenther@suse.de>
21703         PR tree-optimization/56716
21704         * tree-ssa-structalias.c (perform_var_substitution): Adjust
21705         dumping for ref nodes.
21707 2013-03-27  Martin Jambor  <mjambor@suse.cz>
21709         PR tree-optimization/55334
21710         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
21711         restricted pointers to arrays.
21713 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
21715         * Makefile.in (.SUFFIXES): Add .cc.
21716         (.c.o): Apply same recipe for implicit rule .cc.o.
21718 2013-03-27  Richard Biener  <rguenther@suse.de>
21720         PR tree-optimization/37021
21721         * tree-vect-data-refs.c (vect_check_strided_load): Allow
21722         REALPART/IMAGPART_EXPRs around the supported refs.
21723         * tree-ssa-structalias.c (find_func_aliases): Assume that
21724         floating-point values are not used to transfer pointers.
21726 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
21728         * target.def (TARGET_HAS_IFUNC_P): New target hook.
21729         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
21730         * doc/tm.texi: Regenerate.
21731         * targhooks.h (default_has_ifunc_p): New.
21732         * targhooks.c (default_has_ifunc_p): Ditto.
21733         * config/linux-protos.h: New file.
21734         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
21735         hook for linux which disables support of indirect functions in android.
21736         * config/linux-android.c: New file.
21737         * config/t-linux-android.c: Ditto.
21738         * config.gcc: Added new object file linux-android.o.
21739         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
21740         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
21741         * varasm.c (do_assemble_alias): Likewise.
21742         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
21743         doesn't support indirect functions.
21744         * configure: Regenerate.
21746 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
21748         PR target/56102
21749         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
21750         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
21751         mult-word mode.
21753 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
21755         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
21757 2013-03-27  Terry Guo  <terry.guo@arm.com>
21759         * config/arm/arm-cores.def: Added core cortex-r7.
21760         * config/arm/arm-tune.md: Regenerated.
21761         * config/arm/arm-tables.opt: Regenerated.
21762         * doc/invoke.texi: Added entry for core cortex-r7.
21764 2013-03-27  Walter Lee  <walt@tilera.com>
21766         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
21767         double-decrement of next_scratch_regno.
21769 2013-03-27  Walter Lee  <walt@tilera.com>
21771         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
21772         input operands.
21773         (insn_v1mulus): Ditto.
21774         (insn_v2muls): Ditto.
21776 2013-03-27  Walter Lee  <walt@tilera.com>
21778         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
21779         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
21781 2013-03-27  Walter Lee  <walt@tilera.com>
21783         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
21784         (*sibcall_value): Ditto.
21786 2013-03-27  Walter Lee  <walt@tilera.com>
21788         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
21789         (insn_mnz_v8qi): ... this ...
21790         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
21791         vector equivalent.
21792         (insn_v<n>mnz): Replaced by ...
21793         (insn_v1mnz): ... this ...
21794         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
21795         equivalent.
21796         (insn_mz_<mode>): Replaced by ...
21797         (insn_mz_v8qi): ... this ...
21798         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
21799         vector equivalent.
21800         (insn_v<n>mz): Replaced by ...
21801         (insn_v1mz): ... this ...
21802         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
21803         equivalent.
21805 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
21807         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
21809 2013-03-26  Roland McGrath  <mcgrathr@google.com>
21811         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
21812         than fprintf with a non-constant, non-format string.
21814 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21816         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
21817         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
21818         operand 0 predicate.
21819         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
21820         attribute.  Use general_x64nomem_operand as operand 1 predicate.
21821         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
21822         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21823         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
21824         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
21825         (mov<mode>_insv_1): Remove expander.  Merge insn with
21826         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
21827         Use general_x64nomem_operand as operand 1 predicate.
21828         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
21829         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
21830         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21831         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
21832         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
21833         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
21834         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
21835         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
21836         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
21837         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
21838         (general_x64nomem_operand): Ditto.
21840 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
21842         * config/rtems.opt: Add -pthread option.
21844 2013-03-26  Richard Biener  <rguenther@suse.de>
21846         * alias.c (find_base_term): Avoid redundant and not used recursion.
21847         (base_alias_check): Get the initial base term from the caller.
21848         (true_dependence_1): Compute and pass base terms to base_alias_check.
21849         (write_dependence_p): Likewise.
21850         (may_alias_p): Likewise.
21852 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
21854         * config/aarch64/aarch64.c (aarch64_classify_address): Support
21855         PC-relative load in SI modes and above only.
21857 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
21859         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
21860         * config/arm/iwmmxt.md (WCGR0): Update.
21861         (WCGR1, WCGR2, WCGR3): Likewise.
21863 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
21865         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
21866         Use x64 and nox64 isa attributes.
21868 2013-03-26  Richard Biener  <rguenther@suse.de>
21870         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
21871         alignment computations and rely on get_object_alignment_1
21872         for the !TYPE_P case.
21873         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
21875 2013-03-26  Walter Lee  <walt@tilera.com>
21877         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
21878         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
21880 2013-03-25  Jeff Law  <law@redhat.com>
21882         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
21883         check for INTEGRAL_TYPE_P that was missing due to checking in
21884         wrong version of prior patch.
21886 2013-03-25  Walter Lee  <walt@tilera.com>
21888         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
21889         TILEGX_INSN_SHUFFLEBYTES1.
21890         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
21891         shufflebytes1.
21892         (tilegx_builtins): Ditto.
21893         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
21895 2013-03-25  Walter Lee  <walt@tilera.com>
21897         * config/tilegx/tilegx.md (floatsisf2): New pattern.
21898         (floatunssisf2): New pattern.
21899         (floatsidf2): New pattern.
21900         (floatunssidf2): New pattern.
21902 2013-03-25  Walter Lee  <walt@tilera.com>
21904         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
21905         tests for constraint J, K, N, P.
21907 2013-03-25  Walter Lee  <walt@tilera.com>
21909         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
21910         Use indirect/pcrel encoding.
21911         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
21912         Ditto.
21914 2013-03-25  Steve Ellcey  <sellcey@mips.com>
21916         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
21917         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
21918         * config/mips/mips.c (mips_option_override): Set IMADD default.
21919         * config/mips/mips.h (PTF_AVOID_IMADD): New.
21920         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
21921         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
21922         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
21923         * doc/invoke.texi (-mimadd/-mno-imadd): New.
21925 2013-03-25  Jeff Law  <law@redhat.com>
21927         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
21928         slightly to avoid creating and folding useless trees.  Simplify
21929         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
21931 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
21933         * config/i386/i386.md (*zero_extendsidi2): Merge with
21934         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
21935         * config/i386/predicates.md (x86_64_zext_operand): Rename from
21936         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
21937         targets.  Clarify comment.
21939 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21941         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
21942         pass-through jump functions differently.
21943         (ipa_read_jump_function): Likewise.  Also use setter functions to set
21944         up jump functions.
21946 2013-03-25  Martin Jambor  <mjambor@suse.cz>
21948         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
21949         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
21950         process it.
21951         (ipa_get_indirect_edge_target): New function.
21952         (devirtualization_time_bonus): New parameter known_aggs, pass it to
21953         ipa_get_indirect_edge_target.  Update all callers.
21954         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
21955         ipa_get_indirect_edge_target_1 instead of calling
21956         ipa_get_indirect_edge_target.
21957         (create_specialized_node): Pass aggvlas to
21958         ipcp_discover_new_direct_edges.
21960 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21962         * config/arm/arm.md (f_sels, f_seld): New types.
21963         (*cmov<mode>): New pattern.
21964         * config/arm/predicates.md (arm_vsel_comparison_operator): New
21965         predicate.
21967 2013-03-25  Kai Tietz  <ktietz@redhat.com>
21969         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
21970         POSIX-printf for mingw-hosted  builds.
21972 2013-03-25  Richard Biener  <rguenther@suse.de>
21974         PR middle-end/56694
21975         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
21976         must-not-throw stmt location.
21978 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21980         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
21981         Emit load-acquire versions when acq is true.
21982         (arm_emit_store_exclusive): Add rel parameter.
21983         Emit store-release versions when rel is true.
21984         (arm_split_compare_and_swap): Use acquire-release instructions
21985         instead.
21986         of barriers when appropriate.
21987         (arm_split_atomic_op): Likewise.
21988         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
21989         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
21990         (VUNSPEC_SLX): Likewise.
21991         (VUNSPEC_LDA): Likewise.
21992         (VUNSPEC_STL): Likewise.
21993         * config/arm/sync.md (atomic_load<mode>): New pattern.
21994         (atomic_store<mode>): Likewise.
21995         (arm_load_acquire_exclusive<mode>): Likewise.
21996         (arm_load_acquire_exclusivesi): Likewise.
21997         (arm_load_acquire_exclusivedi): Likewise.
21998         (arm_store_release_exclusive<mode>): Likewise.
22000 2013-03-25  Catherine Moore  <clm@codesourcery.com>
22002         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
22003         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
22004         * config/mip/predicates.md (lwsp_swsp_operand,
22005         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
22006         sb16_operand, db4_operand, db7_operand, ib3_operand,
22007         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
22008         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
22009         andi16_operand): New predicates.
22010         * config/mips/mips.md (compression): New attribute.
22011         (enabled): New attribute.
22012         (length): Consider compression in computing length.
22013         (shift_compression): New code attribute.
22014         (*add<mode>3): New operands. Record compression.
22015         (sub<mode>3): Likewise.
22016         (one_cmpl<mode>2): Likewise.
22017         (*and<mode>3): Likewise.
22018         (*ior<mode>3): Likewise.
22019         (unnamed pattern for xor): Likewise.
22020         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
22021         (*<optab><mode>3): Likewise.
22022         (*mov<mode>_internal: Likewise.
22023         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
22024         (mips_unsigned_immediate_p): New.
22025         (umips_lwsp_swsp_address_p): New.
22026         (m16_based_address_p): New.
22027         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
22028         (mips_unsigned_immediate_p): New prototype.
22029         (lwsp_swsp_address_p): New prototype.
22030         (m16_based_address_p): New prototype.
22031         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
22032         (mips_signed_immediate_p): New function.
22033         (m16_based_address_p): New function.
22034         (lwsp_swsp_address_p): New function.
22035         (mips_print_operand_punctuation): Recognize short delay slot insns
22036         for microMIPS.add<mode>3"
22038 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22040         PR target/56720
22041         * config/arm/iterators.md (v_cmp_result): New mode attribute.
22042         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
22044 2013-03-25  Richard Biener  <rguenther@suse.de>
22046         PR tree-optimization/56689
22047         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
22048         any edge.
22050 2013-03-25  Richard Biener  <rguenther@suse.de>
22052         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
22053         of bitmap.
22054         (memory_references): Likewise.
22055         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
22056         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
22057         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
22058         (gather_mem_refs_in_loops): Fold into ...
22059         (analyze_memory_references): ... this.  Move initialization
22060         to tree_ssa_lim_initialize.
22061         (fill_always_executed_in): Rename to ...
22062         (fill_always_executed_in_1): ... this.
22063         (fill_always_executed_in): Move contains_call computation to
22064         this new function from ...
22065         (tree_ssa_lim_initialize): ... here.
22066         (tree_ssa_lim): Call fill_always_executed_in.
22068 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22070         * postreload.c (reload_combine): Fix code detecting returns.
22072 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
22074         * function.c (emit_use_return_register_into_block): On cc0 targets,
22075         do not emit the sequence between cc0 setter and user.
22077 2013-03-25  Kai Tietz  <ktietz@redhat.com>
22079         * config/i386/predicates.md (local_symbolic_operand): Interpret
22080         dll-imported symbols as none-local.
22082 2013-03-25  Richard Biener  <rguenther@suse.de>
22084         * tree-ssa-loop-im.c (struct depend): Remove.
22085         (struct lim_aux_data): Make depends a vec of gimples.
22086         (free_lim_aux_data): Adjust.
22087         (add_dependency): Likewise.
22088         (set_level): Likewise.
22090 2013-03-25  Richard Biener  <rguenther@suse.de>
22092         PR middle-end/56434
22093         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
22094         the pointer returned by calls with ECF_MALLOC set.
22096 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22098         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
22100 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
22102         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
22103         using MMXMODE mode iterator.
22104         (*move<mode>_internal): Merge with *movv2sf_internal and
22105         *movv2sf_internal_rex64 using MMXMODE mode iterator.
22107 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22109         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
22110         (record_last_mem_set_info): Likewise.
22112         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
22113         of XNEWVEC followed by memset.
22114         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
22116 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
22118         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
22119         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
22120         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
22121         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
22122         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
22123         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
22124         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
22125         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
22126         BARRIER_P instead of GET_CODE.
22128 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22130         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
22131         inaccuracy in the probing code.
22133         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
22134         (ctrapdi4): Likewise.
22136 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22138         * calls.c (expand_call): Add missing guard to code handling return
22139         of non-BLKmode structures in MSB.
22140         * function.c (expand_function_end): Likewise.
22142 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
22144         * combine.c (try_combine): Adjust comment.  Do not add the set of
22145         insn #0 if the destination indirectly is set or dies in insn #2.
22146         Tidy up code to distribute a new note.
22148 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22150         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
22151         also for alternatives 16 and 17.
22153 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
22155         * config/i386/sse.md (*mov<mode>_internal): Merge with
22156         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
22157         Emit insn template depending on type attribute.  Use
22158         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22159         movd instead of movq mnemonic for interunit moves.  Rewrite mode
22160         attribute calculation.  Remove unit attribute calculation.
22161         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
22162         Set prefix_data16 attribute for DImode ssemov types.
22163         Use Ym instead of y for SSE-MMX conversion alternatives.
22164         Reorder operand constraints.
22166 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
22168         * df.h (df_insn_delete): Adjust prototype.
22169         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
22170         and let it decide whether mark the basic block dirty.
22171         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
22172         * df-scan.c (df_insn_info_delete): New helper function, split
22173         off from df_insn_delete.
22174         (df_scan_free_bb_info): Use it.
22175         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
22176         Likewise.
22177         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
22178         that the insn is actually an insn and it has a non-NULL basic block.
22179         Do not mark basic block dirty if only deleting a DEBUG_INSN.
22181 2013-03-22  Richard Biener  <rguenther@suse.de>
22183         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
22184         dep_ref members.
22185         (mem_ref_alloc): Do not allocate them.
22186         (refs_independent_p): Do not query or maintain a cache.
22188 2013-03-22  Richard Biener  <rguenther@suse.de>
22190         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
22191         (gather_mem_refs_in_loops): Do not compute it.
22192         (analyze_memory_references): Do not allocate it.
22193         (tree_ssa_lim_finalize): Do not free it.
22194         (for_all_locs_in_loop): Do not query all_refs_in_loop.
22196 2013-03-22  Richard Biener  <rguenther@suse.de>
22198         * is-a.h (as_a): Use gcc_checking_assert.
22200 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
22202         * config/aarch64/aarch64.c (aarch64_print_operand): New
22203         format specifier for printing a constant in hex.
22204         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
22205         format specifier for printing second operand.
22207 2013-03-22  Richard Biener  <rguenther@suse.de>
22209         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
22210         bitmaps.
22211         (gather_mem_refs_in_loops): Perform store accumulation here.
22212         (create_vop_ref_mapping_loop): Remove.
22213         (create_vop_ref_mapping): Likewise.
22214         (analyze_memory_references): Initialize refs_stored_in_loop.
22215         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
22216         (record_indep_loop): Remove.
22217         (record_dep_loop): New function.
22218         (ref_indep_loop_p_1): Adjust to only walk over references
22219         in the loop, not its subloops.
22220         (ref_indep_loop_p): Rename to ...
22221         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
22222         maintaining a more fine-grained cache.
22223         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
22224         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
22226 2013-03-22  Richard Biener  <rguenther@suse.de>
22228         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
22229         (struct mem_ref): Make accesses_in_loop a vec of a vec of
22230         aggregate mem_ref_loc.
22231         (free_mem_ref_locs): Inline into ...
22232         (memref_free): ... this and adjust.
22233         (mem_ref_alloc): Adjust.
22234         (mem_ref_locs_alloc): Remove.
22235         (record_mem_ref_loc): Adjust.
22236         (get_all_locs_in_loop): Rewrite into ...
22237         (for_all_locs_in_loop): ... this iterator.
22238         (rewrite_mem_ref_loc): New functor.
22239         (rewrite_mem_refs): Use for_all_locs_in_loop.
22240         (sm_set_flag_if_changed): New functor.
22241         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
22242         (ref_always_accessed): New functor.
22243         (ref_always_accessed_p): Use for_all_locs_in_loop.
22245 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22247         * tree-pass.h (PROP_gimple_lvec): New.
22248         * passes.c (dump_properties): Handle PROP_gimple_lvec.
22249         (init_optimization_passes): Move pass_lower_vector.
22250         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
22251         PROP_gimple_lvec.
22252         (pass_lower_vector): Provide PROP_gimple_lvec.
22253         (pass_lower_vector_ssa): Likewise.
22254         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
22256 2013-03-21  Mark Wielaard  <mjw@redhat.com>
22258         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
22260 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22262         * config/i386/i386.md (*movdi_internal): Disparage slightly
22263         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
22264         conversion alternatives.
22266 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
22268         PR middle-end/48087
22269         * diagnostic.def (DK_WERROR): New kind.
22270         * diagnostic.h (werrorcount): Define.
22271         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
22272         promoted to DK_ERROR, increment DK_WERROR counter instead of
22273         DK_ERROR counter.
22274         * toplev.c (toplev_main): Call print_ignored_options even if
22275         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
22276         even if just werrorcount is non-zero.
22278         PR debug/55608
22279         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
22280         on failure.
22281         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
22282         (string_cst_pool_decl): New function.
22283         (optimize_one_addr_into_implicit_ptr): New function.
22284         (resolve_addr_in_expr): Optimize DWARF location expression
22285         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
22286         which doesn't live in memory, but has DW_AT_location or
22287         DW_AT_const_value, or refers to a string literal, into
22288         DW_OP_GNU_implicit_pointer.
22289         (optimize_location_into_implicit_ptr): New function.
22290         (resolve_addr): If removing DW_AT_location of a variable because
22291         it was DW_OP_addr of address of the variable, but the variable doesn't
22292         live in memory, try to emit const value attribute for the initializer.
22294 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22296         * tree.h (VECTOR_TYPE_P): New macro.
22297         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
22298         TYPE_MODE): Use it.
22299         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
22300         VEC_COND_EXPR cannot be lvalues.
22301         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
22303 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
22305         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
22306         Restrict the transformation to equal modes.
22308 2013-03-21  Richard Biener  <rguenther@suse.de>
22310         PR tree-optimization/39326
22311         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
22312         (MEM_ANALYZABLE): Adjust.
22313         (record_mem_ref_loc): Move bitmap ops ...
22314         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
22315         unanalyzable refs, do not record locations for it.
22316         (analyze_memory_references): Allocate ref zero as shared
22317         unanalyzable ref.
22318         (refs_independent_p): Do not test for unanalyzed mems here.
22319         (ref_indep_loop_p_1): Special-case disambiguation against
22320         the unanalyzed ref.
22321         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
22323 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
22325         * config/arm/arm-protos.h (tune_params): Add
22326         prefer_neon_for_64bits field.
22327         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
22328         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
22329         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
22330         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
22331         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
22332         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
22333         (arm_option_override): Handle -mneon-for-64bits new option.
22334         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
22335         (prefer_neon_for_64bits): Declare new variable.
22336         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
22337         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
22338         (arch_enabled): Handle new arch types. Remove support for onlya8
22339         and nota8.
22340         (one_cmpldi2): Use new arch names.
22341         (zero_extend<mode>di2, extend<mode>di2): Ditto.
22342         * config/arm/arm.opt (mneon-for-64bits): Add option.
22343         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
22344         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
22345         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
22346         of onlya8.
22347         * doc/invoke.texi (-mneon-for-64bits): Document.
22349 2013-03-21  Richard Biener  <rguenther@suse.de>
22351         PR tree-optimization/39326
22352         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
22353         (sort_bbs_in_loop_postorder_cmp): New function.
22354         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
22356 2013-03-21  Richard Biener  <rguenther@suse.de>
22358         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
22359         (vect_insert_into_interleaving_chain): Likewise.
22360         (vect_drs_dependent_in_basic_block): Inline ...
22361         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
22362         split out from ...
22363         (vect_analyze_data_ref_dependence): ... here.  Simplify.
22364         (vect_check_interleaving): Simplify.
22365         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
22366         (vect_slp_analyze_data_ref_dependences): ... this new function.
22367         (dr_group_sort_cmp): New function.
22368         (vect_analyze_data_ref_accesses): Compute data-reference groups
22369         here instead of in vect_analyze_data_ref_dependence.  Use
22370         a more efficient algorithm.
22371         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
22372         vect_slp_analyze_data_ref_dependences.  Call
22373         vect_analyze_data_ref_accesses earlier.
22374         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
22375         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
22376         (vect_slp_analyze_data_ref_dependences): New prototype.
22378 2013-03-21  Richard Biener  <rguenther@suse.de>
22380         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
22381         ref is stored in the loop.
22382         (find_refs_for_sm): Walk only over all stores.
22383         (store_motion_loop): Allocate from lim_bitmap_obstack.
22384         (store_motion): Likewise.
22386 2013-03-21  Richard Biener  <rguenther@suse.de>
22388         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
22389         Update virtual SSA form.
22391 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22393         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
22394         * configure: Regenerate.
22395         * config.in: Regenerate.
22396         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
22397         if !HAVE_LD_EH_FRAME_CIEV3.
22399 2013-03-21  Richard Biener  <rguenther@suse.de>
22401         * tree-cfg.c (verify_expr_no_block): New function.
22402         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
22403         nor DECL_VALUE_EXPR have locations with associated blocks.
22404         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
22405         (clear_unused_block_pointer): Remove code dealing with
22406         blocks in DECL_DEBUG_EXPR locations.
22408 2013-03-21  Richard Biener  <rguenther@suse.de>
22410         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
22411         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
22412         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
22413         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
22414         instead of DECL_DEBUG_EXPR_IS_FROM.
22415         * gimplify.c (gimplify_modify_expr): Likewise.
22416         * tree-cfg.c (verify_expr_location_1): Likewise.
22417         * tree-complex.c (create_one_component_var): Likewise.
22418         * tree-sra.c (create_access_replacement): Likewise.
22419         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
22420         (clear_unused_block_pointer): Likewise.
22421         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
22422         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
22423         * var-tracking.c (var_debug_decl): Likewise.
22424         (track_expr_p): Likewise.
22425         * tree-inline.c (add_local_variables): Likewise.  Set
22426         DECL_HAS_DEBUG_EXPR_P after copying it.
22427         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
22428         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
22430 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
22432         PR bootstrap/56656
22433         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
22434         * configure: Regenerate.
22435         * config.in: Regenerate.
22436         * config/i386/i386.md (*movdf_internal): Use
22437         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
22438         movd instead of movq mnemonic for interunit moves.
22439         (*movdi_internal): Ditto.
22441 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22443         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
22444         (abd<mode>_3): New pattern.
22445         (aba<mode>_3): New pattern.
22446         (fabd<mode>_3): New pattern.
22448 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22450         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
22451         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
22452         occurrence of REGISTER_PREFIX as its empty string.
22454 2013-03-20  Jeff Law  <law@redhat.com>
22456         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
22457         addititional equivalences for equality comparisons between an SSA_NAME
22458         and a constant where the SSA_NAME was set from a widening conversion.
22460 2013-03-20  Walter Lee  <walt@tilera.com>
22462         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
22464 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22466         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
22467         depending on type attribute.
22468         (*movti_internal): Ditto.
22469         (*movtf_internal): Ditto.
22470         (*movxf_internal): Ditto.
22471         (*movdf_internal): Ditto.
22472         (*movsf_internal): Ditto.
22474 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22476         * config/i386/i386.md (*movti_internal): Set prefix attribute to
22477         maybe_vex for sselog1 and ssemov types.
22478         (*movdi_internal): Reorder operand constraints.
22479         (*movsi_internal): Ditto.  Set prefix attribute to
22480         maybe_vex for sselog1 and ssemov types.
22481         (*movtf_internal): Set prefix attribute to maybe_vex
22482         for sselog1 and ssemov types.
22483         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
22484         DImode ssemov types.  Reorder operand constraints.
22485         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
22486         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
22487         attribute for SImode ssemov types.  Reorder operand constraints.
22489 2013-03-20  Martin Jambor  <mjambor@suse.cz>
22491         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
22492         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
22494 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
22496         * config/rs6000/predicates.md (indexed_address, update_address_mem
22497         update_indexed_address_mem): New predicates.
22498         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
22499         attribute for load/store instructions.
22500         * config/rs6000/dfp.md (movsd_store): Likewise.
22501         (movsd_load): Likewise.
22502         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
22503         (unnamed HI->DI extend define_insn): Likewise.
22504         (unnamed SI->DI extend define_insn): Likewise.
22505         (unnamed QI->SI extend define_insn): Likewise.
22506         (unnamed QI->HI extend define_insn): Likewise.
22507         (unnamed HI->SI extend define_insn): Likewise.
22508         (unnamed HI->SI extend define_insn): Likewise.
22509         (extendsfdf2_fpr): Likewise.
22510         (movsi_internal1): Likewise.
22511         (movsi_internal1_single): Likewise.
22512         (movhi_internal): Likewise.
22513         (movqi_internal): Likewise.
22514         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
22515         attribute for load/store instructions.
22516         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
22517         instructions.
22518         (mov<mode>_softfloat): Likewise.
22519         (mov<mode>_hardfloat32): Likewise.
22520         (mov<mode>_hardfloat64): Likewise.
22521         (mov<mode>_softfloat64): Likewise.
22522         (movdi_internal32): Likewise.
22523         (movdi_internal64): Likewise.
22524         (probe_stack_<mode>): Likewise.
22526 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
22528         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
22529         floating point, and decimal floating point to reload iterator.
22531         * config/rs6000/constraints.md (wl constraint): New constraints to
22532         return FLOAT_REGS if certain options are used to reduce the number
22533         of separate patterns that exist in the file.
22534         (wx constraint): Likewise.
22535         (wz constraint): Likewise.
22537         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22538         -mdebug=reg, print wg, wl, wx, and wz constraints.
22539         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22540         Initialize the reload functions for 64-bit binary/decimal floating
22541         point types.
22542         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22543         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22544         create the buffer on the stack to overcome not having a 32-bit
22545         load and store.
22546         (rs6000_emit_move): Likewise.
22547         (rs6000_secondary_memory_needed_rtx): Likewise.
22548         (rs6000_alloc_sdmode_stack_slot): Likewise.
22549         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22550         via xxlxor, just like DFmode 0.0.
22552         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
22553         define as 1 if we are running on a power7 or newer.
22554         (enum r6000_reg_class_enum): Add new constraints.
22556         * config/rs6000/dfp.md (movsd): Delete, combine with binary
22557         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22558         with other moves by using conditional constraits (wg).  Use LFIWZX
22559         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
22560         (movsd splitter): Likewise.
22561         (movsd_hardfloat): Likewise.
22562         (movsd_softfloat): Likewise.
22564         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
22565         binary and decimal floating point moves.
22566         (fmove_ok): New attributes to combine binary and decimal floating
22567         point moves, and to combine power6x (mfpgpr) moves along normal
22568         floating moves.
22569         (real_value_to_target): Likewise.
22570         (f32_lr): Likewise.
22571         (f32_lm): Likewise.
22572         (f32_li): Likewise.
22573         (f32_sr): Likewise.
22574         (f32_sm): Likewise.
22575         (f32_si): Likewise.
22576         (movsf): Combine binary and decimal floating point moves.  Combine
22577         power6x (mfpgpr) moves with other moves by using conditional
22578         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
22579         (mov<mode> for SFmode/SDmode); Likewise.
22580         (SFmode/SDmode splitters): Likewise.
22581         (movsf_hardfloat): Likewise.
22582         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
22583         (movsf_softfloat): Likewise.
22584         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
22586         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
22587         wx and wz constraints.
22589         * config/rs6000/constraints.md (wg constraint): New constraint to
22590         return FLOAT_REGS if -mmfpgpr (power6x) was used.
22592         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
22593         constraint.
22595         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
22596         -mdebug=reg, print wg, wl, wx, and wz constraints.
22597         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
22598         Initialize the reload functions for 64-bit binary/decimal floating
22599         point types.
22600         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
22601         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
22602         create the buffer on the stack to overcome not having a 32-bit
22603         load and store.
22604         (rs6000_emit_move): Likewise.
22605         (rs6000_secondary_memory_needed_rtx): Likewise.
22606         (rs6000_alloc_sdmode_stack_slot): Likewise.
22607         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
22608         via xxlxor, just like DFmode 0.0.
22610         * config/rs6000/dfp.md (movdd): Delete, combine with binary
22611         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
22612         with other moves by using conditional constraits (wg).  Use LFIWZX
22613         and STFIWX for loading SDmode on power7.
22614         (movdd splitters): Likewise.
22615         (movdd_hardfloat32): Likewise.
22616         (movdd_softfloat32): Likewise.
22617         (movdd_hardfloat64_mfpgpr): Likewise.
22618         (movdd_hardfloat64): Likewise.
22619         (movdd_softfloat64): Likewise.
22621         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
22622         64-bit binary and decimal floating point moves.
22623         (FMOVE64X): Likewise.
22624         (movdf): Combine 64-bit binary and decimal floating point moves.
22625         Combine power6x (mfpgpr) moves with other moves by using
22626         conditional constraits (wg).
22627         (mov<mode> for DFmode/DDmode): Likewise.
22628         (DFmode/DDmode splitters): Likewise.
22629         (movdf_hardfloat32): Likewise.
22630         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
22631         (movdf_softfloat32): Likewise.
22632         (movdf_hardfloat64_mfpgpr): Likewise.
22633         (movdf_hardfloat64): Likewise.
22634         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
22635         (movdf_softfloat64): Likewise.
22636         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
22637         (reload_<mode>_load): Move to later in the file so they aren't in
22638         the middle of the floating point move insns.
22639         (reload_<mode>_store): Likewise.
22641         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
22642         constraint.
22644         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
22645         constraint if -mdebug=reg.
22646         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
22647         Enable using dd reload support if needed.
22649         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
22650         binary and decimal floating point moves in rs6000.md.
22651         (movtd_internal): Likewise.
22653         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
22654         decimal floating point moves.
22655         (movtf): Likewise.
22656         (movtf_internal): Likewise.
22657         (mov<mode>_internal, TDmode/TFmode): Likewise.
22658         (movtf_softfloat): Likewise.
22659         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
22661         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
22662         movdi_internal64, using wg constraint for move direct operations.
22663         (movdi_internal64): Likewise.
22665         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
22666         MODES_TIEABLE_P for selected modes.  Print the numerical value of
22667         the various virtual registers. Use GPR/FPR first/last values,
22668         instead of hard coding the register numbers.  Print which modes
22669         have reload functions registered.
22670         (rs6000_option_override_internal): If -mdebug=reg, trace the options
22671         settings before/after setting cpu, target and subtarget settings.
22672         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
22673         and for secondary reload failures in rs6000_secondary_reload_inner.
22674         (rs6000_secondary_reload_fail): Likewise.
22675         (rs6000_secondary_reload_inner): Likewise.
22677         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
22678         macros for first/last GPR and FPR registers.
22679         (LAST_GPR_REGNO): Likewise.
22680         (FIRST_FPR_REGNO): Likewise.
22681         (LAST_FPR_REGNO): Likewise.
22683         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
22684         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
22685         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
22686         (vcond<mode><mode>): Likewise.
22687         (vcondu<mode><mode>): Likewise.
22688         (vector_gtu<mode>): Likewise.
22689         (vector_gte<mode>): Likewise.
22690         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
22691         to prevent the compiler from converting DImode operations to TImode.
22692         (ior<mode>3): Likewise.
22693         (and<mode>3): Likewise.
22694         (one_cmpl<mode>2): Likewise.
22695         (nor<mode>3): Likewise.
22696         (andc<mode>3): Likewise.
22698         * config/rs6000/constraints.md (wt constraint): New constraint
22699         that returns VSX_REGS if TImode is allowed in VSX registers.
22701         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
22702         constant under VSX.
22704         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
22705         similar to TImode, but it is restricted to being in the GPRs.
22707         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
22708         TImode to occupy a single VSX register.
22710         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
22711         -mvsx-timode for power7/power8.
22712         (power7 cpu): Likewise.
22713         (power8 cpu): Likewise.
22715         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
22716         sure that TFmode/TDmode take up two registers if they are ever
22717         allowed in the upper VSX registers.
22718         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
22719         registers.
22720         (rs6000_init_hard_regno_mode_ok): Likewise.
22721         (rs6000_debug_reg_global): Add debugging for PTImode and wt
22722         constraint.  Print if LRA is turned on.
22723         (rs6000_option_override_internal): Give an error if -mvsx-timode
22724         and VSX is not enabled.
22725         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
22726         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
22727         to reg+offset addressing.  Use PTImode when checking offset
22728         addresses for validity.
22729         (reg_offset_addressing_ok_p): Likewise.
22730         (rs6000_legitimate_offset_address_p): Likewise.
22731         (rs6000_legitimize_address): Likewise.
22732         (rs6000_legitimize_reload_address): Likewise.
22733         (rs6000_legitimate_address_p): Likewise.
22734         (rs6000_eliminate_indexed_memrefs): Likewise.
22735         (rs6000_emit_move): Likewise.
22736         (rs6000_secondary_reload): Likewise.
22737         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
22738         reloads to fpr registers to continue to use reg+offset addressing,
22739         but 64-bit reloads to altivec registers need reg+reg addressing.
22740         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
22741         it.  Treat LO_SUM like a PLUS operation.
22742         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
22743         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
22744         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
22745         registers to share a register with a smaller sized type, since VSX
22746         puts scalars in the upper 64-bits.
22747         (print_operand): Add support for PTImode.
22748         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
22749         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
22750         registers, but don't have arithmetic support.
22751         (rs6000_memory_move_cost): Add test for VSX.
22752         (rs6000_opt_masks): Add -mvsx-timode.
22754         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
22755         for TImode.
22756         (VSs): Likewise.
22757         (VSr): Use wt constraint for TImode.
22758         (VSv): Drop TImode support.
22759         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
22760         (vsx_movti_64bit): Likewise.
22761         (vsx_movti_32bit): Likewise.
22762         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
22763         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
22764         one '?' on the appropriate output constraint.  Do not allow TImode
22765         logical operations on 32-bit systems.
22766         (vsx_ior<mode>3): Likewise.
22767         (vsx_xor<mode>3): Likewise.
22768         (vsx_one_cmpl<mode>2): Likewise.
22769         (vsx_nor<mode>3): Likewise.
22770         (vsx_andc<mode>3): Likewise.
22771         (vsx_concat_<mode>): Likewise.
22772         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
22774         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
22775         OPTION_MASK_VSX_TIMODE.
22776         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
22777         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
22779         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
22780         (TI2 iterator): New iterator for TImode, PTImode.
22781         (wd mode attribute): Add values for vector types.
22782         (movti_string): Replace TI move operations with operations for TImode
22783         and PTImode.  Add support for TImode being allowed in VSX registers.
22784         (mov<mode>_string, TImode/PTImode): Likewise.
22785         (movti_ppc64): Likewise.
22786         (mov<mode>_ppc64, TImode/PTImode): Likewise.
22787         (TI mode splitters): Likewise.
22789         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
22790         constraint.
22792 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
22794         PR tree-optimization/56355
22795         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
22796         Also handle integers with undefined overflow.
22798 2013-03-20  Catherine Moore  <clm@codesourcery.com>
22799             Maciej W. Rozycki  <macro@codesourcery.com>
22800             Tom de Vries  <tom@codesourcery.com>
22801             Nathan Sidwell  <nathan@codesourcery.com>
22802             Iain Sandoe  <iain@codesourcery.com>
22803             Nathan Froyd  <froydnj@codesourcery.com>
22804             Chao-ying Fu  <fu@mips.com>
22806         * doc/extend.texi: (micromips, nomicromips, nocompression):
22807         Document new function attributes.
22808         * doc/invoke.texi (minterlink-compressed, mmicromips,
22809         m14k, m14ke, m14kec): Document new options.
22810         (minterlink-mips16): Update documentation.
22811         * doc/md.texi (ZC, ZD): Document new constraints.
22812         * configure.ac (gcc_cv_as_micromips): Check if linker
22813         supports the .set micromips directive.
22814         * configure: Regenerate.
22815         * config.in: Regenerate.
22816         * config/mips/mips-tables.opt: Regenerate.
22817         * config/mips/micromips.md: New file.
22818         * constraints.md (ZC, ZD): New constraints.
22819         * config/mips/predicates.md (movep_src_register): New predicate.
22820         (movep_src_operand): New predicate.
22821         (non_volatile_mem_operand): New predicate.
22822         * config/mips/mips.md (multimem): New type.
22823         (length): Differentiate between 17-bit and 18-bit branch offsets.
22824         (MOVEP1, MOVEP2): New mode iterator.
22825         (mov_<load>l): Use ZC constraint.
22826         (mov_<load>r): Likewise.
22827         (mov_<store>l): Likewise.
22828         (mov_<store>r): Likewise.
22829         (*branch_equality<mode>_inverted): Add microMIPS support.
22830         (*branch_equality<mode>): Likewise.
22831         (*jump_absolute): Likewise.
22832         (indirect_jump_<mode>): Likewise.
22833         (tablejump_<mode>): Likewise.
22834         (<optab>_internal): Likewise.
22835         (sibcall_internal): Likewise.
22836         (sibcall_value_internal): Likewise.
22837         (prefetch): Use constraint ZD.
22838         * config/mips/mips.opt (minterlink-compressed): New option.
22839         (minterlink-mips16): Now an alias for minterlink-compressed.
22840         (mmicromips): New option.
22841         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
22842         (compare_and_swap_12): Likewise.
22843         (sync_add<mode>): Likewise.
22844         (sync_<optab>_12): Likewise.
22845         (sync_old_<optab>_12): Likewise.
22846         (sync_new_<optab>_12): Likewise.
22847         (sync_nand_12): Likewise.
22848         (sync_old_nand_12): Likewise.
22849         (sync_new_nand_12): Likewise.
22850         (sync_sub<mode>): Likewise.
22851         (sync_old_add<mode>): Likewise.
22852         (sync_old_sub<mode>): Likewise.
22853         (sync_new_add<mode>): Likewise.
22854         (sync_new_sub<mode>): Likewise.
22855         (sync_<optab><mode>): Likewise.
22856         (sync_old_<optab><mode>): Likewise.
22857         (sync_new_<optab><mode>): Likewise.
22858         (sync_nand<mode>): Likewise.
22859         (sync_old_nand<mode>): Likewise.
22860         (sync_new_nand<mode>): Likewise.
22861         (sync_lock_test_and_set<mode>): Likewise.
22862         (test_and_set_12): Likewise.
22863         (atomic_compare_and_swap<mode>): Likewise.
22864         (atomic_exchange<mode>_llsc): Likewise.
22865         (atomic_fetch_add<mode>_llsc): Likewise.
22866         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
22867         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
22868         (umips_save_restore_pattern_p): Likewise.
22869         (umips_load_store_pair_p): Likewise.
22870         (umips_output_load_store_pair): Likewise.
22871         (umips_movep_target_p): Likewise.
22872         (umips_12bit_offset_address_p): Likewise.
22873         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
22874         (mips_base_mips16): Rename this...
22875         (mips_base_compression_flags): ...to this. Update all uses.
22876         (mips_attribute_table): Add micromips, nomicromips and nocompression.
22877         (mips_mips16_decl_p): Delete.
22878         (mips_nomips16_decl_p): Delete.
22879         (mips_get_compress_on_flags): New function.
22880         (mips_get_compress_off_flags): New function.
22881         (mips_get_compress_mode): New function.
22882         (mips_get_compress_on_name): New function.
22883         (mips_get_compress_off_name): New function.
22884         (mips_insert_attributes): Support multiple compression types.
22885         (mips_merge_decl_attributes): Likewise.
22886         (umips_12bit_offset_address_p): New function.
22887         (mips_start_function_definition): Emit .set micromips directive.
22888         (mips_call_may_need_jalx_p): New function.
22889         (mips_function_ok_for_sibcall): Add microMIPS support.
22890         (mips_print_operand_punctuation): Support short delay slots and
22891         compact jumps.
22892         (umips_swm_mask, umips_swm_encoding): New.
22893         (umips_build_save_restore): New function.
22894         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
22895         (was_mips16_p): Remove.
22896         (old_compression_mode): New.
22897         (mips_set_compression_mode): New function.
22898         (mips_set_current_function): Add microMIPS support.
22899         (mips_option_override): Likewise.
22900         (umips_save_restore_pattern_p): New function.
22901         (umips_output_save_restore): New function.
22902         (umips_load_store_pair_p_1): New function.
22903         (umips_load_store_pair_p): New function.
22904         (umips_output_load_store_pair_1): New function.
22905         (umips_output_load_store_pair): New function.
22906         (umips_movep_target_p) New function.
22907         (mips_prepare_pch_save): Add microMIPS support.
22908         * config/mips/mips.h (TARGET_COMPRESSION): New.
22909         (TARGET_CPU_CPP_BUILTINS): Update macro
22910         to use new compression flags and to support microMIPS.
22911         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
22912         (MIPS_ARCH_FLOAT_SPEC): Likewise.
22913         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
22914         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
22915         (ASM_SPEC): Support mmicromips and mno-micromips.
22916         (M16STORE_REG_P): New macro.
22917         (MIPS_CALL): Support TARGET_MICROMIPS.
22918         (MICROMIPS_J): New macro.
22919         (mips_base_mips16): Rename this...
22920         (mips_base_compression_flags): ...to this.
22921         (UMIPS_12BIT_OFFSET_P): New macro.
22922         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
22923         (MULTILIB_DIRNAMES): Likewise.
22924 2013-03-20  Richard Biener  <rguenther@suse.de>
22926         PR tree-optimization/56661
22927         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
22928         the result does not have to be distinct.
22930 2013-03-20  Richard Biener  <rguenther@suse.de>
22932         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
22933         remap_gimple_op_r.
22935 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
22936             Steven Bosscher  <steven@gcc.gnu.org>
22938         PR rtl-optimization/56605
22939         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
22941 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
22943         PR bootstrap/56656
22944         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
22945         that require movd instead of movq.
22947 2013-03-20  Richard Biener  <rguenther@suse.de>
22949         * tree-ssa-structalias.c (struct variable_info): Add pointer
22950         to the first field of an aggregate with sub-vars.  Make
22951         this and the pointer to the next subfield its ID.
22952         (vi_next): New function.
22953         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
22954         storedanything_id, integer_id): Increment by one.
22955         (new_var_info, get_call_vi, lookup_call_clobber_vi,
22956         get_call_clobber_vi): Adjust.
22957         (solution_set_expand): Simplify and speedup.
22958         (solution_set_add): Inline into ...
22959         (set_union_with_increment): ... this.  Adjust accordingly.
22960         (do_sd_constraint): Likewise.
22961         (do_ds_constraint): Likewise.
22962         (do_complex_constraint): Simplify.
22963         (build_pred_graph): Adjust.
22964         (solve_graph): Likewise.  Simplify and speedup.
22965         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
22966         get_constraint_for_component_ref, get_constraint_for_1,
22967         first_vi_for_offset, first_or_preceding_vi_for_offset,
22968         create_function_info_for, create_variable_info_for_1,
22969         create_variable_info_for, intra_create_variable_infos): Adjust.
22970         (init_base_vars): Push NULL for ID zero.
22971         (compute_points_to_sets): Adjust.
22973 2013-03-20  Richard Biener  <rguenther@suse.de>
22975         * cfgloop.c (verify_loop_structure): Streamline and avoid
22976         ICEing on corrupt loop tree.
22977         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
22978         loop tree.
22980 2013-03-20  Richard Biener  <rguenther@suse.de>
22982         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
22983         check whether an SSA update is needed.
22985 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
22987         * config/mips/constraints.md (T): Rename to...
22988         (Yf): ...this.
22989         (U): Rename to...
22990         (Yd): ...this.
22991         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
22992         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
22994 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
22996         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
22997         (*subsi3_carryin_uxtw): Likewise.
22999 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23001         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
23002         (*rorsi3_insn_uxtw): Likewise.
23004 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
23006         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
23007         (*extrsi5_insn_uxtw): Likewise.
23009 2013-03-19  Richard Biener  <rguenther@suse.de>
23011         PR tree-optimization/56273
23012         * passes.c (init_optimization_passes): Move second VRP after DOM.
23014 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
23016         * config/i386/i386.md (*movti_internal): Merge from
23017         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
23018         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
23019         nox64 isa attributes.
23021 2013-03-18  Richard Biener  <rguenther@suse.de>
23023         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
23024         (unite): Likewise.
23025         (merge_node_constraints): Likewise.
23026         (build_succ_graph): Likewise.
23027         (valid_graph_edge): Inline into single caller.
23028         (unify_nodes): Likewise.  Use bitmap_set_bit return value
23029         and cache varinfo.
23030         (scc_visit): Fix formatting and variable use.
23031         (do_sd_constraint): Use gcc_checking_assert.
23032         (do_ds_constraint): Likewise.
23033         (do_complex_constraint): Likewise.
23034         (condense_visit): Likewise.  Cleanup.
23035         (dump_pred_graph): New function.
23036         (perform_var_substitution): Dump the pred-graph before
23037         variable substitution.
23038         (find_equivalent_node): Use gcc_checking_assert.
23039         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
23041 2013-03-18  Richard Biener  <rguenther@suse.de>
23043         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
23044         Remove cond_expr_stmt_list argument and do not gimplify the
23045         built expression.
23046         (vect_loop_versioning): Adjust.
23047         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
23048         Cleanup to use less temporaries.
23049         (vect_create_data_ref_ptr): Cleanup.
23051 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23053         PR tree-optimization/56635
23054         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
23055         require types_compatible_p types.
23057 2013-03-18  Nick Clifton  <nickc@redhat.com>
23059         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
23060         spurious backslash.
23062         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
23063         Add missing line to comment describing function.
23065 2013-03-18  Richard Biener  <rguenther@suse.de>
23067         PR tree-optimization/56210
23068         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
23069         Handle string / character search functions.
23070         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
23072 2013-03-18  Richard Biener  <rguenther@suse.de>
23074         PR middle-end/56483
23075         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
23076         and implement properly.
23077         * gimple.h (gimple_cond_single_var_p): Remove.
23079 2013-03-18  Richard Biener  <rguenther@suse.de>
23081         * tree-data-ref.h (find_data_references_in_loop): Declare.
23082         * tree-data-ref.c (get_references_in_stmt): Use a stack
23083         vector pre-allocated in the callers.
23084         (find_data_references_in_stmt): Adjust.
23085         (graphite_find_data_references_in_stmt): Likewise.
23086         (create_rdg_vertices): Likewise.
23087         (find_data_references_in_loop): Export.
23088         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
23089         Compute dependences here...
23090         (vect_analyze_data_refs): ...not here.  When we encounter
23091         a non-vectorizable data reference in basic-block vectorization
23092         truncate the data reference vector.  Do not bother to
23093         fixup data-dependence information for gather loads.
23094         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
23095         of data references, as reported.
23097 2013-03-18  Richard Biener  <rguenther@suse.de>
23099         PR tree-optimization/3713
23100         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
23101         has_constants and expr.
23102         (stmt_has_constants): Properly valueize SSA names when deciding
23103         whether the stmt has constants.
23105 2013-03-18  Richard Biener  <rguenther@suse.de>
23107         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
23108         whole function when there is nothing to do.
23109         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
23110         * tree-vectorizer.c (vectorize_loops): Update virtual and
23111         loop-closed SSA once.
23112         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
23114 2013-03-18  Richard Biener  <rguenther@suse.de>
23116         PR middle-end/56113
23117         * domwalk.c (bb_postorder): New global static.
23118         (cmp_bb_postorder): New function.
23119         (walk_dominator_tree): Replace scheme imposing an order for
23120         visiting dominator sons by one sorting them at the time they
23121         are pushed on the stack.
23123 2013-03-18  Richard Biener  <rguenther@suse.de>
23125         PR tree-optimization/39326
23126         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
23127         (struct mem_ref): Replace mem member with ao_ref typed member.
23128         (MEM_ANALYZABLE): Adjust.
23129         (memref_eq): Likewise.
23130         (mem_ref_alloc): Likewise.
23131         (gather_mem_refs_stmt): Likewise.
23132         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
23133         (execute_sm_if_changed_flag_set): Adjust.
23134         (execute_sm): Likewise.
23135         (ref_always_accessed_p): Likewise.
23136         (refs_independent_p): Likewise.
23137         (can_sm_ref_p): Likewise.
23139 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
23141         PR c/56566
23142         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
23143         return 1 even for !unsignedp.
23145 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23147         * config/i386/i386.md (isa): Add x64 and nox64.
23148         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
23149         (*pushtf): Enable *roF alternative for x64 isa only.
23150         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
23151         mode attribute of integer alternatives to DImode for TARGET_64BIT.
23152         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
23153         (*movtf_internal): Merge from *movtf_internal_rex64 and
23154         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
23155         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
23156         nox64 isa attributes.
23157         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
23158         nox64 isa attributes.
23159         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
23161 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
23163         * config/alpha/alpha.c (TARGET_LRA_P): New define.
23165 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
23167         PR target/56640
23168         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
23169         class names.  Remove trailing comma after "ALL_REGS".
23171 2013-03-16  Jan Hubicka  <jh@suse.cz>
23173         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
23174         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
23175         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
23176         of cgraph_get_create_node.
23177         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
23179 2013-03-16  Jason Merrill  <jason@redhat.com>
23181         PR debug/49090
23182         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
23183         with DW_AT_default_value.
23185 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
23187         * BASE-VER: Set to 4.9.0.
23189 2013-03-14  Andi Kleen  <ak@linux.intel.com>
23191         PR target/56619
23192         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
23193         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
23194         Document _x* TSX intrinsics.
23196 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
23197             David Holsgrove  <david.holsgrove@xilinx.com>
23199         * configure.ac: Add MicroBlaze TLS support detection.
23200         * configure: Regenerate.
23201         * config/microblaze/microblaze-protos.h
23202         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
23203         symbol_mentioned_p, label_mentioned_p): Add prototypes.
23204         * config/microblaze/microblaze.c (microblaze_address_type): Add
23205         ADDRESS_TLS and tls_reloc address types.
23206         (microblaze_address_info): Add tls_reloc.
23207         (TARGET_HAVE_TLS): Define.
23208         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
23209          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
23210          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
23211         load_tls_operand,  microblaze_call_tls_get_addr,
23212         microblaze_legitimize_tls_address): New functions.
23213         (microblaze_classify_unspec): Handle UNSPEC_TLS.
23214         (get_base_reg): Use microblaze_tls_symbol_p.
23215         (microblaze_classify_address): Handle TLS.
23216         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
23217         label_mentioned_p and microblaze_tls_referenced_p.
23218         (microblaze_legitimize_address): Handle TLS.
23219         (microblaze_address_insns): Handle ADDRESS_TLS.
23220         (pic_address_needs_scratch): Handle TLS.
23221         (print_operand_address): Handle TLS.
23222         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
23223         (microblaze_expand_move): Handle TLS.
23224         (microblaze_legitimate_constant_p): Check
23225         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
23226         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
23227         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
23228         (PIC_OFFSET_TABLE_REGNUM): Set.
23229         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
23230         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
23231         (addsi3, movsi_internal2, movdf_internal): Update constraints
23232         * config/microblaze/predicates.md (arith_plus_operand): Define
23233         (move_operand): Redefine as move_src_operand,
23234         check microblaze_tls_referenced_p.
23236 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23238         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
23239         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
23241 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
23243         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
23244         CC mode for AND.
23246 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
23248         PR tree-optimization/53265
23249         * common.opt (Waggressive-loop-optimizations): New option.
23250         * tree-ssa-loop-niter.c: Include tree-pass.h.
23251         (do_warn_aggressive_loop_optimizations): New function.
23252         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
23253         if number_of_latch_executions returned constant.
23254         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
23255         early.  If number_of_latch_executions returned constant, set
23256         nb_iterations_upper_bound back to it.
23257         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
23258         field.
23259         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
23260         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
23262         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
23263         (MULTILIB_OSDIRNAMES): Set.
23264         * genmultilib: If defaultosdirname doesn't start with :: , set
23265         defaultosdirname2 instead, clear it and emit two . multilib_raw
23266         entries instead of just one.
23268 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
23270         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23271         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
23272         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
23273         (SUBTARGET_OVERRIDE_OPTIONS): New.
23275 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
23277         PR target/49880
23278         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
23279         (musermode): Convert to Var(TARGET_USERMODE).
23280         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
23281         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
23282         * config/sh/sh.c (sh_option_override): Use
23283         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
23284         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
23285         condition.
23286         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
23287         TARGET_SH4.
23288         (udivsi3_i4_single, divsi3_i4_single): Use
23289         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
23291 2013-03-13  Dave Korn  <dave.korn.cygwin@....>
23293         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
23294         default setting.
23296 2013-03-13  Richard Biener  <rguenther@suse.de>
23298         PR tree-optimization/56608
23299         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
23300         calls when vectorizing basic-blocks.
23302 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
23304         PR plugins/45078
23305         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
23306         tm_file.
23308 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
23310         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
23312 2013-03-11  Jan Hubicka  <jh@suse.cz>
23314         PR lto/56557
23315         * lto-streamer-out.c (output_symbol_p): Skip references from
23316         constructors of external variables.
23318 2013-03-11  Jan Hubicka  <jh@suse.cz>
23320         PR middle-end/56571
23321         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
23322         from pseudos.
23323         * emit-rtl.c (verify_rtx_sharing): Likewise.
23324         (copy_insn_1): Likewise.
23325         * rtl.c (copy_rtx): Likewise.
23327 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
23329         PR target/56591
23330         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
23331         output_operand_lossage message.
23333 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
23335         PR target/56470
23336         * arm.c (shift_op): Validate RTL pattern on the fly.
23337         (arm_print_operand, case 'S'): Don't use shift_operator to validate
23338         the RTL.
23340 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23342         PR target/56347
23343         * config/pa/pa.md (call_value): Check for calls to powf and direct to
23344         new call patterns that clobber %fr12.
23345         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
23346         split and postreload patterns.
23347         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
23348         registers %fr12 and %fr12R as call used.
23350 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
23352         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
23353         (canon_address, record_store, replace_read, check_mem_read_rtx,
23354         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
23355         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
23356         rest_of_handle_dse): Likewise.
23358 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
23360         PR middle-end/56524
23361         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
23362         Add base_optabs.
23363         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
23364         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
23365         (save_optabs_if_changed): Replace with...
23366         (init_tree_optimization_optabs): ...this.
23367         * optabs.c (save_optabs_if_changed): Rename to...
23368         (init_tree_optimization_optabs): ...this.  Take the optimization node
23369         as argument.  Do nothing if the base optabs are already correct.
23370         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
23371         to recompute optabs.
23372         * function.h (function): Remove optabs field.
23373         * function.c (invoke_set_current_function_hook): Call
23374         init_tree_optimization_optabs.  Use the result to initialize
23375         this_fn_optabs.
23377 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
23379         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
23380         if GTMA_HAS_NO_INSTRUMENTATION.
23381         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
23382         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
23383         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
23384         * gimple-pretty-print.c (dump_gimple_transaction): Handle
23385         GTMA_HAS_NO_INSTRUMENTATION.
23387 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
23389         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
23390         libasan_preinit.o.
23392 2013-03-08  Marek Polacek  <polacek@redhat.com>
23393             Jakub Jelinek  <jakub@redhat.com>
23395         PR tree-optimization/56478
23396         * predict.c (is_comparison_with_loop_invariant_p): Change the
23397         type of loop_step to tree.
23398         (predict_loops): Adjust.
23399         (predict_iv_comparison): Perform the computations on double_ints.
23401 2013-03-08  Richard Biener  <rguenther@suse.de>
23403         PR tree-optimization/56570
23404         * tree-cfg.c (verify_expr_location_1): Verify locations for
23405         DECL_DEBUG_EXPR.
23406         * tree-sra.c (create_access_replacement): Strip locations
23407         from DECL_DEBUG_EXPRs.
23409 2013-03-08  Richard Biener  <rguenther@suse.de>
23411         * tree-inline.c (expand_call_inline): Do not associate
23412         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
23413         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
23415 2013-03-08  Richard Biener  <rguenther@suse.de>
23417         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
23418         or block changes with -Og.  Fix for location / block encoding
23419         changes and PHI arguments with locations.
23421 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
23423         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
23424         for all counters.
23425         (struct output_info): Likewise.
23426         (register_overhead): Remove bad gcc_assert.
23427         (bitmap_find_bit): If there is only a single bitmap element, do not
23428         count a miss as a search.
23429         (print_statistics): Update for counter type changes.
23430         (dump_bitmap_statistics): Likewise.  Print headers such that they
23431         are properly lined up with the printed counters.
23433 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
23435         PR tree-optimization/56559
23436         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
23437         check that it has only a single use.
23439 2013-03-07  Richard Biener  <rguenther@suse.de>
23441         * doc/invoke.texi (fwhole-program): Discourage use in combination
23442         with -flto.
23444 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
23446         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
23448         PR tree-optimization/56539
23449         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
23450         instead of GSI_CONTINUE_LINKING as last argument to
23451         force_gimple_operand_gsi.  Adjust function comment.
23453         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
23454         aarch64-cores.def.
23456         PR middle-end/56548
23457         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
23458         promoted mode, convert the result back to the original mode.
23460 2013-03-06  Richard Biener  <rguenther@suse.de>
23462         PR middle-end/56294
23463         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
23464         (insert_updated_phi_nodes_compare_uids): New function.
23465         (update_ssa): Sort symbols_to_rename after UID before
23466         traversing it to insert PHI nodes.
23468 2013-03-06  Richard Biener  <rguenther@suse.de>
23470         PR middle-end/50494
23471         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
23472         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
23474         Revert
23475         2013-02-13  Richard Biener  <rguenther@suse.de>
23477         PR lto/50494
23478         * varasm.c (output_constant_def_1): Get the decl representing
23479         the constant as argument.
23480         (output_constant_def): Wrap output_constant_def_1.
23481         (make_decl_rtl): Use output_constant_def_1 with the decl
23482         representing the constant.
23483         (build_constant_desc): Optionally re-use a decl already
23484         representing the constant.
23485         (tree_output_constant_def): Adjust.
23487 2013-03-06  Joey Ye  <joey.ye@arm.com>
23489         PR lto/50293
23490         * gcc.c (convert_white_space): New function.
23491         (main): Handles white space in function name.
23493 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
23495         PR target/56529
23496         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
23497         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
23498         to SH_DIV_CALL_TABLE for TARGET_SH2.
23499         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
23500         list.
23501         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
23502         call-table options.
23504 2013-03-05  Sterling Augustine  <saugustine@google.com>
23505             Cary Coutant  <ccoutant@google.com>
23507         PR debug/55364
23508         * dwarf2out.c (resolve_addr): Don't call
23509         remove_loc_list_addr_table_entries a second time for the same
23510         expression.
23512 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23514         PR debug/56510
23515         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
23516         (avoid_complex_debug_insns): New function.
23517         (expand_debug_locations): Call it.
23519         PR rtl-optimization/56484
23520         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
23521         lifetimes of hard registers on small register class machines.
23523 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23525         * config/microblaze/microblaze-protos.h: Rename
23526         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
23527         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
23528         fast_interrupt.
23529         (microblaze_fast_interrupt_function_p): New function.
23530         (microblaze_is_interrupt_handler): Rename to
23531         microblaze_is_interrupt_variant and add fast_interrupt check.
23532         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
23533         (save_restore_insns): Likewise.
23534         (compute_frame_size): Likewise.
23535         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
23536         (microblaze_globalize_label): Likewise.
23537         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
23538         * config/microblaze/microblaze.md: Use wrapper
23539         microblaze_is_interrupt_variant.
23541 2013-03-05  Kai Tietz  <ktietz@redhat.com>
23543         * sdbout.c (sdbout_one_type): Switch to current function's section
23544         supporting cold/hot.
23546 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
23548         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
23549         -mxl-reorder.
23551 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23553         PR middle-end/56461
23554         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
23555         if VALGRIND_GET_VBITS is defined, temporarily make object
23556         memory all defined, and restore previous valgrind addressability
23557         and definability afterwards.  Free this_object at the end.
23559         PR middle-end/56461
23560         * lra.c (lra): Call lra_clear_live_ranges if live_p,
23561         right before calling lra_create_live_ranges, also call it
23562         when clearing live_p.  Only call lra_clear_live_ranges
23563         at the end if live_p.
23565         PR middle-end/56461
23566         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
23568 2013-03-05  Richard Biener  <rguenther@suse.de>
23570         PR tree-optimization/56521
23571         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
23572         value-id.
23574 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
23576         PR c++/55135
23577         * except.h (remove_unreachable_eh_regions): New prototype.
23578         * except.c (remove_eh_handler_splicer): New function, split out
23579         of remove_eh_handler.
23580         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
23581         warning about running it on many EH regions one at a time.
23582         (remove_unreachable_eh_regions_worker): New function, walk the
23583         EH tree in depth-first order and remove non-marked regions.
23584         (remove_unreachable_eh_regions): New function.
23585         * tree-eh.c (mark_reachable_handlers): New function, split out
23586         from remove_unreachable_handlers.
23587         (remove_unreachable_handlers): Use mark_reachable_handlers and
23588         remove_unreachable_eh_regions.
23589         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
23590         and remove_unreachable_eh_regions.
23592 2013-03-05  Richard Biener  <rguenther@suse.de>
23594         PR middle-end/56525
23595         * loop-init.c (fix_loop_structure): Remove loops in two stages,
23596         not freeing them until the end.
23598 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23600         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
23602 2013-03-05  Richard Biener  <rguenther@suse.de>
23604         PR tree-optimization/56270
23605         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
23606         of loads after scheduling an SLP instance.
23608 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23610         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
23611         tic6x.exp.
23612         (check_gcc_parallelize): Run guality.exp as a separate job from
23613         vect.exp with unsorted.exp and $(dg_target_exps) separately from
23614         struct-layout-1.exp with stackalign.exp.
23616         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
23618         PR middle-end/56461
23619         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
23620         load_index sbitmap even if some bit in it isn't set.
23622         PR middle-end/56461
23623         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
23624         (discover_iteration_bound_by_body_walk): Change queues to
23625         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
23626         spelling in comment.  Call safe_push on queues[bound_index] directly.
23627         Release queues[queue_index] in every iteration unconditionally.
23628         Release bounds vector.
23630         PR middle-end/56461
23631         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
23632         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
23633         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
23634         inner_phis vector.
23636 2013-03-05  Richard Biener  <rguenther@suse.de>
23638         PR lto/56515
23639         * tree-inline.c (remap_blocks_to_null): New function.
23640         (expand_call_inline): When expanding a call stmt without
23641         an associated block inline remap all callee blocks to NULL.
23643 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
23645         PR rtl-optimization/56494
23646         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
23647         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
23648         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
23650         PR middle-end/56461
23651         * sel-sched-ir.c (free_sched_pools): Release
23652         succs_info_pool.stack[succs_info_pool.max_top] vectors too
23653         if succs_info_pool.max_top isn't -1.
23655         PR bootstrap/56509
23656         * opts.c (opts_obstack, opts_concat): Moved to...
23657         * opts-common.c (opts_obstack, opts_concat): ... here.
23659 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23661         PR middle-end/56461
23662         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
23664 2013-03-04  Martin Jambor  <mjambor@suse.cz>
23666         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
23667         all appropriate places.
23669 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
23671         PR tree-optimization/56424
23672         * ipa-split.c (split_function): Do not set the RSO flag if result is
23673         not by reference and its type is a register type.
23675 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
23677         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
23678         (microblaze_legitimate_pic_operand): Likewise
23679         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
23680         new function microblaze_legitimate_pic_operand
23681         * config/microblaze/microblaze-protos.h
23682         (microblaze_legitimate_pic_operand): Declare.
23684 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23686         * config/microblaze/predicates.md (call_insn_simple_operand):
23687         New predicate for supported rtx code types.
23688         * config/microblaze/microblaze.md (call_internal1): Use
23689         call_insn_simple_operand predicate.
23691 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
23693         PR middle-end/56461
23694         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
23695         partitions.ordered_remove.
23697         PR middle-end/56461
23698         * tree-vect-stmts.c (vectorizable_conversion): Don't call
23699         vec_oprnds0.create (1) for modifier == NONE.
23701         PR middle-end/56461
23702         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
23703         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
23704         vec_oprnds1 right before pushing anything to it for
23705         scalar_shift_arg.
23707         PR middle-end/56461
23708         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
23709         set nbbs to 0 instead of having separate code path.
23710         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
23711         instead of false as last argument if returning NULL.
23713 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
23715         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
23716         the attribute is now called "target" instead of "option".
23717         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
23718         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
23719         attribute/pragma name for TARGET_OPTION_VALID_P and
23720         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
23721         * doc/tm.texi: Regenerated.
23723 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
23725         * config/microblaze/microblaze.c:
23726         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
23727         * config/microblaze/microblaze.h: Add -mxl-reorder to
23728         DRIVER_SELF_SPECS.
23729         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
23730         instructions emitted if TARGET_REORDER.
23731         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
23732         or 0 for -m/-mno case, but initialises as 2 to detect default use case
23733         separately.
23735 2013-03-01  Xinliang David Li  <davidxl@google.com>
23737         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
23738         walk length.
23740 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
23742         PR middle-end/56461
23743         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
23744         vector even when returning true.  Fix up function comment formatting.
23746         PR middle-end/56461
23747         * ira-build.c (ira_loop_nodes_count): New variable.
23748         (create_loop_tree_nodes): Initialize it.
23749         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
23751         PR middle-end/56461
23752         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
23753         method on dr_chain and result_chain.
23754         * tree-vect-stmts.c (vectorizable_store): Only call
23755         result_chain.create if j == 0.
23757         PR middle-end/56461
23758         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
23759         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
23760         before overwriting it.
23762 2013-03-01  Tobias Burnus  <burnus@net-b.de>
23764         * doc/extended.texi (C Extensions): Change order in @menu
23765         to match @node.
23766         (Other MIPS Built-in Functions): Move last MIPS entry before
23767         "picoChip Built-in Functions".
23768         (SH Built-in Functions): Move after RX Built-in Functions.
23769         * doc/gcc.texi (Introduction): Change order in @menu
23770         to match @node.
23771         * doc/md.texi (Constraints): Ditto.
23772         * gty.texi (Type Information): Ditto.
23773         (User-provided marking routines for template types): Make
23774         subsection.
23775         * doc/invoke.texi (AArch64 Options): Move before
23776         "Adapteva Epiphany Options".
23778 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
23779             Jakub Jelinek  <jakub@redhat.com>
23781         PR sanitizer/56454
23782         * asan.c (gate_asan): Lookup no_sanitize_address instead of
23783         no_address_safety_analysis attribute.
23784         * doc/extend.texi (no_address_safety_attribute): Rename to
23785         no_sanitize_address attribute, mention no_address_safety_analysis
23786         attribute as deprecated alias.
23788 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
23790         PR middle-end/56461
23791         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
23792         type to vec<vec<tree> > *.
23793         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
23794         to be vec<tree> instead of vec<tree> *, set vec_defs
23795         to vNULL and call vec_defs.create (number_of_vects), adjust other
23796         uses of vec_defs.
23797         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
23798         vectorizable_condition): Adjust vect_get_slp_defs callers.
23800 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23802         * config/aarch64/aarch64.c
23803         (aarch64_float_const_representable): Remove unused variable.
23805 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23807         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
23809 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23811         * config/aarch64/aarch64-builtins.c
23812         (aarch64_init_simd_builtins): Make static.
23814 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
23816         * config/aarch64/aarch64.c
23817         (aarch64_simd_make_constant): Make static.
23819 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23821         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
23822         with no initialization to the RHS of debug statements.
23824 2013-02-28  Martin Jambor  <mjambor@suse.cz>
23826         PR tree-optimization/56294
23827         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
23828         Adjust dumping.
23829         (get_access_replacement): Do not call create_access_replacement.
23830         Assert a replacement exists.
23831         (get_repl_default_def_ssa_name): Create the replacement declaration
23832         itself.
23834 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23836         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
23837         final_end_function.
23839 2013-02-28  Marek Polacek  <polacek@redhat.com>
23841         PR rtl-optimization/56466
23842         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
23843         if we're changing a loop.
23844         (peel_loops_completely): Likewise.
23846 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
23848         PR c++/55813
23849         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
23851 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23853         PR target/56445
23854         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
23855         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
23856         INTX_FTYPE_FX, FX_FTYPE_INTX.
23857         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
23859 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23861         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
23862         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
23863         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
23864         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
23865         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
23866         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
23867         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
23868         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
23869         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
23870         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
23871         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
23872         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
23873         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
23874         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
23875         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
23876         (avrxmega6): Increase max flash segments from 5 to 6.
23877         * config/avr/t-multilib: Regenerate.
23878         * config/avr/avr-tables.opt: Regenerate.
23879         * doc/avr-mmcu.texi: Regenerate.
23881 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
23883         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
23884         (avr_device_to_arch): Rename to avr_device_to_ld.
23885         (avr_device_to_as): New prototype.
23886         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
23887         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
23888         * config/avr/driver-avr.c (avr_device_to_as): New.
23889         (avr_device_to_arch): Rename to avr_device_to_ld.
23891 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23893         PR middle-end/56461
23894         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
23895         method on dr_chain and result_chain.
23897         PR middle-end/56461
23898         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
23899         pointer_set_destroy on not_executed_last_iteration.
23901         PR middle-end/56461
23902         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
23904         PR middle-end/56461
23905         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
23906         FOR_EACH_DEFINED_FUNCTION when freeing state.
23908         PR middle-end/56461
23909         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
23910         pool_free.
23911         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
23912         overwriting it.
23914         PR middle-end/56461
23915         * ipa-cp.c (decide_whether_version_node): Call vec_free on
23916         known_aggs[i].items and release known_aggs vector.
23918         PR middle-end/56461
23919         * ipa-reference.c (propagate): Free node_info even for alias nodes.
23921 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
23923         * config/microblaze/microblaze.c (microblaze_emit_compare):
23924         Use xor for EQ/NE comparisions.
23925         * config/microblaze/microblaze.md (cstoresf4): Add constraints
23926         (cbranchsf4): Adjust operator to comparison_operator.
23928 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23930         PR middle-end/56461
23931         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
23932         vector.
23933         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
23934         vec_safe_push, always update *slot.
23935         (redirect_edge_var_map_clear): Use vec_free.
23936         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
23937         (free_var_map_entry): Use vec_free.
23938         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
23939         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
23941 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
23943         PR middle-end/45472
23944         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
23945         when the may_trap_p bit of the exprs being merged differs.
23946         Reorder tests for speculativeness in the logical and operator.
23948 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
23950         * incpath.c (add_standard_paths): Use reconcat instead of concat
23951         where appropriate and avoid leaking memory.
23953         * opts.h: Include obstack.h.
23954         (opts_concat): New prototype.
23955         (opts_obstack): New declaration.
23956         * opts.c (opts_concat): New function.
23957         (opts_obstack): New variable.
23958         (init_options_struct): Call gcc_init_obstack on opts_obstack.
23959         (finish_options): Use opts_concat instead of concat
23960         and XOBNEWVEC instead of XNEWVEC.
23961         * opts-common.c (generate_canonical_option, decode_cmdline_option,
23962         generate_option): Likewise.
23963         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
23964         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
23966         PR target/56455
23967         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
23968         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
23970 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
23972         PR middle-end/56461
23973         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
23975 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23977         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
23978         (arm_block_move_unaligned_straight): Likewise.
23979         (arm_adjust_block_mem): Likewise.
23981 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
23983         PR target/48901
23984         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
23985         temp, cond and label.
23986         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
23988         PR target/52500
23989         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
23990         * config/c6x/c6x.h (dbx_register_map): Update declaration.
23992         PR target/52501
23993         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
23994         of prologue/epilogue functions.
23996         PR target/52550
23997         * config/tilegx/tilegx.c (tilegx_expand_prologue):
23998         Remove unused variable cfa_offset.
23999         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
24001         PR target/54639
24002         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
24003         type promotion to unsigned.
24005         PR target/54640
24006         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
24007         for HOST_WIDE_INT of 32 bit / same size as int.
24008         (arm_block_move_unaligned_straight): Likewise.
24009         (arm_adjust_block_mem): Likewise.
24011         PR target/54662
24012         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
24013         ALL_CFLAGS.
24015 2013-02-26  Marek Polacek  <polacek@redhat.com>
24017         PR tree-optimization/56426
24018         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
24020 2013-02-26  Richard Biener  <rguenther@suse.de>
24022         PR target/56444
24023         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
24024         unused variable loops.
24026 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
24028         PR tree-optimization/56448
24029         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
24030         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
24031         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
24032         later operands of the references, or even first operand for
24033         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
24035         PR tree-optimization/56443
24036         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
24037         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
24038         to type_for_mode langhook.
24040 2013-02-25  Matt Turner  <mattst88@gmail.com>
24042         * doc/invoke.texi: Document r4700.
24044 2013-02-25  Richard Biener  <rguenther@suse.de>
24046         PR tree-optimization/56175
24047         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
24048         split out from ...
24049         (simplify_bitwise_binary): ... here.  Also guard the conversion
24050         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
24052 2013-02-25  Catherine Moore  <clm@codesourcery.com>
24054         Revert:
24055         2013-02-24  Catherine Moore  <clm@codesourcery.com>
24056             Maciej W. Rozycki  <macro@codesourcery.com>
24057             Tom de Vries  <tom@codesourcery.com>
24058             Nathan Sidwell  <nathan@codesourcery.com>
24059             Iain Sandoe  <iain@codesourcery.com>
24060             Nathan Froyd  <froydnj@codesourcery.com>
24061             Chao-ying Fu  <fu@mips.com>
24063         * doc/extend.texi: (micromips, nomicromips, nocompression):
24064         Document new function attributes.
24065         * doc/invoke.texi (minterlink-compressed, mmicromips,
24066         m14k, m14ke, m14kec): Document new options.
24067         (minterlink-mips16): Update documentation.
24068         * doc/md.texi (ZC, ZD): Document new constraints.
24069         * configure.ac (gcc_cv_as_micromips): Check if linker
24070         supports the .set micromips directive.
24071         * configure: Regenerate.
24072         * config.in: Regenerate.
24073         * config/mips/mips-tables.opt: Regenerate.
24074         * config/mips/micromips.md: New file.
24075         * constraints.md (ZC, AD): New constraints.
24076         * config/mips/predicates.md (movep_src_register): New predicate.
24077         (movep_src_operand): New predicate.
24078         (non_volatile_mem_operand): New predicate.
24079         * config/mips/mips.md (multimem): New type.
24080         (length): Differentiate between 17-bit and 18-bit branch offsets.
24081         (MOVEP1, MOVEP2): New mode iterator.
24082         (mov_<load>l): Use ZC constraint.
24083         (mov_<load>r): Likewise.
24084         (mov_<store>l): Likewise.
24085         (mov_<store>r): Likewise.
24086         (*branch_equality<mode>_inverted): Add microMIPS support.
24087         (*branch_equality<mode>): Likewise.
24088         (*jump_absolute): Likewise.
24089         (indirect_jump_<mode>): Likewise.
24090         (tablejump_<mode>): Likewise.
24091         (<optab>_internal): Likewise.
24092         (sibcall_internal): Likewise.
24093         (sibcall_value_internal): Likewise.
24094         (prefetch): Use constraint ZD.
24095         * config/mips/mips.opt (minterlink-compressed): New option.
24096         (minterlink-mips16): Now an alias for minterlink-compressed.
24097         (mmicromips): New option.
24098         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24099         (compare_and_swap_12): Likewise.
24100         (sync_add<mode>): Likewise.
24101         (sync_<optab>_12): Likewise.
24102         (sync_old_<optab>_12): Likewise.
24103         (sync_new_<optab>_12): Likewise.
24104         (sync_nand_12): Likewise.
24105         (sync_old_nand_12): Likewise.
24106         (sync_new_nand_12): Likewise.
24107         (sync_sub<mode>): Likewise.
24108         (sync_old_add<mode>): Likewise.
24109         (sync_old_sub<mode>): Likewise.
24110         (sync_new_add<mode>): Likewise.
24111         (sync_new_sub<mode>): Likewise.
24112         (sync_<optab><mode>): Likewise.
24113         (sync_old_<optab><mode>): Likewise.
24114         (sync_new_<optab><mode>): Likewise.
24115         (sync_nand<mode>): Likewise.
24116         (sync_old_nand<mode>): Likewise.
24117         (sync_new_nand<mode>): Likewise.
24118         (sync_lock_test_and_set<mode>): Likewise.
24119         (test_and_set_12): Likewise.
24120         (atomic_compare_and_swap<mode>): Likewise.
24121         (atomic_exchange<mode>_llsc): Likewise.
24122         (atomic_fetch_add<mode>_llsc): Likewise.
24123         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24124         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24125         (umips_save_restore_pattern_p): Likewise.
24126         (umips_load_store_pair_p): Likewise.
24127         (umips_output_load_store_pair): Likewise.
24128         (umips_movep_target_p): Likewise.
24129         (umips_12bit_offset_address_p): Likewise.
24130         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24131         (mips_base_mips16): Rename this...
24132         (mips_base_compression_flags): ...to this. Update all uses.
24133         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24134         (mips_mips16_decl_p): Delete.
24135         (mips_nomips16_decl_p): Delete.
24136         (mips_get_compress_on_flags): New function.
24137         (mips_get_compress_off_flags): New function.
24138         (mips_get_compress_mode): New function.
24139         (mips_get_compress_on_name): New function.
24140         (mips_get_compress_off_name): New function.
24141         (mips_insert_attributes): Support multiple compression types.
24142         (mips_merge_decl_attributes): Likewise.
24143         (umips_12bit_offset_address_p): New function.
24144         (mips_start_function_definition): Emit .set micromips directive.
24145         (mips_call_may_need_jalx_p): New function.
24146         (mips_function_ok_for_sibcall): Add microMIPS support.
24147         (mips_print_operand_punctuation): Support short delay slots and
24148         compact jumps.
24149         (umips_swm_mask, umips_swm_encoding): New.
24150         (umips_build_save_restore): New function.
24151         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24152         (was_mips16_p): Remove.
24153         (old_compression_mode): New.
24154         (mips_set_compression_mode): New function.
24155         (mips_set_current_function): Add microMIPS support.
24156         (mips_option_override): Likewise.
24157         (umips_save_restore_pattern_p): New function.
24158         (umips_output_save_restore): New function.
24159         (umips_load_store_pair_p_1): New function.
24160         (umips_load_store_pair_p): New function.
24161         (umips_output_load_store_pair_1): New function.
24162         (umips_output_load_store_pair): New function.
24163         (umips_movep_target_p) New function.
24164         (mips_prepare_pch_save): Add microMIPS support.
24165         * config/mips/mips.h (TARGET_COMPRESSION): New.
24166         (TARGET_CPU_CPP_BUILTINS): Update macro
24167         to use new compression flags and to support microMIPS.
24168         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24169         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24170         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24171         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24172         (ASM_SPEC): Support mmicromips and mno-micromips.
24173         (M16STORE_REG_P): New macro.
24174         (MIPS_CALL): Support TARGET_MICROMIPS.
24175         (MICROMIPS_J): New macro.
24176         (mips_base_mips16): Rename this...
24177         (mips_base_compression_flags): ...to this.
24178         (UMIPS_12BIT_OFFSET_P): New macro.
24179         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24180         (MULTILIB_DIRNAMES): Likewise.
24182 2013-02-25  Tom de Vries  <tom@codesourcery.com>
24184         PR rtl-optimization/56131
24185         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
24186         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
24187         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
24189 2013-02-25  Tobias Burnus  <burnus@net-b.de>
24191         * doc/invoke.texi (-fsanitize=): Move from optimization
24192         to debugging options.
24194 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24196         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
24198 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
24199             Alexander Monakov  <amonakov@ispras.ru>
24201         PR middle-end/56077
24202         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
24203         flush pending lists also on non-jumps.  Adjust comment.
24205 2013-02-24  Catherine Moore  <clm@codesourcery.com>
24206             Maciej W. Rozycki  <macro@codesourcery.com>
24207             Tom de Vries  <tom@codesourcery.com>
24208             Nathan Sidwell  <nathan@codesourcery.com>
24209             Iain Sandoe  <iain@codesourcery.com>
24210             Nathan Froyd  <froydnj@codesourcery.com>
24211             Chao-ying Fu  <fu@mips.com>
24213         * doc/extend.texi: (micromips, nomicromips, nocompression):
24214         Document new function attributes.
24215         * doc/invoke.texi (minterlink-compressed, mmicromips,
24216         m14k, m14ke, m14kec): Document new options.
24217         (minterlink-mips16): Update documentation.
24218         * doc/md.texi (ZC, ZD): Document new constraints.
24219         * configure.ac (gcc_cv_as_micromips): Check if linker
24220         supports the .set micromips directive.
24221         * configure: Regenerate.
24222         * config.in: Regenerate.
24223         * config/mips/mips-tables.opt: Regenerate.
24224         * config/mips/micromips.md: New file.
24225         * constraints.md (ZC, AD): New constraints.
24226         * config/mips/predicates.md (movep_src_register): New predicate.
24227         (movep_src_operand): New predicate.
24228         (non_volatile_mem_operand): New predicate.
24229         * config/mips/mips.md (multimem): New type.
24230         (length): Differentiate between 17-bit and 18-bit branch offsets.
24231         (MOVEP1, MOVEP2): New mode iterator.
24232         (mov_<load>l): Use ZC constraint.
24233         (mov_<load>r): Likewise.
24234         (mov_<store>l): Likewise.
24235         (mov_<store>r): Likewise.
24236         (*branch_equality<mode>_inverted): Add microMIPS support.
24237         (*branch_equality<mode>): Likewise.
24238         (*jump_absolute): Likewise.
24239         (indirect_jump_<mode>): Likewise.
24240         (tablejump_<mode>): Likewise.
24241         (<optab>_internal): Likewise.
24242         (sibcall_internal): Likewise.
24243         (sibcall_value_internal): Likewise.
24244         (prefetch): Use constraint ZD.
24245         * config/mips/mips.opt (minterlink-compressed): New option.
24246         (minterlink-mips16): Now an alias for minterlink-compressed.
24247         (mmicromips): New option.
24248         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
24249         (compare_and_swap_12): Likewise.
24250         (sync_add<mode>): Likewise.
24251         (sync_<optab>_12): Likewise.
24252         (sync_old_<optab>_12): Likewise.
24253         (sync_new_<optab>_12): Likewise.
24254         (sync_nand_12): Likewise.
24255         (sync_old_nand_12): Likewise.
24256         (sync_new_nand_12): Likewise.
24257         (sync_sub<mode>): Likewise.
24258         (sync_old_add<mode>): Likewise.
24259         (sync_old_sub<mode>): Likewise.
24260         (sync_new_add<mode>): Likewise.
24261         (sync_new_sub<mode>): Likewise.
24262         (sync_<optab><mode>): Likewise.
24263         (sync_old_<optab><mode>): Likewise.
24264         (sync_new_<optab><mode>): Likewise.
24265         (sync_nand<mode>): Likewise.
24266         (sync_old_nand<mode>): Likewise.
24267         (sync_new_nand<mode>): Likewise.
24268         (sync_lock_test_and_set<mode>): Likewise.
24269         (test_and_set_12): Likewise.
24270         (atomic_compare_and_swap<mode>): Likewise.
24271         (atomic_exchange<mode>_llsc): Likewise.
24272         (atomic_fetch_add<mode>_llsc): Likewise.
24273         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
24274         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
24275         (umips_save_restore_pattern_p): Likewise.
24276         (umips_load_store_pair_p): Likewise.
24277         (umips_output_load_store_pair): Likewise.
24278         (umips_movep_target_p): Likewise.
24279         (umips_12bit_offset_address_p): Likewise.
24280         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
24281         (mips_base_mips16): Rename this...
24282         (mips_base_compression_flags): ...to this. Update all uses.
24283         (mips_attribute_table): Add micromips, nomicromips and nocompression.
24284         (mips_mips16_decl_p): Delete.
24285         (mips_nomips16_decl_p): Delete.
24286         (mips_get_compress_on_flags): New function.
24287         (mips_get_compress_off_flags): New function.
24288         (mips_get_compress_mode): New function.
24289         (mips_get_compress_on_name): New function.
24290         (mips_get_compress_off_name): New function.
24291         (mips_insert_attributes): Support multiple compression types.
24292         (mips_merge_decl_attributes): Likewise.
24293         (umips_12bit_offset_address_p): New function.
24294         (mips_start_function_definition): Emit .set micromips directive.
24295         (mips_call_may_need_jalx_p): New function.
24296         (mips_function_ok_for_sibcall): Add microMIPS support.
24297         (mips_print_operand_punctuation): Support short delay slots and
24298         compact jumps.
24299         (umips_swm_mask, umips_swm_encoding): New.
24300         (umips_build_save_restore): New function.
24301         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
24302         (was_mips16_p): Remove.
24303         (old_compression_mode): New.
24304         (mips_set_compression_mode): New function.
24305         (mips_set_current_function): Add microMIPS support.
24306         (mips_option_override): Likewise.
24307         (umips_save_restore_pattern_p): New function.
24308         (umips_output_save_restore): New function.
24309         (umips_load_store_pair_p_1): New function.
24310         (umips_load_store_pair_p): New function.
24311         (umips_output_load_store_pair_1): New function.
24312         (umips_output_load_store_pair): New function.
24313         (umips_movep_target_p) New function.
24314         (mips_prepare_pch_save): Add microMIPS support.
24315         * config/mips/mips.h (TARGET_COMPRESSION): New.
24316         (TARGET_CPU_CPP_BUILTINS): Update macro
24317         to use new compression flags and to support microMIPS.
24318         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
24319         (MIPS_ARCH_FLOAT_SPEC): Likewise.
24320         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
24321         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
24322         (ASM_SPEC): Support mmicromips and mno-micromips.
24323         (M16STORE_REG_P): New macro.
24324         (MIPS_CALL): Support TARGET_MICROMIPS.
24325         (MICROMIPS_J): New macro.
24326         (mips_base_mips16): Rename this...
24327         (mips_base_compression_flags): ...to this.
24328         (UMIPS_12BIT_OFFSET_P): New macro.
24329         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
24330         (MULTILIB_DIRNAMES): Likewise.
24332 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
24334         PR target/52555
24335         * target-globals.c (save_target_globals): For init_reg_sets and
24336         target_reinit remporarily set this_fn_optabs to this_target_optabs.
24338 2013-02-22  James Grennahlgh  <james.greenhalgh@arm.com>
24340         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
24341         * config/aarch64/t-aarch64
24342         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
24344 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
24346         PR inline-asm/56148
24347         * lra-constraints.c (process_alt_operands): Reload operand
24348         conflicting with earlier clobber only if no more other conflicting
24349         operands.
24351 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
24353         PR sanitizer/56393
24354         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
24355         if not linking a shared library.
24357 2013-02-22  Seth LaForge  <sethml@google.com>
24359         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
24361 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
24363         * config/arm/arm.md (split for extendsidi): Update condition.
24364         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
24365         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
24366         (qhs_zextenddi_cstr): Likewise.
24368 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24370         PR middle-end/56420
24371         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
24372         avoid signed wrapping.
24373         (expand_mult): Handle properly multiplication by
24374         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
24375         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
24376         in the compiler if coeff is HOST_WIDE_INT_MIN.
24377         (expand_divmod): Don't make ext_op1 static, change it's type to
24378         uhwi.  Avoid undefined behavior in -INTVAL (op1).
24380         PR rtl-optimization/50339
24381         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
24382         field.
24383         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
24384         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
24385         into splitting_ashiftrt field.
24386         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
24387         ASHIFTRT.
24388         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
24389         choices.
24391 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
24393         PR middle-end/56108
24394         * trans-mem.c (execute_tm_mark): Do not expand transactions that
24395         are sure to go irrevocable.
24397 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
24399         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
24400         scalars are valid operands.
24402 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24404         PR tree-optimization/56310
24405         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
24406         only matching indices and non-negative final offsets.
24407         (intersect_aggregates_with_edge): Pass src_idx to
24408         agg_replacements_to_vector.  Pass src_idx insstead of index to
24409         intersect_with_agg_replacements.
24411 2013-02-21  Martin Jambor  <mjambor@suse.cz>
24413         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
24414         instead of hard-wired defaults.
24416 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
24418         * doc/invoke.texi (MIPS Options): Update documentation of the
24419         floating-point multiply-accumulate instruction restrictions.
24421 2013-02-21  Kostya Serebryany  <kcc@google.com>
24423         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
24424         asan_shadow_offset on x86_64 linux.
24426 2013-02-21  Richard Biener  <rguenther@suse.de>
24428         PR tree-optimization/56415
24429         Revert
24430         2013-02-11  Richard Biener  <rguenther@suse.de>
24432         PR tree-optimization/56273
24433         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24434         first VRP run.
24436 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
24438         PR bootstrap/56258
24439         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
24440         instead of @itemx.
24442         PR inline-asm/56405
24443         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
24444         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
24446 2013-02-20  Jan Hubicka  <jh@suse.cz>
24448         PR tree-optimization/56265
24449         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
24450         when target is referenced for first time.
24452 2013-02-20  Richard Biener  <rguenther@suse.de>
24454         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
24455         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
24456         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
24457         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
24458         not return anything.
24459         (rename_ssa_copies): Do not remove unused locals.
24460         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
24461         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
24462         * passes.c (execute_function_todo): Do not schedule unused locals
24463         removal if cleanup_tree_cfg did something.
24464         * tree-ssa-live.c (remove_unused_locals): Dump statistics
24465         about the number of removed locals.
24467 2013-02-20  Richard Biener  <rguenther@suse.de>
24469         PR tree-optimization/56398
24470         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
24472 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24474         PR tree-optimization/55334
24475         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
24476         restricted pointers to arrays.
24478 2013-02-20  Richard Biener  <rguenther@suse.de>
24479             Jakub Jelinek  <jakub@redhat.com>
24481         PR tree-optimization/56396
24482         * tree-ssa-ccp.c (n_const_val): New static variable.
24483         (get_value): Return NULL for SSA names we don't have a lattice
24484         entry for.
24485         (ccp_initialize): Initialize n_const_val.
24486         * tree-ssa-copy.c (n_copy_of): New static variable.
24487         (init_copy_prop): Initialize n_copy_of.
24488         (get_value): Return NULL_TREE for SSA names we don't have a
24489         lattice entry for.
24491 2013-02-20  Martin Jambor  <mjambor@suse.cz>
24493         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
24495 2013-02-20  Richard Biener  <rguenther@suse.de>
24497         * genpreds.c (write_lookup_constraint): Do not compare first
24498         letter of the constraint again.
24500 2013-02-20  Richard Biener  <rguenther@suse.de>
24502         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
24503         and ceil_log2.
24504         (get_use_iv_cost): Terminate hashtable walk when coming across
24505         an empty entry.
24507 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
24509         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
24510         reassociation for avx2 targets.
24512 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24514         * config/microblaze/microblaze.c: microblaze_has_clz = 0
24515         Add version check for v8.10.a to enable microblaze_has_clz
24516         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
24517         version and TARGET_PATTERN_COMPARE check
24518         * config/microblaze/microblaze.md: New clzsi2 instruction
24520 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24522         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
24523         function before branching.
24525 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24527         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
24528         DUMP_INSN_RTX_UID.
24529         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
24531 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
24533         PR middle-end/55889
24534         * sel-sched.c: Include ira.h.
24535         (implicit_clobber_conflict_p): New function.
24536         (moveup_expr): Use it.
24537         * Makefile.in (sel-sched.o): Depend on ira.h.
24539 2013-02-19  Richard Biener  <rguenther@suse.de>
24541         PR tree-optimization/56384
24542         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
24543         (vn_hash_type): Split out from ...
24544         (vn_hash_constant_with_type): ... here.
24545         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
24546         (vn_phi_eq): Compare types from vn_phi_s structure.
24547         (vn_phi_lookup): Populate vn_phi_s type.
24548         (vn_phi_insert): Likewise.
24550 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
24552         PR tree-optimization/56350
24553         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
24554         if haven't found reduction or nested cycle operand, rather than
24555         asserting we must find it.
24557         PR tree-optimization/56381
24558         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
24559         to fold_build3.
24561 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
24562             Jakub Jelinek  <jakub@redhat.com>
24564         PR target/52555
24565         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
24566         (swap_optab_enable): Same.
24567         (init_all_optabs): Use argument instead of global.
24568         * tree.h (struct tree_optimization_option): New field target_optabs.
24569         * expr.h (init_all_optabs): Add argument to prototype.
24570         (TREE_OPTIMIZATION_OPTABS): New.
24571         (save_optabs_if_changed): Protoize.
24572         * optabs.h: Declare this_fn_optabs.
24573         * optabs.c (save_optabs_if_changed): New.
24574         Declare this_fn_optabs.
24575         (init_optabs): Add argument to init_all_optabs() call.
24576         * function.c (invoke_set_current_function_hook): Handle per
24577         function optabs.
24578         * function.h (struct function): New field optabs.
24579         * config/mips/mips.c (mips_set_mips16_mode): Handle when
24580         optimization_current_node has changed.
24581         * target-globals.h (save_target_globals_default_opts): Protoize.
24582         * target-globals.c (save_target_globals_default_opts): New.
24584 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
24586         PR target/56347
24587         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
24588         registers %fr12 and %fr12R as call used.
24590         PR target/56214
24591         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
24592         and HImode, require all displacements to be an integer multiple of
24593         their mode size.
24594         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
24595         only allow QImode and HImode when reload is in progress and strict is
24596         true.  Likewise for symbolic addresses.  Use base14_operand to check
24597         displacements in REG+BASE addresses.
24599 2013-02-18  Richard Biener  <rguenther@suse.de>
24601         PR tree-optimization/56366
24602         * tree-vect-loop.c (get_initial_def_for_induction): Properly
24603         handle sign-conversion of outer-loop initial induction value.
24605 2013-02-18  Richard Biener  <rguenther@suse.de>
24607         PR middle-end/56349
24608         * cfghooks.c (merge_blocks): If we merge a latch into another
24609         block adjust references to it.
24610         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
24611         (verify_loop_structure): Verify that a recorded latch is in fact
24612         a latch.
24614 2013-02-18  Richard Biener  <rguenther@suse.de>
24616         PR tree-optimization/56321
24617         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
24618         order SSA name release and virtual operand unlinking.
24620 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24622         * config/microblaze/microblaze.md (save_stack_block): Define.
24623         (restore_stack_block): Likewise.
24625 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24627         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
24628         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
24629         * config/microblaze/microblaze.c (microblaze_option_override):
24630         Bail out early for PIC modes when target does not support PIC.
24632 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24634         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
24635         Replace with a microblaze version.
24636         (microblaze_trampoline_init): Adapt for microblaze.
24637         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
24638         microblaze.
24640 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
24641             Dodji Seketeli  <dodji@redhat.com>
24643         PR asan/56330
24644         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
24645         (instrument_mem_region_access): Do not forget to always put
24646         instrumentation of the of 'base' and 'base + len' in a "if (len !=
24647         0) statement, even for cases where either 'base' or 'base + len'
24648         are not instrumented -- because they have been previously
24649         instrumented.  Simplify the logic by putting all the statements
24650         instrument 'base + len' inside a sequence, and then insert that
24651         sequence right before the current insertion point.  Then, to
24652         instrument 'base + len', just get an iterator on that statement.
24653         And do not forget to update the pointer to iterator the function
24654         received as argument.
24656 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
24658         PR rtl-optimization/56348
24659         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
24661 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
24663         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
24664         (clean_graph_dump_file): Pass base to start_graph_dump.
24666 2013-02-14  Richard Henderson  <rth@redhat.com>
24668         PR target/55941
24669         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
24671 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
24673         * collect2-aix.h: Define F_LOADONLY.
24675 2013-02-14  Richard Biener  <rguenther@suse.de>
24677         PR lto/50494
24678         * varasm.c (output_constant_def_1): Get the decl representing
24679         the constant as argument.
24680         (output_constant_def): Wrap output_constant_def_1.
24681         (make_decl_rtl): Use output_constant_def_1 with the decl
24682         representing the constant.
24683         (build_constant_desc): Optionally re-use a decl already
24684         representing the constant.
24685         (tree_output_constant_def): Adjust.
24687 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
24689         Fix an asan crash
24690         * asan.c (instrument_builtin_call):  Really put the length of the
24691         second source argument into src1_len.
24693 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24695         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
24696         argument.  If it is false, don't create edge from then_bb to
24697         fallthru_bb.
24698         (insert_if_then_before_iter): Pass true to it.
24699         (build_check_stmt): Pass false to it.
24700         (transform_statements): Flush hash table only on extended basic
24701         block boundaries, rather than at the beginning of every bb.
24702         Don't flush hash table on nonfreeing_call_p calls.
24703         * tree-flow.h (nonfreeing_call_p): New prototype.
24704         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
24706 2013-02-13  David S. Miller  <davem@davemloft.net>
24708         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
24710 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
24712         PR target/56184
24713         * ira.c (max_regno_before_ira): Move from ...
24714         (ira): ... here.
24715         (fix_reg_equiv_init): Use max_regno_before_ira instead of
24716         vec_safe_length.
24718 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
24720         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
24722 2013-02-13  Richard Biener  <rguenther@suse.de>
24724         PR lto/56295
24725         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
24726         globals in MEM_REFs.
24728 2013-02-13  Richard Biener  <rguenther@suse.de>
24730         * loop-init.c (loop_optimizer_init): Clear loop state when
24731         re-initializing preserved loops.
24732         * loop-unswitch.c (unswitch_single_loop): Return whether
24733         we unswitched the loop.  Do not verify loop state here.
24734         (unswitch_loops): When we unswitched a loop discover new loops.
24736 2013-02-13  Kostya Serebryany  <kcc@google.com>
24738         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
24739         on x86_64 linux.
24740         * sanitizer.def: Rename __asan_init to __asan_init_v1.
24742 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
24744         Avoid instrumenting duplicated memory access in the same basic block
24745         * Makefile.in (asan.o): Add new dependency on hash-table.h
24746         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
24747         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
24748         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
24749         (free_mem_ref_resources, has_mem_ref_been_instrumented)
24750         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
24751         (get_mem_ref_of_assignment): New functions.
24752         (get_mem_refs_of_builtin_call): Extract from
24753         instrument_builtin_call and tweak a little bit to make it fit with
24754         the new signature.
24755         (instrument_builtin_call): Use the new
24756         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
24757         of is_gimple_builtin_call.
24758         (instrument_derefs, instrument_mem_region_access): Insert the
24759         instrumented memory reference into the hash table.
24760         (maybe_instrument_assignment): Renamed instrument_assignment into
24761         this, and change it to advance the iterator when instrumentation
24762         actually happened and return true in that case.  This makes it
24763         homogeneous with maybe_instrument_assignment, and thus give a
24764         chance to callers to be more 'regular'.
24765         (transform_statements): Clear the memory reference hash table
24766         whenever we enter a new BB, when we cross a function call, or when
24767         we are done transforming statements.  Use
24768         maybe_instrument_assignment instead of instrumentation.  No more
24769         need to special case maybe_instrument_assignment and advance the
24770         iterator after calling it; it's now handled just like
24771         maybe_instrument_call.  Update comment.
24773 2013-02-13  Richard Biener  <rguenther@suse.de>
24775         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
24776         Fix loop discovery code.
24778 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
24780         PR inline-asm/56148
24781         * lra-constraints.c (process_alt_operands): Match early clobber
24782         operand with itself.  Check conflicts with earlyclobber only if
24783         the operand is not reloaded.  Prefer to reload conflicting operand
24784         if earlyclobber and matching operands are the same.
24786 2013-02-12  Richard Biener  <rguenther@suse.de>
24788         PR lto/56297
24789         * lto-streamer-out.c (write_symbol): Do not output symbols
24790         for hard register variables.
24792 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
24794         PR target/54222
24795         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
24796         (umulsidi3_insn, mulsidi3_insn): New insns.
24798 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
24800         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
24801         (struct tune_params): Add vec_costs field.
24802         * config/arm/arm.c (arm_builtin_vectorization_cost)
24803         (arm_add_stmt_cost): New functions.
24804         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
24805         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
24806         (arm_default_vec_cost): New struct of type cpu_vec_costs.
24807         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
24808         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
24809         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
24810         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
24812 2013-02-12  Richard Biener  <rguenther@suse.de>
24814         PR lto/56295
24815         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
24816         decls again if possible.
24818 2013-02-12  Richard Biener  <rguenther@suse.de>
24820         PR middle-end/56288
24821         * tree-ssa.c (verify_ssa_name): Fix check, move
24822         SSA_NAME_IN_FREE_LIST check up.
24824 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
24825             Steven Bosscher   <steven@gcc.gnu.org>
24827         PR rtl-optimization/56151
24828         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
24829         equal to op0 or op1, and last_insn pattern is CODE operation
24830         with MEM dest and one of the operands matches that MEM.
24832 2013-02-11  Sriraman Tallam  <tmsriramgoogle.com>
24834         * doc/extend.texi: Document Function Multiversioning and "default"
24835         parameter string to target attribute.
24836         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
24837         target attribute parameter is "default".
24838         (ix86_compare_version_priority): Remove checks for target attribute.
24839         (ix86_mangle_function_version_assembler_name): Change error to sorry.
24840         Remove check for target attribute equal to NULL. Add assert.
24841         (ix86_generate_version_dispatcher_body): Change error to sorry.
24843 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
24844             Jack Howarth  <howarth@bromo.med.uc.edu>
24845             Patrick Marlier  <patrick.marlier@gmail.com>
24847         PR libitm/55693
24848         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
24849         define ENDFILE_SPEC as TM_DESTRUCTOR.
24850         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
24852 2013-02-11  Alexander Potapenko  <glider@google.com>
24853             Jack Howarth  <howarth@bromo.med.uc.edu>
24854             Jakub Jelinek  <jakub@redhat.com>
24856         PR sanitizer/55617
24857         * config/darwin.c (cdtor_record): Rename ctor_record.
24858         (sort_cdtor_records): Rename sort_ctor_records.
24859         (finalize_dtors): New routine to sort destructors by
24860         priority before use in assemble_integer.
24861         (machopic_asm_out_destructor): Use finalize_dtors if needed.
24863 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
24865         PR rtl-optimization/56275
24866         * simplify-rtx.c (avoid_constant_pool_reference): Check that
24867         offset is non-negative and less than cmode size before
24868         calling simplify_subreg.
24870 2013-02-11  Richard Biener  <rguenther@suse.de>
24872         PR tree-optimization/56264
24873         * cfgloop.h (fix_loop_structure): Adjust prototype.
24874         * loop-init.c (fix_loop_structure): Return the number of
24875         newly discovered loops.
24876         * tree-cfgcleanup.c (repair_loop_structures): When new loops
24877         are discovered, do a full loop-closed SSA rewrite.
24879 2013-02-11  Richard Biener  <rguenther@suse.de>
24881         PR tree-optimization/56273
24882         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
24883         first VRP run.
24884         (check_array_ref): Fix missing newline in dumps.
24885         (search_for_addr_array): Likewise.
24887 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
24889         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
24891 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
24893         PR target/56256
24894         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
24896 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
24898         PR rtl-optimization/56246
24899         * lra-constraints.c (simplify_operand_subreg): Try to reuse
24900         reload pseudo.
24901         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
24902         constraints are satisfied.
24904 2013-02-08  Jeff Law  <law@redhat.com>
24906         PR debug/53948
24907         * emit-rtl.c (reg_is_parm_p): New function.
24908         * regs.h (reg_is_parm_p): New prototype.
24909         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
24910         callee-clobbered registers.
24912 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
24914         PR target/56043
24915         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
24916         If there is no implicit builtin declaration, just return NULL.
24918 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24920         * config/i386/sse.md (FMAMODEM): New mode iterator.
24921         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
24922         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
24924 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
24926         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
24927         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
24928         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
24930 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
24932         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
24933         (microblaze*-*-elf): Likewise.
24934         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
24935         LINK_SPEC.
24936         * config/microblaze/microblaze-c.c: Add builtin defines for
24937         _LITTLE_ENDIAN and _BIG_ENDIAN.
24938         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
24939         add to TARGET_DEFAULT flags.
24940         Expand ASM_SPEC and LINK_SPEC.
24941         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
24942         * config/microblaze/microblaze.md: Update extendsidi2 and
24943         movdi_internal instructions to use low-order / high-order reg
24944         print_operands.
24945         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
24946         options and inversemask / mask of LITTLE_ENDIAN.
24947         * config/microblaze/t-microblaze: Expand multilib options to
24948         include mlittle-endian (le) and update exceptions patterns.
24950 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
24952         PR rtl-optimization/56195
24953         * lra-constraints.c (get_reload_reg): Don't reuse regs
24954         if they have smaller mode than requested, if they have
24955         wider mode than requested, try to return a SUBREG.
24957         PR tree-optimization/56250
24958         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
24959         if type is unsigned and code isn't MULT_EXPR.
24961 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
24963         PR tree-optimization/56064
24964         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
24965         bits according to mode.
24966         * fixed-value.h (fixed_from_double_int)
24967         (const_fixed_from_double_int): Adjust comments.
24969 2013-02-08  Richard Biener  <rguenther@suse.de>
24971         PR lto/56231
24972         * lto-streamer.h (struct data_in): Remove current_file, current_line
24973         and current_col members.
24974         * lto-streamer-out.c (lto_output_location): Stream changed bits
24975         en-block for efficiency.
24976         * lto-streamer-in.c (clear_line_info): Remove.
24977         (lto_input_location): Cache current file, line and column
24978         globally via local statics.  Read changed bits en-block.
24979         (input_function): Do not call clear_line_info.
24980         (lto_read_body): Likewise.
24981         (lto_input_toplevel_asms): Likewise.
24983 2013-02-08  Michael Matz  <matz@suse.de>
24985         PR tree-optimization/52448
24986         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
24987         (nt_call_phase): New static.
24988         (add_or_mark_expr): Only mark accesses with newer phase than any
24989         call seen.
24990         (nonfreeing_call_p): New.
24991         (nt_init_block): Update nt_call_phase, mark blocks as visited.
24992         (nt_fini_block): Keep blocks marked as visited.
24993         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
24995 2013-02-08  Richard Biener  <rguenther@suse.de>
24997         * ira.c (ira): Free broken dominator information.
24999 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
25001         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
25003 2013-02-08  Marek Polacek  <polacek@redhat.com>
25005         * cfgloop.c (verify_loop_structure): Add more checking of headers.
25007 2013-02-08  Richard Biener  <rguenther@suse.de>
25009         PR middle-end/56181
25010         * cfgloop.h (flow_loops_find): Adjust.
25011         (bb_loop_header_p): Declare.
25012         * cfgloop.c (bb_loop_header_p): New function split out from ...
25013         (flow_loops_find): ... here.  Adjust function signature,
25014         support incremental loop structure update.
25015         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
25016         * cfgloopmanip.c (fix_loop_structure): Move ...
25017         * loop-init.c (fix_loop_structure): ... here.
25018         (apply_loop_flags): Split out from ...
25019         (loop_optimizer_init): ... here.
25020         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
25021         in incremental mode, only remove dead loops here.
25023 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
25025         PR target/54222
25026         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
25027         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
25028         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
25029         (*round<mode>3.libgcc): New insns for fixed-modes.
25030         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
25031         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
25032         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
25033         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
25034         implementations.  Define to __builtin_avr_absFX,
25035         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
25036         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
25037         __builtin_avr_countlsFX, respectively.
25038         * config/avr/avr-c.c (target.h): Include it.
25039         (enum avr_builtin_id): New enum.
25040         (avr_resolve_overloaded_builtin): New static function.
25041         (avr_register_target_pragmas): Use it to set
25042         targetm.resolve_overloaded_builtin.
25043         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
25044         tree nodes used by DEF_BUILTIN.
25045         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
25046         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
25047         <AVR_BUILTIN_xxBITS>: Same.
25049 2013-02-08  Richard Biener  <rguenther@suse.de>
25051         * cfgloop.c (verify_loop_structure): Properly handle
25052         a loop exiting to another loop header.
25053         * ira-int.h (ira_loops): Remove.
25054         * ira.c (ira_loops): Remove.
25055         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
25056         (do_reload): Use loop_optimizer_finalize.
25057         * ira-build.c (create_loop_tree_nodes): Use get_loops and
25058         number_of_loops to access the loop tree.
25059         (more_one_region_p): Likewise.
25060         (finish_loop_tree_nodes): Likewise.
25061         (rebuild_regno_allocno_maps): Likewise.
25062         (mark_loops_for_removal): Likewise.
25063         (mark_all_loops_for_removal): Likewise.
25064         (remove_unnecessary_regions): Likewise.
25065         (ira_build): Likewise.
25066         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
25068 2013-02-08  Richard Biener  <rguenther@suse.de>
25070         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
25071         * ipa-pure-const.c (analyze_function): Avoid calling
25072         mark_irreducible_loops twice.
25073         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
25075 2013-02-07  David S. Miller  <davem@davemloft.net>
25077         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
25078         on 'reg'.
25079         * var-tracking.c (vt_add_function_parameter): Test the presence of
25080         HAVE_window_save properly and do not remap argument registers when
25081         we have a leaf function.
25083 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
25085         PR bootstrap/56227
25086         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
25087         instead of "ll".
25088         * config/i386/i386.c (ix86_print_operand): Ditto.
25090 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25092         * lra-constraints.c (process_alt_operands): Fix recently added comment.
25094 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
25096         PR rtl-optimization/56225
25097         * lra-constraints.c (process_alt_operands): Check that reload hard
25098         reg can hold value for strict_low_part.
25100 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25102         PR debug/56154
25103         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
25104         dwarf2out_end_function.
25105         (in_first_function_p, maybe_at_text_label_p,
25106         first_loclabel_num_not_at_text_label): New variables.
25107         (dwarf2out_var_location): In the first function find out
25108         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
25109         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
25110         functions.
25112 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
25114         PR rtl-optimization/56178
25115         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
25116         SUBREG of a register.  Tidy up related block of code.
25117         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
25118         note if the source is a register or a SUBREG of a register.
25120 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
25122         PR target/56228
25123         * config/rs6000/rs6000.md (ptrm): New mode attr.
25124         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
25125         call_value_indirect_aix<pttrsize>,
25126         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
25127         m in constraints.
25129 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25131         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
25132         if -bnortl. Convert to strcmp and strncmp.
25134 2013-02-07  Alan Modra  <amodra@gmail.com>
25136         PR target/54009
25137         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
25138         addresses won't wrap when offsetting.
25139         (rs6000_secondary_reload): Provide secondary reloads needed for
25140         wrapping LO_SUM addresses.
25142 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
25144         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
25145         MACH, just __MACH__.
25147 2013-02-06  Richard Biener  <rguenther@suse.de>
25149         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
25150         instead of calling fix_loop_structure.
25152 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
25154         PR middle-end/56217
25155         * omp-low.c (use_pointer_for_field): Return false if
25156         lower_send_shared_vars doesn't generate any copy-out code.
25158 2013-02-06  Tom de Vries  <tom@codesourcery.com>
25160         PR rtl-optimization/56131
25161         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
25162         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
25163         of the label is NULL.  Add comment.
25165 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
25167         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
25169         PR sanitizer/55374
25170         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
25171         (STATIC_LIBTSAN_LIBS): Likewise.
25172         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
25173         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
25174         is defined, don't add anything else beyond that.
25175         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
25176         (LINK_COMMAND_SPEC): Use them.
25178         PR tree-optimization/56205
25179         * tree-stdarg.c (check_all_va_list_escapes): Return true if
25180         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
25181         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
25183 2013-02-05  Richard Biener  <rguenther@suse.de>
25185         PR tree-optimization/53342
25186         PR tree-optimization/53185
25187         * tree-vectorizer.h (vect_check_strided_load): Remove.
25188         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
25189         not disallow peeling for vectorized strided loads.
25190         (vect_check_strided_load): Make static and simplify.
25191         (vect_analyze_data_refs): Adjust.
25192         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
25193         correctly when vectorizing strided loads.
25195 2013-02-05  Richard Biener  <rguenther@suse.de>
25197         * doc/install.texi: Refer to ISL, not PPL.
25199 2013-02-05  Jan Hubicka  <jh@suse.cz>
25201         PR tree-optimization/55789
25202         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
25204 2013-02-05  Jan Hubicka  <jh@suse.cz>
25206         PR tree-optimization/55789
25207         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
25208         the dead call anyway.
25210 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
25212         PR sanitizer/55374
25213         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
25215 2013-02-04  Alexander Potapenko  <glider@google.com>
25216             Jack Howarth  <howarth@bromo.med.uc.edu>
25217             Jakub Jelinek  <jakub@redhat.com>
25219         PR sanitizer/55617
25220         * config/darwin.c (sort_ctor_records): Stabilized qsort
25221         on constructor priority by using original position.
25222         (finalize_ctors): New routine to sort constructors by
25223         priority before use in assemble_integer.
25224         (machopic_asm_out_constructor): Use finalize_ctors if needed.
25226 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
25228         PR libstdc++/54314
25229         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
25230         about visibility on artificial decls.
25231         * config/sol2.c (solaris_assemble_visibility): Likewise.
25233 2013-02-04  Kai Tietz  <ktietz@redhat.com>
25235         PR target/56186
25236         * config/i386/i386.c (function_value_ms_64): Add additional valtype
25237         argument and improve checking of return-argument types for 16-byte
25238         modes.
25239         (ix86_function_value_1): Add additional valtype argument on call
25240         of function_value_64.
25241         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
25242         handling infunction_value_64 function.
25244 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
25246         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
25248 2013-02-04  Richard Biener  <rguenther@suse.de>
25250         PR tree-optimization/56188
25251         * tree-ssa-structalias.c (label_visit): Consider case with
25252         initially non-empty points-to set.
25253         (perform_var_substitution): Dump node mapping and clean up.
25255 2013-02-04  Richard Guenther  <rguenther@suse.de>
25257         PR lto/56168
25258         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
25259         node prevail as last resort.
25260         (lto_symtab_merge_decls): Remove guard on LTRANS here.
25261         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
25263 2013-02-04  Richard Biener  <rguenther@suse.de>
25265         PR tree-optimization/56113
25266         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
25267         Merge into ...
25268         (equiv_class_lookup_or_add): ... this.
25269         (label_visit): Adjust and fix error in previous patch.
25270         (perform_var_substitution): Adjust.
25272 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
25274         * config/sh/divtab.c: Fix formatting and comments throughout the file.
25275         * config/sh/sh4-300.md: Likewise.
25276         * config/sh/sh4a.md: Likewise.
25277         * config/sh/constraints.md: Likewise.
25278         * config/sh/sh.md: Likewise.
25279         * config/sh/netbsd-elf.h: Likewise.
25280         * config/sh/predicates.md: Likewise.
25281         * config/sh/sh-protos.h: Likewise.
25282         * config/sh/ushmedia.h: Likewise.
25283         * config/sh/linux.h: Likewise.
25284         * config/sh/sh.c: Likewise.
25285         * config/sh/superh.h: Likewise.
25286         * config/sh/elf.h: Likewise.
25287         * config/sh/sh4.md: Likewise.
25288         * config/sh/sh.h: Likewise.
25290 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
25292         * config/pa/constraints.md: Adjust unused letters.  Change "T"
25293         constraint to match_test floating_point_store_memory_operand().
25294         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
25295         (base14_operand): New.
25296         (floating_point_store_memory_operand): New.
25297         (integer_store_memory_operand): Revise to use base14_operand and
25298         reg_plus_base_memory_operand.
25299         (move_dest_operand): Allow symbolic_memory_operands.
25300         (symbolic_memory_operand): Check for LO_SOM.
25301         (symbolic_operand): Change default case to break.
25302         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
25303         CONST_DOUBLE values to be reloaded by putting them into memory when
25304         the destination is a floating point register.
25305         (movdf): Remove code to handle CONST_DOUBLE.
25306         (movsf): Likewise.
25307         (reload_indf_r1): New.
25308         (reload_insf_r1): New.
25309         Consistently use "Q" and "T" constraints with integer and floating
25310         point move instructions, respectively.
25311         (movdi): Remove FAIL.
25312         Change predicate for source operand unamed DImode move from
25313         general_operand to move_src_operand.
25314         (umulsidi3): Change predicate for destination operand to
25315         register_operand.
25316         Likewise for similar unamed patterns.
25317         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
25318         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
25319         (hppa_legitimize_address): Simplify mask calculation.
25320         (pa_emit_move_sequence): Revised handling of secondary reloads from
25321         REG+D addresses for floating point loads and stores.  Directly handle
25322         loading CONST0_RTX (mode) to a floating point register.
25323         (pa_secondary_reload): Handle reloading DF and SFmode constant values
25324         to floating point registers.  Don't restrict secondary reloads to
25325         floating point registers to integer modes.  Revise some comments and
25326         cleanup some code.
25327         (TARGET_LEGITIMATE_ADDRESS_P): Define.
25328         (pa_legitimate_address_p): New.
25329         (pa_legitimize_reload_address): New.
25330         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
25331         (STRICT_REG_OK_FOR_BASE_P): New.
25332         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
25333         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
25335 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
25336             Andrew Dixie  <andrewd@gentrack.com>
25338         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
25339         flag set.
25341 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
25343         * expmed.c (extract_bit_field_1): Pass the full width of the
25344         structure to get_best_reg_extraction_insn.
25346 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25348         PR target/54601
25349         * configure.ac (use_cxa_atexit): Add AIX.
25350         * configure: Regenerate.
25352         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
25354 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
25356         PR debug/54793
25357         * final.c (need_profile_function): New variable.
25358         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
25359         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
25360         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
25361         notes, targetm.asm_out.function_prologue doesn't emit anything,
25362         HAVE_prologue and profiler should be emitted before prologue,
25363         set need_profile_function instead of emitting it.
25364         (final_scan_insn): If need_profile_function, emit
25365         profile_function on the first NOTE_INSN_BASIC_BLOCK or
25366         NOTE_INSN_FUNCTION_BEG note.
25368 2013-02-01  Richard Henderson  <rth@redhat.com>
25370         * config/rs6000/rs6000.md (smulditi3): New.
25371         (umulditi3): New.
25373         * config/alpha/alpha.md (umulditi3): New.
25375 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
25377         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
25378         (ASM_OUTPUT_ALIGNED_LOCAL): New.
25380 2013-02-01  Richard Biener  <rguenther@suse.de>
25382         PR tree-optimization/56113
25383         * tree-ssa-structalias.c (label_visit): Reduce work for
25384         single-predecessor nodes.
25386 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
25388         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
25389         range isn't testing for zero.
25391 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
25393         PR middle-end/56113
25394         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
25396 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
25397             Nick Clifton  <nickc@redhat.com>
25399         * config/v850/constraints.md (Q): Define as a memory constraint.
25400         * config/v850/predicates.md (label_ref_operand): New predicate.
25401         (e3v5_shift_operand): New predicate.
25402         (ior_operator): New predicate.
25403         * config/v850/t-v850: Add e3v5 multilib.
25404         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
25405         (v850_gen_movdi): Prototype.
25406         * config/v850/v850.c: Add support for e3v5 architecture.
25407         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
25408         TARGET_V850E_UP.
25409         (construct_save_jarl): Add e3v5 long JARL support.
25410         (v850_adjust_insn_length): New function.  Adjust length of call
25411         insns when using e3v5 instructions.
25412         (v850_gen_movdi): New function: Generate instructions to move a
25413         DImode value.
25414         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
25415         (CPP_SPEC): Define __v850e3v5__ as appropriate.
25416         (TARGET_USE_FPU): Enable for e3v5.
25417         (CONST_OK_FOR_W): New macro.
25418         (ADJUST_INSN_LENGTH): Define.
25419         * config/v850/v850.md (UNSPEC_LOOP): Define.
25420         (attr cpu): Add v850e3v5.
25421         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
25422         (movdi): New pattern.
25423         (movdi_internal): New pattern.
25424         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
25425         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
25426         (cstoresf4): Likewise.
25427         (cstoredf4): Likewise.
25428         (insv): New pattern.
25429         (rotlso3_a): New pattern.
25430         (rotlsi3_b): New pattern
25431         (rotlsi3_v850e3v5): New pattern.
25432         (doloop_begin): New pattern.
25433         (fix_loop_counter): New pattern.
25434         (doloop_end): New pattern.
25435         (branch_normal): Add e3v5 long branch support.
25436         (branch_invert): Likewise.
25437         (branch_z_normal): Likewise.
25438         (branch_z_invert): Likewise.
25439         (branch_nz_normal): Likewise.
25440         (branch_nz_invert): Likewise.
25441         (call_internal_short): Add e3v5 register-indirect JARL support.
25442         (call_internal_long): Likewise.
25443         (call_value_internal_short): Likewise.
25444         (call_value_internal_long): Likewise.
25445         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
25446         (mloop): New option.
25447         * config.gcc: Add support for configuring v840e3v5 target.
25448         * doc/invoke.texi: Document new v850 specific command line options.
25450 2013-01-31  Paul Koning  <ni1d@arrl.net>
25452         PR debug/55059
25453         PR debug/54508
25454         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
25455         children if parent is a class.
25456         (prune_unused_types_prune): Don't add DW_AT_declaration.
25458 2013-01-31  Richard Biener  <rguenther@suse.de>
25460         PR tree-optimization/56157
25461         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
25462         match up operand with SLP child.
25464 2013-01-31  Jason Merrill  <jason@redhat.com>
25466         PR debug/54410
25467         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
25468         parameters the first time.
25469         (gen_scheduled_generic_parms_dies): Check completeness here.
25471 2013-01-31  Richard Biener  <rguenther@suse.de>
25473         PR middle-end/53073
25474         * common.opt (faggressive-loop-optimizations): New flag,
25475         enabled by default.
25476         * doc/invoke.texi (faggressive-loop-optimizations): Document.
25477         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
25478         infer_loop_bounds_from_undefined by it.
25480 2013-01-31  Richard Biener  <rguenther@suse.de>
25482         PR tree-optimization/56150
25483         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
25484         visit virtual operands.
25485         (find_uses_to_rename_bb): Likewise.
25487 2013-01-31  Richard Biener  <rguenther@suse.de>
25489         PR tree-optimization/56150
25490         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
25491         mixed store non-store stmts.
25493 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
25495         PR sanitizer/55374
25496         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
25497         LIBASAN_EARLY_SPEC is defined.
25498         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
25499         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
25500         before %o.
25501         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
25503         PR c++/55742
25504         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
25505         invalid args instead of ICEing on it.
25506         (ix86_valid_target_attribute_tree): Return error_mark_node if
25507         ix86_valid_target_attribute_inner_p failed.
25508         (ix86_valid_target_attribute_p): Return false only if
25509         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
25510         target("default") attribute.
25511         (sorted_attr_string): Change argument from const char * to tree,
25512         merge in all target attribute arguments rather than just one.
25513         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
25514         instead of free.  Avoid using strcat.
25515         (ix86_mangle_function_version_assembler_name): Mangle
25516         target("default") as if no target attribute is present.  Adjust
25517         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
25518         instead of xmalloc and XDELETEVEC instead of free.
25519         (ix86_function_versions): Don't return true if one of the decls
25520         doesn't have target attribute.  If they don't and one of the decls
25521         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
25522         sorted_attr_string caller.  Use XDELETEVEC instead of free.
25523         (ix86_supports_function_versions): Remove.
25524         (make_name): Fix up formatting.
25525         (make_dispatcher_decl): Remove resolver_name and its initialization.
25526         Avoid leaking memory.
25527         (is_function_default_version): Return true if there is
25528         target("default") attribute rather than no target attribute at all.
25529         (make_resolver_func): Avoid leaking memory.
25530         (ix86_generate_version_dispatcher_body): Likewise.
25531         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
25532         * target.def (supports_function_versions): Remove.
25533         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
25534         * doc/tm.texi: Regenerated.
25536 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
25538         PR rtl-optimization/56144
25539         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
25540         for values with side effects.
25542 2013-01-30  Richard Biener  <rguenther@suse.de>
25544         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
25545         (sparseset_pop): Likewise.
25546         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
25547         to be able to use quick_push in the worker loop.
25549 2013-01-30  Marek Polacek  <polacek@redhat.com>
25551         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
25553 2013-01-30  Richard Biener  <rguenther@suse.de>
25555         PR lto/56147
25556         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
25558 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
25560         PR tree-optimization/56064
25561         * fixed-value.c (fixed_from_double_int): New function.
25562         * fixed-value.h (fixed_from_double_int): New prototype.
25563         (const_fixed_from_double_int): New static inline function.
25564         * fold-const.c (native_interpret_fixed): New static function.
25565         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
25566         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
25567         (native_encode_fixed): New static function.
25568         (native_encode_expr) <FIXED_CST>: Use it.
25569         (native_interpret_int): Move double_int worker code to...
25570         * double-int.c (double_int::from_buffer): ...this new static method.
25571         * double-int.h (double_int::from_buffer): Prototype it.
25573 2013-01-30  Richard Biener  <rguenther@suse.de>
25575         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
25576         New pointer-map and obstack.
25577         (init_alias_vars): Allocate pointer-map and obstack.
25578         (delete_points_to_sets): Free them.
25579         (find_what_var_points_to): Cache result.
25580         (find_what_p_points_to): Adjust for changed interface of
25581         find_what_var_points_to.
25582         (compute_points_to_sets): Likewise.
25583         (ipa_pta_execute): Likewise.
25585 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25587         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
25588         * configure: Regenerate.
25589         * config.in: Regenerate.
25590         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
25591         #nobits/#progbits if supported.
25593 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
25595         PR target/56121
25596         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
25597         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
25598         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
25600 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25602         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
25603         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
25605 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25607         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
25608         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
25610 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25612         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
25613         declaration.
25614         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
25615         * config/arm/cortex-a7.md: New bypasses using
25616         arm_mac_accumulator_is_result.
25618 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25620         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
25621         (cortex_a7_neon_mla): Likewise.
25622         (cortex_a7_fpfmad): New reservation.
25623         (cortex_a7_fpmacs): Use ffmas and update required units.
25624         (cortex_a7_fpmuld): Update required units and latency.
25625         (cortex_a7_fpmacd): Likewise.
25626         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
25627         (cortex_a7_neon). Likewise.
25628         (bypass) Update participating units.
25630 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
25632         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
25633         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
25634         from fmac to ffma.
25635         * config/arm/vfp11.md (vfp_farith): Use ffmas.
25636         (vfp_fmul): Use ffmad.
25637         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
25638         (cortex_r4_fmacd): Use ffmad.
25639         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
25640         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
25641         (cortex_a9_fmacd): Use ffmad.
25642         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
25643         (cortex_a8_vfp_macd): Use ffmad.
25644         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
25645         (cortex_a5_fpmacd): Use ffmad.
25646         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
25647         (cortex_a15_vfp_macd): Use ffmad.
25648         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
25650 2013-01-29  Jason Merrill  <jason@redhat.com>
25652         PR libstdc++/54314
25653         * varasm.c (default_assemble_visibility): Don't warn about
25654         visibility on artificial decls.
25656 2013-01-29  Richard Biener  <rguenther@suse.de>
25658         PR tree-optimization/56113
25659         * tree-ssa-structalias.c (equiv_class_lookup): Also return
25660         the bitmap leader.
25661         (label_visit): Free duplicate bitmaps and record the leader instead.
25662         (perform_var_substitution): Adjust.
25664 2013-01-29  Richard Biener  <rguenther@suse.de>
25666         PR tree-optimization/55270
25667         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
25668         the CFG, schedule loops for fixup.
25670 2013-01-29  Nick Clifton  <nickc@redhat.com>
25672         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
25673         SP_REG.
25675 2013-01-28  Leif Ekblad  <leif@rdos.net>
25677         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
25678         * config/i386/i386.h (TARGET_RDOS): New macro.
25679         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
25680         * config/i386/i386.c (ix86_option_override_internal): For 64bit
25681         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
25682         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
25683         DEFAULT_LARGE_SECTION_THRESHOLD.
25684         * config/i386/i386.md (R14_REG, R15_REG): New constants.
25685         * config/i386/rdos.h: New file.
25686         * config/i386/rdos64.h: New file.
25688 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
25690         PR other/54814
25691         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
25692         TEST_HARD_REG_BIT.
25694 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25696         PR rtl-optimization/56117
25697         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
25698         call cselib_lookup_from_insn on the MEM before calling
25699         add_insn_mem_dependence.
25701 2013-01-28  Richard Biener  <rguenther@suse.de>
25703         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
25704         to a stmt that didn't have one.
25705         (copy_phis_for_bb): Likewise for PHI arguments.
25706         (copy_debug_stmt): Likewise for debug stmts.
25708 2013-01-28  Richard Biener  <rguenther@suse.de>
25710         PR tree-optimization/56034
25711         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
25712         (partition_builtin_p): Adjust.
25713         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
25714         it is the last partition.
25715         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
25716         up the vertex for the definition.
25717         (classify_partition): Classify whether a partition is a
25718         PKIND_REDUCTION, thus has uses outside of the loop.
25719         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
25720         Merge all PKIND_REDUCTION partitions into the last partition.
25721         (tree_loop_distribution): Seed partitions from reductions as well.
25723 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
25725         PR tree-optimization/56125
25726         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
25727         pow(x,c) into sqrt(x) * powi(x, n/2) or
25728         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
25729         optimizing for size.
25730         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
25731         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
25732         integer.
25734         PR tree-optimization/56094
25735         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
25736         to UNKNOWN_LOCATION while gimplifying expr.
25738 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
25740         PR target/56114
25741         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
25742         operand 0 in movabs insn template for -masm=intel asm alternative.
25743         (*movabs<mode>_2): Ditto for operand 1.
25745 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
25747         PR target/54663
25748         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
25749         of microblaze-c.o
25751 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25753         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
25754         tm_file.
25756 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25758         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
25759         Undef to avoid warning.
25761 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
25763         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
25764         * configure: Regenerate.
25766 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
25768         PR tree-optimization/56098
25769         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
25770         for stmts with volatile ops.
25771         (cond_store_replacement): Don't optimize if assign has volatile ops.
25772         (cond_if_else_store_replacement_1): Don't optimize if either
25773         then_assign or else_assign have volatile ops.
25774         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
25775         volatile ops.
25777 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25779         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
25781 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25783         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
25784         missing ':' in asm example.
25786 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
25788         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
25789         entries into lane and laneq entries.
25790         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
25791         Remove AdvSIMD scalar modes.
25792         (aarch64_sq<r>dmulh_laneq<mode>): New.
25793         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
25794         modes.
25795         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
25796         builtin implementations to relfect changes in RTL in aarch64-simd.md.
25797         * config/aarch64/iterators.md (VCOND): New.
25798         (VCONQ): New.
25800 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
25802         PR target/54222
25803         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
25804         Add NULL LIBNAME argument to existing definitions.
25805         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
25806         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
25807         * config/avr/avr.c (DEF_BUILTIN): Same.
25808         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
25809         (avr_expand_builtin): Expand to a vanilla call if a libgcc
25810         implementation is available (DECL_ASSEMBLER_NAME is set).
25811         (avr_fold_absfx): New static function.
25812         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
25813         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
25814         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
25815         AVR_BUILTIN_ABSLLK.
25816         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
25817         (abshk, absk, abslk, absllk): Provide as static inline functions.
25819 2013-01-25  Marek Polacek  <polacek@redhat.com>
25821         PR tree-optimization/56035
25822         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
25824 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
25826         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
25827         (*movtf_internal_rex64): Add (!o,C) alternative
25828         (*movxf_internal_rex64): Ditto.
25829         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
25831 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
25833         * doc/invoke.texi: fix typo.
25834         * doc/objc.texi: fix typo.
25836 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
25838         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
25839         for the first two alternatives.
25841 2013-01-24  Diego Novillo  <dnovillo@google.com>
25843         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
25844         (ggc-zone.o): Remove.
25845         * configure.ac: Remove option --with-gc.
25846         * configure: Re-generate.
25847         * doc/install.texi: Remove documentation for --with-gc.
25848         * gengtype.c (write_enum_defn): Remove.  Update all users.
25849         (write_Types_process_field): Remove generation of gt_e_* argument.
25850         (output_type_enum): Remove.  Update all users.
25851         (write_enum_defn): Remove.  Update all users.
25852         (enum alloc_zone): Remove.  Update all users.
25853         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
25854         * ggc-common.c (ggc_splay_alloc): Remove first argument.
25855         Update all callers.
25856         (struct ptr_data): Remove field TYPE.  Update all users.
25857         (gt_pch_note_object): Remove argument TYPE.  Update all users.
25858         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
25859         Update all users.
25860         * ggc-none.c (ggc_alloc_typed_stat): Remove.
25861         (struct alloc_zone): Remove.
25862         (ggc_internal_alloc_zone_stat): Remove.
25863         (ggc_internal_cleared_alloc_zone_stat): Remove.
25864         * ggc-page.c (ggc_alloc_typed_stat): Remove.
25865         (ggc_pch_count_object): Remove last argument.  Update all users.
25866         (ggc_pch_alloc_object): Remove last argument.  Update all users.
25867         (struct alloc_zone): Remove.
25868         * ggc-zone.c: Remove.
25869         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
25870         (struct alloc_zone): Remove.
25871         (ggc_alloc_typed_stat): Remove.
25872         (ggc_alloc_typed): Remove.
25873         (ggc_splay_alloc): Remove first argument.
25874         (rtl_zone): Remove.  Update all users.
25875         (tree_zone): Remove.  Update all users.
25876         (tree_id_zone): Remove.  Update all users.
25877         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
25878         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
25879         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
25880         * tree-ssanames.c: Remove references to zone allocator in comments.
25882 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
25884         * config/avr/avr.c (avr_out_fract): Make register numbers that
25885         might be outside of source operand signed.
25887 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
25889         * config/i386/constraints.md (Yf): New constraint.
25890         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
25891         of f constraint to conditionaly disable x87 register preferences.
25892         (*movdf_internal): Ditto.
25893         (*movsf_internal): Ditto.
25895 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
25897         PR inline-asm/55934
25898         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
25899         that have operands with impossible constraints.
25900         Add a FIXME for a speed-up opportunity.
25901         * lra-constraints.c (process_alt_operands): Verify that a class
25902         selected from constraints on asms is valid for the operand mode.
25903         (curr_insn_transform): Remove incorrect comment.
25905 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
25907         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
25908         TOC operand is a valid symbol ref in the constant pool.
25910 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
25912         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
25914 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
25916         PR target/54222
25917         * config/avr/stdfix.h: New file.
25918         * t-avr (stdfix-gcc.h): New rule to build it.
25919         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
25921 2013-01-23  Kostya Serebryany  <kcc@google.com>
25923         * config/darwin.h: remove dependency on
25924         CoreFoundation (asan on Mac OS).
25926 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
25928         PR target/49069
25929         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
25930         instead of cmpdi_operand for first comparison operand.
25931         Don't assert that comparison operands aren't both constants.
25933 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25935         * doc/install.texi (Downloading the Source): Update references to
25936         downloading separate components.
25938 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
25940         * doc/extend.texi (__int128): Improve grammar.
25942 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
25944         PR target/56028
25945         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
25946         alternative to (o,r).
25947         (*movdi_internal_rex64): Remove (!o,n) alternative.
25948         (DImode immediate->memory splitter): Remove.
25949         (DImode immediate->memory peephole2): Remove.
25950         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
25951         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
25952         alternative to (!o,*r).
25953         (*movtf_internal_sse): New pattern.
25954         (*movxf_internal_rex64): New pattern.
25955         (*movxf_internal): Disable for TARGET_64BIT.
25956         (*movdf_internal_rex64): Remove (!o,F) alternative.
25958 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
25960         PR middle-end/56074
25961         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
25962         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
25963         * tree-vect-loop-manip.c (find_loop_location): Also ignore
25964         stmt locations where LOCATION_LOCUS of the stmt location is
25965         UNKNOWN_LOCATION or BUILTINS_LOCATION.
25967         PR target/55686
25968         * config/i386/i386.md (UNSPEC_STOS): New.
25969         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
25970         *strsetqi_1): Add UNSPEC_STOS.
25972 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
25974         PR c++/56067
25975         * doc/invoke.texi: Remove left over -Wsynth example.
25977 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25979         PR tree-optimization/56051
25980         * fold-const.c (fold_binary_loc): Don't fold
25981         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
25982         a narrowing conversion, or widening conversion from signed
25983         to unsigned.
25985 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
25987         PR rtl-optimization/56023
25988         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
25989         dependent on debug instruction.
25991 2013-01-21  Martin Jambor  <mjambor@suse.cz>
25993         PR middle-end/56022
25994         * function.c (allocate_struct_function): Call
25995         invoke_set_current_function_hook earlier.
25997 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
25999         * reload1.c (init_reload): Only initialize reload_obstack
26000         during the first call.
26002 2013-01-21  Marek Polacek  <polacek@redhat.com>
26004         * cfgloop.c (verify_loop_structure): Fix up grammar.
26006 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
26008         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
26009         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
26011 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26013         PR target/56058
26014         * config/arm/marvell-pj4.md: Update copyright year.
26015         Fix up use of alu to alu_reg and simple_alu_imm.
26017 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
26019         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
26021 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
26023         PR target/55433
26024         * lra-constraints.c (curr_insn_transform): Don't reuse original
26025         insn for secondary memory move when memory mode should be different.
26027 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26029         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
26030         atomic_storedi_1): New patterns.
26032 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26034         btver2 pipeline descriptions.
26035         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
26036         descriptions.
26037         * config/i386/i386.md (btver2_decode): New type attributes.
26038         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
26039         type attributes.
26040         * config/i386/btver2.md: New file describing btver2 pipelines.
26042 2013-01-19  Andrew Pinski  <apinski@cavium.com>
26044         PR tree-optimization/52631
26045         * tree-ssa-sccvn (visit_use): Before looking up the original
26046         statement, try looking up the simplified expression.
26048 2013-01-19  Anthony Green  <green@moxielogic.com>
26050         * config/moxie/moxie.c (moxie_expand_prologue): Set
26051         current_function_static_stack_size.
26053 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26055         PR tree-optimization/56029
26056         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
26057         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
26059 2013-01-18  Sharad Singhai  <singhai@google.com>
26061         PR tree-optimization/55995
26062         * dumpfile.c (dump_loc): Print location only if available.
26063         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
26065 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
26067         PR target/55433
26068         * lra-constraints.c (curr_insn_transform): Reuse original insn for
26069         secondary memory move.
26070         (inherit_reload_reg): Use rclass instead of cl for
26071         check_secondary_memory_needed_p.
26073 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
26075         PR middle-end/56015
26076         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
26077         the case where writing real complex part of target modifies op1.
26079 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
26081         * config/aarch64/aarch64-simd.md
26082         (aarch64_vcond_internal<mode>): Handle unordered cases.
26083         * config/aarch64/iterators.md (v_cmp_result): New.
26085 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
26086             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26088         * config/arm/marvell-pj4.md: New file.
26089         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
26090         * config/arm/arm.md (generic_sched): Add marvell_pj4.
26091         (generic_vfp): Likewise.
26092         * config/arm/arm-cores.def: Add marvell-pj4.
26093         * config/arm/arm-tune.md: Regenerate.
26094         * config/arm/arm-tables.opt: Regenerate.
26095         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
26096         * doc/invoke.texi: Document marvell-pj4.
26098 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
26100         * config/aarch64/arm_neon.h: Map scalar types to standard types.
26102 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26104         PR debug/54114
26105         PR debug/54402
26106         PR debug/49888
26107         * var-tracking.c (negative_power_of_two_p): New.
26108         (global_get_addr_cache, local_get_addr_cache): New.
26109         (get_addr_from_global_cache, get_addr_from_local_cache): New.
26110         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
26111         heading comment.
26112         (vt_stack_offset_p): Remove.
26113         (vt_canon_true_dep): Always canonicalize loc's address.
26114         (clobber_overlapping_mems): Make sure we have a MEM.
26115         (local_get_addr_clear_given_value): New.
26116         (val_reset): Clear local cached entries.
26117         (compute_bb_dataflow): Create and release the local cache.
26118         Disable duplicate MEMs clobbering.
26119         (emit_notes_in_bb): Clobber MEMs likewise.
26120         (vt_emit_notes): Create and release the local cache.
26121         (vt_initialize, vt_finalize): Create and release the global
26122         cache, respectively.
26123         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
26125 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26127         PR libmudflap/53359
26128         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
26129         not found in the symtab.
26131 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
26133         PR debug/56006
26134         PR rtl-optimization/55547
26135         PR rtl-optimization/53827
26136         PR debug/53671
26137         PR debug/49888
26138         * alias.c (offset_overlap_p): New, factored out of...
26139         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
26140         the conservative special case for symbolic constants.  Don't
26141         adjust zero sizes on alignment.
26143 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
26145         PR rtl-optimization/52573
26146         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
26147         REG_UNUSED for the same register.
26149 2013-01-17  Richard Biener  <rguenther@suse.de>
26150             Marek Polacek  <polacek@redhat.com>
26152         PR rtl-optimization/55833
26153         * loop-unswitch.c (unswitch_loops): Move loop verification...
26154         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
26155         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
26156         Set it to true when we're removing a loop from hierarchy tree in
26157         an irreducible region.
26158         (fix_bb_placements): Adjust caller.
26159         (fix_loop_placements): Likewise.
26161 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
26163         * config/avr/builtins.def (DEF_BUILTIN): Factor out
26164         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
26165         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
26166         Remove ID.  Adjust comments.
26167         * config/avr/avr-c.c (avr_builtin_name): Remove.
26168         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
26169         * config/avr/avr.c (avr_tolower): New static function.
26170         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
26171         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
26172         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
26173         default expansion.
26175 2013-01-17  Jan Hubicka  <jh@suse.cz>
26177         PR tree-optimization/55273
26178         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
26180 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
26182         PR target/55981
26183         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
26184         store through atomic_store<mode>_1.
26185         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
26187 2013-01-17  Martin Jambor  <mjambor@suse.cz>
26189         PR tree-optimizations/55264
26190         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
26191         for virtual methods.
26192         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
26193         virtual methods before inlining is over.
26194         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
26195         virtual functions.
26196         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
26197         non-virtual.
26199 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
26201         PR rtl-optimization/56005
26202         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
26203         pending reads for prefetch.
26205 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
26207         * config/aarch64/aarch64.md
26208         (*cstoresi_neg_uxtw): New pattern.
26209         (*cmovsi_insn_uxtw): New pattern.
26210         (*<optab>si3_uxtw): New pattern.
26211         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
26212         (*<optab>si3_insn_uxtw): New pattern.
26213         (*bswapsi2_uxtw): New pattern.
26215 2013-01-16  Richard Biener  <rguenther@suse.de>
26217         * tree-inline.c (tree_function_versioning): Remove set but
26218         never used variable.
26220 2013-01-16  Richard Biener  <rguenther@suse.de>
26222         PR tree-optimization/55964
26223         * tree-flow.h (rename_variables_in_loop): Remove.
26224         (rename_variables_in_bb): Likewise.
26225         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
26226         (copy_loop_before): Adjust and delete update-ssa status.
26227         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
26228         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
26229         (rename_variables_in_loop): Remove.
26230         (slpeel_update_phis_for_duplicate_loop): Likewise.
26231         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
26232         use available cfg machinery instead of duplicating it.
26233         Update PHI nodes and perform poor-mans SSA update here.
26234         (slpeel_tree_peel_loop_to_edge): Adjust.
26236 2013-01-16  Richard Biener  <rguenther@suse.de>
26238         PR tree-optimization/54767
26239         PR tree-optimization/53465
26240         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
26241         (vrp_visit_phi_node): For PHI arguments coming via backedges
26242         drop all symbolical range information.
26243         (execute_vrp): Compute backedges.
26245 2013-01-16  Richard Biener  <rguenther@suse.de>
26247         * doc/install.texi: Update CLooG and ISL requirements to
26248         0.18.0 and 0.11.1.
26250 2013-01-16  Christian Bruel  <christian.bruel@st.com>
26252         PR target/55301
26253         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
26254         (broken_move): Handle UNSPECV_SP_SWITCH_B.
26255         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
26257 2013-01-16  DJ Delorie  <dj@redhat.com>
26259         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
26260         (UNSPECV_SP_SWITCH_E): New.
26261         (sp_switch_1): Change to an unspec.
26262         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
26263         replace $r15.
26265 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
26267         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
26268         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
26269         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
26270         (expand_mem_thread_fence): Ditto.
26271         (expand_mem_signal_fence): Ditto.
26272         (expand_atomic_load): Ditto.
26273         (expand_atomic_store): Ditto.
26275 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
26277         PR rtl-optimization/55547
26278         PR rtl-optimization/53827
26279         PR debug/53671
26280         PR debug/49888
26281         * alias.c (memrefs_conflict_p): Set sizes to negative after
26282         AND adjustments.
26284 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26286         PR target/55940
26287         * function.c (thread_prologue_and_epilogue_insns): Always
26288         add crtl->drap_reg to set_up_by_prologue.set, even if
26289         stack_realign_drap is false.
26291 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
26293         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
26294         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
26295         *call): Fix indention.
26297 2013-01-15  Tom de Vries  <tom@codesourcery.com>
26299         PR target/55876
26300         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
26301         Update comment.
26303 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
26305         PR rtl-optimization/55153
26306         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
26308 2013-01-15  Martin Jambor  <mjambor@suse.cz>
26310         PR tree-optimization/55920
26311         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
26312         accesses as grp_to_be_debug_replaced.
26314 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26316         PR tree-optimization/55920
26317         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
26318         there is non-useless type conversion needed from debug rhs to lhs,
26319         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
26321 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
26322             Mikael Pettersson  <mikpe@it.uu.se>
26324         PR target/43961
26325         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
26326         Thumb.
26327         (ASM_OUTPUT_CASE_LABEL): Remove.
26328         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
26329         * final.c (shorten_branches): Update alignment of labels before
26330         jump tables if CASE_VECTOR_SHORTEN_MODE.
26332 2013-01-15  Richard Biener  <rguenther@suse.de>
26334         PR bootstrap/55961
26335         * system.h: Do not include gmp.h for building host tools.
26337 2013-01-15  Richard Biener  <rguenther@suse.de>
26339         PR middle-end/55882
26340         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
26341         account for bitpos when computing alignment.
26343 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
26345         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
26346         (ix86_target_macros_internal): Likewise.
26348         * config/i386/i386.c (m_CORE2I7): Removed.
26349         (m_CORE_HASWELL): New macro.
26350         (m_CORE_ALL): Likewise.
26351         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
26352         (initial_ix86_arch_features): Likewise.
26353         (processor_target_table): Initializations for Core avx2.
26354         (cpu_names): New names "core-avx2".
26355         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
26356         PROCESSOR_CORE_HASWELL.
26357         (ix86_issue_rate): New case.
26358         (ia32_multipass_dfa_lookahead): Likewise.
26359         (ix86_sched_init_global): Likewise.
26361         * config/i386/i386.h (TARGET_HASWELL): New macro.
26362         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
26363         (processor_type): New PROCESSOR_HASWELL.
26365 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
26367         PR tree-optimization/55955
26368         * tree-vect-loop.c (vectorizable_reduction): Give up early on
26369         *SHIFT_EXPR and *ROTATE_EXPR codes.
26371         PR tree-optimization/48766
26372         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
26373         -ftrapv disable -fwrapv.
26375 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26377         PR target/55974
26378         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
26379         etc. to 1 and not to __flash.
26380         Use LL suffix for __INT24_MAX__ with -mint8.
26381         Use ULL suffix for __UINT24_MAX__ with -mint8.
26383 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26385         * config/avr/avr-arch.h
26386         (struct base_arch_s): Use typedef avr_arch_t instead.
26387         (struct arch_info_s): Use typedef avr_arch_info_t instead.
26388         (struct mcu_type_s): Use typedef avr_mcu_t instead.
26389         * config/avr/avr.c: Same.
26390         * config/avr/avr-devices.c: Same.
26391         * config/avr/driver-avr.c: Same.
26392         * config/avr/gen-avr-mmcu-texi.c: Same.
26393         * config/avr/avr-mcus.def: Adjust comment.
26395 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
26397         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
26398         * config/aarch64/iterators.md (VALLDI): New.
26400 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
26401             Andi Kleen  <ak@linux.intel.com>
26403         PR target/55948
26404         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
26405         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
26406         memmodel flag.
26408 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26410         * config/avr/avr-stdint.h: Remove trailing blanks.
26411         * config/avr/avr-log.h: Same.
26412         * config/avr/avr-arch.h: Same.
26413         * config/avr/avr-devices.c: Same.
26414         * config/avr/avr-dimode.md: Same.
26415         * config/avr/predicates.md: Same.
26416         * config/avr/avr-c.c: Same.  And fix typo.
26418         * config/avr/avr-protos.h: Same.  And:
26419         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
26420         (init_cumulative_args): Rename to avr_init_cumulative_args.
26421         (expand_prologue): Rename to avr_expand_prologue.
26422         (expand_epilogue): Rename to avr_expand_epilogue.
26423         (adjust_insn_length): Rename to avr_adjust_insn_length.
26424         (notice_update_cc): Rename to avr_notice_update_cc.
26425         (final_prescan_insn): Rename to avr_final_prescan_insn.
26426         * config/avr/avr.c: Same.
26427         * config/avr/avr.h: Same.
26428         * config/avr/avr.md: Remove trailing blanks.
26429         (prologue): Use avr_expand_prologue.
26430         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
26432 2013-01-14  Richard Biener  <rguenther@suse.de>
26434         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
26435         verify_location, collect_subblocks): New functions.
26436         (verify_gimple_in_cfg): Verify that locations only reference
26437         BLOCKs in the functions BLOCK tree.
26439 2013-01-14  Richard Biener  <rguenther@suse.de>
26441         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
26442         PHI argument.
26443         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
26444         unshare reference.
26445         (insert_out_of_ssa_copy_on_edge): Likewise.
26446         (rewrite_close_phi_out_of_ssa): Likewise.
26447         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
26448         debug expressions.
26449         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
26450         propagated constants.
26451         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
26452         can not be shared.
26454 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
26456         * config/avr/avr-modes.def: Add GPL copyright notice.
26458 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
26460         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
26461         MEMMODEL_MASK to determine memory model.
26462         (atomic_store<mode>): Ditto from operands[2].
26463         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
26465 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
26467         PR fortran/55935
26468         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
26469         (fold_gimple_assign): Don't call unshare_expr here.
26470         (fold_ctor_reference): Call unshare_expr.
26472 2013-01-13  Terry Guo  <terry.guo@arm.com>
26474         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
26475         * doc/fragments.texi: Document MULTILIB_REUSE.
26476         * gcc.c (multilib_reuse): New internal spec.
26477         (set_multilib_dir): Also search multilib from multilib_reuse.
26478         * genmultilib (tmpmultilib3): Refactor code.
26479         (tmpmultilib4): Ditto.
26480         (multilib_reuse): New multilib argument.
26482 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
26484         * Makefile.in: Update copyright.
26486 2013-01-12  Tom de Vries  <tom@codesourcery.com>
26488         PR middle-end/55890
26489         * calls.c (expand_call): Check if arg_nr is valid.
26491 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
26493         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
26494         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
26495         documentation.  Add missing '__' in front of
26496         __builtin_ia32_packssdw256.
26498 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26500         PR target/55719
26501         * config/s390/s390.c (s390_preferred_reload_class): Do not return
26502         NO_REGS for larl operands.
26503         (s390_reload_larl_operand): Use s390_load_address instead of
26504         emit_move_insn.
26506 2013-01-11  Richard Biener  <rguenther@suse.de>
26508         * tree-cfg.c (verify_node_sharing_1): Split out from ...
26509         (verify_node_sharing): ... here.
26510         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
26512 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
26514         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
26515         Substitute TREECHECKING.
26516         * configure: Regenerate.
26517         * Makefile.in (TREECHECKING): New.
26519 2013-01-11  Richard Guenther  <rguenther@suse.de>
26521         PR tree-optimization/44061
26522         * tree-vrp.c (extract_range_basic): Compute zero as
26523         value-range for __builtin_constant_p of function parameters.
26525 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
26527         Update copyright years.
26529 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
26531         PR rtl-optimization/55672
26532         * lra-eliminations.c (mark_not_eliminable): Permit addition with
26533         const to be eliminable.
26535 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
26537         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
26538         * configure: Regenerate.
26540 2013-01-10  Richard Biener  <rguenther@suse.de>
26542         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
26544 2013-01-10  Richard Biener  <rguenther@suse.de>
26546         PR bootstrap/55792
26547         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
26548         locations for virtual PHI arguments.
26549         (rewrite_update_phi_arguments): Likewise.
26551 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
26553         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
26554         on to assembler.
26556 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
26558         PR tree-optimization/55921
26559         * tree-complex.c (expand_complex_asm): New function.
26560         (expand_complex_operations_1): Call it for GIMPLE_ASM.
26562 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26564         PR target/55718
26565         * config/s390/s390.c (s390_symref_operand_p)
26566         (s390_loadrelative_operand_p): Merge the two functions.
26567         (s390_check_qrst_address, print_operand_address): Add parameters
26568         to s390_loadrelative_operand_p invokation.
26569         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
26570         (s390_reload_larl_operand, s390_secondary_reload): Use
26571         s390_loadrelative_operand_p instead of s390_symref_operand_p.
26572         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
26574 2013-01-09  Mike Stump  <mikestump@comcast.net>
26576         * dse.c (record_store): Remove unnecessary assert.
26578 2013-01-09  Jan Hubicka  <jh@suse.cz>
26580         PR tree-optimization/55569
26581         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
26582         * cfgloop.h (scale_loop_profile): Likewise.
26584 2013-01-09  Jan Hubicka  <jh@suse.cz>
26586         PR lto/45375
26587         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
26588         functions.
26589         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
26591 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
26593         PR middle-end/55114
26594         * expr.h (maybe_emit_group_store): Declare.
26595         * expr.c (maybe_emit_group_store): New function.
26596         * builtins.c (expand_builtin_int_roundingfn): Call it.
26597         (expand_builtin_int_roundingfn_2): Likewise.
26599 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
26601         PR rtl-optimization/55829
26602         * lra-constraints.c (match_reload): Add code for absent output.
26603         (curr_insn_transform): Add code for reloads of matched inputs
26604         without output.
26606 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
26608         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
26609         attribute of movddup insn to DF.
26610         (*vec_interleave_lowv2df): Ditto.
26611         (vec_dupv2df): Ditto.
26613 2013-01-09  Jan Hubicka  <jh@suse.cz>
26615         PR tree-optimiation/55875
26616         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
26617         EVERY_ITERATION parameter.
26618         (number_of_iterations_exit): Check if exit is executed every iteration.
26619         (idx_infer_loop_bounds): Similarly here.
26620         (n_of_executions_at_most): Simplify
26621         to only test for cases where statement is dominated by the
26622         particular bound; handle correctly the "postdominance" test.
26623         (scev_probably_wraps_p): Use max loop iterations info
26624         as a global bound first.
26626 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
26627             Nick Clifton  <nickc@redhat.com>
26629         * config/v850/v850.md (cbranchsf4): New pattern.
26630         (cstoresf4): New pattern.
26631         (cbranchdf4): New pattern.
26632         (cstoredf4): New pattern.
26633         (movsicc): Disallow floating point comparisons.
26634         (cmpsf_le_insn): Fix order of operators.
26635         (cmpsf_lt_insn): Likewise.
26636         (cmpsf_eq_insn): Likewise.
26637         (cmpdf_le_insn): Likewise.
26638         (cmpdf_lt_insn): Likewise.
26639         (cmpdf_eq_insn): Likewise.
26640         (cmpsf_ge_insn): Use LE comparison.
26641         (cmpdf_ge_insn): Likewise.
26642         (cmpsf_gt_insn): Use LT comparison.
26643         (cmpdf_gt_insn): Likewise.
26644         (cmpsf_ne_insn): Delete pattern.
26645         (cmpdf_ne_insn): Delete pattern.
26646         * config/v850/v850.c (v850_gen_float_compare): Use
26647         gen_cmpdf_eq_insn for NE comparison.
26648         (v850_float_z_comparison_operator)
26649         (v850_float_nz_comparison_operator): Move from here ...
26650         * config/v850/predicates.md: ... to here.  Move GT and GE
26651         comparisons into v850_float_z_comparison_operator.
26652         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
26653         Delete prototype.
26654         (v850_float_nz_comparison_operator): Likewise.
26656 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26658         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
26659         with calls to gen_insvsi/gen_insvdi.
26661 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
26663         * config/i386/i386.c (initial_ix86_tune_features): Set up
26664         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
26666 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
26667             Jakub Jelinek  <jakub@redhat.com>
26669         PR tree-optimization/48189
26670         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
26671         If nitercst is 0, don't predict the exit edge.
26673 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
26675         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
26676         in asm_fprintf with reg_names.
26677         (aarch64_print_operand_address): Likewise.
26678         (aarch64_return_addr): Likewise.
26679         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
26681 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26683         * config/pa/pa.h (VAL_U6_BITS_P): Define.
26684         (INT_U6_BITS): Likewise.
26685         * config/pa/predicates.md (uint6_operand): New predicate.
26686         (shift5_operand, shift6_operand): Likewise.
26687         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
26688         arith32_operand.
26689         (lshrdi3): Use shift6_operand.
26690         (shrpsi4, shrpdi4): New insn patterns.
26691         (extzv): Delete expander.
26692         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
26693         predicates in unamed zero extract patterns.  Tighten common constraint.
26694         (extv): Delete expander.
26695         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
26696         predicates in unamed sign extract patterns.  Tighten common constraint.
26697         (insv): Delete expander.
26698         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
26699         predicates in unamed insert patterns.  Tighten common constraint.
26700         Change uint32_operand predicate to uint6_operand predicate in unamed
26701         DImode pattern to insert constant values of type 1...1xxxx.
26703 2013-01-04  Jan Hubicka  <jh@suse.cz>
26705         PR tree-optimization/55823
26706         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
26707         issue.
26709 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26710             Uros Bizjak  <ubizjak@gmail.com>
26712         PR rtl-optimization/55845
26713         * df-problems.c (can_move_insns_across): Stop scanning at
26714         volatile_insn_p source instruction or give up if
26715         across_from .. across_to range contains any volatile_insn_p
26716         instructions.
26718 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26720         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
26721         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
26722         Declare.
26723         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
26724         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
26726 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26728         PR fortran/55341
26729         * asan.c (asan_clear_shadow): New function.
26730         (asan_emit_stack_protection): Use it.
26732 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
26734         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
26735         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
26736         with tab instead of space.
26738 2013-01-08  Nick Clifton  <nickc@redhat.com>
26740         * config/rl78/rl78.c (rl78_expand_prologue): Always select
26741         register bank 0 at the start of an interrupt handler.
26742         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
26743         MDBH registers.
26745 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26747         * config/aarch64/aarch64-simd.md
26748         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
26749         (aarch64_simd_bsl): Likewise.
26750         (aarch64_vcond_internal<mode>): Likewise.
26751         (vcond<mode><mode>): Likewise.
26752         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
26753         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
26755 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
26757         * config/aarch64/aarch64-builtins.c
26758         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
26760 2013-01-08  Martin Jambor  <mjambor@suse.cz>
26762         PR debug/55579
26763         * tree-sra.c (analyze_access_subtree): Return true also after
26764         potentially creating a debug-only replacement.
26766 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26768         PR middle-end/55890
26769         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
26771         PR tree-optimization/54120
26772         * tree-vrp.c (range_fits_type_p): Don't allow
26773         src_precision < precision from signed vr to unsigned_p
26774         if vr->min or vr->max is negative.
26775         (simplify_float_conversion_using_ranges): Test can_float_p
26776         against CODE_FOR_nothing.
26778 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
26779             Richard Biener  <rguenther@suse.de>
26781         PR middle-end/55851
26782         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
26783         types instead of just INTEGER_TYPE types.
26785 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
26787         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
26788         TF_SIZE): Define.
26790 2013-01-07  Steve Ellcey  <sellcey@mips.com>
26792         PR target/42661
26793         * config/mips/mips.opt: Change mad to mmad to match documentation.
26795 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26797         PR target/55897
26798         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
26799         .progmemx.data now.
26801 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26803         PR target/55897
26804         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
26805         (avr_addrspace_t): Add .section_name field.
26806         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
26807         array size.
26808         (avr_addrspace): Same.  Initialize .section_name.  Remove last
26809         NULL entry.  Put __memx into .progmemx.data.
26810         (progmem_section_prefix): Remove.
26811         (avr_asm_init_sections): No need to initialize progmem_section.
26812         (avr_asm_named_section): Use avr_addrspace[].section_name to get
26813         section name prefix.
26814         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
26815         retrieve the progmem section.
26816         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
26817         boundary to run over avr_addrspace[].
26818         (avr_register_target_pragmas): Ditto.
26820 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
26822         * varasm.c (output_constant_def_contents): For asan_protect_global
26823         protected strings, adjust DECL_ALIGN if needed, before testing for
26824         anchored symbols.
26825         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
26826         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
26827         normal decls.
26828         (output_object_block): For asan protected decls, emit asan padding
26829         after their contents.
26830         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
26831         (asan_finish_file): Test it here instead.
26833 2013-01-07  Nick Clifton  <nickc@redhat.com>
26834             Matthias Klose  <doko@debian.org>
26835             Doug Kwan  <dougkwan@google.com>
26836             H.J. Lu  <hongjiu.lu@intel.com>
26838         PR driver/55470
26839         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
26841         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
26843         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
26845         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
26847         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
26849 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26851         PR target/54461
26852         * doc/install.texi (Cross-Compiler-Specific Options): Document
26853         --with-avrlibc.
26855 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
26857         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
26858         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
26859         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
26860         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
26861         vqmovun_high_s64): Fix source operand number and update copyright.
26863 2013-01-07  Richard Biener  <rguenther@suse.de>
26865         PR middle-end/55890
26866         * gimple.h (gimple_call_builtin_p): New overload.
26867         * gimple.c (validate_call): New function.
26868         (gimple_call_builtin_p): Likewise.
26869         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26870         Use gimple_call_builtin_p.
26871         (find_func_clobbers): Likewise.
26872         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
26873         (strlen_optimize_stmt): Likewise.
26875 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26877         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
26878         (vld1q_dup_*): Likewise.
26879         (vld1_*): Likewise.
26880         (vld1q_*): Likewise.
26881         (vld1_lane_*): Likewise.
26882         (vld1q_lane_*): Likewise.
26884 2013-01-07  Richard Biener  <rguenther@suse.de>
26886         * lto-streamer.h (LTO_minor_version): Bump to 2.
26888 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
26890         * config/aarch64/aarch64-protos.h
26891         (aarch64_const_double_zero_rtx_p): Rename to...
26892         (aarch64_float_const_zero_rtx_p): ...this.
26893         (aarch64_float_const_representable_p): New.
26894         (aarch64_output_simd_mov_immediate): Likewise.
26895         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
26896         move immediate case.
26897         * config/aarch64/aarch64.c
26898         (aarch64_const_double_zero_rtx_p): Rename to...
26899         (aarch64_float_const_zero_rtx_p): ...this.
26900         (aarch64_print_operand): Allow printing of new constants.
26901         (aarch64_valid_floating_const): New.
26902         (aarch64_legitimate_constant_p): Check for valid floating-point
26903         constants.
26904         (aarch64_simd_valid_immediate): Likewise.
26905         (aarch64_vect_float_const_representable_p): New.
26906         (aarch64_float_const_representable_p): Likewise.
26907         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
26908         (aarch64_output_simd_mov_immediate): New.
26909         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
26910         (*movdf_aarch64): Likewise.
26911         * config/aarch64/constraints.md (Ufc): New.
26912         (Y): call aarch64_float_const_zero_rtx.
26913         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
26915 2013-01-07  Richard Biener  <rguenther@suse.de>
26917         PR tree-optimization/55888
26918         PR tree-optimization/55862
26919         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
26920         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
26921         not if it is contained therein.
26923 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26925         * config/avr/t-avr: Typo.
26927 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
26929         PR55243
26930         * config/avr/t-avr: Don't automatically rebuild
26931         $(srcdir)/config/avr/t-multilib
26932         $(srcdir)/config/avr/avr-tables.opt
26933         $(srcdir)/doc/avr-mmcu.texi
26934         (avr-mcus): New phony target to build them on request.
26935         (s-avr-mlib, s-avr-mmcu-texi): Remove.
26936         * avr/avr-mcus.def: Adjust comments.
26938 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
26940         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
26942 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26944         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
26946 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
26948         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
26950 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
26952         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
26953         to generate profiling.
26954         * config/rs6000/aix64.h (LIB_SPEC): Same.
26956 2013-01-04  Andrew Pinski  <apinski@cavium.com>
26958         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
26959         New function.
26960         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
26962 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
26964         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
26965         unconditionally.
26966         (ix86_expand_move): Ditto.
26967         (ix86_zero_extend_to_Pmode): Ditto.
26968         (ix86_expand_call): Ditto.
26969         (ix86_expand_special_args_builtin): Ditto.
26970         (ix86_expand_builtin): Ditto.
26972 2013-01-04  Richard Biener  <rguenther@suse.de>
26974         PR tree-optimization/55862
26975         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
26976         translating them through PHI nodes.
26978 2013-01-04  Martin Jambor  <mjambor@suse.cz>
26980         PR tree-optimization/55755
26981         * tree-sra.c (sra_modify_assign): Do not check that an access has no
26982         children when trying to avoid producing a VIEW_CONVERT_EXPR.
26984 2013-01-04  Marek Polacek  <polacek@redhat.com>
26986         PR middle-end/55859
26987         * opts.c (default_options_optimization): Clarify error message.
26989 2013-01-04  Richard Biener  <rguenther@suse.de>
26991         PR middle-end/55863
26992         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
26993         reassociation.
26995 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26997         PR target/53789
26998         * config/pa/pa.md (movsi): Revert previous change.
26999         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
27000         references.
27002 2013-01-03  Richard Henderson  <rth@redhat.com>
27004         * config/i386/i386.c (ix86_expand_move): Always assign to op1
27005         after eliminating TLS symbols.
27007 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
27009         PR bootstrap/50167
27010         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
27011         * graphite-poly.c (debug_gmp_value): Likewise.
27013 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
27015         PR target/55712
27016         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
27017         selected code model, define __code_mode_small__, __code_model_medium__,
27018         __code_model_large__, __code_model_32__ or __code_model_kernel__.
27019         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
27020         xchg temporary register with %k.  Declare temporary register as
27021         early clobbered.
27022         [__x86_64__]: For medium and large code models, preserve %rbx register.
27024 2013-01-03  Richard Biener  <rguenther@suse.de>
27026         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
27027         (dump_subscript): Adjust.
27028         (finalize_ddr_dependent): Do not dump redundant info.
27029         (analyze_siv_subscript): Adjust.
27030         (subscript_dependence_tester): Likewise.
27031         (compute_affine_dependence): Likewise.
27033 2013-01-03  Richard Biener  <rguenther@suse.de>
27035         Revert
27036         2013-01-03  Richard Biener  <rguenther@suse.de>
27038         PR tree-optimization/55857
27039         * tree-vect-stmts.c (vectorizable_load): Do not setup
27040         re-alignment for invariant loads.
27042         2013-01-02  Richard Biener  <rguenther@suse.de>
27044         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27045         invariant load do not generate a vector load from the scalar location.
27047 2013-01-03  Richard Biener  <rguenther@suse.de>
27049         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
27050         for not vectorizing.
27051         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
27052         not build INDIRECT_REFs, call get_name once only.
27053         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
27054         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
27056 2013-01-03  Richard Biener  <rguenther@suse.de>
27058         PR tree-optimization/55857
27059         * tree-vect-stmts.c (vectorizable_load): Do not setup
27060         re-alignment for invariant loads.
27062 2013-01-03  Richard Biener  <rguenther@suse.de>
27064         PR lto/55848
27065         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
27066         prefer a built-in decl.
27068 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27070         * gcc.c (process_command): Update copyright notice dates.
27071         * gcov.c (print_version): Likewise.
27072         * gcov-dump.c (print_version): Likewise.
27074         PR rtl-optimization/55838
27075         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
27076         iv0.step, iv1.step and step.
27078 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27079             Marc Glisse  <marc.glisse@inria.fr>
27081         PR tree-optimization/55832
27082         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
27083         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
27084         integer_{one,zero}_node.
27086 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
27088         PR debug/54402
27089         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
27090         * var-tracking.c (reverse_op): Don't add reverse ops to
27091         VALUEs that have already
27092         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
27094 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
27096         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
27098 2013-01-02  Teresa Johnson  <tejohnson@google.com>
27100         * dumpfile.c (dump_loc): Print filename with location.
27101         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
27102         new location_t parameter to emit complete unroll message with
27103         new dump framework.
27104         (canonicalize_loop_induction_variables): Compute loops location
27105         and pass to try_unroll_loop_completely.
27106         * loop-unroll.c (report_unroll_peel): New function.
27107         (peel_loops_completely): Use new dump format with location
27108         for main dumpfile message, and invoke report_unroll_peel on success.
27109         (decide_unrolling_and_peeling): Ditto.
27110         (decide_peel_once_rolling): Remove old dumpfile message subsumed
27111         by report_unroll_peel.
27112         (decide_peel_completely): Ditto.
27113         (decide_unroll_constant_iterations): Ditto.
27114         (decide_unroll_runtime_iterations): Ditto.
27115         (decide_peel_simple): Ditto.
27116         (decide_unroll_stupid): Ditto.
27117         * cfgloop.c (get_loop_location): New function.
27118         * cfgloop.h (get_loop_location): Declare.
27120 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27122         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
27123         NULL.
27125 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27127         PR middle-end/55198
27128         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
27129         BLKmode objects when EXPAND_MEMORY is specified.
27131 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
27133         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
27134         in loop predicate.
27135         (fold_builtin_cpu): Do not share cpu model decls across statements.
27137 2013-01-02  Jason Merrill  <jason@redhat.com>
27139         PR c++/55804
27140         * tree.c (build_array_type_1): Revert earlier change.
27142 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
27144         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
27145         "cortex-a57".
27146         * config/aarch64/aarch64-tune.md: Re-generate.
27148 2013-01-02  Richard Biener  <rguenther@suse.de>
27150         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
27151         invariant load do not generate a vector load from the scalar location.
27153 2013-01-02  Richard Biener  <rguenther@suse.de>
27155         PR bootstrap/55784
27156         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
27157         * configure: Regenerate.
27159 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27161         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
27162         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
27163         (expand_builtin_int_roundingfn_2): Keep the original target around
27164         for the fallback case.
27166 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
27168         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
27169         to be clear for sign changes.
27171 2013-01-01  Jan Hubicka  <jh@suse.cz>
27173         * ipa-inline-analysis.c: Fix formatting.
27175 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
27177         PR tree-optimization/55831
27178         * tree-vect-loop.c (get_initial_def_for_induction): Use
27179         gsi_after_labels instead of gsi_start_bb.
27181 Copyright (C) 2013 Free Software Foundation, Inc.
27183 Copying and distribution of this file, with or without modification,
27184 are permitted in any medium without royalty provided the copyright
27185 notice and this notice are preserved.