* gcc.c (do_spec_1): Do not insert a space after a %* substitution
[official-gcc.git] / gcc / ChangeLog
blobf5409b761514edfe8cb89951e87e524452b0f572
1 2013-11-06  Nick Clifton  <nickc@redhat.com>
3         * gcc.c (do_spec_1): Do not insert a space after a %* substitution
4         unless it is the last part of a spec substring.
5         * doc/invoke.texi (Spec Files): Document space insertion
6         behaviour of %*.
8 2013-11-06  Christian Bruel  <christian.bruel@st.com>
10         * gcc/config/sh/sh-mem.cc (sh_expand_cmpnstr, sh_expand_cmpstr):
11         Factorize probabilities, Use adjust_address instead of
12         adjust_automodify_address when possible. Enable for optimize.
13         (sh_expand_strlen): New function.
14         * gcc/config/sh/sh-protos.h (sh_expand_strlen): Declare.
15         * gcc/config/sh/sh.md (strlensi): New pattern.
16         (UNSPEC_BUILTIN_STRLEN): Define.
18 2013-11-06  Jakub Jelinek  <jakub@redhat.com>
20         PR middle-end/58970
21         * expr.c (get_bit_range): Handle *offset == NULL_TREE.
22         (expand_assignment): If *bitpos is negative, set *offset
23         and adjust *bitpos, so that it is not negative.
25 2013-11-06  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
27         * config/i386/bdver3.md : Added two additional decoder units 
28         to support issue rate of 4 and remodeled vector unit.
29         * config/i386/i386.c (ix86_issue_rate): Issue rate for BD
30         architectures is set to 4.
31         * config/i386/i386.c (ia32_multipass_dfa_lookahead): DFA
32         lookahead is set to 4 for BD architectures.
34 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
36         * config/rs6000/rs6000.c (rs6000_option_override_internal):
37         Remove restriction against use of VSX instructions when generating
38         code for little endian mode.
40 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
42         * config/rs6000/altivec.md (mulv4si3): Ensure we generate vmulouh
43         for both big and little endian.
44         (mulv8hi3): Swap input operands for merge high and merge low
45         instructions for little endian.
47 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
49         * config/rs6000/altivec.md (vec_widen_umult_even_v16qi): Change
50         define_insn to define_expand that uses even patterns for big
51         endian and odd patterns for little endian.
52         (vec_widen_smult_even_v16qi): Likewise.
53         (vec_widen_umult_even_v8hi): Likewise.
54         (vec_widen_smult_even_v8hi): Likewise.
55         (vec_widen_umult_odd_v16qi): Likewise.
56         (vec_widen_smult_odd_v16qi): Likewise.
57         (vec_widen_umult_odd_v8hi): Likewise.
58         (vec_widen_smult_odd_v8hi): Likewise.
59         (altivec_vmuleub): New define_insn.
60         (altivec_vmuloub): Likewise.
61         (altivec_vmulesb): Likewise.
62         (altivec_vmulosb): Likewise.
63         (altivec_vmuleuh): Likewise.
64         (altivec_vmulouh): Likewise.
65         (altivec_vmulesh): Likewise.
66         (altivec_vmulosh): Likewise.
68 2013-11-05  Mike Stump  <mikestump@comcast.net>
70         * Makefile.in (mostlyclean): Remove c-family objects.
72 2013-11-05  Ian Lance Taylor  <iant@google.com>
74         * config/i386/sync.md (atomic_compare_and_swap<dwi>_doubleword):
75         If possible, add .cfi directives to record change to bx.
76         * config/i386/i386.c (ix86_emit_cfi): New function.
77         * config/i386/i386-protos.h (ix86_emit_cfi): Declare.
79 2013-11-05  Steven Bosscher  <steven@gcc.gnu.org>
82         * rtlanal.c (tablejump_p): Expect a JUMP_TABLE_DATA to always follow
83         immediately after a label for a tablejump pattern.
85         * config/arm/arm.c (is_jump_table): Remove.
86         (create_fix_barrier): Use tablejump_p instead.
87         (arm_reorg): Likewise.
88         (thumb1_output_casesi): Expect JUMP_TABLE_DATA to always be NEXT_INSN.
89         (thumb2_output_casesi): Likewise.
90         * config/aarch64/aarch64.c (aarch64_output_casesi): Likewise.
91         * config/sh/sh.md (casesi_worker_1, casesi_worker_2,
92         casesi_shift_media, casesi_load_media): Likewise.
93         * config/iq2000/iq2000.md: Likewise (in anonymous define_insn).
94         * config/microblaze/microblaze.md: Likewise.
96 2013-11-05  Tobias Burnus  <burnus@net-b.de>
98         * doc/invoke.texi (-Wdate-time): Document.
100 2013-11-05  Richard Sandiford  <rdsandiford@googlemail.com>
102         * double-int.c (lshift_double, rshift_double): Remove
103         SHIFT_COUNT_TRUNCATED handling.
105 2013-11-05  Jeff Law  <law@redhat.com>
107         * Makefile.in (OBJS): Add gimple-ssa-isolate-paths.o
108         * common.opt (-fisolate-erroneous-paths): Add option and
109         documentation.
110         * gimple-ssa-isolate-paths.c: New file.
111         * gimple.c (check_loadstore): New function.
112         (infer_nonnull_range): Moved into gimple.c from tree-vrp.c
113         Verify OP is in the argument list and the argument corresponding
114         to OP is a pointer type.  Use operand_equal_p rather than
115         pointer equality when testing if OP is on the nonnull list.
116         Use check_loadstore rather than count_ptr_derefs.  Handle
117         GIMPLE_RETURN statements.
118         * tree-vrp.c (infer_nonnull_range): Remove.
119         * gimple.h (infer_nonnull_range): Declare.
120         * opts.c (default_options_table): Add OPT_fisolate_erroneous_paths.
121         * passes.def: Add pass_isolate_erroneous_paths.
122         * timevar.def (TV_ISOLATE_ERRONEOUS_PATHS): New timevar.
123         * tree-pass.h (make_pass_isolate_erroneous_paths): Declare.
124         * tree-ssa.c (struct count_ptr_d): Remove.
125         (count_ptr_derefs, count_uses_and_derefs): Remove.
126         * tree-ssa.h (count_uses_and_derefs): Remove.
128 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
130         PR rtl-optimization/58997
131         * loop-iv.c (iv_subreg): For IV_UNKNOWN_EXTEND, expect
132         get_iv_value to be in iv->mode rather than iv->extend_mode.
133         (iv_extend): Likewise.  Otherwise, if iv->extend != extend,
134         use lowpart_subreg on get_iv_value before calling simplify_gen_unary.
135         * loop-unswitch.c (may_unswitch_on): Make sure op[i] is in the right
136         mode.
138 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
140         * gimple.h: Move some prototypes to gimple-expr.h and add to include
141         list.
142         (extract_ops_from_tree, gimple_call_addr_fndecl, is_gimple_reg_type):
143         Move to gimple-expr.h.
144         * gimple-expr.h: New file.  Relocate some prototypes from gimple.h.
145         (types_compatible_p, is_gimple_reg_type, is_gimple_variable,
146         is_gimple_id, virtual_operand_p, is_gimple_addressable,
147         is_gimple_constant, extract_ops_from_tree, gimple_call_addr_fndecl):
148         Relocate here.
149         * gimple.c (extract_ops_from_tree_1, gimple_cond_get_ops_from_tree,
150         gimple_set_body, gimple_body, gimple_has_body_p, is_gimple_lvalue,
151         is_gimple_condexpr, is_gimple_addressable, is_gimple_constant,
152         is_gimple_address, is_gimple_invariant_address,
153         is_gimple_ip_invariant_address, is_gimple_min_invariant,
154         is_gimple_ip_invariant, is_gimple_variable, is_gimple_id,
155         virtual_operand_p, is_gimple_reg, is_gimple_val, is_gimple_asm_val,
156         is_gimple_min_lval, is_gimple_call_addr, is_gimple_mem_ref_addr,
157         gimple_decl_printable_name, useless_type_conversion_p,
158         types_compatible_p, gimple_can_coalesce_p, copy_var_decl): Move to 
159         gimple-expr.[ch].
160         * gimple-expr.c: New File.
161         (useless_type_conversion_p, gimple_set_body, gimple_body,
162         gimple_has_body_p, gimple_decl_printable_name, copy_var_decl,
163         gimple_can_coalesce_p, extract_ops_from_tree_1, 
164         gimple_cond_get_ops_from_tree, is_gimple_lvalue, is_gimple_condexpr,
165         is_gimple_address, is_gimple_invariant_address,
166         is_gimple_ip_invariant_address, is_gimple_min_invariant,
167         is_gimple_ip_invariant, is_gimple_reg, is_gimple_val,
168         is_gimple_asm_val, is_gimple_min_lval, is_gimple_call_addr,
169         is_gimple_mem_ref_addr): Relocate here.
170         * Makefile.in (OBJS): Add gimple-expr.o.
172 2013-11-05  David Malcolm  <dmalcolm@redhat.com>
174         * gengtype-parse.c (struct_field_seq): Support empty structs.
176 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
178         * config/i386/t-rtems (MULTILIB_MATCHES): Fix option typos.
180 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
182         * config/i386/i386-c.c (ix86_target_macros): Define _SOFT_FLOAT
183         for !TARGET_80387.
184         * config/i386/rtemself.h (TARGET_OS_CPP_BUILTINS): Do not define
185         _SOFT_FLOAT here.
186         (LONG_DOUBLE_TYPE_SIZE): New define.
187         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Ditto.
189 2013-11-05  Paolo Carlini  <paolo.carlini@oracle.com>
191         PR c++/58724
192         * doc/extend.texi [visibility ("visibility_type")]: Add example
193         about visibility attribute on namespace declaration.
195 2013-11-05  Richard Biener  <rguenther@suse.de>
197         PR ipa/58492
198         * passes.def (all_passes): Start with pass_fixup_cfg again.
200 2013-11-05  Richard Biener  <rguenther@suse.de>
202         PR tree-optimization/58955
203         * tree-loop-distribution.c (pg_add_dependence_edges): Fix
204         edge direction.
206 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
208         * config/rs6000/vector.md (vec_pack_sfix_trunc_v2df): Adjust for
209         little endian.
210         (vec_pack_ufix_trunc_v2df): Likewise.
212 2013-11-05  H.J. Lu  <hongjiu.lu@intel.com>
214         PR middle-end/58981
215         * doc/md.texi (@code{movmem@var{m}}): Specify Pmode as mode of
216         pattern, instead of word_mode.
218         * expr.c (emit_block_move_via_movmem): Don't use mode wider than
219         Pmode for size.
220         (set_storage_via_setmem): Likewise.
222 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
224         * tree-outof-ssa.c (queue_phi_copy_p): Combine phi_ssa_name_p from
225         gimple.h and the rest of the condition in eliminate_build.
226         (eliminate_build): Call new routine.
227         * gimple.h (phi_ssa_name_p): Delete.
229 2013-11-05  Trevor Saunders  <tsaunders@mozilla.com>
231         * vec.c (vec_prefix::calculate_allocation): Don't try to handle the
232         case of no prefix and reserving zero slots, because when that's the
233         case we'll never get here.
234         * vec.h (va_heap::reserve): Don't try and handle
235         vec_prefix::calculate_allocation returning zero because that should
236         never happen.
238 2013-11-05  Richard Biener  <rguenther@suse.de>
240         PR middle-end/58941
241         * tree-dfa.c (get_ref_base_and_extent): Merge common code
242         in MEM_REF and TARGET_MEM_REF handling.  Make sure to
243         process trailing array detection before diving into the
244         view-converted object (and possibly apply some extra offset).
246 2013-11-05  Joseph Myers  <joseph@codesourcery.com>
248         * config/i386/i386.c (ix86_float_exceptions_rounding_supported_p):
249         New function.
250         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Define.
252 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
254         PR tree-optimization/58958
255         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Use
256         get_addr_base_and_unit_offset instead of get_ref_base_and_extent.
258 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
260         * tree-ssa-alias.h (ranges_overlap_p): Handle negative offsets.
261         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
263 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
265         PR tree-optimization/58984
266         * ipa-prop.c (ipa_load_from_parm_agg_1): Add SIZE_P argument,
267         set *SIZE_P if non-NULL on success.
268         (ipa_load_from_parm_agg, ipa_analyze_indirect_call_uses): Adjust
269         callers.
270         (ipcp_transform_function): Likewise.  Punt if size of access
271         is different from TYPE_SIZE on v->value's type.
273 2013-11-05  Tobias Burnus  <burnus@net-b.de>
275         * doc/invoke.texi (-fopenmp-simd): Document new option.
276         * gimplify.c (gimplify_body): Accept -fopenmp-simd.
277         * omp-low.c (execute_expand_omp, execute_lower_omp): Ditto.
278         * tree.c (attribute_value_equal): Ditto.
280 2013-11-04  Wei Mi  <wmi@google.com>
282         * sched-rgn.c (add_branch_dependences): Keep insns in
283         a SCHED_GROUP at the end of BB to remain their location.
285 2013-11-04  Wei Mi  <wmi@google.com>
287         * gcc/config/i386/i386.c (memory_address_length): Extract a part
288         of code to rip_relative_addr_p.
289         (rip_relative_addr_p): New Function.
290         (ix86_macro_fusion_p): Ditto.
291         (ix86_macro_fusion_pair_p): Ditto.
292         * gcc/config/i386/i386.h: Add new tune features about macro-fusion.
293         * gcc/config/i386/x86-tune.def (DEF_TUNE): Ditto.
294         * gcc/doc/tm.texi: Generated.
295         * gcc/doc/tm.texi.in: Ditto.
296         * gcc/haifa-sched.c (try_group_insn): New Function.
297         (group_insns_for_macro_fusion): Ditto.
298         (sched_init): Call group_insns_for_macro_fusion.
299         * gcc/target.def: Add two hooks: macro_fusion_p and
300         macro_fusion_pair_p.
302 2013-11-04  Kostya Serebryany  <kcc@google.com>
304         Update to match the changed asan API.
305         * asan.c (asan_function_start): New function.
306         (asan_emit_stack_protection): Update the string stored in the
307         stack red zone to match new API.  Store the PC of the current
308         function in the red zone.
309         (asan_global_struct): Update the __asan_global definition to match
310         the new API.
311         (asan_add_global): Ditto.
312         * asan.h (asan_function_start): New prototype.
313         * final.c (final_start_function): Call asan_function_start.
314         * sanitizer.def (BUILT_IN_ASAN_INIT): Rename __asan_init_v1
315         to __asan_init_v3.
317 2013-11-04  Wei Mi  <wmi@google.com>
319         * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Separate
320         PROCESSOR_COREI7_AVX out from PROCESSOR_COREI7.
321         * gcc/config/i386/i386.c (ix86_option_override_internal): Ditto.
322         (ix86_issue_rate): Ditto.
323         (ix86_adjust_cost): Ditto.
324         (ia32_multipass_dfa_lookahead): Ditto.
325         (ix86_sched_init_global): Ditto.
326         (get_builtin_code_for_version): Ditto.
327         * gcc/config/i386/i386.h (enum target_cpu_default): Ditto.
328         (enum processor_type): Ditto.
329         * gcc/config/i386/x86-tune.def (DEF_TUNE): Ditto.
331 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
333         PR rtl-optimization/58967
334         * config/rs6000/rs6000.c (legitimate_lo_sum_address_p): Remove
335         !lra_in_progress for mode sizes bigger word.
337 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
339         * config/rs6000/altivec.md (vec_widen_umult_hi_v16qi): Swap
340         arguments to merge instruction for little endian.
341         (vec_widen_umult_lo_v16qi): Likewise.
342         (vec_widen_smult_hi_v16qi): Likewise.
343         (vec_widen_smult_lo_v16qi): Likewise.
344         (vec_widen_umult_hi_v8hi): Likewise.
345         (vec_widen_umult_lo_v8hi): Likewise.
346         (vec_widen_smult_hi_v8hi): Likewise.
347         (vec_widen_smult_lo_v8hi): Likewise.
349 2013-11-04  Ian Lance Taylor  <iant@google.com>
351         * builtins.def (ATTR_NOTHROWCALL_LEAF_LIST): Define.
352         * sync-builtins.def: Use ATTR_NOTHROWCALL_LEAF_LIST for all sync
353         builtins that take pointers.
354         * lto-opts.c (lto_write_options): Write -fnon-call-exceptions if set.
355         * lto-wrapper.c (merge_and_complain): Collect OPT_fnon_call_exceptions.
356         (run_gcc): Pass -fnon-call-exceptions.
358 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
360         * optabs.c (expand_vec_perm): Revert one incorrect line from
361         2013-10-31 change.
363         PR tree-optimization/58978
364         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): Don't modify
365         use_stmt by single_imm_use directly.  Only call single_imm_use
366         on SSA_NAMEs.
368 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
370         PR rtl-optimization/58968
371         * lra-spills.c (return_regno_p): New function.
372         (lra_final_code_change): Use it.
374 2013-11-04  Joseph Myers  <joseph@codesourcery.com>
376         * doc/cpp.texi (__GCC_IEC_559, __GCC_IEC_559_COMPLEX): Document macros.
377         * target.def (float_exceptions_rounding_supported_p): New hook.
378         * targhooks.c (default_float_exceptions_rounding_supported_p): New
379         function.
380         * targhooks.h (default_float_exceptions_rounding_supported_p): Declare.
381         * doc/tm.texi.in (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P):
382         New @hook.
383         * doc/tm.texi: Regenerate.
384         * config.gcc (powerpc*-*-linux*): Set extra_objs.
385         * config/rs6000/rs6000-linux.c: New file.
386         * config/rs6000/rs6000-protos.h
387         (rs6000_linux_float_exceptions_rounding_supported_p): Declare.
388         * config/rs6000/linux.h
389         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): New macro.
390         * config/rs6000/linux64.h
391         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Likewise.
392         * config/rs6000/t-linux (rs6000-linux.o): New rule.
393         * config/rs6000/t-linux64 (rs6000-linux.o): Likewise.
395 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
397         * config/rs6000/vsx.md (*vsx_le_perm_store_<mode> for VSX_D):
398         Replace the define_insn_and_split with a define_insn and two
399         define_splits, with the split after reload re-permuting the source
400         register to its original value.
401         (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
402         (*vsx_le_perm_store_v8hi): Likewise.
403         (*vsx_le_perm_store_v16qi): Likewise.
405 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
407         * config/rs6000/vector.md (vec_pack_trunc_v2df):  Adjust for
408         little endian.
410 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
412         PR tree-optimization/58946
413         * tree-ssa-reassoc.c (maybe_optimize_range_tests): Update all
414         bbs with bbinfo[idx].op != NULL before all blocks with
415         bbinfo[idx].op == NULL.
417 2013-11-04  Richard Sandiford  <rdsandiford@googlemail.com>
419         * config/avr/avr-log.c (avr_double_int_pop_digit): Delete.
420         (avr_dump_double_int_hex): Likewise.
421         (avr_log_vadump): Remove %D and %X handling.
422         * config/avr/avr.c (avr_double_int_push_digit): Delete.
423         (avr_map_op_t): Change map from double_int to unsigned int.
424         (avr_map_op): Update accordingly.
425         (avr_map, avr_map_metric, avr_has_nibble_0xf, avr_map_decompose)
426         (avr_move_bits, avr_out_insert_bits, avr_fold_builtin): Operate on
427         unsigned ints rather than double_ints.
429 2013-11-03  Marek Polacek  <polacek@redhat.com>
431         Implement -fsanitize=vla-bound.
432         * opts.c (common_handle_option): Handle vla-bound.
433         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE): Define.
434         * flag-types.h (enum sanitize_code): Add SANITIZE_VLA.
435         * asan.c (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR.
437 2013-11-02  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
439         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Adjust for
440         little endian.
442 2013-11-02  Uros Bizjak  <ubizjak@gmail.com>
444         * config/i386/constraints.md (Ts, Tv): New address constrains.
445         * config/i386/i386.md (*lea<mode>, *<mode>_<bndcheck>): Use Ts
446         constraint for address_no_seg_operand.
447         * config/i386/sse.md (*avx512pf_gatherpf<mode>_mask)
448         (*avx512pf_gatherpf<mode>, *avx512pf_scatterpf<mode>_mask)
449         (*avx512pf_scatterpf<mode>, *avx2_gathersi<mode>)
450         (*avx2_gathersi<mode>_2, *avx2_gatherdi<mode>, *avx2_gatherdi<mode>_2)
451         (*avx2_gatherdi<mode>_3, *avx2_gatherdi<mode>_4)
452         (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2)
453         (*avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2)
454         (*avx512f_scattersi<mode> *avx512f_scatterdi<mode>): Use Tv
455         constraint for vsib_address_operand.
457 2013-11-02  Steven Bosscher  <steven@gcc.gnu.org>
459         * gcse.c (pre_delete): Remove references to regmove from comments.
460         * recog.c: (validate_replace_rtx_1): Likewise.
461         * config/rl78/rl78.c: Likewise.
462         * config/v850/v850.h: Likewise, and remove unused ENABLE_REGMOVE_PASS.
463         * common/config/m32r/m32r-common.c: Don't manipulate OPT_fregmove.
464         * common/config/mmix/mmix-common.c: Likewise.
466 2013-11-01  Trevor Saunders  <tsaunders@mozilla.com>
468         * function.c (reorder_blocks): Convert block_stack to a stack_vec.
469         * gimplify.c (gimplify_compound_lval): Likewise.
470         * graphite-clast-to-gimple.c (gloog): Likewise.
471         * graphite-dependences.c (loop_is_parallel_p): Likewise.
472         * graphite-scop-detection.c (scopdet_basic_block_info): Likewise.
473         (limit_scop); Likewise.
474         (build_scops): Likewise.
475         (dot_scop): Likewise.
476         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
477         (build_scop_drs): Likewise.
478         (insert_stmts): Likewise.
479         (insert_out_of_ssa_copy): Likewise.
480         (remove_phi): Likewise.
481         (rewrite_commutative_reductions_out_of_ssa_close_phi): Likewise.
482         * hw-doloop.c (discover_loop): Likewise.
483         * tree-call-cdce.c (shrink_wrap_one_built_in_call): Likewise.
484         * tree-dfa.c (dump_enumerated_decls): Likewise.
485         * tree-if-conv.c (if_convertable_loop_p): Likewise.
486         * tree-inline.c (tree_function_versioning): Likewise.
487         * tree-loop-distribution.c (build_rdg): Likewise.
488         (rdg_flag_vertex_and_dependent): Likewise.
489         (distribute_loop): Likewise.
490         * tree-parloops.c (loop_parallel_p): Likewise.
491         (eliminate_local_variables): Likewise.
492         (separate_decls_in_region): Likewise.
493         * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
494         * tree-ssa-phiopt.c (cond_if_else_store_replacement): Likewise.
495         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
496         * tree-vect-loop.c (vect_analyze_scaler_cycles_1): Likewise.
497         * tree-vect-patterns.c (vect_pattern_recog): Likewise.
498         * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
499         (vectorizable_condition): Likewise.
501 2013-11-01  Uros Bizjak  <ubizjak@gmail.com>
503         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): Always define as 0/1.
504         * configure: Regenerate.
505         * config/i386/i386.md (*movdi_internal): Change
506         HAVE_AS_IX86_INTERUNIT_MOVQ to runtime check.
507         (*movdf_internal): Ditto.
508         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
509         * config/i386/sse.md (vec_concatv2di): Output interunit movq
510         for HAVE_AS_IX86_INTERUNIT_MOVQ targets.
512 2013-10-31  Robert Suchanek  <Robert.Suchanek@imgtec.com>
514         * lra-spills.c (assign_spill_hard_regs): Remove statement terminator
515         after comment.
517 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
519         Automated part of renaming of symtab_node_base to symtab_node.
521         Patch autogenerated by rename_symtab.py from
522         https://github.com/davidmalcolm/gcc-refactoring-scripts
523         revision 58bb219cc090b2f4516a9297d868c245495ee622
524         with ChangeLog entry fixed up by hand.
526         * cgraph.c (x_cgraph_nodes_queue): Rename symtab_node_base to
527         symtab_node.
528         (cgraph_node_for_asm): Likewise.
529         * cgraph.h (symtab_node_base): Likewise.
530         (cgraph_node): Likewise.
531         (varpool_node): Likewise.
532         (is_a_helper <cgraph_node>::test): Likewise.
533         (is_a_helper <varpool_node>::test): Likewise.
534         (symtab_nodes): Likewise.
535         (symtab_register_node): Likewise.
536         (symtab_unregister_node): Likewise.
537         (symtab_remove_node): Likewise.
538         (symtab_get_node): Likewise.
539         (symtab_node_for_asm): Likewise.
540         (symtab_node_asm_name): Likewise.
541         (symtab_node_name): Likewise.
542         (symtab_insert_node_to_hashtable): Likewise.
543         (symtab_add_to_same_comdat_group): Likewise.
544         (symtab_dissolve_same_comdat_group_list): Likewise.
545         (dump_symtab_node): Likewise.
546         (debug_symtab_node): Likewise.
547         (dump_symtab_base): Likewise.
548         (verify_symtab_node): Likewise.
549         (verify_symtab_base): Likewise.
550         (symtab_used_from_object_file_p): Likewise.
551         (symtab_alias_ultimate_target): Likewise.
552         (symtab_resolve_alias): Likewise.
553         (fixup_same_cpp_alias_visibility): Likewise.
554         (symtab_for_node_and_aliases): Likewise.
555         (symtab_nonoverwritable_alias): Likewise.
556         (availability symtab_node_availability): Likewise.
557         (symtab_semantically_equivalent_p): Likewise.
558         (fixup_same_cpp_alias_visibility): Likewise.
559         (symtab_prevail_in_asm_name_hash): Likewise.
560         (cgraph): Likewise.
561         (varpool): Likewise.
562         (varpool_first_variable): Likewise.
563         (varpool_next_variable): Likewise.
564         (varpool_first_static_initializer): Likewise.
565         (varpool_next_static_initializer): Likewise.
566         (varpool_first_defined_variable): Likewise.
567         (varpool_next_defined_variable): Likewise.
568         (cgraph_first_defined_function): Likewise.
569         (cgraph_next_defined_function): Likewise.
570         (cgraph_first_function): Likewise.
571         (cgraph_next_function): Likewise.
572         (cgraph_first_function_with_gimple_body): Likewise.
573         (cgraph_next_function_with_gimple_body): Likewise.
574         (symtab_alias_target): Likewise.
575         (symtab_real_symbol_p): Likewise.
576         (symtab_can_be_discarded): Likewise.
577         * cgraphbuild.c (mark_address): Likewise.
578         (mark_load): Likewise.
579         (mark_store): Likewise.
580         * cgraphunit.c (decide_is_symbol_needed): Likewise.
581         (first): Likewise.
582         (enqueue_node): Likewise.
583         (referred_to_p): Likewise.
584         (cgraph_process_same_body_aliases): Likewise.
585         (analyze_functions): Likewise.
586         (handle_alias_pairs): Likewise.
587         (output_weakrefs): Likewise.
588         (compile): Likewise.
589         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
590         * ipa-inline-analysis.c (inline_write_summary): Likewise.
591         * ipa-prop.c (remove_described_reference): Likewise.
592         (try_decrement_rdesc_refcount): Likewise.
593         (ipa_edge_duplication_hook): Likewise.
594         * ipa-ref.c (ipa_record_reference): Likewise.
595         (ipa_maybe_record_reference): Likewise.
596         (ipa_clone_ref): Likewise.
597         (ipa_clone_references): Likewise.
598         (ipa_clone_referring): Likewise.
599         (ipa_find_reference): Likewise.
600         (ipa_remove_stmt_references): Likewise.
601         (ipa_clear_stmts_in_references): Likewise.
602         * ipa-ref.h (symtab_node_base): Likewise.
603         (ipa_ref): Likewise.
604         (ipa_record_reference): Likewise.
605         (ipa_maybe_record_reference): Likewise.
606         (ipa_clone_references): Likewise.
607         (ipa_clone_referring): Likewise.
608         (ipa_clone_ref): Likewise.
609         (ipa_find_reference): Likewise.
610         (ipa_remove_stmt_references): Likewise.
611         (ipa_clear_stmts_in_references): Likewise.
612         * ipa-reference.c (ipa_reference_write_optimization_summary):
613         Likewise.
614         * ipa.c (enqueue_node): Likewise.
615         (process_references): Likewise.
616         (walk_polymorphic_call_targets): Likewise.
617         (symtab_remove_unreachable_nodes): Likewise.
618         (address_taken_from_non_vtable_p): Likewise.
619         (comdat_can_be_unshared_p_1): Likewise.
620         (comdat_can_be_unshared_p): Likewise.
621         (can_replace_by_local_alias): Likewise.
622         (function_and_variable_visibility): Likewise.
623         * is-a.h: Likewise (within example in comment).
624         * lto-cgraph.c (input_cgraph_opt_summary): Likewise.
625         (lto_symtab_encoder_encode): Likewise.
626         (lto_symtab_encoder_delete_node): Likewise.
627         (lto_symtab_encoder_in_partition_p): Likewise.
628         (lto_set_symtab_encoder_in_partition): Likewise.
629         (output_refs): Likewise.
630         (compute_ltrans_boundary): Likewise.
631         (output_symtab): Likewise.
632         (input_node): Likewise.
633         (input_ref): Likewise.
634         (input_edge): Likewise.
635         (input_cgraph_1): Likewise.
636         (input_refs): Likewise.
637         (output_cgraph_opt_summary): Likewise.
638         (input_node_opt_summary): Likewise.
639         (input_cgraph_opt_section): Likewise.
640         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
641         Likewise.
642         * lto-streamer-out.c (lto_output): Likewise.
643         (output_symbol_p): Likewise.
644         (produce_symtab): Likewise.
645         * lto-streamer.h (lto_encoder_entry): Likewise.
646         (lto_free_function_in_decl_state_for_node): Likewise.
647         (lto_symtab_encoder_encode): Likewise.
648         (lto_symtab_encoder_delete_node): Likewise.
649         (lto_symtab_encoder_in_partition_p): Likewise.
650         (lto_set_symtab_encoder_in_partition): Likewise.
651         (lto_symtab_encoder_lookup): Likewise.
652         (lsei_node): Likewise.
653         (lto_symtab_encoder_deref): Likewise.
654         * symtab.c (symtab_hash): Likewise.
655         (assembler_name_hash): Likewise.
656         (symtab_nodes): Likewise.
657         (hash_node): Likewise.
658         (eq_node): Likewise.
659         (hash_node_by_assembler_name): Likewise.
660         (eq_assembler_name): Likewise.
661         (insert_to_assembler_name_hash): Likewise.
662         (unlink_from_assembler_name_hash): Likewise.
663         (symtab_prevail_in_asm_name_hash): Likewise.
664         (symtab_register_node): Likewise.
665         (symtab_insert_node_to_hashtable): Likewise.
666         (symtab_unregister_node): Likewise.
667         (symtab_get_node): Likewise.
668         (symtab_remove_node): Likewise.
669         (symtab_initialize_asm_name_hash): Likewise.
670         (symtab_node_for_asm): Likewise.
671         (symtab_add_to_same_comdat_group): Likewise.
672         (symtab_dissolve_same_comdat_group_list): Likewise.
673         (symtab_node_asm_name): Likewise.
674         (symtab_node_name): Likewise.
675         (dump_symtab_base): Likewise.
676         (dump_symtab_node): Likewise.
677         (dump_symtab): Likewise.
678         (debug_symtab_node): Likewise.
679         (verify_symtab_base): Likewise.
680         (verify_symtab_node): Likewise.
681         (verify_symtab): Likewise.
682         (symtab_used_from_object_file_p): Likewise.
683         (symtab_node_availability): Likewise.
684         (symtab_alias_ultimate_target): Likewise.
685         (fixup_same_cpp_alias_visibility): Likewise.
686         (symtab_resolve_alias): Likewise.
687         (symtab_for_node_and_aliases): Likewise.
688         (symtab_for_node_and_aliases): Likewise.
689         (symtab_nonoverwritable_alias_1): Likewise.
690         (symtab_nonoverwritable_alias): Likewise.
691         (symtab_semantically_equivalent_p): Likewise.
692         * value-prof.c (init_node_map): Likewise.
693         * varasm.c (find_decl): Likewise.
694         * varpool.c (varpool_node_for_asm): Likewise.
695         (varpool_remove_unreferenced_decls): Likewise.
697 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
699         Manual part of renaming of symtab_node_base to symtab_node.
701         * ipa-ref.h (symtab_node): Remove typedef to pointer type, as it
702         clashes with the preferred name for the base class.
703         (const_symtab_node): Remove redundant typedef.
705 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
707         * optabs.c (expand_vec_perm): Avoid vector mode punning
708         SUBREGs in SET_DEST.
709         * expmed.c (store_bit_field_1): Likewise.
710         * config/i386/sse.md (movdi_to_sse, vec_pack_sfix_trunc_v2df,
711         vec_pack_sfix_v2df, vec_shl_<mode>, vec_shr_<mode>,
712         vec_interleave_high<mode>, vec_interleave_low<mode>): Likewise.
713         * config/i386/i386.c (ix86_expand_vector_move_misalign,
714         ix86_expand_sse_movcc, ix86_expand_int_vcond, ix86_expand_vec_perm,
715         ix86_expand_sse_unpack, ix86_expand_args_builtin,
716         ix86_expand_vector_init_duplicate, ix86_expand_vector_set,
717         emit_reduc_half, expand_vec_perm_blend, expand_vec_perm_pshufb,
718         expand_vec_perm_interleave2, expand_vec_perm_pshufb2,
719         expand_vec_perm_vpshufb2_vpermq,
720         expand_vec_perm_vpshufb2_vpermq_even_odd, expand_vec_perm_even_odd_1,
721         expand_vec_perm_broadcast_1, expand_vec_perm_vpshufb4_vpermq2,
722         ix86_expand_sse2_mulv4si3, ix86_expand_pinsr): Likewise.
723         (expand_vec_perm_palignr): Likewise.  Modify a copy of *d rather
724         than *d itself.
726 2013-10-31  Uros Bizjak  <ubizjak@gmail.com>
728         * config/i386/i386.c (ix86_expand_sse2_abs): Rename function arguments.
729         Use gcc_unreachable for unhandled modes.  Do not check results of
730         expand_simple_binop.  If not expanded to target, move the result.
732 2013-10-31  Chung-Ju Wu  <jasonwucj@gmail.com>
733             Shiva Chen  <shiva0217@gmail.com>
735         * config.gcc (nds32*-*-*): Add nds32 target.
736         * config/nds32/nds32.c: New file.
737         * config/nds32/nds32.h: New file.
738         * config/nds32/nds32.md: New file.
739         * config/nds32/constants.md: New file.
740         * config/nds32/constraints.md: New file.
741         * config/nds32/iterators.md: New file.
742         * config/nds32/nds32-doubleword.md: New file.
743         * config/nds32/nds32-intrinsic.md: New file.
744         * config/nds32/nds32_intrinsic.h: New file.
745         * config/nds32/nds32-modes.def: New file.
746         * config/nds32/nds32-multiple.md: New file.
747         * config/nds32/nds32.opt: New file.
748         * config/nds32/nds32-opts.h: New file.
749         * config/nds32/nds32-protos.h: New file.
750         * config/nds32/nds32-peephole2.md: New file.
751         * config/nds32/pipelines.md: New file.
752         * config/nds32/predicates.md: New file.
753         * config/nds32/t-mlibs: New file.
754         * common/config/nds32: New directory and files.
756         * doc/invoke.texi (NDS32 options): Document nds32 specific options.
757         * doc/md.texi (NDS32 family): Document nds32 specific constraints.
758         * doc/install.texi (Cross-Compiler-Specific Options): Document
759         --with-nds32-lib for nds32 target.
760         * doc/extend.texi (Function Attributes, Target Builtins): Document
761         nds32 specific attributes.
763 2013-10-31  Vladimir Makarov  <vmakarov@redhat.com>
765         * lra-constraints (process_alt_operands): Use the result
766         elimination register for operand when matching constraints.
768 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
770         * tree-vrp.c (maybe_set_nonzero_bits): New function.
771         (remove_range_assertions): Call it.
773         * tree.c (tree_ctz): New function.
774         * tree.h (tree_ctz): New prototype.
775         * tree-ssanames.h (get_range_info, get_nonzero_bits): Change
776         first argument from tree to const_tree.
777         * tree-ssanames.c (get_range_info, get_nonzero_bits): Likewise.
778         * tree-vectorizer.h (vect_generate_tmps_on_preheader): New prototype.
779         * tree-vect-loop-manip.c (vect_generate_tmps_on_preheader): No longer
780         static.
781         * expr.c (highest_pow2_factor): Reimplemented using tree_ctz.
782         * tree-vect-loop.c (vect_analyze_loop_operations,
783         vect_transform_loop): Don't force scalar loop for bound just because
784         number of iterations is unknown, only do it if it is not known to be
785         a multiple of vectorization_factor.
786         * builtins.c (get_object_alignment_2): Use tree_ctz on offset.
788         * gimple-pretty-print.c (dump_ssaname_info): Print newline also
789         in case of VR_VARYING.  Print get_nonzero_bits if not all ones.
790         * tree-ssanames.h (struct range_info_def): Add nonzero_bits field.
791         (set_nonzero_bits, get_nonzero_bits): New prototypes.
792         * tree-ssa-ccp.c (get_default_value): Use get_range_info to see if
793         a default def isn't partially constant.
794         (ccp_finalize): If after IPA, set_range_info if integral SSA_NAME
795         is known to be partially zero.
796         (evaluate_stmt): If we'd return otherwise VARYING, use get_range_info
797         to see if a default def isn't partially constant.
798         * tree-ssanames.c (set_range_info): Initialize nonzero_bits upon
799         creation of a range, if VR_RANGE, try to improve nonzero_bits from
800         the range.
801         (set_nonzero_bits, get_nonzero_bits): New functions.
803         * tree-cfg.c (assert_unreachable_fallthru_edge_p): New function.
804         * tree-cfg.h (assert_unreachable_fallthru_edge_p): New prototype.
805         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): New function.
806         (remove_range_assertions): If ASSERT_EXPR_VAR has no other immediate
807         uses but in the condition and ASSERT_EXPR and the other successor of
808         the predecessor bb is __builtin_unreachable (), set_range_info of the
809         ASSERT_EXPR_VAR to the range info of the ASSERT_EXPR's lhs.
811 2013-10-31  Martin Jambor  <mjambor@suse.cz>
813         PR rtl-optimization/58934
814         Revert:
815         2013-10-30  Martin Jambor  <mjambor@suse.cz>
816         PR rtl-optimization/10474
817         * ira.c (find_moveable_pseudos): Do not calculate dominance info
818         nor df analysis.
819         (interesting_dest_for_shprep): New function.
820         (split_live_ranges_for_shrink_wrap): Likewise.
821         (ira): Calculate dominance info and df analysis. Call
822         split_live_ranges_for_shrink_wrap.
824 2013-10-31  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
825             Yury Gribov  <y.gribov@samsung.com>
827         PR sanitizer/58543
828         * asan.c (asan_clear_shadow): Allocate a new vreg for temporary
829         shadow pointer to avoid clobbering the main one.
831 2013-10-31  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
833         * lower-subreg.c (resolve_simple_move): Copy REG_INC note.
835 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
837         PR bootstrap/58933
838         * ira-color.c (update_costs_from_copies): Add new parameter.  Use
839         it for calling update_costs_from_allocno.
840         (assign_hard_reg): Call restore_costs_from_copies only for
841         !retry_p.  Pass new argument to update_costs_from_copies.
842         (color_pass): Pass new argument to update_costs_from_copies.
843         (ira_mark_allocation_change): Ditto.
845 2013-10-30  Sharad Singhai  <singhai@google.com>
847         PR middle-end/58134
848         * opts.c (common_handle_option): Remove deprecated option
849         -ftree-vectorizer-verbose.
850         * doc/invoke.texi (Debugging Options): Ditto.
851         * opts-global.c (handle_common_deferred_options): Ditto.
852         (dump_remap_tree_vectorizer_verbose): Delete.
853         * common.opt: Set -ftree-vectorizer-verbose as an ignored option.
855 2013-10-30  DJ Delorie  <dj@redhat.com>
857         * config/rx/rx.c (ADD_RX_BUILTIN0): New macro, used for builtins
858         that take no arguments.
860 2013-10-30  Joern Rennecke  <joern.rennecke@embecosm.com>
862         PR other/58545
863         * reload1.c (update_eliminables_and_spill): New function, broken
864         out of reload.
865         (reload): Use it.  Check for frame size change after frame size
866         alignment, and call update_eliminables_and_spill first if continue-ing.
868 2013-10-30  Cong Hou  <congh@google.com>
870         PR target/58762
871         * config/i386/i386-protos.h (ix86_expand_sse2_abs): New function.
872         * config/i386/i386.c (ix86_expand_sse2_abs): New function.
873         * config/i386/sse.md: Add SSE2 support to abs (8/16/32-bit-int).
875 2013-10-18  Mikael Pettersson  <mikpelinux@gmail.com>
877         PR rtl-optimization/58369
878         * reload1.c (compute_reload_subreg_offset): New function.
879         (choose_reload_regs): Use it to pass endian-correct
880         offset to subreg_regno_offset.
882 2013-10-30  Tobias Burnus  <burnus@net-b.de>
884         PR other/33426
885         * gcc/tree-cfg.c (replace_loop_annotate): Replace warning by
886         warning_at.
888 2013-10-30  Jason Merrill  <jason@redhat.com>
890         * configure.ac (loose_warn): Add -Wno-format if
891         --disable-build-format-warnings.
893 2013-10-30  David Malcolm  <dmalcolm@redhat.com>
895         * cgraphunit.c (analyze_functions): Split symtab_node declarations
896         onto multiple lines to make things easier for rename_symtab.py.
898         * symtab.c (symtab_dissolve_same_comdat_group_list): Likewise.
899         (symtab_semantically_equivalent_p): Likewise.
901 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
903         PR target/58784
904         * lra.c (check_rtl): Remove address check before LRA work.
906 2013-10-30  Marc Glisse  <marc.glisse@inria.fr>
908         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for a
909         POINTER_PLUS_EXPR in the defining statement.
911 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
913         * regmove.c: Remove.
914         * tree-pass.h (make_pass_regmove): Remove.
915         * timevar.def (TV_REGMOVE): Remove.
916         * passes.def (pass_regmove): Remove.
917         * opts.c (default_options_table): Remove entry for regmove.
918         * doc/passes.texi: Remove regmove pass description.
919         * doc/invoke.texi (-foptimize-register-move, -fregmove): Remove
920         options.
921         (-fdump-rtl-regmove): Ditto.
922         * common.opt (foptimize-register-move, fregmove): Ignore.
923         * Makefile.in (OBJS): Remove regmove.o.
924         * regmove.c: Remove.
925         * ira-int.h (struct ira_allocno_pref, ira_pref_t): New structure
926         and type.
927         (struct ira_allocno) New member allocno_prefs.
928         (ALLOCNO_PREFS): New macro.
929         (ira_prefs, ira_prefs_num): New external vars.
930         (ira_setup_alts, ira_get_dup_out_num, ira_debug_pref): New prototypes.
931         (ira_debug_prefs, ira_debug_allocno_prefs, ira_create_pref): Ditto.
932         (ira_add_allocno_pref, ira_remove_pref, ira_remove_allocno_prefs):
933         Ditto.
934         (ira_add_allocno_copy_to_list): Remove prototype.
935         (ira_swap_allocno_copy_ends_if_necessary): Ditto.
936         (ira_pref_iterator): New type.
937         (ira_pref_iter_init, ira_pref_iter_cond): New functions.
938         (FOR_EACH_PREF): New macro.
939         * ira.c (commutative_constraint_p): Move from ira-conflicts.c.
940         (ira_get_dup_out_num): Ditto. Rename from get_dup_num.  Modify the
941         code.
942         (ira_setup_alts): New function.
943         (decrease_live_ranges_number): New function.
944         (ira): Call the above function.
945         * ira-build.c (ira_prefs, ira_prefs_num): New global vars.
946         (ira_create_allocno): Initialize allocno prefs.
947         (pref_pool, pref_vec): New static vars.
948         (initiate_prefs, find_allocno_pref, ira_create_pref): New
949         functions.
950         (add_allocno_pref_to_list, ira_add_allocno_pref, print_pref): Ditto.
951         (ira_debug_pref, print_prefs, ira_debug_prefs): Ditto.
952         (print_allocno_prefs, ira_debug_allocno_prefs, finish_pref): Ditto.
953         (ira_remove_pref, ira_remove_allocno_prefs, finish_prefs): Ditto.
954         (ira_add_allocno_copy_to_list): Make static.  Rename to
955         add_allocno_copy_to_list.
956         (ira_swap_allocno_copy_ends_if_necessary): Make static.  Rename to
957         swap_allocno_copy_ends_if_necessary.
958         (remove_unnecessary_allocnos, remove_low_level_allocnos): Call
959         ira_remove_allocno_prefs.
960         (ira_flattening): Ditto.
961         (ira_build): Call initiate_prefs, print_prefs.
962         (ira_destroy): Call finish_prefs.
963         * ira-color.c (struct update_cost_record): New.
964         (struct allocno_color_data): Add new member update_cost_records.
965         (update_cost_record_pool): New static var.
966         (init_update_cost_records, get_update_cost_record): New functions.
967         (free_update_cost_record_list, finish_update_cost_records): Ditto.
968         (struct update_cost_queue_elem): Add member from.
969         (initiate_cost_update): Call init_update_cost_records.
970         (finish_cost_update): Call finish_update_cost_records.
971         (queue_update_cost, get_next_update_cost): Add new param from.
972         (Update_allocno_cost, update_costs_from_allocno): New functions.
973         (update_costs_from_prefs): Ditto.
974         (update_copy_costs): Rename to update_costs_from_copies.
975         (restore_costs_from_copies): New function.
976         (update_conflict_hard_regno_costs): Don't go back.
977         (assign_hard_reg): Call restore_costs_from_copies.  Add printing
978         more debug info.
979         (pop_allocnos): Add priniting more debug info.
980         (color_allocnos): Remove prefs for conflicting hard regs.
981         Call update_costs_from_prefs.
982         * ira-conflicts.c (commutative_constraint_p): Move to ira.c
983         (get_dup_num): Rename, modify, and move to ira.c
984         (process_regs_for_copy): Add prefs.
985         (add_insn_allocno_copies): Put src as first arg of
986         process_regs_for_copy.  Remove dead code.  Call ira_setup_alts.
987         * ira-costs.c (record_reg_classes): Modify and move code into
988         record_operands_costs.
989         (find_costs_and_classes): Create prefs for the hard reg of small
990         reg class.
991         (process_bb_node_for_hard_reg_moves): Add prefs.
993 2013-10-30  Richard Biener  <rguenther@suse.de>
995         PR middle-end/57100
996         * basic-block.h (pre_and_rev_post_order_compute_fn): New function.
997         * cfganal.c (pre_and_rev_post_order_compute_fn): New function
998         as worker for ...
999         (pre_and_rev_post_order_compute): ... which now wraps it.
1000         * graph.c (draw_cfg_nodes_no_loops): Use
1001         pre_and_rev_post_order_compute_fn to avoid ICEing and dependence
1002         on cfun.
1004 2013-10-30  Christian Bruel  <christian.bruel@st.com>
1006         * gcc/config/sh/sh-mem.cc (sh_expand_cmpnstr): New function.
1007         (sh_expand_cmpstr): Handle known align and schedule improvements.
1008         * gcc/config/sh/sh-protos.h (sh_expand_cmpstrn): Declare.
1009         * gcc/config/sh/sh.md (cmpstrnsi): New pattern.
1011 2013-10-30  Martin Jambor  <mjambor@suse.cz>
1013         PR rtl-optimization/10474
1014         * ira.c (find_moveable_pseudos): Do not calculate dominance info
1015         nor df analysis.
1016         (interesting_dest_for_shprep): New function.
1017         (split_live_ranges_for_shrink_wrap): Likewise.
1018         (ira): Calculate dominance info and df analysis. Call
1019         split_live_ranges_for_shrink_wrap.
1021 2013-10-30  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1023         PR target/58854
1024         * config/arm/arm.c (arm_expand_epilogue_apcs_frame): Emit blockage.
1026 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1028         * tree-core.h (tree_index): Add TI_POINTER_BOUNDS_TYPE.
1029         * tree.h (POINTER_BOUNDS_P): New.
1030         (BOUNDED_TYPE_P): New.
1031         (BOUNDED_P): New.
1032         (pointer_bounds_type_node): New.
1033         * tree.c (build_common_tree_nodes): Initialize
1034         pointer_bounds_type_node.
1035         * gimple.h (gimple_call_get_nobnd_arg_index): New.
1036         (gimple_call_num_nobnd_args): New.
1037         (gimple_call_nobnd_arg): New.
1038         (gimple_return_retbnd): New.
1039         (gimple_return_set_retbnd): New
1040         * gimple.c (gimple_build_return): Increase number of ops
1041         for return statement.
1042         (gimple_call_get_nobnd_arg_index): New.
1043         * gimple-pretty-print.c (dump_gimple_return): Print second op.
1045 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1047         * ipa.c (cgraph_build_static_cdtor_1): Support contructors
1048         with "chkp ctor" and "bnd_legacy" attributes.
1049         * gimplify.c (gimplify_init_constructor): Avoid infinite
1050         loop during gimplification of bounds initializer.
1052 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1054         * c-family/c-common.c (handle_bnd_variable_size_attribute): New.
1055         (handle_bnd_legacy): New.
1056         (c_common_attribute_table): Add bnd_variable_size and bnd_legacy.
1057         * doc/extend.texi: Document bnd_variable_size and bnd_legacy
1058         attributes.
1060 2013-10-29  Ilya Enkovich  <ilya.enkovich@intel.com>
1062         * builtin-types.def (BT_FN_VOID_CONST_PTR): New.
1063         (BT_FN_PTR_CONST_PTR): New.
1064         (BT_FN_CONST_PTR_CONST_PTR): New.
1065         (BT_FN_PTR_CONST_PTR_SIZE): New.
1066         (BT_FN_PTR_CONST_PTR_CONST_PTR): New.
1067         (BT_FN_VOID_PTRPTR_CONST_PTR): New.
1068         (BT_FN_VOID_CONST_PTR_SIZE): New.
1069         (BT_FN_PTR_CONST_PTR_CONST_PTR_SIZE): New.
1070         * chkp-builtins.def: New.
1071         * builtins.def: include chkp-builtins.def.
1072         (DEF_CHKP_BUILTIN): New.
1073         * builtins.c (expand_builtin): Support BUILT_IN_CHKP_INIT_PTR_BOUNDS,
1074         BUILT_IN_CHKP_NULL_PTR_BOUNDS, BUILT_IN_CHKP_COPY_PTR_BOUNDS,
1075         BUILT_IN_CHKP_CHECK_PTR_LBOUNDS, BUILT_IN_CHKP_CHECK_PTR_UBOUNDS,
1076         BUILT_IN_CHKP_CHECK_PTR_BOUNDS, BUILT_IN_CHKP_SET_PTR_BOUNDS,
1077         BUILT_IN_CHKP_NARROW_PTR_BOUNDS, BUILT_IN_CHKP_STORE_PTR_BOUNDS,
1078         BUILT_IN_CHKP_GET_PTR_LBOUND, BUILT_IN_CHKP_GET_PTR_UBOUND,
1079         BUILT_IN_CHKP_BNDMK, BUILT_IN_CHKP_BNDSTX, BUILT_IN_CHKP_BNDCL,
1080         BUILT_IN_CHKP_BNDCU, BUILT_IN_CHKP_BNDLDX, BUILT_IN_CHKP_BNDRET,
1081         BUILT_IN_CHKP_INTERSECT, BUILT_IN_CHKP_ARG_BND, BUILT_IN_CHKP_NARROW,
1082         BUILT_IN_CHKP_EXTRACT_LOWER, BUILT_IN_CHKP_EXTRACT_UPPER.
1083         * common.opt (fcheck-pointer-bounds): New.
1084         * toplev.c (process_options): Check Pointer Bounds Checker is
1085         supported.
1086         * doc/extend.texi: Document Pointer Bounds Checker built-in functions.
1088 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1090         * target.def (builtin_chkp_function): New.
1091         (chkp_bound_type): New.
1092         (chkp_bound_mode): New.
1093         (fn_abi_va_list_bounds_size): New.
1094         (load_bounds_for_arg): New.
1095         (store_bounds_for_arg): New.
1096         * targhooks.h (default_load_bounds_for_arg): New.
1097         (default_store_bounds_for_arg): New.
1098         (default_fn_abi_va_list_bounds_size): New.
1099         (default_chkp_bound_type): New.
1100         (default_chkp_bound_mode): New.
1101         (default_builtin_chkp_function): New.
1102         * targhooks.c (default_load_bounds_for_arg): New.
1103         (default_store_bounds_for_arg): New.
1104         (default_fn_abi_va_list_bounds_size): New.
1105         (default_chkp_bound_type): New.
1106         (default_chkp_bound_mode); New.
1107         (default_builtin_chkp_function): New.
1108         * doc/tm.texi.in (TARGET_FN_ABI_VA_LIST_BOUNDS_SIZE): New.
1109         (TARGET_LOAD_BOUNDS_FOR_ARG): New.
1110         (TARGET_STORE_BOUNDS_FOR_ARG): New.
1111         (TARGET_BUILTIN_CHKP_FUNCTION): New.
1112         (TARGET_CHKP_BOUND_TYPE): New.
1113         (TARGET_CHKP_BOUND_MODE): New.
1114         * doc/tm.texi: Regenerated.
1115         * langhooks.h (lang_hooks): Add chkp_supported field.
1116         * langhooks-def.h (LANG_HOOKS_CHKP_SUPPORTED): New.
1117         (LANG_HOOKS_INITIALIZER); Add LANG_HOOKS_CHKP_SUPPORTED.
1119 2013-10-29  Andrew Pinski <apinski@cavium.com>
1121         * tree-ssa-ifcombine.c: Include rtl.h and tm_p.h.
1122         (ifcombine_ifandif): Handle cases where maybe_fold_and_comparisons
1123         fails, combining the branches anyways.
1124         (tree_ssa_ifcombine): Inverse the order of the basic block walk,
1125         increases the number of combinings.
1126         * gimple.h (gsi_start_nondebug_after_labels_bb): New function.
1128 2013-10-29  Mike Stump  <mikestump@comcast.net>
1130         * machmode.def (PARTIAL_INT_MODE): Add precision and name.
1131         * genmodes.c (PARTIAL_INT_MODE): Add precision and name.
1132         (make_vector_mode): Increase namebuf to 16.
1133         (emit_insn_modes_h): When processing BImode, don't
1134         also match partial int modes.
1135         (emit_class_narrowest_mode): Likewise.
1137         * config/bfin/bfin-modes.def: Add precision to PDI.
1138         * config/m32c/m32c-modes.def: Add precision to PSI.
1139         * config/msp430/msp430-modes.def: Add precision to PSI.
1140         * config/rs6000/rs6000-modes.def: Add precision to PTI.
1141         * config/sh/sh-modes.def: Add precision to PSI and PDI.
1143 2013-10-29  Oleg Endo  <olegendo@gcc.gnu.org>
1145         PR target/54236
1146         * config/sh/sh.md (*addc): Rename existing variations to ...
1147         (*addc_r_r_1, *addc_2r_1, *addc_r_1): ... these.
1148         (*addc_r_lsb, *addc_r_r_lsb, *addc_r_lsb_r, *addc_2r_lsb, *addc_r_msb,
1149         *addc_r_r_msb, *addc_2r_msb): New insn_and_split patterns.
1150         * config/sh/sh.c (addsubcosts): Handle some addc special cases.
1152 2013-10-29  Teresa Johnson  <tejohnson@google.com>
1154         PR ipa/58862
1155         * tree-ssa-tail-merge.c (replace_block_by): Tolerate profile
1156         insanities when updating probabilities.
1158 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1160         * gdbhooks.py (CGraphNodePrinter.to_string): Update gdb
1161         prettyprinter for cgraph_node to reflect the conversion of the
1162         symtable types to a C++ class hierarchy: it now *is* a
1163         symtab_node_base, rather than having one (named "symbol").
1165 2013-10-29  Balaji V. Iyer  <balaji.v.iyer@intel.com>
1167         * builtins.c (is_builtin_name): Added a check for __cilkrts_detach and
1168         __cilkrts_pop_frame.  If matched, then return true for built-in
1169         function name.
1170         (expand_builtin): Added BUILT_IN_CILK_DETACH and
1171         BUILT_IN_CILK_POP_FRAME case.
1172         * langhooks-def.h (lhd_install_body_with_frame_cleanup): New prototype.
1173         (lhs_cilk_detect_spawn): Likewise.
1174         (LANG_HOOKS_DECLS): Added LANG_HOOKS_CILKPLUS.
1175         (LANG_HOOKS_CILKPLUS_DETECT_SPAWN_AND_UNWRAP): New #define.
1176         (LANG_HOOKS_CILKPLUS_FRAME_CLEANUP): Likewise.
1177         (LANG_HOOKS_CILKPLUS_GIMPLIFY_SPAWN): Likewise.
1178         (LANG_HOOKS_CILKPLUS): Likewise.
1179         * tree.h (CILK_SPAWN_FN): Likewise.
1180         * builtin.def (DEF_CILK_BUILTIN_STUB): Likewise.
1181         * Makefile.in (C_COMMON_OBJS): Added c-family/cilk.o.
1182         (OBJS): Added cilk-common.o.
1183         (BUILTINS_DEF): Added cilk-builtins.def.
1184         * langhooks.c (lhd_install_body_with_frame_cleanup): New function.
1185         (lhd_cilk_detect_spawn): Likewise.
1186         * langhooks.h (lang_hooks_for_cilkplus): New struct.
1187         (struct lang_hooks): Added new field called "cilkplus."
1188         * cilk-common.c: New file.
1189         * cilk.h: Likewise.
1190         * cilk-builtins.def: Likewise.
1191         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Added
1192         "__cilk" macro and set it to 200.
1193         * function.h (struct function::cilk_frame_decl): New field.
1194         (struct function::is_cilk_function): Likewise.
1195         (struct function::calls_cilk_spawn): Likewise.
1196         * gimplify.c (gimplify_call_expr): Added a check if the function call
1197         being gimplified is a spawn detach point.  If so, then add pop_frame
1198         and detach function calls.
1199         (gimplify_expr): Added a CILK_SPAWN_STMT and CILK_SYNC_STMT case
1200         for gimplifying _Cilk_spawn and _Cilk_sync statements.
1201         (gimplify_return_expr): Added a check for _Cilk_spawn usage in
1202         function.  If so, added a _Cilk_sync and gimplified it.
1203         (gimplify_modify_expr): Added a check for _Cilk_spawn in MODIFY and
1204         INIT_EXPRs.  If so, then call gimplify_cilk_spawn.
1205         * ipa-inline-analysis (initialize_inline_failed): Prevent inlining of
1206         spawner function.
1207         (can_inline_edge_p): Prevent inling of spawnee function.
1208         * ira.c (ira_setup_eliminable_regset): Force usage of frame pointer
1209         for functions that use Cilk keywords.
1210         * tree-inline.h (struct copy_body_data::remap_var_for_cilk): New field.
1211         * tree-pretty-print.c (dump_generic_node): Added CILK_SPAWN_STMT and
1212         CILK_SYNC_STMT cases.
1213         * tree.def (DEFTREECODE): Added CILK_SPAWN_STMT and CILK_SYNC_STMT
1214         trees.
1215         * generic.texi (CILK_SPAWN_STMT): Added documentation for _Cilk_spawn.
1216         (CILK_SYNC_STMT): Added documentation for _Cilk_sync.
1217         * passes.texi (Cilk Keywords): New section that describes the compiler
1218         code changes for handling Cilk Keywords.
1220 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1222         Patch autogenerated by refactor_symtab.py from
1223         https://github.com/davidmalcolm/gcc-refactoring-scripts
1224         revision 58bb219cc090b2f4516a9297d868c245495ee622
1226         * asan.c (asan_finish_file): Update for conversion of symtab types to
1227         a true class hierarchy.
1228         * cfgexpand.c (estimated_stack_frame_size): Likewise.
1229         * cgraph.c (cgraph_get_body): Likewise.
1230         (cgraph_get_create_real_symbol_node): Likewise.
1231         (verify_cgraph_node): Likewise.
1232         (verify_edge_corresponds_to_fndecl): Likewise.
1233         (verify_edge_count_and_frequency): Likewise.
1234         (cgraph_will_be_removed_from_program_if_no_direct_calls): Likewise.
1235         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
1236         (cgraph_can_remove_if_no_direct_calls_and_refs_p): Likewise.
1237         (cgraph_node_cannot_return): Likewise.
1238         (cgraph_set_pure_flag_1): Likewise.
1239         (cgraph_set_const_flag_1): Likewise.
1240         (cgraph_set_nothrow_flag_1): Likewise.
1241         (cgraph_make_node_local_1): Likewise.
1242         (cgraph_for_node_and_aliases): Likewise.
1243         (cgraph_for_node_thunks_and_aliases): Likewise.
1244         (cgraph_node_can_be_local_p): Likewise.
1245         (cgraph_node_cannot_be_local_p_1): Likewise.
1246         (cgraph_function_body_availability): Likewise.
1247         (dump_cgraph_node): Likewise.
1248         (cgraph_rtl_info): Likewise.
1249         (cgraph_mark_address_taken_node): Likewise.
1250         (cgraph_remove_node): Likewise.
1251         (cgraph_release_function_body): Likewise.
1252         (cgraph_update_edges_for_call_stmt_node): Likewise.
1253         (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
1254         (cgraph_make_edge_direct): Likewise.
1255         (cgraph_resolve_speculation): Likewise.
1256         (cgraph_speculative_call_info): Likewise.
1257         (cgraph_turn_edge_to_speculative): Likewise.
1258         (cgraph_create_edge_1): Likewise.
1259         (cgraph_set_call_stmt): Likewise.
1260         (cgraph_node_for_asm): Likewise.
1261         (cgraph_add_thunk): Likewise.
1262         (cgraph_same_body_alias): Likewise.
1263         (cgraph_create_function_alias): Likewise.
1264         (cgraph_create_node): Likewise.
1265         (cgraph_create_empty_node): Likewise.
1266         (record_function_versions): Likewise.
1267         (used_from_object_file_p): Likewise.
1268         * cgraph.h (symtab_can_be_discarded): Likewise.
1269         (symtab_real_symbol_p): Likewise.
1270         (cgraph_mark_force_output_node): Likewise.
1271         (cgraph_edge_recursive_p): Likewise.
1272         (symtab_alias_target): Likewise.
1273         (varpool_all_refs_explicit_p): Likewise.
1274         (varpool_can_remove_if_no_refs): Likewise.
1275         (cgraph_only_called_directly_or_aliased_p): Likewise.
1276         (cgraph_next_function_with_gimple_body): Likewise.
1277         (cgraph_first_function_with_gimple_body): Likewise.
1278         (cgraph_function_with_gimple_body_p): Likewise.
1279         (cgraph_next_function): Likewise.
1280         (cgraph_first_function): Likewise.
1281         (cgraph_next_defined_function): Likewise.
1282         (cgraph_first_defined_function): Likewise.
1283         (varpool_next_defined_variable): Likewise.
1284         (varpool_first_defined_variable): Likewise.
1285         (varpool_next_static_initializer): Likewise.
1286         (varpool_first_static_initializer): Likewise.
1287         (varpool_next_variable): Likewise.
1288         (varpool_first_variable): Likewise.
1289         (varpool_node_name): Likewise.
1290         (varpool): Likewise.
1291         (cgraph): Likewise.
1292         (is_a_helper <varpool_node>::test): Likewise.
1293         (is_a_helper <cgraph_node>::test): Likewise.
1294         (varpool_variable_node): Likewise.
1295         (cgraph_function_or_thunk_node): Likewise.
1296         (varpool_alias_target): Likewise.
1297         (cgraph_alias_target): Likewise.
1298         (cgraph_node_name): Likewise.
1299         (varpool_node_asm_name): Likewise.
1300         (cgraph_node_asm_name): Likewise.
1301         * cgraphbuild.c (remove_cgraph_callee_edges): Likewise.
1302         (cgraph_rebuild_references): Likewise.
1303         (rebuild_cgraph_edges): Likewise.
1304         (record_eh_tables): Likewise.
1305         (build_cgraph_edges): Likewise.
1306         (mark_store): Likewise.
1307         (mark_load): Likewise.
1308         (mark_address): Likewise.
1309         (record_type_list): Likewise.
1310         (record_reference): Likewise.
1311         * cgraphclones.c (cgraph_materialize_all_clones): Likewise.
1312         (cgraph_materialize_clone): Likewise.
1313         (cgraph_function_versioning): Likewise.
1314         (cgraph_copy_node_for_versioning): Likewise.
1315         (update_call_expr): Likewise.
1316         (cgraph_find_replacement_node): Likewise.
1317         (cgraph_create_virtual_clone): Likewise.
1318         (cgraph_clone_node): Likewise.
1319         * cgraphunit.c (compile): Likewise.
1320         (output_weakrefs): Likewise.
1321         (output_in_order): Likewise.
1322         (expand_function): Likewise.
1323         (assemble_thunks_and_aliases): Likewise.
1324         (expand_thunk): Likewise.
1325         (mark_functions_to_output): Likewise.
1326         (handle_alias_pairs): Likewise.
1327         (analyze_functions): Likewise.
1328         (walk_polymorphic_call_targets): Likewise.
1329         (varpool_finalize_decl): Likewise.
1330         (process_function_and_variable_attributes): Likewise.
1331         (cgraph_process_same_body_aliases): Likewise.
1332         (analyze_function): Likewise.
1333         (cgraph_add_new_function): Likewise.
1334         (cgraph_finalize_function): Likewise.
1335         (referred_to_p): Likewise.
1336         (cgraph_reset_node): Likewise.
1337         (cgraph_process_new_functions): Likewise.
1338         (enqueue_node): Likewise.
1339         (decide_is_symbol_needed): Likewise.
1340         * coverage.c (coverage_compute_profile_id): Likewise.
1341         * dbxout.c (dbxout_expand_expr): Likewise.
1342         * dwarf2out.c (premark_types_used_by_global_vars_helper): Likewise.
1343         (reference_to_unused): Likewise.
1344         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
1345         * gimplify.c (unvisit_body): Likewise.
1346         (unshare_body): Likewise.
1347         * ipa-cp.c (ipcp_generate_summary): Likewise.
1348         (ipcp_decision_stage): Likewise.
1349         (identify_dead_nodes): Likewise.
1350         (decide_whether_version_node): Likewise.
1351         (decide_about_value): Likewise.
1352         (perhaps_add_new_callers): Likewise.
1353         (create_specialized_node): Likewise.
1354         (update_profiling_info): Likewise.
1355         (ipcp_propagate_stage): Likewise.
1356         (estimate_local_effects): Likewise.
1357         (good_cloning_opportunity_p): Likewise.
1358         (devirtualization_time_bonus): Likewise.
1359         (propagate_constants_accross_call): Likewise.
1360         (initialize_node_lattices): Likewise.
1361         (ipcp_cloning_candidate_p): Likewise.
1362         (determine_versionability): Likewise.
1363         (print_all_lattices): Likewise.
1364         (print_lattice): Likewise.
1365         (ipcp_discover_new_direct_edges): Likewise.
1366         * ipa-devirt.c (ipa_devirt): Likewise.
1367         (likely_target_p): Likewise.
1368         (update_type_inheritance_graph): Likewise.
1369         (possible_polymorphic_call_target_p): Likewise.
1370         (dump_possible_polymorphic_call_targets): Likewise.
1371         (devirt_variable_node_removal_hook): Likewise.
1372         (record_binfo): Likewise.
1373         (maybe_record_node): Likewise.
1374         (build_type_inheritance_graph): Likewise.
1375         * ipa-inline-analysis.c (inline_write_summary): Likewise.
1376         (inline_generate_summary): Likewise.
1377         (inline_analyze_function): Likewise.
1378         (do_estimate_growth): Likewise.
1379         (simple_edge_hints): Likewise.
1380         (estimate_node_size_and_time): Likewise.
1381         (estimate_edge_devirt_benefit): Likewise.
1382         (compute_inline_parameters): Likewise.
1383         (estimate_function_body_sizes): Likewise.
1384         (compute_bb_predicates): Likewise.
1385         (initialize_inline_failed): Likewise.
1386         (dump_inline_summary): Likewise.
1387         (dump_inline_edge_summary): Likewise.
1388         * ipa-inline-transform.c (inline_transform): Likewise.
1389         (preserve_function_body_p): Likewise.
1390         (save_inline_function_body): Likewise.
1391         (inline_call): Likewise.
1392         (clone_inlined_nodes): Likewise.
1393         (can_remove_node_now_p): Likewise.
1394         (can_remove_node_now_p_1): Likewise.
1395         * ipa-inline.c (early_inliner): Likewise.
1396         (early_inline_small_functions): Likewise.
1397         (inline_always_inline_functions): Likewise.
1398         (ipa_inline): Likewise.
1399         (flatten_function): Likewise.
1400         (inline_small_functions): Likewise.
1401         (speculation_useful_p): Likewise.
1402         (recursive_inlining): Likewise.
1403         (update_caller_keys): Likewise.
1404         (reset_edge_caches): Likewise.
1405         (update_edge_key): Likewise.
1406         (edge_badness): Likewise.
1407         (relative_time_benefit): Likewise.
1408         (want_inline_self_recursive_call_p): Likewise.
1409         (want_inline_small_function_p): Likewise.
1410         (want_early_inline_function_p): Likewise.
1411         (num_calls): Likewise.
1412         (can_early_inline_edge_p): Likewise.
1413         (can_inline_edge_p): Likewise.
1414         (report_inline_failed_reason): Likewise.
1415         * ipa-profile.c (ipa_profile): Likewise.
1416         (ipa_propagate_frequency): Likewise.
1417         (ipa_propagate_frequency_1): Likewise.
1418         (ipa_profile_generate_summary): Likewise.
1419         * ipa-prop.c (ipcp_transform_function): Likewise.
1420         (read_replacements_section): Likewise.
1421         (ipa_prop_read_section): Likewise.
1422         (ipa_modify_call_arguments): Likewise.
1423         (ipa_print_node_params): Likewise.
1424         (propagate_controlled_uses): Likewise.
1425         (update_indirect_edges_after_inlining): Likewise.
1426         (remove_described_reference): Likewise.
1427         (ipa_make_edge_direct_to_target): Likewise.
1428         (ipa_analyze_node): Likewise.
1429         (ipa_analyze_params_uses): Likewise.
1430         (ipa_compute_jump_functions): Likewise.
1431         (ipa_get_callee_param_type): Likewise.
1432         (ipa_print_node_jump_functions): Likewise.
1433         (ipa_initialize_node_params): Likewise.
1434         (ipa_populate_param_decls): Likewise.
1435         (ipa_func_spec_opts_forbid_analysis_p): Likewise.
1436         (write_agg_replacement_chain): Likewise.
1437         (ipa_write_node_info): Likewise.
1438         (ipa_edge_duplication_hook): Likewise.
1439         (try_decrement_rdesc_refcount): Likewise.
1440         * ipa-pure-const.c (propagate_nothrow): Likewise.
1441         (propagate_pure_const): Likewise.
1442         (pure_const_read_summary): Likewise.
1443         (pure_const_write_summary): Likewise.
1444         (analyze_function): Likewise.
1445         * ipa-ref-inline.h (ipa_ref_referred_ref_list): Likewise.
1446         (ipa_ref_referring_ref_list): Likewise.
1447         * ipa-ref.c (ipa_clear_stmts_in_references): Likewise.
1448         (ipa_remove_stmt_references): Likewise.
1449         (ipa_find_reference): Likewise.
1450         (ipa_dump_referring): Likewise.
1451         (ipa_dump_references): Likewise.
1452         (ipa_record_reference): Likewise.
1453         * ipa-reference.c (ipa_reference_read_optimization_summary): Likewise.
1454         (ipa_reference_write_optimization_summary): Likewise.
1455         (write_node_summary_p): Likewise.
1456         (propagate): Likewise.
1457         (read_write_all_from_decl): Likewise.
1458         (generate_summary): Likewise.
1459         (analyze_function): Likewise.
1460         (propagate_bits): Likewise.
1461         (ipa_reference_get_not_written_global): Likewise.
1462         (ipa_reference_get_not_read_global): Likewise.
1463         * ipa-split.c (execute_split_functions): Likewise.
1464         (split_function): Likewise.
1465         * ipa-utils.c (ipa_merge_profiles): Likewise.
1466         (dump_cgraph_node_set): Likewise.
1467         (ipa_reverse_postorder): Likewise.
1468         (ipa_edge_within_scc): Likewise.
1469         (ipa_get_nodes_in_cycle): Likewise.
1470         (ipa_free_postorder_info): Likewise.
1471         (ipa_reduced_postorder): Likewise.
1472         (searchc): Likewise.
1473         (recursive_call_p): Likewise.
1474         * ipa.c (ipa_cdtor_merge): Likewise.
1475         (record_cdtor_fn): Likewise.
1476         (function_and_variable_visibility): Likewise.
1477         (varpool_externally_visible_p): Likewise.
1478         (cgraph_externally_visible_p): Likewise.
1479         (comdat_can_be_unshared_p): Likewise.
1480         (comdat_can_be_unshared_p_1): Likewise.
1481         (address_taken_from_non_vtable_p): Likewise.
1482         (ipa_discover_readonly_nonaddressable_vars): Likewise.
1483         (symtab_remove_unreachable_nodes): Likewise.
1484         (walk_polymorphic_call_targets): Likewise.
1485         (process_references): Likewise.
1486         (enqueue_node): Likewise.
1487         (has_addr_references_p): Likewise.
1488         (cgraph_non_local_node_p_1): Likewise.
1489         * is-a.h (varpool_analyze_node): Likewise.
1490         * lto-cgraph.c (input_symtab): Likewise.
1491         (merge_profile_summaries): Likewise.
1492         (input_cgraph_1): Likewise.
1493         (input_edge): Likewise.
1494         (input_varpool_node): Likewise.
1495         (input_node): Likewise.
1496         (input_overwrite_node): Likewise.
1497         (compute_ltrans_boundary): Likewise.
1498         (output_refs): Likewise.
1499         (lto_output_varpool_node): Likewise.
1500         (lto_output_node): Likewise.
1501         (reachable_from_other_partition_p): Likewise.
1502         (referenced_from_other_partition_p): Likewise.
1503         (lto_output_edge): Likewise.
1504         (output_node_opt_summary): Likewise.
1505         (add_node_to): Likewise.
1506         (reachable_from_this_partition_p): Likewise.
1507         (lto_set_symtab_encoder_in_partition): Likewise.
1508         (lto_symtab_encoder_in_partition_p): Likewise.
1509         (lto_set_symtab_encoder_encode_initializer): Likewise.
1510         (lto_symtab_encoder_encode_initializer_p): Likewise.
1511         (lto_set_symtab_encoder_encode_body): Likewise.
1512         (lto_symtab_encoder_encode_body_p): Likewise.
1513         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
1514         Likewise.
1515         * lto-streamer-in.c (lto_read_body): Likewise.
1516         (fixup_call_stmt_edges): Likewise.
1517         (fixup_call_stmt_edges_1): Likewise.
1518         * lto-streamer-out.c (produce_symtab): Likewise.
1519         (output_symbol_p): Likewise.
1520         (write_symbol): Likewise.
1521         (lto_output): Likewise.
1522         (copy_function): Likewise.
1523         (output_function): Likewise.
1524         * passes.c (function_called_by_processed_nodes_p): Likewise.
1525         (ipa_write_optimization_summaries): Likewise.
1526         (ipa_write_summaries): Likewise.
1527         (do_per_function_toporder): Likewise.
1528         (do_per_function): Likewise.
1529         (dump_passes): Likewise.
1530         * symtab.c (symtab_semantically_equivalent_p): Likewise.
1531         (symtab_nonoverwritable_alias): Likewise.
1532         (symtab_nonoverwritable_alias_1): Likewise.
1533         (symtab_for_node_and_aliases): Likewise.
1534         (symtab_resolve_alias): Likewise.
1535         (fixup_same_cpp_alias_visibility): Likewise.
1536         (symtab_alias_ultimate_target): Likewise.
1537         (symtab_used_from_object_file_p): Likewise.
1538         (verify_symtab_base): Likewise.
1539         (dump_symtab_base): Likewise.
1540         (symtab_node_name): Likewise.
1541         (symtab_node_asm_name): Likewise.
1542         (symtab_dissolve_same_comdat_group_list): Likewise.
1543         (symtab_add_to_same_comdat_group): Likewise.
1544         (symtab_unregister_node): Likewise.
1545         (symtab_insert_node_to_hashtable): Likewise.
1546         (symtab_register_node): Likewise.
1547         (unlink_from_assembler_name_hash): Likewise.
1548         (insert_to_assembler_name_hash): Likewise.
1549         (eq_assembler_name): Likewise.
1550         (hash_node_by_assembler_name): Likewise.
1551         (eq_node): Likewise.
1552         (hash_node): Likewise.
1553         * toplev.c (wrapup_global_declaration_2): Likewise.
1554         * trans-mem.c (ipa_tm_execute): Likewise.
1555         (ipa_tm_transform_clone): Likewise.
1556         (ipa_tm_transform_transaction): Likewise.
1557         (ipa_tm_transform_calls_redirect): Likewise.
1558         (ipa_tm_insert_gettmclone_call): Likewise.
1559         (ipa_tm_insert_irr_call): Likewise.
1560         (ipa_tm_create_version): Likewise.
1561         (ipa_tm_create_version_alias): Likewise.
1562         (ipa_tm_mark_forced_by_abi_node): Likewise.
1563         (ipa_tm_mark_force_output_node): Likewise.
1564         (ipa_tm_diagnose_tm_safe): Likewise.
1565         (ipa_tm_mayenterirr_function): Likewise.
1566         (ipa_tm_scan_irr_function): Likewise.
1567         (ipa_tm_note_irrevocable): Likewise.
1568         (ipa_tm_scan_calls_clone): Likewise.
1569         (get_cg_data): Likewise.
1570         * tree-eh.c (tree_could_trap_p): Likewise.
1571         * tree-emutls.c (ipa_lower_emutls): Likewise.
1572         (create_emultls_var): Likewise.
1573         (lower_emutls_function_body): Likewise.
1574         (gen_emutls_addr): Likewise.
1575         (emutls_decl): Likewise.
1576         (new_emutls_decl): Likewise.
1577         * tree-inline.c (tree_function_versioning): Likewise.
1578         (optimize_inline_calls): Likewise.
1579         (expand_call_inline): Likewise.
1580         (estimate_num_insns): Likewise.
1581         (copy_bb): Likewise.
1582         (delete_unreachable_blocks_update_callgraph): Likewise.
1583         * tree-nested.c (gimplify_all_functions): Likewise.
1584         (create_nesting_tree): Likewise.
1585         (check_for_nested_with_variably_modified): Likewise.
1586         * tree-pretty-print.c (dump_function_header): Likewise.
1587         * tree-profile.c (tree_profiling): Likewise.
1588         * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
1589         (modify_function): Likewise.
1590         (convert_callers): Likewise.
1591         (convert_callers_for_node): Likewise.
1592         * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
1593         (associate_varinfo_to_alias): Likewise.
1594         (create_variable_info_for): Likewise.
1595         (get_constraint_for_ssa_var): Likewise.
1596         * tree-vectorizer.c (increase_alignment): Likewise.
1597         * tree.c (find_decls_types_in_var): Likewise.
1598         (find_decls_types_in_node): Likewise.
1599         (free_lang_data_in_decl): Likewise.
1600         * value-prof.c (gimple_ic_transform): Likewise.
1601         (gimple_ic): Likewise.
1602         (check_ic_target): Likewise.
1603         (init_node_map): Likewise.
1604         * varasm.c (decl_binds_to_current_def_p): Likewise.
1605         (default_binds_local_p_1): Likewise.
1606         (dump_tm_clone_pairs): Likewise.
1607         (assemble_alias): Likewise.
1608         (find_decl): Likewise.
1609         (mark_decl_referenced): Likewise.
1610         * varpool.c (varpool_for_node_and_aliases): Likewise.
1611         (varpool_extra_name_alias): Likewise.
1612         (varpool_create_variable_alias): Likewise.
1613         (add_new_static_var): Likewise.
1614         (varpool_finalize_named_section_flags): Likewise.
1615         (varpool_remove_unreferenced_decls): Likewise.
1616         (enqueue_node): Likewise.
1617         (varpool_assemble_decl): Likewise.
1618         (assemble_aliases): Likewise.
1619         (varpool_analyze_node): Likewise.
1620         (cgraph_variable_initializer_availability): Likewise.
1621         (varpool_add_new_variable): Likewise.
1622         (ctor_for_folding): Likewise.
1623         (dump_varpool_node): Likewise.
1624         (varpool_remove_initializer): Likewise.
1625         (varpool_remove_node): Likewise.
1626         (varpool_node_for_decl): Likewise.
1627         (varpool_create_empty_node): Likewise.
1628         * config/i386/i386.c (ix86_generate_version_dispatcher_body): Likewise.
1629         (ix86_get_function_versions_dispatcher): Likewise.
1631 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1633         * cgraph.h (symtab_node_base): Convert to a class;
1634         add GTY((desc ("%h.type"), tag ("SYMTAB_SYMBOL"))), and take
1635         chain_next/prev from symtab_node_def.
1636         (cgraph_node): Inherit from symtab_node; add GTY option
1637         tag ("SYMTAB_FUNCTION").
1638         (varpool_node): Inherit from symtab_node; add GTY option
1639         tag ("SYMTAB_VARIABLE").
1640         (symtab_node_def): Remove.
1641         (is_a_helper <cgraph_node>::test (symtab_node_def *)): Convert to...
1642         (is_a_helper <cgraph_node>::test (symtab_node_base *)): ...this.
1643         (is_a_helper <varpool_node>::test (symtab_node_def *)): Convert to...
1644         (is_a_helper <varpool_node>::test (symtab_node_base *)): ...this.
1646         * ipa-ref.h (symtab_node_def): Drop.
1647         (symtab_node): Change underlying type from symtab_node_def to
1648         symtab_node_base.
1649         (const_symtab_node): Likwise.
1651         * is-a.h: Update examples in comment.
1653         * symtab.c (symtab_hash): Change symtab_node_def to symtab_node_base.
1654         (assembler_name_hash): Likewise.
1656 2013-10-29  Martin Liska  <marxin.liska@gmail.com>
1658         * doc/tree-ssa.texi (gimple_phi_result): Document.
1659         (gimple_phi_num_args, gimple_phi_arg): Likewise.
1660         (gimple_phi_arg_edge, gimple_phi_arg_def): Likewise.
1661         (PHI_RESULT, PHI_NUM_ARGS): Remove.
1662         (PHI_ARG_ELT, PHI_ARG_EDGE, PHI_ARG_DEF): Likewise.
1664 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1666         * expr.h: Revert change and include tree-core.h.
1667         * rtl.h: Revert change and don't include tree-core.h.
1669 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1671         * config/darwin.c: Include gimple.h.
1672         * config/i386/winnt.c: Likewise.
1674 2013-10-29  Marc Glisse  <marc.glisse@inria.fr>
1676         PR tree-optimization/19831
1677         * tree-ssa-alias.c (stmt_kills_ref_p_1): Handle BUILT_IN_FREE.
1679 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1681         * tree-outof-ssa.h: Remove include files.
1682         * tree-outof-ssa.c: Add required include files from tree-outof-ssa.h.
1683         * expr.c: Likewise.
1684         * tree-ssa-coalesce.c: Likewise.
1685         * cfgexpand.c: Likewise.
1686         * tree-ssa-ter.c: Likewise.
1687         * ipa-prop.h: Remove gimple.h and tree-core.h from include list.
1688         * lto-streamer.h: Likewise.
1689         * cgraphbuild.c: Add gimple.h to include list.
1690         * data-streamer-in.c: Likewise.
1691         * ipa-cp.c: Likewise.
1692         * tree-streamer.c: Likewise.
1693         * lto-compress.c: Likewise.
1694         * ipa-reference.c: Likewise.
1695         * data-streamer-out.c: Likewise.
1696         * lto-cgraph.c: Likewise.
1697         * cgraphclones.c: Likewise.
1698         * ipa-utils.c: Likewise.
1699         * data-streamer.c: Likewise.
1700         * ipa-split.c: Likewise.
1701         * lto-section-in.c: Likewise.
1702         * tree-streamer-out.c: Likewise.
1703         * ipa-prop.c: Likewise.
1704         * tree-streamer-in.c: Likewise.
1705         * symtab.c: Likewise.
1706         * opts-global.c: Likewise.
1707         * lto-opts.c: Likewise.
1708         * lto-section-out.c: Likewise.
1709         * lto-streamer.c: Likewise.
1710         * rtl.h: Add tree-core.h to include list.
1711         * expr.h: Remove tree-core.h from include list.
1712         * gimple.h: Likewise.
1713         * ipa-utils.h: Likewise.
1714         * streamer-hooks.h: Likewise.
1715         * streamer-hooks.c: Include input.h.
1717 2013-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1719         * config/arm/arm.c (cortexa7_extra_costs): New table.
1720         (arm_cortex_a7_tune): New.
1721         * config/arm/arm-cores.def: Use cortex_a7 tuning for cortex-a7.
1723 2013-10-29  Eric Botcazou  <ebotcazou@adacore.com>
1725         * expr.c (expand_expr_real_1) <MEM_EXPR>: Eliminate small redundancy.
1727 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1729         * doc/gty.texi ("Inheritance and GTY"): Make it clear that
1730         to use autogenerated markers for a class-hierarchy, every class
1731         must have a GTY marker.
1732         * gengtype.h (struct type): Add linked list of subclasses to
1733         the "s" member of the union.
1734         (add_subclass): New decl.
1735         * gengtype-state.c (read_state_struct_type): Set up subclass
1736         linked list.
1737         * gengtype.c (get_ultimate_base_class): New.
1738         (add_subclass): New.
1739         (new_structure): Set up subclass linked list.
1740         (set_gc_used_type): Propagate usage information to subclasses.
1741         (output_mangled_typename): Use get_ultimate_base_class.
1742         (walk_subclasses): Use the subclass linked list, avoiding an
1743         O(N^2) when writing out all types.
1744         (walk_type): Issue an error if the base class is missing a tag,
1745         rather than generating bogus C code.  Add a gcc_unreachable
1746         default case, in case people omit tags from concrete subclasses,
1747         or get the values wrong.
1748         (write_func_for_structure): Issue an error for subclasses for
1749         which the base doesn't have a "desc", since otherwise the
1750         autogenerated routines for the base would silently fail to visit
1751         any subclass fields.
1752         (write_root): Use get_ultimate_base_class, tweaking constness of
1753         tp to match that function's signature.
1755 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1757         * doc/gty.texi (GTY Options): Add note about inheritance to
1758         description of desc and tag.
1759         (Inheritance and GTY): New.
1761 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1763         * gengtype-parse.c (opts_have): Drop "static" so that
1764         we can use this from gengtype.c.
1765         * gengtype.c (set_gc_used_type): Mark any base class as used;
1766         update field traversal to visit inherited fields.
1767         (output_mangled_typename):  Convert references to classes within
1768         an inheritance hierarchy to reference the ultimate base class,
1769         since only it will have gt_ functions.
1770         (get_string_option): New.
1771         (walk_subclasses): New.
1772         (walk_type): Treat GTY structs that have a "desc" as being the
1773         root of an inheritance hierarchy.  Generate a switch on it
1774         within the marking function which walks all subclasses, adding
1775         cases for them via walk_subclasses.  For subclasses, visit all
1776         fields of the type (including inherited ones).
1777         (write_func_for_structure): Don't write fns for subclasses, only
1778         for the ultimate base class within an inheritance hierarchy.
1779         Subclasses-marking will be handled by the base class marking functions.
1780         (write_types): Likewise.
1781         (write_local_func_for_structure): Likewise.
1782         (USED_BY_TYPED_GC_P): Emit allocators for subclasses that have
1783         a "tag" option (and are thus concrete subclasses).
1784         (write_root): Use the marker function for the ultimate base class.
1785         * gengtype.h (FOR_ALL_INHERITED_FIELDS): New.
1786         (opts_have): Add declaration.
1788 2013-10-28  Vladimir Makarov  <vmakarov@redhat.com>
1790         * lra-spills.c (lra_final_code_change): Remove useless move insns
1791         originated from moves of pseudos.
1793 2013-10-28  Jeff Law  <law@redhat.com>
1795         * lower-subreg.c (resolve_simple_move): Fix comment typo.
1797 2013-10-28  Trevor Saunders  <tsaunders@mozilla.com>
1799         * df-scan.c (df_collection_rec): Adjust.
1800         (copy_defs): New constant.
1801         (copy_uses): Likewise.
1802         (copy_eq_uses): Likewise.
1803         (copy_mw): Likewise.
1804         (copy_all): Likewise.
1805         (df_insn_rescan): Adjust.
1806         (df_notes_rescan): Likewise.
1807         (df_swap_refs): Likewise.
1808         (df_sort_and_compress_refs): Likewise.
1809         (df_sort_and_compress_mws): Likewise.
1810         (df_install_refs): Likewise.
1811         (df_install_mws): Likewise.
1812         (df_refs_add_to_chains): Add flags parameter controlling which vectors
1813         are coppied.
1814         (df_bb_refs_record): Adjust.
1815         (df_record_entry_block_defs): Likewise.
1816         (df_record_exit_block_defs): Likewise.
1817         (df_refs_verify): Likewise.
1818         (df_mws_verify): Likewise.
1819         (df_insn_refs_verify): Likewise.
1820         (df_bb_verify): Likewise.
1821         * ipa-pure-const.c (finish_state): Remove.
1822         (propagate): Adjust.
1823         * tree-data-ref.c tree-ssa-alias.c tree-ssa-loop-ivcanon.c
1824         tree-ssa-threadedge.c tree-vect-loop-manip.c tree-vect-slp.c
1825         var-tracking.c: Adjust.
1826         * vec.c (stack_vecs): Remove.
1827         (register_stack_vec): Likewise.
1828         (stack_vec_register_index): Likewise.
1829         (unregister_stack_vec): Likewise.
1830         * vec.h (struct va_stack): Remove.
1831         (struct vec<T, A, vl_ptr>): Specialize as
1832         struct vec<T, va_heap, vl_ptr> instead since va_heap is the only
1833         allocation strategy compatable with the vl_ptr layout.
1834         (struct vec<T, va_gc, vl_ptr>): Remove because it now gets an empty
1835         specialization anyway.
1836         (class stack_vec): New class.
1837         (vec_stack_alloc): Remove.
1838         (vec<T, va_heap, vl_ptr>::using_auto_storage): New method.
1840 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1841             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1842             Sergey Lega  <sergey.s.lega@intel.com>
1843             Anna Tikhonova  <anna.tikhonova@intel.com>
1844             Ilya Tocar  <ilya.tocar@intel.com>
1845             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1846             Ilya Verbin  <ilya.verbin@intel.com>
1847             Kirill Yukhin  <kirill.yukhin@intel.com>
1848             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1850         * config/i386/i386.md (prefetch): Allow TARGET_AVX512PF.
1851         (*prefetch_avx512pf_<mode>): New.
1852         * config/i386/sse.md (avx512f_vmcmp<mode>3): Ditto.
1853         (avx512f_maskcmp<mode>3): Ditto.
1854         (vashrv16si3): Ditto.
1856 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1857             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1858             Sergey Lega  <sergey.s.lega@intel.com>
1859             Anna Tikhonova  <anna.tikhonova@intel.com>
1860             Ilya Tocar  <ilya.tocar@intel.com>
1861             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1862             Ilya Verbin  <ilya.verbin@intel.com>
1863             Kirill Yukhin  <kirill.yukhin@intel.com>
1864             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1866         * config/i386/i386.md (any_truncate): New.
1867         (trunsuffix): Ditto.
1868         * config/i386/predicates.md (const_8_to_9_operand): New.
1869         (const_10_to_11_operand): Ditto.
1870         (const_12_to_13_operand): Ditto.
1871         (const_14_to_15_operand): Ditto.
1872         (const_16_to_19_operand): Ditto.
1873         (const_20_to_23_operand): Ditto.
1874         (const_24_to_27_operand): Ditto.
1875         (const_28_to_31_operand): Ditto.
1876         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_FIX_NOTRUNC.
1877         (cvtusi2<ssescalarmodesuffix>32): New.
1878         (cvtusi2<ssescalarmodesuffix>64): Ditto.
1879         (ufloatv16siv16sf2): Ditto.
1880         (avx512f_fix_notruncv16sfv16si): Ditto.
1881         (avx512f_ufix_notruncv16sfv16si): Ditto.
1882         (avx512f_vcvtss2usi): Ditto.
1883         (avx512f_vcvtss2usiq): Ditto.
1884         (avx512f_vcvttss2usi): Ditto.
1885         (avx512f_vcvttss2usiq): Ditto.
1886         (avx512f_vcvtsd2usi): Ditto.
1887         (avx512f_vcvtsd2usiq): Ditto.
1888         (avx512f_vcvttsd2usi): Ditto.
1889         (avx512f_vcvttsd2usiq): Ditto.
1890         (ufloatv8siv8df): Ditto.
1891         (avx512f_cvtdq2pd512_2): Ditto.
1892         (avx512f_cvtpd2dq512): Ditto.
1893         (avx512f_ufix_notruncv8dfv8si): Ditto.
1894         (avx512f_cvtpd2ps512): Ditto.
1895         (vec_unpacks_lo_v16sf): Ditto.
1896         (vec_unpacks_hi_v16sf): Ditto.
1897         (vec_unpacks_float_hi_v16si): Ditto.
1898         (vec_unpacks_float_lo_v16si): Ditto.
1899         (avx512f_unpckhps512): Ditto.
1900         (avx512f_unpcklps512): Ditto.
1901         (avx512f_movshdup512): Ditto.
1902         (avx512f_movsldup512): Ditto.
1903         (vec_extract_lo_v32hi): Ditto.
1904         (vec_extract_hi_v32hi): Ditto.
1905         (vec_extract_lo_v64qi): Ditto.
1906         (vec_extract_hi_v64qi): Ditto.
1907         (avx512f_unpckhpd512): Ditto.
1908         (avx512f_movddup512): Ditto.
1909         (avx512f_unpcklpd512): Ditto.
1910         (*avx512f_unpcklpd512): Ditto.
1911         (avx512f_shufps512_1): Ditto.
1912         (avx512f_shufpd512_1): Ditto.
1913         (avx512f_interleave_highv8di): Ditto.
1914         (avx512f_interleave_lowv8di): Ditto.
1915         (PMOV_DST_MODE): Ditto.
1916         (pmov_src_mode): Ditto.
1917         (pmov_src_lower): Ditto.
1918         (pmov_suff): Ditto.
1919         (*avx512f_<code><pmov_src_lower><mode>2): Ditto.
1920         (*avx512f_<code>v8div16qi2): Ditto.
1921         (*avx512f_<code>v8div16qi2_store): Ditto.
1922         (vec_widen_umult_even_v16si): Ditto.
1923         (*vec_widen_umult_even_v16si): Ditto.
1924         (vec_widen_smult_even_v16si): Ditto.
1925         (*vec_widen_smult_even_v16si): Ditto.
1926         (avx512f_interleave_highv16si): Ditto.
1927         (avx512f_interleave_lowv16si): Ditto.
1928         (avx512f_<code>v16qiv16si2): Ditto.
1929         (avx512f_<code>v16hiv16si2): Ditto.
1930         (avx512f_<code>v8qiv8di2): Ditto.
1931         (avx512f_<code>v8hiv8di2): Ditto.
1932         (avx512f_<code>v8siv8di2): Ditto.
1933         (avx512cd_maskb_vec_dupv8di): Ditto.
1934         (avx512cd_maskw_vec_dupv16si): Ditto.
1935         (avx512f_vcvtph2ps512): Ditto.
1936         (avx512f_vcvtps2ph512): Ditto.
1937         (VEC_EXTRACT_MODE): Extened with wider modes.
1938         (VEC_PERM_AVX2): Ditto.
1939         (VEC_PERM_CONST): Ditto.
1941 2013-10-28  Joern Rennecke  <joern.rennecke@embecosm.com>
1943         * config/arc/arc.c (arc_ccfsm_post_advance):
1944         Add comment about TYPE_RETURN.
1946 2013-10-28  Bin Cheng  <bin.cheng@arm.com>
1948         * tree-ssa-loop-ivopts.c (strip_offset_1): Change parameter type.
1949         Count DECL_FIELD_BIT_OFFSET for COMPONENT_REF.
1950         (strip_offset): Convert offset to unsigned number.
1952 2013-10-27  Tom de Vries  <tom@codesourcery.com>
1954         * cfgexpand.c (gimple_expand_cfg): Remove test for parm_birth_insn.
1955         Don't commit insertions after NOTE_INSN_FUNCTION_BEG.
1957 2013-10-27  Oleg Endo  <olegendo@gcc.gnu.org>
1959         * config/sh/sh.c (MSW, LSW): Move and rename macros to...
1960         * config/sh/sh.h (SH_REG_MSW_OFFSET, SH_REG_LSW_OFFSET): ... here.
1961         (TARGET_BIG_ENDIAN): New macro.
1962         * config/sh/sh.md: Use it instead of !TARGET_LITTLE_ENDIAN.
1963         Use SH_REG_MSW_OFFSET and SH_REG_LSW_OFFSET.
1964         * config/sh/sh.c: Likewise.
1965         * config/sh/sh.h: Likewise.
1967 2013-10-27  Hans-Peter Nilsson  <hp@axis.com>
1969         * config/cris/cris.c (cris_emit_trap_for_misalignment): Replace the
1970         removed PRED_MUDFLAP with PRED_NORETURN.  Correct file-path in comment.
1972 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1974         * config/sh/sh.md (movmemsi): Remove empty constraints and predicates.
1975         Fix formatting.
1976         (cmpstr_t, cmpstrsi): Fix formatting.
1978 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
1980         PR target/52483
1981         * config/sh/predicates.md (general_movdst_operand): Allow reg+reg
1982         addressing, do not use general_operand for memory operands.
1984 2013-10-26  Vladimir Makarov  <vmakarov@redhat.com>
1986         Revert:
1987         2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
1988         * lra-spills.c (lra_final_code_change): Remove useless move insns.
1990 2013-10-26  Jeff Law  <law@redhat.com>
1992         * predict.c (PRED_MUDFLAP): Remove.
1993         * targhooks.c (build_va_arg_indirect_ref): Remove mudflap support.
1995         * Makefile.in (C_COMMON_OBJS): Remove tree-mudflap.
1996         (OBJS): Remove tree-nomudflap.o
1997         (GTFILES): Remove tree-mudflap.c
1998         * builtins.c (expand_builtin_alloc): Remove mudflap support.
1999         * gcc.c (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2000         (mfwrap_spec, mflib_spec): Likewise.
2001         (cpp_unique_options, cc1_options, static_specs): Likewise.
2002         * gimplify (gimplify_vla_decl, build_va_arg_indirect_ref): Likewise.
2003         * passes.def: Likewise.
2004         * toplev.c (compile_file, process_options): Likewise.
2005         * tree-inline.c (copy_tree_r): Likewise.
2006         * tree-pass.,h (make_pass_mudflap_1, make_pass_mudflap_2): Likewise.
2007         * varasm.c (make_decl_rtl, make_decl_rtl_for_debug): Likewise.
2008         (build_constant_desc, output_constant_def_contents): Likewise.
2009         (categorize_decl_for_section): Likewise.
2010         * tree-mudflap.c: Removed.
2011         * tree-mudflap.h: Removed.
2012         * tree-nomudflap.c: Removed.
2013         * bfin/uclinux.h (MFWRAP_SPEC): Remove.
2014         * moxie/uclinux.h (MFWRAP_SPEC): Likewise.
2015         * rs6000/aix.h (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2016         * config/sol2.h (MFLIB_SPEC): Likewise.
2017         * doc/install.texi: Remove mudflap references.
2018         * doc/passes.texi: Similarly.
2019         * doc/sourcebuild.texi: Similarly.
2020         * doc/invoke.texi: Remove mudlfap related options.
2022 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2024         PR rtl-optimization/58759
2025         * lra-constraints.c (lra_constraints): Remove wrong condition to
2026         remove insn setting up an equivalent pseudo.
2028 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2030         * config/rs6000/rs6000-protos.h
2031         (rs6000_secondary_memory_needed_mode): New prototype.
2032         * config/rs6000/rs6000.c: Include ira.h.
2033         (TARGET_LRA_P): Redefine.
2034         (rs6000_legitimate_offset_address_p): Call
2035         legitimate_constant_pool_address_p in strict mode for LRA.
2036         (rs6000_legitimate_address_p): Ditto.
2037         (legitimate_lo_sum_address_p): Add code for LRA.  Use lra_in_progress.
2038         (rs6000_emit_move): Add LRA version of code to generate load/store
2039         of SDmode values.
2040         (rs6000_secondary_memory_needed_mode): New.
2041         (rs6000_alloc_sdmode_stack_slot): Do nothing for LRA.
2042         (rs6000_secondary_reload_class): Return NO_REGS for LRA for
2043         constants, memory, and FP registers.
2044         (rs6000_lra_p): New.
2045         * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): New macro.
2046         * config/rs6000/rs6000.opt (mlra): New option.
2047         * lra-spills.c (lra_final_code_change): Remove useless move insns.
2049 2013-10-25  Yufeng Zhang  <yufeng.zhang@arm.com>
2051         * tree-ssa-math-opts.c (convert_plusminus_to_widen): Call
2052         has_single_use () and not do the conversion if has_single_use ()
2053         returns false for the multiplication result.
2055 2013-10-25  David Malcolm  <dmalcolm@redhat.com>
2057         * tree.h (EXCEPTIONAL_CLASS_P): Rename parameter from "CODE"
2058         to "NODE", since this works on a "tree", not an
2059         "enum tree_code".
2060         (CONSTANT_CLASS_P): Likewise.
2061         (TYPE_P): Likewise.
2062         (DECL_P): Likewise.
2063         (INDIRECT_REF_P): Likewise.
2064         (REFERENCE_CLASS_P): Likewise.
2065         (COMPARISON_CLASS_P): Likewise.
2066         (UNARY_CLASS_P): Likewise.
2067         (BINARY_CLASS_P): Likewise.
2068         (STATEMENT_CLASS_P): Likewise.
2069         (VL_EXP_CLASS_P): Likewise.
2070         (EXPRESSION_CLASS_P): Likewise.
2071         (IS_TYPE_OR_DECL_P): Likewise.
2073 2013-10-25  Marc Glisse  <marc.glisse@inria.fr>
2075         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for an
2076         ADDR_EXPR in the defining statement.
2078 2013-10-25  Richard Biener  <rguenther@suse.de>
2080         PR tree-optimization/58626
2081         * tree-loop-distribution.c (enum rdg_dep_type): Remove
2082         anti_dd, output_dd and input_dd.
2083         (struct rdg_edge): Remove level and relation members.
2084         (RDGE_LEVEL, RDGE_RELATION): Remove.
2085         (dot_rdg_1): Adjust.
2086         (create_rdg_edge_for_ddr): Remove.
2087         (create_rdg_edges_for_scalar): Adjust.
2088         (create_edge_for_control_dependence): Likewise.
2089         (create_rdg_edges): Split into ...
2090         (create_rdg_flow_edges): ... this
2091         (create_rdg_cd_edges): ... and this.
2092         (free_rdg): Adjust.
2093         (build_rdg): Likewise, do not compute data dependences or
2094         add edges for them.
2095         (pg_add_dependence_edges): New function.
2096         (pgcmp): Likewise.
2097         (distribute_loop): First apply all non-dependence based
2098         partition mergings.  Then compute dependences between partitions
2099         and merge and order partitions according to them.
2101 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2103         PR rtl-optimization/58831
2104         * alias.c (init_alias_analysis): At the beginning of each iteration,
2105         set the reg_seen[N] bit if static_reg_base_value[N] is non-null.
2107 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2109         * recog.c (search_ofs): New static variable moved from...
2110         (peep2_find_free_register): ...here.
2111         (peephole2_optimize): Initialize it.
2113 2013-10-25  Tobias Burnus  <burnus@net-b.de>
2115         * doc/invoke.texi (fopenmp): Change supported OpenMP version to 4.0.
2117 2013-10-25  Uros Bizjak  <ubizjak@gmail.com>
2119         * config/i386/i386.h (TARGET_MPX): New define.
2120         (TARGET_MPX_P): Ditto.
2122 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2124         * config/i386/constraints.md (B): New.
2125         (Ti): New.
2126         (Tb): New.
2127         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
2128         * config/i386/i386-modes.def (BND32): New.
2129         (BND64): New.
2130         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
2131         * config/i386/i386.c (isa_opts): Add mmpx.
2132         (regclass_map): Add bound registers.
2133         (dbx_register_map): Likewise.
2134         (dbx64_register_map): Likewise.
2135         (svr4_dbx_register_map): Likewise.
2136         (PTA_MPX): New.
2137         (ix86_option_override_internal): Support MPX ISA.
2138         (ix86_conditional_register_usage): Support bound registers.
2139         (print_reg): Likewise.
2140         (ix86_code_end): Add MPX bnd prefix.
2141         (output_set_got): Likewise.
2142         (ix86_output_call_insn): Likewise.
2143         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
2144         (ix86_print_operand_punct_valid_p): Likewise.
2145         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
2146         UNSPEC_BNDMK_ADDR.
2147         (ix86_class_likely_spilled_p): Add bound regs support.
2148         (ix86_hard_regno_mode_ok): Likewise.
2149         (x86_order_regs_for_local_alloc): Likewise.
2150         (ix86_bnd_prefixed_insn_p): New.
2151         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
2152         (FIXED_REGISTERS): Add bound registers.
2153         (CALL_USED_REGISTERS): Likewise.
2154         (REG_ALLOC_ORDER): Likewise.
2155         (HARD_REGNO_NREGS): Likewise.
2156         (TARGET_MPX): New.
2157         (VALID_BND_REG_MODE): New.
2158         (FIRST_BND_REG): New.
2159         (LAST_BND_REG): New.
2160         (reg_class): Add BND_REGS.
2161         (REG_CLASS_NAMES): Likewise.
2162         (REG_CLASS_CONTENTS): Likewise.
2163         (BND_REGNO_P): New.
2164         (ANY_BND_REG_P): New.
2165         (BNDmode): New.
2166         (HI_REGISTER_NAMES): Add bound registers.
2167         * config/i386/i386.md (UNSPEC_BNDMK): New.
2168         (UNSPEC_BNDMK_ADDR): New.
2169         (UNSPEC_BNDSTX): New.
2170         (UNSPEC_BNDLDX): New.
2171         (UNSPEC_BNDLDX_ADDR): New.
2172         (UNSPEC_BNDCL): New.
2173         (UNSPEC_BNDCU): New.
2174         (UNSPEC_BNDCN): New.
2175         (UNSPEC_MPX_FENCE): New.
2176         (BND0_REG): New.
2177         (BND1_REG): New.
2178         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
2179         (length_immediate): Likewise.
2180         (prefix_0f): Likewise.
2181         (memory): Likewise.
2182         (prefix_rep): Check for bnd prefix.
2183         (length_nobnd): New.
2184         (length): Use length_nobnd if specified.
2185         (BND): New.
2186         (bnd_ptr): New.
2187         (BNDCHECK): New.
2188         (bndcheck): New.
2189         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
2190         (*jcc_2): Likewise.
2191         (jump): Likewise.
2192         (simple_return_internal): Likewise.
2193         (simple_return_pop_internal): Likewise.
2194         (*indirect_jump): Add MPX bnd prefix.
2195         (*tablejump_1): Likewise.
2196         (simple_return_internal_long): Likewise.
2197         (simple_return_indirect_internal): Likewise.
2198         (<mode>_mk): New.
2199         (*<mode>_mk): New.
2200         (mov<mode>): New.
2201         (*mov<mode>_internal_mpx): New.
2202         (<mode>_<bndcheck>): New.
2203         (*<mode>_<bndcheck>): New.
2204         (<mode>_ldx): New.
2205         (*<mode>_ldx): New.
2206         (<mode>_stx): New.
2207         (*<mode>_stx): New.
2208         * config/i386/predicates.md (lea_address_operand): Rename to...
2209         (address_no_seg_operand): ... this.
2210         (address_mpx_no_base_operand): New.
2211         (address_mpx_no_index_operand): New.
2212         (bnd_mem_operator): New.
2213         * config/i386/i386.opt (mmpx): New.
2214         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
2215         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
2217 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2219         * mode-classes.def (MODE_POINTER_BOUNDS): New.
2220         * tree.def (POINTER_BOUNDS_TYPE): New.
2221         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
2222         (POINTER_BOUNDS_MODE): New.
2223         (make_pointer_bounds_mode): New.
2224         * machmode.h (POINTER_BOUNDS_MODE_P): New.
2225         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
2226         (layout_type): Support POINTER_BOUNDS_TYPE.
2227         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
2228         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
2229         (type_contains_placeholder_1): Likewise.
2230         * tree.h (POINTER_BOUNDS_TYPE_P): New.
2231         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
2232         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
2234 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2236         * expr.c (expand_expr_real_1): Use mode of memory reference rather than
2237         mode of address computation when calling memory_address_addr_space.
2239 2013-08-24  Richard Henderson  <rth@twiddle.net>
2241         PR rtl/58542
2242         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
2243         instead of create_convert_operand_to.
2244         (maybe_emit_sync_lock_test_and_set): Likewise.
2245         (expand_atomic_compare_and_swap): Likewise.
2246         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
2248 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
2250         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
2251         Change param_is to use the struct and not the pointer to the struct.
2253 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
2255         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
2256         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2257         Move to targhooks.c.
2258         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
2259         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2260         Relocate from gimplify.c.
2261         * targhooks.h: Add 2 prototypes.
2262         * tree.h: Delete 2 prototypes.
2264 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2266         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
2267         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
2269 2013-10-24  Cong Hou  <congh@google.com>
2271         * convert.c (convert_to_real): Guard those unsafe math function
2272         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
2273         specially.
2275 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
2277         PR ipa/58712
2278         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
2279         as argument.
2280         (cgraph_create_edge): Use the new argument.
2281         (cgraph_create_indirect_edge): Likewise.
2283 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2285         * gcc/config/arc/arc.c (arc_ccfsm_post_advance): Also handle
2286         TYPE_UNCOND_BRANCH.
2287         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
2288         changes statep->state.
2290 2013-10-24  Tobias Burnus  <burnus@net-b.de>
2292         PR other/33426
2293         * tree-cfg.c (replace_loop_annotate): New function.
2294         (execute_build_cfg): Call it.
2295         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
2296         * internal-fn.c (expand_ANNOTATE): New function.
2297         * internal-fn.def (ANNOTATE): Define as new internal function.
2298         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
2299         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
2300         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
2301         * doc/extend.texi (Pragmas): Document #pragma ivdep.
2302         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
2304 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
2305             Marcus Shawcroft  <marcus.shawcroft@arm.com>
2307         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
2308         Special case reload SP+C into none GENERAL_REGS.
2310 2013-10-24  Michael Matz  <matz@suse.de>
2312         * gengtype.c (is_file_equal): Check that files will be same length.
2314 2013-10-25  Christian Bruel  <christian.bruel@st.com>
2316         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
2317         * gcc/config/sh/t-sh (sh-mem.o): New rule.
2318         * gcc/config/sh/sh-mem.cc (expand_block_move): Moved here.
2319         (sh_expand_cmpstr): New function.
2320         * gcc/config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c.
2321         * gcc/config/sh/sh-protos.h (sh_expand_cmpstr): Declare.
2322         * gcc/config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
2323         (rotlhi3_8): Rename.
2325 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2327         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
2328         * Makefile.in (CXXDEPMODE): Assign and change users.
2329         (CCDEPMODE): Delete.
2330         * configure: Regenerate.
2332 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
2334         * gengtype-parse.c (require_without_advance): New.
2335         (type): For GTY-marked types that are not GTY((user)), parse any
2336         base classes, requiring them to be single-inheritance, and not
2337         be templates.  For non-GTY-marked types and GTY((user)),
2338         continue to skip over any C++ inheritance specification.
2339         * gengtype-state.c (state_writer::write_state_struct_type):
2340         Write base class of type (if any).
2341         (read_state_struct_type): Read base class of type (if any).
2342         * gengtype.c (new_structure): Add a "base_class" parameter.
2343         (create_optional_field_): Update for new parameter to new_structure.
2344         (adjust_field_rtx_def): Likewise.
2345         (adjust_field_tree_exp): Likewise.
2346         * gengtype.h (struct type): Add "base_class" field to the s
2347         union field.
2348         (new_structure): Add "base" parameter.
2350 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
2352         PR target/57756
2353         * config/i386/i386.c (ix86_option_override_internal):
2354         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
2355         (ix86_valid_target_attribute_tree):
2356         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
2357         Change TARGET_SSE to TARGET_SSE_P (opts->...)
2359 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2361         * tree-ssa-loop.h: Remove include files.
2362         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2363         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
2364         * graphite-clast-to-gimple.c: Likewise.
2365         * graphite-scop-detection.c: Likewise.
2366         * graphite-sese-to-poly.c: Likewise.
2367         * ipa-inline-analysis.c: Likewise.
2368         * ipa-pure-const.c: Likewise.
2369         * loop-init.c: Likewise.
2370         * passes.c: Likewise.
2371         * predict.c: Likewise.
2372         * tree-cfg.c: Likewise.
2373         * tree-cfgcleanup.c: Likewise.
2374         * tree-chrec.c: Likewise.
2375         * tree-data-ref.c: Likewise.
2376         * tree-loop-distribution.c: Likewise.
2377         * tree-parloops.c: Likewise.
2378         * tree-predcom.c: Likewise.
2379         * tree-scalar-evolution.c: Likewise.
2380         * tree-ssa-address.c: Likewise.
2381         * tree-ssa.c: Likewise.
2382         * tree-ssa-dce.c: Likewise.
2383         * tree-ssa-loop.c: Likewise.
2384         * tree-ssa-loop-im.c: Likewise.
2385         * tree-ssa-loop-ivcanon.c: Likewise.
2386         * tree-ssa-loop-ivopts.c: Likewise.
2387         * tree-ssa-loop-manip.c: Likewise.
2388         * tree-ssa-loop-niter.c: Likewise.
2389         * tree-ssa-loop-prefetch.c: Likewise.
2390         * tree-ssa-loop-unswitch.c: Likewise.
2391         * tree-ssa-reassoc.c: Likewise.
2392         * tree-vect-data-refs.c: Likewise.
2393         * tree-vect-loop.c: Likewise.
2394         * tree-vect-loop-manip.c: Likewise.
2395         * tree-vectorizer.c: Likewise.
2396         * tree-vect-stmts.c: Likewise.
2397         * tree-vrp.c: Likewise.
2399 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2401         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
2403 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2405         PR tree-optimization/58775
2406         PR tree-optimization/58791
2407         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
2408         (insert_stmt_after): Rewritten, don't move the stmt, but really
2409         insert it.
2410         (get_stmt_uid_with_default): Remove.
2411         (build_and_add_sum): Use insert_stmt_after and
2412         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only labels.
2413         (update_range_test): Set uid on stmts added by
2414         force_gimple_operand_gsi.  Don't immediately modify statements
2415         in inter-bb optimization, just update oe->op values.
2416         (optimize_range_tests): Return bool whether any changed have been made.
2417         (update_ops): New function.
2418         (struct inter_bb_range_test_entry): New type.
2419         (maybe_optimize_range_tests): Perform statement changes here.
2420         (not_dominated_by, appears_later_in_bb, get_def_stmt,
2421         ensure_ops_are_available): Remove.
2422         (find_insert_point): Rewritten.
2423         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
2424         return LHS of the (new resp. old) stmt.  Don't call
2425         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
2426         instead of last, move new stmt at the right place.
2427         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
2428         (negate_value): Likewise.  Set uids.
2429         (break_up_subtract_bb): Initialize uids.
2430         (reassociate_bb): Adjust rewrite_expr_tree caller.
2431         (do_reassoc): Don't call renumber_gimple_stmt_uids.
2433 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
2435         PR target/58838
2436         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
2437         TARGET_32BIT final condition.
2438         (mulsi3_internal2 and splitter): Same.
2440 2013-10-23  Jeff Law  <law@redhat.com>
2442         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
2443         through joiner blocks with abnormal outgoing edges.
2445         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
2446         Add parameter JOINERS, to allow/disallow threading through joiner
2447         blocks.
2448         (thread_block): New.  Call thread_block_1.
2449         (mark_threaded_blocks): Remove code to filter out certain cases
2450         of threading through joiner blocks.
2451         (thread_through_all_blocks): Document how we can have a dangling
2452         edge AUX field and clear it.
2454 2013-10-23  Ian Lance Taylor  <iant@google.com>
2456         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
2457         (C++ Dialect Options): Likewise.
2458         (Optimize Options): Likewise.
2460 2013-10-23  Tom de Vries  <tom@codesourcery.com>
2462         PR tree-optimization/58805
2463         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
2465 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2467         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
2468         sequence based on get_range_info returned range.
2470 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2472         * tree-ssa.h: Remove all #include's
2473         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2474         * alias.c: Move required includes from tree-ssa.h.
2475         * asan.c: Likewise.
2476         * builtins.c: Likewise.
2477         * calls.c: Likewise.
2478         * cfgexpand.c: Likewise.
2479         * cfghooks.c: Likewise.
2480         * cfgloop.c: Likewise.
2481         * cfgloopmanip.c: Likewise.
2482         * cgraph.c: Likewise.
2483         * cgraphbuild.c: Likewise.
2484         * cgraphclones.c: Likewise.
2485         * cgraphunit.c: Likewise.
2486         * dse.c: Likewise.
2487         * except.c: Likewise.
2488         * expr.c: Likewise.
2489         * final.c: Likewise.
2490         * fold-const.c: Likewise.
2491         * ggc-page.c: Likewise.
2492         * gimple-builder.c: Likewise.
2493         * gimple-fold.c: Likewise.
2494         * gimple-iterator.c: Likewise.
2495         * gimple-low.c: Likewise.
2496         * gimple-pretty-print.c: Likewise.
2497         * gimple-ssa-strength-reduction.c: Likewise.
2498         * gimple-streamer-in.c: Likewise.
2499         * gimple-streamer-out.c: Likewise.
2500         * gimplify.c: Likewise.
2501         * graphite-blocking.c: Likewise.
2502         * graphite-clast-to-gimple.c: Likewise.
2503         * graphite-dependences.c: Likewise.
2504         * graphite-interchange.c: Likewise.
2505         * graphite-optimize-isl.c: Likewise.
2506         * graphite-poly.c: Likewise.
2507         * graphite-scop-detection.c: Likewise.
2508         * graphite-sese-to-poly.c: Likewise.
2509         * graphite.c: Likewise.
2510         * ipa-cp.c: Likewise.
2511         * ipa-inline-analysis.c: Likewise.
2512         * ipa-inline-transform.c: Likewise.
2513         * ipa-inline.c: Likewise.
2514         * ipa-prop.c: Likewise.
2515         * ipa-pure-const.c: Likewise.
2516         * ipa-reference.c: Likewise.
2517         * ipa-split.c: Likewise.
2518         * ipa-utils.c: Likewise.
2519         * loop-init.c: Likewise.
2520         * lto-cgraph.c: Likewise.
2521         * lto-section-in.c: Likewise.
2522         * lto-section-out.c: Likewise.
2523         * lto-streamer-in.c: Likewise.
2524         * lto-streamer-out.c: Likewise.
2525         * lto-streamer.c: Likewise.
2526         * omp-low.c: Likewise.
2527         * passes.c: Likewise.
2528         * predict.c: Likewise.
2529         * print-tree.c: Likewise.
2530         * profile.c: Likewise.
2531         * sese.c: Likewise.
2532         * targhooks.c: Likewise.
2533         * tracer.c: Likewise.
2534         * trans-mem.c: Likewise.
2535         * tree-call-cdce.c: Likewise.
2536         * tree-cfg.c: Likewise.
2537         * tree-cfgcleanup.c: Likewise.
2538         * tree-chrec.c: Likewise.
2539         * tree-complex.c: Likewise.
2540         * tree-data-ref.c: Likewise.
2541         * tree-dfa.c: Likewise.
2542         * tree-eh.c: Likewise.
2543         * tree-emutls.c: Likewise.
2544         * tree-if-conv.c: Likewise.
2545         * tree-inline.c: Likewise.
2546         * tree-into-ssa.c: Likewise.
2547         * tree-loop-distribution.c: Likewise.
2548         * tree-mudflap.c: Likewise.
2549         * tree-nested.c: Likewise.
2550         * tree-nrv.c: Likewise.
2551         * tree-object-size.c: Likewise.
2552         * tree-outof-ssa.c: Likewise.
2553         * tree-parloops.c: Likewise.
2554         * tree-phinodes.c: Likewise.
2555         * tree-predcom.c: Likewise.
2556         * tree-pretty-print.c: Likewise.
2557         * tree-profile.c: Likewise.
2558         * tree-scalar-evolution.c: Likewise.
2559         * tree-sra.c: Likewise.
2560         * tree-ssa-address.c: Likewise.
2561         * tree-ssa-alias.c: Likewise.
2562         * tree-ssa-ccp.c: Likewise.
2563         * tree-ssa-coalesce.c: Likewise.
2564         * tree-ssa-copy.c: Likewise.
2565         * tree-ssa-copyrename.c: Likewise.
2566         * tree-ssa-dce.c: Likewise.
2567         * tree-ssa-dom.c: Likewise.
2568         * tree-ssa-dse.c: Likewise.
2569         * tree-ssa-forwprop.c: Likewise.
2570         * tree-ssa-ifcombine.c: Likewise.
2571         * tree-ssa-live.c: Likewise.
2572         * tree-ssa-loop-ch.c: Likewise.
2573         * tree-ssa-loop-im.c: Likewise.
2574         * tree-ssa-loop-ivcanon.c: Likewise.
2575         * tree-ssa-loop-ivopts.c: Likewise.
2576         * tree-ssa-loop-manip.c: Likewise.
2577         * tree-ssa-loop-niter.c: Likewise.
2578         * tree-ssa-loop-prefetch.c: Likewise.
2579         * tree-ssa-loop-unswitch.c: Likewise.
2580         * tree-ssa-loop.c: Likewise.
2581         * tree-ssa-math-opts.c: Likewise.
2582         * tree-ssa-operands.c: Likewise.
2583         * tree-ssa-phiopt.c: Likewise.
2584         * tree-ssa-phiprop.c: Likewise.
2585         * tree-ssa-pre.c: Likewise.
2586         * tree-ssa-propagate.c: Likewise.
2587         * tree-ssa-reassoc.c: Likewise.
2588         * tree-ssa-sccvn.c: Likewise.
2589         * tree-ssa-sink.c: Likewise.
2590         * tree-ssa-strlen.c: Likewise.
2591         * tree-ssa-structalias.c: Likewise.
2592         * tree-ssa-tail-merge.c: Likewise.
2593         * tree-ssa-ter.c: Likewise.
2594         * tree-ssa-threadedge.c: Likewise.
2595         * tree-ssa-threadupdate.c: Likewise.
2596         * tree-ssa-uncprop.c: Likewise.
2597         * tree-ssa-uninit.c: Likewise.
2598         * tree-ssa.c: Likewise.
2599         * tree-ssanames.c: Likewise.
2600         * tree-stdarg.c: Likewise.
2601         * tree-streamer-in.c: Likewise.
2602         * tree-switch-conversion.c: Likewise.
2603         * tree-tailcall.c: Likewise.
2604         * tree-vect-data-refs.c: Likewise.
2605         * tree-vect-generic.c: Likewise.
2606         * tree-vect-loop-manip.c: Likewise.
2607         * tree-vect-loop.c: Likewise.
2608         * tree-vect-patterns.c: Likewise.
2609         * tree-vect-slp.c: Likewise.
2610         * tree-vect-stmts.c: Likewise.
2611         * tree-vectorizer.c: Likewise.
2612         * tree-vrp.c: Likewise.
2613         * tree.c: Likewise.
2614         * tsan.c: Likewise.
2615         * value-prof.c: Likewise.
2616         * var-tracking.c: Likewise.
2617         * varpool.c: Likewise.
2618         * vtable-verify.c: Likewise.
2620 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2622         * config/tilegx/tilegx.c: Include "tree.h".
2624 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2626         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
2627         the info, not after it.
2628         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
2629         dump_ssaname_info call.
2630         (pp_gimple_stmt_1): Adjust caller.
2631         (dump_phi_nodes): Likewise.  Don't print "# " here.
2633 2013-10-22  Jan Hubicka  <jh@suse.cz>
2635         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New
2636         tuning flag.
2637         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
2638         * i386.c (expand_small_movmem_or_setmem): New function.
2639         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New
2640         function.
2641         (alg_usable_p): Add support for value ranges; cleanup.
2642         (ix86_expand_set_or_movmem): Add support for misaligned moves.
2644 2013-10-22  Sterling Augustine  <saugustine@google.com>
2646         * doc/invoke.texi: Document -ggnu-pubnames.
2647         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
2648         logic.
2649         * dwarf2out.c: Include gdb/gdb-index.h.
2650         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
2651         debug_generate_pub_sections.
2652         (is_java, output_pubtables, output_pubname): New functions.
2653         (include_pubname_in_output): Handle debug_generate_pub_sections at
2654         level 2.
2655         (size_of_pubnames): Use new local space_for_flags based on
2656         debug_generate_pub_sections.
2657         (output_pubnames): Unify pubnames and pubtypes output logic.
2658         Genericize comments.  Call output_pubname.
2659         (dwarf2out_finish): Move logic to output_pubnames and call it.
2661 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
2663         PR target/58779
2664         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
2665         Remove CCCmode handling.
2666         <case LTU>: Return 'c' suffix for CCCmode.
2667         <case GEU>: Return 'nc' suffix for CCCmode.
2668         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
2669         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
2670         (*sub<mode>3_cc_overflow): Ditto.
2671         (*subsi3_zext_cc_overflow): Ditto.
2673 2013-10-22  Steve Ellcey  <sellcey@mips.com>
2675         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
2676         (AND (NOT OP1) (NOT OP2)).
2678 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
2680         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2681         meaning of merge-high and merge-low masks for little endian; avoid
2682         use of vector-pack masks for little endian for mismatched modes.
2684 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2686         * config/tilepro/tilepro.c: Include "tree.h".
2688 2013-10-22  Andreas Schwab  <schwab@suse.de>
2690         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
2691         with PRE_DEC.
2693 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
2695         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
2696         [Ed Smith-Rowland]): New entries.
2697         ([Stephen M. Webb]): Update.
2699 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
2701         * tree-ssa-ter.h: Remove duplicate copy of file contents.
2703 2013-10-21  Marek Polacek  <polacek@redhat.com>
2705         PR middle-end/58809
2706         * fold-const.c (fold_range_test): Return 0 if the type is not
2707         an integral type.
2709 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
2711         * system.h: Move hwint.h include further down.
2712         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
2713         gcc_checking_asserts.
2714         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
2716 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2718         Fix volatile issues in optimize_bit_field_compare.
2719         * fold-const.c (optimize_bit_field_compare): Bail out if
2720         lvolatilep or rvolatilep.
2722 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2724         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
2725         and get_inner_reference returning different pmode for non-volatile
2726         bit-field members dependent on flag_strict_volatile_bitfields.
2727         * stor-layout.c (layout_decl): Remove special handling of
2728         flag_strict_volatile_bitfields.
2729         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
2730         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
2732 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
2734         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
2735         calculations.
2737 2013-10-21  Jeff Law  <law@redhat.com>
2739         * tree-ssa-threadedge.c (thread_through_normal_block): New
2740         argument VISITED.  Remove VISISTED as a local variable.  When we
2741         have a threadable jump, verify the destination of the jump has not
2742         been visised.
2743         (thread_across_edge): Allocate VISITED bitmap once at function
2744         scope and use it throughout.  Make sure to set appropriate bits in
2745         VISITED for E (start of jump thread path).
2746         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
2747         through a joiner if any edge on the path has a recorded jump thread.
2749 2013-10-21  Ian Lance Taylor  <iant@google.com>
2751         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
2752         don't imply that attributes can solve all problems.
2753         (Directory Options): Fix typo.
2755 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2757         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
2758         extend and extend_add.
2760 2013-10-21  Richard Biener  <rguenther@suse.de>
2762         PR tree-optimization/58794
2763         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
2764         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
2766 2013-10-21  Richard Biener  <rguenther@suse.de>
2768         PR middle-end/58742
2769         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
2770         to (T) X for sign-changing conversions (or no conversion).
2772 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
2774         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
2776 2013-10-20  Jan Hubicka  <jh@suse.cz>
2778         * config/i386/i386-tune.def: Add comment; organize into categories
2780 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
2782         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
2783         argument.  Update function comment.
2784         (expand_set_or_movmem_via_rep): New function combining
2785         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
2786         (expand_movmem_via_rep_mov): Remove.
2787         expand_setmem_via_rep_stos): Remove.
2788         (expand_movmem_epilogue): Update calls correspondingly.
2789         (expand_setmem_epilogue_via_loop): Likewise.
2790         (emit_memset): New.
2791         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
2792         (expand_set_or_movmem_prologue): New function combining
2793         expand_movmem_prologue and expand_setmem_prologue.
2794         (expand_movmem_prologue): Remove.
2795         (expand_setmem_prologue): Remove.
2796         (expand_set_or_movmem_constant_prologue): New function combining
2797         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
2798         (expand_constant_movmem_prologue): Remove.
2799         (expand_constant_setmem_prologue): Remove.
2800         (promote_duplicated_reg): Allow vector-const0 value.
2801         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
2802         and ix86_expand_setmem.
2803         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
2804         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
2806 2013-10-21  Diego Novillo  <dnovillo@google.com>
2808         * asan.c: Include tree.h
2809         * bb-reorder.c: Likewise.
2810         * cfgcleanup.c: Likewise.
2811         * cfgloopmanip.c: Likewise.
2812         * data-streamer-in.c: Likewise.
2813         * data-streamer-out.c: Likewise.
2814         * data-streamer.c: Likewise.
2815         * dwarf2cfi.c: Likewise.
2816         * graphite-blocking.c: Likewise.
2817         * graphite-clast-to-gimple.c: Likewise.
2818         * graphite-dependences.c: Likewise.
2819         * graphite-interchange.c: Likewise.
2820         * graphite-optimize-isl.c: Likewise.
2821         * graphite-poly.c: Likewise.
2822         * graphite-scop-detection.c: Likewise.
2823         * graphite-sese-to-poly.c: Likewise.
2824         * graphite.c: Likewise.
2825         * ipa-devirt.c: Likewise.
2826         * ipa-profile.c: Likewise.
2827         * ipa.c: Likewise.
2828         * ira.c: Likewise.
2829         * loop-init.c: Likewise.
2830         * loop-unroll.c: Likewise.
2831         * lower-subreg.c: Likewise.
2832         * lto/lto-object.c: Likewise.
2833         * recog.c: Likewise.
2834         * reginfo.c: Likewise.
2835         * tree-loop-distribution.c: Likewise.
2836         * tree-parloops.c: Likewise.
2837         * tree-ssa-strlen.c: Likewise.
2838         * tree-streamer.c: Likewise.
2839         * value-prof.c: Likewise.
2840         * target-globals.c: Likewise.
2841         * expr.h: Include tree-core.h instead of tree.h.
2842         * gimple.h: Likewise.
2843         * ipa-prop.h: Likewise.
2844         * ipa-utils.h: Likewise.
2845         * lto-streamer.h: Likewise.
2846         * streamer-hooks.h: Likewise.
2847         * ipa-reference.h: Include cgraph.h instead of tree.h.
2848         * cgraph.h: Include basic-block.h instead of tree.h.
2849         * tree-streamer.h: Do not include tree.h.
2850         * genattrtab.c (write_header): Generate inclusion of
2851         tree.h.
2852         * genautomata.c (main): Likewise.
2853         * genemit.c: Likewise.
2854         * genopinit.c: Likewise.
2855         * genoutput.c (output_prologue): Likewise.
2856         * genpeep.c: Likewise.
2858 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2860         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
2861         little endian.
2862         (vec_unpacku_hi_v8hi): Likewise.
2863         (vec_unpacku_lo_v16qi): Likewise.
2864         (vec_unpacku_lo_v8hi): Likewise.
2866 2013-10-20  Jan Hubicka  <jh@suse.cz>
2868         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
2869         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
2870         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
2872 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
2874         * config/mips/mips.h (ISA_HAS_WSBH): Define.
2875         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
2876         constants.
2877         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
2879 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
2881         PR target/58603
2882         * system.h: Undef m_slot.
2884 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2886         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
2887         all elements for both endian flavors.
2889 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2891         PR target/58792
2892         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
2893         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
2894         and SI_REG for 64bit SYSV ABI targets.
2896 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2898         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
2899         to multi_reg_return.  Clarify that we are skipping USEs of multiple
2900         return registers.  Use bool type where appropriate.
2902 2013-10-18  Jan Hubicka  <jh@suse.cz>
2904         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
2905         for cold functions.
2906         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
2907         (X86_TUNE_PUSH_MEMORY): Likewise.
2908         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
2909         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
2910         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
2911         New.
2912         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
2913         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
2914         Remove.
2915         (ix86_option_override_internal): Update to use tune features instead
2916         of variables.
2918 2013-10-18  Cong Hou  <congh@google.com>
2920         PR tree-optimization/58508
2921         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
2922         statement that contains data refs with zero-step.
2924 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
2926         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
2927         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
2928         * gimple-low.c (gimple_check_call_arg,
2929         gimple_check_call_matching_types): Move to cgraph.c.
2930         * gimple-low.h: Remove prototype.
2931         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
2932         Relocate from gimple-low.c.
2933         * cgraph.h: Add prototype.  Don't include basic-block.h.
2934         * gimplify.c: Add gimple-low to include list.
2935         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
2936         * tree-eh.c: Add gimple-low to include list.
2937         * tree-nested.c: Likewise.
2938         * cfgexpand.c: Add tree-ssa-address.h to include list.
2939         * expr.c: Likewise.
2940         * gimple-fold.c: Likewise.
2941         * gimple-ssa-strength-reduction.c: Likewise.
2942         * trans-mem.c: Likewise.
2943         * tree-mudflap.c: Likewise.
2944         * tree-ssa-loop-ivopts.c: Likewise.
2945         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
2946         (degenerate_phi_result): Move to tree-phinodes.c.
2947         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
2948         * tree-ssa-threadedge.c: Likewise.
2949         * tree-vrp.c: Likewise.
2950         * tree-phinodes.c (degenerate_phi_result): Relocate here.
2951         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
2952         * tree-phinodes.h (degenerate_phi_result): Add prototype.
2953         * tree-ssa-copy.c: Include tree-ssa-dom.h.
2954         * tree-ssa-forwprop.c: Likewise.
2955         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
2956         pass_data_cleanup_cfg_post_optimizing,
2957         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
2958         * tree-optimize.c: Delete File.
2959         * graphite.c: Include tree-cfgcleanup.h.
2960         * passes.c: Likewise.
2961         * tree-cfg.c: Likewise.
2962         * tree-profile.c: Likewise.
2963         * tree-ssa-dse.c: Likewise.
2964         * tree-ssa-loop-ivcanon.c: Likewise.
2965         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
2966         * tree-outof-ssa.c: Include sbitmap.h.
2967         * tree-ssa-live.c: Likewise.
2968         * tree-ssa-propagate.c: Likewise.
2969         * tree-ssa-structalias.c: Likewise.
2970         * tree-stdarg.c: Likewise.
2971         * Makefile.in (OBJS): Delete tree-optimize.o.
2972         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
2973         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
2974         * varasm.c: Include basic-block.h.
2975         * cfgloop.h: Include function.h instead of basic-block.h
2976         (bb_loop_depth): Move to cfgloop.c.
2977         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
2979 2013-10-18  Teresa Johnson  <tejohnson@google.com>
2981         * predict.c (probably_never_executed): Compare frequency-based
2982         count to number of training runs.
2983         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
2985 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2987         * config/arm/arm.c (cortexa9_extra_costs): New table.
2988         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
2990 2013-10-18  Jeff Law  <law@redhat.com>
2992         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
2994         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
2995         (register_jump_thread): Add "registered_jump_thread" debug
2996         counter support.
2997         * dbgcnt.def (registered_jump_thread): New debug counter.
2999 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3001         * config/rs6000/rs6000.c: Include cgraph.h.
3003 2013-10-18  Teresa Johnson  <tejohnson@google.com>
3005         * tree-ssa-tail-merge.c (replace_block_by): Update edge
3006         weights during merging.
3008 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3010         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
3011         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
3012         * tree-ssa-operands.h: Remove prototype.
3013         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
3014         * gimple.c (virtual_operand_p): Relocate from gimple.c.
3015         * gimple.h: Add prototype.
3016         * gimple-ssa.h: Include tree-ssa-operands.h.
3017         * tree-dump.c: Add tree-cfg.h to include list.
3018         * tree-ssa-alias.c: Add ipa-reference.h to include list.
3019         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
3020         * config/i386/i386.c: Don't include tree-flow.h.
3021         * config/rs6000/rs6000.c: Likewise.
3023 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3025         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
3027 2013-10-18  Richard Biener  <rguenther@suse.de>
3029         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
3030         or TYPE_UNSIGNED of integral types.
3031         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
3032         NULL_TREE for zero-precision integral types.
3034 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
3036         * config/aarch64/arm_neon.h
3037         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
3039 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
3041         PR target/57756
3042         * opth-gen.awk: Define target_flags_explicit.
3044 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3046         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
3047         fields to the reg_addr array that describes the valid addressing
3048         mode for any register, general purpose registers, floating point
3049         registers, and Altivec registers.
3050         (FIRST_RELOAD_REG_CLASS): Likewise.
3051         (LAST_RELOAD_REG_CLASS): Likewise.
3052         (struct reload_reg_map_type): Likewise.
3053         (reload_reg_map_type): Likewise.
3054         (RELOAD_REG_VALID): Likewise.
3055         (RELOAD_REG_MULTIPLE): Likewise.
3056         (RELOAD_REG_INDEXED): Likewise.
3057         (RELOAD_REG_OFFSET): Likewise.
3058         (RELOAD_REG_PRE_INCDEC): Likewise.
3059         (RELOAD_REG_PRE_MODIFY): Likewise.
3060         (reg_addr): Likewise.
3061         (mode_supports_pre_incdec_p): New helper functions to say whether
3062         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
3063         (mode_supports_pre_modify_p): Likewise.
3064         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
3065         print the valid address mode bits for each mode.
3066         (rs6000_debug_print_mode): Likewise.
3067         (rs6000_debug_reg_global): Likewise.
3068         (rs6000_setup_reg_addr_masks): New function to set up the address
3069         mask bits for each type.
3070         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
3071         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
3072         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
3073         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
3074         PRE_MODIFY are supported.
3075         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
3076         registers, instead of {src,dest}_av_p.
3077         (rs6000_print_options_internal): Tweak the debug output slightly.
3079 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3081         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
3082         isa attribute.
3084 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
3086         * tree-flow.h (struct omp_region): Move to omp-low.c.
3087         Remove omp_ prototypes and variables.
3088         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
3089         (copy_var_decl): Relocate prototype from tree-flow.h.
3090         * gimple.c (copy_var_decl): Relocate from omp-low.c.
3091         * tree.h: Move prototype to omp-low.h.
3092         * omp-low.h: New File.  Relocate prototypes here.
3093         * omp-low.c (struct omp_region): Make local here.
3094         (root_omp_region): Make static.
3095         (copy_var_decl) Move to gimple.c.
3096         (new_omp_region): Make static.
3097         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
3098         * tree-cfg.c: Include omp-low.h.
3099         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
3100         * gimplify.c: Include omp-low.h.
3101         * tree-parloops.c: Likewise.
3103 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3105         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
3106         operands are in memory, prefer to force non-matched operand 1 to
3107         the register.
3109 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3111         PR target/58673
3112         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
3113         restrict TImode addresses to single indirect registers if both
3114         -mquad-memory and -mvsx-timode are used.
3115         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
3116         we should emit load/store quad.  Remove using %y for quad memory
3117         addresses.
3119         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
3120         constraints to allow load/store quad on machines where TImode is
3121         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
3122         for TImode to load integer constants.
3124 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3126         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
3128 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3130         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
3131         handling of STACK_REG.
3133 2013-10-17  Richard Biener  <rguenther@suse.de>
3135         PR tree-optimization/58143
3136         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
3137         New function.
3138         (rewrite_to_defined_overflow): Likewise.
3139         (move_computations_dom_walker::before_dom): Rewrite stmts
3140         with undefined signed overflow that are not always executed
3141         into unsigned arithmetic.
3143 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3145         PR target/57756
3146         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
3147         explicit isa flag to be an options variable, instead of using
3148         global_options_set.  Remove define from rs6000.h.
3149         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
3151         * config/rs6000/rs6000.c (rs6000_option_override_internal):
3152         Initialize rs6000_isa_flags_explicit.
3153         (rs6000_function_specific_save): Add gcc_options* parameter, so
3154         that the powerpc builds after the 2013-10-15 changes.
3155         (rs6000_function_specific_restore): Likewise.
3157 2013-10-16  DJ Delorie  <dj@redhat.com>
3159         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
3160         op is a REG before checking REGNO.
3161         (rl78_alloc_physical_registers): Verify pattern is a SET before
3162         checking SET_SRC.
3164 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3166         * gcc/config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
3167         endianness.
3168         (vec_unpacks_lo_v4sf): Likewise.
3169         (vec_unpacks_float_hi_v4si): Likewise.
3170         (vec_unpacks_float_lo_v4si): Likewise.
3171         (vec_unpacku_float_hi_v4si): Likewise.
3172         (vec_unpacku_float_lo_v4si): Likewise.
3174 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3176         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
3177         (vsx_concat_v2sf): Likewise.
3179 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
3181         * config/aarch64/aarch64.md
3182         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
3184 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
3186         PR tree-optimization/58697
3187         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
3188         estimated_loop_iterations_int.
3189         (max_stmt_executions_int): Call get_max_loop_iterations_int.
3190         (get_max_loop_iterations_int): New.  HWINT version of
3191         get_max_loop_iterations.
3192         * cfgloop.h: Add prototypes.
3193         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
3194         * loop-unroll.c (decide_peel_once_rolling): Call
3195         get_estimated_loop_iterations_int.
3196         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
3197         * tree-ssa-loop-niter.h: Tweak prototypes.
3199 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
3201         * gengtype-parse.c (struct_field_seq): Ignore access-control
3202         keywords ("public:" etc).
3204 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3206         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
3207         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
3209 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
3211         * config/arm/arm.opt (mlra): New option.
3212         * config/arm/arm.c (arm_lra_p): New function.
3213         (TARGET_LRA_P): Define.
3215 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
3217         * tree-core.h (tree_code_name): Remove.
3218         * tree.h (get_tree_code_name): New prototype.
3219         * tree.c (tree_code_name): Make static.
3220         (get_tree_code_name): New function.
3221         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
3222         tree_class_check_failed, tree_range_check_failed,
3223         tree_not_class_check_failed, omp_clause_check_failed,
3224         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
3225         wrapper get_tree_code_name instead of calling tree_code_name directly.
3226         * tree-vrp.c (dump_asserts_for): Likewise.
3227         * tree-dump.c (dequeue_and_dump): Likewise.
3228         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
3229         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
3230         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
3231         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
3232         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
3233         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
3234         dump_gimple_omp_for): Likewise.
3235         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3236         * tree-ssa-pre.c (print_pre_expr): Likewise.
3237         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3238         * print-tree.c (print_node_brief, print_node): Likewise.
3239         * gimple.c (gimple_check_failed): Likewise.
3240         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
3241         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
3242         * config/mep/mep.c (mep_validate_vliw): Likewise.
3243         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
3244         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
3246 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
3248         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
3249         for AMD bdver3.
3251 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
3253         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
3254         (MULTILIB_MATCHES): Add multilib for -march=v8.
3256 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
3258         PR target/57756
3259         * optc-save-gen.awk: Add extra parameter to the save and restore
3260         target calls.
3261         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
3262         * tree.c (build_optimization_node): New parameter.  Add extra parameter
3263         to call to cl_optimization_save.
3264         (build_target_option_node): New parameter. Add extra parameter
3265         to call to cl_target_option_save.
3266         * tree.h (build_optimization_node): New parameter.
3267         (build_target_option_node): New parameter.
3268         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
3269         build_optimization_node and build_target_option_node.
3270         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
3271         (handle_pragma_push_options): Ditto.
3272         * toplev.c (process_options): Ditto.
3273         * opts.c (init_options_struct): Check for opts_set non-null.
3274         * target.def (target_option.save): New parameter.
3275         (target_option.restore): New parameter.
3276         * tm.texi: Generate.
3277         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3278         (ix86_pragma_target_parse): Ditto.
3279         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
3280         parameters.
3281         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
3282         to  build_optimization_node and build_target_option_node.
3283         (rs6000_valid_attribute_p): Ditto.
3284         (rs6000_pragma_target_parse): Ditto.
3285         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
3286         data.
3287         * config/i386/i386.h:
3288         TARGET_64BIT_P: New Macro
3289         TARGET_MMX_P: New Macro.
3290         TARGET_3DNOW_P: New Macro.
3291         TARGET_3DNOW_A_P: New Macro.
3292         TARGET_SSE_P: New Macro.
3293         TARGET_SSE2_P: New Macro.
3294         TARGET_SSE3_P: New Macro.
3295         TARGET_SSSE3_P: New Macro.
3296         TARGET_SSE4_1_P: New Macro.
3297         TARGET_SSE4_2_P: New Macro.
3298         TARGET_AVX_P: New Macro.
3299         TARGET_AVX2_P: New Macro.
3300         TARGET_AVX512F_P: New Macro.
3301         TARGET_AVX512PF_P: New Macro.
3302         TARGET_AVX512ER_P: New Macro.
3303         TARGET_AVX512CD_P: New Macro.
3304         TARGET_FMA_P: New Macro.
3305         TARGET_SSE4A_P: New Macro.
3306         TARGET_FMA4_P: New Macro.
3307         TARGET_XOP_P: New Macro.
3308         TARGET_LWP_P: New Macro.
3309         TARGET_ABM_P: New Macro.
3310         TARGET_BMI_P: New Macro.
3311         TARGET_BMI2_P: New Macro.
3312         TARGET_LZCNT_P: New Macro.
3313         TARGET_TBM_P: New Macro.
3314         TARGET_POPCNT_P: New Macro.
3315         TARGET_SAHF_P: New Macro.
3316         TARGET_MOVBE_P: New Macro.
3317         TARGET_CRC32_P: New Macro.
3318         TARGET_AES_P: New Macro.
3319         TARGET_PCLMUL_P: New Macro.
3320         TARGET_CMPXCHG16B_P: New Macro.
3321         TARGET_FSGSBASE_P: New Macro.
3322         TARGET_RDRND_P: New Macro.
3323         TARGET_F16C_P: New Macro.
3324         TARGET_RTM_P: New Macro.
3325         TARGET_HLE_P: New Macro.
3326         TARGET_RDSEED_P: New Macro.
3327         TARGET_PRFCHW_P: New Macro.
3328         TARGET_ADX_P: New Macro.
3329         TARGET_FXSR_P: New Macro.
3330         TARGET_XSAVE_P: New Macro.
3331         TARGET_XSAVEOPT_P: New Macro.
3332         TARGET_LP64_P: New Macro.
3333         TARGET_X32_P: New Macro.
3334         TARGET_FPMATH_DEFAULT_P: New Macro.
3335         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
3336         * config/i386/i386.c (ix86_option_override_internal): New parameters.
3337         opts and opts_set.
3338         Change ix86_tune_string to access opts->x_ix86_tune_string.
3339         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
3340         Change ix86_arch_string to access opts->x_ix86_arch_string.
3341         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
3342         Change ix86_pmode to access opts->x_ix86_pmode.
3343         Change ix86_abi to access opts->x_ix86_abi.
3344         Change ix86_cmodel to access opts->x_ix86_cmodel.
3345         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
3346         Change ix86_isa_flags_explicit to access
3347         opts->x_ix86_isa_flags_explicit.
3348         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
3349         Change ix86_regparm to access opts->x_ix86_regparm.
3350         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
3351         Change ix86_preferred_stack_boundary_arg to access
3352         opts->x_ix86_preferred_stack_boundary_arg.
3353         Change ix86_force_align_arg_pointer to access
3354         opts->x_ix86_force_align_arg_pointer.
3355         Change ix86_incoming_stack_boundar_arg to access
3356         opts->x_ix86_incoming_stack_boundar_arg.
3357         Change ix86_fpmath to access opts->x_ix86_fpmath.
3358         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
3359         Change ix86_recip_name to access opts->x_ix86_recip_name.
3360         Change ix86_stack_protector_guard to access
3361         opts->x_ix86_stack_protector_guard.
3362         Change ix86_tune_memcpy_strategy to access
3363         opts->x_ix86_tune_memcpy_strategy.
3364         Change ix86_tune_memset_strategy to access
3365         opts->x_ix86_tune_memset_strategy.
3366         Change global_options to access opts.
3367         Change global_options_set to access opts_set.
3368         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
3369         Change TARGET_MMX to TARGET_MMX_P (opts->...).
3370         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
3371         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
3372         Change TARGET_SSE to TARGET_SSE_P (opts->...).
3373         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
3374         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
3375         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
3376         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
3377         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
3378         Change TARGET_AVX to TARGET_AVX_P (opts->...).
3379         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
3380         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
3381         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
3382         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
3383         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
3384         Change TARGET_FMA to TARGET_FMA_P (opts->...).
3385         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
3386         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
3387         Change TARGET_XOP to TARGET_XOP_P (opts->...).
3388         Change TARGET_LWP to TARGET_LWP_P (opts->...).
3389         Change TARGET_ABM to TARGET_ABM_P (opts->...).
3390         Change TARGET_BMI to TARGET_BMI_P (opts->...).
3391         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
3392         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
3393         Change TARGET_TBM to TARGET_TBM_P (opts->...).
3394         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
3395         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
3396         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
3397         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
3398         Change TARGET_AES to TARGET_AES_P (opts->...).
3399         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
3400         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
3401         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
3402         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
3403         Change TARGET_F16C to TARGET_F16C_P (opts->...).
3404         Change TARGET_RTM to TARGET_RTM_P (opts->...).
3405         Change TARGET_HLE to TARGET_HLE_P (opts->...).
3406         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
3407         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
3408         Change TARGET_ADX to TARGET_ADX_P (opts->...).
3409         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
3410         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
3411         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
3412         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
3413         Change TARGET_X32 to TARGET_X32_P (opts->...).
3414         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
3415         Change TARGET_FLOAT_RETURNS_IN_80387 to
3416         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
3417         (ix86_function_specific_save): New parameter. Use opts-> fields
3418         to replace global fields.
3419         (ix86_function_specific_restore): Ditto.
3420         (ix86_valid_target_attribute_inner_p): New parameters.
3421         Fix recursive call.
3422         Fix call to ix86_handle_option and set_option.
3423         (ix86_valid_target_attribute_tree): New parameters.
3424         Change global_options to access opts.
3425         Change global_options_set to access opts_set.
3426         Fix call to ix86_valid_target_attribute_inner_p.
3427         Change ix86_tune_string to access opts->x_ix86_tune_string.
3428         Change ix86_arch_string to access opts->x_ix86_arch_string.
3429         Change ix86_fpmath to access opts->x_ix86_fpmath
3430         Fix call to ix86_option_override_internal.
3431         Fix call to ix86_add_new_builtins.
3432         Fix calls to build_optimization_node and build_target_option_node.
3433         (ix86_valid_target_attribute_p): Remove access to global_options.
3434         Use new gcc_options structure func_options.
3435         Fix call to ix86_valid_target_attribute_tree.
3436         Fix call to  build_optimization_node.
3437         (get_builtin_code_for_version): Fix call to
3438         ix86_valid_target_attribute_tree.
3440 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
3442         * Makefile.in (PICFLAG): New.
3443         (enable_host_shared): New.
3444         (INTERNAL_CFLAGS): Use PICFLAG.
3445         (LIBIBERTY): Use pic build of libiberty.a if configured with
3446         --enable-host-shared.
3447         * configure.ac: Add --enable-host-shared, setting up new
3448         PICFLAG variable.
3449         * configure: Regenerate.
3450         * doc/install.texi (--enable-shared): Add note contrasting it with ...
3451         (--enable-host-shared): New option.
3453 2013-10-15  Richard Biener  <rguenther@suse.de>
3455         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
3456         for built-in functions.
3458 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3460         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
3461         (optimize_range_tests_1): New function,
3462         extracted from optimize_range_tests.
3463         (optimize_range_tests_xor): Similarly.
3464         (optimize_range_tests_diff): New function.
3465         (optimize_range_tests): Use optimize_range_tests_1.
3467 2013-10-15  Cong Hou  <congh@google.com>
3469         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
3470         requirement of the reduction pattern so that one operand of the
3471         reduction operation can come from outside of the loop.
3473 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3475         * config/arm/neon-schedgen.ml: Remove.
3476         * config/arm/cortex-a9-neon.md: Remove comment regarding
3477         neon-schedgen.ml.
3479 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3481         * config/arm/types: Remove old neon types.
3483 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3485         * config/arm/cortex-a7.md
3486         (cortex_a7_neon_type): New.
3487         (cortex_a7_neon_mul): Update for new types.
3488         (cortex_a7_neon_mla): Likewise.
3489         (cortex_a7_neon): Likewise.
3491 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3493         * config/arm/cortex-a15-neon.md
3494         (cortex_a15_neon_type): New,
3496         (cortex_a15_neon_int_1): Remove.
3497         (cortex_a15_neon_int_2): Likewise.
3498         (cortex_a15_neon_int_3): Likewise.
3499         (cortex_a15_neon_int_4): Likewise.
3500         (cortex_a15_neon_int_5): Likewise.
3501         (cortex_a15_neon_vqneg_vqabs): Likewise.
3502         (cortex_a15_neon_vmov): Likewise.
3503         (cortex_a15_neon_vaba): Likewise.
3504         (cortex_a15_neon_vaba_qqq): Likewise.
3505         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3506         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3507         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3508         Likewise.
3509         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3510         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3511         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
3512         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3513         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3514         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3515         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3516         (cortex_a15_neon_shift_1): Likewise.
3517         (cortex_a15_neon_shift_2): Likewise.
3518         (cortex_a15_neon_shift_3): Likewise.
3519         (cortex_a15_neon_vshl_ddd): Likewise.
3520         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3521         (cortex_a15_neon_vsra_vrsra): Likewise.
3522         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3523         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3524         (cortex_a15_neon_bp_3cycle): Likewise.
3525         (cortex_a15_neon_ldm_2): Likewise.
3526         (cortex_a15_neon_stm_2): Likewise.
3527         (cortex_a15_neon_mcr): Likewise.
3528         (cortex_a15_neon_mrc): Likewise.
3529         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3530         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3531         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3532         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3533         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3534         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3535         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3536         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3537         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3538         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3539         (cortex_a15_neon_bp_simple): Likewise.
3540         (cortex_a15_neon_bp_2cycle): Likewise.
3541         (cortex_a15_neon_bp_3cycle): Likewise.
3542         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3543         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3544         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3545         (cortex_a15_neon_vld2_4_regs): Likewise.
3546         (cortex_a15_neon_vld3_vld4): Likewise.
3547         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3548         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3549         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
3550         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
3551         (cortex_a15_neon_vst3_vst4): Rename to...
3552         (cortex_a15_neon_vst4): This, update for new attributes.
3553         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
3554         (cortex_a15_neon_vld3_vld4_lane): Likewise.
3555         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3556         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3557         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3558         (cortex_a15_neon_ldm_2): Likewise.
3559         (cortex_a15_neon_stm_2): Likewise.
3560         (cortex_a15_neon_mcr): Likewise.
3561         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3562         (cortex_a15_neon_mrc): Likewise.
3563         (cortex_a15_neon_mrrc): Likewise.
3565         (cortex_a15_neon_abd): New.
3566         (cortex_a15_neon_abd_q): Likewise.
3567         (cortex_a15_neon_aba): Likewise.
3568         (cortex_a15_neon_aba_q): Likewise.
3569         (cortex_a15_neon_acc): Likewise.
3570         (cortex_a15_neon_acc_q): Likewise.
3571         (cortex_a15_neon_arith_basic): Likewise.
3572         (cortex_a15_neon_arith_complex): Likewise.
3573         (cortex_a15_neon_multiply): Likewise.
3574         (cortex_a15_neon_multiply_q): Likewise.
3575         (cortex_a15_neon_mla): Likewise.
3576         (cortex_a15_neon_mla_q): Likewise.
3577         (cortex_a15_neon_sat_mla_long): Likewise.
3578         (cortex_a15_neon_shift_acc): Likewise.
3579         (cortex_a15_neon_shift_imm_basic): Likewise.
3580         (cortex_a15_neon_shift_imm_complex): Likewise.
3581         (cortex_a15_neon_shift_reg_basic): Likewise.
3582         (cortex_a15_neon_shift_reg_basic_q): Likewise.
3583         (cortex_a15_neon_shift_reg_complex): Likewise.
3584         (cortex_a15_neon_shift_reg_complex_q): Likewise.
3585         (cortex_a15_neon_fp_negabs): Likewise
3586         (cortex_a15_neon_fp_arith): Likewise
3587         (cortex_a15_neon_fp_arith_q): Likewise
3588         (cortex_a15_neon_fp_cvt_int): Likewise
3589         (cortex_a15_neon_fp_cvt_int_q): Likewise
3590         (cortex_a15_neon_fp_cvt_16): Likewise
3591         (cortex_a15_neon_fp_mul): Likewise
3592         (cortex_a15_neon_fp_mul_q): Likewise
3593         (cortex_a15_neon_fp_mla): Likewise
3594         (cortex_a15_neon_fp_mla_q): Likewise
3595         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
3596         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
3597         (cortex_a15_neon_bitops): Likewise.
3598         (cortex_a15_neon_bitops_q): Likewise.
3599         (cortex_a15_neon_from_gp): Likewise.
3600         (cortex_a15_neon_from_gp_q): Likewise.
3601         (cortex_a15_neon_tbl3_tbl4): Likewise.
3602         (cortex_a15_neon_zip_q): Likewise.
3603         (cortex_a15_neon_to_gp): Likewise.
3604         (cortex_a15_neon_load_a): Likewise.
3605         (cortex_a15_neon_load_b): Likewise.
3606         (cortex_a15_neon_load_c): Likewise.
3607         (cortex_a15_neon_load_d): Likewise.
3608         (cortex_a15_neon_load_e): Likewise.
3609         (cortex_a15_neon_load_f): Likewise.
3610         (cortex_a15_neon_store_a): Likewise.
3611         (cortex_a15_neon_store_b): Likewise.
3612         (cortex_a15_neon_store_c): Likewise.
3613         (cortex_a15_neon_store_d): Likewise.
3614         (cortex_a15_neon_store_e): Likewise.
3615         (cortex_a15_neon_store_f): Likewise.
3616         (cortex_a15_neon_store_g): Likewise.
3617         (cortex_a15_neon_store_h): Likewise.
3618         (cortex_a15_vfp_to_from_gp): Likewise.
3620 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3622         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
3624         (cortex_a9_neon_vshl_ddd): Remove.
3625         (cortex_a9_neon_vst3_vst4): Likewise.
3626         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3628         (cortex_a9_neon_bit_ops_q): New.
3630         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
3631         (cortex_a9_neon_int_2): Likewise.
3632         (cortex_a9_neon_int_3): Likewise.
3633         (cortex_a9_neon_int_4): Likewise.
3634         (cortex_a9_neon_int_5): Likewise.
3635         (cortex_a9_neon_vqneg_vqabs): Likewise.
3636         (cortex_a9_neon_vmov): Likewise.
3637         (cortex_a9_neon_vaba): Likewise.
3638         (cortex_a9_neon_vaba_qqq): Likewise.
3639         (cortex_a9_neon_shift_1): Likewise.
3640         (cortex_a9_neon_shift_2): Likewise.
3641         (cortex_a9_neon_shift_3): Likewise.
3642         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3643         (cortex_a9_neon_vsra_vrsra): Likewise.
3644         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3645         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3646         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3647         Likewise.
3648         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3649         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3650         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3651         Likewise.
3652         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3653         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3654         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3655         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3656         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3657         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3658         (cortex_a9_neon_fp_vsum): Likewise.
3659         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3660         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3661         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3662         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3663         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3664         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3665         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3666         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3667         (cortex_a9_neon_bp_simple): Likewise.
3668         (cortex_a9_neon_bp_2cycle): Likewise.
3669         (cortex_a9_neon_bp_3cycle): Likewise.
3670         (cortex_a9_neon_ldr): Likewise.
3671         (cortex_a9_neon_str): Likewise.
3672         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3673         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3674         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3675         (cortex_a9_neon_vld2_4_regs): Likewise.
3676         (cortex_a9_neon_vld3_vld4): Likewise.
3677         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3678         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3679         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3680         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3681         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3682         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3683         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3684         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3685         (cortex_a9_neon_mcr): Likewise.
3686         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3687         (cortex_a9_neon_mrc): Likewise.
3688         (cortex_a9_neon_mrrc): Likewise.
3690 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3692         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
3694         (cortex_a8_neon_vshl_ddd): Remove.
3695         (cortex_a8_neon_vst3_vst4): Likewise.
3696         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3698         (cortex_a8_neon_bit_ops_q): New.
3700         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
3701         (cortex_a8_neon_int_2): Likewise..
3702         (cortex_a8_neon_int_3): Likewise.
3703         (cortex_a8_neon_int_5): Likewise.
3704         (cortex_a8_neon_vqneg_vqabs): Likewise.
3705         (cortex_a8_neon_int_4): Likewise.
3706         (cortex_a8_neon_vaba): Likewise.
3707         (cortex_a8_neon_vaba_qqq): Likewise.
3708         (cortex_a8_neon_shift_1): Likewise.
3709         (cortex_a8_neon_shift_2): Likewise.
3710         (cortex_a8_neon_shift_3): Likewise.
3711         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3712         (cortex_a8_neon_vsra_vrsra): Likewise.
3713         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3714         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3715         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3716         Likewise.
3717         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3718         (cortex_a8_neon_mla_qqq_8_16): Likewise.
3719         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3720         Likewise.
3721         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3722         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3723         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3724         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3725         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3726         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3727         (cortex_a8_neon_fp_vsum): Likewise.
3728         (cortex_a8_neon_fp_vmul_ddd): Likewise.
3729         (cortex_a8_neon_fp_vmul_qqd): Likewise.
3730         (cortex_a8_neon_fp_vmla_ddd): Likewise.
3731         (cortex_a8_neon_fp_vmla_qqq): Likewise.
3732         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3733         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3734         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3735         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3736         (cortex_a8_neon_bp_simple): Likewise.
3737         (cortex_a8_neon_bp_2cycle): Likewise.
3738         (cortex_a8_neon_bp_3cycle): Likewise.
3739         (cortex_a8_neon_ldr): Likewise.
3740         (cortex_a8_neon_str): Likewise.
3741         (cortex_a8_neon_vld1_1_2_regs): Likewise.
3742         (cortex_a8_neon_vld1_3_4_regs): Likewise.
3743         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3744         (cortex_a8_neon_vld2_4_regs): Likewise.
3745         (cortex_a8_neon_vld3_vld4): Likewise.
3746         (cortex_a8_neon_vld1_vld2_lane): Likewise.
3747         (cortex_a8_neon_vld3_vld4_lane): Likewise.
3748         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3749         (cortex_a8_neon_vst1_3_4_regs): Likewise.
3750         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3751         (cortex_a8_neon_vst1_vst2_lane): Likewise.
3752         (cortex_a8_neon_vst3_vst4_lane): Likewise.
3753         (cortex_a8_neon_mcr): Likewise.
3754         (cortex_a8_neon_mcr_2_mcrr): Likewise.
3755         (cortex_a8_neon_mrc): Likewise.
3756         (cortex_a8_neon_mrrc): Likewise.
3758 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3760         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
3761         (fp): New.
3762         * config/aarch64/aarch64-simd.md (neon_type): Remove.
3763         (aarch64_simd_dup<mode>): Add "type" attribute.
3764         (aarch64_dup_lane<mode>): Likewise.
3765         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
3766         (*aarch64_simd_mov<mode>): Likewise.
3767         (aarch64_simd_mov_from_<mode>low): Likewise.
3768         (aarch64_simd_mov_from_<mode>high): Likewise.
3769         (orn<mode>3): Likewise.
3770         (bic<mode>3): Likewise.
3771         (add<mode>3): Likewise.
3772         (sub<mode>3): Likewise.
3773         (mul<mode>3): Likewise.
3774         (*aarch64_mul3_elt<mode>): Likewise.
3775         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3776         (*aarch64_mul3_elt_to_128df): Likewise.
3777         (*aarch64_mul3_elt_to_64v2df): Likewise.
3778         (neg<mode>2): Likewise.
3779         (abs<mode>2): Likewise.
3780         (abd<mode>_3): Likewise.
3781         (aba<mode>_3): Likewise.
3782         (fabd<mode>_3): Likewise.
3783         (*fabd_scalar<mode>3): Likewise.
3784         (and<mode>3): Likewise.
3785         (ior<mode>3): Likewise.
3786         (xor<mode>3): Likewise.
3787         (one_cmpl<mode>2): Likewise.
3788         (aarch64_simd_vec_set<mode>): Likewise.
3789         (aarch64_simd_lshr<mode>): Likewise.
3790         (aarch64_simd_ashr<mode>): Likewise.
3791         (aarch64_simd_imm_shl<mode>): Likewise.
3792         (aarch64_simd_reg_sshl<mode): Likewise.
3793         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
3794         (aarch64_simd_reg_shl<mode>_signed): Likewise.
3795         (aarch64_simd_vec_setv2di): Likewise.
3796         (aarch64_simd_vec_set<mode>): Likewise.
3797         (aarch64_mla<mode>): Likewise.
3798         (*aarch64_mla_elt<mode>): Likewise.
3799         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3800         (aarch64_mls<mode>): Likewise.
3801         (*aarch64_mls_elt<mode>): Likewise.
3802         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3803         (<su><maxmin><mode>3): Likewise.
3804         (move_lo_quad_<mode>): Likewise.
3805         (aarch64_simd_move_hi_quad_<mode>): Likewise.
3806         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
3807         (vec_pack_trunc_<mode>): Likewise.
3808         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
3809         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
3810         (*aarch64_<su>mlal_lo<mode>): Likewise.
3811         (*aarch64_<su>mlal_hi<mode>): Likewise.
3812         (*aarch64_<su>mlsl_lo<mode>): Likewise.
3813         (*aarch64_<su>mlsl_hi<mode>): Likewise.
3814         (*aarch64_<su>mlal<mode>): Likewise.
3815         (*aarch64_<su>mlsl<mode>): Likewise.
3816         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
3817         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
3818         (add<mode>3): Likewise.
3819         (sub<mode>3): Likewise.
3820         (mul<mode>3): Likewise.
3821         (div<mode>3): Likewise.
3822         (neg<mode>2): Likewise.
3823         (abs<mode>2): Likewise.
3824         (fma<mode>4): Likewise.
3825         (*aarch64_fma4_elt<mode>): Likewise.
3826         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3827         (*aarch64_fma4_elt_to_128df): Likewise.
3828         (*aarch64_fma4_elt_to_64v2df): Likewise.
3829         (fnma<mode>4): Likewise.
3830         (*aarch64_fnma4_elt<mode>): Likewise.
3831         (*aarch64_fnma4_elt_<vswap_width_name><mode>
3832         (*aarch64_fnma4_elt_to_128df): Likewise.
3833         (*aarch64_fnma4_elt_to_64v2df): Likewise.
3834         (<frint_pattern><mode>2): Likewise.
3835         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
3836         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
3837         (vec_unpacks_lo_v4sf): Likewise.
3838         (aarch64_float_extend_lo_v2df): Likewise.
3839         (vec_unpacks_hi_v4sf): Likewise.
3840         (aarch64_float_truncate_lo_v2sf): Likewise.
3841         (aarch64_float_truncate_hi_v4sf): Likewise.
3842         (aarch64_vmls<mode>): Likewise.
3843         (<su><maxmin><mode>3): Likewise.
3844         (<maxmin_uns><mode>3): Likewise.
3845         (reduc_<sur>plus_<mode>): Likewise.
3846         (reduc_<sur>plus_v2di): Likewise.
3847         (reduc_<sur>plus_v2si): Likewise.
3848         (reduc_<sur>plus_<mode>): Likewise.
3849         (aarch64_addpv4sf): Likewise.
3850         (clz<mode>2): Likewise.
3851         (reduc_<maxmin_uns>_<mode>): Likewise.
3852         (reduc_<maxmin_uns>_v2di): Likewise.
3853         (reduc_<maxmin_uns>_v2si): Likewise.
3854         (reduc_<maxmin_uns>_<mode>): Likewise.
3855         (reduc_<maxmin_uns>_v4sf): Likewise.
3856         (aarch64_simd_bsl<mode>_internal): Likewise.
3857         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3858         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3859         (aarch64_get_lane<mode>): Likewise.
3860         (*aarch64_combinez<mode>): Likewise.
3861         (aarch64_combine<mode>): Likewise.
3862         (aarch64_simd_combine<mode>): Likewise.
3863         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
3864         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
3865         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
3866         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
3867         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
3868         (aarch64_<sur>h<addsub><mode>): Likewise.
3869         (aarch64_<sur><addsub>hn<mode>): Likewise.
3870         (aarch64_<sur><addsub>hn2<mode>): Likewise.
3871         (aarch64_pmul<mode>): Likewise.
3872         (aarch64_<su_optab><optab><mode>): Likewise.
3873         (aarch64_<sur>qadd<mode>): Likewise.
3874         (aarch64_sqmovun<mode>): Likewise.
3875         (aarch64_<sur>qmovn<mode>): Likewise.
3876         (aarch64_s<optab><mode>): Likewise.
3877         (aarch64_sq<r>dmulh<mode>): Likewise.
3878         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3879         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3880         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3881         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
3882         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3883         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3884         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
3885         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
3886         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3887         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
3888         (aarch64_sqdmull<mode>): Likewise.
3889         (aarch64_sqdmull_lane<mode>_internal): Likewise.
3890         (aarch64_sqdmull_n<mode>): Likewise.
3891         (aarch64_sqdmull2<mode>_internal): Likewise.
3892         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3893         (aarch64_sqdmull2_n<mode>_internal): Likewise.
3894         (aarch64_<sur>shl<mode>): Likewise.
3895         (aarch64_<sur>q<r>shl<mode>
3896         (aarch64_<sur>shll_n<mode>): Likewise.
3897         (aarch64_<sur>shll2_n<mode>): Likewise.
3898         (aarch64_<sur>shr_n<mode>): Likewise.
3899         (aarch64_<sur>sra_n<mode>): Likewise.
3900         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
3901         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
3902         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
3903         (aarch64_cm<optab><mode>): Likewise.
3904         (aarch64_cm<optab>di): Likewise.
3905         (aarch64_cm<optab><mode>): Likewise.
3906         (aarch64_cm<optab>di): Likewise.
3907         (aarch64_cmtst<mode>): Likewise.
3908         (aarch64_cmtstdi): Likewise.
3909         (aarch64_cm<optab><mode>): Likewise.
3910         (*aarch64_fac<optab><mode>): Likewise.
3911         (aarch64_addp<mode>): Likewise.
3912         (aarch64_addpdi): Likewise.
3913         (sqrt<mode>2): Likewise.
3914         (vec_load_lanesoi<mode>): Likewise.
3915         (vec_store_lanesoi<mode>): Likewise.
3916         (vec_load_lanesci<mode>): Likewise.
3917         (vec_store_lanesci<mode>): Likewise.
3918         (vec_load_lanesxi<mode>): Likewise.
3919         (vec_store_lanesxi<mode>): Likewise.
3920         (*aarch64_mov<mode>): Likewise.
3921         (aarch64_ld2<mode>_dreg): Likewise.
3922         (aarch64_ld2<mode>_dreg): Likewise.
3923         (aarch64_ld3<mode>_dreg): Likewise.
3924         (aarch64_ld3<mode>_dreg): Likewise.
3925         (aarch64_ld4<mode>_dreg): Likewise.
3926         (aarch64_ld4<mode>_dreg): Likewise.
3927         (aarch64_tbl1<mode>): Likewise.
3928         (aarch64_tbl2v16qi): Likewise.
3929         (aarch64_combinev16qi): Likewise.
3930         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
3931         (aarch64_st2<mode>_dreg): Likewise.
3932         (aarch64_st2<mode>_dreg): Likewise.
3933         (aarch64_st3<mode>_dreg): Likewise.
3934         (aarch64_st3<mode>_dreg): Likewise.
3935         (aarch64_st4<mode>_dreg): Likewise.
3936         (aarch64_st4<mode>_dreg): Likewise.
3937         (*aarch64_simd_ld1r<mode>): Likewise.
3938         (aarch64_frecpe<mode>): Likewise.
3939         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
3940         (aarch64_frecps<mode>): Likewise.
3942 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3944         * config/arm/iterators.md (V_elem_ch): New.
3945         (q): Likewise.
3946         (VQH_type): Likewise.
3947         * config/arm/arm.md (is_neon_type): New.
3948         (conds): Use is_neon_type.
3949         (anddi3_insn): Update type attribute.
3950         (xordi3_insn): Likewise.
3951         (one_cmpldi2): Likewise.
3952         * gcc/config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
3953         * gcc/config/arm/neon.md (neon_mov): Update type attribute.
3954         (*movmisalign<mode>_neon_store): Likewise.
3955         (*movmisalign<mode>_neon_load): Likewise.
3956         (vec_set<mode>_internal): Likewise.
3957         (vec_set<mode>_internal): Likewise.
3958         (vec_setv2di_internal): Likewise.
3959         (vec_extract<mode>): Likewise.
3960         (vec_extract<mode>): Likewise.
3961         (vec_extractv2di): Likewise.
3962         (*add<mode>3_neon): Likewise.
3963         (adddi3_neon): Likewise.
3964         (*sub<mode>3_neon): Likewise.
3965         (subdi3_neon): Likewise.
3966         (fma<VCVTF:mode>4): Likewise.
3967         (fma<VCVTF:mode>4_intrinsic): Likewise.
3968         (*fmsub<VCVTF:mode>4): Likewise.
3969         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
3970         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
3971         (ior<mode>3): Likewise.
3972         (and<mode>3): Likewise.
3973         (orn<mode>3_neon): Likewise.
3974         (orndi3_neon): Likewise.
3975         (bic<mode>3_neon): Likewise.
3976         (bicdi3_neon): Likewise.
3977         (xor<mode>3): Likewise.
3978         (one_cmpl<mode>2): Likewise.
3979         (abs<mode>2): Likewise.
3980         (neg<mode>2): Likewise.
3981         (negdi2_neon): Likewise.
3982         (*umin<mode>3_neon): Likewise.
3983         (*umax<mode>3_neon): Likewise.
3984         (*smin<mode>3_neon): Likewise.
3985         (*smax<mode>3_neon): Likewise.
3986         (vashl<mode>3): Likewise.
3987         (vashr<mode>3_imm): Likewise.
3988         (vlshr<mode>3_imm): Likewise.
3989         (ashl<mode>3_signed): Likewise.
3990         (ashl<mode>3_unsigned): Likewise.
3991         (neon_load_count): Likewise.
3992         (ashldi3_neon_noclobber): Likewise.
3993         (ashldi3_neon): Likewise.
3994         (signed_shift_di3_neon): Likewise.
3995         (unsigned_shift_di3_neon): Likewise.
3996         (ashrdi3_neon_imm_noclobber): Likewise.
3997         (lshrdi3_neon_imm_noclobber): Likewise.
3998         (<shift>di3_neon): Likewise.
3999         (widen_ssum<mode>3): Likewise.
4000         (widen_usum<mode>3): Likewise.
4001         (quad_halves_<code>v4si): Likewise.
4002         (quad_halves_<code>v4sf): Likewise.
4003         (quad_halves_<code>v8hi): Likewise.
4004         (quad_halves_<code>v16qi): Likewise.
4005         (reduc_splus_v2di): Likewise.
4006         (neon_vpadd_internal<mode>): Likewise.
4007         (neon_vpsmin<mode>): Likewise.
4008         (neon_vpsmax<mode>): Likewise.
4009         (neon_vpumin<mode>): Likewise.
4010         (neon_vpumax<mode>): Likewise.
4011         (*ss_add<mode>_neon): Likewise.
4012         (*us_add<mode>_neon): Likewise.
4013         (*ss_sub<mode>_neon): Likewise.
4014         (*us_sub<mode>_neon): Likewise.
4015         (neon_vadd<mode>_unspec): Likewise.
4016         (neon_vaddl<mode>): Likewise.
4017         (neon_vaddw<mode>): Likewise.
4018         (neon_vhadd<mode>): Likewise.
4019         (neon_vqadd<mode>): Likewise.
4020         (neon_vaddhn<mode>): Likewise.
4021         (neon_vmul<mode>): Likewise.
4022         (neon_vfms<VCVTF:mode>): Likewise.
4023         (neon_vmlal<mode>): Likewise.
4024         (neon_vmls<mode>): Likewise.
4025         (neon_vmlsl<mode>): Likewise.
4026         (neon_vqdmulh<mode>): Likewise.
4027         (neon_vqdmlal<mode>): Likewise.
4028         (neon_vqdmlsl<mode>): Likewise.
4029         (neon_vmull<mode>): Likewise.
4030         (neon_vqdmull<mode>): Likewise.
4031         (neon_vsub<mode>_unspec): Likewise.
4032         (neon_vsubl<mode>): Likewise.
4033         (neon_vsubw<mode>): Likewise.
4034         (neon_vqsub<mode>): Likewise.
4035         (neon_vhsub<mode>): Likewise.
4036         (neon_vsubhn<mode>): Likewise.
4037         (neon_vceq<mode>): Likewise.
4038         (neon_vcge<mode>): Likewise.
4039         (neon_vcgeu<mode>): Likewise.
4040         (neon_vcgt<mode>): Likewise.
4041         (neon_vcgtu<mode>): Likewise.
4042         (neon_vcle<mode>): Likewise.
4043         (neon_vclt<mode>): Likewise.
4044         (neon_vcage<mode>): Likewise.
4045         (neon_vcagt<mode>): Likewise.
4046         (neon_vtst<mode>): Likewise.
4047         (neon_vabd<mode>): Likewise.
4048         (neon_vabdl<mode>): Likewise.
4049         (neon_vaba<mode>): Likewise.
4050         (neon_vabal<mode>): Likewise.
4051         (neon_vmax<mode>): Likewise.
4052         (neon_vmin<mode>): Likewise.
4053         (neon_vpaddl<mode>): Likewise.
4054         (neon_vpadal<mode>): Likewise.
4055         (neon_vpmax<mode>): Likewise.
4056         (neon_vpmin<mode>): Likewise.
4057         (neon_vrecps<mode>): Likewise.
4058         (neon_vrsqrts<mode>): Likewise.
4059         (neon_vqabs<mode>): Likewise.
4060         (neon_vqneg<mode>): Likewise.
4061         (neon_vcls<mode>): Likewise.
4062         (clz<mode>2): Likewise.
4063         (popcount<mode>2): Likewise.
4064         (neon_vrecpe<mode>): Likewise.
4065         (neon_vrsqrte<mode>): Likewise.
4066         (neon_vget_lane<mode>_sext_internal): Likewise.
4067         (neon_vget_lane<mode>_zext_internal): Likewise.
4068         (neon_vdup_n<mode>): Likewise.
4069         (neon_vdup_n<mode>): Likewise.
4070         (neon_vdup_nv2di): Likewise.
4071         (neon_vdup_lane<mode>_interal): Likewise.
4072         (*neon_vswp<mode>): Likewise.
4073         (neon_vcombine<mode>): Likewise.
4074         (float<mode><V_cvtto>2): Likewise.
4075         (floatuns<mode><V_cvtto>2): Likewise.
4076         (fix_trunc<mode><V_cvtto>2): Likewise.
4077         (fixuns_trunc<mode><V_cvtto>2
4078         (neon_vcvt<mode>): Likewise.
4079         (neon_vcvt<mode>): Likewise.
4080         (neon_vcvtv4sfv4hf): Likewise.
4081         (neon_vcvtv4hfv4sf): Likewise.
4082         (neon_vcvt_n<mode>): Likewise.
4083         (neon_vcvt_n<mode>): Likewise.
4084         (neon_vmovn<mode>): Likewise.
4085         (neon_vqmovn<mode>): Likewise.
4086         (neon_vqmovun<mode>): Likewise.
4087         (neon_vmovl<mode>): Likewise.
4088         (neon_vmul_lane<mode>): Likewise.
4089         (neon_vmul_lane<mode>): Likewise.
4090         (neon_vmull_lane<mode>): Likewise.
4091         (neon_vqdmull_lane<mode>): Likewise.
4092         (neon_vqdmulh_lane<mode>): Likewise.
4093         (neon_vqdmulh_lane<mode>): Likewise.
4094         (neon_vmla_lane<mode>): Likewise.
4095         (neon_vmla_lane<mode>): Likewise.
4096         (neon_vmlal_lane<mode>): Likewise.
4097         (neon_vqdmlal_lane<mode>): Likewise.
4098         (neon_vmls_lane<mode>): Likewise.
4099         (neon_vmls_lane<mode>): Likewise.
4100         (neon_vmlsl_lane<mode>): Likewise.
4101         (neon_vqdmlsl_lane<mode>): Likewise.
4102         (neon_vext<mode>): Likewise.
4103         (neon_vrev64<mode>): Likewise.
4104         (neon_vrev32<mode>): Likewise.
4105         (neon_vrev16<mode>): Likewise.
4106         (neon_vbsl<mode>_internal): Likewise.
4107         (neon_vshl<mode>): Likewise.
4108         (neon_vqshl<mode>): Likewise.
4109         (neon_vshr_n<mode>): Likewise.
4110         (neon_vshrn_n<mode>): Likewise.
4111         (neon_vqshrn_n<mode>): Likewise.
4112         (neon_vqshrun_n<mode>): Likewise.
4113         (neon_vshl_n<mode>): Likewise.
4114         (neon_vqshl_n<mode>): Likewise.
4115         (neon_vqshlu_n<mode>): Likewise.
4116         (neon_vshll_n<mode>): Likewise.
4117         (neon_vsra_n<mode>): Likewise.
4118         (neon_vsri_n<mode>): Likewise.
4119         (neon_vsli_n<mode>): Likewise.
4120         (neon_vtbl1v8qi): Likewise.
4121         (neon_vtbl2v8qi): Likewise.
4122         (neon_vtbl3v8qi): Likewise.
4123         (neon_vtbl4v8qi): Likewise.
4124         (neon_vtbl1v16qi): Likewise.
4125         (neon_vtbl2v16qi): Likewise.
4126         (neon_vcombinev16qi): Likewise.
4127         (neon_vtbx1v8qi): Likewise.
4128         (neon_vtbx2v8qi): Likewise.
4129         (neon_vtbx3v8qi): Likewise.
4130         (neon_vtbx4v8qi): Likewise.
4131         (*neon_vtrn<mode>_insn): Likewise.
4132         (*neon_vzip<mode>_insn): Likewise.
4133         (*neon_vuzp<mode>_insn): Likewise.
4134         (neon_vld1<mode>): Likewise.
4135         (neon_vld1_lane<mode>): Likewise.
4136         (neon_vld1_lane<mode>): Likewise.
4137         (neon_vld1_dup<mode>): Likewise.
4138         (neon_vld1_dup<mode>): Likewise.
4139         (neon_vld1_dupv2di): Likewise.
4140         (neon_vst1<mode>): Likewise.
4141         (neon_vst1_lane<mode>): Likewise.
4142         (neon_vst1_lane<mode>): Likewise.
4143         (neon_vld2<mode>): Likewise.
4144         (neon_vld2<mode>): Likewise.
4145         (neon_vld2_lane<mode>): Likewise.
4146         (neon_vld2_lane<mode>): Likewise.
4147         (neon_vld2_dup<mode>): Likewise.
4148         (neon_vst2<mode>): Likewise.
4149         (neon_vst2<mode>): Likewise.
4150         (neon_vst2_lane<mode>): Likewise.
4151         (neon_vst2_lane<mode>): Likewise.
4152         (neon_vld3<mode>): Likewise.
4153         (neon_vld3qa<mode>): Likewise.
4154         (neon_vld3qb<mode>): Likewise.
4155         (neon_vld3_lane<mode>): Likewise.
4156         (neon_vld3_lane<mode>): Likewise.
4157         (neon_vld3_dup<mode>): Likewise.
4158         (neon_vst3<mode>): Likewise.
4159         (neon_vst3qa<mode>): Likewise.
4160         (neon_vst3qb<mode>): Likewise.
4161         (neon_vst3_lane<mode>): Likewise.
4162         (neon_vst3_lane<mode>): Likewise.
4163         (neon_vld4<mode>): Likewise.
4164         (neon_vld4qa<mode>): Likewise.
4165         (neon_vld4qb<mode>): Likewise.
4166         (neon_vld4_lane<mode>): Likewise.
4167         (neon_vld4_lane<mode>): Likewise.
4168         (neon_vld4_dup<mode>): Likewise.
4169         (neon_vst4<mode>): Likewise.
4170         (neon_vst4qa<mode>): Likewise.
4171         (neon_vst4qb<mode>): Likewise.
4172         (neon_vst4_lane<mode>): Likewise.
4173         (neon_vst4_lane<mode>): Likewise.
4174         (neon_vec_unpack<US>_lo_<mode>): Likewise.
4175         (neon_vec_unpack<US>_hi_<mode>): Likewise.
4176         (neon_vec_<US>mult_lo_<mode>): Likewise.
4177         (neon_vec_<US>mult_hi_<mode>): Likewise.
4178         (neon_vec_<US>shiftl_<mode>): Likewise.
4179         (neon_unpack<US>_<mode>): Likewise.
4180         (neon_vec_<US>mult_<mode>): Likewise.
4181         (vec_pack_trunc_<mode>): Likewise.
4182         (neon_vec_pack_trunc_<mode>): Likewise.
4183         (neon_vabd<mode>_2): Likewise.
4184         (neon_vabd<mode>_3): Likewise.
4186 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4188         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
4189         (load_pair): Update type attribute.
4190         (store_pair): Update type attribute.
4191         * config/aarch64/iterators.md (q): New.
4193 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4195         * config/arm/types.md: Add new types for Neon insns.
4197 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4198             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4199             Sergey Lega  <sergey.s.lega@intel.com>
4200             Anna Tikhonova  <anna.tikhonova@intel.com>
4201             Ilya Tocar  <ilya.tocar@intel.com>
4202             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4203             Ilya Verbin  <ilya.verbin@intel.com>
4204             Kirill Yukhin  <kirill.yukhin@intel.com>
4205             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4207         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
4208         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
4209         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
4210         (rcp14<mode>): New.
4211         (srcp14<mode>): Ditto.
4212         (rsqrt14<mode>): Ditto.
4213         (rsqrt14<mode>): Ditto.
4214         (avx512f_vmscalef<mode>): Ditto.
4215         (avx512f_scalef<mode>): Ditto.
4216         (avx512f_getexp<mode>): Ditto.
4217         (avx512f_sgetexp<mode>): Ditto.
4218         (avx512f_fixupimm<mode>): Ditto.
4219         (avx512f_sfixupimm<mode>): Ditto.
4220         (avx512f_rndscale<mode>): Ditto.
4221         (*avx512er_exp2<mode>): Ditto.
4222         (*avx512er_rcp28<mode>): Ditto.
4223         (avx512er_rsqrt28<mode>): Ditto.
4224         (avx512f_getmant<mode>): Ditto.
4225         (avx512f_getmant<mode>): Ditto.
4226         (avx512f_rndscale<mode>): Fix formatting.
4228 2013-10-15  Martin Jambor  <mjambor@suse.cz>
4230         * ipa-utils.h (ipa_edge_within_scc): Declare.
4231         * ipa-cp.c (edge_within_scc): Moved...
4232         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
4234 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4235             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4236             Sergey Lega  <sergey.s.lega@intel.com>
4237             Anna Tikhonova  <anna.tikhonova@intel.com>
4238             Ilya Tocar  <ilya.tocar@intel.com>
4239             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4240             Ilya Verbin  <ilya.verbin@intel.com>
4241             Kirill Yukhin  <kirill.yukhin@intel.com>
4242             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4244         * config/i386/predicates.md (const_8_to_15_operand): New.
4245         (const_16_to_31_operand): Ditto.
4246         * config/i386/sse.md (V8FI): New.
4247         (V16FI): Ditto.
4248         (reduc_splus_v8df): Ditto.
4249         (reduc_splus_v16sf): Ditto.
4250         (avx512f_vextract<shuffletype>32x4_1): Ditto.
4251         (vec_extract_hi_<mode>): Ditto.
4252         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
4253         (vec_set_lo_<mode>): Ditto.
4254         (vec_set_hi_<mode>): Ditto.
4255         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
4256         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
4257         (avx512f_pshufd_1): Ditto.
4258         (avx512f_broadcast<mode>): Ditto.
4259         (avx512f_broadcast<mode>): Ditto.
4260         (define_split): Split vec_extract_lo into move.
4261         (ssequartermode): Ditto.
4262         (ssedoublemode): Extened with wider modes.
4263         (vec_extract_lo_<mode>): Ditto.
4265 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4266             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4267             Sergey Lega  <sergey.s.lega@intel.com>
4268             Anna Tikhonova  <anna.tikhonova@intel.com>
4269             Ilya Tocar  <ilya.tocar@intel.com>
4270             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4271             Ilya Verbin  <ilya.verbin@intel.com>
4272             Kirill Yukhin  <kirill.yukhin@intel.com>
4273             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4275         * config/i386/predicates.md (register_or_constm1_operand): New.
4276         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
4277         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
4278         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
4279         UNSPEC_SCATTER_PREFETCH
4280         (VI48_512): New.
4281         (avx512f_ucmp<mode>3): Ditto.
4282         (avx512f_vternlog<mode>): Ditto.
4283         (avx512f_align<mode>): Ditto.
4284         (<shift_insn><mode>3): Ditto.
4285         (avx512f_<rotate>v<mode>): Ditto.
4286         (avx512f_<rotate><mode>): Ditto.
4287         (avx512f_eq<mode>3): Ditto.
4288         (avx512f_eq<mode>3_1): Ditto.
4289         (avx512f_gt<mode>3): Ditto.
4290         (avx512f_testm<mode>3): Ditto.
4291         (avx512f_testnm<mode>3): Ditto.
4292         (avx512pf_gatherpf<mode>): Ditto.
4293         (*avx512pf_gatherpf<mode>_mask): Ditto.
4294         (*avx512pf_gatherpf<mode>): Ditto.
4295         (avx512pf_scatterpf<mode>): Ditto.
4296         (*avx512pf_scatterpf<mode>_mask): Ditto.
4297         (*avx512pf_scatterpf<mode>): Ditto.
4298         (avx512f_vec_dup_gpr<mode>): Ditto.
4299         (clz<mode>2): Ditto.
4300         (conflict<mode>): Ditto.
4301         (REDUC_SMINMAX_MODE): Extened with wider modes.
4302         (reduc_<code>_<mode>): Ditto.
4303         (vlshr<mode>3): Ditto.
4304         (vashl<mode>3): Ditto.
4306 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4307             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4308             Sergey Lega  <sergey.s.lega@intel.com>
4309             Anna Tikhonova  <anna.tikhonova@intel.com>
4310             Ilya Tocar  <ilya.tocar@intel.com>
4311             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4312             Ilya Verbin  <ilya.verbin@intel.com>
4313             Kirill Yukhin  <kirill.yukhin@intel.com>
4314             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4316         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
4317         UNSPEC_SCATTER.
4318         (VI48F_512): New.
4319         (avx512fmaskmode): Ditto.
4320         (bcstscalarsuff): Ditto.
4321         (avx512f_blendm<mode>): Ditto.
4322         (cmp_imm_predicate): Ditto.
4323         (avx512f_cmp<mode>3): Ditto.
4324         (avx512f_vec_dup<mode>): Ditto.
4325         (avx512f_vec_dup_mem<mode>): Ditto.
4326         (avx512f_vpermi2var<mode>3): Ditto.
4327         (avx512f_vpermt2var<mode>3): Ditto.
4328         (vec_init<mode>): Ditto.
4329         (avx512f_gathersi<mode>): Ditto.
4330         (*avx512f_gathersi<mode>): Ditto.
4331         (*avx512f_gathersi<mode>_2): Ditto.
4332         (avx512f_gatherdi<mode>): Ditto.
4333         (*avx512f_gatherdi<mode>): Ditto.
4334         (*avx512f_gatherdi<mode>_2): Ditto.
4335         (avx512f_scattersi<mode>): Ditto.
4336         (*avx512f_scattersi<mode>): Ditto.
4337         (avx512f_scatterdi<mode>): Ditto.
4338         (*avx512f_scatterdi<mode>): Ditto.
4339         (sseintprefix): Extened with wider modes.
4340         (VEC_GATHER_IDXSI): Ditto.
4341         (VEC_GATHER_IDXDI): Ditto.
4342         (VEC_GATHER_SRCDI): Ditto.
4344 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
4345             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4347         * config/arm/t-aprofile: New file.
4348         * config.gcc: Handle --with-multilib-list option.
4350 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
4352         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
4353         the input for r2.
4355 2013-10-15  Richard Biener  <rguenther@suse.de>
4357         * tree-loop-distribution.c (build_empty_rdg): Inline into
4358         single user.
4359         (rdg_flag_vertex): Inline into single user.
4360         (rdg_flag_vertex_and_dependent): Likewise.
4361         (build_rdg_partition_for_vertex): Remove processed bitmap.
4362         (rdg_build_partitions): Simplify.
4364 2013-10-15  Richard Biener  <rguenther@suse.de>
4366         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
4367         Restructure forwarding through conversions and copies to
4368         avoid performing copy-propagation the wrong way.  Adjust
4369         recursion invocations.
4370         (forward_propagate_addr_expr): Add argument stating if we
4371         are recursing from a single-use.
4372         (ssa_forward_propagate_and_combine): Adjust.
4374 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
4376         * dumpfile.h (gcc::dump_manager): New class, to hold state
4377         relating to dumpfile management.
4378         (get_dump_file_name): Remove in favor of method of dump_manager.
4379         (dump_initialized_p): Likewise.
4380         (dump_start): Likewise.
4381         (dump_finish): Likewise.
4382         (dump_switch_p): Likewise.
4383         (dump_register): Likewise.
4384         (get_dump_file_info): Likewise.
4385         * context.c (gcc::context::context): Construct the dump_manager
4386         instance.
4387         * context.h (gcc::context::get_dumps): New.
4388         (gcc::context::m_dumps): New.
4389         * coverage.c (coverage_init): Port to dump_manager API.
4390         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
4391         (extra_dump_files_in_use): Likewise.
4392         (extra_dump_files_alloced): Likewise.
4393         (gcc::dump_manager::dump_manager): New.
4394         (dump_register): Convert to...
4395         (gcc::dump_manager::dump_register): ...method, replacing
4396         function-static next_dump with m_next_dump field.
4397         (get_dump_file_info): Convert to...
4398         (gcc::dump_manager::get_dump_file_info): ...method.
4399         (get_dump_file_name): Convert to...
4400         (gcc::dump_manager::get_dump_file_name): ...method.
4401         (dump_start): Convert to...
4402         (gcc::dump_manager::dump_start): ...method.
4403         (dump_finish): Convert to...
4404         (gcc::dump_manager::dump_finish): ...method.
4405         (dump_begin): Replace body with...
4406         (gcc::dump_manager::dump_begin): ...new method.
4407         (dump_phase_enabled_p): Convert to...
4408         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4409         (dump_phase_enabled_p): Convert to...
4410         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4411         (dump_initialized_p):  Convert to...
4412         (gcc::dump_manager::dump_initialized_p): ...method.
4413         (dump_flag_name): Replace body with...
4414         (gcc::dump_manager::dump_flag_name): ...new method.
4415         (dump_enable_all): Convert to...
4416         (gcc::dump_manager::dump_enable_all): ...new method.
4417         (opt_info_enable_passes): Convert to...
4418         (gcc::dump_manager::opt_info_enable_passes): ...new method.
4419         (dump_switch_p_1): Convert to...
4420         (gcc::dump_manager::dump_switch_p_1): ...new method.
4421         (dump_switch_p):  Convert to...
4422         (gcc::dump_manager::dump_switch_p): ...new method.
4423         (opt_info_switch_p): Port to dump_manager API.
4424         (enable_rtl_dump_file): Likewise.
4425         * opts-global.c (handle_common_deferred_options): Port to new
4426         dump_manager API.
4427         * passes.c (pass_manager::finish_optimization_passes): Likewise.
4428         (pass_manager::register_one_dump_file): Likewise.
4429         (pass_manager::register_pass): Likewise.
4430         (pass_init_dump_file): Likewise.
4431         (pass_fini_dump_file): Likewise.
4432         * statistics.c (statistics_early_init): Likewise.
4434 2013-10-14  Richard Biener  <rguenther@suse.de>
4436         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
4437         iterative_hash_canonical_type, gimple_canonical_type_hash,
4438         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
4439         gimple_register_canonical_type, print_gimple_types_stats,
4440         free_gimple_type_tables): Move to lto/lto.c
4441         (gt-gimple.h): Do not include.
4442         * gimple.h (gimple_register_canonical_type,
4443         print_gimple_types_stats, free_gimple_type_tables): Remove.
4444         * Makefile.in (GTFILES): Remove gimple.c.
4446 2013-10-14  Travis Snoozy  <quandary@remstate.com>
4448         PR target/58716
4449         * config/msp430/msp430.c (msp430_option_override): Correct thinko
4450         scanning for msp430x targets.
4452 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
4454         PR bootstrap/58509
4455         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
4456         (registers_ok_for_ldd_peep): Move around.
4457         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
4458         * config/sparc/sparc.md (widening peepholes): Use it.
4460 2013-10-14  Richard Biener  <rguenther@suse.de>
4462         PR middle-end/58712
4463         PR middle-end/55358
4464         * gimple.c (iterative_hash_canonical_type): Make sure to
4465         record the hash into the correct hashtable slot.
4467 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
4469         PR rtl-optimization/58662
4470         * combine.c (try_combine): Take into account death nodes on I2 when
4471         splitting a PARALLEL of two independent SETs.  Fix dump message.
4473 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
4475         PR target/51244
4476         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
4477         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
4478         * config/sh/t-sh (sh_treg_combine.o): New entry.
4479         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
4480         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
4481         (register_sh_passes): New function.  Register sh_treg_combine pass.
4482         (sh_option_override): Invoke it.
4483         (sh_canonicalize_comparison): Handle op0_preserve_value.
4484         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
4485         opportunities.  Canonicalize branch condition.
4486         (nott): Allow only if pseudos can be created for non-SH2A.
4488 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
4490         PR target/58690
4491         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
4492         (ix86_expand_movmem): Replace copy_addr_to_reg with
4493         ix86_copy_addr_to_reg.
4494         (ix86_expand_setmem): Likewise.
4496 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
4498         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
4499         provided by ix86_fp_compare_mode instead of CCFPUmode.
4501 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
4503         * config/aarch64/arm_neon.h
4504         (vtbx<1,3>_<psu>8): Fix register constriants.
4506 2013-10-11  Jeff Law  <law@redhat.com>
4508         PR tree-optimization/58640
4509         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
4510         threading paths that cross over two loop entry points.
4512 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4514         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
4515         handle vector float as well.
4516         (*vsx_le_perm_load_v4si): Likewise.
4517         (*vsx_le_perm_store_v2di): Likewise.
4518         (*vsx_le_perm_store_v4si): Likewise.
4520 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4522         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
4523         directly to circumvent subtract from splat{31} workaround.
4524         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
4525         prototype.
4526         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
4527         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
4528         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
4529         (altivec_vperm_<mode>): Convert to define_insn_and_split to
4530         separate big and little endian logic.
4531         (*altivec_vperm_<mode>_internal): New define_insn.
4532         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
4533         separate big and little endian logic.
4534         (*altivec_vperm_<mode>_uns_internal): New define_insn.
4535         (vec_permv16qi): Add little endian logic.
4537 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
4539         * doc/extend.texi (returns_nonnull): Remove arguments.
4541 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4542             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4543             Sergey Lega  <sergey.s.lega@intel.com>
4544             Anna Tikhonova  <anna.tikhonova@intel.com>
4545             Ilya Tocar  <ilya.tocar@intel.com>
4546             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4547             Ilya Verbin  <ilya.verbin@intel.com>
4548             Kirill Yukhin  <kirill.yukhin@intel.com>
4549             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4551         * config/i386/sse.md (VI48F_256_512): New.
4552         (avx2_permvar<mode>): Change to ...
4553         (<avx2_avx512f>_permvar<mode>): This.
4555 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4556             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4557             Sergey Lega  <sergey.s.lega@intel.com>
4558             Anna Tikhonova  <anna.tikhonova@intel.com>
4559             Ilya Tocar  <ilya.tocar@intel.com>
4560             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4561             Ilya Verbin  <ilya.verbin@intel.com>
4562             Kirill Yukhin  <kirill.yukhin@intel.com>
4563             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4565         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
4566         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
4567         * config/i386/sse.md (VI4_AVX): New.
4568         (sf2simodelower): Ditto.
4569         (sse2_cvtps2dq): Change to ...
4570         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
4572 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4573             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4574             Sergey Lega  <sergey.s.lega@intel.com>
4575             Anna Tikhonova  <anna.tikhonova@intel.com>
4576             Ilya Tocar  <ilya.tocar@intel.com>
4577             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4578             Ilya Verbin  <ilya.verbin@intel.com>
4579             Kirill Yukhin  <kirill.yukhin@intel.com>
4580             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4582         * config/i386/sse.md (V_512): New.
4583         (VI_512): Ditto.
4584         (vcond<V_512:mode><VF_512:mode>): Ditto.
4585         (vcond<V_512:mode><VI_512:mode>): Ditto.
4586         (vcondu<V_512:mode><VI_512:mode>): Ditto.
4588 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4589             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4590             Sergey Lega  <sergey.s.lega@intel.com>
4591             Anna Tikhonova  <anna.tikhonova@intel.com>
4592             Ilya Tocar  <ilya.tocar@intel.com>
4593             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4594             Ilya Verbin  <ilya.verbin@intel.com>
4595             Kirill Yukhin  <kirill.yukhin@intel.com>
4596             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4598         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
4599         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
4600         (FMAMODE): Ditto.
4601         (fma<mode>4): Removed condition.
4602         (fms<mode>4): Ditto.
4603         (fnma<mode>4): Ditto.
4604         (fnms<mode>4): Ditto.
4605         (fma4i_fmadd_<mode>): Ditto.
4606         (*fma_fmadd_<mode>): Ditto.
4607         (*fma_fmsub_<mode>): Ditto.
4608         (*fma_fnmadd_<mode>): Ditto.
4609         (*fma_fnmsub_<mode>): Ditto.
4610         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
4611         (*fma_fmaddsub_<mode>): Ditto.
4612         (*fma_fmsubadd_<mode>): Ditto.
4613         (*fmai_fmadd_<mode>): Ditto.
4614         (*fmai_fmsub_<mode>): Ditto.
4615         (*fmai_fnmadd_<mode>): Ditto.
4616         (*fmai_fnmsub_<mode>): Ditto.
4618 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4619             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4620             Sergey Lega  <sergey.s.lega@intel.com>
4621             Anna Tikhonova  <anna.tikhonova@intel.com>
4622             Ilya Tocar  <ilya.tocar@intel.com>
4623             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4624             Ilya Verbin  <ilya.verbin@intel.com>
4625             Kirill Yukhin  <kirill.yukhin@intel.com>
4626             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4628         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
4629         (VI124_256): Changed to ...
4630         (VI124_256_48_512): This.
4631         (ssepackmode): Extended with wider modes.
4632         (<code><mode>3): Changed iterator.
4633         (*avx2_<code><mode>3): Ditto.
4634         (vec_pack_trunc_<mode>): Ditto.
4636 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4637             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4638             Sergey Lega  <sergey.s.lega@intel.com>
4639             Anna Tikhonova  <anna.tikhonova@intel.com>
4640             Ilya Tocar  <ilya.tocar@intel.com>
4641             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4642             Ilya Verbin  <ilya.verbin@intel.com>
4643             Kirill Yukhin  <kirill.yukhin@intel.com>
4644             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4646         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
4647         (VI8F_256_512): Ditto.
4648         (abs<mode>2): Changed iterator.
4649         (avx2_perm<mode>): Changed to ...
4650         (<avx2_avx512f>_perm<mode>): This.
4651         (avx2_perm<mode>_1): Changed to ...
4652         (<avx2_avx512f>_perm<mode>_1): This.
4654 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4655             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4656             Sergey Lega  <sergey.s.lega@intel.com>
4657             Anna Tikhonova  <anna.tikhonova@intel.com>
4658             Ilya Tocar  <ilya.tocar@intel.com>
4659             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4660             Ilya Verbin  <ilya.verbin@intel.com>
4661             Kirill Yukhin  <kirill.yukhin@intel.com>
4662             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4664         * config/i386/sse.md (VI48_AVX512F): New.
4665         (VI48_AVX2): Changed to ...
4666         (VI48_AVX2_48_AVX512F): This.
4667         (avx2_ashrv<mode>): Changed to ...
4668         (<avx2_avx512f>_ashrv<mode>): This.
4669         (avx2_<shift_insn>v<mode>): Changed to ...
4670         (<avx2_avx512f>_<shift_insn>v<mode>): This.
4672 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4673             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4674             Sergey Lega  <sergey.s.lega@intel.com>
4675             Anna Tikhonova  <anna.tikhonova@intel.com>
4676             Ilya Tocar  <ilya.tocar@intel.com>
4677             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4678             Ilya Verbin  <ilya.verbin@intel.com>
4679             Kirill Yukhin  <kirill.yukhin@intel.com>
4680             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4682         * config/i386/sse.md (VI4_AVX512F): New.
4683         (VI8_AVX2_AVX512F): Ditto.
4684         (mul<mode>3): Extended with wider modes.
4685         (*<sse4_1_avx2>_mul<mode>3): Ditto.
4686         (mul<mode>3): Ditto.
4687         (vec_widen_<s>mult_odd_<mode>): Ditto.
4689 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4690             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4691             Sergey Lega  <sergey.s.lega@intel.com>
4692             Anna Tikhonova  <anna.tikhonova@intel.com>
4693             Ilya Tocar  <ilya.tocar@intel.com>
4694             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4695             Ilya Verbin  <ilya.verbin@intel.com>
4696             Kirill Yukhin  <kirill.yukhin@intel.com>
4697             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4699         * config/i386/sse.md (VI2_AVX512F): New.
4700         (VI124_AVX512F): Ditto.
4701         (sseunpackmode): Extended with wider modes.
4702         (sseunpackfltmode): Ditto.
4703         (vec_unpacks_float_hi_<mode>): Ditto.
4704         (vec_unpacks_float_lo_<mode>): Ditto.
4705         (vec_unpacku_float_hi_<mode>): Ditto.
4706         (vec_unpacku_float_lo_<mode>): Ditto.
4707         (vec_unpacks_lo_<mode>): Ditto.
4708         (vec_unpacks_hi_<mode>): Ditto.
4709         (vec_unpacku_lo_<mode>): Ditto.
4710         (vec_unpacku_hi_<mode>): Ditto.
4712 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4713             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4714             Sergey Lega  <sergey.s.lega@intel.com>
4715             Anna Tikhonova  <anna.tikhonova@intel.com>
4716             Ilya Tocar  <ilya.tocar@intel.com>
4717             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4718             Ilya Verbin  <ilya.verbin@intel.com>
4719             Kirill Yukhin  <kirill.yukhin@intel.com>
4720             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4722         * config/i386/i386.md (multdiv): New.
4723         (multdiv_mnemonic): Ditto.
4724         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
4725         (<sse>_vm<multdiv_mnemonic><mode>3): This.
4726         (<sse>_vmdiv<mode>3): Removed.
4728 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4729             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4730             Sergey Lega  <sergey.s.lega@intel.com>
4731             Anna Tikhonova  <anna.tikhonova@intel.com>
4732             Ilya Tocar  <ilya.tocar@intel.com>
4733             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4734             Ilya Verbin  <ilya.verbin@intel.com>
4735             Kirill Yukhin  <kirill.yukhin@intel.com>
4736             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4738         * config/i386/sse.md (V): Extended with wider modes.
4739         (VF2): Ditto.
4740         (ssehalfvecmode): Ditto.
4741         (i128): Ditto.
4742         (ssepackfltmode): Ditto.
4743         (avx_vec_concat<mode>): Ditto.
4744         (V_256_512): New iterator.
4745         (VF2_512_256): Ditto.
4746         (si2dfmode): New attribute.
4747         (si2dfmodelower): Ditto.
4748         (sf2dfmode): Ditto.
4749         (concat_tg_mode): Ditto.
4750         (floatv4siv4df2): Changed to ...
4751         (float<si2dfmodelower><mode>2): This.
4752         (avx_cvtps2pd256): Changed to ...
4753         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
4754         (vec_pack_trunc_v4df): Changed to ...
4755         (vec_pack_trunc_<mode>): This.
4756         (avx_vpermil<mode>): Changed to ...
4757         (<sse2_avx_avx512f>_vpermil<mode>): This.
4758         (<fixsuffix>fix_truncv8dfv8si2): New.
4759         (vec_pack_sfix_trunc_v8df): Ditto.
4760         (avx512f_rndscale<mode>): Ditto.
4761         (avx512f_roundpd512): Ditto.
4762         (vec_pack_ufix_trunc_<mode>): Updated iterator.
4764 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4765             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4766             Sergey Lega  <sergey.s.lega@intel.com>
4767             Anna Tikhonova  <anna.tikhonova@intel.com>
4768             Ilya Tocar  <ilya.tocar@intel.com>
4769             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4770             Ilya Verbin  <ilya.verbin@intel.com>
4771             Kirill Yukhin  <kirill.yukhin@intel.com>
4772             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4774         * config/i386/i386.md (any_fix): New iterator.
4775         (fixsuffix): New attribute.
4776         * config/i386/sse.md (VF1): Extened with wider modes.
4777         (VI): Ditto.
4778         (VI_AVX2): Ditto.
4779         (VI8): Ditto.
4780         (sseintvecmodelower): Ditto.
4781         (ssescalarmode): Ditto.
4782         (ssescalarnum): Ditto.
4783         (VF1_128_256): New.
4784         (ssexmmmode): Ditto.
4785         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
4786         (<sse>_rcp<mode>2): Change iterator.
4787         (rsqrt<mode>2): Ditto.
4788         (<sse>_rsqrt<mode>2): Ditto.
4789         (avx2_vec_dup<mode>): Ditto.
4790         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
4791         (round<mode>2_sfix): Ditto.
4792         (avx2_pbroadcast<mode>): Ditto.
4793         (*andnot<mode>3): Handle XI mode.
4794         (*<code><mode>3): Ditto.
4795         (AVXTOSSEMODE): Removed.
4796         (avx_vpermil<mode>): Changed to ...
4797         (<sse2_avx_avx512f>_vpermil<mode>): This.
4799 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4800             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4801             Sergey Lega  <sergey.s.lega@intel.com>
4802             Anna Tikhonova  <anna.tikhonova@intel.com>
4803             Ilya Tocar  <ilya.tocar@intel.com>
4804             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4805             Ilya Verbin  <ilya.verbin@intel.com>
4806             Kirill Yukhin  <kirill.yukhin@intel.com>
4807             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4809         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
4810         (<sse>_comi): Ditto.
4811         (<sse>_ucomi): Ditto.
4812         (sse_cvtss2siq_2): Ditto.
4813         (sse2_cvtsd2si): Ditto.
4814         (sse2_cvtsd2siq): Ditto.
4815         (sse2_cvttsd2si): Ditto.
4816         (sse2_cvttsd2siq): Ditto.
4817         (<shift_insn><mode>3): Ditto.
4818         (sse2_cvtsi2sdq): Update constraint and prefix.
4819         (sse_cvtsi2ss): Update prefix.
4820         (sse_cvtsi2ssq): Ditto.
4822 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4824         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
4825         ignore internal calls.
4827 2013-10-11  Richard Biener  <rguenther@suse.de>
4829         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
4830         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
4831         TREE_OVERFLOW is set.
4833 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
4835         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
4837         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
4838         * gimple.h: Likewise.
4840         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
4842         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
4843         macros.
4845         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
4847         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
4848         * configure: Regenerate.
4850 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4852         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
4853         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
4854         formatting fixes, use pp_colon instead of pp_character (..., ':'),
4855         similarly pp_right_paren.
4856         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
4857         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
4858         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
4859         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
4860         clauses.
4861         (omp_declare_simd_clauses_equal,
4862         omp_remove_redundant_declare_simd_attrs): New functions.
4863         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
4864         (walk_tree_1): Handle new OpenMP 4.0 clauses.
4865         * tree.h (OMP_LOOP_CHECK): Define.
4866         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
4867         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
4868         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
4869         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
4870         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
4871         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
4872         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
4873         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
4874         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
4875         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
4876         OMP_CLAUSE_SIMDLEN_EXPR): Define.
4877         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
4878         (omp_remove_redundant_declare_simd_attrs): New prototype.
4879         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
4880         GIMPLE_OMP_TEAMS): New codes.
4881         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
4882         * omp-low.c (struct omp_context): Add cancel_label and cancellable
4883         fields.
4884         (target_nesting_level): New variable.
4885         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
4886         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
4887         for collapse > 1 static schedule unless ordered.
4888         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
4889         (determine_parallel_type): Adjust get_ws_args_for caller.
4890         (install_var_field): Handle mask & 4 for double indirection.
4891         (scan_sharing_clauses): Ignore shared clause on teams construct.
4892         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
4893         (create_omp_child_function): If inside target or declare target
4894         constructs, set "omp declare target" attribute on the child function.
4895         (find_combined_for): New function.
4896         (scan_omp_parallel): Handle combined loops.
4897         (scan_omp_target, scan_omp_teams): New functions.
4898         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
4899         restrictions and set ctx->cancellable for cancellable constructs.
4900         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
4901         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
4902         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
4903         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
4904         (omp_clause_aligned_alignment): New function.
4905         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
4906         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
4907         on teams constructs.  Handle user defined reductions and new
4908         OpenMP 4.0 clauses.
4909         (lower_reduction_clauses): Don't set placeholder to address of ref
4910         if it has already the right type.
4911         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
4912         (expand_parallel_call): Use the new non-_start suffixed builtins,
4913         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
4914         and GOMP_parallel_end after the call.
4915         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
4916         (expand_omp_for_init_counts): Handle combined loops.
4917         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
4918         loops.
4919         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
4920         end of cancellable loops.
4921         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
4922         Likewise.  Handle collapse > 1 loops.
4923         (expand_omp_simd): Handle combined loops.
4924         (expand_omp_for): Add inner_stmt argument, adjust callers of
4925         expand_omp_for* functions, use expand_omp_for_static*chunk even
4926         for collapse > 1 unless ordered.
4927         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
4928         of cancellable sections.
4929         (expand_omp_single): Remove need_barrier variable, just rely on
4930         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
4931         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
4932         (expand_omp_atomic_load, expand_omp_atomic_store,
4933         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
4934         (expand_omp_target): New function.
4935         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
4936         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
4937         (build_omp_regions_1): Immediately close region for
4938         GF_OMP_TARGET_KIND_UPDATE.
4939         (maybe_add_implicit_barrier_cancel): New function.
4940         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
4941         cancellation.
4942         (lower_omp_single): Likewise.  Add clobber after the barrier.
4943         (lower_omp_taskgroup): New function.
4944         (lower_omp_for): Handle combined loops.  Adjust
4945         lower_rec_input_clauses caller.  Handle cancellation.
4946         (lower_depend_clauses): New function.
4947         (lower_omp_taskreg): Lower depend clauses.  Adjust
4948         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
4949         cancellation.
4950         (lower_omp_target, lower_omp_teams): New functions.
4951         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
4952         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
4953         and GOMP_cancellation_point calls.
4954         (lower_omp): Fold stmts inside of target region.
4955         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
4956         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4957         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
4958         (BT_FN_VOID_OMPFN_PTR_UINT,
4959         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
4960         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
4961         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
4962         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
4963         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
4964         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
4965         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
4966         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
4967         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
4968         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
4969         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
4970         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
4971         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
4972         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
4973         BUILT_IN_GOMP_PARALLEL_END.
4974         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
4975         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
4976         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
4977         GF_OMP_FOR_KIND_DISTRIBUTE.
4978         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
4979         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
4980         (dump_gimple_omp_return): Print lhs if it has any.
4981         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
4982         gimple_omp_atomic_seq_cst_p.
4983         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
4984         and GIMPLE_OMP_TEAMS.
4985         * langhooks.c (lhd_omp_mappable_type): New function.
4986         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
4987         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
4988         hook.
4989         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
4990         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
4991         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
4992         (struct gimplify_omp_ctx): Add combined_loop field.
4993         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
4994         on stmts inside of target region.
4995         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
4996         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
4997         ORT_TARGET and ORT_TARGET_DATA.
4998         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
4999         Handle GOVD_MAP.
5000         (omp_notice_threadprivate_variable): Complain about threadprivate
5001         variables in target region.
5002         (omp_notice_variable): Complain about vars with non-mappable type
5003         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
5004         (omp_check_private): Ignore ORT_TARGET* regions.
5005         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
5006         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
5007         (find_combined_omp_for): New function.
5008         (gimplify_omp_for): Handle gimplification of combined loops.
5009         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
5010         OMP_TEAMS.
5011         (gimplify_omp_target_update): New function.
5012         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
5013         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
5014         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
5015         (gimplify_body): If fndecl has "omp declare target" attribute, add
5016         implicit ORT_TARGET context around it.
5017         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
5018         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
5019         * tree-nested.c (convert_nonlocal_reference_stmt,
5020         convert_local_reference_stmt, convert_gimple_call): Handle
5021         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5022         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
5023         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
5024         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
5025         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
5026         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
5027         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
5028         BUILT_IN_GOMP_LOOP_END_CANCEL,
5029         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
5030         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
5031         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
5032         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
5033         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
5034         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
5035         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
5036         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
5037         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
5038         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
5039         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
5040         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
5041         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
5042         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
5043         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
5044         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5045         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
5046         gimple_build_omp_teams): New functions.
5047         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
5048         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
5049         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
5050         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5051         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
5052         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
5053         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
5054         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
5055         GF_OMP_ATOMIC_SEQ_CST): New.
5056         (gimple_build_omp_taskgroup, gimple_build_omp_target,
5057         gimple_build_omp_teams): New prototypes.
5058         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
5059         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5060         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
5061         GIMPLE_OMP_SINGLE as end of range.
5062         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
5063         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
5064         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
5065         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
5066         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
5067         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
5068         gimple_omp_target_kind, gimple_omp_target_set_kind,
5069         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
5070         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
5071         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
5072         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
5073         gimple_omp_teams_set_clauses): New inlines.
5074         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
5075         and GIMPLE_OMP_TASKGROUP.
5076         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
5077         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
5078         enum omp_clause_proc_bind_kind): New.
5079         (union omp_clause_subcode): Add depend_kind, map_kind and
5080         proc_bind_kind fields.
5081         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
5082         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5083         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
5084         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
5085         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
5087 2013-10-10  Teresa Johnson  <tejohnson@google.com>
5089         * predict.c (tree_estimate_probability): Add new parameter
5090         for estimate_bb_frequencies.
5091         (estimate_bb_frequencies): Add new parameter to force estimation.
5092         (rebuild_frequencies): When max frequency in function is small,
5093         recompute counts from frequencies.
5094         * predict.h (estimate_bb_frequencies): New parameter.
5096 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5098         * ipa-inline.c (ipa_inline): Fix leak of "order" when
5099         optimizations are disabled.
5101 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5103         * coverage.c (coverage_finish): Fix leak of da_file_name.
5105 2013-10-10  Jan Hubicka  <jh@suse.cz>
5107         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
5108         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
5109         Bobcat and generic.
5111 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
5113         PR middle-end/58670
5114         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
5115         if any labels are in FALLTHRU_BB, use a special label emitted
5116         immediately after the asm goto insn rather than label_rtx
5117         of the LABEL_DECL.
5118         (expand_asm_stmt): Adjust caller.
5119         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
5120         edge if the last insn in predecessor is a jump with single successor,
5121         but it isn't simplejump_p.
5123 2013-10-10  Richard Biener  <rguenther@suse.de>
5125         PR tree-optimization/58656
5126         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
5128 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5130         * gimplify.c: Include expr.h and tm_p.h for targets with special
5131         va-arg padding requirements.
5133 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5135         * tree-flow.h: Move some prototypes to gimple.h.
5136         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
5137         * gimple.h: Relocate some prototypes from tree-flow.h
5138         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
5139         Move to gimplify.c.
5140         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
5141         (build_va_arg_indirect_ref): Relocate and make static.
5142         (std_gimplify_va_arg_expr): Relocate here.
5143         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
5144         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
5146 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5148         * doc/md.texi: Document the mnemonic attribute.
5150 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5152         PR target/57377
5153         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
5154         (set_attr_alternative x ...) when searching for user defined
5155         mnemonic attribute.
5157 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5159         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
5161 2013-10-09  Easwaran Raman <eraman@google.com>
5163         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
5164         * cfgexpand.c (defer_stack_allocation): ...use here
5165         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
5167 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
5169         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
5170         (operand_equal_for_value_replacement): New function, extracted from
5171         value_replacement and enhanced to catch more cases.
5172         (value_replacement): Use operand_equal_for_value_replacement.
5174 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5176         * loop-doloop.c (doloop_modify, doloop_optimize): Use
5177         get_max_loop_iterations.
5179 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5181         * config/arm/aarch-common.c (arm_early_load_addr_dep):
5182         Place comment above function.
5184 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5186         * tree-flow.h: Remove all remaining prototypes, enums and structs that
5187         are not related to tree-cfg.c.
5188         * tree-ssa-address.h: New file.  Relocate prototypes.
5189         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
5190         (addr_for_mem_ref): New.  Combine call to get_address_description and
5191         return addr_for_mem_ref.
5192         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
5193         * tree-ssa-live.h: Adjust prototypes.
5194         * passes.c: Include tree-ssa-live.h.
5195         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
5196         * graphite.c (graphite_transform_loops): Make static.
5197         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5198         make_pass_graphite, pass_data_graphite_transforms,
5199         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
5200         * ipa-pure-const.c (warn_function_noreturn): Make static.
5201         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5202         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5203         Relocate from tree-cfg.c
5204         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
5205         static.
5206         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5207         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5208         Move to ipa-pure-const.c.
5209         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
5210         Relocate from tree-optimize.c.
5211         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
5212         make_pass_fixup_cfg): Move to tree-cfg.c.
5213         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
5214         Relocate some prototypes.
5215         * tree-data-ref.h (tree_check_data_deps) Add prototype.
5216         * tree-dump.c (dump_function_to_file): Remove prototype.
5217         Add tree-flow.h to the include file.
5218         * tree-dump.h: Remove prototype.
5219         * tree-parloops.h: New File.  Add prototypes.
5220         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
5221         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
5222         from tree-ssa-loop.c.
5223         * tree-predcom.c (run_tree_predictive_commoning,
5224         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
5225         Relocate here from tree-ssa-loop.c.
5226         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
5227         ssa_name_values.release ().
5228         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
5229         (ssa_name_values): Relocate from tree-flow.h.
5230         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
5231         * tree-ssa-loop.c (run_tree_predictive_commoning,
5232         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
5233         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5234         make_pass_graphite, pass_data_graphite_transforms,
5235         make_pass_graphite_transforms, gate_tree_parallelize_loops,
5236         tree_parallelize_loops, pass_data_parallelize_loops,
5237         make_pass_parallelize_loops): Move to other files.
5238         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
5239         moved here.
5240         * tree.h: Remove prototypes from tree-address.c.
5242 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5244         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
5245         (struct int_tree_map): Move to tree-hasher.h
5246         (SCALE, LABEL, PERCENT): Move to gimple.h
5247         * tree-flow-inline.h: Delete.  Move functions to other files.
5248         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
5249         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
5250         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
5251         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
5252         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
5253         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
5254         * tree-hasher.h: Don't include tree-flow.h.
5255         (struct int_tree_map): Relocate from tree-flow.h.
5256         * tree-sra.c (contains_view_convert_expr_p): Relocate from
5257         tree-flow-inline.h and make static.
5258         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
5259         tree-flow-inline.h.
5260         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
5261         tree-flow-inline.h and make static.
5262         * tree.h (is_global_var, may_be_aliased): Relocate from
5263         tree-flow-inline.h.
5264         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
5265         * value-prof.c: No longer include tree-flow-inline.h.
5266         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
5268 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5270         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
5271         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
5272         (enum move_pos): Move to tree-ssa-loop-im.h
5273         * cfgloop.h: Move some prototypes.
5274         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
5275         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
5276         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
5277         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
5278         (loop_containing_stmt): Relocate from tree-flow-inline.h.
5279         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
5280         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
5281         (enum move_pos): Relocate here.
5282         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
5283         tree-ssa-loop.c.
5284         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
5285         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
5286         make_pass_lim): Relocate here from tree-ssa-loop.c.
5287         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
5288         tree-ssa-loop.c.
5289         (loop_edge_to_cancel, unloop_loops): Make static.
5290         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5291         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
5292         (tree_complete_unroll, gate_tree_complete_unroll,
5293         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
5294         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5295         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
5296         * tree-ssa-loop-ivopts.c: Remove local prototypes.
5297         (stmt_invariant_in_loop_p): Remove unused function.
5298         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
5299         * tree-ssa-loop-manip.h: New file.  Add prototypes.
5300         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
5301         (gcov_type_to_double_int): Move to cfgloop.h.
5302         (double_int_cmp, bound_index,
5303         estimate_numbers_of_iterations_loop): Make static.
5304         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
5305         (max_loop_iterations): Factor out get_max_loop_iterations.
5306         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
5307         cfgloop.c.
5308         * tree-ssa-loop-niter.h: New file.  Add prototypes.
5309         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
5310         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
5311         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
5312         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
5313         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
5314         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
5315         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
5316         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
5317         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
5318         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
5319         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5320         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
5321         pass_data_complete_unroll, make_pass_complete_unroll,
5322         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5323         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
5324         tree-ssa-loop-ivcanon.c.
5325         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
5326         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
5327         tree-ssa-loop-prefetch.c.
5328         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
5329         tree-ssa-loop-im.c.
5330         (get_lsm_tmp_name): Relocate and add suffix parameter.
5331         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
5332         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
5333         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
5334         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
5335         (get_estimated_loop_iterations): Factor out accessor from
5336         estimated_loop_iterations in tree-ssa-loop-niter.c.
5337         (get_max_loop_iterations): Factor out accessor from
5338         _max_loop_iterations in tree-ssa-niter.c.
5339         * loop-unroll.c (decide_unroll_constant_iterations,
5340         decide_unroll_runtime_iterations, decide_peel_simple,
5341         decide_unroll_stupid): Use new get_* accessors.
5343 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
5345         PR tree-optimization/20318
5346         * doc/extend.texi (returns_nonnull): New function attribute.
5347         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
5348         attribute.
5349         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
5350         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
5352 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
5354         PR middle-end/58570
5355         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
5356         false if both components are bitfields.
5358 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5360         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
5361         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
5362         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
5363         * config/aarch64/aarch64.h
5364         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
5365         * config/aarch64/aarch64-simd-builtins.def
5366         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
5368 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5370         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
5372 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5374         * config/aarch64/arm_neon.h (vdiv_f64): Added.
5376 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5378         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
5379         (vneg_f64): New intrinsic.
5380         (vneg_s8): Asm replaced with C.
5381         (vneg_s16): Likewise.
5382         (vneg_s32): Likewise.
5383         (vneg_s64): New intrinsic.
5384         (vnegq_f32): Asm replaced with C.
5385         (vnegq_f64): Likewise.
5386         (vnegq_s8): Likewise.
5387         (vnegq_s16): Likewise.
5388         (vnegq_s32): Likewise.
5389         (vnegq_s64): Likewise.
5391 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
5393         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
5395 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5397         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
5398         packed stack special handling.
5399         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
5400         back to fixed stack slots for FPRs saved due to stdarg.
5402 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5404         * config/s390/s390.c (s390_frame_info): Restructure function.
5406 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5408         * config/s390/s390.c (struct s390_frame_layout): New field
5409         gpr_save_slots.
5410         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
5411         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
5412         regs_ever_clobbered to char*.
5413         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
5414         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
5415         (s390_register_info): Enable FPR save slots.  Move/Copy some
5416         functionality into ...
5417         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
5418         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
5419         function.
5420         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
5421         not imply a stack frame.
5422         (s390_init_frame_layout): Remove variable clobbered_regs.
5423         (s390_update_register_info): Remove function.
5424         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
5425         cannot be used for register renaming.
5426         (s390_hard_regno_scratch_ok): New function.
5427         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
5428         (s390_initial_elimination_offset): Change offset calculation of
5429         the return address pointer.
5430         (save_gprs): Deal with only r6 being saved from the call-saved regs.
5431         (restore_gprs): Set frame related flag.
5432         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
5433         (s390_emit_prologue): Call s390_register_info instead of
5434         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
5435         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
5436         (s390_optimize_prologue): Call s390_optimize_register_info.
5437         Try to remove also FPR slot save/restore INSNs.  Remove frame
5438         related flags from restore INSNs.
5440 2013-10-08  DJ Delorie  <dj@redhat.com>
5442         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
5443         (movhi): Likewise.
5445         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
5446         avoid conflict with the MI use of %c.
5447         * config/rl78/rl78-real.md: change %c to %C throughout.
5448         * config/rl78/rl78-virt.md: Likewise.
5450 2013-10-08  Jan Hubicka  <jh@suse.cz>
5452         * config/i386/i386.c (ix86_option_override_internal): Switch
5453         to SSE math for -ffast-math when target ISA supports SSE2.
5455 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
5457         * tree-flow.h: Remove some prototypes.
5458         * tree.h: Remove some protypes, add a couple.
5459         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
5460         using_eh_for_cleanups_p): Add interface routines for front ends.
5461         * tree-eh.h: New file.  Add protoptyes.
5462         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
5463         (add_stmt_to_eh_lp_fn): Make static.
5464         (lower_try_finally): Use new using_eh_for_cleanups_p.
5465         * emit-rtl.c: Include tree-eh.h.
5466         * gimple.h: Include tree-eh.h.
5468 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
5470         PR tree-optimization/58480
5471         * tree-vrp.c (infer_nonnull_range): New function.
5472         (infer_value_range): Call infer_nonnull_range.
5474 2013-10-08  Dehao Chen  <dehao@google.com>
5476         PR tree-optimization/58619
5477         * tree-inline.c (copy_phis_for_bb): Combine location data
5478         only if non-null.
5480 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5482         PR target/58423
5483         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
5484         RTX_FRAME_RELATED_P on INSN.
5486 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5488         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
5489         (altivec_expand_vec_perm_const): Call it.
5491 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5493         * config/rs6000/vector.md (mov<mode>): Emit permuted move
5494         sequences for LE VSX loads and stores at expand time.
5495         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
5496         prototype.
5497         * config/rs6000/rs6000.c (rs6000_const_vec): New.
5498         (rs6000_gen_le_vsx_permute): New.
5499         (rs6000_gen_le_vsx_load): New.
5500         (rs6000_gen_le_vsx_store): New.
5501         (rs6000_gen_le_vsx_move): New.
5502         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
5503         (*vsx_le_perm_load_v4si): New.
5504         (*vsx_le_perm_load_v8hi): New.
5505         (*vsx_le_perm_load_v16qi): New.
5506         (*vsx_le_perm_store_v2di): New.
5507         (*vsx_le_perm_store_v4si): New.
5508         (*vsx_le_perm_store_v8hi): New.
5509         (*vsx_le_perm_store_v16qi): New.
5510         (*vsx_xxpermdi2_le_<mode>): New.
5511         (*vsx_xxpermdi4_le_<mode>): New.
5512         (*vsx_xxpermdi8_le_V8HI): New.
5513         (*vsx_xxpermdi16_le_V16QI): New.
5514         (*vsx_lxvd2x2_le_<mode>): New.
5515         (*vsx_lxvd2x4_le_<mode>): New.
5516         (*vsx_lxvd2x8_le_V8HI): New.
5517         (*vsx_lxvd2x16_le_V16QI): New.
5518         (*vsx_stxvd2x2_le_<mode>): New.
5519         (*vsx_stxvd2x4_le_<mode>): New.
5520         (*vsx_stxvd2x8_le_V8HI): New.
5521         (*vsx_stxvd2x16_le_V16QI): New.
5523 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
5525         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
5527 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5529         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
5530         loop to work also for 31bit ABI.
5531         Save the stack pointer for frame_size > 0.
5533 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5535         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
5536         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
5537         constraint letters from expanders.
5538         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
5539         retry count to general_operand.
5540         ("tabort"): Give operand 0 a mode.
5541         ("tabort_1"): Add mode and constraint letter for operand 0.
5542         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
5544 2013-10-04  Jeff Law  <law@redhat.com>
5546         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
5548         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
5549         out of ...
5550         (thread_across_edge): Here.  Call it.
5552 2013-10-04  Cary Coutant  <ccoutant@google.com>
5554         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
5555         discarding a location list expression (or a piece of one).
5557 2013-10-03  Jan Hubicka  <jh@suse.cz>
5559         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
5560         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
5561         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
5563 2013-10-03  Jan Hubicka  <jh@suse.cz>
5565         * config/i386/i386.c (ix86_option_override_internal): Do not enable
5566         accumulate-outgoing-args when producing unwind info.
5568 2013-10-03  Wei Mi  <wmi@google.com>
5570         * lra-constraints.c (insert_move_for_subreg): New function
5571         extracted from simplify_operand_subreg.
5572         (simplify_operand_subreg): Add reload for paradoxical subreg.
5574 2013-10-03  Rong Xu  <xur@google.com>
5576         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
5577         the candidate of builtin_expect such that we should fix the
5578         size/time estimation.
5579         (estimate_function_body_sizes): Do the acutally size/time fix-up
5580         for builtin_expect.
5582 2013-10-03  Rong Xu  <xur@google.com>
5584         * predict.c (tree_predict_by_opcode): Get the probability
5585         for builtin_expect from param builtin_expect_probability.
5586         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
5587         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
5588         * doc/invoke.texi: Add documentation for builtin-expect-probability.
5590 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
5592         PR c++/19476
5593         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
5594         * calls.c (alloca_call_p): Use get_callee_fndecl.
5595         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
5596         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
5597         Likewise.
5598         (vrp_visit_stmt): Remove duplicated code.
5600 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
5602         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
5603         ceildf2, btruncdf2, instead of vsx_* name.
5605         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
5606         iterators to only do V2DF and V4SF here.  Move the DF code to
5607         rs6000.md where it is combined with SF mode.  Replace <VSv> with
5608         just 'v' since only vector operations are handled with these insns
5609         after moving the DF support to rs6000.md.
5610         (vsx_sub<mode>3): Likewise.
5611         (vsx_mul<mode>3): Likewise.
5612         (vsx_div<mode>3): Likewise.
5613         (vsx_fre<mode>2): Likewise.
5614         (vsx_neg<mode>2): Likewise.
5615         (vsx_abs<mode>2): Likewise.
5616         (vsx_nabs<mode>2): Likewise.
5617         (vsx_smax<mode>3): Likewise.
5618         (vsx_smin<mode>3): Likewise.
5619         (vsx_sqrt<mode>2): Likewise.
5620         (vsx_rsqrte<mode>2): Likewise.
5621         (vsx_fms<mode>4): Likewise.
5622         (vsx_nfma<mode>4): Likewise.
5623         (vsx_copysign<mode>3): Likewise.
5624         (vsx_btrunc<mode>2): Likewise.
5625         (vsx_floor<mode>2): Likewise.
5626         (vsx_ceil<mode>2): Likewise.
5627         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
5628         (vsx_sminsf3): Likewise.
5629         (vsx_fmadf4): Likewise.
5630         (vsx_fmsdf4): Likewise.
5631         (vsx_nfmadf4): Likewise.
5632         (vsx_nfmsdf4): Likewise.
5633         (vsx_cmpdf_internal1): Likewise.
5635         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
5636         simpler to select whether a target has SPE or traditional floating
5637         point support in iterators.
5638         (TARGET_DF_SPE): Likewise.
5639         (TARGET_SF_FPR): Likewise.
5640         (TARGET_DF_FPR): Likewise.
5641         (TARGET_SF_INSN): Macros to say whether floating point support
5642         exists for a given operation for expanders.
5643         (TARGET_DF_INSN): Likewise.
5645         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
5646         combining of SF/DF mode operations, using both traditional and VSX
5647         registers.
5648         (Fvsx): Likewise.
5649         (Ff): Likewise.
5650         (Fv): Likewise.
5651         (Fs): Likewise.
5652         (Ffre): Likewise.
5653         (FFRE): Likewise.
5654         (abs<mode>2): Combine SF/DF modes using traditional floating point
5655         instructions.  Add support for using the upper DF registers with
5656         VSX support, and SF registers with power8-vector support.  Update
5657         expanders for operations supported by both the SPE and traditional
5658         floating point units.
5659         (abs<mode>2_fpr): Likewise.
5660         (nabs<mode>2): Likewise.
5661         (nabs<mode>2_fpr): Likewise.
5662         (neg<mode>2): Likewise.
5663         (neg<mode>2_fpr): Likewise.
5664         (add<mode>3): Likewise.
5665         (add<mode>3_fpr): Likewise.
5666         (sub<mode>3): Likewise.
5667         (sub<mode>3_fpr): Likewise.
5668         (mul<mode>3): Likewise.
5669         (mul<mode>3_fpr): Likewise.
5670         (div<mode>3): Likewise.
5671         (div<mode>3_fpr): Likewise.
5672         (sqrt<mode>3): Likewise.
5673         (sqrt<mode>3_fpr): Likewise.
5674         (fre<Fs>): Likewise.
5675         (rsqrt<mode>2): Likewise.
5676         (cmp<mode>_fpr): Likewise.
5677         (smax<mode>3): Likewise.
5678         (smin<mode>3): Likewise.
5679         (smax<mode>3_vsx): Likewise.
5680         (smin<mode>3_vsx): Likewise.
5681         (negsf2): Delete SF operations that are merged with DF.
5682         (abssf2): Likewise.
5683         (addsf3): Likewise.
5684         (subsf3): Likewise.
5685         (mulsf3): Likewise.
5686         (divsf3): Likewise.
5687         (fres): Likewise.
5688         (fmasf4_fpr): Likewise.
5689         (fmssf4_fpr): Likewise.
5690         (nfmasf4_fpr): Likewise.
5691         (nfmssf4_fpr): Likewise.
5692         (sqrtsf2): Likewise.
5693         (rsqrtsf_internal1): Likewise.
5694         (smaxsf3): Likewise.
5695         (sminsf3): Likewise.
5696         (cmpsf_internal1): Likewise.
5697         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
5698         (negdf2): Delete DF operations that are merged with SF.
5699         (absdf2): Likewise.
5700         (nabsdf2): Likewise.
5701         (adddf3): Likewise.
5702         (subdf3): Likewise.
5703         (muldf3): Likewise.
5704         (divdf3): Likewise.
5705         (fred): Likewise.
5706         (rsqrtdf_internal1): Likewise.
5707         (fmadf4_fpr): Likewise.
5708         (fmsdf4_fpr): Likewise.
5709         (nfmadf4_fpr): Likewise.
5710         (nfmsdf4_fpr): Likewise.
5711         (sqrtdf2): Likewise.
5712         (smaxdf3): Likewise.
5713         (smindf3): Likewise.
5714         (cmpdf_internal1): Likewise.
5715         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
5716         (btrunc<mode>2): Delete separate expander, and combine with the
5717         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
5718         (btrunc<mode>2_fpr): Likewise.
5719         (ceil<mode>2): Likewise.
5720         (ceil<mode>2_fpr): Likewise.
5721         (floor<mode>2): Likewise.
5722         (floor<mode>2_fpr): Likewise.
5723         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
5724         Add support for using the upper registers with VSX and
5725         power8-vector.  Move insns to be closer to the define_expands. On
5726         VSX systems, prefer the traditional form of FMA over the VSX
5727         version, since the traditional form allows the target not to
5728         overlap with the inputs.
5729         (fms<mode>4_fpr): Likewise.
5730         (nfma<mode>4_fpr): Likewise.
5731         (nfms<mode>4_fpr): Likewise.
5733 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5734             Richard Earnshaw  <richard.earnshaw@arm.com>
5736         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
5737         (struct mult_cost_table): Likewise.
5738         (struct mem_cost_table): Likewise.
5739         (struct fp_cost_table): Likewise.
5740         (struct vector_cost_table): Likewise.
5741         (cpu_cost_table): Likewise.
5742         * config/arm/arm.opt (mold-rts-costs): New option.
5743         (mnew-generic-costs): Likewise.
5744         * config/arm/arm.c (generic_extra_costs): New table.
5745         (cortexa15_extra_costs): Likewise.
5746         (arm_slowmul_tune): Use NULL as new costs.
5747         (arm_fastmul_tune): Likewise.
5748         (arm_strongarm_tune): Likewise.
5749         (arm_xscale_tune): Likewise.
5750         (arm_9e_tune): Likewise.
5751         (arm_v6t2_tune): Likewise.
5752         (arm_cortex_a5_tune): Likewise.
5753         (arm_cortex_a9_tune): Likewise.
5754         (arm_v6m_tune): Likewise.
5755         (arm_fa726te_tune): Likewise.
5756         (arm_cortex_a15_tune): Use cortex15_extra_costs.
5757         (arm_cortex_tune): Use generict_extra_costs.
5758         (shifter_op_p): New function.
5759         (arm_unspec_cost): Likewise.
5760         (LIBCALL_COST): Define.
5761         (arm_new_rtx_costs): New function.
5762         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
5763         table is available. Use old costs otherwise unless mnew-generic-costs
5764         is specified.
5765         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
5766         (cpu_cost_table): Declare.
5768 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5770         PR target/58460
5771         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
5772         (*subs_mul_imm_<mode>)
5773         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
5774         (*sub_<shift>_<mode>)
5775         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
5776         Remove k constraint.
5778 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
5780         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
5781         code.
5782         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
5784 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5786         * predict.c (probably_never_executed): New function.
5787         (probably_never_executed_bb_p): Invoke probably_never_executed.
5788         (probably_never_executed_edge_p): Ditto.
5789         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5790         Treat profile insanities conservatively.
5792 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
5794         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
5796 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
5798         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
5799         use it.  Use smaller increase for scratch.  Don't increase reject
5800         for early clobber scratch.
5801         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
5802         setting eliminated regs except setting fp from hfp.
5803         (lra_eliminate): Check lra_insn_recog_data on NULL.
5805 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5807         PR target/58587
5808         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
5809         setting -mvsx-timode by default until the underlying problem is fixed.
5810         (RS6000_CPU, power7 defaults): Likewise.
5812 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5814         * config/x-linux (host-linux.o): Remove header dependencies.
5815         Use $(COMPILE) and $(POSTCOMPILE).
5816         * config/t-linux-android (linux-android.o): Ditto.
5818 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5820         * Makefile.in (expmed.o-warn): Remove.
5822 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5824         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
5825         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
5827 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5829         * dojump.c (do_jump_1): Divide probability between
5830         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
5832 2013-10-02  Tom Tromey  <tromey@redhat.com>
5834         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
5836 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5838         * tree-flow.h: Remove some prototypes.
5839         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
5840         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
5841         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
5842         mark_virtual_phi_result_for_renaming): Relocate here.
5843         * tree-into-ssa.h: Add prototypes.
5844         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
5845         single_pred_before_succ_order.
5846         (blocks_in_phiopt_order): Rename and move to cfganal.c.
5847         (nonfreeing_call_p) Move to gimple.c.
5848         * cfganal.c (single_pred_before_succ_order): Move and renamed from
5849         tree-ssa-phiopt.c.
5850         * basic-block.h (single_pred_before_succ_order): Add prototype.
5851         * gimple.c (nonfreeing_call_p): Relocate here.
5852         * gimple.h: Add prototype.
5853         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
5854         * tree-ssa-dom.h: New file.  Relocate prototypes here.
5855         * tree-ssa.h: Include tree-ssa-dom.h.
5857 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5859         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
5860         Use $(COMPILE) and $(POSTCOMPILE).
5862         * config/alpha/x-alpha (driver-alpha.o): Ditto.
5864 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5866         * tree-flow.h: Remove some prototypes.
5867         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
5868         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5869         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
5870         propagate_tree_value*): Move from here to...
5871         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
5872         propagate_tree_value*): Relocate here.
5873         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5874         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
5875         * gimple-fold.c: Remove gimple-fold.h from include list.
5876         * tree-vrp.c: Remove gimple-fold.h from include list.
5877         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
5878         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
5879         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
5880         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
5881         * sese.c: Add tree-ssa-propagate.h to include list.
5883 2013-10-02  Richard Biener  <rguenther@suse.de>
5885         * tree-loop-distribution.c: Include tree-vectorizer.h for
5886         find_loop_location.
5887         (enum partition_kind): Remove PKIND_REDUCTION.
5888         (struct partition_s): Remove has_writes member, add reduction_p member.
5889         (partition_alloc): Adjust.
5890         (partition_builtin_p): Likewise.
5891         (partition_has_writes): Remove.
5892         (partition_reduction_p): New function.
5893         (partition_merge_into): Likewise.
5894         (generate_code_for_partition): Commonize builtin partition
5895         handling tail.
5896         (rdg_cannot_recompute_vertex_p): Remove.
5897         (already_processed_vertex_p): Likewise.
5898         (rdg_flag_vertex): Do not set has_writes.
5899         (classify_partition): Adjust.
5900         (rdg_build_partitions): Do not set has_writes, treat all
5901         partitions as useful.
5902         (distribute_loop): Record number of library calls generated.  Adjust.
5903         (tree_loop_distribution): Report number of loops and library
5904         calls generated as opt-info.
5906 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5908         * tree-flow.h: Include new .h files.  Move prototypes.
5909         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
5910         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
5911         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
5912         * tree-pretty-print.h: Add prototypes from tree-flow.h.
5913         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
5914         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
5915         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
5916         * tree.h (get_ref_base_and_extent): Move prototype out.
5917         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
5918         tree-dfa.h.
5919         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
5920         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
5921         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
5922         * tree-scalar-evolution.c: Include tree.h.
5923         * sese.c: Include tree.h.
5924         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
5925         * dwarf2out.c: Include tree-dfa.h.
5926         * tree-chrec.c: Include tree.h.
5927         * tree-data-ref.c: Include tree.h.
5929 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
5931         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
5932         Fix whitespace.
5934 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
5936         * config/t-sol2 (sol2-c.o): Remove header dependencies.
5937         Use $(COMPILE) and $(POSTCOMPILE).
5938         (sol2-cxx.o): Likewise.
5939         (sol2-stubs.o): Likewise.
5940         (sol2.o): Likewise.
5941         * config/x-solaris (host-solaris.o): Likewise.
5943         * config/sparc/t-sparc (sparc.o): Remove.
5944         (sparc-c.o): Remove header dependencies.
5945         Use $(COMPILE) and $(POSTCOMPILE).
5946         * config/sparc/x-sparc: Likewise.
5948 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
5950         * config/arc/arc-opts.h: Add 2013 to Copyright years.
5951         * config/arc/arc700.md: Likewise.
5952         * config/arc/arc-modes.def: Likewise.
5953         * config/arc/arc-simd.h: Likewise.
5954         * config/arc/t-arc-uClibc: Likewise.
5955         * config/arc/t-arc-newlib: Likewise.
5957 2013-10-02  Renlin Li  <renlin.li@arm.com>
5959         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5960         plus_constant.
5961         (aarch64_expand_epilogue): Likewise.
5963 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5964             Yufeng Zhang  <yufeng.zhang@arm.com>
5966         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
5967         declaration.
5968         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
5969         'base_in' represent a conversion and legal_cast_p_1 holds; set
5970         'base_in' with the returned value from get_unwidened.
5972 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5974         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
5975         plus_constant is not used.
5977 2013-10-01  Wei Mi  <wmi@google.com>
5979         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
5980         * config/i386/i386.md: Add define_peephole2 to
5981         break partial reg stall for cvtss2sd/cvtsd2ss.
5983 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5985         * config/arc/arc.c (pass_arc_ifcvt::clone):
5986         Update for ctxt_ -> m_ctxt change.
5988 2013-10-01  Jeff Law  <law@redhat.com>
5990         * tree-ssa-threadupdate.c (struct redirection_data): Delete
5991         outgoing_edge and intermediate_edge fields.  Instead store the path.
5992         (redirection_data::hash): Hash on the last edge's destination index.
5993         (redirection_data::equal): Check the entire thread path.
5994         (lookup_redirectio_data): Corresponding changes.
5995         (create_edge_and_update_destination_phis): Likewise.
5996         (thread_single_edge): Likewise.
5998 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
5999             Diego Novillo <dnovillo@google.com>
6001         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
6002         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
6003         (vld32wh_insn, vld32wl_insn): Delete commented-out old
6004         versions of these patterns.
6006         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
6007         (__builtin_arc_aligned): Likewise.
6009         * config/arc/arc.md: Expand adc_0 comment stating the intended
6010         purpose and why it isn't ready.
6011         Replace commented out call_value_via_label_mixed with a
6012         plain comment about bl_s.
6014         * config/arc/arc.c (stdio.h): Don't include directly.
6015         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
6016         Remove if (1) condition.
6017         (arc_encode_section_info): Fix comment.
6019 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6021         * config/arc/arc.c (arc_conditional_register_usage):
6022         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
6023         Also set reg_alloc_order for DMA config regs.
6025 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6026             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6028         * doc/install.texi (--with-cpu): Mention ARC.
6029         (arc-*-elf32): New paragraph.
6030         (arc-linux-uclibc): Likewise.
6031         * doc/md.texi (Machine Constraints): Add ARC part.
6032         * doc/invoke.texi: (menu): Add ARC Options.
6033         (Machine Dependent Options) <ARC Options>: Add synopsis.
6034         (node ARC Options): Add.
6035         * doc/extend.texi (long_call / short_call attribute): Add ARC.
6036         (ARC Built-in Functions): New section defining
6037         generic ARC built-in functions.
6038         (ARC SIMD Built-in Functions): New section defining SIMD specific
6039         built-in functions.
6040         (Declaring Attributes of Functions): Extended
6041         description of short_call and long_call attributes for ARC and
6042         added index entries.
6044 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
6045             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
6046             Joern Rennecke  <joern.rennecke@embecosm.com>
6047             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
6048             Brendan Kehoe  <brendan@zen.org>
6049             Michael Eager  <eager@eagercon.com>
6050             Simon Cook  <simon.cook@embecosm.com>
6051             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6053         * config/arc, common/config/arc: New directories.
6055 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6056             Brendan Kehoe  <brendan@zen.org>
6057             Simon Cook  <simon.cook@embecosm.com>
6059         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
6061 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6063         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
6064         * tree-ssa-coalesce.h: New. Move prototype to here.
6065         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
6066         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
6067         (gimple_can_coalesce_p): Move to...
6068         * gimple.c (gimple_can_coalesce_p): Here.
6070 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6072         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
6073         (dump_decl_set): Move to gimple.c.
6074         * gimple.h: Don't include tree-ssa-operands.h.
6075         (dump_decl_set): Add prototype.
6076         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6077         Move to gimple-ssa.h.
6078         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
6079         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
6080         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
6081         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
6082         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
6083         than PHI_ARG_DEF.
6084         (dump_decl_set): Relocate here.
6085         * gimple-ssa.h: New file.
6086         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6087         Relocate from gimple.h.
6088         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
6089         * tree-ssa-operands.c (swap_ssa_operands): Rename from
6090         swap_tree_operands and remove non-ssa path.
6091         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
6092         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
6093         swap_ssa_operands.
6094         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
6095         vect_is_simple_reduction_1): Use swap_ssa_operands.
6096         * tree-flow.h: Move various prototypes to tree-phinodes.h.
6097         (enum need_phi_state): Move to tree-into-ssa.c.
6098         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6099         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
6100         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
6101         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
6102         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6103         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6104         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6105         num_imm_uses): Move to ssa-iterators.h.
6106         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
6107         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
6108         tree-phinodes.h.
6109         (op_iter_done, op_iter_next_def, op_iter_next_tree,
6110         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
6111         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
6112         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
6113         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
6114         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
6115         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
6116         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
6117         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
6118         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
6119         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
6120         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
6121         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
6122         (set_phi_nodes): Move to tree-phinodes.h.
6123         * tree-ssa-operands.h (enum ssa_op_iter_type,
6124         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
6125         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
6126         (dump_decl_set): Remove prototype.
6127         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
6128         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
6129         (set_phi_nodes): Relocate from tree-flow-inline.h.
6130         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
6131         tree-flow-inline.h
6132         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
6133         include list.  Temporarily add gimple.h to include list.
6134         * ssa-iterators.h: New file.
6135         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6136         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
6137         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
6138         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
6139         Relocate from tree-ssa-operands.h.
6140         (delink_imm_use, link_imm_use_to_list, link_imm_use,
6141         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6142         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6143         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6144         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
6145         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
6146         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
6147         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
6148         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
6149         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
6150         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
6151         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
6152         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
6153         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
6154         Relocate from tree-flow-inline.h.
6155         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
6157 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
6159         * aarch64-simd.md
6160         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
6161         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
6162         Insert '\t' to output template.
6163         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
6164         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
6165         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
6166         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
6168 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
6170         * doc/install.texi (Host/target specific installation notes for GCC):
6171         Put @anchor before @heading.
6172         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
6173         Use @email for email addresses.
6175 2013-10-01  Jeff Law  <law@redhat.com>
6177         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
6178         a vec.  Only delete the path if we create one without successfully
6179         registering a jump thread.
6180         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
6181         as a pointer.
6182         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
6183         (paths): New vector of jump threading paths.
6184         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
6185         (THREAD_PATH): New accessor macro for the entire thread path.
6186         (lookup_redirection_data): Get intermediate and final outgoing edge
6187         from the thread path.
6188         (create_edge_and_update_destination_phis): Copy the threading path.
6189         (ssa_fix_duplicate_block_edges): Get edges and block types from the
6190         jump threading path.
6191         (ssa_redirect_edges): Get edges and block types from the jump threading
6192         path.  Free the path vector.
6193         (thread_block): Get edges from the jump threading path.  Look at the
6194         entire path to see if we thread to a loop exit.  If we cancel a jump
6195         thread request, then free the path vector.
6196         (thread_single_edge): Get edges and block types from the jump threading
6197         path.  Free the path vector.
6198         (thread_through_loop_header): Get edges and block types from the jump
6199         threading path.  Free the path vector.
6200         (mark_threaded_blocks): Iterate over the vector of paths and store
6201         the path on the appropriate edge.  Get edges and block types from the
6202         jump threading path.
6203         (mark_threaded_blocks): Get edges and block types from the jump
6204         threading path.  Free the path vector.
6205         (thread_through_all_blocks): Use the vector of paths rather than
6206         a vector of 3-edge sets.
6207         (register_jump_thread): Accept pointer to a path vector rather
6208         than the path vector itself.  Store the path vector for later use.
6209         Simplify.
6211 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
6212             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6214         PR target/58574
6215         * config/s390/s390.c (s390_split_branches): Modify check for table
6216         jump insns.
6217         (s390_chunkify_start): Rearrange table jump insn check in order to
6218         deal with compare and branch insns correctly.
6220 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
6222         PR target/58578
6223         Revert
6224         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
6225         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
6226         define_insn_and_split.
6227         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
6228         (shiftsi3_compare): New pattern.
6229         (rrx): New pattern.
6230         * config/arm/unspecs.md (UNSPEC_RRX): New.
6232 2013-10-01  Alan Modra  <amodra@gmail.com>
6234         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
6235         casing inout operands.
6237 2013-10-01  Richard Biener  <rguenther@suse.de>
6239         PR tree-optimization/58553
6240         * tree-loop-distribution.c (struct partition_s): Add niter member.
6241         (classify_partition): Populate niter member for the partition
6242         and properly identify whether the relevant store happens before
6243         or after the loop exit.
6244         (generate_memset_builtin): Use niter member from the partition.
6245         (generate_memcpy_builtin): Likewise.
6247 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6249         * vec.h (vec_prefix, vec): Prefix member names with "m_".
6250         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
6252 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6254         * basic-block.h (edge_list): Prefix member names with "m_".
6255         * context.h (context): Likewise.
6256         * domwalk.h (dom_walker): Likewise.
6257         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
6258         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
6259         * hash-table.h (hash_table): Likewise.
6260         * machmode.h (bit_field_mode_iterator): Likewise.
6261         * pass_manager.h (pass_list): Likewise.
6262         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
6263         * tree-pass.h (pass_data): Likewise.
6264         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
6265         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
6266         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
6267         * asan.c (pass_data_asan): Update accordingly.
6268         * cfganal.c (control_dependences::find_control_dependence): Likewise.
6269         (control_dependences::control_dependences): Likewise.
6270         (control_dependences::~control_dependences): Likewise.
6271         (control_dependences::~control_dependences): Likewise.
6272         (control_dependences::get_edges_dependent_on): Likewise.
6273         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
6274         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
6275         * context.c (gcc::context::context): Likewise.
6276         * cprop.c (pass_rtl_cprop::clone): Likewise.
6277         * domwalk.c (dom_walker::walk): Likewise.
6278         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
6279         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
6280         * mode-switching.c (pass_mode_switching::clone): Likewise.
6281         * passes.c (opt_pass::opt_pass): Likewise.
6282         (pass_manager::pass_manager): Likewise.
6283         * predict.c (pass_strip_predict_hints::clone): Likewise.
6284         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
6285         (pass_split_all_insns::clone): Likewise.
6286         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6287         Likewise.
6288         (bit_field_mode_iterator::next_mode): Likewise.
6289         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
6290         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
6291         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
6292         * tree-complex.c (pass_lower_complex::clone): Likewise.
6293         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
6294         * tree-object-size.c (pass_object_sizes::clone): Likewise.
6295         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
6296         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
6297         (pass_fold_builtins::clone): Likewise.
6298         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
6299         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
6300         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
6301         (pass_cd_dce::clone): Likewise.
6302         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
6303         (pass_phi_only_cprop::clone): Likewise.
6304         * tree-ssa-dse.c (pass_dse::clone): Likewise.
6305         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
6306         * tree-ssa-loop.c (pass_lim::clone): Likewise.
6307         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
6308         * tree-ssa-pre.c (pass_fre::clone): Likewise.
6309         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
6310         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
6311         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
6312         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
6313         * tree-vrp.c (pass_vrp::clone): Likewise.
6314         * tsan.c (pass_tsan::clone): Likewise.
6316 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
6318         PR middle-end/58564
6319         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
6320         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
6322         PR middle-end/58564
6323         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
6324         optimization, punt if sign_bit_p looked through any zero extension.
6326 2013-09-30  Teresa Johnson  <tejohnson@google.com>
6328         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
6329         Update redirected out edge count in joiner case.
6330         (ssa_redirect_edges): Common the joiner and non-joiner cases
6331         so that joiner case gets profile updates.
6333 2013-09-30  Richard Biener  <rguenther@suse.de>
6335         PR tree-optimization/58554
6336         * tree-loop-distribution.c (classify_partition): Require
6337         unconditionally executed stores for memcpy and memset recognition.
6338         (tree_loop_distribution): Calculate dominance info.
6340 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
6342         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
6343         (NO_PROFILE_COUNTERS): Likewise.
6344         (PROFILE_HOOK): Likewise.
6345         (FUNCTION_PROFILER): Likewise.
6346         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
6348 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
6350         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
6351         calls and defines in TARGET_MACHO conditional.
6352         (load_macho_picbase_di): Likewise.
6353         (reload_macho_picbase): Likewise.
6354         (reload_macho_picbase_si): Likewise.
6355         (reload_macho_picbase_di): Likewise.
6356         (nonlocal_goto_receiver): Likewise.
6358 2013-09-30  Nick Clifton  <nickc@redhat.com>
6360         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
6361         that use the MSP430X ISA.
6362         (msp430_option_override): Scan -mmcu command line option for any
6363         MCU name that supports the MSP430X ISA.
6364         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
6365         -mmcu options which enable the MSP430X ISA.
6367 2013-09-30  Richard Biener  <rguenther@suse.de>
6369         PR middle-end/58532
6370         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
6371         before looking for setjmp-like calls.
6373 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6375         PR target/10901
6376         * config/darwin-protos.h (machopic_get_function_picbase): New.
6377         * config/darwin.c (machopic_get_function_picbase): New.
6378         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
6379         label for a new func.  (load_macho_picbase_di): Likewise.
6380         (reload_macho_picbase): New expand.
6381         (reload_macho_picbase_si): New insn.
6382         (reload_macho_picbase_di): New insn.
6383         (nonlocal_goto_receiver): New define and split.
6384         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
6385         (unspecv enum): Add UNSPECV_NLGR.
6387 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6389         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
6390         that altivec registers are correctly sized on Darwin.
6392 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6394         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
6395         darwin-driver.o): Use COMPILE and POSTCOMPILE.
6396         * config/x-darwin (host-darwin.o): Likewise.
6397         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
6398         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
6399         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
6401 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
6403         * doc/invoke.texi: Fix usage of @tie{} command.
6405 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
6407         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
6409 2013-09-28  Jan Hubicka  <jh@suse.cz>
6411         * config/i386/x86-tune.def: Add documentation for each of the options;
6412         add whitespace.
6414 2013-09-28  Jan Hubicka  <jh@suse.cz>
6416         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
6417         generic.
6418         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
6419         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
6420         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
6421         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
6422         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
6424 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
6426         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
6427         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
6428         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
6429         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
6430         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
6431         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
6432         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
6433         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
6434         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
6435         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
6436         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
6437         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
6438         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
6439         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
6440         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
6441         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
6442         gimple.h, godump.c, graphite-clast-to-gimple.c,
6443         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
6444         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
6445         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
6446         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
6447         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
6448         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
6449         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
6450         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
6451         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
6452         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
6453         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
6454         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
6455         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
6456         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
6457         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
6458         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
6459         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
6460         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
6461         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
6462         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
6463         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
6464         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
6465         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
6466         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
6467         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
6468         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
6469         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
6470         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
6471         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
6472         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
6473         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
6474         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
6475         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
6476         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
6477         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6478         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
6479         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
6480         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
6481         whitespace before "(".
6483 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
6485         * expr.h (extract_bit_field): Remove packedp parameter.
6486         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
6487         from forward declaration.
6488         (store_split_bit_field): Remove packedp arg from calls to
6489         extract_fixed_bit_field.
6490         (extract_bit_field_1): Remove packedp parameter and packedp
6491         argument from recursive calls and calls to extract_fixed_bit_field.
6492         (extract_bit_field): Remove packedp parameter and corresponding
6493         arg to extract_bit_field_1.
6494         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
6495         to issue warnings.
6496         (extract_split_bit_field): Remove packedp arg from call to
6497         extract_fixed_bit_field.
6498         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
6499         (copy_blkmode_from_reg): Likewise.
6500         (copy_blkmode_to_reg): Likewise.
6501         (read_complex_part): Likewise.
6502         (store_field): Likewise.
6503         (expand_expr_real_1): Likewise.
6504         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
6505         to extract_bit_field.
6506         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
6507         call to extract_bit_field.
6508         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
6509         call to extract_bit_field.
6510         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
6511         and special packedp behavior from -fstrict-volatile-bitfields
6512         documentation.
6514 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6516         * lra-eliminations.c (init_elim_table): Guard value_p.
6518 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
6520         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
6521         DFmode, DImode, and SFmode in the upper VSX registers based on the
6522         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
6523         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
6524         of the VSX register handling.
6526         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
6527         (f32_sv): Likewise.
6528         (zero_extendsidi2_lfiwzx): Add support for loading into the
6529         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
6530         only do VSX memory options on Altivec registers.
6531         (extendsidi2_lfiwax): Likewise.
6532         (extendsfdf2_fpr): Likewise.
6533         (mov<mode>_hardfloat, SF/SD modes): Likewise.
6534         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
6535         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
6536         (movdi_internal64): Likewise.
6538 2013-09-27  Xinliang David Li  <davidxl@google.com>
6540         * opts.c (finish_options): Adjust parameters
6541         according to vect cost model.
6542         (common_handle_option): Set dynamic vect cost
6543         model for FDO.
6544         targhooks.c (default_add_stmt_cost): Compute stmt cost
6545         unconditionally.
6546         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
6547         Use helper function.
6548         * tree-vectorizer.h (unlimited_cost_model): New function.
6549         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
6550         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
6551         function.
6552         (vect_enhance_data_refs_alignment): Ditto.
6553         * flag-types.h: New enum.
6554         * common/config/i386/i386-common.c (ix86_option_init_struct):
6555         No need to initialize vect_cost_model flag.
6556         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
6557         unconditionally.
6559 2013-09-27  Diego Novillo  <dnovillo@google.com>
6561         * gimple.h (enum ssa_mode): Remove.
6563 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
6565         * cfgloop.h (number_of_loops): Fix typo in check for null.
6567 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
6569         PR middle-end/58551
6570         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
6571         are children of outermost saved_cfun's loop, and set it up to
6572         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
6573         if loop != loop0 and SESE region contains bbs that belong to loop0.
6575 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6577         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
6578         (binary_scale_code_p, get_base_term, get_index_term): New functions.
6579         (set_address_segment, set_address_base, set_address_index)
6580         (set_address_disp): Accept the argument unconditionally.
6581         (baseness): Remove must_be_base_p and must_be_index_p checks.
6582         (decompose_normal_address): Classify as much as possible in the
6583         main loop.
6585 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6587         * cse.c (count_reg_usage): Handle INT_LIST.
6588         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
6589         * reginfo.c (reg_scan_mark_refs): Likewise.
6590         * reload1.c (eliminate_regs_1): Likewise.
6592 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
6594         PR middle-end/58547
6595         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
6596         signed.
6598 2013-09-27  Richard Biener  <rguenther@suse.de>
6600         PR tree-optimization/58459
6601         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
6602         restriction not propagating into loops.
6604 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
6606         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
6607         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
6608         * doc/tree-ssa.texi (Walking use-def chains): Delete.
6610 2013-09-26  Richard Biener  <rguenther@suse.de>
6612         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
6614 2013-09-26  Richard Biener  <rguenther@suse.de>
6616         * alias.h (component_uses_parent_alias_set): Rename to ...
6617         (component_uses_parent_alias_set_from): ... this.
6618         * alias.c (component_uses_parent_alias_set): Rename to ...
6619         (component_uses_parent_alias_set_from): ... this and return
6620         the desired parent.
6621         (reference_alias_ptr_type_1): Use the result from
6622         component_uses_parent_alias_set_from instead of stripping
6623         components one at a time.
6624         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
6626 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
6628         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
6629         Move prototypes to...
6630         * tree-ssa-ter.h: New File.  Move prototypes here.
6631         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
6632         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
6633         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
6634         from is_replaceable_p.
6635         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
6636         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
6637         (process_replaceable): Use ter_is_replaceable_p.
6638         (find_replaceable_in_bb): Use ter_is_replaceable_p.
6639         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
6640         newly refactored ssa_is_replaceable_p.
6641         * cfgexpand.c: Include tree-outof-ssa.h.
6642         * ssaexpand.h: Delete.
6644 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
6646         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
6647         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
6648         tree-ssa.c
6649         (create_gimple_tmp): Delete.
6650         (get_expr_type, build_assign, build_type_cast): Move to...
6651         * gimple-builder.c: New File.
6652         (get_expr_type): Relocate from gimple.c.
6653         (build_assign, build_type_cast): Change to only create ssanames.
6654         * gimple.h: Move prototypes to...
6655         * gimple-builder.h: New File. Here.
6656         * tree-ssa.h: And here.
6657         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
6658         count_uses_and_derefs): Relocate from gimple.c.
6659         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
6660         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
6661         * tree-ssa-math-opts (execute_cse_reciprocals): Use
6662         gimple_replace_ssa_lhs.
6663         * asan.c: Include gimple-builder.h.
6664         * Makefile.in: Add gimple-builder.o.
6666 2013-09-26  Richard Biener  <rguenther@suse.de>
6668         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
6669         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
6670         (loe_visit_block): Use gcc_checking_assert.
6671         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
6672         gimple_assign_ssa_name_copy_p.
6673         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
6674         change.
6676 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
6678         * config/rs6000/t-rs6000 (rs6000.o): Remove.
6679         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
6681 2013-09-26  Richard Biener  <rguenther@suse.de>
6683         PR tree-optimization/58539
6684         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
6685         the fact that debug statements are not taking part in loop-closed
6686         SSA construction.
6688 2013-09-26  Nick Clifton  <nickc@redhat.com>
6690         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
6691         time warning message.
6692         (msp430_print_operand_raw): Delete unused letter parameter.
6693         (TARGET_PRINT_OPERAND_ADDRESS): Define.
6694         (msp430_print_operand_address): New function.
6695         (msp430_print_operand): Move address printing code from here to
6696         new function.
6697         * config/msp430/msp430.md (movsipsi2): Add comment in generated
6698         assembler.
6699         (zero_extendpsisi2): Likewise.
6700         (extendpsisi2): New pattern.
6701         (andneghi3): New pattern.
6703 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
6705         * config/aarch64/aarch64.opt (mlra): New option.
6706         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
6707         (TARGET_LRA_P): Define.
6709 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
6711         * expr.c (expand_assignment): Remove obsolete comment.
6713 2013-09-25  Jeff Law  <law@redhat.com>
6715         * tree-flow.h (thread_through_all_blocks): Prototype moved into
6716         tree-ssa-threadupdate.h.
6717         (register_jump_thread): Similarly.
6718         * tree-ssa-threadupdate.h: New header file.
6719         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
6720         * tree-vrp.c: Likewise.
6721         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
6722         (thread_around_empty_blocks): Change type of path vector argument to
6723         an edge,type pair from just an edge.  Initialize both elements when
6724         appending to a jump threading path.  Tweak references to elements
6725         appropriately.
6726         (thread_across_edge): Similarly.  Release memory for the elements
6727         as needed.
6728         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
6729         (dump_jump_thread_path): New function broken out from
6730         register_jump_thread.
6731         (register_jump_thread): Use dump_jump_thread_path.  Change type of
6732         path vector entries.  Search the path for NULL edges and dump
6733         the path if one is found.  Tweak the conversion of path to 3-edge
6734         form to use the block copy type information embedded in the path.
6736 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6738         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
6740 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6741             Vladimir Makarov  <vmakarov@redhat.com>
6743         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
6744         from the least significant bit.
6745         (strip_address_mutations): Add bitfield operations handling.
6746         (must_be_index_p): Add shifting and rotate operations handling.
6747         (set_address_base): Use must_be_base_p predicate.
6748         (set_address_index): Use must_be_index_p predicate.
6750 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6751             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
6752             Sergey Lega  <sergey.s.lega@intel.com>
6753             Anna Tikhonova  <anna.tikhonova@intel.com>
6754             Ilya Tocar  <ilya.tocar@intel.com>
6755             Andrey Turetskiy  <andrey.turetskiy@intel.com>
6756             Ilya Verbin  <ilya.verbin@intel.com>
6757             Kirill Yukhin  <kirill.yukhin@intel.com>
6758             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
6760         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
6761         Use new names.
6762         (ix86_expand_vector_move_misalign): Support new unaligned load and
6763         stores and use new names.
6764         (CODE_FOR_sse2_storedqu): Rename to ...
6765         (CODE_FOR_sse2_storedquv16qi): ... this.
6766         (CODE_FOR_sse2_loaddqu): Rename to ...
6767         (CODE_FOR_sse2_loaddquv16qi): ... this.
6768         (CODE_FOR_avx_loaddqu256): Rename to ...
6769         (CODE_FOR_avx_loaddquv32qi): ... this.
6770         (CODE_FOR_avx_storedqu256): Rename to ...
6771         (CODE_FOR_avx_storedquv32qi): ... this.
6772         * config/i386/i386.md (fpint_logic): New.
6773         * config/i386/sse.md (VMOVE): Extend for AVX512.
6774         (VF): Ditto.
6775         (VF_128_256): New.
6776         (VF_512): Ditto.
6777         (VI_UNALIGNED_LOADSTORE): Ditto.
6778         (sse2_avx_avx512f): Ditto.
6779         (sse2_avx2): Extend for AVX512.
6780         (sse4_1_avx2): Ditto.
6781         (avx2_avx512f): New.
6782         (sse): Extend for AVX512.
6783         (sse2): Ditto.
6784         (sse4_1): Ditto.
6785         (avxsizesuffix): Ditto.
6786         (sseintvecmode): Ditto.
6787         (ssePSmode): Ditto.
6788         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
6789         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
6790         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6791         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
6792         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6793         (<sse2_avx_avx512f>_storedqu<mode): ... this.
6794         (<sse>_movnt<mode>): Replace constraint "x" with "v".
6795         (STORENT_MODE): Extend for AVX512.
6796         (*absneg<mode>2): Replace constraint "x" with "v".
6797         (*mul<mode>3): Ditto.
6798         (*ieee_smin<mode>3): Ditto.
6799         (*ieee_smax<mode>3): Ditto.
6800         (avx_cmp<mode>3): Replace VF with VF_128_256.
6801         (*<sse>_maskcmp<mode>3_comm): Ditto.
6802         (<sse>_maskcmp<mode>3): Ditto.
6803         (<sse>_andnot<mode>3): Extend for AVX512.
6804         (<code><mode>3, anylogic): Replace VF with VF_128_256.
6805         (<code><mode>3, fpint_logic): New.
6806         (*<code><mode>3): Extend for AVX512.
6807         (avx512flogicsuff): New.
6808         (avx512f_<logic><mode>): Ditto.
6809         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
6810         VF_128_256.
6811         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
6812         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
6813         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
6814         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
6815         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
6816         (xop_vpermil2<mode>3): Ditto.
6817         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
6818         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
6819         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
6820         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
6822 2013-09-25  Tom Tromey  <tromey@redhat.com>
6824         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
6825         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
6826         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
6827         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
6828         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
6829         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
6830         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
6831         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
6832         (GRAPHITE_HTAB_H): Remove.
6834 2013-09-25  Tom Tromey  <tromey@redhat.com>
6836         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
6838 2013-09-25  Tom Tromey  <tromey@redhat.com>
6840         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
6842 2013-09-25  Tom Tromey  <tromey@redhat.com>
6844         * config/i386/t-i386 (i386.o): Remove.
6845         (i386-c.o): Use COMPILE and POSTCOMPILE.
6847 2013-09-25  Tom Tromey  <tromey@redhat.com>
6849         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
6851 2013-09-25  Tom Tromey  <tromey@redhat.com>
6853         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
6854         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
6855         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
6856         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
6857         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
6858         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
6859         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
6860         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
6861         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
6862         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
6863         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
6864         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
6865         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
6866         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
6867         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
6868         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
6869         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
6870         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
6871         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
6872         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
6873         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
6874         (tree-ssa-pre.o, tree-ssa-sccvn.o)
6875         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
6876         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
6877         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
6878         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
6879         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
6880         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
6881         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
6882         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
6883         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
6884         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
6885         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
6886         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
6887         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
6888         (graphite-clast-to-gimple.o, graphite-dependences.o)
6889         (graphite-interchange.o, graphite-poly.o)
6890         (graphite-scop-detection.o, graphite-sese-to-poly.o)
6891         (graphite-optimize-isl.o, tree-vect-loop.o)
6892         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
6893         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
6894         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
6895         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
6896         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
6897         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
6898         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
6899         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
6900         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
6901         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
6902         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
6903         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
6904         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
6905         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
6906         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
6907         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
6908         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
6909         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
6910         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
6911         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
6912         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
6913         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
6914         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
6915         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
6916         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
6917         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
6918         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
6919         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
6920         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
6921         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
6922         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
6923         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
6924         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
6925         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
6926         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
6927         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
6928         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
6929         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
6930         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
6931         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
6932         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
6933         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
6934         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
6935         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
6936         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
6937         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
6938         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
6939         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
6940         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
6941         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
6942         (gcov-dump.o): Remove.
6943         (default-c.o): Use COMPILE and POSTCOMPILE.
6944         (CFLAGS-gcc.o): New variable.
6945         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
6947 2013-09-25  Tom Tromey  <tromey@redhat.com>
6949         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
6950         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
6951         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
6952         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
6953         (c-family/c-pragma.o, c-family/c-pretty-print.o)
6954         (c-family/c-semantics.o, c-family/c-ada-spec.o)
6955         (c-family/array-notation-common.o, c-family/stub-objc.o)
6956         (c-family/c-ubsan.o): Remove.
6958 2013-09-25  Tom Tromey  <tromey@redhat.com>
6960         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
6962 2013-09-25  Tom Tromey  <tromey@redhat.com>
6964         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
6965         to add -DENABLE_SHARED_LIBGCC.
6966         (gcc.o): Don't use subshell.
6968 2013-09-25  Tom Tromey  <tromey@redhat.com>
6970         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
6971         * configure.ac: Don't invoke AM_PROG_CC_C_O.
6972         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
6973         * configure, config.in: Rebuild.
6975 2013-09-25  Tom Tromey  <tromey@redhat.com>
6977         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
6978         (COMPILE, POSTCOMPILE): New variables.
6979         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
6980         (DEPFILES): New variable.
6981         Include ".Po" files.
6982         * configure.ac: Add checks for dependency checking.
6983         * configure, aclocal.m4: Regenerate.
6985 2013-09-25  Tom Tromey  <tromey@redhat.com>
6987         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
6988         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
6990 2013-09-25  Tom Tromey  <tromey@redhat.com>
6992         * Makefile.in (generated_files): Add options.h,
6993         target-hooks-def.h, insn-opinit.h,
6994         common/common-target-hooks-def.h, pass-instances.def,
6995         c-family/c-target-hooks-def.h.
6997 2013-09-25  Jeff Law  <law@redhat.com>
6999         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
7000         than foo[foo.length () - 1] to access last member in a vec.
7001         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
7003 2013-09-25  Richard Biener  <rguenther@suse.de>
7005         PR middle-end/58521
7006         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
7008 2013-09-25  Jan Hubicka  <jh@suse.cz>
7010         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
7011         test.
7013 2013-09-25  Marek Polacek  <polacek@redhat.com>
7015         PR sanitizer/58420
7016         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
7017         when determining the type name.
7019 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
7021         * config/sh/sh.md: Fix formatting.
7023 2013-09-24  Xinliang David Li  <davidxl@google.com>
7025         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
7026         max peel iterations parameter.
7027         * param.def: New parameter.
7028         * doc/invoke.texi: Document New parameter.
7030 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
7032         * gimple-pretty-print.c: Various whitespace tweaks.
7033         * tree-core.h: Likewise.
7034         * tree-pretty-print.c: Likewise.
7035         * tree-ssa-alias.c: Likewise.
7036         * tree-ssa-copy.c: Likewise.
7037         * tree-ssanames.c: Likewise.
7038         * tree-ssanames.h: Likewise.
7039         * tree-vrp.c: Likewise.
7041 2013-09-24  Alan Modra  <amodra@gmail.com>
7043         PR middle-end/57134
7044         PR middle-end/57586
7045         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
7046         for output operands that disallow regs.  Don't use EXPAND_WRITE on
7047         inout operands.
7049 2013-09-24  Richard Biener  <rguenther@suse.de>
7051         PR middle-end/58513
7052         * tree.c (reference_alias_ptr_type): Move ...
7053         * alias.c (reference_alias_ptr_type): ... here and implement
7054         in terms of the new reference_alias_ptr_type_1.
7055         (ref_all_alias_ptr_type_p): New helper.
7056         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
7057         use ref_all_alias_ptr_type_p.
7058         (get_deref_alias_set): Add flag_strict_aliasing check here.
7059         (reference_alias_ptr_type_1): New function, split out from ...
7060         (get_alias_set): ... here.
7061         (alias_ptr_types_compatible_p): New function.
7062         * alias.h (reference_alias_ptr_type): Declare.
7063         (alias_ptr_types_compatible_p): Likewise.
7064         * tree.h (reference_alias_ptr_type): Remove.
7065         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
7066         to compare MEM_REF alias types.
7068 2013-09-24  Richard Biener  <rguenther@suse.de>
7070         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
7072 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7074         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
7075         reload helper function arrays into a single array reg_addr.
7076         (reload_fpr_gpr): Likewise.
7077         (reload_gpr_vsx): Likewise.
7078         (reload_vsx_gpr): Likewise.
7079         (struct rs6000_reg_addr): Likewise.
7080         (reg_addr): Likewise.
7081         (rs6000_debug_reg_global): Change rs6000_vector_reload,
7082         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
7083         (rs6000_init_hard_regno_mode_ok): Likewise.
7084         (rs6000_secondary_reload_direct_move): Likewise.
7085         (rs6000_secondary_reload): Likewise.
7087         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
7088         constraints: wu, ww, and wy.  Repurpose wv constraint added during
7089         power8 changes.  Put wg constraint in alphabetical order.
7091         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
7092         for future work to add ISA 2.07 VSX single precision support.
7093         (-mvsx-scalar-double): Change default from -1 to 1, update
7094         documentation comment.
7095         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
7096         (-mupper-regs-df): New debug switch to control whether DF values
7097         can go in the traditional Altivec registers.
7098         (-mupper-regs-sf): New debug switch to control whether SF values
7099         can go in the traditional Altivec registers.
7101         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
7102         and wy constraints.
7103         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
7104         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
7105         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
7106         (rs6000_debug_legitimate_address_p): Print if we are running
7107         before, during, or after reload.
7108         (rs6000_secondary_reload): Add a comment.
7109         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
7111         * config/rs6000/constraints.md (wa constraint): Sort w<x>
7112         constraints.  Update documentation string.
7113         (wd constraint): Likewise.
7114         (wf constraint): Likewise.
7115         (wg constraint): Likewise.
7116         (wn constraint): Likewise.
7117         (ws constraint): Likewise.
7118         (wt constraint): Likewise.
7119         (wx constraint): Likewise.
7120         (wz constraint): Likewise.
7121         (wu constraint): New constraint for ISA 2.07 SFmode scalar
7122         instructions.
7123         (ww constraint): Likewise.
7124         (wy constraint): Likewise.
7125         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
7126         in the previous submissions.
7127         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
7129 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
7131         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
7132         (REG_BR_PROB): Say that the probability is stored in an int_list.
7133         * reg-notes.def: Update commentary to mention INT_LIST.
7134         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
7135         (INT_LIST): New rtx.
7136         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
7137         * rtlanal.c (int_reg_note_p): New function.
7138         (alloc_reg_note): Assert that the note does not have an int argument.
7139         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
7140         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
7141         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
7142         rather than an INSN_LIST.  Handle INT_LIST.
7143         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
7144         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
7145         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
7146         Manipulate them as ints rather than rtxes.
7147         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
7148         * regmove.c (copy_src_to_dest): Likewise.
7149         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
7151         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
7152         into the cases that need it.
7153         * config/arm/arm.c (arm_unwind_emit): Likewise.
7155         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
7156         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
7157         * loop-doloop.c (add_test, doloop_modify): Likewise.
7158         * loop-unswitch.c (compare_and_jump_seq): Likewise.
7159         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
7160         * predict.c (combine_predictions_for_insn): Likewise.
7161         * print-rtl.c (print_rtx): Handle INT_LIST.
7162         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
7163         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
7164         * config/arm/arm.c (emit_unlikely_jump): Likewise.
7165         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
7166         (ix86_split_fp_branch, predict_jump): Likewise.
7167         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
7168         * config/sh/sh.c (expand_cbranchsi4): Likewise.
7169         * config/spu/spu.c (ea_load_store_inline): Likewise.
7171         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
7172         value of a REG_BR_PROB note.
7173         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
7174         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
7175         * emit-rtl.c (try_split): Likewise.
7176         * predict.c (br_prob_note_reliable_p): Likewise.
7177         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
7178         * reorg.c (mostly_true_jump): Likewise.
7179         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
7180         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
7181         * config/i386/i386.c (ix86_print_operand): Likewise.
7182         * config/ia64/ia64.c (ia64_print_operand): Likewise.
7183         * config/mmix/mmix.c (mmix_print_operand): Likewise.
7184         * config/rs6000/rs6000.c (output_cbranch): Likewise.
7185         * config/s390/s390.c (s390_expand_tbegin): Likewise.
7186         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
7187         * config/sparc/sparc.c (output_cbranch): Likewise.
7188         * config/spu/spu.c (get_branch_target): Likewise.
7189         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
7190         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
7192 2013-09-23  Jan Hubicka  <jh@suse.cz>
7194         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
7195         for ipa-devirt.
7196         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
7197         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
7198         of external calls
7199         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
7200         (gimple_fold_call): Dump inconsistent devirtualizations; add
7201         sanity check for type based devirtualizations.
7202         * ipa-prop.c: Include ipa-utils.h
7203         (ipa_intraprocedural_devirtualization): Add sanity check.
7204         (try_make_edge_direct_virtual_call): Likewise.
7206 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
7208         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
7209         assignment statements.
7211 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
7213         * gimple-pretty-print.c (dump_ssaname_info): New function.
7214         (dump_gimple_phi): Call it.
7215         (pp_gimple_stmt_1): Likewise.
7216         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
7217         (range_info_def): Declare.
7218         * tree-pretty-print.c (pp_double_int): New function.
7219         (dump_generic_node): Call it.
7220         * tree-pretty-print.h (pp_double_int): Declare.
7221         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
7222         * tree-ssanames.h (range_info_def): New structure.
7223         (value_range_type): Move definition here.
7224         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
7225         Declare.
7226         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
7227         initialization.
7228         (set_range_info): New function.
7229         (get_range_info): Likewise.
7230         (duplicate_ssa_name_range_info): Likewise.
7231         (duplicate_ssa_name_fn): Check pointer type and call
7232         duplicate_ssa_name_range_info.
7233         * tree-ssa-copy.c (fini_copy_prop): Likewise.
7234         * tree-vrp.c (value_range_type): Remove definition, now in
7235         tree-ssanames.h.
7236         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
7237         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
7238         (SSA_NAME_RANGE_INFO): New macro.
7240 2013-09-23  Richard Biener  <rguenther@suse.de>
7242         PR tree-optimization/58464
7243         * tree-ssa-pre.c (phi_trans_lookup): Remove.
7244         (phi_trans_add): Change to add conditionally on being not present.
7245         (phi_translate_1): Remove recursion detection here.
7246         (phi_translate): Pre-seed the cache with NULL to catch
7247         recursion here in a more generic way.
7248         (bitmap_find_leader): Adjust comment.
7249         (get_representative_for): Dump value-numbers.
7250         (create_expression_by_pieces): Likewise.
7251         (insert_into_preds_of_block): Likewise.
7253 2013-09-23  Christian Bruel  <christian.bruel@st.com>
7255         PR target/58475
7256         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
7257         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
7259 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
7261         Revert r202780:
7262         2013-09-20  Renlin Li  <renlin.li@arm.com>
7264         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7265         plus_constant.
7266         (aarch64_expand_epilogue): Likewise.
7267         (aarch64_legitimize_reload_address): Likewise.
7269 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
7271         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
7272         NULL_TREE before pushing them onto the vector.  Likewise for labels.
7274 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
7276         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
7277         comparison operators when -fno-trapping-math is in effect.
7278         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
7279         comparison operators in TFmode and assert that unsupported operators
7280         cannot reach here.
7281         (ia64_print_operand): Likewise.
7283 2013-09-21  Jan Hubicka  <jh@suse.cz>
7285         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
7286         (sse_typeless_stores): Enable for core
7287         (sse_load0_by_pxor): Likewise.
7288         (four_jump_limit): Disable for core.
7289         (pad_returns): Likewise.
7290         (avoid_vector_decode): Likewise.
7291         (fuse_cmp_and_branch): Enable for cores.
7292         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
7294 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7296         PR middle-end/56791
7297         * config/pa/pa.c (pa_option_override): Disable auto increment and
7298         decrement instructions until reload is completed.
7300         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
7301         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
7302         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
7304 2013-09-20  DJ Delorie  <dj@redhat.com>
7305             Nick Clifton  <nickc@redhat.com>
7307         * config/rl78/rl78.c: Various whitespace and comment tweaks.
7308         (need_to_save): Save bank 0 on interrupts.
7309         (characterize_address): Strip far address wrappers.
7310         (rl78_as_legitimate_address): Likewise.
7311         (transcode_memory_rtx): Likewise.
7312         (rl78_peep_movhi_p): Disable this peephole after devirt.
7313         (rl78_propogate_register_origins): Forget all origins when a
7314         CLOBBER is seen.
7315         * config/rl78/rl78-virt.md: Various whitespace tweaks.
7316         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
7317         peephole2's.
7318         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
7319         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
7320         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
7321         relocating.
7322         * config/rl78/constraints.md: Various whitespace and paren tweaks.
7324 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7326         * config/pa/pa.md: In "scc" insn patterns, change output template to
7327         handle const0_rtx in reg_or_0_operand operands.
7329 2013-09-20  Martin Husemann  <martin@NetBSD.org>
7331         PR target/56875
7332         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
7333         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
7335 2013-09-20  Richard Biener  <rguenther@suse.de>
7337         PR middle-end/58484
7338         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
7339         remembering SSA name version and block index.
7340         (new_scev_info_str): Adjust.
7341         (hash_scev_info): Likewise.  Also hash the block index.
7342         (eq_scev_info): Adjust.
7343         (find_var_scev_info): Likewise.
7344         (struct instantiate_cache_entry): Remove.
7345         (struct instantiate_cache_type): Use a htab to map name, block
7346         to chrec.
7347         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7348         (get_instantiated_value_entry): Likewise.
7349         (hash_idx_scev_info, eq_idx_scev_info): New functions.
7350         (instantiate_scev_name): Adjust.
7352 2013-09-20  Jeff Law  <law@redhat.com>
7354         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
7356 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
7358         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7359         Call aarch64_simd_expand_args to update op[argc].
7361 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7363         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
7364         plugin argument.
7366 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7368         * gengtype.c (file_rules): Added rule for *.cc files.
7369         (get_output_file_with_visibility): Give fatal message when no
7370         rules found.
7372 2013-09-20  Renlin Li  <renlin.li@arm.com>
7374         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
7375         (aarch64_expand_epilogue): Likewise.
7376         (aarch64_legitimize_reload_address): Likewise.
7378 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7380         PR middle-end/57748
7381         * expr.c (expand_assignment): Remove misalignp code path.
7383 2013-09-20  Marek Polacek  <polacek@redhat.com>
7385         PR sanitizer/58413
7386         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
7387         TYPE_PRECISION.  Add asserts.
7389 2013-09-20  Richard Biener  <rguenther@suse.de>
7391         PR tree-optimization/58453
7392         * tree-loop-distribution.c (distribute_loop): Apply the cost
7393         model for -ftree-loop-distribute-patterns, too.
7395 2013-09-20  Richard Biener  <rguenther@suse.de>
7397         PR middle-end/58473
7398         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
7399         make type comparison less strict.
7401 2013-09-20  Alan Modra  <amodra@gmail.com>
7403         * configure: Regenerate.
7404         * aclocal.m4: Regenerate.
7406 2013-09-20  Marek Polacek  <polacek@redhat.com>
7408         PR other/58467
7409         * doc/extend.texi: Document that attribute used is meant to be used
7410         on variables with static storage duration.
7412 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7414         PR tree-optimization/58472
7415         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
7416         simd_lane_access set inv_p = false.
7417         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
7418         the simduid magic VAR_DECL.
7420 2013-09-19  Jan Hubicka  <jh@suse.cz>
7422         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
7424 2013-09-17  Jeff Law  <law@redhat.com>
7426         * tree-ssa-dom.c (record_temporary_equivalences): New function
7427         split out of dom_opt_dom_walker::after_dom_children.
7428         (dom_opt_dom_walker::thread_across_edge): Move common code
7429         in here from dom_opt_dom_walker::after_dom_children.
7430         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
7432 2013-09-19  Jan Hubicka  <jh@suse.cz>
7434         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
7435         (TARGET_GENERIC): Use PROCESOR_GENERIC
7436         (enum processor_type): Unify generic32 and 64.
7437         * i386.md (cpu): Likewise.
7438         * x86-tune.def (use_leave): Enable for generic32.
7439         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
7440         * athlon.md: Change generic64 to generic in all occurences.
7441         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
7442         (ix86_target_macros_internal): Likewise.
7443         * driver-i386.c (host_detect_local_cpu): Likewise.
7444         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
7445         to ..
7446         (generic_memcpy, generic_memset, generic_cost): This one.
7447         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
7448         (m_GENERIC32, m_GENERIC64): Remove.
7449         (m_GENERIC): Turn into one flag.
7450         (processor_target): Unify generic tunnings.
7451         (ix86_option_override_internal): Replace generic32/64 by generic.
7452         (ix86_issue_rate): Likewise.
7453         (ix86_adjust_cost): Likewise.
7455 2013-09-19  Jan Hubicka  <jh@suse.cz>
7457         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
7458         of speculative flag.
7460 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7462         * omp-low.c (expand_omp_sections): Always pass len - 1 to
7463         GOMP_sections_start, even if !exit_reachable.
7465 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7467         * lra-constraints.c (need_for_all_save_p): Use macro
7468         HARD_REGNO_CALL_PART_CLOBBERED.
7469         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
7470         set up pseudo conflict hard regs.
7472 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7474         PR target/58452
7475         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
7476         operands.
7478 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7480         PR rtl-optimization/58438
7481         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
7482         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
7483         from previous subpasses.
7485 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
7487         * arm.c (arm_get_frame_offsets): Validate architecture supports
7488         LDRD/STRD before accepting the tuning preference.
7489         (arm_expand_prologue): Likewise.
7490         (arm_expand_epilogue): Likewise.
7492 2013-09-18  Richard Biener  <rguenther@suse.de>
7494         PR tree-optimization/58417
7495         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
7496         have chrecs with symbols defined in the loop as operands.
7497         (chrec_fold_multiply): Likewise.
7498         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
7499         parameters before folding binary operations.
7500         (struct instantiate_cache_entry_hasher): Remove.
7501         (struct instantiate_cache_type): Use a pointer-map.
7502         (instantiate_cache_type::instantiate_cache_type): New function.
7503         (instantiate_cache_type::get): Likewise.
7504         (instantiate_cache_type::set): Likewise.
7505         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7506         (get_instantiated_value_entry): Likewise.
7507         (global_cache): New global.
7508         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
7509         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
7510         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
7511         (instantiate_scev_name): Adjust.
7512         (instantiate_scev): Construct global instead of local cache.
7513         (resolve_mixers): Likewise.
7515 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
7516             Paolo Carlini  <paolo.carlini@oracle.com>
7518         PR c++/58458
7519         * doc/implement-cxx.texi: Fix references to the C++ standards.
7521 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
7523         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
7524         * tree-vect-data-refs.c (vect_analyze_data_refs): For
7525         simd_lane_access drs, update also DR_ALIGNED_TO.
7527 2013-09-18  Marek Polacek  <polacek@redhat.com>
7529         PR sanitizer/58411
7530         * doc/extend.texi: Document no_sanitize_undefined attribute.
7531         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
7532         no_sanitize_undefined attribute.
7534 2013-09-18  Nick Clifton  <nickc@redhat.com>
7536         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
7537         (ASM_DECLARE_FUNCTION_NAME): Define.
7539 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
7541         * compare-elim.c (find_comparison_dom_walker): New class
7542         (find_comparisons_in_bb): Rename to
7543         find_comparison_dom_walker::before_dom_children
7544         (find_comparisons): Adjust
7545         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
7546         adjust.
7547         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
7548         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
7549         (init_walk_dominator_tree): Remove declaration.
7550         (fini_walk_dominator_tree): Remove declaration.
7551         * fwprop.c (single_def_use_dom_walker): New class
7552         (single_def_use_enter_block): Convert to
7553         single_def_use_dom_walker::before_dom_children.
7554         (single_def_use_leave_block): Convert to
7555         single_def_use_dom_walker::after_dom_children.
7556         (build_single_def_use_links): Adjust.
7557         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
7558         class.
7559         (find_candidates_in_block): Convert to
7560         find_candidates_dom_walker::before_dom_children.
7561         (execute_strength_reduction): Adjust.
7562         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
7563         (sese_dom_walker): New class.
7564         (sese_dom_walker::sese_dom_walker): New constructor.
7565         (sese_dom_walker::~sese_dom_walker): New destructor.
7566         (build_sese_conditions_before): Convert to
7567         sese_dom_walker::before_dom_children.
7568         (build_sese_conditions_after): Convert to
7569         sese_dom_walker::after_dom_children.
7570         (build_poly_scop): Adjust
7571         * tree-into-ssa.c (rewrite_dom_walker): New class
7572         (rewrite_enter_block): Convert to
7573         rewrite_dom_walker::before_dom_children.
7574         (rewrite_leave_block): Convert to
7575         rewrite_dom_walker::after_dom_children.
7576         (rewrite_update_dom_walker): New class.
7577         (rewrite_update_enter_block): Convert to
7578         rewrite_update_dom_walker::before_dom_children.
7579         (rewrite_update_leave_block): Convert to
7580         rewrite_update_dom_walker::after_dom_children.
7581         (rewrite_blocks, rewrite_into_ssa): Adjust.
7582         (mark_def_dom_walker): New class.
7583         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
7584         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
7585         (mark_def_sites_blocks): Convert to
7586         mark_def_dom_walker::before_dom_children.
7587         (mark_def_site_blocks): Remove.
7588         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
7589         (tree_ssa_dominator_optimize): Adjust.
7590         (dom_thread_across_edge): Convert to method
7591         dom_opt_dom_walker::thread_across_edge.
7592         (dom_opt_enter_block): Convert to member function
7593         dom_opt_dom_walker::before_dom_children.
7594         (dom_opt_leave_block): Convert to member function
7595         dom_opt_dom_walker::after_dom_children.
7596         * tree-ssa-dse.c (dse_dom_walker): New class.
7597         (dse_enter_block): Convert to member function
7598         dse_dom_walker::before_dom_children.
7599         (tree_ssa_dse): Adjust.
7600         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
7601         (determine_invariantness_stmt): Convert to method
7602         invariantness_dom_walker::before_dom_children.
7603         (determine_invariantness): Remove
7604         (move_computations_dom_walker): New class.
7605         (move_computations_stmt): Convert to method
7606         move_computations_dom_walker::before_dom_children.
7607         (move_computations, tree_ssa_lim): Adjust.
7608         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
7609         (nt_init_block): Convert to method
7610         notrappping_dom_walker::before_dom_children.
7611         (nt_fini_block): Convert to method
7612         method nontrapping_dom_walker::after_dom_children.
7613         (get_non_trapping): Adjust.
7614         * tree-ssa-pre.c (eliminate_dom_walker): New class.
7615         (eliminate_bb): Convert to method
7616         eliminate_dom_walker::before_dom_children.
7617         (eliminate_leave_block): Convert to method
7618         eliminate_dom_walker::after_dom_children.
7619         (eliminate): Adjust.
7620         * tree-ssa-strlen.c (strlen_dom_walker): New class.
7621         (strlen_enter_block): Convert to method
7622         strlen_dom_walker::before_dom_children.
7623         (strlen_leave_block): Convert to method
7624         method strlen_dom_walker::after_dom_children.
7625         (tree_ssa_strlen): Adjust.
7626         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
7627         (tree_ssa_uncprop): Adjust.
7628         (uncprop_leave_block): Convert to method
7629         uncprop_dom_walker::after_dom_children.
7630         (uncprop_leave_block): Convert to method
7631         uncprop_dom_walker::before_dom_children.
7633 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
7635         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
7636         branch in same basic block.  Check both src and dest of the move insn.
7638 2013-09-17  Nick Clifton  <nickc@redhat.com>
7640         * config/rl78/rl78-real.md (bf): New pattern.
7641         (bt): New pattern.
7642         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
7643         (rl78_print_operand): Do not put a # before a %B.
7644         * config/rl78/rl78.opt: Tweak doc strings.
7646 2013-09-17  DJ Delorie  <dj@redhat.com>
7648         * config/rl78/constraints.md (Wcv): Allow up to $r31.
7649         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
7650         (rl78_option_override): Likewise, if -mallregs.
7651         (is_virtual_register): Likewise.
7652         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
7653         (REGNO_OK_FOR_BASE_P): Likewise.
7654         * config/rl78/rl78.opt (-mallregs): New.
7656 2013-09-17  Nick Clifton  <nickc@redhat.com>
7658         * config/rl78/rl78.c (need_to_save): Change return type to bool.
7659         For interrupt functions: save all call clobbered registers if the
7660         interrupt handler is not a leaf function.
7661         (rl78_expand_prologue): Always recompute the frame information.
7662         For interrupt functions: only select bank 0 if one of the bank 0
7663         registers is going to be psuhed.
7665 2013-09-17  DJ Delorie  <dj@redhat.com>
7667         * config/rl78/constraints.md: For each W* constraint, rename to C*
7668         and create a W* constraint that checks for an optional ES: prefix
7669         pattern also.
7670         * config/rl78/rl78.md (UNS_ES_ADDR): New.
7671         (es_addr): New.  Used to wrap far addresses.
7672         * config/rl78/rl78-protos.h (rl78_es_addr): New.
7673         (rl78_es_base): New.
7674         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
7675         wrapped far addresses.
7676         (rl78_print_operand_1): Unwrap far addresses before processing.
7677         (rl78_lo16): Wrap far addresses in unspecs.
7678         (rl78_es_addr): New.
7679         (rl78_es_base): New.
7680         (insn_ok_now): Check for not-yet-wrapped far addresses.
7681         (transcode_memory_rtx): Properly re-wrap far addresses.
7683 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7685         * config/sparc/t-rtems: Add leon3 multilibs.
7687 2013-09-17  Cong Hou  <congh@google.com>
7689         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
7690         when checking the dot production pattern. The type of rhs operand
7691         of multiply is now checked correctly.
7693 2013-09-17  Jeff Law  <law@redhat.com>
7695         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7696         edge implied equivalences into successor phis.
7697         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
7698         here from tree-ssa-threadedge.c.
7699         (mark_threaded_blocks): When threading through a joiner, if both
7700         successors of the joiner's clone reach the same block, verify the
7701         PHI arguments are equal.  If not, cancel the jump threading request.
7702         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
7703         tree-ssa-threadupdate.c
7704         (thread_across_edge): Don't check PHI argument equality when
7705         threading through joiner block here.
7707 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
7709         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
7710         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
7711         (warn_uninit, warn_uninitialized_vars,
7712         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7713         Move to tree-ssa-uninit.c.
7714         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
7715         (has_undefined_value_p): New.  Pass dependant parts of
7716         ssa_undefined_value_p.
7717         (uninit_undefined_value_p): Use has_undefined_value_p.
7718         (warn_uninit, warn_uninitialized_vars,
7719         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7720         Move from tree-ssa.c.
7721         * tree-ssa.h: Adjust prototypes.
7723 2013-09-17  Jan Hubicka  <jh@suse.cz>
7725         PR middle-end/58332
7726         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
7727         * ipa-inline.c (can_inline_edge_p): Do not downgrade
7728         FUNCTION_NOT_OPTIMIZED.
7729         * ipa-inline-analysis.c (compute_inline_parameters): Function
7730         not optimized is not inlinable unless it is alwaysinline.
7731         (inline_analyze_function): Force calls in not optimized
7732         function not inlinable.
7734 2013-09-17  Jan Hubicka  <jh@suse.cz>
7736         PR middle-end/58329
7737         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
7738         to return NULL.
7739         * ipa.c (function_and_variable_visibility): Likewise.
7740         * ipa-profile.c (ipa_profile): Likewise.
7742 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7744         PR ipa/58398
7745         * cgraph.c (cgraph_function_body_availability): Check for ifunc
7746         attribute, and don't inline the resolver in this case.
7748 2013-09-17  Teresa Johnson  <tejohnson@google.com>
7750         * coverage.c (get_coverage_counts): Add missing newline.
7752 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7754         PR tree-optimization/58088
7755         * fold-const.c (mask_with_trailing_zeros): New function.
7756         (fold_binary_loc): Make sure we don't recurse infinitely
7757         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
7758         Use mask_with_trailing_zeros where appropriate.
7760 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
7762         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
7763         of 'prev' var to get better distance estimation.
7765 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
7767         * tree-inline.h (struct copy_body_data): Add transform_parameter.
7768         * tree-inline.c (is_parameter_of): New predicate.
7769         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
7770         a parameter has been remapped.
7771         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
7772         (optimize_inline_calls): Initialize transform_parameter.
7773         (copy_gimple_seq_and_replace_locals): Likewise.
7774         (tree_function_versioning): Likewise.
7775         (maybe_inline_call_in_expr): Likewise.
7777 2013-09-17  Nick Clifton  <nickc@redhat.com>
7779         * config/msp430/msp430-protos.h: Add prototypes for new functions.
7780         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
7781         interrupt handlers.
7782         (is_attr_func): New function.
7783         (msp430_is_interrupt_func): New function.
7784         (is_naked_func): New function.
7785         (is_reentrant_func): New function.
7786         (is_critical_func): New function.
7787         (msp430_start_function): Add annotations for function attributes.
7788         (msp430_attr): New function.
7789         (msp430_attribute_table): New.
7790         (msp430_function_section): New function.
7791         (TARGET_ASM_FUNCTION_SECTION): Define.
7792         (msp430_builtin): New enum.
7793         (msp430_init_builtins): New function.
7794         (msp430_builtin_devl): New function.
7795         (msp430_expand_builtin): New function.
7796         (TARGET_INIT_BUILTINS): Define.
7797         (TARGET_EXPAND_BUILTINS): Define.
7798         (TARGET_BUILTIN_DECL): Define.
7799         (msp430_expand_prologue): Add support for naked, interrupt,
7800         critical and reentrant functions.
7801         (msp430_expand_epilogue): Likewise.
7802         (msp430_print_operand): Handle 'O' character.
7803         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
7804         NO_TRAMPOLINES.
7805         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
7806         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
7807         (pushm): Use a 'n' rather than an 'i' constraint.
7808         (msp_return): Add generation of the interrupt return instruction.
7809         (disable_interrupts): New pattern.
7810         (enable_interrupts): New pattern.
7811         (push_intr_state): New pattern.
7812         (pop_intr_state): New pattern.
7813         (bic_SR): New pattern.
7814         (bis_SR): New pattern.
7815         * doc/extend.texi: Document MSP430 function attributes and builtin
7816         functions.
7818 2013-09-17  Richard Biener  <rguenther@suse.de>
7820         PR tree-optimization/58432
7821         * tree-loop-distribution.c (tree_loop_distribution): Also
7822         scan PHIs for outside loop uses and seed a partition from them.
7824 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
7826         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
7827         (restructure_reference): Call backtrace_base_for_ref.
7829 2013-09-17  Alan Modra  <amodra@gmail.com>
7831         PR target/57589
7832         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
7833         patch.
7835 2013-09-16  DJ Delorie  <dj@redhat.com>
7837         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
7838         vregs location for RL78/G10.
7839         (rl78_expand_prologue): Avoid SEL on G10.
7840         (rl78_expand_epilogue): Likewise.
7841         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
7842         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
7843         __RL78_G10__ when appropriate.
7844         (ASM_SPEC): Pass -mg10 along to the assembler.
7845         * config/rl78/rl78.md (sel_rb): Disable for G10.
7846         * config/rl78/rl78.opt: Add -mg10 option.
7847         * config/rl78/t-rl78: Add -mg10 multilib.
7849 2013-09-16  Xinliang David Li  <davidxl@google.com>
7851         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
7852         * omp-low.c (omp_max_vf): Ditto.
7853         (expand_omp_simd): Ditto.
7854         * tree-vectorizer.c (vectorize_loops): Ditto.
7855         (gate_vect_slp): Ditto.
7856         (gate_increase_alignment): Ditto.
7857         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
7858         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
7859         (gate_tree_vectorize): Name change.
7860         (tree_vectorize): Ditto.
7861         (pass_vectorize::gate): Call new function.
7862         (pass_vectorize::execute): Ditto.
7863         * opts.c: O3 default setting change.
7864         (finish_options): Check new flag.
7865         * doc/invoke.texi: Document new flags.
7866         * common.opt: New flags.
7868 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
7870         * doc/tm.texi.in (Cond Exec Macros): Remove node.
7871         (Condition Code): Don't reference it.
7872         * doc/tm.texi: Regenerate.
7874 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
7876         PR middle-end/58418
7877         * lra-constraints.c (undo_optional_reloads): Consider all optional
7878         reload even if it did not get a hard reg.
7880 2013-09-16  Teresa Johnson  <tejohnson@google.com>
7882         * dumpfile.c (dump_loc): Remove newline emission.
7883         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
7884         emission to dump_printf_loc calls where missing.
7885         (vect_mark_for_runtime_alias_test): Ditto.
7886         (vect_analyze_data_ref_dependence): Ditto.
7887         (vect_analyze_data_ref_dependences): Ditto.
7888         (vect_slp_analyze_data_ref_dependence): Ditto.
7889         (vect_slp_analyze_data_ref_dependences): Ditto.
7890         (vect_compute_data_ref_alignment): Ditto.
7891         (vect_update_misalignment_for_peel): Ditto.
7892         (vect_verify_datarefs_alignment): Ditto.
7893         (vector_alignment_reachable_p): Ditto.
7894         (vect_get_data_access_cost): Ditto.
7895         (vect_enhance_data_refs_alignment): Ditto.
7896         (vect_find_same_alignment_drs): Ditto.
7897         (vect_analyze_data_refs_alignment): Ditto.
7898         (vect_analyze_group_access): Ditto.
7899         (vect_analyze_data_ref_access): Ditto.
7900         (vect_analyze_data_ref_accesses): Ditto.
7901         (vect_prune_runtime_alias_test_list): Ditto.
7902         (vect_analyze_data_refs): Ditto.
7903         (vect_create_addr_base_for_vector_ref): Ditto.
7904         (vect_create_data_ref_ptr): Ditto.
7905         (vect_grouped_store_supported): Ditto.
7906         (vect_grouped_load_supported): Ditto.
7907         * value-prof.c (check_counter): Ditto.
7908         (check_ic_target): Ditto.
7909         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
7910         (vect_recog_widen_mult_pattern): Ditto.
7911         (vect_recog_widen_sum_pattern): Ditto.
7912         (vect_recog_over_widening_pattern): Ditto.
7913         (vect_recog_widen_shift_pattern): Ditto.
7914         (vect_recog_rotate_pattern): Ditto.
7915         (vect_recog_vector_vector_shift_pattern): Ditto.
7916         (vect_recog_divmod_pattern): Ditto.
7917         (vect_recog_mixed_size_cond_pattern): Ditto.
7918         (vect_recog_bool_pattern): Ditto.
7919         (vect_pattern_recog_1): Ditto.
7920         (vect_pattern_recog): Ditto.
7921         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
7922         (vect_is_simple_iv_evolution): Ditto.
7923         (vect_analyze_scalar_cycles_1): Ditto.
7924         (vect_get_loop_niters): Ditto.
7925         (vect_analyze_loop_1): Ditto.
7926         (vect_analyze_loop_form): Ditto.
7927         (vect_analyze_loop_operations): Ditto.
7928         (vect_analyze_loop_2): Ditto.
7929         (vect_analyze_loop): Ditto.
7930         (report_vect_op): Ditto.
7931         (vect_is_slp_reduction): Ditto.
7932         (vect_is_simple_reduction_1): Ditto.
7933         (vect_get_known_peeling_cost): Ditto.
7934         (vect_estimate_min_profitable_iters): Ditto.
7935         (vect_model_reduction_cost): Ditto.
7936         (vect_model_induction_cost): Ditto.
7937         (get_initial_def_for_induction): Ditto.
7938         (vect_create_epilog_for_reduction): Ditto.
7939         (vectorizable_reduction): Ditto.
7940         (vectorizable_induction): Ditto.
7941         (vectorizable_live_operation): Ditto.
7942         (vect_loop_kill_debug_uses): Ditto.
7943         (vect_transform_loop): Ditto.
7944         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
7945         (vect_stmt_relevant_p): Ditto.
7946         (process_use): Ditto.
7947         (vect_mark_stmts_to_be_vectorized): Ditto.
7948         (vect_model_simple_cost): Ditto.
7949         (vect_model_promotion_demotion_cost): Ditto.
7950         (vect_model_store_cost): Ditto.
7951         (vect_get_store_cost): Ditto.
7952         (vect_model_load_cost): Ditto.
7953         (vect_get_load_cost): Ditto.
7954         (vect_init_vector_1): Ditto.
7955         (vect_get_vec_def_for_operand): Ditto.
7956         (vect_finish_stmt_generation): Ditto.
7957         (vectorizable_call): Ditto.
7958         (vectorizable_conversion): Ditto.
7959         (vectorizable_assignment): Ditto.
7960         (vectorizable_shift): Ditto.
7961         (vectorizable_operation): Ditto.
7962         (vectorizable_store): Ditto.
7963         (vectorizable_load): Ditto.
7964         (vectorizable_condition): Ditto.
7965         (vect_analyze_stmt): Ditto.
7966         (vect_transform_stmt): Ditto.
7967         (vect_is_simple_use): Ditto.
7968         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
7969         (vect_can_advance_ivs_p): Ditto.
7970         (vect_update_ivs_after_vectorizer): Ditto.
7971         (vect_do_peeling_for_loop_bound): Ditto.
7972         (vect_gen_niters_for_prolog_loop): Ditto.
7973         (vect_update_inits_of_drs): Ditto.
7974         (vect_create_cond_for_alias_checks): Ditto.
7975         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
7976         (vect_build_slp_tree_1): Ditto.
7977         (vect_supported_load_permutation_p): Ditto.
7978         (vect_analyze_slp_instance): Ditto.
7979         (vect_analyze_slp): Ditto.
7980         (vect_make_slp_decision): Ditto.
7981         (vect_detect_hybrid_slp): Ditto.
7982         (vect_bb_vectorization_profitable_p): Ditto.
7983         (vect_slp_analyze_bb_1): Ditto.
7984         (vect_update_slp_costs_according_to_vf): Ditto.
7985         (vect_get_mask_element): Ditto.
7986         (vect_transform_slp_perm_load): Ditto.
7987         (vect_schedule_slp_instance): Ditto.
7988         (vect_schedule_slp): Ditto.
7989         (vect_slp_transform_bb): Ditto.
7990         * profile.c (read_profile_edge_counts): Ditto.
7991         (compute_branch_probabilities): Ditto.
7992         * coverage.c (get_coverage_counts): Ditto.
7994 2013-09-16  Diego Novillo  <dnovillo@google.com>
7996         * tree-core.h: Add missing comment lines from refactoring of tree.h.
7998 2013-09-16  Jan Hubicka  <jh@suse.cz>
8000         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
8001         abstract functions; for static functions check the presence of body.
8003 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8005         * config/aarch64/aarch64-simd-builtins.def (fma): New.
8006         * config/aarch64/aarch64-simd.md
8007         (aarch64_mla_elt<mode>): New.
8008         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
8009         (aarch64_mls_elt<mode>): Likewise.
8010         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
8011         (aarch64_fma4_elt<mode>): Likewise.
8012         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
8013         (aarch64_fma4_elt_to_128v2df): Likewise.
8014         (aarch64_fma4_elt_to_64df): Likewise.
8015         (fnma<mode>4): Likewise.
8016         (aarch64_fnma4_elt<mode>): Likewise.
8017         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8018         (aarch64_fnma4_elt_to_128v2df): Likewise.
8019         (aarch64_fnma4_elt_to_64df): Likewise.
8020         * config/aarch64/iterators.md (VDQSF): New.
8021         * config/aarch64/arm_neon.h
8022         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
8023         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
8025 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8027         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
8028         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8029         (aarch64_mul3_elt_to_128df): Likewise.
8030         (aarch64_mul3_elt_to_64v2df): Likewise.
8031         * config/aarch64/iterators.md (VEL): Also handle DFmode.
8032         (VMUL): New.
8033         (VMUL_CHANGE_NLANES) Likewise.
8034         (h_con): Likewise.
8035         (f): Likewise.
8036         * config/aarch64/arm_neon.h
8037         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
8039 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8041         * config/aarch64/arm_neon.h
8042         (vcvtx_high_f32_f64): Fix parameters.
8044 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8045             Uros Bizjak  <ubizjak@gmail.com>
8047         * config/alpha.c: Include tree-ssanames.h.
8049 2013-09-16  Richard Biener  <rguenther@suse.de>
8051         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
8052         (dot_rdg_1): Handle control_dd.
8053         (create_edge_for_control_dependence): New function.
8054         (create_rdg_edges): Add control dependences if asked for.
8055         (build_rdg): Likewise.
8056         (generate_loops_for_partition): If there are not necessary
8057         control stmts remove all their dependencies.
8058         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
8059         (distribute_loop): Pass on control dependences.
8060         (tree_loop_distribution): Compute control dependences and remove
8061         restriction on number of loop nodes.
8063 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
8065         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
8066         for internal calls.
8068 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
8070         * cse.c (try_const_anchors): Punt on CC modes.
8072 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8074         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
8076 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
8078         PR target/58382
8079         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
8080         calls to word_mode.
8082 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8084         PR target/48094
8085         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
8086         seen.
8087         (darwin_objc1_section): Likewise.
8088         (darwin_file_end): Emit Image Info section when required.
8090 2013-09-14  Jan Hubicka  <jh@suse.cz>
8092         * tree-into-ssa.c (gate_into_ssa): New.
8093         (pass_data_build_ssa): Use it.
8094         * cgraph.h (expand_thunk): Update prototype.
8095         * cgraphunit.c (analyze_function): Expand thunks early.
8096         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
8097         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
8098         set lowered flag; do not add new function.
8099         (assemble_thunks_and_aliases): Update.
8100         * tree-ssa.c (gate_init_datastructures): New gate.
8101         (pass_data_init_datastructures): Use it.
8103 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8105         PR target/58269
8106         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
8107         xmm register set described in the psABI.
8109 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
8111         * dwarf2out.c (should_emit_struct_debug): Add check
8112         for type_decl variable is not NULL.
8114 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
8116         * config.gcc: Use new winnt-c.c target hooks
8117         * config/t-winnt: New file
8118         * config/winnt-c.c: New file
8119         * doc/tm.texi.in: Document new hook
8120         * doc/tm.texi: Regenerated
8122 2013-09-13  Jan Hubicka  <jh@suse.cz>
8124         PR middle-end/58094
8125         * ipa-inline.c (check_callers): New function.
8126         (check_caller_edge): Remove.
8127         (want_inline_function_to_all_callers_p): Also permit alises that are
8128         called dirrectly.
8129         (inline_to_all_callers): Terminate the walk when devirtualization
8130         introduce new calls.
8132 2013-09-13  Jan Hubicka  <jh@suse.cz>
8134         * ipa-inline-analysis.c (struct growth_data): Add node.
8135         (do_estimate_growth_1): Fix detection of recursion.
8137 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
8139         PR tree-optimization/58392
8140         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
8141         to avoid shadowing of outer loop variable.  If
8142         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
8143         replace_by_duplicate_decl simduid of loops that have it set and
8144         set dest_cfun->has_simduid_loops and/or
8145         dest_cfun->has_force_vect_loops.
8146         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
8147         instead of maybe_lookup_decl.
8148         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
8149         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
8150         call.  Copy over force_vect and copy and remap simduid.  Set
8151         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
8152         (copy_cfg_body): Remove blocks_to_copy argument.  Use
8153         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
8154         caller.  Don't set cfun->has_simduid_loops and/or
8155         cfun->has_force_vect_loops here.
8156         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
8157         caller.
8158         (expand_call_inline, tree_function_versioning): Adjust copy_body
8159         callers.
8161 2013-09-13  Martin Jambor  <mjambor@suse.cz>
8163         PR bootstrap/58388
8164         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
8165         the assert if the edge was a speculative one.
8167 2013-09-13  Richard Biener  <rguenther@suse.de>
8169         * tree-data-ref.h (known_dependences_p): Move here ...
8170         * tree-loop-distribution.c (known_dependences_p): ... from here.
8171         (dump_rdg_component, debug_rdg_component): Remove.
8172         (dump_rdg): Adjust.
8173         (generate_loops_for_partition): Use gimple_uid instead of
8174         relying on matching stmt visit order.
8175         (rdg_build_partitions): Take starting stmt vector.
8176         (ldist_gen): Merge into ...
8177         (distribute_loop): ... this function.  Do not compute starting
8178         vertices vector.
8179         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
8181 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8183         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
8184         Set type attribute correctly. Set predicable_short_it attribute.
8185         (cmpsi_shiftsi): Remove %? from output template.
8187 2013-09-13  Richard Biener  <rguenther@suse.de>
8189         * tree-loop-distribution.c (struct rdg_component,
8190         rdg_defs_used_in_other_loops_p, free_rdg_components,
8191         rdg_build_components): Remove.
8192         (stmts_from_loop): Do not record virtual PHIs.
8193         (generate_loops_for_partition): Skip virtual PHIs.
8194         (build_rdg_partition_for_component): Rename to ...
8195         (build_rdg_partition_for_vertex): ... this and adjust.
8196         (rdg_build_partitions): Take a vector of starting vertices
8197         instead of components.  Remove unnecessary leftover handling.
8198         (ldist_gen): Do not build components or record other stores.
8199         (distribute_loop): Do not distribute loops containing stmts
8200         with side-effects.
8202 2013-09-13  Christian Bruel  <christian.bruel@st.com>
8204         PR target/58314
8205         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
8207 2013-09-13  Kai Tietz  <ktietz@redhat.com>
8209         * config.gcc: Separate cases for mingw and cygwin targets,
8210         and add 64-bit cygwin target case.
8212         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
8213         dll-export inline-functions.
8214         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
8217 2013-09-13  Jeff Law  <law@redhat.com>
8219         PR middle-end/58387
8220         Revert:
8221         2013-09-06  Jeff Law  <law@redhat.com>
8223         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8224         edge implied equivalences into successor phis.
8226 2013-09-12  DJ Delorie  <dj@redhat.com>
8228         * config/rl78/rl78-virt.md: Change from | to \; for asm line
8229         separators.
8231 2013-09-12  Brooks Moses  <bmoses@google.com>
8233         PR driver/42955
8234         * Makefile.in: Do not install driver binaries in $(target)/bin.
8236 2013-09-12  DJ Delorie  <dj@redhat.com>
8238         * config/rl78/rl78.opt (mrelax): New.
8239         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
8240         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
8242         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
8243         between SP and FP.
8244         (rl78_expand_epilogue): Likewise.
8246 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
8248         PR middle-end/58335
8249         * lra-eliminations.c (remove_reg_equal_offset_note): New.
8250         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
8251         pointer elimination with using remove_reg_equal_offset_note.
8253 2013-09-12  DJ Delorie  <dj@redhat.com>
8255         * config/msp430/: New port.
8256         * config.gcc (msp430): Added.
8257         * doc/invoke.texi: Document MSP430 options.
8258         * doc/install.texi: Document msp430-elf
8259         * doc/md.texi: Document msp430-elf
8260         * doc/contrib.texi: Document msp430-elf
8262         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
8263         PSImode.
8265 2013-09-12  Martin Jambor  <mjambor@suse.cz>
8267         PR ipa/58389
8268         * ipa-prop.c (remove_described_reference): Give up if the edge in the
8269         reference descriptor is NULL.
8270         (ipa_edge_removal_hook): If owning a reference descriptor, set its
8271         edge to NULL.
8273 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8275         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
8276         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
8277         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
8278         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
8279         make_temp_ssa_name): move to tree-ssanames.h
8280         * tree-ssa-alias.h: Move prototype.
8281         * tree-ssa.h: Include tree-ssanames.h.
8282         * tree-ssanames.c (FREE_SSANAMES): Move to here.
8283         * tree-ssanames.h: New.  Move items from tree-flow*.h
8284         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
8286 2013-09-12  Richard Biener  <rguenther@suse.de>
8288         PR tree-optimization/58404
8289         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
8290         propagate non-invariant addresses into dereferences wrapped
8291         in component references.
8293 2013-09-12  Richard Biener  <rguenther@suse.de>
8295         PR tree-optimization/58402
8296         * passes.def: Move pass_late_warn_uninitialized later.
8298 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8300         * tree-ssa.h: New.  Move content from tree-flow.h and
8301         tree-flow-inline.h.
8302         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
8303         Move prototypes belonging to tree-ssa.c.
8304         * tree-flow-inline.h (redirect_edge_var_map_def,
8305         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
8306         tree-ssa.h.
8307         * gimple.h: Adjust prototypes.
8308         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
8309         to...
8310         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
8311         * tree.h: Move prototype to tree-ssa.h.
8312         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
8313         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
8314         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
8315         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
8316         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
8317         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
8318         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
8319         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
8320         graphite-blocking.c, graphite-clast-to-gimple.c,
8321         graphite-dependences.c, graphite-interchange.c,
8322         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
8323         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
8324         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
8325         ipa-reference.c, ipa-split.c, ipa-utils.c,
8326         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
8327         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
8328         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
8329         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
8330         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
8331         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
8332         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
8333         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
8334         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
8335         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
8336         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
8337         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
8338         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
8339         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
8340         value-prof.c, var-tracking.c,
8341         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
8343 2013-09-12  Richard Biener  <rguenther@suse.de>
8345         PR tree-optimization/58396
8346         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
8347         (build_rdg): Take a loop-nest parameter, fix memleaks.
8348         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
8350 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8352         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
8353         for SLM.
8355 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
8357         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
8358         The FMA instruction names should have a 'v' prefix.
8360 2013-09-12  Richard Biener  <rguenther@suse.de>
8362         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
8363         (dot_rdg): Use popen instead of system in optional code.
8364         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
8365         (already_processed_vertex_p): Adjust.
8366         (has_anti_or_output_dependence, predecessor_has_mem_write,
8367         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
8368         rdg_flag_uses): Remove.
8369         (rdg_flag_vertex): Simplify.
8370         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
8371         remove recursion.
8372         (build_rdg_partition_for_component): Process the first vertex
8373         of a component only.
8374         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
8376 2013-09-12  Alan Modra  <amodra@gmail.com>
8378         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
8380 2013-09-11  DJ Delorie  <dj@redhat.com>
8381             Nick Clifton  <nickc@redhat.com>
8383         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
8384         (rl78_stack_based_mem): New.
8385         * config/rl78/constraints.md (Iv08): New.
8386         (Iv16): New.
8387         (Iv24): New.
8388         (Is09): New.
8389         (Is17): New.
8390         (Is25): New.
8391         (ISsi): New.
8392         (IShi): New.
8393         (ISqi): New.
8394         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
8395         (movhi): Likewise.
8396         (movsi): Change from expand to insn-and-split.
8397         (ashrsi3): Clobber AX.
8398         (lshrsi3): New.
8399         (ashlsi3): New.
8400         (cbranchsi4): New.
8401         * config/rl78/rl78.md (CC_REG): Fix.
8402         (addsi3): Allow memory and immediate operands.
8403         (addsi3_internal): Split into...
8404         (addsi3_internal_virt): ...new, and ...
8405         (addsi3_internal_real): ...new.
8406         (subsi): New.
8407         (subsi3_internal_virt): New.
8408         (subsi3_internal_real): New.
8409         (mulsi3): Add memory operand.
8410         (mulsi3_rl78): Likewise.
8411         (mulsi3_g13): Likewise.
8412         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
8413         (cbranchqi4_real): Add more constraint options.
8414         (cbranchhi4_real): Expand pattern.
8415         (cbranchhi4_real_signed): New.
8416         (cbranchhi4_real_inverted): New.
8417         (cbranchsi4_real_lt): New.
8418         (cbranchsi4_real_ge): New.
8419         (cbranchsi4_real_signed): New.
8420         (cbranchsi4_real): New.
8421         (peephole2): New.
8422         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
8423         constant shifts.
8424         (lshrsi3_virt): Likewise.
8425         (ashlsi3_virt): Likewise.
8426         (cbranchqi4_virt_signed): New.
8427         (cbranchhi4_virt_signed): New.
8428         (cbranchsi4_virt): New.
8429         * config/rl78/rl78.c: Whitespace fixes throughout.
8430         (move_elim_pass): New.
8431         (pass_data_rl78_move_elim): New.
8432         (pass_rl78_move_elim): New.
8433         (make_pass_rl78_move_elim): New.
8434         (rl78_devirt_info): Run devirt earlier.
8435         (rl78_move_elim_info): New.
8436         (rl78_asm_file_start): Register it.
8437         (rl78_split_movsi): New.
8438         (rl78_as_legitimate_address): Allow virtual base registers when
8439         appropriate.
8440         (rl78_addr_space_convert): Remove spurious debug stuff.
8441         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
8442         (rl78_print_operand): More cases for not printing '#'.
8443         (rl78_expand_compare): Remove most of the logic.
8444         (content_memory): New.
8445         (clear_content_memory): New.
8446         (get_content_index): New.
8447         (get_content_name): New.
8448         (display_content_memory): New.
8449         (update_content): New.
8450         (record_content): New.
8451         (already_contains): New.
8452         (insn_ok_now): Re-recog insns with virtual registers.
8453         (add_postponed_content_update): New.
8454         (process_postponed_content_update): New.
8455         (gen_and_emit_move): New.
8456         (transcode_memory_rtx): Record new location content.
8457         Use gen_and_emit_move.
8458         (force_into_acc): New.
8459         (move_to_acc): Use gen_and_emit_move.
8460         (move_from_acc): Likewise.
8461         (move_acc_to_reg): Likewise.
8462         (move_to_x): Likewise.
8463         (move_to_hl): Likewise.
8464         (move_to_de): Likewise.
8465         (rl78_alloc_physical_registers_op1): Record location content.
8466         (has_constraint): New.
8467         (rl78_alloc_physical_registers_op2): Record location content.
8468         Optimize use of HL.
8469         (rl78_alloc_physical_registers_ro1): Likewise.
8470         (rl78_alloc_physical_registers_cmp): Likewise.
8471         (rl78_alloc_physical_registers_umul): Likewise.
8472         (rl78_alloc_address_registers_macax): New.
8473         (rl78_alloc_physical_registers): Initialize and set location
8474         content memory as needed.
8475         (rl78_reorg): Make sure split2 is called.
8476         (rl78_rtx_costs): New.
8478 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
8480         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
8481         for (not (neg ...)) and (neg (not ...)) cases.
8483 2013-09-11  Richard Biener  <rguenther@suse.de>
8485         PR middle-end/58377
8486         * passes.def: Split critical edges before late uninit warning passes.
8487         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
8489 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
8491         PR tree-optimization/58385
8492         * fold-const.c (build_range_check): If both low and high are NULL,
8493         use omit_one_operand_loc to preserve exp side-effects.
8495 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8497         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
8499 2013-09-11  Richard Biener  <rguenther@suse.de>
8501         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
8502         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
8503         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
8504         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
8505         stmts_from_loop, known_dependences_p, build_empty_rdg,
8506         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
8507         * tree-loop-distribution.c: ... here.
8508         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
8509         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
8510         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
8511         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
8512         * tree-loop-distribution.c: ... here.
8513         * tree-loop-distribution.c: Include gimple-pretty-print.h.
8514         (struct partition_s): Add loops member.
8515         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
8516         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
8517         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
8519 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8520             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8521             Sergey Lega  <sergey.s.lega@intel.com>
8522             Anna Tikhonova  <anna.tikhonova@intel.com>
8523             Ilya Tocar  <ilya.tocar@intel.com>
8524             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8525             Ilya Verbin  <ilya.verbin@intel.com>
8526             Kirill Yukhin  <kirill.yukhin@intel.com>
8527             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8529         * config/i386/constraints.md (k): New.
8530         (Yk): Ditto.
8531         * config/i386/i386.c (const regclass_map): Add new mask registers.
8532         (dbx_register_map): Ditto.
8533         (dbx64_register_map): Ditto.
8534         (svr4_dbx_register_map): Ditto.
8535         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
8536         disabled.
8537         (ix86_preferred_reload_class): Disable constants for mask registers.
8538         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
8539         (ix86_hard_regno_mode_ok): Support new mask registers.
8540         (x86_order_regs_for_local_alloc): Ditto.
8541         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
8542         (FIXED_REGISTERS): Add new mask registers.
8543         (CALL_USED_REGISTERS): Ditto.
8544         (REG_ALLOC_ORDER): Ditto.
8545         (VALID_MASK_REG_MODE): New.
8546         (FIRST_MASK_REG): Ditto.
8547         (LAST_MASK_REG): Ditto.
8548         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
8549         (MAYBE_MASK_CLASS_P): New.
8550         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
8551         (REG_CLASS_CONTENTS): Ditto.
8552         (MASK_REGNO_P): New.
8553         (ANY_MASK_REG_P): Ditto.
8554         (HI_REGISTER_NAMES): Add new mask registers.
8555         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
8556         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
8557         mask registers.
8558         (attribute "type"): Add mskmov, msklog.
8559         (attribute "length_immediate"): Support them.
8560         (attribute "memory"): Ditto.
8561         (attribute "prefix_0f"): Ditto.
8562         (*movhi_internal): Support new mask registers.
8563         (*movqi_internal): Ditto.
8564         (define_split): Split out clobber pattern is a logic
8565         insn on mask registers.
8566         (*k<logic><mode>): New.
8567         (*andhi_1): Extend to support mask regs.
8568         (*andqi_1): Extend to support mask regs.
8569         (kandn<mode>): New.
8570         (define_split): Split and-not to and and not if operands
8571         are not mask regs.
8572         (*<code><mode>_1): Separate HI mode to new pattern...
8573         (*<code>hi_1): This.
8574         (*<code>qi_1): Extend to support mask regs.
8575         (kxnor<mode>): New.
8576         (kortestzhi): Ditto.
8577         (kortestchi): Ditto.
8578         (kunpckhi): Ditto.
8579         (*one_cmpl<mode>2_1): Remove HImode and handle it...
8580         (*one_cmplhi2_1): ...Here, now with mask registers support.
8581         (*one_cmplqi2_1): Support new mask registers.
8582         (HI/QImode arithmetics splitter): Don't split if mask registers
8583         are used.
8584         (HI/QImode not splitter): Ditto.
8585         * config/i386/predicated.md (mask_reg_operand): New.
8586         (general_reg_operand): Ditto.
8588 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8590         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
8591         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
8593 2013-09-10  Jeff Law  <law@redhat.com>
8595         PR tree-optimization/58380
8596         * tree-ssa-threadupdate.c (thread_block): Recognize another case
8597         of threading through a buried loop header.
8599         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
8600         return value for single successor case.
8602 2013-09-10  Jan Hubicka  <jh@suse.cz>
8604         * ipa-devirt.c (ipa_devirt): Enable with LTO.
8606 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
8608         PR target/58361
8609         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
8610         support conditional execution.
8611         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
8613 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
8615         * lra.c (lra): Clear lra_optional_reload_pseudos before every
8616         constraint pass.
8617         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
8618         Check destination too to check move insn.
8619         (undo_optional_reloads): Add check that the original peudo did not
8620         changed its allocation and the optional reload was inherited on last
8621         inheritance pass.  Break loop after deciding to keep optional reload.
8622         (lra_undo_inheritance): Add check that inherited pseudo still in
8623         memory.
8625 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
8627         * config/aarch64/aarch64.md (generic_sched): New.
8628         * config/aarch64/aarch64-generic.md (load): Make conditional
8629         on generic_sched attribute.
8630         (nonload): Likewise.
8632 2013-09-10  Jan Hubicka  <jh@suse.cz>
8634         * lto-cgraph.c: Include ipa-utils.h.
8635         (compute_ltrans_boundary): Also add possible targets into the boundary.
8637 2013-09-10  Jan Hubicka  <jh@suse.cz>
8639         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
8640         VAR_DECL of vtable rather than full expression.
8642 2013-09-10  Jan Hubicka  <jh@suse.cz>
8643             Paolo Carlini  <paolo.carlini@oracle.com>
8645         * cgraphunit.c (analyze_functions): Save input_location, set it
8646         to UNKNOWN_LOCATION and restore it at the end.
8648 2013-09-10  Martin Jambor  <mjambor@suse.cz>
8650         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
8651         represented by a thunk.
8653 2013-09-10  Jeff Law  <law@redhat.com>
8655         PR tree-optimization/58343
8656         * tree-ssa-threadupdate.c (thread_block): Identify and disable
8657         jump threading requests through loop headers buried in the middle
8658         of a jump threading path.
8660         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
8661         in return value/type.
8663 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
8665         PR rtl-optimization/58365
8666         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
8667         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
8668         it differs.
8670 2013-09-10  Richard Biener  <rguenther@suse.de>
8672         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
8673         * tree-data-ref.c (create_rdg_vertices): Collect all data
8674         references, signal failure to the caller, use data-ref API.
8675         (build_rdg): Compute data references only once.  Maintain lifetime
8676         of data references and data dependences from within RDG.
8677         (free_rdg): Free dependence relations.
8678         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
8679         inventing extra dependences.
8680         (distribute_loop): Update for RDG API changes.
8682 2013-09-10  Kai Tietz  <ktietz@redhat.com>
8684         * doc/invoke.texi (fms-extensions): Document changed
8685         behavior for ms-abi targets.
8686         * config/i386/i386.c (ix86_option_override_internal):
8687         Set default value of option -fms-extension for ms-abi targets.
8689 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8691         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
8693 2013-09-10  Alan Modra  <amodra@gmail.com>
8695         PR target/58330
8696         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
8698 2013-09-10  Alan Modra  <amodra@gmail.com>
8700         * config/rs6000/predicates.md (add_cint_operand): New.
8701         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
8702         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
8703         using add_cint_operand.
8704         (largetoc_high_plus_aix): Likewise.
8706 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
8708         PR tree-optimization/58364
8709         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
8710         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
8711         the current range can't be an unconditional true or false.
8713 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
8715         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
8717 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
8719         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
8721 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
8723         PR c++/43452
8724         * doc/invoke.texi (-Wdelete-incomplete): Document it.
8726 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
8728         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
8729         NO_REGS for immediate that can't be moved directly into FP_REGS.
8731 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8733         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
8734         comparison with negated operand.
8735         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
8736         RTL form.
8738 2013-09-09  Richard Biener  <rguenther@suse.de>
8740         PR middle-end/58326
8741         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
8742         of a subloop record all its block as affecting loop-closed SSA form.
8744 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8746         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
8747         of an rtx/bitpos pair.
8748         (store_fixed_bit_field): Update accordingly.
8750 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8752         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
8753         GEN_INT.
8754         * builtins.c (expand_errno_check): Likewise.
8755         * dwarf2cfi.c (init_return_column_size): Likewise.
8756         * except.c (sjlj_mark_call_sites): Likewise.
8757         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
8758         * lra-constraints.c (emit_inc): Likewise.
8759         * ree.c (combine_set_extension): Likewise.
8760         * regmove.c (fixup_match_2): Likewise.
8761         * reload1.c (inc_for_reload): Likewise.
8763 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8765         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
8766         (force_to_mode, simplify_shift_const_1, simplify_comparison):
8767         Use gen_int_mode with the mode of the associated simplify_* call.
8768         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
8769         * expmed.c (expand_shift_1): Likewise.
8770         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8771         * loop-iv.c (iv_number_of_iterations): Likewise.
8772         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8773         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
8775 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8777         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
8778         of the associated expand_* call.
8779         (asan_emit_stack_protection): Likewise.
8780         * builtins.c (round_trampoline_addr): Likewise.
8781         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
8782         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
8783         (emit_store_flag): Likewise.
8784         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
8785         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
8786         Likewise.
8787         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8788         * ifcvt.c (noce_try_store_flag_constants): Likewise.
8789         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8790         * modulo-sched.c (generate_prolog_epilog): Likewise.
8791         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
8792         (expand_ctz, expand_ffs, expand_unop): Likewise.
8794 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8796         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
8797         of the associated gen_rtx_* call.
8798         * caller-save.c (init_caller_save): Likewise.
8799         * combine.c (find_split_point, make_extraction): Likewise.
8800         (make_compound_operation): Likewise.
8801         * dwarf2out.c (mem_loc_descriptor): Likewise.
8802         * explow.c (plus_constant, probe_stack_range): Likewise.
8803         * expmed.c (expand_mult_const): Likewise.
8804         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
8805         * reload1.c (init_reload): Likewise.
8806         * valtrack.c (cleanup_auto_inc_dec): Likewise.
8807         * var-tracking.c (adjust_mems): Likewise.
8808         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
8809         rather than gen_rtx_fmt_ee.
8811 2013-09-09  Jan Hubicka  <jh@suse.cz>
8813         PR middle-end/58294
8814         * value-prof.c (gimple_ic): Copy also abnormal edges.
8816 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8818         * asan.c (asan_shadow_cst): Use gen_int_mode.
8820 2013-09-08  Jan Hubicka  <jh@suse.cz>
8822         * ipa-profile.c: Add toplevel comment.
8823         (ipa_propagate_frequency_1): Be more conservative when profile is read.
8824         (contains_hot_call_p): New function.
8825         (ipa_propagate_frequency): Set frequencies based on counts when
8826         profile is read.
8827         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
8828         profile; do not tamper with profile after inlining if it is read.
8830 2013-09-08  Jan Hubicka  <jh@suse.cz>
8832         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
8833         speculative edges.
8835 2013-09-08  Jan Hubicka  <jh@suse.cz>
8837         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
8838         summary generation.
8840 2013-09-08  Jeff Law  <law@redhat.com>
8842         PR bootstrap/58340
8843         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
8844         of 'found'.
8846 2013-09-08  Andi Kleen  <ak@linux.intel.com>
8848         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
8850 2013-09-08  Jan Hubicka  <jh@suse.cz>
8852         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
8854 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
8856         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
8857         for non-debug insns.
8858         * lra.c (new_insn_reg): Take the containing insn as a parameter.
8859         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
8860         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
8861         accordingly.
8863 2013-09-08  Jan Hubicka  <jh@suse.cz>
8865         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
8866         targets and devirtualize to BUILT_IN_UNREACHABLE.
8867         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
8868         * ipa.c (walk_polymorphic_call_targets): New function.
8869         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
8870         functions; use the new timevar.
8871         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
8872         was removed from the program.
8873         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
8874         not consider it in the walk when its vtable is dead.
8875         (possible_polymorphic_call_targets_1): Pass anonymous flag to
8876         record_binfo.
8877         (devirt_variable_node_removal_hook): New function.
8878         (possible_polymorphic_call_targets): Also register
8879         devirt_variable_node_removal_hook.
8880         (ipa_devirt): Do not do non-speculative devirtualization.
8881         (gate_ipa_devirt): One execute if devirtualizing speculatively.
8883 2013-09-08  Jan Hubicka  <jh@suse.cz>
8885         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
8886         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
8887         varpool_remove_variable_insertion_hook): Declare.
8888         * varpool.c (varpool_node_hook_list): New structure.
8889         (first_varpool_node_removal_hook,
8890         first_varpool_variable_insertion_hook): New variables.
8891         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
8892         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
8893         varpool_remove_variable_insertion_hook,
8894         varpool_call_variable_insertion_hooks): New functions.
8895         (varpool_remove_node): Use it.
8897 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
8899         PR c++/54941
8900         * diagnostic.c (diagnostic_build_prefix): When s.file is
8901         "<built-in>" don't output line and column numbers.
8903 2013-09-06  Jan Hubicka  <jh@suse.cz>
8905         * cgraphunit.c (expand_thunk): Get body before touching arguments.
8906         * lto-streamer-out.c: Stream thunks, too.
8907         * lto-streamer-in.c (input_function): Pop cfun here
8908         (lto_read_body): Instead of here.
8910 2013-09-06  Caroline Tice  <cmtice@google.com>
8912         * doc/install.texi: Add documentation for the --enable-vtable-verify
8913         and the --disable-libvtv configure options.
8915 2013-09-06  Jeff Law  <law@redhat.com>
8917         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8918         edge implied equivalences into successor phis.
8920 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
8922         * resource.c (mark_referenced_resources): Handle COND_EXEC.
8924 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
8926         * resource.c (mark_target_live_regs): Compute resources taking
8927         into account if a call is predicated or not.
8929 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
8931         * toplev.c (output_stack_usage): Be prepared for suffixes created by
8932         the compiler in the function names.
8934 2013-09-06  Jan Hubicka  <jh@suse.cz>
8936         PR middle-end/58094
8937         * ipa-inline.c (has_caller_p): New function.
8938         (want_inline_function_to_all_callers_p): Use it.
8939         (sum_callers, inline_to_all_callers): Break out from ...
8940         (ipa_inline): ... here.
8942 2013-09-06  Jan Hubicka  <jh@suse.cz>
8944         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
8945         only when AVX is enabled.
8947 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8949         * config/aarch64/aarch64.md
8950         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
8951         is fpsimd_<load/store>2.
8952         (load_pair<mode>): Likewise.
8953         (store_pair<mode>): Likewise.
8955 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8957         * config/arm/types.md (type): Add "mrs" type.
8958         * config/aarch64/aarch64.md
8959         (aarch64_load_tp_hard): Make type "mrs".
8960         * config/arm/arm.md
8961         (load_tp_hard): Make type "mrs".
8962         * config/arm/cortex-a15.md: Update with new attributes.
8963         * config/arm/cortex-a5.md: Update with new attributes.
8964         * config/arm/cortex-a53.md: Update with new attributes.
8965         * config/arm/cortex-a7.md: Update with new attributes.
8966         * config/arm/cortex-a8.md: Update with new attributes.
8967         * config/arm/cortex-a9.md: Update with new attributes.
8968         * config/arm/cortex-m4.md: Update with new attributes.
8969         * config/arm/cortex-r4.md: Update with new attributes.
8970         * config/arm/fa526.md: Update with new attributes.
8971         * config/arm/fa606te.md: Update with new attributes.
8972         * config/arm/fa626te.md: Update with new attributes.
8973         * config/arm/fa726te.md: Update with new attributes.
8975 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8977         * config/aarch64/aarch64.md
8978         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
8979         (*movtf_aarch64): Likewise.
8980         * config/arm/arm.md
8981         (thumb1_movdi_insn): Use "multiple" for type where more than one
8982         instruction is used for a move.
8983         (*arm32_movhf): Likewise.
8984         (*thumb_movdf_insn): Likewise.
8986 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
8988         * config/arm/types.md (type): Rename fcpys to fmov.
8989         * config/arm/vfp.md
8990         (*arm_movsi_vfp): Rename type fcpys as fmov.
8991         (*thumb2_movsi_vfp): Likewise
8992         (*movhf_vfp_neon): Likewise
8993         (*movhf_vfp): Likewise
8994         (*movsf_vfp): Likewise
8995         (*thumb2_movsf_vfp): Likewise
8996         (*movsfcc_vfp): Likewise
8997         (*thumb2_movsfcc_vfp): Likewise
8998         * config/aarch64/aarch64-simd.md
8999         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
9000         * config/aarch64/aarch64.md
9001         (*movsi_aarch64): Replace type mov_reg with fmovs.
9002         (*movdi_aarch64): Likewise
9003         (*movsf_aarch64): Likewise
9004         (*movdf_aarch64): Likewise
9005         * config/arm/arm.c
9006         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
9007         * config/arm/iwmmxt.md
9008         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
9009         * config/arm/arm1020e.md: Update with new attributes.
9010         * config/arm/cortex-a15-neon.md: Update with new attributes.
9011         * config/arm/cortex-a5.md: Update with new attributes.
9012         * config/arm/cortex-a53.md: Update with new attributes.
9013         * config/arm/cortex-a7.md: Update with new attributes.
9014         * config/arm/cortex-a8-neon.md: Update with new attributes.
9015         * config/arm/cortex-a9.md: Update with new attributes.
9016         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9017         * config/arm/cortex-r4f.md: Update with new attributes.
9018         * config/arm/marvell-pj4.md: Update with new attributes.
9019         * config/arm/vfp11.md: Update with new attributes.
9021 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9023         * config/aarch64/aarch64.md
9024         (*madd<mode>): Fix type attribute.
9025         (*maddsi_uxtw): Likewise.
9026         (*msub<mode>): Likewise.
9027         (*msubsi_uxtw): Likewise.
9028         (<su_optab>maddsidi4): Likewise.
9029         (<su_optab>msubsidi4): Likewise.
9031 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9033         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
9034         * config/arm/arm.md (core_cycles): Remove fdiv.
9035         * config/arm/vfp.md:
9036         (*sqrtsf2_vfp): Update for attribute changes.
9037         (*sqrtdf2_vfp): Likewise.
9038         * config/aarch64/aarch64.md:
9039         (sqrt<mode>2): Update for attribute changes.
9040         * config/arm/arm1020e.md: Update with new attributes.
9041         * config/arm/cortex-a15-neon.md: Update with new attributes.
9042         * config/arm/cortex-a5.md: Update with new attributes.
9043         * config/arm/cortex-a53.md: Update with new attributes.
9044         * config/arm/cortex-a7.md: Update with new attributes.
9045         * config/arm/cortex-a8-neon.md: Update with new attributes.
9046         * config/arm/cortex-a9.md: Update with new attributes.
9047         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9048         * config/arm/cortex-r4f.md: Update with new attributes.
9049         * config/arm/marvell-pj4.md: Update with new attributes.
9050         * config/arm/vfp11.md: Update with new attributes.
9052 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9054         * config/arm/types.md
9055         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
9056         * config/aarch64/aarch64.md
9057         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
9058         new attributes.
9059         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
9060         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
9061         (float<GPI:mode><GPF:mode>2): Likewise.
9062         * config/arm/vfp.md
9063         (*truncsisf2_vfp): Update with new attributes.
9064         (*truncsidf2_vfp): Likewise.
9065         (fixuns_truncsfsi2): Likewise.
9066         (fixuns_truncdfsi2): Likewise.
9067         (*floatsisf2_vfp): Likewise.
9068         (*floatsidf2_vfp): Likewise.
9069         (floatunssisf2): Likewise.
9070         (floatunssidf2): Likewise.
9071         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
9072         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
9073         * config/arm/arm1020e.md: Update with new attributes.
9074         * config/arm/cortex-a15-neon.md: Update with new attributes.
9075         * config/arm/cortex-a5.md: Update with new attributes.
9076         * config/arm/cortex-a53.md: Update with new attributes.
9077         * config/arm/cortex-a7.md: Update with new attributes.
9078         * config/arm/cortex-a8-neon.md: Update with new attributes.
9079         * config/arm/cortex-a9.md: Update with new attributes.
9080         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9081         * config/arm/cortex-r4f.md: Update with new attributes.
9082         * config/arm/marvell-pj4.md: Update with new attributes.
9083         * config/arm/vfp11.md: Update with new attributes.
9085 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9087         * config/aarch64/arm_neon.h
9088         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
9089         (vqtbx<1,2,3,4><q>_s8): Likewise.
9091 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9093         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
9094         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
9095         (add<mode>3): Add type attribute.
9096         (add<mode>3): Likewise.
9097         (usadd<mode>3): Likewise.
9098         (ssadd<mode>3): Likewise.
9099         (sub<mode>3): Likewise.
9100         (sub<mode>3): Likewise.
9101         (ussub<mode>3): Likewise.
9102         (sssub<mode>3): Likewise.
9103         (ssmulsa3): Likewise.
9104         (usmulusa3): Likewise.
9105         (arm_usatsihi): Likewise.
9106         * config/arm/vfp.md
9107         (*movdi_vfp): Add types for all instructions.
9108         (*movdi_vfp_cortexa8): Likewise.
9109         (*movhf_vfp_neon): Likewise.
9110         (*movhf_vfp): Likewise.
9111         (*movdf_vfp): Likewise.
9112         (*thumb2_movdf_vfp): Likewise.
9113         (*thumb2_movdfcc_vfp): Likewise.
9114         * config/arm/arm.md: Add type attribute to all insn patterns.
9115         (*thumb1_adddi3): Add type attribute.
9116         (*arm_adddi3): Likewise.
9117         (*adddi_sesidi_di): Likewise.
9118         (*adddi_zesidi_di): Likewise.
9119         (*thumb1_addsi3): Likewise.
9120         (addsi3_compare0): Likewise.
9121         (*addsi3_compare0_scratch): Likewise.
9122         (*compare_negsi_si): Likewise.
9123         (cmpsi2_addneg): Likewise.
9124         (*addsi3_carryin_<optab>): Likewise.
9125         (*addsi3_carryin_alt2_<optab>): Likewise.
9126         (*addsi3_carryin_clobercc_<optab>): Likewise.
9127         (*subsi3_carryin): Likewise.
9128         (*subsi3_carryin_const): Likewise.
9129         (*subsi3_carryin_compare): Likewise.
9130         (*subsi3_carryin_compare_const): Likewise.
9131         (*arm_subdi3): Likewise.
9132         (*thumb_subdi3): Likewise.
9133         (*subdi_di_zesidi): Likewise.
9134         (*subdi_di_sesidi): Likewise.
9135         (*subdi_zesidi_di): Likewise.
9136         (*subdi_sesidi_di): Likewise.
9137         (*subdi_zesidi_ze): Likewise.
9138         (thumb1_subsi3_insn): Likewise.
9139         (*arm_subsi3_insn): Likewise.
9140         (*anddi3_insn): Likewise.
9141         (*anddi_zesidi_di): Likewise.
9142         (*anddi_sesdi_di): Likewise.
9143         (*ne_zeroextracts): Likewise.
9144         (*ne_zeroextracts): Likewise.
9145         (*ite_ne_zeroextr): Likewise.
9146         (*ite_ne_zeroextr): Likewise.
9147         (*anddi_notdi_di): Likewise.
9148         (*anddi_notzesidi): Likewise.
9149         (*anddi_notsesidi): Likewise.
9150         (andsi_notsi_si): Likewise.
9151         (thumb1_bicsi3): Likewise.
9152         (*iordi3_insn): Likewise.
9153         (*iordi_zesidi_di): Likewise.
9154         (*iordi_sesidi_di): Likewise.
9155         (*thumb1_iorsi3_insn): Likewise.
9156         (*xordi3_insn): Likewise.
9157         (*xordi_zesidi_di): Likewise.
9158         (*xordi_sesidi_di): Likewise.
9159         (*arm_xorsi3): Likewise.
9160         (*andsi_iorsi3_no): Likewise.
9161         (*smax_0): Likewise.
9162         (*smax_m1): Likewise.
9163         (*arm_smax_insn): Likewise.
9164         (*smin_0): Likewise.
9165         (*arm_smin_insn): Likewise.
9166         (*arm_umaxsi3): Likewise.
9167         (*arm_uminsi3): Likewise.
9168         (*minmax_arithsi): Likewise.
9169         (*minmax_arithsi_): Likewise.
9170         (*satsi_<SAT:code>): Likewise.
9171         (arm_ashldi3_1bit): Likewise.
9172         (arm_ashrdi3_1bit): Likewise.
9173         (arm_lshrdi3_1bit): Likewise.
9174         (*arm_negdi2): Likewise.
9175         (*thumb1_negdi2): Likewise.
9176         (*arm_negsi2): Likewise.
9177         (*thumb1_negsi2): Likewise.
9178         (*negdi_extendsid): Likewise.
9179         (*negdi_zero_extend): Likewise.
9180         (*arm_abssi2): Likewise.
9181         (*thumb1_abssi2): Likewise.
9182         (*arm_neg_abssi2): Likewise.
9183         (*thumb1_neg_abss): Likewise.
9184         (one_cmpldi2): Likewise.
9185         (extend<mode>di2): Likewise.
9186         (*compareqi_eq0): Likewise.
9187         (*arm_extendhisi2addsi): Likewise.
9188         (*arm_movdi): Likewise.
9189         (*thumb1_movdi_insn): Likewise.
9190         (*arm_movt): Likewise.
9191         (*thumb1_movsi_insn): Likewise.
9192         (pic_add_dot_plus_four): Likewise.
9193         (pic_add_dot_plus_eight): Likewise.
9194         (tls_load_dot_plus_eight): Likewise.
9195         (*thumb1_movhi_insn): Likewise.
9196         (*thumb1_movsf_insn): Likewise.
9197         (*movdf_soft_insn): Likewise.
9198         (*thumb_movdf_insn): Likewise.
9199         (cbranchsi4_insn): Likewise.
9200         (cbranchsi4_scratch): Likewise.
9201         (*negated_cbranchsi4): Likewise.
9202         (*tbit_cbranch): Likewise.
9203         (*tlobits_cbranch): Likewise.
9204         (*tstsi3_cbranch): Likewise.
9205         (*cbranchne_decr1): Likewise.
9206         (*addsi3_cbranch): Likewise.
9207         (*addsi3_cbranch_scratch): Likewise.
9208         (*arm_cmpdi_insn): Likewise.
9209         (*arm_cmpdi_unsig): Likewise.
9210         (*arm_cmpdi_zero): Likewise.
9211         (*thumb_cmpdi_zero): Likewise.
9212         (*deleted_compare): Likewise.
9213         (*mov_scc): Likewise.
9214         (*mov_negscc): Likewise.
9215         (*mov_notscc): Likewise.
9216         (*cstoresi_eq0_thumb1_insn): Likewise.
9217         (cstoresi_nltu_thumb1): Likewise.
9218         (cstoresi_ltu_thu): Likewise.
9219         (thumb1_addsi3_addgeu): Likewise.
9220         (*arm_jump): Likewise.
9221         (*thumb_jump): Likewise.
9222         (*check_arch2): Likewise.
9223         (arm_casesi_internal): Likewise.
9224         (thumb1_casesi_dispatch): Likewise.
9225         (*arm_indirect_jump): Likewise.
9226         (*thumb1_indirect_jump): Likewise.
9227         (nop): Likewise.
9228         (*and_scc): Likewise.
9229         (*ior_scc): Likewise.
9230         (*compare_scc): Likewise.
9231         (*cond_move): Likewise.
9232         (*cond_arith): Likewise.
9233         (*cond_sub): Likewise.
9234         (*cmp_ite0): Likewise.
9235         (*cmp_ite1): Likewise.
9236         (*cmp_and): Likewise.
9237         (*cmp_ior): Likewise.
9238         (*ior_scc_scc): Likewise.
9239         (*ior_scc_scc_cmp): Likewise.
9240         (*and_scc_scc): Likewise.
9241         (*and_scc_scc_cmp): Likewise.
9242         (*and_scc_scc_nod): Likewise.
9243         (*negscc): Likewise.
9244         (movcond_addsi): Likewise.
9245         (movcond): Likewise.
9246         (*ifcompare_plus_move): Likewise.
9247         (*if_plus_move): Likewise.
9248         (*ifcompare_move_plus): Likewise.
9249         (*if_move_plus): Likewise.
9250         (*ifcompare_arith_arith): Likewise.
9251         (*if_arith_arith): Likewise.
9252         (*ifcompare_arith_move): Likewise.
9253         (*if_arith_move): Likewise.
9254         (*ifcompare_move_arith): Likewise.
9255         (*if_move_arith): Likewise.
9256         (*ifcompare_move_not): Likewise.
9257         (*if_move_not): Likewise.
9258         (*ifcompare_not_move): Likewise.
9259         (*if_not_move): Likewise.
9260         (*ifcompare_shift_move): Likewise.
9261         (*if_shift_move): Likewise.
9262         (*ifcompare_move_shift): Likewise.
9263         (*if_move_shift): Likewise.
9264         (*ifcompare_shift_shift): Likewise.
9265         (*ifcompare_not_arith): Likewise.
9266         (*ifcompare_arith_not): Likewise.
9267         (*if_arith_not): Likewise.
9268         (*ifcompare_neg_move): Likewise.
9269         (*if_neg_move): Likewise.
9270         (*ifcompare_move_neg): Likewise.
9271         (*if_move_neg): Likewise.
9272         (prologue_thumb1_interwork): Likewise.
9273         (*cond_move_not): Likewise.
9274         (*sign_extract_onebit): Likewise.
9275         (*not_signextract_onebit): Likewise.
9276         (stack_tie): Likewise.
9277         (align_4): Likewise.
9278         (align_8): Likewise.
9279         (consttable_end): Likewise.
9280         (consttable_1): Likewise.
9281         (consttable_2): Likewise.
9282         (consttable_4): Likewise.
9283         (consttable_8): Likewise.
9284         (consttable_16): Likewise.
9285         (*thumb1_tablejump): Likewise.
9286         (prefetch): Likewise.
9287         (force_register_use): Likewise.
9288         (thumb_eh_return): Likewise.
9289         (load_tp_hard): Likewise.
9290         (load_tp_soft): Likewise.
9291         (tlscall): Likewise.
9292         (*arm_movtas_ze): Likewise.
9293         (*arm_rev): Likewise.
9294         (*arm_revsh): Likewise.
9295         (*arm_rev16): Likewise.
9296         * config/arm/thumb2.md
9297         (*thumb2_smaxsi3): Likewise.
9298         (*thumb2_sminsi3): Likewise.
9299         (*thumb32_umaxsi3): Likewise.
9300         (*thumb2_uminsi3): Likewise.
9301         (*thumb2_negdi2): Likewise.
9302         (*thumb2_abssi2): Likewise.
9303         (*thumb2_neg_abss): Likewise.
9304         (*thumb2_movsi_insn): Likewise.
9305         (tls_load_dot_plus_four): Likewise.
9306         (*thumb2_movhi_insn): Likewise.
9307         (*thumb2_mov_scc): Likewise.
9308         (*thumb2_mov_negs): Likewise.
9309         (*thumb2_mov_negs): Likewise.
9310         (*thumb2_mov_nots): Likewise.
9311         (*thumb2_mov_nots): Likewise.
9312         (*thumb2_movsicc_): Likewise.
9313         (*thumb2_movsfcc_soft_insn): Likewise.
9314         (*thumb2_indirect_jump): Likewise.
9315         (*thumb2_and_scc): Likewise.
9316         (*thumb2_ior_scc): Likewise.
9317         (*thumb2_ior_scc_strict_it): Likewise.
9318         (*thumb2_cond_move): Likewise.
9319         (*thumb2_cond_arith): Likewise.
9320         (*thumb2_cond_ari): Likewise.
9321         (*thumb2_cond_sub): Likewise.
9322         (*thumb2_negscc): Likewise.
9323         (*thumb2_movcond): Likewise.
9324         (thumb2_casesi_internal): Likewise.
9325         (thumb2_casesi_internal_pic): Likewise.
9326         (*thumb2_alusi3_short): Likewise.
9327         (*thumb2_mov<mode>_shortim): Likewise.
9328         (*thumb2_addsi_short): Likewise.
9329         (*thumb2_subsi_short): Likewise.
9330         (thumb2_addsi3_compare0): Likewise.
9331         (*thumb2_cbz): Likewise.
9332         (*thumb2_cbnz): Likewise.
9333         (*thumb2_one_cmplsi2_short): Likewise.
9334         (*thumb2_negsi2_short): Likewise.
9335         (*orsi_notsi_si): Likewise.
9336         * config/arm/arm1020e.md: Update with new attributes.
9337         * config/arm/arm1026ejs.md: Update with new attributes.
9338         * config/arm/arm1136jfs.md: Update with new attributes.
9339         * config/arm/arm926ejs.md: Update with new attributes.
9340         * config/arm/cortex-a15.md: Update with new attributes.
9341         * config/arm/cortex-a5.md: Update with new attributes.
9342         * config/arm/cortex-a53.md: Update with new attributes.
9343         * config/arm/cortex-a7.md: Update with new attributes.
9344         * config/arm/cortex-a8.md: Update with new attributes.
9345         * config/arm/cortex-a9.md: Update with new attributes.
9346         * config/arm/cortex-m4.md: Update with new attributes.
9347         * config/arm/cortex-r4.md: Update with new attributes.
9348         * config/arm/fa526.md: Update with new attributes.
9349         * config/arm/fa606te.md: Update with new attributes.
9350         * config/arm/fa626te.md: Update with new attributes.
9351         * config/arm/fa726te.md: Update with new attributes.
9353 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9355         * config/aarch64/aarch64-simd.md
9356         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
9357         <vwx> iterator to ensure correct register choice.
9358         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
9359         (aarch64_sqdmull_n<mode>): Likewise.
9360         (aarch64_sqdmull2_n<mode>_internal): Likewise.
9361         * config/aarch64/arm_neon.h
9362         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
9363         (vml<as><q>_n_<su>16): Likewise.
9364         (vml<as>l_high_lane<q>_<su>16): Likewise.
9365         (vml<as>l_high_n_<su>16): Likewise.
9366         (vml<as>l_lane<q>_<su>16): Likewise.
9367         (vml<as>l_n_<su>16): Likewise.
9368         (vmul<q>_lane<q>_<su>16): Likewise.
9369         (vmul<q>_n_<su>16): Likewise.
9370         (vmull_lane<q>_<su>16): Likewise.
9371         (vmull_n_<su>16): Likewise.
9372         (vmull_high_lane<q>_<su>16): Likewise.
9373         (vmull_high_n_<su>16): Likewise.
9374         (vqrdmulh<q>_n_s16): Likewise.
9376 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
9378         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
9379         have the correct lane parameter.
9381 2013-09-06  Richard Biener <rguenther@suse.de>
9383         * cfganal.c (control_dependences::~control_dependences):
9384         Properly free all of the vector.
9386 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
9388         PR target/58269
9389         * config/i386/i386.c (ix86_conditional_register_usage):
9390         Proper initialize extended SSE registers.
9392 2013-09-06  Jan Hubicka  <jh@suse.cz>
9394         PR tree-optimization/58311
9395         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
9397 2013-09-06  Jan Hubicka  <jh@suse.cz>
9399         * Makefile.in (tree-sra.o): Update dependencies.
9400         * tree-sra.c: Include ipa-utils.h
9401         (scan_function): Use recursive_call_p.
9402         (has_caller_p): New function.
9403         (cgraph_for_node_and_aliases): Count also callers of aliases.
9405 2013-09-06  Jan Hubicka  <jh@suse.cz>
9407         PR middle-end/58094
9408         * cgraph.h (symtab_semantically_equivalent_p): Declare.
9409         * tree-tailcall.c: Include ipa-utils.h.
9410         (find_tail_calls): Use it.
9411         * ipa-pure-const.c (check_call): Likewise.
9412         * ipa-utils.c (recursive_call_p): New function.
9413         * ipa-utils.h (recursive_call_p): Dclare.
9414         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
9415         (symtab_semantically_equivalent_p): New function.
9416         * Makefile.in (tree-tailcall.o): Update dependencies.
9418 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9420         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
9421         non-inlinable part.
9423 2013-09-06  Richard Biener  <rguenther@suse.de>
9425         * lto-streamer.h (lto_global_var_decls): Remove.
9426         * Makefile.in (OBJS): Remove lto-symtab.o.
9427         (lto-symtab.o): Remove.
9428         (GTFILES): Remove lto-symtab.c
9429         * lto-symtab.c: Move to lto/
9431 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9433         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
9434         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
9435         (UNSPEC_FPINT_RINT): New constant definitions.
9436         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
9437         definition with 2 attributes.
9438         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
9439         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
9440         definitions.
9442 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9444         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
9445         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
9446         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
9447         attribute to "z196_alone".
9448         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
9449         "zEC12_simple".
9451 2013-09-06  Richard Biener  <rguenther@suse.de>
9453         * basic-block.h (class control_dependences): New.
9454         * tree-ssa-dce.c (control_dependence_map): Remove.
9455         (cd): New global.
9456         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
9457         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
9458         find_pdom, find_control_dependence, find_all_control_dependences):
9459         Move to cfganal.c.
9460         (mark_control_dependent_edges_necessary,
9461         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
9462         tree_dce_done, perform_tree_ssa_dce): Adjust.
9463         * cfganal.c (set_control_dependence_map_bit,
9464         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
9465         find_all_control_dependences): Move from tree-ssa-dce.c and
9466         implement as methods of control_dependences class.
9467         (control_dependences::control_dependences): New.
9468         (control_dependences::~control_dependences): Likewise.
9469         (control_dependences::get_edges_dependent_on): Likewise.
9470         (control_dependences::get_edge): Likewise.
9472 2013-09-04  Jan Hubicka  <jh@suse.cz>
9474         * tree.c (types_same_for_odr): Drop overactive check.
9475         * ipa-devirt.c (hash_type_name): Likewise.
9477 2013-09-04  Jan Hubicka  <jh@suse.cz>
9479         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
9480         (analyze_functions): ... here.
9482 2013-09-04  Jan Hubicka  <jh@suse.cz>
9484         PR middle-end/58201
9485         * cgraphunit.c (analyze_functions): Clear AUX fields
9486         after processing; initialize assembler name has.
9488 2013-09-05  Jeff Law  <law@redhat.com>
9490         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
9491         from thread_around_empty_block.  Record threading path into PATH.
9492         Recurse if threading through the initial block is successful.
9493         (thread_across_edge): Corresponding changes to slightly simplify.
9495 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9497         * config/aarch64/aarch64.md
9498         (type): Remove frecpe, frecps, frecpx.
9499         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
9500         fix to be a TARGET_SIMD instruction.
9501         (aarch64_frecps): Remove.
9502         * config/aarch64/aarch64-simd.md
9503         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
9504         (aarch64_frecps<mode>): Handle all float/vector of float modes.
9506 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9507             Sofiane Naci  <sofiane.naci@arm.com>
9509         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
9510         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
9511         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
9512         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
9513         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
9514         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
9515         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
9516         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
9517         into "clz, "rbit".  Rename "shift" to "shift_imm".
9518         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
9519         changes.  Update for attribute changes all occurrences of arlo_* and
9520         shift* types.
9521         * config/arm/arm-fixed.md: Update for attribute changes
9522         all occurrences of arlo_* types.
9523         * config/arm/thumb2.md: Update for attribute changes all occurrences
9524         of arlo_* types.
9525         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
9526         (cortexa7_older_only): Likewise.
9527         (cortexa7_younger):  Likewise.
9528         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
9529         (1020alu_shift_op): Likewise.
9530         (1020alu_shift_reg_op): Likewise.
9531         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
9532         (alu_shift_op): Likewise.
9533         (alu_shift_reg_op): Likewise.
9534         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
9535         (11_alu_shift_op): Likewise.
9536         (11_alu_shift_reg_op): Likewise.
9537         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
9538         (9_alu_shift_reg_op): Likewise.
9539         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
9540         attribute changes.
9541         (cortex_a15_alu_shift): Likewise.
9542         (cortex_a15_alu_shift_reg): Likewise.
9543         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
9544         attribute changes.
9545         (cortex_a5_alu_shift): Likewise.
9546         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
9547         attribute changes.
9548         (cortex_a53_alu_shift): Likewise.
9549         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
9550         attribute changes.
9551         (cortex_a7_alu_reg): Likewise.
9552         (cortex_a7_alu_shift): Likewise.
9553         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
9554         attribute changes.
9555         (cortex_a8_alu_shift): Likewise.
9556         (cortex_a8_alu_shift_reg): Likewise.
9557         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
9558         (cortex_a9_dp_shift): Likewise.
9559         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
9560         attribute changes.
9561         * config/arm/cortex-r4.md
9562         (cortex_r4_alu): Update for attribute changes.
9563         (cortex_r4_mov): Likewise.
9564         (cortex_r4_alu_shift_reg): Likewise.
9565         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
9566         (526_alu_shift_op): Likewise.
9567         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
9568         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
9569         (626te_alu_shift_op): Likewise.
9570         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
9571         (726te_alu_shift_op): Likewise.
9572         (726te_alu_shift_reg_op): Likewise.
9573         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
9574         (mp626_alu_shift_op): Likewise.
9575         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
9576         (pj4_alu_conds): Likewise.
9577         (pj4_shift): Likewise.
9578         (pj4_shift_conds): Likewise.
9579         (pj4_alu_shift): Likewise.
9580         (pj4_alu_shift_conds): Likewise.
9581         * config/aarch64/aarch64.md: Update for attribute change
9582         all occurrences of arlo_* and shift* types.
9584 2013-09-05  Mike Stump  <mikestump@comcast.net>
9586         * tree.h: Move documentation for tree_function_decl to tree-core.h
9587         with the declaration.
9589 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
9591         PR target/58139
9592         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
9593         looking for widest mode.
9595 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
9597         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
9599 2013-09-05  Richard Biener  <rguenther@suse.de>
9601         PR tree-optimization/58137
9602         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
9603         Do not create vectors of pointers.
9604         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
9605         types for the components of the vector initializer.
9606         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
9607         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
9609 2013-09-05  Martin Jambor  <mjambor@suse.cz>
9611         * ipa-prop.c (remove_described_reference): Accept missing references,
9612         return false if that hppens, otherwise return true.
9613         (cgraph_node_for_jfunc): New function.
9614         (try_decrement_rdesc_refcount): Likewise.
9615         (try_make_edge_direct_simple_call): Use them.
9616         (ipa_edge_removal_hook): Remove references from rdescs.
9617         (ipa_edge_duplication_hook): Clone rdescs and their references
9618         when the new edge has the same caller as the old one.
9619         * cgraph.c (cgraph_resolve_speculation): Remove speculative
9620         reference before removing any edges.
9622 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
9624         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
9625         initial store.
9626         * thumb2.md (thumb2_storewb_parisi): New pattern.
9628 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
9630         * config/aarch64/aarch64-option-extensions.def: Add
9631         AARCH64_OPT_EXTENSION of 'crc'.
9632         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
9633         (AARCH64_ISA_CRC): Ditto.
9634         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
9635         description of the CRC extension.
9637 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9639         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
9640         * config/rs6000/linux.h: Ditto.
9641         * alpha/linux.h: Ditto.
9642         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
9643         no_c99_libc_has_function.
9644         * config/c6x/uclinux-elf.h: Ditto.
9645         * config/lm32/uclinux-elf.h: Ditto.
9646         * config/m68k/uclinux.h: Ditto.
9647         * config/moxie/uclinux.h: Ditto.
9648         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
9649         (crisv32-*-linux*, cris-*-linux*): Ditto.
9650         * config/bfin/bfin.c: Include "tm_p.h".
9652 2013-09-05  Richard Biener  <rguenther@suse.de>
9654         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
9655         check for a definition without a basic-block.
9657 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9658             Sofiane Naci <sofiane.naci@arm.com>
9660         * config/aarch64/aarch64.md
9661         (*movti_aarch64): Rename r_2_f and f_2_r.
9662         (*movsf_aarch64): Likewise.
9663         (*movdf_aarch64): Likewise.
9664         (*movtf_aarch64): Likewise.
9665         (aarch64_movdi_<mode>low): Likewise.
9666         (aarch64_movdi_<mode>high): Likewise.
9667         (aarch64_mov<mode>high_di): Likewise.
9668         (aarch64_mov<mode>low_di): Likewise.
9669         (aarch64_movtilow_tilow): Likewise.
9670         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
9671         values to config/arm/types.md
9672         (attribute "conds"): Update for attribute change.
9673         (anddi3_insn): Likewise.
9674         (iordi3_insn): Likewise.
9675         (xordi3_insn): Likewise.
9676         (one_cmpldi2): Likewise.
9677         * config/arm/types.md (type): Add Neon types.
9678         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
9679         use "type" attribute.
9680         (movmisalign<mode>_neon_store): Likewise.
9681         (movmisalign<mode>_neon_load): Likewise.
9682         (vec_set<mode>_internal): Likewise.
9683         (vec_setv2di_internal): Likewise.
9684         (vec_extract<mode>): Likewise.
9685         (vec_extractv2di): Likewise.
9686         (add<mode>3_neon): Likewise.
9687         (adddi3_neon): Likewise.
9688         (sub<mode>3_neon): Likewise.
9689         (subdi3_neon): Likewise.
9690         (mul<mode>3_neon): Likewise.
9691         (mul<mode>3add<mode>_neon): Likewise.
9692         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
9693         (fma<VCVTF:mode>4)): Likewise.
9694         (fma<VCVTF:mode>4_intrinsic): Likewise.
9695         (fmsub<VCVTF:mode>4)): Likewise.
9696         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
9697         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
9698         (ior<mode>3): Likewise.
9699         (and<mode>3): Likewise.
9700         (anddi3_neon): Likewise.
9701         (orn<mode>3_neon): Likewise.
9702         (orndi3_neon): Likewise.
9703         (bic<mode>3_neon): Likewise.
9704         (bicdi3_neon): Likewise.
9705         (xor<mode>3): Likewise.
9706         (one_cmpl<mode>2): Likewise.
9707         (abs<mode>2): Likewise.
9708         (neg<mode>2): Likewise.
9709         (umin<mode>3_neon): Likewise.
9710         (umax<mode>3_neon): Likewise.
9711         (smin<mode>3_neon): Likewise.
9712         (smax<mode>3_neon): Likewise.
9713         (vashl<mode>3): Likewise.
9714         (vashr<mode>3_imm): Likewise.
9715         (vlshr<mode>3_imm): Likewise.
9716         (ashl<mode>3_signed): Likewise.
9717         (ashl<mode>3_unsigned): Likewise.
9718         (neon_load_count): Likewise.
9719         (ashldi3_neon_noclobber): Likewise.
9720         (signed_shift_di3_neon): Likewise.
9721         (unsigned_shift_di3_neon): Likewise.
9722         (ashrdi3_neon_imm_noclobber): Likewise.
9723         (lshrdi3_neon_imm_noclobber): Likewise.
9724         (widen_ssum<mode>3): Likewise.
9725         (widen_usum<mode>3): Likewise.
9726         (quad_halves_<code>v4si): Likewise.
9727         (quad_halves_<code>v4sf): Likewise.
9728         (quad_halves_<code>v8hi): Likewise.
9729         (quad_halves_<code>v16qi): Likewise.
9730         (reduc_splus_v2di): Likewise.
9731         (neon_vpadd_internal<mode>): Likewise.
9732         (neon_vpsmin<mode>): Likewise.
9733         (neon_vpsmax<mode>): Likewise.
9734         (neon_vpumin<mode>): Likewise.
9735         (neon_vpumax<mode>): Likewise.
9736         (ss_add<mode>_neon): Likewise.
9737         (us_add<mode>_neon): Likewise.
9738         (ss_sub<mode>_neon): Likewise.
9739         (us_sub<mode>_neon): Likewise.
9740         (neon_vadd<mode>_unspec): Likewise.
9741         (neon_vaddl<mode>): Likewise.
9742         (neon_vaddw<mode>): Likewise.
9743         (neon_vhadd<mode>): Likewise.
9744         (neon_vqadd<mode>): Likewise.
9745         (neon_vaddhn<mode>): Likewise.
9746         (neon_vmul<mode>): Likewise.
9747         (neon_vmla<mode>): Likewise.
9748         (neon_vmlal<mode>): Likewise.
9749         (neon_vmls<mode>): Likewise.
9750         (neon_vmlsl<mode>): Likewise.
9751         (neon_vqdmulh<mode>): Likewise.
9752         (neon_vqdmlal<mode>): Likewise.
9753         (neon_vqdmlsl<mode>): Likewise.
9754         (neon_vmull<mode>): Likewise.
9755         (neon_vqdmull<mode>): Likewise.
9756         (neon_vsub<mode>_unspec): Likewise.
9757         (neon_vsubl<mode>): Likewise.
9758         (neon_vsubw<mode>): Likewise.
9759         (neon_vqsub<mode>): Likewise.
9760         (neon_vhsub<mode>): Likewise.
9761         (neon_vsubhn<mode>): Likewise.
9762         (neon_vceq<mode>): Likewise.
9763         (neon_vcge<mode>): Likewise.
9764         (neon_vcgeu<mode>): Likewise.
9765         (neon_vcgt<mode>): Likewise.
9766         (neon_vcgtu<mode>): Likewise.
9767         (neon_vcle<mode>): Likewise.
9768         (neon_vclt<mode>): Likewise.
9769         (neon_vcage<mode>): Likewise.
9770         (neon_vcagt<mode>): Likewise.
9771         (neon_vtst<mode>): Likewise.
9772         (neon_vabd<mode>): Likewise.
9773         (neon_vabdl<mode>): Likewise.
9774         (neon_vaba<mode>): Likewise.
9775         (neon_vabal<mode>): Likewise.
9776         (neon_vmax<mode>): Likewise.
9777         (neon_vmin<mode>): Likewise.
9778         (neon_vpaddl<mode>): Likewise.
9779         (neon_vpadal<mode>): Likewise.
9780         (neon_vpmax<mode>): Likewise.
9781         (neon_vpmin<mode>): Likewise.
9782         (neon_vrecps<mode>): Likewise.
9783         (neon_vrsqrts<mode>): Likewise.
9784         (neon_vqabs<mode>): Likewise.
9785         (neon_vqneg<mode>): Likewise.
9786         (neon_vcls<mode>): Likewise.
9787         (clz<mode>2): Likewise.
9788         (popcount<mode>2): Likewise.
9789         (neon_vrecpe): Likewise.
9790         (neon_vrsqrte): Likewise.
9791         (neon_vget_lane<mode>_sext_internal): Likewise.
9792         (neon_vget_lane<mode>_zext_internal): Likewise.
9793         (neon_vdup_n<mode>): Likewise.
9794         (neon_vdup_nv2di): Likewise.
9795         (neon_vdpu_lane<mode>_internal): Likewise.
9796         (neon_vswp<mode>): Likewise.
9797         (float<mode><V_cvtto>2): Likewise.
9798         (floatuns<mode><V_cvtto>2): Likewise.
9799         (fix_trunc<mode><V_cvtto>)2): Likewise
9800         (fixuns_trunc<mode><V_cvtto)2): Likewise.
9801         (neon_vcvt<mode>): Likewise.
9802         (neon_vcvtv4sfv4hf): Likewise.
9803         (neon_vcvtv4hfv4sf): Likewise.
9804         (neon_vcvt_n<mode>): Likewise.
9805         (neon_vmovn<mode>): Likewise.
9806         (neon_vqmovn<mode>): Likewise.
9807         (neon_vqmovun<mode>): Likewise.
9808         (neon_vmovl<mode>): Likewise.
9809         (neon_vmul_lane<mode>): Likewise.
9810         (neon_vmull_lane<mode>): Likewise.
9811         (neon_vqdmull_lane<mode>): Likewise.
9812         (neon_vqdmulh_lane<mode>): Likewise.
9813         (neon_vmla_lane<mode>): Likewise.
9814         (neon_vmlal_lane<mode>): Likewise.
9815         (neon_vqdmlal_lane<mode>): Likewise.
9816         (neon_vmls_lane<mode>): Likewise.
9817         (neon_vmlsl_lane<mode>): Likewise.
9818         (neon_vqdmlsl_lane<mode>): Likewise.
9819         (neon_vext<mode>): Likewise.
9820         (neon_vrev64<mode>): Likewise.
9821         (neon_vrev32<mode>): Likewise.
9822         (neon_vrev16<mode>): Likewise.
9823         (neon_vbsl<mode>_internal): Likewise.
9824         (neon_vshl<mode>): Likewise.
9825         (neon_vqshl<mode>): Likewise.
9826         (neon_vshr_n<mode>): Likewise.
9827         (neon_vshrn_n<mode>): Likewise.
9828         (neon_vqshrn_n<mode>): Likewise.
9829         (neon_vqshrun_n<mode>): Likewise.
9830         (neon_vshl_n<mode>): Likewise.
9831         (neon_vqshl_n<mode>): Likewise.
9832         (neon_vqshlu_n<mode>): Likewise.
9833         (neon_vshll_n<mode>): Likewise.
9834         (neon_vsra_n<mode>): Likewise.
9835         (neon_vsri_n<mode>): Likewise.
9836         (neon_vsli_n<mode>): Likewise.
9837         (neon_vtbl1v8qi): Likewise.
9838         (neon_vtbl2v8qi): Likewise.
9839         (neon_vtbl3v8qi): Likewise.
9840         (neon_vtbl4v8qi): Likewise.
9841         (neon_vtbx1v8qi): Likewise.
9842         (neon_vtbx2v8qi): Likewise.
9843         (neon_vtbx3v8qi): Likewise.
9844         (neon_vtbx4v8qi): Likewise.
9845         (neon_vtrn<mode>_internal): Likewise.
9846         (neon_vzip<mode>_internal): Likewise.
9847         (neon_vuzp<mode>_internal): Likewise.
9848         (neon_vld1<mode>): Likewise.
9849         (neon_vld1_lane<mode>): Likewise.
9850         (neon_vld1_dup<mode>): Likewise.
9851         (neon_vld1_dupv2di): Likewise.
9852         (neon_vst1<mode>): Likewise.
9853         (neon_vst1_lane<mode>): Likewise.
9854         (neon_vld2<mode>): Likewise.
9855         (neon_vld2_lane<mode>): Likewise.
9856         (neon_vld2_dup<mode>): Likewise.
9857         (neon_vst2<mode>): Likewise.
9858         (neon_vst2_lane<mode>): Likewise.
9859         (neon_vld3<mode>): Likewise.
9860         (neon_vld3qa<mode>): Likewise.
9861         (neon_vld3qb<mode>): Likewise.
9862         (neon_vld3_lane<mode>): Likewise.
9863         (neon_vld3_dup<mode>): Likewise.
9864         (neon_vst3<mode>): Likewise.
9865         (neon_vst3qa<mode>): Likewise.
9866         (neon_vst3qb<mode>): Likewise.
9867         (neon_vst3_lane<mode>): Likewise.
9868         (neon_vld4<mode>): Likewise.
9869         (neon_vld4qa<mode>): Likewise.
9870         (neon_vld4qb<mode>): Likewise.
9871         (neon_vld4_lane<mode>): Likewise.
9872         (neon_vld4_dup<mode>): Likewise.
9873         (neon_vst4<mode>): Likewise.
9874         (neon_vst4qa<mode>): Likewise.
9875         (neon_vst4qb<mode>): Likewise.
9876         (neon_vst4_lane<mode>): Likewise.
9877         (neon_vec_unpack<US>_lo_<mode>): Likewise.
9878         (neon_vec_unpack<US>_hi_<mode>): Likewise.
9879         (neon_vec_<US>mult_lo_<mode>): Likewise.
9880         (neon_vec_<US>mult_hi_<mode>): Likewise.
9881         (neon_vec_<US>shiftl_<mode>): Likewise.
9882         (neon_unpack<US>_<mode>): Likewise.
9883         (neon_vec_<US>mult_<mode>): Likewise.
9884         (vec_pack_trunc_<mode>): Likewise.
9885         (neon_vec_pack_trunk_<mode>): Likewise.
9886         (neon_vabd<mode>_2): Likewise.
9887         (neon_vabd<mode>_3): Likewise.
9888         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
9889         (thumb2_movsi_vfp): Likewise.
9890         (movdi_vfp): Likewise.
9891         (movdi_vfp_cortexa8): Likewise.
9892         (movhf_vfp_neon): Likewise.
9893         (movhf_vfp): Likewiwse.
9894         (movsf_vfp): Likewiwse.
9895         (thumb2_movsf_vfp): Likewiwse.
9896         (movdf_vfp): Likewise.
9897         (thumb2_movdf_vfp): Likewise.
9898         (movsfcc_vfp): Likewise.
9899         (thumb2_movsfcc_vfp): Likewise.
9900         (movdfcc_vfp): Likewise.
9901         (thumb2_movdfcc_vfp): Likewise.
9902         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
9903         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
9904         (v10_v2c): Likewise.
9905         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
9906         attribute change.
9907         (cortex_a15_neon_int_2): Likewise.
9908         (cortex_a15_neon_int_3): Likewise.
9909         (cortex_a15_neon_int_4): Likewise.
9910         (cortex_a15_neon_int_5): Likewise.
9911         (cortex_a15_neon_vqneg_vqabs): Likewise.
9912         (cortex_a15_neon_vmov): Likewise.
9913         (cortex_a15_neon_vaba): Likewise.
9914         (cortex_a15_neon_vaba_qqq): Likewise.
9915         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9916         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9917         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
9918         scalar_64_32_long_scalar): Likewise.
9919         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9920         (cortex_a15_neon_mla_qqq_8_16): Likewise.
9921         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
9922         lotype_qdd_64_32_long): Likewise.
9923         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
9924         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
9925         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
9926         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
9927         (cortex_a15_neon_shift_1): Likewise.
9928         (cortex_a15_neon_shift_2): Likewise.
9929         (cortex_a15_neon_shift_3): Likewise.
9930         (cortex_a15_neon_vshl_ddd): Likewise.
9931         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
9932         (cortex_a15_neon_vsra_vrsra): Likewise.
9933         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
9934         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
9935         (cortex_a15_neon_fp_vmul_ddd): Likewise.
9936         (cortex_a15_neon_fp_vmul_qqd): Likewise.
9937         (cortex_a15_neon_fp_vmla_ddd): Likewise.
9938         (cortex_a15_neon_fp_vmla_qqq): Likewise.
9939         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
9940         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
9941         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
9942         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
9943         (cortex_a15_neon_bp_simple): Likewise.
9944         (cortex_a15_neon_bp_2cycle): Likewise.
9945         (cortex_a15_neon_bp_3cycle): Likewise.
9946         (cortex_a15_neon_vld1_1_2_regs): Likewise.
9947         (cortex_a15_neon_vld1_3_4_regs): Likewise.
9948         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
9949         (cortex_a15_neon_vld2_4_regs): Likewise.
9950         (cortex_a15_neon_vld3_vld4): Likewise.
9951         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
9952         (cortex_a15_neon_vst1_3_4_regs): Likewise.
9953         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
9954         (cortex_a15_neon_vst3_vst4): Likewise.
9955         (cortex_a15_neon_vld1_vld2_lane): Likewise.
9956         (cortex_a15_neon_vld3_vld4_lane" 10
9957         (cortex_a15_neon_vst1_vst2_lane): Likewise.
9958         (cortex_a15_neon_vst3_vst4_lane): Likewise.
9959         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
9960         (cortex_a15_neon_ldm_2): Likewise.0
9961         (cortex_a15_neon_stm_2): Likewise.
9962         (cortex_a15_neon_mcr): Likewise.
9963         (cortex_a15_neon_mcr_2_mcrr): Likewise.
9964         (cortex_a15_neon_mrc): Likewise.
9965         (cortex_a15_neon_mrrc): Likewise.
9966         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
9967         change.
9968         (cortex_a15_alu_shift): Likewise.
9969         (cortex_a15_alu_shift_reg): Likewise.
9970         (cortex_a15_mult32): Likewise.
9971         (cortex_a15_mult64): Likewise.
9972         (cortex_a15_block): Likewise.
9973         (cortex_a15_branch): Likewise.
9974         (cortex_a15_load1): Likewise.
9975         (cortex_a15_load3): Likewise.
9976         (cortex_a15_store1): Likewise.
9977         (cortex_a15_store3): Likewise.
9978         (cortex_a15_call): Likewise.
9979         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
9980         (cortex_a5_f2r): Likewise.
9981         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
9982         change.
9983         (cortex_a53_f2r): Likewise.
9984         * config/arm/cortex-a7.md
9985         (cortex_a7_branch): Update for attribute change.
9986         (cortex_a7_call): Likewise.
9987         (cortex_a7_alu_imm): Likewise.
9988         (cortex_a7_alu_reg): Likewise.
9989         (cortex_a7_alu_shift): Likewise.
9990         (cortex_a7_mul): Likewise.
9991         (cortex_a7_load1): Likewise.
9992         (cortex_a7_store1): Likewise.
9993         (cortex_a7_load2): Likewise.
9994         (cortex_a7_store2): Likewise.
9995         (cortex_a7_load3): Likewise.
9996         (cortex_a7_store3): Likewise.
9997         (cortex_a7_load4): Likewise.
9998         (cortex_a7_store4): Likewise.
9999         (cortex_a7_fpalu): Likewise.
10000         (cortex_a7_fconst): Likewise.
10001         (cortex_a7_fpmuls): Likewise.
10002         (cortex_a7_neon_mul): Likewise.
10003         (cortex_a7_fpmacs): Likewise.
10004         (cortex_a7_neon_mla: Likewise.
10005         (cortex_a7_fpmuld: Likewise.
10006         (cortex_a7_fpmacd: Likewise.
10007         (cortex_a7_fpfmad: Likewise.
10008         (cortex_a7_fdivs: Likewise.
10009         (cortex_a7_fdivd: Likewise.
10010         (cortex_a7_r2f: Likewise.
10011         (cortex_a7_f2r: Likewise.
10012         (cortex_a7_f_flags: Likewise.
10013         (cortex_a7_f_loads: Likewise.
10014         (cortex_a7_f_loadd: Likewise.
10015         (cortex_a7_f_stores: Likewise.
10016         (cortex_a7_f_stored: Likewise.
10017         (cortex_a7_neon): Likewise.
10018         * config/arm/cortex-a8-neon.md
10019         (cortex_a8_neon_mrc): Update for attribute change.
10020         (cortex_a8_neon_mrrc): Likewise.
10021         (cortex_a8_neon_int_1): Likewise.
10022         (cortex_a8_neon_int_2): Likewise.
10023         (cortex_a8_neon_int_3): Likewise.
10024         (cortex_a8_neon_int_4): Likewise.
10025         (cortex_a8_neon_int_5): Likewise.
10026         (cortex_a8_neon_vqneg_vqabs): Likewise.
10027         (cortex_a8_neon_vmov): Likewise.
10028         (cortex_a8_neon_vaba): Likewise.
10029         (cortex_a8_neon_vaba_qqq): Likewise.
10030         (cortex_a8_neon_vsma): Likewise.
10031         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10032         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10033         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10034         long_scalar): Likewise.
10035         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10036         (cortex_a8_neon_mla_qqq_8_16): Likewise.
10037         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10038         long_scalar_qdd_64_32_long): Likewise.
10039         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10040         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10041         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
10042         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10043         (cortex_a8_neon_shift_1): Likewise.
10044         (cortex_a8_neon_shift_2): Likewise.
10045         (cortex_a8_neon_shift_3): Likewise.
10046         (cortex_a8_neon_vshl_ddd): Likewise.
10047         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10048         (cortex_a8_neon_vsra_vrsra): Likewise.
10049         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
10050         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
10051         (cortex_a8_neon_fp_vsum): Likewise.
10052         (cortex_a8_neon_fp_vmul_ddd): Likewise.
10053         (cortex_a8_neon_fp_vmul_qqd): Likewise.
10054         (cortex_a8_neon_fp_vmla_ddd): Likewise.
10055         (cortex_a8_neon_fp_vmla_qqq): Likewise.
10056         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
10057         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
10058         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10059         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10060         (cortex_a8_neon_bp_simple): Likewise.
10061         (cortex_a8_neon_bp_2cycle): Likewise.
10062         (cortex_a8_neon_bp_3cycle): Likewise.
10063         (cortex_a8_neon_ldr): Likewise.
10064         (cortex_a8_neon_str): Likewise.
10065         (cortex_a8_neon_vld1_1_2_regs): Likewise.
10066         (cortex_a8_neon_vld1_3_4_regs): Likewise.
10067         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10068         (cortex_a8_neon_vld2_4_regs): Likewise.
10069         (cortex_a8_neon_vld3_vld4): Likewise.
10070         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10071         (cortex_a8_neon_vst1_3_4_regs): Likewise.
10072         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
10073         (cortex_a8_neon_vst3_vst4): Likewise.
10074         (cortex_a8_neon_vld1_vld2_lane): Likewise.
10075         (cortex_a8_neon_vld3_vld4_lane): Likewise.
10076         (cortex_a8_neon_vst1_vst2_lane): Likewise.
10077         (cortex_a8_neon_vst3_vst4_lane): Likewise.
10078         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
10079         (cortex_a8_neon_mcr): Likewise.
10080         (cortex_a8_neon_mcr_2_mcrr): Likewise.
10081         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
10082         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
10083         change.
10084         (ca9_neon_mrrc): Likewise.
10085         (cortex_a9_neon_int_1): Likewise.
10086         (cortex_a9_neon_int_2): Likewise.
10087         (cortex_a9_neon_int_3): Likewise.
10088         (cortex_a9_neon_int_4): Likewise.
10089         (cortex_a9_neon_int_5): Likewise.
10090         (cortex_a9_neon_vqneg_vqabs): Likewise.
10091         (cortex_a9_neon_vmov): Likewise.
10092         (cortex_a9_neon_vaba): Likewise.
10093         (cortex_a9_neon_vaba_qqq): Likewise.
10094         (cortex_a9_neon_vsma): Likewise.
10095         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10096         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10097         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10098         long_scalar): Likewise.
10099         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10100         (cortex_a9_neon_mla_qqq_8_16): Likewise.
10101         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10102         long_scalar_qdd_64_32_long): Likewise.
10103         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10104         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10105         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
10106         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10107         (cortex_a9_neon_shift_1): Likewise.
10108         (cortex_a9_neon_shift_2): Likewise.
10109         (cortex_a9_neon_shift_3): Likewise.
10110         (cortex_a9_neon_vshl_ddd): Likewise.
10111         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10112         (cortex_a9_neon_vsra_vrsra): Likewise.
10113         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
10114         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
10115         (cortex_a9_neon_fp_vsum): Likewise.
10116         (cortex_a9_neon_fp_vmul_ddd): Likewise.
10117         (cortex_a9_neon_fp_vmul_qqd): Likewise.
10118         (cortex_a9_neon_fp_vmla_ddd): Likewise.
10119         (cortex_a9_neon_fp_vmla_qqq): Likewise.
10120         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
10121         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
10122         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10123         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10124         (cortex_a9_neon_bp_simple): Likewise.
10125         (cortex_a9_neon_bp_2cycle): Likewise.
10126         (cortex_a9_neon_bp_3cycle): Likewise.
10127         (cortex_a9_neon_ldr): Likewise.
10128         (cortex_a9_neon_str): Likewise.
10129         (cortex_a9_neon_vld1_1_2_regs): Likewise.
10130         (cortex_a9_neon_vld1_3_4_regs): Likewise.
10131         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10132         (cortex_a9_neon_vld2_4_regs): Likewise.
10133         (cortex_a9_neon_vld3_vld4): Likewise.
10134         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10135         (cortex_a9_neon_vst1_3_4_regs): Likewise.
10136         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
10137         (cortex_a9_neon_vst3_vst4): Likewise.
10138         (cortex_a9_neon_vld1_vld2_lane): Likewise.
10139         (cortex_a9_neon_vld3_vld4_lane): Likewise.
10140         (cortex_a9_neon_vst1_vst2_lane): Likewise.
10141         (cortex_a9_neon_vst3_vst4_lane): Likewise.
10142         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
10143         (cortex_a9_neon_mcr): Likewise.
10144         (cortex_a9_neon_mcr_2_mcrr): Likewise.
10145         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
10146         (cortex_a9_fps): Likewise.
10147         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
10148         change.
10149         (cortex_m4_fmuls): Likewise.
10150         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
10151         change.
10152         (cortex_r4_mrc): Likewise.
10153         * config/arm/iterators.md: Update comment referring to neon_type.
10154         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
10155         (iwmmxt_movsi_insn): Likewise.
10156         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
10157         attribute change.
10158         (pj4_core_to_vfp): Likewise.
10159         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
10160         attribute change.
10161         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
10162         (vfp_fstore): Likewise.
10163         * doc/md.texi: Change references to neon_type to refer to type.
10165 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10167         * tree.h (DECL_BUILT_IN): Fix typo in comment.
10169 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
10171         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
10172         lglobl if not weak.
10174 2013-09-04  Easwaran Raman  <eraman@google.com>
10176         PR middle-end/57370
10177         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
10178         (build_and_add_sum): Use it.
10179         (appears_later_in_bb): Simplify code.
10181 2013-09-04  Teresa Johnson  <tejohnson@google.com>
10183         * dumpfile.c (dump_finish): Don't close stderr/stdout.
10185 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
10187         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
10189 2013-09-04  Jan Hubicka  <jh@suse.cz>
10191         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
10192         * ipa-devirt.c: Include diganostic.h
10193         (odr_type_d): Add types and types_set.
10194         (hash_type_name): Work for types with vtables during LTO.
10195         (odr_hasher::remove): Fix comment; destroy types_set.
10196         (add_type_duplicate): New function,
10197         (get_odr_type): Use it.
10198         (dump_type_inheritance_graph): Dump type duplicates.
10199         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
10200         graph.
10201         * tree.c (types_same_for_odr): Give exact answers on types with
10202         virtual tables.
10204 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10206         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
10207         explaining their differences.
10209 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
10211         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
10213 2013-09-03  Jeff Law  <law@redhat.com>
10215         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
10216         when not threading through a joiner block.  Pass joiner/no joiner
10217         state to register_jump_thread.
10218         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
10219         state from argument rather than implying on path length.
10220         Dump the entire jump thread path into debugging dump.
10221         * tree-flow.h (register_jump_thread): Update prototype.
10223 2013-08-29  Xinliang David Li  <davidxl@google.com>
10225         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10226         Remove a trivial gcc_assert.
10228 2013-08-29  Xinliang David Li  <davidxl@google.com>
10230         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
10231         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
10232         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10233         Delay base decl alignment adjustment.
10234         * tree-vectorizer.c (vect_destroy_datarefs): New function.
10235         * tree-vectorizer.h: New data structure.
10236         (set_dr_misalignment): New function.
10237         (dr_misalignment): Ditto.
10238         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
10239         (vectorizable_load): Ditto.
10240         (ensure_base_align): New function.
10241         (vectorize_loops): Add dbg_cnt support.
10242         (execute_vect_slp): Ditto.
10243         * dbgcnt.def: New debug counter.
10244         * Makefile: New dependency.
10246 2013-09-03  Meador Inge  <meadori@codesourcery.com>
10248         Revert:
10250         2013-08-30  Meador Inge  <meadori@codesourcery.com>
10252         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10254 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
10256         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
10257         function descriptor.
10259 2013-09-03  Richard Biener  <rguenther@suse.de>
10261         * tree-affine.c (add_elt_to_tree): Fix association issue,
10262         avoid useless converts and make sure to always return a
10263         properly typed result.
10265 2013-09-03  Richard Biener  <rguenther@suse.de>
10267         PR middle-end/57656
10268         * fold-const.c (negate_expr_p): Fix division case.
10269         (negate_expr): Likewise.
10271 2013-09-03  Richard Biener  <rguenther@suse.de>
10273         PR lto/58285
10274         * tree-streamer-out.c: Include tm.h.
10275         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
10277 2013-09-03  Jan Hubicka  <jh@suse.cz>
10279         * tree-profile.c (tree_profiling): Cleanup CFG when done.
10281 2013-09-03  Alan Modra  <amodra@gmail.com>
10283         * config.gcc (powerpc*-*-linux*): Add support for little-endian
10284         multilibs to big-endian target and vice versa.
10285         * config/rs6000/t-linux64: Use := assignment on all vars.
10286         (MULTILIB_EXTRA_OPTS): Remove fPIC.
10287         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
10288         * config/rs6000/t-linux64le: New file.
10289         * config/rs6000/t-linux64bele: New file.
10290         * config/rs6000/t-linux64lebe: New file.
10292 2013-09-02  Jan Hubicka  <jh@suse.cz>
10294         * ipa-inline-transform.c (inline_transform): Do not
10295         optimize_inline_calls when not optimizing.
10297 2013-09-02  Jan Hubicka  <jh@suse.cz>
10299         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
10300         duplicated nodes for assembler names.
10301         * symtab.c (symtab_unregister_node): Do not attempt to unlink
10302         hard registers from assembler name hash.
10304 2013-09-02  Jan Hubicka  <jh@suse.cz>
10306         * ipa-split.c (execute_split_functions): Split externally visible
10307         functions called once.
10309 2013-09-02  Martin Jambor  <mjambor@suse.cz>
10311         PR ipa/58106
10312         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
10313         linked list.  When finding the correct duplicate, also consider also
10314         the caller in additon to its inlined_to node.
10316 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
10318         * config/aarch64/aarch64-simd-builtins.def
10319         (dup_lane_scalar): Remove.
10320         * config/aarch64/aarch64-simd.md
10321         (aarch64_simd_dup): Add 'w->w' alternative.
10322         (aarch64_dup_lane<mode>): Allow for VALL.
10323         (aarch64_dup_lane_scalar<mode>): Remove.
10324         (aarch64_dup_lane_<vswap_width_name><mode>): New.
10325         (aarch64_get_lane_signed<mode>): Add w->w altenative.
10326         (aarch64_get_lane_unsigned<mode>): Likewise.
10327         (aarch64_get_lane<mode>): Likewise.
10328         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
10329         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
10330         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
10331         (VCON): Change container of V2SF.
10332         (vswap_width_name): Likewise.
10333         * config/aarch64/arm_neon.h
10334         (__aarch64_vdup_lane_any): New.
10335         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10336         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
10337         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10339 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
10341         PR middle-end/56382
10342         * expr.c (emit_move_complex): Do not move complex FP values as parts if
10343         the source or the destination is a single hard register.
10345 2013-09-02  Richard Biener  <rguenther@suse.de>
10347         PR middle-end/57511
10348         * tree-scalar-evolution.c (instantiate_scev_name): Allow
10349         non-linear SCEVs.
10351 2013-09-02  Richard Biener  <rguenther@suse.de>
10353         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
10354         arithmetic to sizetype.
10356 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
10358         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
10359         Find auto-increment use both before and after candidate.
10361 2013-09-02  Marek Polacek  <polacek@redhat.com>
10363         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
10365 2013-09-01  Jan Hubicka  <jh@suse.cz>
10367         * Makefile.in: Add ipa-profile.o
10368         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
10369         * cgraph.c (struct cgraph_propagate_frequency_data,
10370         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
10371         ipa-profile.c; replace cgraph_ by ipa_ prefix.
10372         * cgraph.h (cgraph_propagate_frequency): Remove.
10373         * ipa-inline-analysis.c: Include ipa-utils.h;
10374         drop duplicated cfgloop.h.
10375         (inline_update_callee_summaries): Update.
10376         * ipa-profile.c: New file.
10377         * ipa-utils.h (ipa_propagate_frequency): Declare.
10378         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
10379         data-streamer.h, value-prof.h.
10380         (symtab_remove_unreachable_nodes): Update profile.
10381         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
10382         account_time_size, cmp_counts, dump_histogram,
10383         ipa_profile_generate_summary, ipa_profile_write_summary,
10384         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
10385         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
10386         Move to ipa-profile.c.
10388 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
10390         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
10392 2013-09-01  Jan Hubicka  <jh@suse.cz>
10394         * common.opt (fdevirtualize-speculatively): New function.
10395         * invoke.texi (fdevirtualize-speculatively): Document.
10396         * ipa-devirt.c: Include ipa-inline.h
10397         (likely_target_p): New function.
10398         (ipa_devirt): New function.
10399         (gate_ipa_devirt): New function.
10400         (pass_data_ipa_devirt): New static var.
10401         (pass_ipa_devirt): Likewise.
10402         (make_pass_ipa_devirt): New function.
10403         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
10404         (common_handle_option): Disable devirtualization when
10405         value range profiling is available.
10406         * passes.def (pass_ipa_devirt): Add.
10407         * timever.def (TV_IPA_DEVIRT): New timevar.
10408         * tree-pass.h (make_pass_ipa_devirt):
10410 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
10412         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
10413         include sanitize(undefined).
10415 2013-08-31  Diego Novillo  <dnovillo@google.com>
10417         * Makefile.in (TREE_CORE_H): Define.
10418         (TREE_H): Use.
10419         (GTFILES): Add tree-core.h.
10420         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
10421         size the array.
10422         * tree-core.h: New file.
10423         Move all data structures, enum, typedefs, global
10424         declarations and constants from ...
10425         * tree.h: ... here.
10427 2013-08-31  Jan Hubicka  <jh@suse.cz>
10429         * bulitins.c (expand_builtin): Do not early exit for gcov
10430         instrumented functions.
10432 2013-08-31  Marek Polacek  <polacek@redhat.com>
10434         * ubsan.c: Include tm_p.h.
10436 2013-08-31  Jan Hubicka  <jh@suse.cz>
10438         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
10439         warning.
10441         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
10442         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
10443         * tree-cfg.c (verify_gimple_label): ... here.
10444         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
10445         (ipa_merge_profiles): New function.
10446         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
10447         (lto_input_function_body): Likewise.
10448         * ipa-utils.h (ipa_merge_profiles): Declare.
10449         * lto-streamer.h (lto_input_function_body): Update prototype.
10450         (emit_label_in_global_context_p): Remove.
10451         * lto-symtab.c: Include ipa-utils.h
10452         (lto_cgraph_replace_node): Use ipa_merge_profiles.
10454 2013-08-31  Jan Hubicka  <jh@suse.cz>
10456         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
10458 2013-08-31  Jan Hubicka  <jh@suse.cz>
10460         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
10462 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
10464         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
10465         "cmp" RTX before signed_comparison_operator check to account
10466         for "code" changes.
10468 2013-08-30  Jan Hubicka  <jh@suse.cz>
10470         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
10471         (detect_type_change_1): Rename to ...
10472         (detect_type_change): ... this one; early return on non-polymorphic
10473         types.
10474         (detect_type_change_ssa): Add comp_type parameter; update
10475         use of detect_type_change.
10476         (compute_complex_assign_jump_func): Add param_type parameter;
10477         update use of detect_type_change_ssa.
10478         (compute_complex_ancestor_jump_func): Likewise.
10479         (ipa_get_callee_param_type): New function.
10480         (ipa_compute_jump_functions_for_edge): Compute parameter type;
10481         update calls to the jump function computation functions.
10483 2013-08-30  Teresa Johnson  <tejohnson@google.com>
10484             Steven Bosscher  <steven@gcc.gnu.org>
10486         * cfgrtl.c (fixup_new_cold_bb): New routine.
10487         (commit_edge_insertions): Invoke fixup_partitions.
10488         (find_partition_fixes): New routine.
10489         (fixup_partitions): Ditto.
10490         (verify_hot_cold_block_grouping): Update comments.
10491         (rtl_verify_edges): Invoke find_partition_fixes.
10492         (rtl_verify_bb_pointers): Update comments.
10493         (rtl_verify_bb_layout): Ditto.
10494         * basic-block.h (probably_never_executed_edge_p): Declare.
10495         (fixup_partitions): Ditto.
10496         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
10497         * bb-reorder.c (sanitize_hot_paths): New function.
10498         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
10499         sanitize_hot_paths.
10500         * predict.c (probably_never_executed_edge_p): New routine.
10501         * cfg.c (check_bb_profile): Add partition insanity warnings.
10503 2013-08-30  Meador Inge  <meadori@codesourcery.com>
10505         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10507 2013-08-30  Marek Polacek  <polacek@redhat.com>
10509         * Makefile.in (ubsan.o): Add.
10510         (c-family/c-ubsan.o): Add.
10511         (builtins.o): Add ubsan.h dependency.
10512         * ubsan.h: New file.
10513         * ubsan.c: New file.
10514         * common.opt: Add -fsanitize=undefined option.
10515         (flag_sanitize): Add variable.
10516         (fsanitize=): Add option.  Add Driver.
10517         (fsanitize=thread): Remove option.
10518         (fsanitize=address): Likewise.
10519         (static-libubsan): New option.
10520         * doc/invoke.texi: Document the new flag and -static-libubsan.
10521         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
10522         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
10523         * builtin-attrs.def (ATTR_COLD): Define.
10524         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10525         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
10526         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
10527         * flag-types.h (sanitize_code): New enum.
10528         * opts.c (common_handle_option): Parse command line arguments
10529         of -fsanitize=.  Add -fsanitize=unreachable option.
10530         * varasm.c (get_variable_section): Adjust.
10531         (assemble_noswitch_variable): Likewise.
10532         (assemble_variable): Likewise.
10533         (output_constant_def_contents): Likewise.
10534         (categorize_decl_for_section): Likewise.
10535         (place_block_symbol): Likewise.
10536         (output_object_block): Likewise.
10537         * builtins.def: Likewise.
10538         * toplev.c (compile_file): Likewise.
10539         (process_options): Likewise.
10540         * cppbuiltin.c: Likewise.
10541         * tsan.c (tsan_pass): Likewise.
10542         (tsan_gate): Likewise.
10543         (tsan_gate_O0): Likewise.
10544         * cfgexpand.c (partition_stack_vars): Likewise.
10545         (expand_stack_vars): Likewise.
10546         (defer_stack_allocation): Likewise.
10547         (expand_used_vars): Likewise.
10548         * cfgcleanup.c (old_insns_match_p): Likewise.
10549         * asan.c (asan_finish_file): Likewise.
10550         (asan_instrument): Likewise.
10551         (gate_asan): Likewise.
10552         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
10553         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10554         (asan_global_struct): Use pointer_sized_int_node instead
10555         calling build_nonstandard_integer_type.
10556         (initialize_sanitizer_builtins): Likewise.
10557         (asan_finish_file): Likewise.
10558         * gcc.c: Document %{%:function(args):X}.
10559         (static_spec_functions): Add sanitize.
10560         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
10561         store funcval != NULL there.
10562         (do_spec_1): Adjust handle_spec_function caller.
10563         (handle_braces): Allow %:function(args) as condition.
10564         (sanitize_spec_function): New function.
10565         (ADD_STATIC_LIBUBSAN_LIBS): Define.
10566         (LIBUBSAN_SPEC): Likewise.
10567         (LIBUBSAN_EARLY_SPEC): Likewise.
10568         (SANITIZER_SPEC): Handle libubsan.
10569         (SANITIZER_EARLY_SPEC): Likewise.
10570         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
10571         instead of fsanitize=address.
10572         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
10573         instead of fsanitize=address*.
10574         * builtins.c: Include ubsan.h.
10575         (fold_builtin_0): Instrument __builtin_unreachable.
10576         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
10577         instead of flag_asan.
10578         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
10579         (pointer_sized_int_node): Define.
10580         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
10582 2013-08-30  Mike Stump  <mikestump@comcast.net>
10584         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
10585         with RE patterns.
10587 2013-08-29  Jan Hubicka  <jh@suse.cz>
10589         * cgraph.c (cgraph_function_body_availability): Handle weakref
10590         correctly.
10591         * passes.def: Remove pass_fixup_cfg.
10592         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
10593         track when we need to remove functions.
10594         (gate_ipa_inline): Execute inlining always; add comment why.
10595         (pass_data_ipa_inline): Remove TODO_remove_functions.
10596         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
10597         do not produce summaries.
10598         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
10599         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
10600         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
10601         constant pool and vtable.
10603 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
10605         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
10606         New arm_neon.h's internal macros to specify 64-bit constants.
10607         Avoid using stdint.h's macros.
10609 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10611         * recog.c (verify_changes): Verify that changes[i].old is non-zero
10612         before applying REG_P.
10614 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
10616         PR tree-optimization/58277
10617         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
10618         after seeing too many stmts with vdef in between dombb and current
10619         bb, invalidate everything.
10621 2013-08-30  Richard Biener  <rguenther@suse.de>
10623         * fold-const.c (fold_single_bit_test): Fix overflow test.
10625 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
10627         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
10628         and which can live in a register, always retrieve the value on entry.
10629         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
10630         passed by invisible reference specially.
10631         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
10632         (vt_add_function_parameter): Correctly deal with a parameter passed by
10633         invisible reference.
10635 2013-08-30  Jan Hubicka  <jh@suse.cz>
10637         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
10639 2013-08-30  Richard Biener  <rguenther@suse.de>
10641         PR tree-optimization/58228
10642         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
10643         allow invariant loads in nested loop vectorization.
10645 2013-08-30  Richard Biener  <rguenther@suse.de>
10647         PR tree-optimization/58223
10648         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
10649         (has_anti_or_output_dependence): ... this and adjust to also
10650         look for output dependences.
10651         (mark_nodes_having_upstream_mem_writes): Adjust.
10652         (rdg_flag_uses): Likewise.
10654 2013-08-30  Richard Biener  <rguenther@suse.de>
10656         PR tree-optimization/58010
10657         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
10658         assert that we have a loop-closed PHI.
10660 2013-08-29  Jan Hubicka  <jh@suse.cz>
10662         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
10663         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
10664         * lto-section-in.c (lto_free_function_in_decl_state): New function.
10665         (lto_free_function_in_decl_state_for_node): New function.
10667 2013-08-29  Xinliang David Li  <davidxl@google.com>
10669         * loop-unroll.c (report_unroll_peel): Minor message change.
10670         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
10671         Emit alignment peeling message with default -fopt-info.
10672         (vect_loop_versioning): Emit loop version info message.
10673         * tree-vectorizer.c (vectorize_loops): Minor message change.
10674         (execute_vect_slp): Ditto.
10676 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
10678         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
10679         of the clone from the DECL_NAME of the original function.
10681 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
10683         * passes.c (register_pass): Add overload.
10684         * tree-pass.h (register_pass): Forward declare it.  Add comment.
10686 2013-08-29  Jan Hubicka  <jh@suse.cz>
10688         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
10689         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
10690         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
10691         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10692         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
10693         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
10694         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10695         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
10697 2013-08-29  Teresa Johnson  <tejohnson@google.com>
10699         * dumpfile.c (dump_loc): Output column number.
10700         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
10701         * doc/invoke.texi: Document optall -fopt-info flag.
10702         * profile.c (read_profile_edge_counts): Use new dump framework.
10703         (compute_branch_probabilities): Ditto.
10704         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
10705         when pass not in any opt group.
10706         * pass_manager.h (pass_manager::get_pass_profile): New method.
10707         * value-prof.c (check_counter): Use new dump framework.
10708         (check_ic_target): Ditto.
10709         * coverage.c (get_coverage_counts): Ditto.
10710         (coverage_init): Setup new dump framework.
10712 2013-08-29  Richard Biener  <rguenther@suse.de>
10714         PR tree-optimization/58246
10715         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
10716         handle the dominance check inside a basic-block.
10718 2013-08-29  Richard Biener  <rguenther@suse.de>
10720         PR middle-end/57287
10721         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
10722         of default defs that appear in abnormal PHI nodes.
10724 2013-08-29  Richard Biener  <rguenther@suse.de>
10726         PR tree-optimization/57685
10727         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
10728         single-use operands to avoid exponential complexity.
10730 2013-08-28  Dehao Chen  <dehao@google.com>
10732         * ipa-inline.c (edge_badness): Fix integer underflow.
10734 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
10736         * gtm-builtins.def (_ITM_free): Declare leaf.
10738 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10740         PR target/58067
10741         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
10742         (*tls_local_dynamic_base_64_largepic): Likewise.
10743         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
10744         Remove predicate from call operand.
10745         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
10746         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
10748 2013-08-28  Jeff Law  <law@redhat.com>
10750         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
10751         checks for the number of predecessors and successors allowed.
10752         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
10753         which require copying a joiner block if there is a request which
10754         is a subpath that requires no joiner block copying.
10756 2013-08-28  Jan Hubicka  <jh@suse.cz>
10758         * lto-streamer-out.c (DFS_write_tree_body): Drop
10759         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
10760         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
10761         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
10762         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
10763         Do not read DECL_ERROR_ISSUED.
10764         (unpack_ts_decl_with_vis_value_fields): Do not read
10765         DECL_DEFER_OUTPUT.
10766         (lto_input_ts_binfo_tree_pointers): Do not read
10767         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
10768         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
10769         write DECL_ERROR_ISSUED..
10770         (pack_ts_decl_with_vis_value_fields): Do not write
10771         DECL_DEFER_OUTPUT.
10772         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
10773         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
10774         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
10775         * tree.h (tree_decl_common): Update comment.
10776         (DECL_ERROR_ISSUED): Remove.
10778 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10780         PR middle-end/58257
10781         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
10783 2013-08-28  Jan Hubicka  <jh@suse.cz>
10785         * builtins.def (free): Declare leaf.
10787 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
10789         * gdbhooks.py: New.
10790         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
10791         * configure: Regenerate.
10793 2013-08-27  Martin Jambor  <mjambor@suse.cz>
10795         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
10796         (ipa_ancestor_jf_data): Likewise.
10797         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
10798         (ipa_get_jf_pass_through_type_preserved): New function.
10799         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
10800         (ipa_get_jf_ancestor_type_preserved): New function.
10801         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
10802         (ipa_get_jf_ancestor_result): Likewise.
10803         (propagate_vals_accross_pass_through): Use
10804         ipa_get_jf_pass_through_result to do all the value mappings.
10805         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
10806         type_preserved flag.
10807         (ipa_set_jf_cst_copy): New function.
10808         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
10809         (ipa_set_jf_arith_pass_through): Likewise.
10810         (ipa_set_ancestor_jf): Likewise.
10811         (compute_complex_assign_jump_func): Set type_preserved instead of
10812         punting.
10813         (ipa_compute_jump_functions_for_edge): Likewise.
10814         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
10815         (update_jump_functions_after_inlining): Update type_preserved.
10816         Explicitely create jump functions when combining one with pass_through.
10817         (ipa_write_jump_function): Stream the type_preserved flags.
10818         (ipa_read_jump_function): Likewise.
10820 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
10821             Aldy Hernandez  <aldyh@redhat.com>
10823         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
10824         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
10825         * function.h (struct function): Add has_force_vect_loops and
10826         has_simduid_loops.
10827         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
10828         * gimple.c (gimple_build_omp_critical): Add KIND argument and
10829         handle it.
10830         * gimple.def: Update CLAUSES comments.
10831         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
10832         (gimple_build_omp_for): Add argument to prototype.
10833         (gimple_omp_for_kind): New.
10834         (gimple_omp_for_set_kind): New.
10835         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
10836         GOVD_DATA_SHARE_CLASS.
10837         (enum omp_region_type): Add ORT_SIMD.
10838         (gimple_add_tmp_var): Handle ORT_SIMD.
10839         (gimplify_var_or_parm_decl): Same.
10840         (is_gimple_stmt): Same.
10841         (omp_firstprivatize_variable): Same.
10842         (omp_add_variable): Only use splay_tree_insert if lookup failed.
10843         (omp_notice_variable): Handle ORT_SIMD.
10844         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
10845         (omp_check_private): Handle ORT_SIMD.
10846         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10847         OMP_CLAUSE_SAFELEN.
10848         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
10849         Handle OMP_CLAUSE_LASTPRIVATE.
10850         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10851         OMP_CLAUSE_SAFELEN.
10852         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
10853         (gimplify_expr): Handle OMP_SIMD.
10854         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
10855         (expand_GOMP_SIMD_VF): New.
10856         (expand_GOMP_SIMD_LAST_LANE): New.
10857         * internal-fn.def (GOMP_SIMD_LANE): New.
10858         (GOMP_SIMD_VF): New.
10859         (GOMP_SIMD_LAST_LANE): New.
10860         * omp-low.c: Include target.h.
10861         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
10862         OMP_CLAUSE_SAFELEN.
10863         (check_omp_nesting_restrictions): Same.
10864         (omp_max_vf): New.
10865         (lower_rec_simd_input_clauses): New.
10866         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
10867         OMP_CLAUSE_LINEAR.
10868         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
10869         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
10870         (expand_omp_build_assign): New.
10871         (expand_omp_for_init_counts): New.
10872         (expand_omp_for_init_vars): New.
10873         (extract_omp_for_update_vars): New.
10874         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
10875         and rewrite accordingly.
10876         (expand_omp_simd): New.
10877         (expand_omp_for): Use expand_omp_simd.
10878         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
10879         (lower_omp_for): Do not lower the body.
10880         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
10881         in their own loops.
10882         * tree-flow.h (find_omp_clause): Remove prototype.
10883         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
10884         forcing vectorization of the loop, or if flag_tree_vectorize.
10885         (gate_tree_if_conversion): Similarly.
10886         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
10887         gimple_build_omp_for.
10888         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
10889         * tree-parloops (create_parallel_loop): Pass kind argument to
10890         gimple_build_omp_for.
10891         * tree-pretty-print.c (dump_omp_clause): Add cases for
10892         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
10893         OMP_CLAUSE__SIMDUID_.
10894         (dump_generic_node): Handle OMP_SIMD.
10895         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
10896         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
10897         unroll OMP_SIMD loops here.
10898         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
10899         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
10900         loop->safelen.
10901         (vect_analyze_data_refs): Handle simd loops.
10902         * tree-vect-loop.c (vectorizable_live_operation): Handle
10903         IFN_GOMP_SIMD*.
10904         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
10905         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
10906         (vectorizable_load): Same.
10907         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
10908         (struct simduid_to_vf): New.
10909         (simduid_to_vf::hash): New.
10910         (simduid_to-vf::equal): New.
10911         (struct simd_array_to_simduid): New.
10912         (simd_array_to_simduid::hash): New.
10913         (simd_array_to_simduid::equal): New.
10914         (adjust_simduid_builtins): New.
10915         (struct note_simd_array_uses_struct): New.
10916         (note_simd_array_uses_cb): New.
10917         (note_simd_array_uses): New.
10918         (vectorize_loops): Handle simd hints and adjust simd builtins
10919         accordingly.
10920         * tree-vectorizer.h (struct _stmt_vec_info): Add
10921         simd_lane_access_p field.
10922         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
10923         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
10924         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
10925         (omp_clause_code_name): Same.
10926         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
10927         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
10928         * tree.def (OMP_SIMD): New entry.
10929         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
10930         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
10931         (OMP_CLAUSE_DECL): Adjust range for new clauses.
10932         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
10933         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
10934         (OMP_CLAUSE_LINEAR_STEP): New.
10935         (OMP_CLAUSE_SAFELEN_EXPR): New.
10936         (OMP_CLAUSE__SIMDUID__DECL): New.
10937         (find_omp_clause): New prototype.
10939 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
10941         * config/i386/driver-i386.c (host_detect_local_cpu): Update
10942         Haswell processor detection.
10944 2013-08-27  Christian Widmer  <shadow@umbrox.de>
10946         PR target/57927
10947         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
10948         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
10949         AVX2 capable processors.
10951 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
10953         * config/aarch64/arm_neon.h: Replace all inline asm implementations
10954         of vget_low_* with implementations in terms of other intrinsics.
10956 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
10958         PR middle-end/57219
10959         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
10960         values to -1, 0 and 1.
10962 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
10964         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
10965         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
10966         (<optab><mode>3_insn): Remove.
10967         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
10968         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
10969         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
10970         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
10971         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
10972         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
10973         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
10974         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
10975         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
10976         (ror<mode>3_insn): Likewise.
10977         * config/aarch64/predicates.md (aarch64_simd_register): New.
10979 2013-08-27  Richard Biener  <rguenther@suse.de>
10981         PR tree-optimization/57521
10982         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
10983         one edge is non-critical.
10984         (find_phi_replacement_condition): Make sure to use a non-critical
10985         edge.  Cleanup and remove old bug workarounds.
10986         (bb_postdominates_preds): Remove.
10987         (if_convertible_loop_p_1): Do not compute post-dominators.
10988         (combine_blocks): Do not free post-dominators.
10989         (main_tree_if_conversion): Likewise.
10990         (pass_data_if_conversion): Add TODO_verify_ssa.
10992 2013-08-27  DJ Delorie  <dj@redhat.com>
10994         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
10996 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
10998         * function.c (assign_parm_find_data_types): Set passed_mode and
10999         nominal_mode to the TYPE_MODE of nominal_type for the built
11000         pointer type in case of the struct-pass-by-reference.
11002 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11004         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
11005         (UINT16_TYPE): Change default to "unsigned int".
11007         * config/avr/avr.opt (mfract-convert-truncate): New option.
11008         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
11009         is set, round negative fractional integers according to n1169
11010         when converting to integer types.
11012 2013-08-26  Jan Hubicka  <jh@suse.cz>
11014         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
11015         methods can not be called indirectly when their address is not taken.
11017 2013-08-26  Jan Hubicka  <jh@suse.cz>
11019         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
11020         ctor_for_folding.
11022 2013-08-26  Jan Hubicka  <jh@suse.cz>
11024         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
11025         can be unshared.
11027 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11029         * reload.c (find_valid_class): Allow classes that do not include
11030         FIRST_PSEUDO_REGISTER - 1.
11032 2013-08-26  Jan Hubicka  <jh@suse.cz>
11034         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
11035         fix edge count/frequency when speculation failed; fix type check
11036         for the direct call.
11038 2013-08-26  Jan Hubicka  <jh@suse.cz>
11040         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
11042 2013-08-26  Jan Hubicka  <jh@suse.cz>
11044         * ipa-inline-transform.c (inline_transform): Be ready for basic block
11045         to be changed by edge redirection.
11047 2013-08-26  Jan Hubicka  <jh@suse.cz>
11049         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
11050         formating; add sanity check.
11051         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
11052         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
11053         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
11054         (resolve_noninline_speculation): Update callee keys, too.
11056 2013-08-26  Jan Hubicka  <jh@suse.cz>
11058         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
11059         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
11061 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11063         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
11064         into proper place.
11066 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
11068         * config/i386/i386.c (ix86_debug_options): Remove prototype.
11069         (x86_64_elf_select_section): Ditto.
11070         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
11071         arguments.
11072         (ix86_pass_by_reference): Ditto.
11073         (output_set_got): Ditto.
11074         (ix86_unary_operator_ok): Ditto.
11075         (ix86_expand_builtin): Ditto.
11077 2013-08-23  Jan Hubicka  <jh@suse.cz>
11079         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
11081 2013-08-23  Jan Hubicka  <jh@suse.cz>
11083         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
11084         (tree_decl_with_vis): Add FINAL field.
11086 2013-08-23  Jeff Law  <law@redhat.com>
11088         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
11089         the debugging dump when the expression is fully redundant.
11091 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11093         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
11094         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
11095         * pretty-print.c (pp_formatted_text_data): Likewise.
11096         (pp_write_text_to_stream): Likewise.
11097         (pp_write_text_as_dot_label_to_stream): Likewise.
11098         (pp_append_r): Likewise.
11099         (pp_format): Likewise.
11100         (pp_flush): Likewise.
11101         (pp_clear_output_area): Likewise.
11102         (pp_append_text): Likewise.
11103         (pp_formatted_text): Likewise.
11104         (pp_remaining_character_count_for_line): Likewise.
11105         (pp_newline): Likewise.
11106         (pp_character): Likewise.
11107         (output_buffer::~output_buffer): Define.
11108         (pretty_printer::~pretty_printer): Destruct output buffer.
11109         * pretty-print.h (output_buffer::~output_buffer): Declare.
11110         (pretty_printer::~pretty_printer): Declare virtual.
11112 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
11114         PR other/57324
11115         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
11116         HOST_WIDE_INT_M1U): New macros.
11117         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
11118         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
11119         unsigned -1 for lshift.
11120         * cse.c (cse_insn): Likewise.
11121         * double-int.c (rshift_double, lshift_double): Likewise.
11122         * builtins.c (fold_builtin_bitop): Likewise.
11123         * combine.c (force_to_mode): Likewise.
11124         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
11125         * simplify-rtx.c (simplify_const_unary_operation,
11126         simplify_const_binary_operation): Likewise.
11127         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
11128         check_va_list_escapes): Likewise.
11129         * rtlanal.c (nonzero_bits1): Likewise.
11130         * expmed.c (expand_smod_pow2): Likewise.
11131         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
11133 2013-08-23  Jan Hubicka  <jh@suse.cz>
11135         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
11136         as having address taken.
11138 2013-08-23  Jan Hubicka  <jh@suse.cz>
11140         * ipa-utils.h (method_class_type): Declare.
11141         * ipa-devirt.c (method_class_type): Export.
11143         * cgraphunit.c (analyze_functions): Do basic devirtualization;
11144         do not walk base classes of anonymous types.
11146 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
11148         PR rtl-optimization/58220
11149         PR regression/58221
11150         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
11151         handle SEQUENCE insns properly.
11153 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11155         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
11156         definition.
11157         (pp_newline_and_indent): Likewise.
11158         (pp_separate_with): Likewise.
11159         * pretty-print.c (pp_newline_and_flush): Define.
11160         (pp_newline_and_indent): Likewise.
11161         (pp_separate_with): Likewise.
11163 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11165         PR target/58218
11166         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
11167         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
11169 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
11171         * config/i386/predicates.md (ext_sse_reg_operand): New.
11172         * config/i386/i386.md (*movti_internal): Use
11173         predicate to determine if EVEX is needed.
11174         (*movsi_internal): Ditto.
11175         (*movdf_internal): Ditto.
11176         (*movsf_internal): Ditto.
11177         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
11179 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11181         PR tree-optimization/58209
11182         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
11183         (find_tail_calls): Give up for pointer result types if m is non-NULL.
11184         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
11185         emit POINTER_PLUS_EXPR.
11186         (create_tailcall_accumulator): For pointer result type accumulate in
11187         sizetype type.
11189 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
11191         * configure.ac: Add backslashes missing from the last change.
11192         * configure: Regenerate.
11194 2013-08-22  Jan Hubicka  <jh@susue.cz>
11196         * ipa.c (function_and_variable_visibility): First remember function
11197         was global and then make it local.
11199 2013-08-22  Julian Brown  <julian@codesourcery.com>
11201         * configure.ac: Add aarch64 to list of arches which use "nop" in
11202         debug_line test.
11203         * configure: Regenerate.
11205 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11207         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
11208         gnu_libc_has_function.
11209         * config/s390/tpf.h: Likewise.
11211 2013-08-22  Jan Hubicka  <jh@susue.cz>
11213         * timevar.c (validate_phases): Add cast.
11215 2013-08-22  Jan Hubicka  <jh@susue.cz>
11217         * timevar.c (validate_phases): Use size_t for memory.
11218         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
11220 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11222         * pretty-print.h (output_buffer::output_buffer): Declare.
11223         (pretty_printer::pretty_printer): Likewise.
11224         (pp_construct): Remove.
11225         * pretty-print.c (output_buffer::output_buffer): Define.
11226         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
11227         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
11228         (print_gimple_expr): Likewise.
11229         (print_gimple_seq): Likewise.
11230         (gimple_dump_bb): Likewise.
11231         * sched-vis.c (dump_value_slim): Likewise.
11232         (dump_insn_slim): Likewise.
11233         (dump_rtl_slim): Likewise.
11234         (str_pattern_slim): Likewise.
11235         * tree-mudflap.c (mf_varname_tree): Likewise.
11236         * graph.c (print_graph_cfg): Likewise.
11237         (start_graph_dump): Likewise.
11238         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
11239         placement-new.
11240         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
11241         pretty printer initialization.
11242         * coretypes.h (diagnostic_context): Remove superflous type alias
11243         declaration.
11244         (pretty_printer): Likewise.  Declare directly as a class.
11245         (pretty_print_info): Remove declaration as class.
11246         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
11247         and pp_clear_output_area.
11248         (asan_add_global): Likewise.
11250 2013-08-22  Jan Hubicka  <jh@suse.cz>
11252         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
11253         * ipa-utils.h (update_type_inheritance_graph): Declare.
11254         (possible_polymorphic_call_target_p): Declare.
11255         (possible_polymorphic_call_target_p): New.
11256         * ipa-devirt.c: Update toplevel comments.
11257         (cached_polymorphic_call_targets): Move up.
11258         (odr_type_d): Move ID down.
11259         (polymorphic_type_binfo_p): Update comment.
11260         (odr_hasher::remove): Likewise;
11261         (get_odr_type): Set anonymous_namespace.
11262         (dump_odr_type): Dump it.
11263         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
11264         (maybe_record_node): Record node in cached_polymorphic_call_targets.
11265         (record_binfo): Add comment.
11266         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
11267         built.
11268         (devirt_node_removal_hook): Do not iCE when cache is freed.
11269         (possible_polymorphic_call_target_p): New predicate.
11270         (update_type_inheritance_graph): New function.
11272 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11273             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
11274             Sergey Lega  <sergey.s.lega@intel.com>
11275             Anna Tikhonova  <anna.tikhonova@intel.com>
11276             Ilya Tocar  <ilya.tocar@intel.com>
11277             Andrey Turetskiy  <andrey.turetskiy@intel.com>
11278             Ilya Verbin  <ilya.verbin@intel.com>
11279             Kirill Yukhin  <kirill.yukhin@intel.com>
11280             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11282         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
11283         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
11284         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
11285         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
11286         (OPTION_MASK_ISA_AVX2_UNSET): Update.
11287         (OPTION_MASK_ISA_AVX512F_UNSET): New.
11288         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
11289         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
11290         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
11291         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
11292         OPT_mavx512pf, OPT_mavx512er cases.
11293         * config/i386/constraints.md (v): New constraint.
11294         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
11295         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
11296         (bit_AVX512CD): New.
11297         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11298         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
11299         * config/i386/i386-c.c (ix86_target_macros_internal):
11300         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
11301         __AVX512PF__.
11302         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
11303         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
11304         * config/i386/i386.c (regclass_map, dbx_register_map)
11305         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
11306         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
11307         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
11308         -mavx512pf options.
11309         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
11310         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
11311         -mavx512pf options.  Fix formatting.
11312         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
11313         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
11314         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
11315         -mavx512cd, -mavx512pf options.
11316         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
11317         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
11318         (ix86_preferred_output_reload_class): Replace SSE_REGS with
11319         ALL_SSE_REGS.
11320         (ix86_hard_regno_mode_ok): Support 512-bit registers.
11321         (ix86_set_reg_reg_cost): Ditto.
11322         (x86_order_regs_for_local_alloc): Ditto.
11323         (MAX_VECT_LEN): Extend to 64-byte.
11324         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
11325         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
11326         (TARGET_AVX512ER, TARGET_AVX512CD): New.
11327         (BIGGEST_ALIGNMENT): Extend to 512-bits.
11328         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
11329         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
11330         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
11331         (SSE_REG_MODE_P): Support new modes.
11332         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
11333         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
11334         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
11335         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
11336         (REG_CLASS_CONTENTS): Add new registers.
11337         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
11338         (EXT_REX_SSE_REGNO_P): New.
11339         (HI_REGISTER_NAMES): Add new registers.
11340         * config/i386/i386.md: Define constants for new registers.
11341         (mode): Add new 512-bit modes.
11342         (prefix): Support evex prefix.
11343         (isa): Support avx512f, noavx512f, fma_avx512f.
11344         (ssemodesuffix): Add new 512-bit modes.
11345         (movxi): New.
11346         (*movxi_internal_avx512f): Ditto.
11347         (*movdi_internal): Replace constraint "x" with the new constraint "v".
11348         Support MODE_XI.
11349         (*movsi_internal): Likewise.
11350         (*movdf_internal): Likewise.
11351         (*movsf_internal): Likewise.
11352         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
11353         (<code><mode>3): Likewise.
11354         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
11355         New.
11356         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
11357         with the new constraint "v".
11358         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
11359         modes.
11360         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
11361         with the new constraint "v".
11362         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
11363         (<sse2>_storedqu<avxsizesuffix>): Likewise.
11364         (*<plusminus_insn><mode>3): Likewise.
11365         (<sse>_vm<plusminus_insn><mode>3): Likewise.
11366         (*mul<mode>3): Likewise.
11367         (<sse>_vmmul<mode>3): Likewise.
11368         (<sse>_div<mode>3): Likewise.
11369         (<sse>_vmdiv<mode>3): Likewise.
11370         (<sse>_sqrt<mode>2): Likewise.
11371         (<sse>_vmsqrt<mode>2): Likewise.
11372         (*<code><mode>3_finite): Likewise.
11373         (*<code><mode>3) <smaxmin>: Likewise.
11374         (<sse>_vm<code><mode>3): Likewise.
11375         (*<code><mode>3) <any_logic>: Likewise.
11376         (*fma_fmadd_<mode>): Likewise.
11377         (*fma_fmsub_<mode>): Likewise.
11378         (*fma_fnmadd_<mode>): Likewise.
11379         (*fma_fnmsub_<mode>): Likewise.
11380         (*fma_fmaddsub_<mode>): Likewise.
11381         (*fma_fmsubadd_<mode>): Likewise.
11382         (*fmai_fmadd_<mode>): Likewise.
11383         (*fmai_fmsub_<mode>): Likewise.
11384         (*fmai_fnmadd_<mode>): Likewise.
11385         (*fmai_fnmsub_<mode>): Likewise.
11386         (sse_cvtsi2ss): Likewise.
11387         (sse_cvtsi2ssq): Likewise.
11388         (sse_cvtss2si): Likewise.
11389         (sse_cvtss2si_2): Likewise.
11390         (sse_cvtss2siq): Likewise.
11391         (sse_cvtss2siq_2): Likewise.
11392         (sse_cvttss2si): Likewise.
11393         (sse_cvtss2siq_2): Likewise.
11394         (float<sseintvecmodelower><mode>2): Likewise.
11395         (sse2_cvtsd2si_2): Likewise.
11396         (sse2_cvtsd2siq_2): Likewise.
11397         (*<plusminus_insn><mode>3): Likewise.
11398         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
11399         (*<sse4_1_avx2>_mul<mode>3): Likewise.
11400         (ashr<mode>3): Likewise.
11401         (<shift_insn><mode>3): Likewise.
11402         (avx2_<code><mode>3): Likewise.
11403         (*avx2_<code><mode>3): Likewise.
11404         (*andnot<mode>3): Likewise.
11405         (*<code><mode>3) <any_logic>: Likewise.
11406         (abs<mode>2): Likewise.
11407         (avx2_permvar<mode>): Likewise.
11408         (avx2_perm<mode>_1): Likewise.
11409         (*avx_vpermilp<mode>): Likewise.
11410         (avx_vpermilvar<mode>3): Likewise.
11411         (avx2_ashrv<mode>): Likewise.
11412         (avx2_<shift_insn>v<mode>): Likewise.
11413         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
11414         -mavx512cd.
11415         * doc/rtl.texi: Document XImode.
11417 2013-08-21  Jeff Law  <law@redhat.com>
11419         * tree-flow.h (register_jump_thread): Pass vector of edges
11420         instead of each important edge.
11421         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
11422         thread path into a vector and pass that to register_jump_thread.
11423         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
11424         passed in edge vector to the current 3-edge form.
11426         Revert:
11427         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11429         * dce.c (fini_dce): Call df_analyze again just in case
11430         delete_unmarked_insns removed anything.
11432 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
11434         * reload.h (struct reg_equivs): Rename to ..
11435         (struct reg_equivs_s): .. this.
11437 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
11439         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
11441 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11443         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
11445 2013-08-21  Jeff Law  <law@redhat.com>
11447         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
11448         simplify assignments too.  If the RHS collapses to a singleton
11449         range, then return the value for the range.
11451 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
11453         * config/i386/sse.md (V16): Rename to...
11454         (VMOVE): this.
11455         (mov<mode>): Update iterator name.
11456         (*mov<mode>_internal): Ditto.
11457         (push<mode>1): Ditto.
11458         (movmisalign<mode>): Ditto.
11460 2013-08-20  Jan Hubicka  <jh@suse.cz>
11462         PR bootstrap/58186
11463         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
11464         entry for direct edges.
11465         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
11467 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11469         Revert my last two changes, r201865 and r201864:
11471         Revert r201865:
11472         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11474         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11475         instances can own GC refs.
11477         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11478         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11479         (gcc::context::gt_pch_nx): Likewise.
11480         (gcc::context::gt_pch_nx):  Likewise.
11481         * ggc.h (gt_ggc_mx <T>): New.
11482         (gt_pch_nx_with_op <T>): New.
11483         (gt_pch_nx <T>): New.
11484         * passes.c (opt_pass::gt_ggc_mx): New.
11485         (opt_pass::gt_pch_nx): New.
11486         (opt_pass::gt_pch_nx_with_op): New.
11487         (pass_manager::gt_ggc_mx): New.
11488         (pass_manager::gt_pch_nx): New.
11489         (pass_manager::gt_pch_nx_with_op): New.
11490         (pass_manager::operator new): Use
11491         ggc_internal_cleared_alloc_stat rather than xcalloc.
11492         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11493         (pass_manager::gt_ggc_mx): New.
11494         (pass_manager::gt_pch_nx): New.
11495         (pass_manager::gt_pch_nx_with_op): New.
11496         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11497         (opt_pass::operator new): New.
11498         (opt_pass::gt_ggc_mx): New.
11499         (opt_pass::gt_pch_nx): New.
11500         (opt_pass::gt_pch_nx_with_op): New.
11502         Revert r201864:
11503         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11505         * Makefile.in (GTFILES): Add context.h.
11506         * context.c (gcc::context::operator new): New.
11507         (gcc::context::gt_ggc_mx): New.
11508         (gcc::context::gt_pch_nx): New.
11509         (gcc::context::gt_pch_nx): New.
11510         * context.h (gcc::context): Add GTY((user)) marking.
11511         (gcc::context::operator new): New.
11512         (gcc::context::gt_ggc_mx): New.
11513         (gcc::context::gt_pch_nx): New.
11514         (gcc::context::gt_pch_nx): New.
11515         (g): Add GTY marking.
11516         (gt_ggc_mx (gcc::context *)): New.
11517         (gt_pch_nx (gcc::context *)): New.
11518         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11519         void *cookie)): New.
11520         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11522 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11524         * dce.c (fini_dce): Call df_analyze again just in case
11525         delete_unmarked_insns removed anything.
11527 2013-08-20  Teresa Johnson  <tejohnson@google.com>
11529         PR rtl-optimizations/57451
11530         * final.c (reemit_insn_block_notes): Prevent lexical blocks
11531         from crossing split section boundaries.
11533 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
11535         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
11536         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
11537         with MULTLIB_DEFAULTS.
11539 2013-08-20  Nick Clifton  <nickc@redhat.com>
11541         * target.def (narrow_volatile_bitfield): Note that the default
11542         value is false, not !TARGET_STRICT_ALIGN.
11543         * doc/tm.texi: Regenerate.
11545 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
11547         Fix LIB_SPEC for systems without libpthread.
11549         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
11550         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
11551         for Android.
11552         * config/i386/linux-common.h: Likewise.
11553         * config/mips/linux-common.h: Likewise.
11555 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
11557         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
11558         checks.
11560 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11562         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11563         instances can own GC refs.
11565         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11566         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11567         (gcc::context::gt_pch_nx): Likewise.
11568         (gcc::context::gt_pch_nx):  Likewise.
11569         * ggc.h (gt_ggc_mx <T>): New.
11570         (gt_pch_nx_with_op <T>): New.
11571         (gt_pch_nx <T>): New.
11572         * passes.c (opt_pass::gt_ggc_mx): New.
11573         (opt_pass::gt_pch_nx): New.
11574         (opt_pass::gt_pch_nx_with_op): New.
11575         (pass_manager::gt_ggc_mx): New.
11576         (pass_manager::gt_pch_nx): New.
11577         (pass_manager::gt_pch_nx_with_op): New.
11578         (pass_manager::operator new): Use
11579         ggc_internal_cleared_alloc_stat rather than xcalloc.
11580         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11581         (pass_manager::gt_ggc_mx): New.
11582         (pass_manager::gt_pch_nx): New.
11583         (pass_manager::gt_pch_nx_with_op): New.
11584         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11585         (opt_pass::operator new): New.
11586         (opt_pass::gt_ggc_mx): New.
11587         (opt_pass::gt_pch_nx): New.
11588         (opt_pass::gt_pch_nx_with_op): New.
11590 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11592         * Makefile.in (GTFILES): Add context.h.
11593         * context.c (gcc::context::operator new): New.
11594         (gcc::context::gt_ggc_mx): New.
11595         (gcc::context::gt_pch_nx): New.
11596         (gcc::context::gt_pch_nx): New.
11597         * context.h (gcc::context): Add GTY((user)) marking.
11598         (gcc::context::operator new): New.
11599         (gcc::context::gt_ggc_mx): New.
11600         (gcc::context::gt_pch_nx): New.
11601         (gcc::context::gt_pch_nx): New.
11602         (g): Add GTY marking.
11603         (gt_ggc_mx (gcc::context *)): New.
11604         (gt_pch_nx (gcc::context *)): New.
11605         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11606         void *cookie)): New.
11607         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11609 2013-08-20  Alan Modra  <amodra@gmail.com>
11611         PR target/57865
11612         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
11613         (rs6000_emit_epilogue): Likewise.
11615 2013-08-19  Dehao Chen  <dehao@google.com>
11617         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
11619 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
11620             Jakub Jelinek  <jakub@redhat.com>
11622         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
11623         (BUILT_IN_FABSD64): Likewise.
11624         (BUILT_IN_FABSD128): Likewise.
11625         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
11626         (fold_builtin_1): Likewise.
11627         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
11628         destination and source operands.
11629         (*abstd2_fpr): Likewise.
11630         (*nabstd2_fpr): Likewise.
11632 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
11634         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
11635         JUMP_P and INSN_P.
11637 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
11639         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
11640         incomplete.
11642 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11644         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
11645         * builtins.c (default_libc_has_function): New.
11646         (gnu_libc_has_function): Ditto.
11647         (no_c99_libc_has_function): Ditto.
11648         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
11649         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
11650         (fold_builtin_sincos): Likewise.
11651         (fold_builtin_cexp): Likewise.
11652         * builtins.def (DEF_C94_BUILTIN): Likewise.
11653         (DEF_C99_BUILTIN): Likewise.
11654         (DEF_C99_C90RES_BUILTIN): Likewise.
11655         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
11656         definitions to using this define.
11657         * config/darwin-protos.h (darwin_libc_has_function): New.
11658         * config/darwin.c (darwin_libc_has_function): Ditto.
11659         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
11660         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
11661         * config/darwin.h: Ditto.
11662         * config/elfos.h: Ditto.
11663         * config/freebsd.h: Ditto.
11664         * config/i386/cygming.h: Ditto.
11665         * config/i386/djgpp.h: Ditto.
11666         * config/i386/i386-interix.h: Ditto.
11667         * config/microblaze/microblaze.h: Ditto.
11668         * config/mmix/mmix.h: Ditto.
11669         * config/gnu-user.h: Ditto.
11670         * config/ia64/hpux.h: Ditto.
11671         * config/pa/pa-hpux.h: Ditto.
11672         * config/pdp11/pdp11.h: Ditto.
11673         * config/picochip/picochip.h: Ditto.
11674         * config/linux.h: Ditto.
11675         * config/netbsd.h: Ditto.
11676         * config/openbsd.h: Ditto.
11677         * config/rs6000/aix43.h: Ditto.
11678         * config/rs6000/aix51.h: Ditto.
11679         * config/rs6000/aix52.h: Ditto.
11680         * config/rs6000/aix53.h: Ditto.
11681         * config/rs6000/aix61.h: Ditto.
11682         * config/rs6000/darwin.h: Ditto.
11683         * config/rs6000/linux.h: Ditto.
11684         * config/rs6000/linux64.h: Ditto.
11685         * config/s390/tpf.h: Ditto.
11686         * config/sol2-10.h: Ditto.
11687         * config/sol2.h: Ditto.
11688         * config/vms/vms.h: Ditto.
11689         * config/vxworks.h: Ditto.
11690         * config/linux-android.c (linux_android_libc_has_function):
11691         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
11692         * config/linux-protos.h (linux_android_libc_has_function):
11693         New declaration.
11694         * config/i386/i386.c (ix86_libc_has_function): New.
11695         * config/i386/i386-protos.h
11696         (ix86_libc_has_function): New declaration.
11697         * config/i386/i386.md
11698         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
11699         ("isinf<mode>2): Likewise.
11700         * convert.c (convert_to_integer): Using new target hook
11701         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
11702         TARGET_C99_FUNCTIONS.
11703         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
11704         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
11705         * coretypes.h (function_class): New enum for different
11706         classes of functions.
11707         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
11708         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
11709         (TARGET_HAS_SINCOS): Likewise.
11710         (TARGET_LIBC_HAS_FUNCTION): New.
11711         * doc/tm.texi: Regenerated.
11712         * targhooks.h (default_libc_has_function): New declaration.
11713         (no_c99_libc_has_function): Ditto.
11714         (gnu_libc_has_function): Ditto.
11715         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
11716         and TARGET_HAS_SINCOS.
11718 2013-08-18  Jan Hubicka  <jh@suse.cz>
11720         * Makeifle-in (ipa-devirt.o): New.
11721         (GTFILES): Add ipa-utils.h and ipa-devirt.c
11722         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
11723         (analyze_functions): Look into possible targets of polymorphic call.
11724         * dumpfile.c (dump_files): Add type-inheritance dump.
11725         * dumpfile.h (TDI_inheritance): New.
11726         * ipa-devirt.c: New file.
11727         * ipa-utils.h (odr_type_d): Forward declare.
11728         (odr_type): New type.
11729         (build_type_inheritance_graph): Declare.
11730         (possible_polymorphic_call_targets): Declare and introduce inline
11731         variant when only edge is pased.
11732         (dump_possible_polymorphic_call_targets): Likewise.
11733         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
11734         * tree.c (type_in_anonymous_namespace_p): Break out from ...
11735         (types_same_for_odr): ... here.
11736         * tree.h (type_in_anonymous_namespace_p): Declare.
11738 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
11740         PR tree-optimization/58006
11741         * tree-parloops.c (take_address_of): Don't ICE if get_name
11742         returns NULL.
11743         (eliminate_local_variables_stmt): Remove clobber stmts.
11745 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
11747         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
11748         error message is issued for an alias to undefined symbol.
11750 2013-08-18  Jan Hubicka  <jh@suse.cz>
11752         * cgraph.c (cgraph_create_indirect_edge): Discover
11753         polymorphic calls and record basic info into indirect_info.
11754         * gimple-fold.c (gimple_fold_call): When doing BINFO based
11755         devirtualization, ignore objc function calls.
11756         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
11757         call with no parm index info.
11758         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
11759         * tree.c (virtual_method_call_p): New function.
11760         * tree.h (virtual_method_call_p): Declare.
11762 2013-08-16  Jan Hubicka  <jh@suse.cz>
11764         PR middle-end/58179
11765         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
11767 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
11769         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
11770         attribute.
11772 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11774         * gengtype.c (type_for_name): Add special-case support for
11775         locating types within the "gcc::" namespace.
11776         (open_base_files): Emit a "using namespace gcc" directive.
11778 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
11780         PR target/58160
11781         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
11782         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
11784         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
11785         array instead of each individual operand as a separate argument.
11786         (emit_fusion_gpr_load): Likewise.
11787         (expand_fusion_gpr_load): Add new function declaration.
11789         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
11790         signature to have the operands passed as an array, instead of as
11791         separate arguments.  Allow ZERO_EXTEND to be in the memory
11792         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
11793         depend on the register live/dead flags when peepholes are run.
11794         (expand_fusion_gpr_load): New function to be called from the
11795         peephole2 pass, to change the register that addis sets to be the
11796         target register.
11797         (emit_fusion_gpr_load): Change the calling signature to have the
11798         operands passed as an array, instead of as separate arguments.
11799         Allow ZERO_EXTEND to be in the memory address, and also
11800         SIGN_EXTEND if -mpower8-fusion-sign.
11802         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
11803         unspec enumeration.
11804         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
11805         adjust the register addis loads up in the peephole2 pass.  Do not
11806         depend on the register live/dead state when the peephole pass is done.
11808 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11810         * gengtype.c (create_user_defined_type): Ensure that the kind
11811         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
11812         declaration is seen before the GTY((user)) marking.
11814 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
11816         PR target/58105
11817         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
11819 2013-08-16  Jan Hubicka  <jh@suse.cz>
11821         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
11822         arugment expected_type.
11823         (gimple_fold_call): Use it.
11824         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
11825         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
11826         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
11827         (try_make_edge_direct_virtual_call): Likewise.
11828         * tree.c (obj_type_ref_class): New.
11829         * tree.h (obj_type_ref_class): Use it.
11831 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11833         * sched-vis.c (rtl_slim_pp_initialized): Remove.
11834         (rtl_slim_pp): Likewise.
11835         (init_rtl_slim_pretty_print): Likewise.
11836         (dump_value_slim):  Don't call it.  Use local pretty printer.
11837         (dump_insn_slim): Likewise.
11838         (dump_rtl_slim): Likewise.
11839         (str_pattern_slim): Likewise.
11840         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
11841         Simplify.
11843 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
11845         PR tree-optimization/58164
11846         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
11847         walk gimple_goto_dest of GIMPLE_GOTO.
11849         PR tree-optimization/58165
11850         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
11851         bi_call must be the last stmt in a bb, don't split_block, instead
11852         use fallthru edge from it and give up if there is none.
11853         Release conds vector when returning early.
11855 2013-08-14  Xinliang David Li  <davidxl@google.com>
11857         * config/i386/i386.c (ix86_option_override_internal):
11858         Remove unused variable and field.
11860 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11862         PR target/57949
11863         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
11864         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
11865         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
11866         and Linux, correct BLKmode alignment when 128-bit alignment is
11867         required and compatibility flag is not set.
11868         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
11869         for zero-size arguments when compatibility flag is not set.
11871 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11873         PR tree-optimization/58145
11874         * tree-sra.c (build_ref_for_offset): If prev_base has
11875         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
11877 2013-08-14  Xinliang David Li  <davidxl@google.com>
11879         * config/i386/i386.c (ix86_option_override_internal):
11880         Fix uninitialized variable error.
11882 2013-08-14  Xinliang David Li  <davidxl@google.com>
11884         * config/i386/i386.opt: Define two new options.
11885         * config/i386/x86-tune.def: Add arch selector field in macros.
11886         * config/i386/i386.h: Adjust macro definition.
11887         * config/i386/i386.c (ix86_option_override_internal):
11888         Refactor the code.
11889         (parse_mtune_ctrl_str): New function.
11890         (set_ix86_tune_features): New function.
11891         (ix86_function_specific_restore): Call the new helper function.
11893 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
11895         PR rtl-optimization/57662
11896         * sel-sched.c (code_motion_process_successors): When the current insn
11897         is removed after the recursive traversal, break from the loop.
11898         Add comments and debug printouts.
11900 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11901             Alexandre Oliva  <aoliva@redhat.com>
11903         PR target/58067
11904         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
11905         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
11906         there also UNSPEC_PLTOFF.
11908 2013-08-14  Marek Polacek  <polacek@redhat.com>
11910         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
11911         NUM_CONDITIONS bit positions.
11913 2013-08-13  Cary Coutant  <ccoutant@google.com>
11915         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
11916         (attr_checksum): Hash vector contents instead of pointer.
11917         (attr_checksum_ordered): Likewise.
11919 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
11921         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
11922         when Pmode != word_mode.  Add length_address attribute.
11923         (sse3_monitor_<mode>): Merge from sse3_monitor and
11924         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
11925         Pmode != word_mode.  Update insn length attribute.
11926         * config/i386/i386.c (ix86_option_override_internal): Update
11927         ix86_gen_monitor selection for merged sse3_monitor insn.
11929 2013-08-13  Julian Brown  <julian@codesourcery.com>
11931         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
11932         perform invalid legitimization on greater-than-word-size modes for
11933         TARGET_E500_DOUBLE.
11935 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
11937         * ira.c (setup_class_translate_array): Use aclass instead of cl
11938         for classes not fully covered by allocno classes.
11940 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
11942         PR tree-optimization/57661
11943         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
11944         * tree-inline.c (tree_function_versioning): Initialize it.
11945         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
11946         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
11947         that is not being copied.
11949         PR sanitizer/56417
11950         * asan.c (instrument_strlen_call): Fix typo in comment.
11951         Use char * type even for the lhs of POINTER_PLUS_EXPR.
11953 2013-08-13  Steve Ellcey  <sellcey@mips.com>
11955         * config/mips/mips.md (prefetch): Use lw instead of ld on
11956         loongson in 32bit mode.
11958 2013-08-13  Nick Clifton  <nickc@redhat.com>
11960         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
11962 2013-08-13  Jan Hubicka  <jh@suse.cz>
11964         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
11965         introduced edge; fix typo in sanity check.
11966         (cgraph_resolve_speculation): Export; improve diagnostic.
11967         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
11968         speculation at type mismatch.
11969         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
11970         (cgraph_resolve_speculation): Declare.
11971         (symtab_can_be_discarded): New function.
11972         * value-prof.c (gimple_ic_transform): Remove actual transform code.
11973         * ipa-inline-transform.c (speculation_removed): New global var.
11974         (clone_inlined_nodes): See if speculation can be removed.
11975         (inline_call): If speculations was removed, we growths may not match.
11976         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
11977         (speculation_useful_p): New function.
11978         (resolve_noninline_speculation): New function.
11979         (inline_small_functions): Resolve useless speculations.
11980         * ipa-inline.h (speculation_useful_p): Declare
11981         * ipa.c (can_replace_by_local_alias): Simplify.
11982         (ipa_profile): Produce speculative calls in non-lto, too;
11983         add simple cost model; produce local aliases.
11985 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
11987         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
11988         PASS_MANAGER_H.
11990 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
11992         * config/i386/i386.c (ix86_function_versions): Use error + inform.
11994 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
11996         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
11997         iterator instead of X87MODEF.
11999 2013-08-12  Perez Read  <netfirewall@gmail.com>
12001         PR target/58132
12002         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
12003         operand 0 for intel asm alternative.
12004         (*movabs<mode>_2): Ditto for operand 1.
12006 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
12008         * config/aarch64/arm_none.h
12009         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
12011 2013-08-12  Nick Clifton  <nickc@redhat.com>
12013         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
12015 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
12017         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
12018         expand for QI/HImode operand to produce more effictive code for
12019         unsigned char(short) --> float(double) conversion.
12021 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
12023         * doc/invoke.texi: Mention that -ftls-model does not force the final
12024         model.
12026 2013-08-12  Marek Polacek  <polacek@redhat.com>
12027             Marc Glisse  <marc.glisse@inria.fr>
12029         PR tree-optimization/57980
12030         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
12031         when creating -1 constant.
12033 2013-08-10  Jan Hubicka  <jh@suse.cz>
12035         Workaround binutils PR14342.
12036         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
12037         (gimple_init_edge_profiler): Likewise.
12038         (gimple_gen_ic_func_profiler): Likewise.
12040 2013-08-09  Jan Hubicka  <jh@suse.cz>
12042         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
12044 2013-08-09  Xinliang David Li  <davidxl@google.com>
12046         * config/i386/stringop.def: New file.
12047         * config/i386/stringop.opt: New file.
12048         * config/i386/i386-opts.h: Include stringopt.def.
12049         * config/i386/i386.opt: Include stringopt.opt.
12050         * config/i386/i386.c (ix86_option_override_internal):
12051         Override default size based stringop inline strategies with options.
12052         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
12053         New function.
12055 2013-08-09  Jan Hubicka  <jh@suse.cz>
12057         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
12059 2013-08-09  Jan Hubicka  <jh@suse.cz>
12061         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
12062         CGRAPH_FREQ_MAX.
12063         (dump_cgraph_node): Dump profile-id.
12064         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
12065         and common_target_probability.
12066         * lto-cgraph.c (lto_output_edge): Stream common targets.
12067         (lto_output_node): Stream profile ids.
12068         (input_node): Stream profile ids.
12069         (input_edge): Stream common targets.
12070         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
12071         * ipa.c: Include value-prof.h
12072         (ipa_profile_generate_summary): Turn indirect call statement histograms
12073         into common targets.
12074         (ipa_profile): Turn common targets into speculative edges.
12076 2013-08-09  Jan Hubicka  <jh@suse.cz>
12078         * cgraph.h (cgraph_node): Add profile_id.
12079         * value-prof.c (cgraph_node_map): Turn into pointer_map.
12080         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
12081         (del_node_map): Update.
12082         (find_func_by_funcdef_no): Replace by ...
12083         (find_func_by_profile_id): ... this one.
12084         (gimple_ic_transform): Do not remove useful histograms when
12085         speculation is not done; dump info when indirect call removal
12086         can happen at LTO.
12087         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
12088         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
12089         (__gcov_indirect_call_profiler_v2): .. this one.
12090         * profile.h (init_node_map): Update.
12091         * coverage.c (coverage_compute_profile_id): New function.
12092         * coverage.h (coverage_compute_profile_id): Declare.
12093         * tree-profile.c (init_ic_make_global_vars): Make
12094         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
12095         (gimple_init_edge_profiler): Update prototype of
12096         __gcov_indirect_call_profiler.
12097         (gimple_gen_ic_func_profiler): Simplify.
12098         (tree_profiling): Use init_node_map
12100 2013-08-09  Jan Hubicka  <jh@suse.cz>
12102         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
12103         non-speculative refs.
12104         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
12105         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
12106         (cgraph_set_call_stmt): Likewise.
12107         (cgraph_create_edge_1): Fix release checking compilatoin;
12108         clear lto_stmt_uid.
12109         (cgraph_free_edge): Free indirect info.
12110         (cgraph_turn_edge_to_speculative): New function.
12111         (cgraph_speculative_call_info): New function.
12112         (cgraph_make_edge_direct): Return direct edge; handle speculation.
12113         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
12114         (dump_cgraph_node): Dump speculation.
12115         (verify_edge_count_and_frequency): Accept speculative edges.
12116         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
12117         (verify_cgraph_node): Handle speculation.
12118         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
12119         (cgraph_set_call_stmt): Update prototype.
12120         (cgraph_make_edge_direct): Update prototype.
12121         (cgraph_speculative_call_info): Declare.
12122         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
12123         to change; update call of ipa_find_references.
12124         * ipa-ref.c (ipa_record_reference): Fix return value; clear
12125         lto_stmt_uid and speculative flags.
12126         (ipa_dump_references): Dump speculation.
12127         (ipa_clone_references): Clone speculative flag.
12128         (ipa_clone_referring): Likewise.
12129         (ipa_clone_ref): New function.
12130         (ipa_find_reference): Look into lto_stmt_uids
12131         (ipa_clear_stmts_in_references): Do not clear speculative calls.
12132         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
12133         (ipa_find_reference): Update declaration.
12134         (ipa_clone_ref): Declare.
12135         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
12136         stream speculative flag.
12137         (lto_output_ref): Stream statements uids and speculation.
12138         (input_ref): Likewise.
12139         (input_edge): Stream speuclation.
12140         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
12141         (cgraph_set_call_stmt_including_clones): Handle speculation.
12142         * ipa-inline.c (heap_edge_removal_hook): New function.
12143         (inline_small_functions): Register it.
12144         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
12145         also initialize refs.
12146         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
12147         edge to change.
12148         (try_make_edge_direct_simple_call): Likewise.
12149         (try_make_edge_direct_simple_call): Likewise.
12150         (update_indirect_edges_after_inlining): Likewise.
12151         (remove_described_reference): Look proper lto_stmt_uid.
12152         (propagate_controlled_uses): Likewise.
12153         (propagate_controlled_uses): Liekwise.
12154         * tree-inline.c (copy_bb): Copy speculative edges.
12155         (redirect_all_calls): New function.
12156         (copy_cfg_body): Do redirection after loop info is updated.
12157         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
12159 2013-08-09  Jan Hubicka  <jh@suse.cz>
12161         * lto-streamer-out.c (output_function): Renumber PHIs.
12162         * lto-streamer-in.c (input_function): Likewise.
12164 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
12166         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
12167         (get_lane_unsigned): Likewise.
12168         (dup_lane_scalar): Likewise.
12169         (get_lane): enable for VALL.
12170         * config/aarch64/aarch64-simd.md
12171         (aarch64_dup_lane_scalar<mode>): Remove.
12172         (aarch64_get_lane_signed<mode>): Likewise.
12173         (aarch64_get_lane_unsigned<mode>): Likewise.
12174         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
12175         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
12176         (aarch64_get_lane<mode>): Enable for all vector modes.
12177         (aarch64_get_lanedi): Remove misleading constraints.
12178         * config/aarch64/arm_neon.h
12179         (__aarch64_vget_lane_any): Define.
12180         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
12181         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
12182         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
12183         * config/aarch64/iterators.md (VDQQH): New.
12184         (VDQQHS): Likewise.
12185         (vwcore): Likewise.
12187 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
12189         * configure.ac: Add GAS check for LEON instructions on SPARC.
12190         * configure: Regenerate.
12191         * config.in: Likewise.
12192         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
12193         sparc*-*-* block.
12194         * config/sparc/sparc.opt (LEON, LEON3): New masks.
12195         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
12196         for LEON or LEON3.
12197         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
12198         (AS_LEON_FLAG): New macro.
12199         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
12200         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
12201         Deal with LEON and LEON3 for the memory model.
12202         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
12203         (atomic_compare_and_swap<mode>_1): Likewise.
12204         (*atomic_compare_and_swap<mode>_1): Likewise.
12206 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
12208         * config/arm/neon.md (vcond): Fix floating-point vector
12209         comparisons against 0.
12211 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
12213         * lra-constraints.c (emit_spill_move): Remove assert.
12214         (process_alt_operands): Add more debugging
12215         output.  Increase reject for spilling into memory.  Decrease
12216         reject for reloading scratch.
12217         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
12219 2013-08-08  Steve Ellcey  <sellcey@mips.com>
12221         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
12222         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
12223         micromips incompatible.  Add nan2008.
12224         (MULTILIB_DIRNAMES): Add nan2008.
12225         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12226         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
12227         and micromips incompatible.  Add nan2008.
12228         (MULTILIB_DIRNAMES): Add nan2008.
12229         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12231 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
12233         PR rtl-optimization/58079
12234         * combine.c (combine_simplify_rtx): Avoid using SUBST if
12235         simplify_comparison has widened a comparison with an integer.
12237 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12239         * config/arm/neon.md (movmisalign<mode>): Disable when we
12240         don't allow unaligned accesses.
12241         (*movmisalign<mode>_neon_store): Likewise.
12242         (*movmisalign<mode>_neon_load): Likewise.
12243         (*movmisalign<mode>_neon_store): Likewise.
12244         (*movmisalign<mode>_neon_load): Likewise.
12246 2013-08-08  Jan Hubicka  <jh@suse.cz>
12248         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
12249         (make_pass_rebuild_cgraph_edges): Also clear references.
12250         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
12251         * ipa-inline-transform.c (inline_transform): Remove all references
12252         after inlining.
12253         * cgraphunit.c (expand_function): Remove all references after
12254         expansion.
12255         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
12256         (ipa_find_reference): Rewrite to iterator.
12257         (remove_stmt_references): Likewise.
12258         (ipa_clear_stmts_in_references): New function.
12259         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
12260         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
12261         clear references.
12262         * ipa-split.c (split_function): Remove references in split function.
12264 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12266         PR target/57431
12267         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
12268         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
12270 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12272         PR target/56979
12273         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
12274         suggested mode for the assignment isn't compatible with the
12275         registers required.
12277 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12279         PR target/58065
12280         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
12282 2013-08-07  Xinliang David Li  <davidxl@google.com>
12284         * config/i386/i386.opt: New option -mtune-ctrl=.
12285         * config/i386/x86-tune.def: New file.
12286         * config/i386/i386.h: include x86-tune.def.
12287         * config/i386/i386.c (ix86_option_override_internal):
12288         Parsing -mtune-ctrl= option and set tune features.
12290 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
12292         PR other/12081
12293         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
12294         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
12295         to gen_2arg_fn_t.
12297 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12299         * rtl.h (update_alignments): Declare.
12300         * final.c (grow_label_align): New function extracted from...
12301         (shorten_branches): ...here.  Call it.
12302         (update_alignments): New function.
12303         * reorg.c (sibling_labels): New variable.
12304         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
12305         the new label along with it onto the sibling_labels vector.
12306         (fill_simple_delay_slots): Adjust call to get_label_before.
12307         (fill_slots_from_thread): Likewise.
12308         (relax_delay_slots): Likewise.
12309         (make_return_insns): Likewise.
12310         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
12312 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12314         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
12315         document its semantics.
12316         (diagnostic_report_diagnostic): Adjust accordingly.
12318 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12320         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
12321         (sparc_option_override): ...and port to new C++ pass API.
12322         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
12324 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
12326         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
12328 2013-08-06  Caroline Tice  <cmtice@google.com>
12330         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
12331         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
12332         * tree-pass.h: Add pass_vtable_verify.
12333         * varasm.c (assemble_variable): Add code to properly set the comdat
12334         section and name for the .vtable_map_vars section.
12335         (assemble_vtyv_preinit_initializer): New function.
12336         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
12337         LINK_ONCE flag.
12338         * output.h: Add function decl for assemble_vtv_preinit_initializer.
12339         * vtable-verify.c: New file.
12340         * vtable-verify.h: New file.
12341         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
12342         initialiation levels.
12343         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
12344         * passes.def: Insert pass_vtable_verify.
12345         * aclocal.m4: Reorder includes.
12346         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
12347         -fvtv-counts options.
12348         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
12349         as appropriate, if -fvtable-verify=... is used.
12350         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
12351         -fvtable-verify=... is used.
12352         * Makefile.in (OBJS):  Add vtable-verify.o to list.
12353         (vtable-verify.o): Add new build rule.
12354         (GTFILES): Add vtable-verify.c to list.
12355         * common.opt (fvtable-verify=): New flag.
12356         (vtv_priority): Values for fvtable-verify= flag.
12357         (fvtv-counts): New flag.
12358         (fvtv-debug): New flag.
12359         * tree.h (save_vtable_map_decl): New extern function decl.
12361 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12363         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
12364         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
12365         (pass_data_rl78_devirt): ...new pass_data instance and...
12366         (make_pass_rl78_devirt): ...new function.
12367         (rl78_asm_file_start): Port pass registration to new C++ API.
12369 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12371         * coretypes.h (rtl_opt_pass): Add.
12372         (gcc::context): Add.
12373         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
12374         (epiphany_init): Port to new C++ pass API.
12375         (epiphany_optimize_mode_switching): Likewise.
12376         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
12377         (pass_manager::get_pass_mode_switching): New.
12378         (pass_manager::get_pass_peephole2): New.
12379         * mode-switching.c (pass_mode_switching): Add clone method.
12380         * recog.c (pass_peephole2): Add clone method.
12381         (pass_split_all_insns): Add clone method.
12383 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
12385         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
12386         (mips_option_override): ...here, porting to new C++ API for passes.
12388 2013-08-06  Jan Hubicka  <jh@suse.cz>
12390         * cgraph.c (cgraph_get_body): New function based on lto.c
12391         implementation.
12392         * cgraph.h (cgraph_get_body): Declare.
12393         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
12394         LTO paths.
12395         * cgraphunit.c (expand_function): Get body prior expanding.
12396         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
12397         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
12398         really need.
12399         * passes.c (do_per_function_toporder): Get body.
12400         * tree-inline.c (expand_call_inline): Get body prior inlining it.
12401         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
12403 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12405         PR fortran/57987
12406         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
12407         is not re-finalized.  Rename second parameter to no_collect.
12409 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12411         PR middle-end/58041
12412         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
12413         MEM_REF has proper alignment information.
12415 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
12417         PR other/12081
12418         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
12419         class insn_gen_fn.
12420         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
12421         rtx (*) (rtx, ...) with insn_gen_fn.
12422         * genoutput.c (output_insn_data): Cast gen_? function pointers to
12423         insn_gen_fn::stored_funcptr.  Add initializer braces.
12425 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12427         Rewrite how instances of passes are cloned to remove assumptions
12428         about their sizes (thus allowing pass subclasses to have
12429         additional data fields, albeit non-GC-managed ones at this point).
12431         * passes.c (make_pass_instance): Now that passes have clone
12432         methods, rewrite this function to eliminate XNEW and memcpy
12433         calls that used hardcoded sizes.  Since this function no longer
12434         creates pass instances, rename it to...
12435         (add_pass_instance): ...this.  Document the old way that passes were
12436         numbered and flagged, and rework this function to continue using it.
12437         (next_pass_1): Add an initial_pass argument for use by
12438         add_pass_instance.
12439         (position_pass): When adding multiple instances of a pass, use
12440         the pass's clone method, rather than relying on the XNEW/memcpy
12441         within the former make_pass_instance (now add_pass_instance).
12442         (pass_manager::pass_manager): When invoking next_pass_1, also supply
12443         the initial instance of the current pass within the pass manager.
12445 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12447         This is the automated part of the conversion of passes from C
12448         structs to C++ classes.
12450         Patch autogenerated by refactor_passes.py from
12451         https://github.com/davidmalcolm/gcc-refactoring-scripts
12452         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
12454         * asan.c (pass_asan): Convert from a global struct to a subclass of
12455         gimple_opt_pass along with...
12456         (pass_data_asan): ...new pass_data instance and...
12457         (make_pass_asan): ...new function.
12458         (pass_asan_O0): Convert from a global struct to a subclass of
12459         gimple_opt_pass along with...
12460         (pass_data_asan_O0): ...new pass_data instance and...
12461         (make_pass_asan_O0): ...new function.
12462         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
12463         subclass of rtl_opt_pass along with...
12464         (pass_data_inc_dec): ...new pass_data instance and...
12465         (make_pass_inc_dec): ...new function.
12466         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
12467         a subclass of rtl_opt_pass along with...
12468         (pass_data_reorder_blocks): ...new pass_data instance and...
12469         (make_pass_reorder_blocks): ...new function.
12470         (pass_duplicate_computed_gotos): Convert from a global struct to a
12471         subclass of rtl_opt_pass along with...
12472         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
12473         (make_pass_duplicate_computed_gotos): ...new function.
12474         (pass_partition_blocks): Convert from a global struct to a subclass of
12475         rtl_opt_pass along with...
12476         (pass_data_partition_blocks): ...new pass_data instance and...
12477         (make_pass_partition_blocks): ...new function.
12478         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
12479         struct to a subclass of rtl_opt_pass along with...
12480         (pass_data_branch_target_load_optimize1): ...new pass_data instance
12481         and...
12482         (make_pass_branch_target_load_optimize1): ...new function.
12483         (pass_branch_target_load_optimize2): Convert from a global struct to a
12484         subclass of rtl_opt_pass along with...
12485         (pass_data_branch_target_load_optimize2): ...new pass_data instance
12486         and...
12487         (make_pass_branch_target_load_optimize2): ...new function.
12488         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
12489         of rtl_opt_pass along with...
12490         (pass_data_jump): ...new pass_data instance and...
12491         (make_pass_jump): ...new function.
12492         (pass_jump2): Convert from a global struct to a subclass of
12493         rtl_opt_pass along with...
12494         (pass_data_jump2): ...new pass_data instance and...
12495         (make_pass_jump2): ...new function.
12496         * cfgexpand.c (pass_expand): Convert from a global struct to a
12497         subclass of rtl_opt_pass along with...
12498         (pass_data_expand): ...new pass_data instance and...
12499         (make_pass_expand): ...new function.
12500         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
12501         of rtl_opt_pass along with...
12502         (pass_data_free_cfg): ...new pass_data instance and...
12503         (make_pass_free_cfg): ...new function.
12504         (pass_into_cfg_layout_mode): Convert from a global struct to a
12505         subclass of rtl_opt_pass along with...
12506         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
12507         (make_pass_into_cfg_layout_mode): ...new function.
12508         (pass_outof_cfg_layout_mode): Convert from a global struct to a
12509         subclass of rtl_opt_pass along with...
12510         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
12511         (make_pass_outof_cfg_layout_mode): ...new function.
12512         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
12513         struct to a subclass of gimple_opt_pass along with...
12514         (pass_data_build_cgraph_edges): ...new pass_data instance and...
12515         (make_pass_build_cgraph_edges): ...new function.
12516         (pass_rebuild_cgraph_edges): Convert from a global struct to a
12517         subclass of gimple_opt_pass along with...
12518         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
12519         (make_pass_rebuild_cgraph_edges): ...new function.
12520         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
12521         subclass of gimple_opt_pass along with...
12522         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
12523         and...
12524         (make_pass_remove_cgraph_callee_edges): ...new function.
12525         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
12526         struct to a subclass of rtl_opt_pass along with...
12527         (pass_data_stack_adjustments): ...new pass_data instance and...
12528         (make_pass_stack_adjustments): ...new function.
12529         * combine.c (pass_combine): Convert from a global struct to a subclass
12530         of rtl_opt_pass along with...
12531         (pass_data_combine): ...new pass_data instance and...
12532         (make_pass_combine): ...new function.
12533         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
12534         global struct to a subclass of rtl_opt_pass along with...
12535         (pass_data_compare_elim_after_reload): ...new pass_data instance
12536         and...
12537         (make_pass_compare_elim_after_reload): ...new function.
12538         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
12539         of rtl_opt_pass along with...
12540         (pass_data_rtl_cprop): ...new pass_data instance and...
12541         (make_pass_rtl_cprop): ...new function.
12542         * cse.c (pass_cse): Convert from a global struct to a subclass of
12543         rtl_opt_pass along with...
12544         (pass_data_cse): ...new pass_data instance and...
12545         (make_pass_cse): ...new function.
12546         (pass_cse2): Convert from a global struct to a subclass of
12547         rtl_opt_pass along with...
12548         (pass_data_cse2): ...new pass_data instance and...
12549         (make_pass_cse2): ...new function.
12550         (pass_cse_after_global_opts): Convert from a global struct to a
12551         subclass of rtl_opt_pass along with...
12552         (pass_data_cse_after_global_opts): ...new pass_data instance and...
12553         (make_pass_cse_after_global_opts): ...new function.
12554         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
12555         of rtl_opt_pass along with...
12556         (pass_data_ud_rtl_dce): ...new pass_data instance and...
12557         (make_pass_ud_rtl_dce): ...new function.
12558         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
12559         rtl_opt_pass along with...
12560         (pass_data_fast_rtl_dce): ...new pass_data instance and...
12561         (make_pass_fast_rtl_dce): ...new function.
12562         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
12563         a subclass of rtl_opt_pass along with...
12564         (pass_data_df_initialize_opt): ...new pass_data instance and...
12565         (make_pass_df_initialize_opt): ...new function.
12566         (pass_df_initialize_no_opt): Convert from a global struct to a
12567         subclass of rtl_opt_pass along with...
12568         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
12569         (make_pass_df_initialize_no_opt): ...new function.
12570         (pass_df_finish): Convert from a global struct to a subclass of
12571         rtl_opt_pass along with...
12572         (pass_data_df_finish): ...new pass_data instance and...
12573         (make_pass_df_finish): ...new function.
12574         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
12575         rtl_opt_pass along with...
12576         (pass_data_rtl_dse1): ...new pass_data instance and...
12577         (make_pass_rtl_dse1): ...new function.
12578         (pass_rtl_dse2): Convert from a global struct to a subclass of
12579         rtl_opt_pass along with...
12580         (pass_data_rtl_dse2): ...new pass_data instance and...
12581         (make_pass_rtl_dse2): ...new function.
12582         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
12583         subclass of rtl_opt_pass along with...
12584         (pass_data_dwarf2_frame): ...new pass_data instance and...
12585         (make_pass_dwarf2_frame): ...new function.
12586         * except.c (pass_set_nothrow_function_flags): Convert from a global
12587         struct to a subclass of rtl_opt_pass along with...
12588         (pass_data_set_nothrow_function_flags): ...new pass_data instance
12589         and...
12590         (make_pass_set_nothrow_function_flags): ...new function.
12591         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
12592         subclass of rtl_opt_pass along with...
12593         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
12594         and...
12595         (make_pass_convert_to_eh_region_ranges): ...new function.
12596         * final.c (pass_compute_alignments): Convert from a global struct to a
12597         subclass of rtl_opt_pass along with...
12598         (pass_data_compute_alignments): ...new pass_data instance and...
12599         (make_pass_compute_alignments): ...new function.
12600         (pass_final): Convert from a global struct to a subclass of
12601         rtl_opt_pass along with...
12602         (pass_data_final): ...new pass_data instance and...
12603         (make_pass_final): ...new function.
12604         (pass_shorten_branches): Convert from a global struct to a subclass of
12605         rtl_opt_pass along with...
12606         (pass_data_shorten_branches): ...new pass_data instance and...
12607         (make_pass_shorten_branches): ...new function.
12608         (pass_clean_state): Convert from a global struct to a subclass of
12609         rtl_opt_pass along with...
12610         (pass_data_clean_state): ...new pass_data instance and...
12611         (make_pass_clean_state): ...new function.
12612         * function.c (pass_instantiate_virtual_regs): Convert from a global
12613         struct to a subclass of rtl_opt_pass along with...
12614         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
12615         (make_pass_instantiate_virtual_regs): ...new function.
12616         (pass_leaf_regs): Convert from a global struct to a subclass of
12617         rtl_opt_pass along with...
12618         (pass_data_leaf_regs): ...new pass_data instance and...
12619         (make_pass_leaf_regs): ...new function.
12620         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
12621         subclass of rtl_opt_pass along with...
12622         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
12623         and...
12624         (make_pass_thread_prologue_and_epilogue): ...new function.
12625         (pass_match_asm_constraints): Convert from a global struct to a
12626         subclass of rtl_opt_pass along with...
12627         (pass_data_match_asm_constraints): ...new pass_data instance and...
12628         (make_pass_match_asm_constraints): ...new function.
12629         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
12630         subclass of rtl_opt_pass along with...
12631         (pass_data_rtl_fwprop): ...new pass_data instance and...
12632         (make_pass_rtl_fwprop): ...new function.
12633         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
12634         rtl_opt_pass along with...
12635         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
12636         (make_pass_rtl_fwprop_addr): ...new function.
12637         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
12638         rtl_opt_pass along with...
12639         (pass_data_rtl_pre): ...new pass_data instance and...
12640         (make_pass_rtl_pre): ...new function.
12641         (pass_rtl_hoist): Convert from a global struct to a subclass of
12642         rtl_opt_pass along with...
12643         (pass_data_rtl_hoist): ...new pass_data instance and...
12644         (make_pass_rtl_hoist): ...new function.
12645         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
12646         subclass of gimple_opt_pass along with...
12647         (pass_data_lower_cf): ...new pass_data instance and...
12648         (make_pass_lower_cf): ...new function.
12649         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
12650         from a global struct to a subclass of gimple_opt_pass along with...
12651         (pass_data_strength_reduction): ...new pass_data instance and...
12652         (make_pass_strength_reduction): ...new function.
12653         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
12654         of rtl_opt_pass along with...
12655         (pass_data_rtl_ifcvt): ...new pass_data instance and...
12656         (make_pass_rtl_ifcvt): ...new function.
12657         (pass_if_after_combine): Convert from a global struct to a subclass of
12658         rtl_opt_pass along with...
12659         (pass_data_if_after_combine): ...new pass_data instance and...
12660         (make_pass_if_after_combine): ...new function.
12661         (pass_if_after_reload): Convert from a global struct to a subclass of
12662         rtl_opt_pass along with...
12663         (pass_data_if_after_reload): ...new pass_data instance and...
12664         (make_pass_if_after_reload): ...new function.
12665         * init-regs.c (pass_initialize_regs): Convert from a global struct to
12666         a subclass of rtl_opt_pass along with...
12667         (pass_data_initialize_regs): ...new pass_data instance and...
12668         (make_pass_initialize_regs): ...new function.
12669         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
12670         of ipa_opt_pass_d along with...
12671         (pass_data_ipa_cp): ...new pass_data instance and...
12672         (make_pass_ipa_cp): ...new function.
12673         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
12674         global struct to a subclass of gimple_opt_pass along with...
12675         (pass_data_inline_parameters): ...new pass_data instance and...
12676         (make_pass_inline_parameters): ...new function.
12677         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
12678         subclass of gimple_opt_pass along with...
12679         (pass_data_early_inline): ...new pass_data instance and...
12680         (make_pass_early_inline): ...new function.
12681         (pass_ipa_inline): Convert from a global struct to a subclass of
12682         ipa_opt_pass_d along with...
12683         (pass_data_ipa_inline): ...new pass_data instance and...
12684         (make_pass_ipa_inline): ...new function.
12685         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
12686         struct to a subclass of gimple_opt_pass along with...
12687         (pass_data_local_pure_const): ...new pass_data instance and...
12688         (make_pass_local_pure_const): ...new function.
12689         (pass_ipa_pure_const): Convert from a global struct to a subclass of
12690         ipa_opt_pass_d along with...
12691         (pass_data_ipa_pure_const): ...new pass_data instance and...
12692         (make_pass_ipa_pure_const): ...new function.
12693         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
12694         to a subclass of ipa_opt_pass_d along with...
12695         (pass_data_ipa_reference): ...new pass_data instance and...
12696         (make_pass_ipa_reference): ...new function.
12697         * ipa-split.c (pass_split_functions): Convert from a global struct to
12698         a subclass of gimple_opt_pass along with...
12699         (pass_data_split_functions): ...new pass_data instance and...
12700         (make_pass_split_functions): ...new function.
12701         (pass_feedback_split_functions): Convert from a global struct to a
12702         subclass of gimple_opt_pass along with...
12703         (pass_data_feedback_split_functions): ...new pass_data instance and...
12704         (make_pass_feedback_split_functions): ...new function.
12705         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
12706         global struct to a subclass of simple_ipa_opt_pass along with...
12707         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
12708         instance and...
12709         (make_pass_ipa_function_and_variable_visibility): ...new function.
12710         (pass_ipa_free_inline_summary): Convert from a global struct to a
12711         subclass of simple_ipa_opt_pass along with...
12712         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
12713         (make_pass_ipa_free_inline_summary): ...new function.
12714         (pass_ipa_whole_program_visibility): Convert from a global struct to a
12715         subclass of ipa_opt_pass_d along with...
12716         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
12717         and...
12718         (make_pass_ipa_whole_program_visibility): ...new function.
12719         (pass_ipa_profile): Convert from a global struct to a subclass of
12720         ipa_opt_pass_d along with...
12721         (pass_data_ipa_profile): ...new pass_data instance and...
12722         (make_pass_ipa_profile): ...new function.
12723         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
12724         ipa_opt_pass_d along with...
12725         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
12726         (make_pass_ipa_cdtor_merge): ...new function.
12727         * ira.c (pass_ira): Convert from a global struct to a subclass of
12728         rtl_opt_pass along with...
12729         (pass_data_ira): ...new pass_data instance and...
12730         (make_pass_ira): ...new function.
12731         (pass_reload): Convert from a global struct to a subclass of
12732         rtl_opt_pass along with...
12733         (pass_data_reload): ...new pass_data instance and...
12734         (make_pass_reload): ...new function.
12735         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
12736         subclass of rtl_opt_pass along with...
12737         (pass_data_cleanup_barriers): ...new pass_data instance and...
12738         (make_pass_cleanup_barriers): ...new function.
12739         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
12740         of rtl_opt_pass along with...
12741         (pass_data_loop2): ...new pass_data instance and...
12742         (make_pass_loop2): ...new function.
12743         (pass_rtl_loop_init): Convert from a global struct to a subclass of
12744         rtl_opt_pass along with...
12745         (pass_data_rtl_loop_init): ...new pass_data instance and...
12746         (make_pass_rtl_loop_init): ...new function.
12747         (pass_rtl_loop_done): Convert from a global struct to a subclass of
12748         rtl_opt_pass along with...
12749         (pass_data_rtl_loop_done): ...new pass_data instance and...
12750         (make_pass_rtl_loop_done): ...new function.
12751         (pass_rtl_move_loop_invariants): Convert from a global struct to a
12752         subclass of rtl_opt_pass along with...
12753         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
12754         (make_pass_rtl_move_loop_invariants): ...new function.
12755         (pass_rtl_unswitch): Convert from a global struct to a subclass of
12756         rtl_opt_pass along with...
12757         (pass_data_rtl_unswitch): ...new pass_data instance and...
12758         (make_pass_rtl_unswitch): ...new function.
12759         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
12760         subclass of rtl_opt_pass along with...
12761         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
12762         and...
12763         (make_pass_rtl_unroll_and_peel_loops): ...new function.
12764         (pass_rtl_doloop): Convert from a global struct to a subclass of
12765         rtl_opt_pass along with...
12766         (pass_data_rtl_doloop): ...new pass_data instance and...
12767         (make_pass_rtl_doloop): ...new function.
12768         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
12769         a subclass of rtl_opt_pass along with...
12770         (pass_data_lower_subreg): ...new pass_data instance and...
12771         (make_pass_lower_subreg): ...new function.
12772         (pass_lower_subreg2): Convert from a global struct to a subclass of
12773         rtl_opt_pass along with...
12774         (pass_data_lower_subreg2): ...new pass_data instance and...
12775         (make_pass_lower_subreg2): ...new function.
12776         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
12777         struct to a subclass of ipa_opt_pass_d along with...
12778         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
12779         (make_pass_ipa_lto_gimple_out): ...new function.
12780         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
12781         of ipa_opt_pass_d along with...
12782         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
12783         (make_pass_ipa_lto_finish_out): ...new function.
12784         * mode-switching.c (pass_mode_switching): Convert from a global struct
12785         to a subclass of rtl_opt_pass along with...
12786         (pass_data_mode_switching): ...new pass_data instance and...
12787         (make_pass_mode_switching): ...new function.
12788         * modulo-sched.c (pass_sms): Convert from a global struct to a
12789         subclass of rtl_opt_pass along with...
12790         (pass_data_sms): ...new pass_data instance and...
12791         (make_pass_sms): ...new function.
12792         * omp-low.c (pass_expand_omp): Convert from a global struct to a
12793         subclass of gimple_opt_pass along with...
12794         (pass_data_expand_omp): ...new pass_data instance and...
12795         (make_pass_expand_omp): ...new function.
12796         (pass_lower_omp): Convert from a global struct to a subclass of
12797         gimple_opt_pass along with...
12798         (pass_data_lower_omp): ...new pass_data instance and...
12799         (make_pass_lower_omp): ...new function.
12800         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
12801         of gimple_opt_pass along with...
12802         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
12803         (make_pass_diagnose_omp_blocks): ...new function.
12804         * passes.c (pass_early_local_passes): Convert from a global struct to
12805         a subclass of simple_ipa_opt_pass along with...
12806         (pass_data_early_local_passes): ...new pass_data instance and...
12807         (make_pass_early_local_passes): ...new function.
12808         (pass_all_early_optimizations): Convert from a global struct to a
12809         subclass of gimple_opt_pass along with...
12810         (pass_data_all_early_optimizations): ...new pass_data instance and...
12811         (make_pass_all_early_optimizations): ...new function.
12812         (pass_all_optimizations): Convert from a global struct to a subclass
12813         of gimple_opt_pass along with...
12814         (pass_data_all_optimizations): ...new pass_data instance and...
12815         (make_pass_all_optimizations): ...new function.
12816         (pass_all_optimizations_g): Convert from a global struct to a subclass
12817         of gimple_opt_pass along with...
12818         (pass_data_all_optimizations_g): ...new pass_data instance and...
12819         (make_pass_all_optimizations_g): ...new function.
12820         (pass_rest_of_compilation): Convert from a global struct to a subclass
12821         of rtl_opt_pass along with...
12822         (pass_data_rest_of_compilation): ...new pass_data instance and...
12823         (make_pass_rest_of_compilation): ...new function.
12824         (pass_postreload): Convert from a global struct to a subclass of
12825         rtl_opt_pass along with...
12826         (pass_data_postreload): ...new pass_data instance and...
12827         (make_pass_postreload): ...new function.
12828         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
12829         subclass of rtl_opt_pass along with...
12830         (pass_data_gcse2): ...new pass_data instance and...
12831         (make_pass_gcse2): ...new function.
12832         * postreload.c (pass_postreload_cse): Convert from a global struct to
12833         a subclass of rtl_opt_pass along with...
12834         (pass_data_postreload_cse): ...new pass_data instance and...
12835         (make_pass_postreload_cse): ...new function.
12836         * predict.c (pass_profile): Convert from a global struct to a subclass
12837         of gimple_opt_pass along with...
12838         (pass_data_profile): ...new pass_data instance and...
12839         (make_pass_profile): ...new function.
12840         (pass_strip_predict_hints): Convert from a global struct to a subclass
12841         of gimple_opt_pass along with...
12842         (pass_data_strip_predict_hints): ...new pass_data instance and...
12843         (make_pass_strip_predict_hints): ...new function.
12844         * recog.c (pass_peephole2): Convert from a global struct to a subclass
12845         of rtl_opt_pass along with...
12846         (pass_data_peephole2): ...new pass_data instance and...
12847         (make_pass_peephole2): ...new function.
12848         (pass_split_all_insns): Convert from a global struct to a subclass of
12849         rtl_opt_pass along with...
12850         (pass_data_split_all_insns): ...new pass_data instance and...
12851         (make_pass_split_all_insns): ...new function.
12852         (pass_split_after_reload): Convert from a global struct to a subclass
12853         of rtl_opt_pass along with...
12854         (pass_data_split_after_reload): ...new pass_data instance and...
12855         (make_pass_split_after_reload): ...new function.
12856         (pass_split_before_regstack): Convert from a global struct to a
12857         subclass of rtl_opt_pass along with...
12858         (pass_data_split_before_regstack): ...new pass_data instance and...
12859         (make_pass_split_before_regstack): ...new function.
12860         (pass_split_before_sched2): Convert from a global struct to a subclass
12861         of rtl_opt_pass along with...
12862         (pass_data_split_before_sched2): ...new pass_data instance and...
12863         (make_pass_split_before_sched2): ...new function.
12864         (pass_split_for_shorten_branches): Convert from a global struct to a
12865         subclass of rtl_opt_pass along with...
12866         (pass_data_split_for_shorten_branches): ...new pass_data instance
12867         and...
12868         (make_pass_split_for_shorten_branches): ...new function.
12869         * ree.c (pass_ree): Convert from a global struct to a subclass of
12870         rtl_opt_pass along with...
12871         (pass_data_ree): ...new pass_data instance and...
12872         (make_pass_ree): ...new function.
12873         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
12874         subclass of rtl_opt_pass along with...
12875         (pass_data_stack_regs): ...new pass_data instance and...
12876         (make_pass_stack_regs): ...new function.
12877         (pass_stack_regs_run): Convert from a global struct to a subclass of
12878         rtl_opt_pass along with...
12879         (pass_data_stack_regs_run): ...new pass_data instance and...
12880         (make_pass_stack_regs_run): ...new function.
12881         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
12882         subclass of rtl_opt_pass along with...
12883         (pass_data_cprop_hardreg): ...new pass_data instance and...
12884         (make_pass_cprop_hardreg): ...new function.
12885         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
12886         subclass of rtl_opt_pass along with...
12887         (pass_data_reginfo_init): ...new pass_data instance and...
12888         (make_pass_reginfo_init): ...new function.
12889         * regmove.c (pass_regmove): Convert from a global struct to a subclass
12890         of rtl_opt_pass along with...
12891         (pass_data_regmove): ...new pass_data instance and...
12892         (make_pass_regmove): ...new function.
12893         * regrename.c (pass_regrename): Convert from a global struct to a
12894         subclass of rtl_opt_pass along with...
12895         (pass_data_regrename): ...new pass_data instance and...
12896         (make_pass_regrename): ...new function.
12897         * reorg.c (pass_delay_slots): Convert from a global struct to a
12898         subclass of rtl_opt_pass along with...
12899         (pass_data_delay_slots): ...new pass_data instance and...
12900         (make_pass_delay_slots): ...new function.
12901         (pass_machine_reorg): Convert from a global struct to a subclass of
12902         rtl_opt_pass along with...
12903         (pass_data_machine_reorg): ...new pass_data instance and...
12904         (make_pass_machine_reorg): ...new function.
12905         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
12906         of rtl_opt_pass along with...
12907         (pass_data_sched): ...new pass_data instance and...
12908         (make_pass_sched): ...new function.
12909         (pass_sched2): Convert from a global struct to a subclass of
12910         rtl_opt_pass along with...
12911         (pass_data_sched2): ...new pass_data instance and...
12912         (make_pass_sched2): ...new function.
12913         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
12914         to a subclass of rtl_opt_pass along with...
12915         (pass_data_stack_ptr_mod): ...new pass_data instance and...
12916         (make_pass_stack_ptr_mod): ...new function.
12917         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
12918         to a subclass of rtl_opt_pass along with...
12919         (pass_data_rtl_store_motion): ...new pass_data instance and...
12920         (make_pass_rtl_store_motion): ...new function.
12921         * tracer.c (pass_tracer): Convert from a global struct to a subclass
12922         of gimple_opt_pass along with...
12923         (pass_data_tracer): ...new pass_data instance and...
12924         (make_pass_tracer): ...new function.
12925         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
12926         to a subclass of gimple_opt_pass along with...
12927         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
12928         (make_pass_diagnose_tm_blocks): ...new function.
12929         (pass_lower_tm): Convert from a global struct to a subclass of
12930         gimple_opt_pass along with...
12931         (pass_data_lower_tm): ...new pass_data instance and...
12932         (make_pass_lower_tm): ...new function.
12933         (pass_tm_init): Convert from a global struct to a subclass of
12934         gimple_opt_pass along with...
12935         (pass_data_tm_init): ...new pass_data instance and...
12936         (make_pass_tm_init): ...new function.
12937         (pass_tm_mark): Convert from a global struct to a subclass of
12938         gimple_opt_pass along with...
12939         (pass_data_tm_mark): ...new pass_data instance and...
12940         (make_pass_tm_mark): ...new function.
12941         (pass_tm_edges): Convert from a global struct to a subclass of
12942         gimple_opt_pass along with...
12943         (pass_data_tm_edges): ...new pass_data instance and...
12944         (make_pass_tm_edges): ...new function.
12945         (pass_tm_memopt): Convert from a global struct to a subclass of
12946         gimple_opt_pass along with...
12947         (pass_data_tm_memopt): ...new pass_data instance and...
12948         (make_pass_tm_memopt): ...new function.
12949         (pass_ipa_tm): Convert from a global struct to a subclass of
12950         simple_ipa_opt_pass along with...
12951         (pass_data_ipa_tm): ...new pass_data instance and...
12952         (make_pass_ipa_tm): ...new function.
12953         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
12954         subclass of gimple_opt_pass along with...
12955         (pass_data_call_cdce): ...new pass_data instance and...
12956         (make_pass_call_cdce): ...new function.
12957         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
12958         subclass of gimple_opt_pass along with...
12959         (pass_data_build_cfg): ...new pass_data instance and...
12960         (make_pass_build_cfg): ...new function.
12961         (pass_split_crit_edges): Convert from a global struct to a subclass of
12962         gimple_opt_pass along with...
12963         (pass_data_split_crit_edges): ...new pass_data instance and...
12964         (make_pass_split_crit_edges): ...new function.
12965         (pass_warn_function_return): Convert from a global struct to a
12966         subclass of gimple_opt_pass along with...
12967         (pass_data_warn_function_return): ...new pass_data instance and...
12968         (make_pass_warn_function_return): ...new function.
12969         (pass_warn_function_noreturn): Convert from a global struct to a
12970         subclass of gimple_opt_pass along with...
12971         (pass_data_warn_function_noreturn): ...new pass_data instance and...
12972         (make_pass_warn_function_noreturn): ...new function.
12973         (pass_warn_unused_result): Convert from a global struct to a subclass
12974         of gimple_opt_pass along with...
12975         (pass_data_warn_unused_result): ...new pass_data instance and...
12976         (make_pass_warn_unused_result): ...new function.
12977         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
12978         a subclass of gimple_opt_pass along with...
12979         (pass_data_merge_phi): ...new pass_data instance and...
12980         (make_pass_merge_phi): ...new function.
12981         * tree-complex.c (pass_lower_complex): Convert from a global struct to
12982         a subclass of gimple_opt_pass along with...
12983         (pass_data_lower_complex): ...new pass_data instance and...
12984         (make_pass_lower_complex): ...new function.
12985         (pass_lower_complex_O0): Convert from a global struct to a subclass of
12986         gimple_opt_pass along with...
12987         (pass_data_lower_complex_O0): ...new pass_data instance and...
12988         (make_pass_lower_complex_O0): ...new function.
12989         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
12990         subclass of gimple_opt_pass along with...
12991         (pass_data_lower_eh): ...new pass_data instance and...
12992         (make_pass_lower_eh): ...new function.
12993         (pass_refactor_eh): Convert from a global struct to a subclass of
12994         gimple_opt_pass along with...
12995         (pass_data_refactor_eh): ...new pass_data instance and...
12996         (make_pass_refactor_eh): ...new function.
12997         (pass_lower_resx): Convert from a global struct to a subclass of
12998         gimple_opt_pass along with...
12999         (pass_data_lower_resx): ...new pass_data instance and...
13000         (make_pass_lower_resx): ...new function.
13001         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
13002         of gimple_opt_pass along with...
13003         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
13004         (make_pass_lower_eh_dispatch): ...new function.
13005         (pass_cleanup_eh): Convert from a global struct to a subclass of
13006         gimple_opt_pass along with...
13007         (pass_data_cleanup_eh): ...new pass_data instance and...
13008         (make_pass_cleanup_eh): ...new function.
13009         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
13010         to a subclass of simple_ipa_opt_pass along with...
13011         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
13012         (make_pass_ipa_lower_emutls): ...new function.
13013         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
13014         a subclass of gimple_opt_pass along with...
13015         (pass_data_if_conversion): ...new pass_data instance and...
13016         (make_pass_if_conversion): ...new function.
13017         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
13018         subclass of gimple_opt_pass along with...
13019         (pass_data_build_ssa): ...new pass_data instance and...
13020         (make_pass_build_ssa): ...new function.
13021         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
13022         global struct to a subclass of gimple_opt_pass along with...
13023         (pass_data_loop_distribution): ...new pass_data instance and...
13024         (make_pass_loop_distribution): ...new function.
13025         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
13026         subclass of gimple_opt_pass along with...
13027         (pass_data_mudflap_1): ...new pass_data instance and...
13028         (make_pass_mudflap_1): ...new function.
13029         (pass_mudflap_2): Convert from a global struct to a subclass of
13030         gimple_opt_pass along with...
13031         (pass_data_mudflap_2): ...new pass_data instance and...
13032         (make_pass_mudflap_2): ...new function.
13033         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
13034         subclass of gimple_opt_pass along with...
13035         (pass_data_mudflap_1): ...new pass_data instance and...
13036         (make_pass_mudflap_1): ...new function.
13037         (pass_mudflap_2): Convert from a global struct to a subclass of
13038         gimple_opt_pass along with...
13039         (pass_data_mudflap_2): ...new pass_data instance and...
13040         (make_pass_mudflap_2): ...new function.
13041         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
13042         gimple_opt_pass along with...
13043         (pass_data_nrv): ...new pass_data instance and...
13044         (make_pass_nrv): ...new function.
13045         (pass_return_slot): Convert from a global struct to a subclass of
13046         gimple_opt_pass along with...
13047         (pass_data_return_slot): ...new pass_data instance and...
13048         (make_pass_return_slot): ...new function.
13049         * tree-object-size.c (pass_object_sizes): Convert from a global struct
13050         to a subclass of gimple_opt_pass along with...
13051         (pass_data_object_sizes): ...new pass_data instance and...
13052         (make_pass_object_sizes): ...new function.
13053         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
13054         global struct to a subclass of gimple_opt_pass along with...
13055         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
13056         and...
13057         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13058         (pass_fixup_cfg): Convert from a global struct to a subclass of
13059         gimple_opt_pass along with...
13060         (pass_data_fixup_cfg): ...new pass_data instance and...
13061         (make_pass_fixup_cfg): ...new function.
13062         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
13063         (make_pass_mudflap_1): ...new function.
13064         (pass_mudflap_2): Replace declaration with that of...
13065         (make_pass_mudflap_2): ...new function.
13066         (pass_asan): Replace declaration with that of...
13067         (make_pass_asan): ...new function.
13068         (pass_asan_O0): Replace declaration with that of...
13069         (make_pass_asan_O0): ...new function.
13070         (pass_tsan): Replace declaration with that of...
13071         (make_pass_tsan): ...new function.
13072         (pass_tsan_O0): Replace declaration with that of...
13073         (make_pass_tsan_O0): ...new function.
13074         (pass_lower_cf): Replace declaration with that of...
13075         (make_pass_lower_cf): ...new function.
13076         (pass_refactor_eh): Replace declaration with that of...
13077         (make_pass_refactor_eh): ...new function.
13078         (pass_lower_eh): Replace declaration with that of...
13079         (make_pass_lower_eh): ...new function.
13080         (pass_lower_eh_dispatch): Replace declaration with that of...
13081         (make_pass_lower_eh_dispatch): ...new function.
13082         (pass_lower_resx): Replace declaration with that of...
13083         (make_pass_lower_resx): ...new function.
13084         (pass_build_cfg): Replace declaration with that of...
13085         (make_pass_build_cfg): ...new function.
13086         (pass_early_tree_profile): Replace declaration with that of...
13087         (make_pass_early_tree_profile): ...new function.
13088         (pass_cleanup_eh): Replace declaration with that of...
13089         (make_pass_cleanup_eh): ...new function.
13090         (pass_sra): Replace declaration with that of...
13091         (make_pass_sra): ...new function.
13092         (pass_sra_early): Replace declaration with that of...
13093         (make_pass_sra_early): ...new function.
13094         (pass_early_ipa_sra): Replace declaration with that of...
13095         (make_pass_early_ipa_sra): ...new function.
13096         (pass_tail_recursion): Replace declaration with that of...
13097         (make_pass_tail_recursion): ...new function.
13098         (pass_tail_calls): Replace declaration with that of...
13099         (make_pass_tail_calls): ...new function.
13100         (pass_tree_loop): Replace declaration with that of...
13101         (make_pass_tree_loop): ...new function.
13102         (pass_tree_loop_init): Replace declaration with that of...
13103         (make_pass_tree_loop_init): ...new function.
13104         (pass_lim): Replace declaration with that of...
13105         (make_pass_lim): ...new function.
13106         (pass_tree_unswitch): Replace declaration with that of...
13107         (make_pass_tree_unswitch): ...new function.
13108         (pass_predcom): Replace declaration with that of...
13109         (make_pass_predcom): ...new function.
13110         (pass_iv_canon): Replace declaration with that of...
13111         (make_pass_iv_canon): ...new function.
13112         (pass_scev_cprop): Replace declaration with that of...
13113         (make_pass_scev_cprop): ...new function.
13114         (pass_empty_loop): Replace declaration with that of...
13115         (make_pass_empty_loop): ...new function.
13116         (pass_record_bounds): Replace declaration with that of...
13117         (make_pass_record_bounds): ...new function.
13118         (pass_graphite): Replace declaration with that of...
13119         (make_pass_graphite): ...new function.
13120         (pass_graphite_transforms): Replace declaration with that of...
13121         (make_pass_graphite_transforms): ...new function.
13122         (pass_if_conversion): Replace declaration with that of...
13123         (make_pass_if_conversion): ...new function.
13124         (pass_loop_distribution): Replace declaration with that of...
13125         (make_pass_loop_distribution): ...new function.
13126         (pass_vectorize): Replace declaration with that of...
13127         (make_pass_vectorize): ...new function.
13128         (pass_slp_vectorize): Replace declaration with that of...
13129         (make_pass_slp_vectorize): ...new function.
13130         (pass_complete_unroll): Replace declaration with that of...
13131         (make_pass_complete_unroll): ...new function.
13132         (pass_complete_unrolli): Replace declaration with that of...
13133         (make_pass_complete_unrolli): ...new function.
13134         (pass_parallelize_loops): Replace declaration with that of...
13135         (make_pass_parallelize_loops): ...new function.
13136         (pass_loop_prefetch): Replace declaration with that of...
13137         (make_pass_loop_prefetch): ...new function.
13138         (pass_iv_optimize): Replace declaration with that of...
13139         (make_pass_iv_optimize): ...new function.
13140         (pass_tree_loop_done): Replace declaration with that of...
13141         (make_pass_tree_loop_done): ...new function.
13142         (pass_ch): Replace declaration with that of...
13143         (make_pass_ch): ...new function.
13144         (pass_ccp): Replace declaration with that of...
13145         (make_pass_ccp): ...new function.
13146         (pass_phi_only_cprop): Replace declaration with that of...
13147         (make_pass_phi_only_cprop): ...new function.
13148         (pass_build_ssa): Replace declaration with that of...
13149         (make_pass_build_ssa): ...new function.
13150         (pass_build_alias): Replace declaration with that of...
13151         (make_pass_build_alias): ...new function.
13152         (pass_build_ealias): Replace declaration with that of...
13153         (make_pass_build_ealias): ...new function.
13154         (pass_dominator): Replace declaration with that of...
13155         (make_pass_dominator): ...new function.
13156         (pass_dce): Replace declaration with that of...
13157         (make_pass_dce): ...new function.
13158         (pass_dce_loop): Replace declaration with that of...
13159         (make_pass_dce_loop): ...new function.
13160         (pass_cd_dce): Replace declaration with that of...
13161         (make_pass_cd_dce): ...new function.
13162         (pass_call_cdce): Replace declaration with that of...
13163         (make_pass_call_cdce): ...new function.
13164         (pass_merge_phi): Replace declaration with that of...
13165         (make_pass_merge_phi): ...new function.
13166         (pass_split_crit_edges): Replace declaration with that of...
13167         (make_pass_split_crit_edges): ...new function.
13168         (pass_pre): Replace declaration with that of...
13169         (make_pass_pre): ...new function.
13170         (pass_profile): Replace declaration with that of...
13171         (make_pass_profile): ...new function.
13172         (pass_strip_predict_hints): Replace declaration with that of...
13173         (make_pass_strip_predict_hints): ...new function.
13174         (pass_lower_complex_O0): Replace declaration with that of...
13175         (make_pass_lower_complex_O0): ...new function.
13176         (pass_lower_complex): Replace declaration with that of...
13177         (make_pass_lower_complex): ...new function.
13178         (pass_lower_vector): Replace declaration with that of...
13179         (make_pass_lower_vector): ...new function.
13180         (pass_lower_vector_ssa): Replace declaration with that of...
13181         (make_pass_lower_vector_ssa): ...new function.
13182         (pass_lower_omp): Replace declaration with that of...
13183         (make_pass_lower_omp): ...new function.
13184         (pass_diagnose_omp_blocks): Replace declaration with that of...
13185         (make_pass_diagnose_omp_blocks): ...new function.
13186         (pass_expand_omp): Replace declaration with that of...
13187         (make_pass_expand_omp): ...new function.
13188         (pass_expand_omp_ssa): Replace declaration with that of...
13189         (make_pass_expand_omp_ssa): ...new function.
13190         (pass_object_sizes): Replace declaration with that of...
13191         (make_pass_object_sizes): ...new function.
13192         (pass_strlen): Replace declaration with that of...
13193         (make_pass_strlen): ...new function.
13194         (pass_fold_builtins): Replace declaration with that of...
13195         (make_pass_fold_builtins): ...new function.
13196         (pass_stdarg): Replace declaration with that of...
13197         (make_pass_stdarg): ...new function.
13198         (pass_early_warn_uninitialized): Replace declaration with that of...
13199         (make_pass_early_warn_uninitialized): ...new function.
13200         (pass_late_warn_uninitialized): Replace declaration with that of...
13201         (make_pass_late_warn_uninitialized): ...new function.
13202         (pass_cse_reciprocals): Replace declaration with that of...
13203         (make_pass_cse_reciprocals): ...new function.
13204         (pass_cse_sincos): Replace declaration with that of...
13205         (make_pass_cse_sincos): ...new function.
13206         (pass_optimize_bswap): Replace declaration with that of...
13207         (make_pass_optimize_bswap): ...new function.
13208         (pass_optimize_widening_mul): Replace declaration with that of...
13209         (make_pass_optimize_widening_mul): ...new function.
13210         (pass_warn_function_return): Replace declaration with that of...
13211         (make_pass_warn_function_return): ...new function.
13212         (pass_warn_function_noreturn): Replace declaration with that of...
13213         (make_pass_warn_function_noreturn): ...new function.
13214         (pass_cselim): Replace declaration with that of...
13215         (make_pass_cselim): ...new function.
13216         (pass_phiopt): Replace declaration with that of...
13217         (make_pass_phiopt): ...new function.
13218         (pass_forwprop): Replace declaration with that of...
13219         (make_pass_forwprop): ...new function.
13220         (pass_phiprop): Replace declaration with that of...
13221         (make_pass_phiprop): ...new function.
13222         (pass_tree_ifcombine): Replace declaration with that of...
13223         (make_pass_tree_ifcombine): ...new function.
13224         (pass_dse): Replace declaration with that of...
13225         (make_pass_dse): ...new function.
13226         (pass_nrv): Replace declaration with that of...
13227         (make_pass_nrv): ...new function.
13228         (pass_rename_ssa_copies): Replace declaration with that of...
13229         (make_pass_rename_ssa_copies): ...new function.
13230         (pass_sink_code): Replace declaration with that of...
13231         (make_pass_sink_code): ...new function.
13232         (pass_fre): Replace declaration with that of...
13233         (make_pass_fre): ...new function.
13234         (pass_check_data_deps): Replace declaration with that of...
13235         (make_pass_check_data_deps): ...new function.
13236         (pass_copy_prop): Replace declaration with that of...
13237         (make_pass_copy_prop): ...new function.
13238         (pass_vrp): Replace declaration with that of...
13239         (make_pass_vrp): ...new function.
13240         (pass_uncprop): Replace declaration with that of...
13241         (make_pass_uncprop): ...new function.
13242         (pass_return_slot): Replace declaration with that of...
13243         (make_pass_return_slot): ...new function.
13244         (pass_reassoc): Replace declaration with that of...
13245         (make_pass_reassoc): ...new function.
13246         (pass_rebuild_cgraph_edges): Replace declaration with that of...
13247         (make_pass_rebuild_cgraph_edges): ...new function.
13248         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
13249         (make_pass_remove_cgraph_callee_edges): ...new function.
13250         (pass_build_cgraph_edges): Replace declaration with that of...
13251         (make_pass_build_cgraph_edges): ...new function.
13252         (pass_local_pure_const): Replace declaration with that of...
13253         (make_pass_local_pure_const): ...new function.
13254         (pass_tracer): Replace declaration with that of...
13255         (make_pass_tracer): ...new function.
13256         (pass_warn_unused_result): Replace declaration with that of...
13257         (make_pass_warn_unused_result): ...new function.
13258         (pass_diagnose_tm_blocks): Replace declaration with that of...
13259         (make_pass_diagnose_tm_blocks): ...new function.
13260         (pass_lower_tm): Replace declaration with that of...
13261         (make_pass_lower_tm): ...new function.
13262         (pass_tm_init): Replace declaration with that of...
13263         (make_pass_tm_init): ...new function.
13264         (pass_tm_mark): Replace declaration with that of...
13265         (make_pass_tm_mark): ...new function.
13266         (pass_tm_memopt): Replace declaration with that of...
13267         (make_pass_tm_memopt): ...new function.
13268         (pass_tm_edges): Replace declaration with that of...
13269         (make_pass_tm_edges): ...new function.
13270         (pass_split_functions): Replace declaration with that of...
13271         (make_pass_split_functions): ...new function.
13272         (pass_feedback_split_functions): Replace declaration with that of...
13273         (make_pass_feedback_split_functions): ...new function.
13274         (pass_strength_reduction): Replace declaration with that of...
13275         (make_pass_strength_reduction): ...new function.
13276         (pass_ipa_lower_emutls): Replace declaration with that of...
13277         (make_pass_ipa_lower_emutls): ...new function.
13278         (pass_ipa_function_and_variable_visibility): Replace declaration with
13279         that of...
13280         (make_pass_ipa_function_and_variable_visibility): ...new function.
13281         (pass_ipa_tree_profile): Replace declaration with that of...
13282         (make_pass_ipa_tree_profile): ...new function.
13283         (pass_early_local_passes): Replace declaration with that of...
13284         (make_pass_early_local_passes): ...new function.
13285         (pass_ipa_whole_program_visibility): Replace declaration with that
13286         of...
13287         (make_pass_ipa_whole_program_visibility): ...new function.
13288         (pass_ipa_lto_gimple_out): Replace declaration with that of...
13289         (make_pass_ipa_lto_gimple_out): ...new function.
13290         (pass_ipa_increase_alignment): Replace declaration with that of...
13291         (make_pass_ipa_increase_alignment): ...new function.
13292         (pass_ipa_inline): Replace declaration with that of...
13293         (make_pass_ipa_inline): ...new function.
13294         (pass_ipa_free_lang_data): Replace declaration with that of...
13295         (make_pass_ipa_free_lang_data): ...new function.
13296         (pass_ipa_free_inline_summary): Replace declaration with that of...
13297         (make_pass_ipa_free_inline_summary): ...new function.
13298         (pass_ipa_cp): Replace declaration with that of...
13299         (make_pass_ipa_cp): ...new function.
13300         (pass_ipa_reference): Replace declaration with that of...
13301         (make_pass_ipa_reference): ...new function.
13302         (pass_ipa_pure_const): Replace declaration with that of...
13303         (make_pass_ipa_pure_const): ...new function.
13304         (pass_ipa_pta): Replace declaration with that of...
13305         (make_pass_ipa_pta): ...new function.
13306         (pass_ipa_lto_finish_out): Replace declaration with that of...
13307         (make_pass_ipa_lto_finish_out): ...new function.
13308         (pass_ipa_tm): Replace declaration with that of...
13309         (make_pass_ipa_tm): ...new function.
13310         (pass_ipa_profile): Replace declaration with that of...
13311         (make_pass_ipa_profile): ...new function.
13312         (pass_ipa_cdtor_merge): Replace declaration with that of...
13313         (make_pass_ipa_cdtor_merge): ...new function.
13314         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
13315         of...
13316         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13317         (pass_init_datastructures): Replace declaration with that of...
13318         (make_pass_init_datastructures): ...new function.
13319         (pass_fixup_cfg): Replace declaration with that of...
13320         (make_pass_fixup_cfg): ...new function.
13321         (pass_expand): Replace declaration with that of...
13322         (make_pass_expand): ...new function.
13323         (pass_instantiate_virtual_regs): Replace declaration with that of...
13324         (make_pass_instantiate_virtual_regs): ...new function.
13325         (pass_rtl_fwprop): Replace declaration with that of...
13326         (make_pass_rtl_fwprop): ...new function.
13327         (pass_rtl_fwprop_addr): Replace declaration with that of...
13328         (make_pass_rtl_fwprop_addr): ...new function.
13329         (pass_jump): Replace declaration with that of...
13330         (make_pass_jump): ...new function.
13331         (pass_jump2): Replace declaration with that of...
13332         (make_pass_jump2): ...new function.
13333         (pass_lower_subreg): Replace declaration with that of...
13334         (make_pass_lower_subreg): ...new function.
13335         (pass_cse): Replace declaration with that of...
13336         (make_pass_cse): ...new function.
13337         (pass_fast_rtl_dce): Replace declaration with that of...
13338         (make_pass_fast_rtl_dce): ...new function.
13339         (pass_ud_rtl_dce): Replace declaration with that of...
13340         (make_pass_ud_rtl_dce): ...new function.
13341         (pass_rtl_dce): Replace declaration with that of...
13342         (make_pass_rtl_dce): ...new function.
13343         (pass_rtl_dse1): Replace declaration with that of...
13344         (make_pass_rtl_dse1): ...new function.
13345         (pass_rtl_dse2): Replace declaration with that of...
13346         (make_pass_rtl_dse2): ...new function.
13347         (pass_rtl_dse3): Replace declaration with that of...
13348         (make_pass_rtl_dse3): ...new function.
13349         (pass_rtl_cprop): Replace declaration with that of...
13350         (make_pass_rtl_cprop): ...new function.
13351         (pass_rtl_pre): Replace declaration with that of...
13352         (make_pass_rtl_pre): ...new function.
13353         (pass_rtl_hoist): Replace declaration with that of...
13354         (make_pass_rtl_hoist): ...new function.
13355         (pass_rtl_store_motion): Replace declaration with that of...
13356         (make_pass_rtl_store_motion): ...new function.
13357         (pass_cse_after_global_opts): Replace declaration with that of...
13358         (make_pass_cse_after_global_opts): ...new function.
13359         (pass_rtl_ifcvt): Replace declaration with that of...
13360         (make_pass_rtl_ifcvt): ...new function.
13361         (pass_into_cfg_layout_mode): Replace declaration with that of...
13362         (make_pass_into_cfg_layout_mode): ...new function.
13363         (pass_outof_cfg_layout_mode): Replace declaration with that of...
13364         (make_pass_outof_cfg_layout_mode): ...new function.
13365         (pass_loop2): Replace declaration with that of...
13366         (make_pass_loop2): ...new function.
13367         (pass_rtl_loop_init): Replace declaration with that of...
13368         (make_pass_rtl_loop_init): ...new function.
13369         (pass_rtl_move_loop_invariants): Replace declaration with that of...
13370         (make_pass_rtl_move_loop_invariants): ...new function.
13371         (pass_rtl_unswitch): Replace declaration with that of...
13372         (make_pass_rtl_unswitch): ...new function.
13373         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
13374         (make_pass_rtl_unroll_and_peel_loops): ...new function.
13375         (pass_rtl_doloop): Replace declaration with that of...
13376         (make_pass_rtl_doloop): ...new function.
13377         (pass_rtl_loop_done): Replace declaration with that of...
13378         (make_pass_rtl_loop_done): ...new function.
13379         (pass_web): Replace declaration with that of...
13380         (make_pass_web): ...new function.
13381         (pass_cse2): Replace declaration with that of...
13382         (make_pass_cse2): ...new function.
13383         (pass_df_initialize_opt): Replace declaration with that of...
13384         (make_pass_df_initialize_opt): ...new function.
13385         (pass_df_initialize_no_opt): Replace declaration with that of...
13386         (make_pass_df_initialize_no_opt): ...new function.
13387         (pass_reginfo_init): Replace declaration with that of...
13388         (make_pass_reginfo_init): ...new function.
13389         (pass_inc_dec): Replace declaration with that of...
13390         (make_pass_inc_dec): ...new function.
13391         (pass_stack_ptr_mod): Replace declaration with that of...
13392         (make_pass_stack_ptr_mod): ...new function.
13393         (pass_initialize_regs): Replace declaration with that of...
13394         (make_pass_initialize_regs): ...new function.
13395         (pass_combine): Replace declaration with that of...
13396         (make_pass_combine): ...new function.
13397         (pass_if_after_combine): Replace declaration with that of...
13398         (make_pass_if_after_combine): ...new function.
13399         (pass_ree): Replace declaration with that of...
13400         (make_pass_ree): ...new function.
13401         (pass_partition_blocks): Replace declaration with that of...
13402         (make_pass_partition_blocks): ...new function.
13403         (pass_match_asm_constraints): Replace declaration with that of...
13404         (make_pass_match_asm_constraints): ...new function.
13405         (pass_regmove): Replace declaration with that of...
13406         (make_pass_regmove): ...new function.
13407         (pass_split_all_insns): Replace declaration with that of...
13408         (make_pass_split_all_insns): ...new function.
13409         (pass_fast_rtl_byte_dce): Replace declaration with that of...
13410         (make_pass_fast_rtl_byte_dce): ...new function.
13411         (pass_lower_subreg2): Replace declaration with that of...
13412         (make_pass_lower_subreg2): ...new function.
13413         (pass_mode_switching): Replace declaration with that of...
13414         (make_pass_mode_switching): ...new function.
13415         (pass_sms): Replace declaration with that of...
13416         (make_pass_sms): ...new function.
13417         (pass_sched): Replace declaration with that of...
13418         (make_pass_sched): ...new function.
13419         (pass_ira): Replace declaration with that of...
13420         (make_pass_ira): ...new function.
13421         (pass_reload): Replace declaration with that of...
13422         (make_pass_reload): ...new function.
13423         (pass_clean_state): Replace declaration with that of...
13424         (make_pass_clean_state): ...new function.
13425         (pass_branch_prob): Replace declaration with that of...
13426         (make_pass_branch_prob): ...new function.
13427         (pass_value_profile_transformations): Replace declaration with that
13428         of...
13429         (make_pass_value_profile_transformations): ...new function.
13430         (pass_postreload_cse): Replace declaration with that of...
13431         (make_pass_postreload_cse): ...new function.
13432         (pass_gcse2): Replace declaration with that of...
13433         (make_pass_gcse2): ...new function.
13434         (pass_split_after_reload): Replace declaration with that of...
13435         (make_pass_split_after_reload): ...new function.
13436         (pass_branch_target_load_optimize1): Replace declaration with that
13437         of...
13438         (make_pass_branch_target_load_optimize1): ...new function.
13439         (pass_thread_prologue_and_epilogue): Replace declaration with that
13440         of...
13441         (make_pass_thread_prologue_and_epilogue): ...new function.
13442         (pass_stack_adjustments): Replace declaration with that of...
13443         (make_pass_stack_adjustments): ...new function.
13444         (pass_peephole2): Replace declaration with that of...
13445         (make_pass_peephole2): ...new function.
13446         (pass_if_after_reload): Replace declaration with that of...
13447         (make_pass_if_after_reload): ...new function.
13448         (pass_regrename): Replace declaration with that of...
13449         (make_pass_regrename): ...new function.
13450         (pass_cprop_hardreg): Replace declaration with that of...
13451         (make_pass_cprop_hardreg): ...new function.
13452         (pass_reorder_blocks): Replace declaration with that of...
13453         (make_pass_reorder_blocks): ...new function.
13454         (pass_branch_target_load_optimize2): Replace declaration with that
13455         of...
13456         (make_pass_branch_target_load_optimize2): ...new function.
13457         (pass_leaf_regs): Replace declaration with that of...
13458         (make_pass_leaf_regs): ...new function.
13459         (pass_split_before_sched2): Replace declaration with that of...
13460         (make_pass_split_before_sched2): ...new function.
13461         (pass_compare_elim_after_reload): Replace declaration with that of...
13462         (make_pass_compare_elim_after_reload): ...new function.
13463         (pass_sched2): Replace declaration with that of...
13464         (make_pass_sched2): ...new function.
13465         (pass_stack_regs): Replace declaration with that of...
13466         (make_pass_stack_regs): ...new function.
13467         (pass_stack_regs_run): Replace declaration with that of...
13468         (make_pass_stack_regs_run): ...new function.
13469         (pass_df_finish): Replace declaration with that of...
13470         (make_pass_df_finish): ...new function.
13471         (pass_compute_alignments): Replace declaration with that of...
13472         (make_pass_compute_alignments): ...new function.
13473         (pass_duplicate_computed_gotos): Replace declaration with that of...
13474         (make_pass_duplicate_computed_gotos): ...new function.
13475         (pass_variable_tracking): Replace declaration with that of...
13476         (make_pass_variable_tracking): ...new function.
13477         (pass_free_cfg): Replace declaration with that of...
13478         (make_pass_free_cfg): ...new function.
13479         (pass_machine_reorg): Replace declaration with that of...
13480         (make_pass_machine_reorg): ...new function.
13481         (pass_cleanup_barriers): Replace declaration with that of...
13482         (make_pass_cleanup_barriers): ...new function.
13483         (pass_delay_slots): Replace declaration with that of...
13484         (make_pass_delay_slots): ...new function.
13485         (pass_split_for_shorten_branches): Replace declaration with that of...
13486         (make_pass_split_for_shorten_branches): ...new function.
13487         (pass_split_before_regstack): Replace declaration with that of...
13488         (make_pass_split_before_regstack): ...new function.
13489         (pass_convert_to_eh_region_ranges): Replace declaration with that
13490         of...
13491         (make_pass_convert_to_eh_region_ranges): ...new function.
13492         (pass_shorten_branches): Replace declaration with that of...
13493         (make_pass_shorten_branches): ...new function.
13494         (pass_set_nothrow_function_flags): Replace declaration with that of...
13495         (make_pass_set_nothrow_function_flags): ...new function.
13496         (pass_dwarf2_frame): Replace declaration with that of...
13497         (make_pass_dwarf2_frame): ...new function.
13498         (pass_final): Replace declaration with that of...
13499         (make_pass_final): ...new function.
13500         (pass_rtl_seqabstr): Replace declaration with that of...
13501         (make_pass_rtl_seqabstr): ...new function.
13502         (pass_release_ssa_names): Replace declaration with that of...
13503         (make_pass_release_ssa_names): ...new function.
13504         (pass_early_inline): Replace declaration with that of...
13505         (make_pass_early_inline): ...new function.
13506         (pass_inline_parameters): Replace declaration with that of...
13507         (make_pass_inline_parameters): ...new function.
13508         (pass_update_address_taken): Replace declaration with that of...
13509         (make_pass_update_address_taken): ...new function.
13510         (pass_convert_switch): Replace declaration with that of...
13511         (make_pass_convert_switch): ...new function.
13512         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
13513         to a subclass of simple_ipa_opt_pass along with...
13514         (pass_data_ipa_tree_profile): ...new pass_data instance and...
13515         (make_pass_ipa_tree_profile): ...new function.
13516         * tree-sra.c (pass_sra_early): Convert from a global struct to a
13517         subclass of gimple_opt_pass along with...
13518         (pass_data_sra_early): ...new pass_data instance and...
13519         (make_pass_sra_early): ...new function.
13520         (pass_sra): Convert from a global struct to a subclass of
13521         gimple_opt_pass along with...
13522         (pass_data_sra): ...new pass_data instance and...
13523         (make_pass_sra): ...new function.
13524         (pass_early_ipa_sra): Convert from a global struct to a subclass of
13525         gimple_opt_pass along with...
13526         (pass_data_early_ipa_sra): ...new pass_data instance and...
13527         (make_pass_early_ipa_sra): ...new function.
13528         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
13529         subclass of gimple_opt_pass along with...
13530         (pass_data_ccp): ...new pass_data instance and...
13531         (make_pass_ccp): ...new function.
13532         (pass_fold_builtins): Convert from a global struct to a subclass of
13533         gimple_opt_pass along with...
13534         (pass_data_fold_builtins): ...new pass_data instance and...
13535         (make_pass_fold_builtins): ...new function.
13536         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
13537         subclass of gimple_opt_pass along with...
13538         (pass_data_copy_prop): ...new pass_data instance and...
13539         (make_pass_copy_prop): ...new function.
13540         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
13541         global struct to a subclass of gimple_opt_pass along with...
13542         (pass_data_rename_ssa_copies): ...new pass_data instance and...
13543         (make_pass_rename_ssa_copies): ...new function.
13544         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
13545         subclass of gimple_opt_pass along with...
13546         (pass_data_dce): ...new pass_data instance and...
13547         (make_pass_dce): ...new function.
13548         (pass_dce_loop): Convert from a global struct to a subclass of
13549         gimple_opt_pass along with...
13550         (pass_data_dce_loop): ...new pass_data instance and...
13551         (make_pass_dce_loop): ...new function.
13552         (pass_cd_dce): Convert from a global struct to a subclass of
13553         gimple_opt_pass along with...
13554         (pass_data_cd_dce): ...new pass_data instance and...
13555         (make_pass_cd_dce): ...new function.
13556         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
13557         subclass of gimple_opt_pass along with...
13558         (pass_data_dominator): ...new pass_data instance and...
13559         (make_pass_dominator): ...new function.
13560         (pass_phi_only_cprop): Convert from a global struct to a subclass of
13561         gimple_opt_pass along with...
13562         (pass_data_phi_only_cprop): ...new pass_data instance and...
13563         (make_pass_phi_only_cprop): ...new function.
13564         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
13565         subclass of gimple_opt_pass along with...
13566         (pass_data_dse): ...new pass_data instance and...
13567         (make_pass_dse): ...new function.
13568         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
13569         a subclass of gimple_opt_pass along with...
13570         (pass_data_forwprop): ...new pass_data instance and...
13571         (make_pass_forwprop): ...new function.
13572         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
13573         struct to a subclass of gimple_opt_pass along with...
13574         (pass_data_tree_ifcombine): ...new pass_data instance and...
13575         (make_pass_tree_ifcombine): ...new function.
13576         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
13577         subclass of gimple_opt_pass along with...
13578         (pass_data_ch): ...new pass_data instance and...
13579         (make_pass_ch): ...new function.
13580         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
13581         subclass of gimple_opt_pass along with...
13582         (pass_data_tree_loop): ...new pass_data instance and...
13583         (make_pass_tree_loop): ...new function.
13584         (pass_tree_loop_init): Convert from a global struct to a subclass of
13585         gimple_opt_pass along with...
13586         (pass_data_tree_loop_init): ...new pass_data instance and...
13587         (make_pass_tree_loop_init): ...new function.
13588         (pass_lim): Convert from a global struct to a subclass of
13589         gimple_opt_pass along with...
13590         (pass_data_lim): ...new pass_data instance and...
13591         (make_pass_lim): ...new function.
13592         (pass_tree_unswitch): Convert from a global struct to a subclass of
13593         gimple_opt_pass along with...
13594         (pass_data_tree_unswitch): ...new pass_data instance and...
13595         (make_pass_tree_unswitch): ...new function.
13596         (pass_predcom): Convert from a global struct to a subclass of
13597         gimple_opt_pass along with...
13598         (pass_data_predcom): ...new pass_data instance and...
13599         (make_pass_predcom): ...new function.
13600         (pass_vectorize): Convert from a global struct to a subclass of
13601         gimple_opt_pass along with...
13602         (pass_data_vectorize): ...new pass_data instance and...
13603         (make_pass_vectorize): ...new function.
13604         (pass_graphite): Convert from a global struct to a subclass of
13605         gimple_opt_pass along with...
13606         (pass_data_graphite): ...new pass_data instance and...
13607         (make_pass_graphite): ...new function.
13608         (pass_graphite_transforms): Convert from a global struct to a subclass
13609         of gimple_opt_pass along with...
13610         (pass_data_graphite_transforms): ...new pass_data instance and...
13611         (make_pass_graphite_transforms): ...new function.
13612         (pass_check_data_deps): Convert from a global struct to a subclass of
13613         gimple_opt_pass along with...
13614         (pass_data_check_data_deps): ...new pass_data instance and...
13615         (make_pass_check_data_deps): ...new function.
13616         (pass_iv_canon): Convert from a global struct to a subclass of
13617         gimple_opt_pass along with...
13618         (pass_data_iv_canon): ...new pass_data instance and...
13619         (make_pass_iv_canon): ...new function.
13620         (pass_scev_cprop): Convert from a global struct to a subclass of
13621         gimple_opt_pass along with...
13622         (pass_data_scev_cprop): ...new pass_data instance and...
13623         (make_pass_scev_cprop): ...new function.
13624         (pass_record_bounds): Convert from a global struct to a subclass of
13625         gimple_opt_pass along with...
13626         (pass_data_record_bounds): ...new pass_data instance and...
13627         (make_pass_record_bounds): ...new function.
13628         (pass_complete_unroll): Convert from a global struct to a subclass of
13629         gimple_opt_pass along with...
13630         (pass_data_complete_unroll): ...new pass_data instance and...
13631         (make_pass_complete_unroll): ...new function.
13632         (pass_complete_unrolli): Convert from a global struct to a subclass of
13633         gimple_opt_pass along with...
13634         (pass_data_complete_unrolli): ...new pass_data instance and...
13635         (make_pass_complete_unrolli): ...new function.
13636         (pass_parallelize_loops): Convert from a global struct to a subclass
13637         of gimple_opt_pass along with...
13638         (pass_data_parallelize_loops): ...new pass_data instance and...
13639         (make_pass_parallelize_loops): ...new function.
13640         (pass_loop_prefetch): Convert from a global struct to a subclass of
13641         gimple_opt_pass along with...
13642         (pass_data_loop_prefetch): ...new pass_data instance and...
13643         (make_pass_loop_prefetch): ...new function.
13644         (pass_iv_optimize): Convert from a global struct to a subclass of
13645         gimple_opt_pass along with...
13646         (pass_data_iv_optimize): ...new pass_data instance and...
13647         (make_pass_iv_optimize): ...new function.
13648         (pass_tree_loop_done): Convert from a global struct to a subclass of
13649         gimple_opt_pass along with...
13650         (pass_data_tree_loop_done): ...new pass_data instance and...
13651         (make_pass_tree_loop_done): ...new function.
13652         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
13653         struct to a subclass of gimple_opt_pass along with...
13654         (pass_data_cse_reciprocals): ...new pass_data instance and...
13655         (make_pass_cse_reciprocals): ...new function.
13656         (pass_cse_sincos): Convert from a global struct to a subclass of
13657         gimple_opt_pass along with...
13658         (pass_data_cse_sincos): ...new pass_data instance and...
13659         (make_pass_cse_sincos): ...new function.
13660         (pass_optimize_bswap): Convert from a global struct to a subclass of
13661         gimple_opt_pass along with...
13662         (pass_data_optimize_bswap): ...new pass_data instance and...
13663         (make_pass_optimize_bswap): ...new function.
13664         (pass_optimize_widening_mul): Convert from a global struct to a
13665         subclass of gimple_opt_pass along with...
13666         (pass_data_optimize_widening_mul): ...new pass_data instance and...
13667         (make_pass_optimize_widening_mul): ...new function.
13668         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
13669         subclass of gimple_opt_pass along with...
13670         (pass_data_phiopt): ...new pass_data instance and...
13671         (make_pass_phiopt): ...new function.
13672         (pass_cselim): Convert from a global struct to a subclass of
13673         gimple_opt_pass along with...
13674         (pass_data_cselim): ...new pass_data instance and...
13675         (make_pass_cselim): ...new function.
13676         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
13677         subclass of gimple_opt_pass along with...
13678         (pass_data_phiprop): ...new pass_data instance and...
13679         (make_pass_phiprop): ...new function.
13680         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
13681         subclass of gimple_opt_pass along with...
13682         (pass_data_pre): ...new pass_data instance and...
13683         (make_pass_pre): ...new function.
13684         (pass_fre): Convert from a global struct to a subclass of
13685         gimple_opt_pass along with...
13686         (pass_data_fre): ...new pass_data instance and...
13687         (make_pass_fre): ...new function.
13688         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
13689         subclass of gimple_opt_pass along with...
13690         (pass_data_reassoc): ...new pass_data instance and...
13691         (make_pass_reassoc): ...new function.
13692         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
13693         subclass of gimple_opt_pass along with...
13694         (pass_data_sink_code): ...new pass_data instance and...
13695         (make_pass_sink_code): ...new function.
13696         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
13697         subclass of gimple_opt_pass along with...
13698         (pass_data_strlen): ...new pass_data instance and...
13699         (make_pass_strlen): ...new function.
13700         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
13701         struct to a subclass of gimple_opt_pass along with...
13702         (pass_data_build_alias): ...new pass_data instance and...
13703         (make_pass_build_alias): ...new function.
13704         (pass_build_ealias): Convert from a global struct to a subclass of
13705         gimple_opt_pass along with...
13706         (pass_data_build_ealias): ...new pass_data instance and...
13707         (make_pass_build_ealias): ...new function.
13708         (pass_ipa_pta): Convert from a global struct to a subclass of
13709         simple_ipa_opt_pass along with...
13710         (pass_data_ipa_pta): ...new pass_data instance and...
13711         (make_pass_ipa_pta): ...new function.
13712         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
13713         subclass of gimple_opt_pass along with...
13714         (pass_data_uncprop): ...new pass_data instance and...
13715         (make_pass_uncprop): ...new function.
13716         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
13717         global struct to a subclass of gimple_opt_pass along with...
13718         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
13719         (make_pass_late_warn_uninitialized): ...new function.
13720         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
13721         to a subclass of gimple_opt_pass along with...
13722         (pass_data_init_datastructures): ...new pass_data instance and...
13723         (make_pass_init_datastructures): ...new function.
13724         (pass_early_warn_uninitialized): Convert from a global struct to a
13725         subclass of gimple_opt_pass along with...
13726         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
13727         (make_pass_early_warn_uninitialized): ...new function.
13728         (pass_update_address_taken): Convert from a global struct to a
13729         subclass of gimple_opt_pass along with...
13730         (pass_data_update_address_taken): ...new pass_data instance and...
13731         (make_pass_update_address_taken): ...new function.
13732         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
13733         struct to a subclass of gimple_opt_pass along with...
13734         (pass_data_release_ssa_names): ...new pass_data instance and...
13735         (make_pass_release_ssa_names): ...new function.
13736         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
13737         subclass of gimple_opt_pass along with...
13738         (pass_data_stdarg): ...new pass_data instance and...
13739         (make_pass_stdarg): ...new function.
13740         * tree-switch-conversion.c (pass_convert_switch): Convert from a
13741         global struct to a subclass of gimple_opt_pass along with...
13742         (pass_data_convert_switch): ...new pass_data instance and...
13743         (make_pass_convert_switch): ...new function.
13744         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
13745         to a subclass of gimple_opt_pass along with...
13746         (pass_data_tail_recursion): ...new pass_data instance and...
13747         (make_pass_tail_recursion): ...new function.
13748         (pass_tail_calls): Convert from a global struct to a subclass of
13749         gimple_opt_pass along with...
13750         (pass_data_tail_calls): ...new pass_data instance and...
13751         (make_pass_tail_calls): ...new function.
13752         * tree-vect-generic.c (pass_lower_vector): Convert from a global
13753         struct to a subclass of gimple_opt_pass along with...
13754         (pass_data_lower_vector): ...new pass_data instance and...
13755         (make_pass_lower_vector): ...new function.
13756         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
13757         gimple_opt_pass along with...
13758         (pass_data_lower_vector_ssa): ...new pass_data instance and...
13759         (make_pass_lower_vector_ssa): ...new function.
13760         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
13761         to a subclass of gimple_opt_pass along with...
13762         (pass_data_slp_vectorize): ...new pass_data instance and...
13763         (make_pass_slp_vectorize): ...new function.
13764         (pass_ipa_increase_alignment): Convert from a global struct to a
13765         subclass of simple_ipa_opt_pass along with...
13766         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
13767         (make_pass_ipa_increase_alignment): ...new function.
13768         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
13769         gimple_opt_pass along with...
13770         (pass_data_vrp): ...new pass_data instance and...
13771         (make_pass_vrp): ...new function.
13772         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
13773         subclass of simple_ipa_opt_pass along with...
13774         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
13775         (make_pass_ipa_free_lang_data): ...new function.
13776         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
13777         gimple_opt_pass along with...
13778         (pass_data_tsan): ...new pass_data instance and...
13779         (make_pass_tsan): ...new function.
13780         (pass_tsan_O0): Convert from a global struct to a subclass of
13781         gimple_opt_pass along with...
13782         (pass_data_tsan_O0): ...new pass_data instance and...
13783         (make_pass_tsan_O0): ...new function.
13784         * var-tracking.c (pass_variable_tracking): Convert from a global
13785         struct to a subclass of rtl_opt_pass along with...
13786         (pass_data_variable_tracking): ...new pass_data instance and...
13787         (make_pass_variable_tracking): ...new function.
13788         * web.c (pass_web): Convert from a global struct to a subclass of
13789         rtl_opt_pass along with...
13790         (pass_data_web): ...new pass_data instance and...
13791         (make_pass_web): ...new function.
13792         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
13793         declaration with that of...
13794         (make_pass_mode_switch_use): ...new function.
13795         (pass_resolve_sw_modes): Replace declaration with that of...
13796         (make_pass_resolve_sw_modes): ...new function.
13797         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
13798         from a global struct to a subclass of rtl_opt_pass along with...
13799         (pass_data_mode_switch_use): ...new pass_data instance and...
13800         (make_pass_mode_switch_use): ...new function.
13801         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
13802         from a global struct to a subclass of rtl_opt_pass along with...
13803         (pass_data_resolve_sw_modes): ...new pass_data instance and...
13804         (make_pass_resolve_sw_modes): ...new function.
13805         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
13806         struct to a subclass of rtl_opt_pass along with...
13807         (pass_data_insert_vzeroupper): ...new pass_data instance and...
13808         (make_pass_insert_vzeroupper): ...new function.
13809         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
13810         global struct to a subclass of rtl_opt_pass along with...
13811         (pass_data_work_around_errata): ...new pass_data instance and...
13812         (make_pass_work_around_errata): ...new function.
13813         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
13814         struct to a subclass of rtl_opt_pass along with...
13815         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
13816         (make_pass_mips_machine_reorg2): ...new function.
13818 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13820         * passes.c (pass_manager::operator new): New.
13822 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13824         Handwritten part of conversion of passes to C++ classes.
13826         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
13827         (toplev.o): Add dep on PASS_MANAGER_H.
13828         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
13829         of early local pases to reflect this moving from a global to a
13830         member of gcc::pass_manager.
13831         (cgraph_add_new_function): Likewise.
13832         * lto-cgraph.c (lto_output_node): Update for conversion of
13833         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
13834         * passes.c (opt_pass::clone): New.
13835         (opt_pass::gate): New.
13836         (opt_pass::execute): New.
13837         (opt_pass::opt_pass): New.
13838         (pass_manager::execute_early_local_passes): New.
13839         (pass_manager::execute_pass_mode_switching): new.
13840         (finish_optimization_passes): Convert to...
13841         (pass_manager::finish_optimization_passes): ...this.
13842         (finish_optimization_passes): Update for conversion of passes to
13843         C++ classes.
13844         (register_dump_files_1): Use has_gate since we cannot portably
13845         check a vtable entry against NULL.
13846         (dump_one_pass): Likewise.
13847         (ipa_write_summaries_2): Likewise.
13848         (ipa_write_optimization_summaries_1): Likewise.
13849         (ipa_read_summaries_1): Likewise.
13850         (ipa_read_optimization_summaries_1): Likewise.
13851         (execute_ipa_stmt_fixups): Likewise.
13852         (pass_manager::pass_manager): Rewrite pass-creation, invoking
13853         pass-creation functions rather than wiring up globals, and
13854         storing the results in fields of pass_manager generated using
13855         pass-instances.def.
13856         (pass_manager::dump_profile_report): Update for conversion of
13857         passes to C++ classes.
13858         (pass_manager::execute_ipa_summary_passes): Likewise.
13859         (execute_one_ipa_transform_pass): Likewise.
13860         (execute_one_pass): Use has_gate and has_execute since we cannot
13861         portably check a vtable entry against NULL.
13862         * pass_manager.h (pass_manager::finish_optimization_passes): New.
13863         (pass_manager): Use pass-instances.def to add fields for the
13864         various pass instances.
13865         * toplev.c (finalize): Update for move of
13866         finish_optimization_passes to a method of gcc::pass_manager.
13867         * toplev.h (finish_optimization_passes): Move to method of class
13868         pass_manager.
13869         * tree-pass.h (struct pass_data): New.
13870         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
13871         (opt_pass::gate): Convert to virtual function.
13872         (opt_pass::~opt_pass): New.
13873         (opt_pass::clone): New.
13874         (opt_pass::execute): Convert to virtual function.
13875         (opt_pass::opt_pass): New.
13876         (opt_pass::ctxt_): new.
13877         (gimple_opt_pass): Convert to subclass of opt_pass.
13878         (gimple_opt_pass::gimple_opt_pass): New.
13879         (rtl_opt_pass): Convert to subclass of opt_pass.
13880         (rtl_opt_pass::rtl_opt_pass): New.
13881         (ipa_opt_pass_d): Convert to subclass of opt_pass.
13882         (ipa_opt_pass_d::ipa_opt_pass_d): New.
13883         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
13884         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
13885         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
13886         invocation of pass_mode_switching to reflect this moving from a
13887         global to a member of gcc::pass_manager.
13888         (ix86_option_override): Rework how pass_insert_vzeroupper is
13889         added to the pass_manager to reflect autogenerated changes.
13890         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
13892 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
13894         PR rtl-optimization/57708
13895         * recog.c (peep2_find_free_register): Validate all regs in a
13896         multi-reg mode.
13898 2013-08-05  Jan Hubicka  <jh@suse.cz>
13900         PR lto/57602
13901         * cgraph.c (verify_cgraph_node): Accept local flags from other
13902         partitions.
13903         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
13904         (function_and_variable_visibility): Likewise.
13905         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
13907 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13909         * graph.c (init_graph_slim_pretty_print): Remove.
13910         (print_graph_cfg): Do not call it.  Use local pretty printer.
13911         (start_graph_dump): Likewise.
13913 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13915         * gimple-pretty-print.c (buffer): Remove.
13916         (initialized): Likewise.
13917         (maybe_init_pretty_print): Likewise.
13918         (print_gimple_stmt): Do not call it.  Use non-static local
13919         pretty_printer variable.
13920         (print_gimple_expr): Likewise.
13921         (print_gimple_seq): Likewise.
13922         (gimple_dump_bb): Likewise.
13924 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13926         * asan.c (asan_pp): Remove.
13927         (asan_pp_initialized): Likewise.
13928         (asan_pp_initialize): Likewise.
13929         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
13930         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
13931         (asan_add_global): Likewise.
13933 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13935         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
13936         * diagnostic.h (diagnostic_flush_buffer): Adjust.
13937         * pretty-print.c (pp_formatted_text_data): Likewise.
13938         (pp_indent): Rename from pp_base_indent.
13939         (pp_format): Rename from pp_base_format.
13940         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
13941         (pp_format_verbatim): Rename from pp_base_format_verbatim.
13942         (pp_flush): Rename from pp_base_flush.
13943         (pp_set_line_maximum_length): Rename from
13944         pp_base_set_line_maximum_length.
13945         (pp_clear_output_area): Rename from pp_base_clear_output_area.
13946         (pp_set_prefix): Rename from pp_base_set_prefix.
13947         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
13948         (pp_emit_prefix): Rename from pp_base_emit_prefix.
13949         (pp_append_text): Rename from pp_base_append_text.
13950         (pp_formatted_text): Rename from pp_base_formatted_text.
13951         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
13952         (pp_remaining_character_count_for_line): Rename from
13953         pp_base_remaining_character_count_for_line.
13954         (pp_newline): Rename from pp_base_newline.
13955         (pp_character): Rename from pp_base_character.
13956         (pp_string): Rename from pp_base_string.
13957         (pp_maybe_space): Rename from pp_base_maybe_space.
13958         * asan.c (asan_pp_string): Adjust.
13959         (asan_emit_stack_protection): Likewise.
13960         (asan_add_global): Likewise.
13961         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
13962         * tree-mudflap.c (mf_varname_tree): Likewise.
13963         * tree-pretty-print.c (pp_tree_identifier): Rename from
13964         pp_base_tree_identifier.
13965         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
13966         Declare as function.
13968 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13970         * pretty-print.h (pp_bar_bar): New.
13971         (pp_ampersand_ampersand): Likewise.
13972         (pp_less_equal): Likewise.
13973         (pp_greater_equal): Likewise.
13974         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
13975         printer functions instead of pp_string or operators and punctuators.
13976         (dump_gimple_call): Likewise.
13977         (dump_gimple_omp_for): Likewise.
13978         (dump_gimple_transaction): Likewise.
13979         (dump_gimple_phi): Likewise.
13980         (pp_gimple_stmt_1): Likewise.
13981         * sched-vis.c (print_insn): Likewise.
13982         * tree-mudflap.c (mf_varname_tree): Likewise.
13983         * tree-pretty-print.c (dump_block_node): Likewise.
13984         (dump_generic_node): Likewise.
13986 2013-08-02  Jan Hubicka  <jh@suse.cz>
13988         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
13989         boundaries.
13990         * lto-streamer-out.c (tree_is_indexable): Results decls and
13991         parm decls are not indexable.
13992         (DFS_write_tree_body): Do not follow args and results.
13993         (hash_tree): Likewise.
13994         (output_functions): Rearrange so struct function is needed
13995         only when real body is output; be able to also ouptut abstract
13996         functions; output DECL_ARGUMENTS and DECL_RESULT.
13997         (lto_output): When not in WPA, ale store abstract functions.
13998         (write_symbol): Do not care about RESULT_DECL.
13999         (output_symbol_p): Handle correctly sbtract decls.
14000         * lto-streamer-in.c (input_function): Rearrange so struct
14001         function can be NULL at entry; allow streaming of
14002         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
14003         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
14004         sanity check during LTO.
14005         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
14006         RESULT_DECl and DECL_ARGUMENTS.
14007         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14008         Likewise.
14010 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14012         * pretty-print.h (pp_underscore): New.
14013         (pp_comma): Tidy.
14014         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
14015         printer functions instead of pp_character.
14016         (dump_binary_rhs): Likewise.
14017         (dump_ternary_rhs): Likewise.
14018         (dump_gimple_call_args): Likewise.
14019         (pp_points_to_solution): Likewise.
14020         (dump_gimple_call): Likewise.
14021         (dump_gimple_switch): Likewise.
14022         (dump_gimple_cond): Likewise.
14023         (dump_gimple_bind): Likewise.
14024         (dump_gimple_try): Likewise.
14025         (dump_gimple_omp_for): Likewise.
14026         (dump_gimple_omp_continue): Likewise.
14027         (dump_gimple_omp_single): Likewise.
14028         (dump_gimple_omp_sections): Likewise.
14029         (dump_gimple_omp_block): Likewise.
14030         (dump_gimple_omp_critical): Likewise.
14031         (dump_gimple_transaction): Likewise.
14032         (dump_gimple_asm): Likewise.
14033         (dump_gimple_phi): Likewise.
14034         (dump_gimple_omp_parallel): Likewise.
14035         (dump_gimple_omp_task): Likewise.
14036         (dump_gimple_omp_atomic_load): Likewise.
14037         (dump_gimple_omp_atomic_store): Likewise.
14038         (dump_gimple_mem_ops): Likewise.
14039         (pp_gimple_stmt_1): Likewise.
14040         (pp_cfg_jump): Likewise.
14041         (dump_implicit_edges): Likewise.
14042         (gimple_dump_bb_for_graph): Likewise.
14043         * graph.c (draw_cfg_node): Likewise.
14044         * langhooks.c (lhd_print_error_function): Likewise.
14045         * sched-vis.c (print_exp): Likewise.
14046         (print_value): Likewise.
14047         (print_pattern): Likewise.
14048         (print_insn): Likewise.
14049         (rtl_dump_bb_for_graph): Likewise.
14050         * tree-pretty-print.c (dump_function_declaration): Likewise.
14051         (dump_array_domain): Likewise.
14052         (dump_omp_clause): Likewise.
14053         (dump_location): Likewise.
14054         (dump_generic_node): Likewise.
14055         (print_struct_decl): Likewise.
14056         * diagnostic.c (diagnostic_show_locus): Use pp_space.
14058 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14060         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
14061         candidate table when replacing a candidate statement.
14062         (replace_rhs_if_not_dup): Likewise.
14063         (replace_one_candidate): Likewise.
14065 2013-08-02  Jan Hubicka  <jh@suse.cz>
14066             Martin Liska  <marxin.liska@gmail.com>
14068         * cgraphunit.c (add_new_function): Fix logic when adding from
14069         late IPA pass.
14070         (assemble_thunk): Rename to ...
14071         (expand_thunk); .. this one; export; get it working with
14072         general functions; make produced gimple valid.
14073         * cgraph.h (expand_thunk): Declare.
14075 2013-08-02  Jan Hubicka  <jh@suse.cz>
14077         * ipa-cp.c (gather_context_independent_values): Use
14078         ipa_get_param_move_cost.
14079         (get_replacement_map): Remove PARAM; move parameter folding
14080         into tree-inline.c
14081         (create_specialized_node): Update.
14082         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
14083         assert that we have gimple body; update move_cost.
14084         (count_formal_params): Assert that we have gimple body.
14085         (ipa_dump_param): New function.
14086         (ipa_alloc_node_params): Break out from ...
14087         (ipa_initialize_node_params): ... here.
14088         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
14089         (ipa_write_node_info): Stream move costs.
14090         (ipa_read_node_info): Read move costs.
14091         (ipa_update_after_lto_read): Do not recompute node params.
14092         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
14093         (ipa_get_param): Check we are not in WPA.
14094         (ipa_get_param_move_cost): New.
14095         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
14096         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
14097         parm numbers to be present.
14099 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14101         PR rtl-optimization/58048
14102         * lra-constraints.c (process_alt_operands): Don't check asm
14103         operand on register.
14105 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
14107         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
14108         the implied StoreLoad barrier for atomic operations if before.
14110 2013-08-02  Jan Hubicka  <jh@suse.cz>
14111             Martin Liska  <marxin.liska@gmail.com>
14113         * cgraph.c (cgraph_function_body_availability): Do not check
14114         cgraph flags.
14115         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
14116         symtab_node_availability): Declare.
14117         * ipa.c (can_replace_by_local_alias): New.
14118         (function_and_variable_visibility): Use it.
14119         * symtab.c (symtab_for_node_and_aliases,
14120         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
14122 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14124         PR rtl-optimization/57963
14125         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
14126         (lra_constraints): Use them.
14128 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
14130         * config/arm/types.md (define_attr "type"): Add "load_acq"
14131         and "store_rel".
14132         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
14133         changes.
14134         (cortex_a53_store1): Likewise.
14136 2013-08-01  Jan Hubicka  <jh@suse.cz>
14138         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
14139         partitions are not needed.
14141 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
14143         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
14144         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
14145         MAYBE_NON_Q_CLASS_P where appropriate.
14147 2013-08-01  Jan Hubicka  <jh@suse.cz>
14149         * cgraph.h (release_function_body): Declare.
14150         * tree.c (free_lang_data_in_decl): Free, parameters and return values
14151         of unused delcarations.
14153 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14155         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
14156         RTL form when subtracting a constant.
14158 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14160         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
14161         Generate canonical plus rtx with negated immediate instead of minus
14162         where appropriate.
14163         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
14165 2013-08-01  Jan Hubicka  <jh@suse.cz>
14167         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
14168         (cgraph_release_function_body): Likewise.
14169         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
14170         * cgraph.h (cgrpah_node): Rename abstract_and_needed
14171         to used_as_abstract_origin.
14172         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
14173         symbols used as abstract origins.
14174         * cgraphunit.c (analyze_functions): Update.
14175         * ipa.c (symtab_remove_unreachable_nodes): Recompute
14176         used_as_abstract_origin.
14177         * tree-inline.c (tree_function_versioning): Update
14178         used_as_abstract_origin; be ready for DECL_RESULT and
14179         DECL_ARGUMENTS to be NULL.
14181         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
14182         for abstract functions.
14183         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
14184         real symbols.
14186 2013-08-01  Jan Hubicka  <jh@suse.cz>
14188         * profile.c (compute_value_histograms): Fix thinko.
14190 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14192         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
14193         aarch-common-protos.h to extra_headers.
14194         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14195         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
14196         * config/aarch64/t-aarch64 (aarch-common.o): Define.
14198 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14200         * config/aarch64/aarch64.md (define_attr "type"): Delete.
14201         Include "../arm/types.md".  Define "type" attribute for all patterns.
14202         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
14203         attribute changes.
14205 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
14207         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
14208         to support power8 load fusion.
14209         (fusion_gpr_mem_load): Likewise.
14211         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
14213         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
14214         declarations for power8 load fusion.
14215         (emit_fusion_gpr_load): Likewise.
14217         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
14218         tuning for power8, turn on fusion mode by default.  Turn on sign
14219         extending fusion mode if normal fusion mode is on, and we are at
14220         -O2 or -O3.
14221         (fusion_gpr_load_p): New function, return true if we can fuse an
14222         addis instruction with a dependent load to a GPR.
14223         (emit_fusion_gpr_load): Emit the instructions for power8 load
14224         fusion to GPRs.
14226         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
14227         (VSX load fusion peepholes): New peepholes to fuse together an
14228         addi instruction with a VSX load instruction.
14230         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
14231         peepholes to fuse an addis instruction with a load to a GPR base
14232         register.  If we are supporting sign extending fusions, convert
14233         sign extending loads to zero extending loads and add an explicit
14234         sign extension.
14236 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14238         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
14239         aarch-common-protos.h to extra_headers.
14240         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14241         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
14242         (arm_early_store_addr_dep): Likewise.
14243         (arm_no_early_alu_shift_dep): Likewise.
14244         (arm_no_early_alu_shift_value_dep): Likewise.
14245         (arm_no_early_mul_dep): Likewise.
14246         (arm_no_early_store_addr_dep): Likewise.
14247         (arm_mac_accumulator_is_mul_result): Likewise.
14248         (arm_mac_accumulator_is_result): Likewise.
14249         * config/arm/aarch-common.c: ... here.  New file.
14250         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
14251         here to ...
14252         (arm_early_store_addr_dep): Likewise.
14253         (arm_no_early_alu_shift_dep): Likewise.
14254         (arm_no_early_alu_shift_value_dep): Likewise.
14255         (arm_no_early_mul_dep): Likewise.
14256         (arm_no_early_store_addr_dep): Likewise.
14257         (arm_mac_accumulator_is_mul_result): Likewise.
14258         (arm_mac_accumulator_is_result): Likewise.
14259         * config/arm/aarch-common-protos.h: ... here.  New file.
14260         * config/arm/t-arm (aarch-common.o): Define.
14262 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14264         * config/arm/arm.md: Include new file "types.md".
14265         (define_attr "type"): Move from here to ...
14266         (define_attr "mul32"): Likewise.
14267         (define_attr "mul64"): Likewise.
14268         * config/arm/types.md: ... here.  New file.
14270 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14272         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
14273         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
14275 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
14277         * gen-pass-instances.awk: Fix offset of substr().
14279 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
14281         * Makefile.in (pass-instances.def): New.
14282         (passes.o): Replace dependency on passes.def with one on
14283         pass-instances.def
14285         * gen-pass-instances.awk: New.
14287         * passes.c (pass_manager::pass_manager): Use pass-instances.def
14288         rather than passes.def, updating local definition of NEXT_PASS
14289         macro to add an extra NUM parameter (currently unused).
14291 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
14293         * Makefile.in (PASS_MANAGER_H): New.
14294         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
14295         (passes.o): Likewise.
14296         (statistics.o): Likewise.
14297         (cgraphunit.o): Likewise.
14298         (context.o): Depend on PASS_MANAGER_H.
14300         * pass_manager.h: New.
14302         * cgraphunit.c (cgraph_add_new_function): Update for moves
14303         of globals to fields of pass_manager.
14304         (analyze_function): Likewise.
14305         (expand_function): Likewise.
14306         (ipa_passes): Likewise.
14307         (compile): Likewise.
14309         * context.c (context::context): New.
14310         * context.h  (context::context): New.
14311         (context::get_passes): New.
14312         (context::passes_): New.
14314         * lto-cgraph.c (input_node): Update for moves of globals to
14315         fields of pass_manager.
14317         * passes.c (all_passes): Remove, in favor of a field of the
14318         same name within the new class pass_manager.
14319         (all_small_ipa_passes): Likewise.
14320         (all_lowering_passes): Likewise.
14321         (all_regular_ipa_passes): Likewise.
14322         (all_late_ipa_passes): Likewise.
14323         (all_lto_gen_passes): Likewise.
14324         (passes_by_id): Likewise.
14325         (passes_by_id_size): Likewise.
14326         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14327         the new class pass_manager.
14328         (set_pass_for_id): Convert to...
14329         (pass_manager::set_pass_for_id): ...method.
14330         (get_pass_for_id): Convert to...
14331         (pass_manager::get_pass_for_id): ...method.
14332         (register_one_dump_file): Move body of implementation into...
14333         (pass_manager::register_one_dump_file): ...here.
14334         (register_dump_files_1): Convert to...
14335         (pass_manager::register_dump_files_1): ...method.
14336         (register_dump_files): Convert to...
14337         (pass_manager::register_dump_files): ...method.
14338         (create_pass_tab): Update for moves of globals to fields of
14339         pass_manager.
14340         (dump_passes): Move body of implementation into...
14341         (pass_manager::dump_passes): ...here.
14342         (register_pass): Move body of implementation into...
14343         (pass_manager::register_pass): ...here.
14344         (init_optimization_passes): Convert into...
14345         (pass_manager::pass_manager): ...constructor for new
14346         pass_manager class, and initialize the pass_lists array.
14347         (check_profile_consistency): Update for moves of globals to
14348         fields of pass_manager.
14349         (dump_profile_report): Move body of implementation into...
14350         (pass_manager::dump_profile_report): ...here.
14351         (ipa_write_summaries_1): Update for moves of pass lists from
14352         being globals to fields of pass_manager.
14353         (ipa_write_optimization_summaries): Likewise.
14354         (ipa_read_summaries):  Likewise.
14355         (ipa_read_optimization_summaries): Likewise.
14356         (execute_all_ipa_stmt_fixups): Likewise.
14358         * statistics.c (statistics_fini): Update for moves of globals to
14359         fields of pass_manager.
14361         * toplev.c (general_init): Replace call to
14362         init_optimization_passes with construction of the pass_manager
14363         instance.
14365         * tree-pass.h (all_passes): Remove, in favor of a field of the
14366         same name within the new class pass_manager.
14367         (all_small_ipa_passes): Likewise.
14368         (all_lowering_passes): Likewise.
14369         (all_regular_ipa_passes): Likewise.
14370         (all_lto_gen_passes): Likewise.
14371         (all_late_ipa_passes): Likewise.
14372         (passes_by_id): Likewise.
14373         (passes_by_id_size): Likewise.
14374         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14375         the new class pass_manager.
14376         (get_pass_for_id): Remove.
14378 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14380         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
14381         configs.
14383 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14385         * arm.md (mulhi3): New expand pattern.
14387 2013-07-30  Jan Hubicka  <jh@suse.cz>
14388             Martin Liska  <marxin.liska@gmail.com>
14390         * profile.c (compute_value_histograms): Do not ICE when
14391         there is mismatch only on some counters.
14393 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
14395         PR rtl-optimization/57637
14396         * function.c (move_insn_for_shrink_wrap): Also check the
14397         GEN set of the LIVE problem for the liveness analysis
14398         if it exists, otherwise give up.
14400 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14402         PR tree-optimization/57993
14403         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
14404         replaced statement in the candidate table.
14405         (phi_add_costs): Return infinite cost when the hidden basis does
14406         not dominate all phis on which the candidate is dependent.
14407         (replace_one_candidate): Record replaced statement in the
14408         candidate table.
14410 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
14412         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
14413         (ashlv2si3): New expander.
14414         (*ashlv2si3_i): New define_insn_and_split.
14415         * predicates.md (float_operation): Allow patterns with three
14416         basic sub-patterns.
14418         PR rtl-optimization/58021
14419         * mode-switching.c (create_pre_exit): Always split off preceding
14420         insns if we are not at the basic block head.
14422 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
14424         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
14425         (UCLIBC_DYNAMIC_LINKER): New macro.
14426         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
14427         `-mnan=2008'.
14428         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
14429         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
14430         `-mnan=2008'.
14431         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
14432         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
14433         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
14434         for SF and DF modes.  Use ieee_quad_format for TF mode.
14435         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
14436         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
14437         (mips_option_override): Handle `-mnan=legacy'.
14438         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
14439         `-mabs=2008' and `-mnan=2008'.
14440         (OPTION_DEFAULT_SPECS): Add "nan" default.
14441         (ASM_SPEC): Handle `-mnan='.
14442         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
14443         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
14444         comment accordingly.
14445         (neg<mode>2): Likewise.
14446         * config/mips/mips.opt (mabs, mnan): New options.
14447         * doc/install.texi (Configuration): Document `--with-nan=' option.
14448         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
14449         `-mnan=' options.
14450         (MIPS Options): Document them.
14451         * config.gcc <mips*-*-*>: Handle `--with-nan='.
14452         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
14453         * configure: Regenerate.
14454         * config.in: Regenerate.
14456 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14458         * config/i386/i386.md (float post-reload splitters): Do not check
14459         for subregs of SSE registers.
14461 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14462             H.J. Lu  <hongjiu.lu@intel.com>
14464         PR target/57954
14465         PR target/57988
14466         * config/i386/i386.md (post-reload splitter
14467         to avoid partial SSE reg dependency stalls): New pattern.
14469 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
14471         * config/s390/s390.md ("movcc"): Swap load and store instructions.
14473 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
14475         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
14476         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
14478 2013-07-26  Cary Coutant  <ccoutant@google.com>
14480         * dwarf2out.c (die_checksum_ordered): Don't include template
14481         instantiations in signature.
14482         (is_template_parameter): New function.
14483         (is_template_instantiation): New function.
14484         (generate_skeleton_bottom_up): Don't include template instantiations
14485         in type unit DIE.
14486         (generate_skeleton): Likewise.
14487         (break_out_comdat_types): Move recursive call to break out nested
14488         types earlier.
14489         (prune_unused_types_mark_generic_parms_dies): Call
14490         is_template_parameter.
14492 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
14494         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
14495         uses vector registers.
14496         * config/aarch64/iterators.md: Add attributes rtn and vas.
14498 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14499             Richard Earnshaw  <richard.earnshaw@arm.com>
14501         * combine.c (simplify_comparison): Re-canonicalize operands
14502         where appropriate.
14503         * config/arm/arm.md (movcond_addsi): New splitter.
14505 2013-07-25  Sterling Augustine  <saugustine@google.com>
14507         * dwarf2out.c (size_of_pubnames): Move code to...
14508         (include_pubname_in_output): ...here.  New.
14509         (want_pubnames): Rearrange.
14510         (output_pubnames): Call include_pubname_in_output.  Move assertion.
14512 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14514         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
14516 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14518         PR target/38836
14519         * doc/extend.texi: Remove obsolete builtins. Fix
14520         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
14522 2013-07-25  Jan Hubicka  <jh@suse.cz>
14524         * cgraph.c (release_function_body): Break out from ...
14525         (cgraph_release_function_body): ... this one; also release DECL_RESULT
14526         and DECL_ARGUMENTS.
14527         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
14528         old_tree in the map.
14529         (create_specialized_node): Update.
14530         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
14531         into index.
14532         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
14533         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
14534         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
14535         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
14536         DECL_RESULT.
14538 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14540         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
14541         addsi3_carryin_alt2_<optab>): Correct output template.
14543 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14545         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
14546         Adjust for arm_restrict_it.
14547         Remove trailing whitespace.
14549 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
14551         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
14552         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
14554         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
14556 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
14558         PR rtl-optimization/57960
14559         * lra-constraints.c (process_alt_operands): Use the right mode
14560         when checking strict_low.
14562 2013-07-25  Jan Hubicka  <jh@suse.cz>
14564         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
14565         * cgraph.c (cgraph_remove_node): Do not release function body
14566         when in cgraph streaming.
14567         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
14568         in other partitions are not considered reachable; fix handling of
14569         clones.
14571 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14573         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
14575 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14577         PR target/19599
14578         PR target/57731
14579         PR target/57837
14580         * config/arm/arm.md ("*sibcall_insn): Replace use of
14581         Ss with US. Adjust output for v5 and v4t.
14582         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
14584         * config/arm/constraints.md ("Ss"): Rename to US.
14586 2013-07-25  Terry Guo  <terry.guo@arm.com>
14588         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
14589         shift_add/shift_sub0/shift_sub1 RTXs.
14591 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
14592             Anton Blanchard  <anton@au1.ibm.com>
14594         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
14595         (altivec_vpks<VI_char>ss): Likewise.
14596         (altivec_vpks<VI_char>us): Likewise.
14597         (altivec_vpku<VI_char>us): Likewise.
14598         (altivec_vpku<VI_char>um): Likewise.
14600 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
14602         Introduce context class.
14604         * Makefile.in (CONTEXT_H): New.
14605         (OBJS): Add context.o.
14606         (toplev.o): Add CONTEXT_H to dependencies.
14607         (context.o): New.
14609         * toplev.c (general_init): Create the singleton gcc::context instance.
14611         * context.c: New.
14613         * context.h: New.
14615 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
14617         PR rtl-optimization/57968
14618         * mode-switching.c (create_pre_exit): Allow instructions that
14619         don't set a return register to need a non-exit mode.
14621 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14622             Anton Blanchard  <anton@au1.ibm.com>
14624         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
14625         operands to vperm for little endian.
14626         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
14627         of lvsl to create the control mask for a vperm for little endian.
14629 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14630             Anton Blanchard  <anton@au1.ibm.com>
14632         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
14633         two operands for little-endian.
14635 2013-07-23  Steve Ellcey  <sellcey@mips.com>
14637         * config/mips/mips.c (mips_case_values_threshold): New.
14638         (TARGET_CASE_VALUES_THRESHOLD): Define.
14640 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14641             Anton Blanchard  <anton@au1.ibm.com>
14643         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
14644         selection of field for vector splat in little endian mode.
14646 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
14648         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
14649         expanders to rs6000.md.
14650         (ior<mode>3): Likewise.
14651         (and<mode>3): Likewise.
14652         (one_cmpl<mode>2): Likewise.
14653         (nor<mode>3): Likewise.
14654         (andc<mode>3): Likewise.
14655         (eqv<mode>3): Likewise.
14656         (nand<mode>3): Likewise.
14657         (orc<mode>3): Likewise.
14659         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
14660         declaration.
14662         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
14663         to split multi-word logical operations.
14664         (rs6000_split_logical_di): Likewise.
14665         (rs6000_split_logical): Likewise.
14667         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
14668         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
14669         and allow TImode operations in 32-bit.
14670         (vsx_and<mode>3_64bit): Likewise.
14671         (vsx_ior<mode>3_32bit): Likewise.
14672         (vsx_ior<mode>3_64bit): Likewise.
14673         (vsx_xor<mode>3_32bit): Likewise.
14674         (vsx_xor<mode>3_64bit): Likewise.
14675         (vsx_one_cmpl<mode>2_32bit): Likewise.
14676         (vsx_one_cmpl<mode>2_64bit): Likewise.
14677         (vsx_nor<mode>3_32bit): Likewise.
14678         (vsx_nor<mode>3_64bit): Likewise.
14679         (vsx_andc<mode>3_32bit): Likewise.
14680         (vsx_andc<mode>3_64bit): Likewise.
14681         (vsx_eqv<mode>3_32bit): Likewise.
14682         (vsx_eqv<mode>3_64bit): Likewise.
14683         (vsx_nand<mode>3_32bit): Likewise.
14684         (vsx_nand<mode>3_64bit): Likewise.
14685         (vsx_orc<mode>3_32bit): Likewise.
14686         (vsx_orc<mode>3_64bit): Likewise.
14688         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
14689         logical types in GPRs.
14691         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
14692         logical insns to rs6000.md, and allow TImode operations in
14693         32-bit.
14694         (altivec_ior<mode>3): Likewise.
14695         (altivec_xor<mode>3): Likewise.
14696         (altivec_one_cmpl<mode>2): Likewise.
14697         (altivec_nor<mode>3): Likewise.
14698         (altivec_andc<mode>3): Likewise.
14700         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
14701         attributes for moving the 128-bit logical operations into
14702         rs6000.md.
14703         (BOOL_REGS_OUTPUT): Likewise.
14704         (BOOL_REGS_OP1): Likewise.
14705         (BOOL_REGS_OP2): Likewise.
14706         (BOOL_REGS_UNARY): Likewise.
14707         (BOOL_REGS_AND_CR0): Likewise.
14708         (one_cmpl<mode>2): Add support for DI logical operations on
14709         32-bit, splitting the operations to 32-bit.
14710         (anddi3): Likewise.
14711         (iordi3): Likewise.
14712         (xordi3): Likewise.
14713         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
14714         changes to combine the 32/64-bit code, allow logical operations on
14715         TI mode in 32-bit, and to use similar match_operator patterns like
14716         scalar mode uses.  Combine the Altivec and VSX code for logical
14717         operations, and move it here.
14718         (ior<mode>3, 128-bit types): Likewise.
14719         (xor<mode>3, 128-bit types): Likewise.
14720         (one_cmpl<mode>3, 128-bit types): Likewise.
14721         (nor<mode>3, 128-bit types): Likewise.
14722         (andc<mode>3, 128-bit types): Likewise.
14723         (eqv<mode>3, 128-bit types): Likewise.
14724         (nand<mode>3, 128-bit types): Likewise.
14725         (orc<mode>3, 128-bit types): Likewise.
14726         (and<mode>3_internal): Likewise.
14727         (bool<mode>3_internal): Likewise.
14728         (boolc<mode>3_internal1): Likewise.
14729         (boolc<mode>3_internal2): Likewise.
14730         (boolcc<mode>3_internal1): Likewise.
14731         (boolcc<mode>3_internal2): Likewise.
14732         (eqv<mode>3_internal1): Likewise.
14733         (eqv<mode>3_internal2): Likewise.
14734         (one_cmpl1<mode>3_internal): Likewise.
14736 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14738         * config/microblaze/microblaze.c (microblaze_expand_prologue):
14739         Rename flag_stack_usage to flag_stack_usage_info.
14741 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14743         * config/microblaze/sync.md: New file.
14744         * config/microblaze/microblaze.md: Include sync.md
14745         * config/microblaze/microblaze.c: Add print_operand 'y'.
14746         * config/microblaze/constraints.md: Add memory_contraint
14747         'Q' which is a single register.
14749 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
14751         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
14753 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14755         * reload.c (find_reloads): Exit loop once we find this operand
14756         cannot be reloaded somehow for this alternative.
14758         * reload.c (find_reloads): Exit loop once we find a hard register.
14760         * rtlanal.c (computed_jump_p): Exit loop once we find label
14761         reference is used.
14763         * i386.c (ix86_pad_returns): Exit loop after setting replace.
14765         * cfgloopmanip.c (remove_path): Exit loop after setting
14766         irred_invalidated.
14768         * gensupport.c (subst_dup): Avoid loop if code is not
14769         MATCH_DUP nor MATCH_OP_DUP.
14771 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
14773         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
14775 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14777         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
14778         true for SP_REGNUM if mode == ptr_mode.
14779         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
14780         with value R0_REGNUM + 31.
14782 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14784         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
14785         pad pointer-typed argument downward.
14787 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14789         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
14790         and __ILP32__ when the ILP32 model is in use.
14792 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14794         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
14795         (aarch64_load_symref_appropriately): In the case of
14796         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
14797         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
14798         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
14799         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
14800         if 'mode' doesn't equal to 'ptr_mode'.
14801         (aarch64_output_mi_thunk): Add an assertion on the alignment of
14802         'vcall_offset'; change to call aarch64_emit_move differently depending
14803         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
14804         to calculate the upper bound of 'vcall_offset'.
14805         (aarch64_cannot_force_const_mem): Change to also return true if
14806         mode != ptr_mode.
14807         (aarch64_legitimize_reload_address): In the case of large
14808         displacements, add new local variable 'xmode' and an assertion
14809         based on it; change to use 'xmode' to generate the new rtx and
14810         reload.
14811         (aarch64_asm_trampoline_template): Change to generate the template
14812         differently depending on TARGET_ILP32 or not; change to use
14813         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
14814         (aarch64_trampoline_size): Removed.
14815         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
14816         and replace immediate literals with it.  Change to use 'ptr_mode'
14817         instead of 'DImode' and call convert_memory_address if the mode
14818         of 'fnaddr' doesn't equal to 'ptr_mode'.
14819         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
14820         to output symbol.
14821         (aarch64_elf_asm_destructor): Likewise.
14822         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
14823         on TARGET_ILP32 instead of aarch64_trampoline_size.
14824         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
14825         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
14826         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
14827         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
14828         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
14829         (storewb_pair<GPI:mode>_<P:mode>): ... this.
14830         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
14831         depending on the value of 'mode'.
14832         (add_losym_<mode>): New.
14833         (ldr_got_small_<mode>): New, based on ldr_got_small.
14834         (ldr_got_small): Remove.
14835         (ldr_got_small_sidi): New.
14836         * config/aarch64/iterators.md (P): New.
14837         (PTR): Change to 'ptr_mode' in the condition.
14839 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14841         * config.gcc (aarch64*-*-*): Support --with-abi.
14842         (aarch64*-*-elf): Support --with-multilib-list.
14843         (aarch64*-*-linux*): Likewise.
14844         (supported_defaults): Add abi to aarch64*-*-*.
14845         * configure.ac: Mention AArch64 for --with-multilib-list.
14846         * configure: Re-generated.
14847         * config/aarch64/biarchilp32.h: New file.
14848         * config/aarch64/biarchlp64.h: New file.
14849         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
14850         (ABI_SPEC): Ditto.
14851         (MULTILIB_DEFAULTS): Ditto.
14852         (DRIVER_SELF_SPECS): Ditto.
14853         (ASM_SPEC): Update to also substitute -mabi.
14854         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
14855         file whose name depends on -mabi= and -mbig-endian.
14856         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
14857         TARGET_ILP32.
14858         (POINTER_SIZE): New define.
14859         (POINTERS_EXTEND_UNSIGNED): Ditto.
14860         (enum aarch64_abi_type): New enumeration tag.
14861         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
14862         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
14863         (TARGET_ILP32): New define.
14864         * config/aarch64/aarch64.opt (mabi): New.
14865         (aarch64_abi): New.
14866         (ilp32, lp64): New values for -mabi.
14867         * config/aarch64/t-aarch64 (comma): New define.
14868         (MULTILIB_OPTIONS): Ditto.
14869         (MULTILIB_DIRNAMES): Ditto.
14870         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
14871         * doc/invoke.texi: Document -mabi for AArch64.
14873 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
14875         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
14877 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14878             Anton Blanchard  <anton@au1.ibm.com>
14880         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
14881         endianness when selecting field to splat.
14883 2013-07-22  Eric Christopher  <echristo@gmail.com>
14885         * dwarf2out.c (die_odr_checksum): New function to use
14886         CHECKSUM_ macros and ULEB128 for DIE tag.
14887         (generate_type_signature): Use.
14889 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
14891         * config.gcc (sparc*-*-*): Accept leon3 processor.
14892         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
14893         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
14894         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
14895         * config/sparc/sparc.opt (enum processor_type): Add leon3.
14896         (mfix-ut699): Adjust comment.
14897         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
14898         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
14899         (CPP_CPU_SPEC): Likewise.
14900         (ASM_CPU_SPEC): Likewise.
14901         * config/sparc/sparc.c (leon3_cost): New constant.
14902         (sparc_option_override): Add leon3 support.
14903         (mem_ref): New function.
14904         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
14905         (sparc_do_work_around_errata): Look into the instruction in the delay
14906         slot and adjust accordingly.  Add fix for the data cache nullify issues
14907         of the UT699.  Change insertion position for the NOP.
14908         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
14909         (leon3_load): New reservation.
14910         (leon_store): Bump latency to 2.
14911         (grfpu): New automaton.
14912         (grfpu_alu): New unit.
14913         (grfpu_ds): Likewise.
14914         (leon_fp_alu): Adjust.
14915         (leon_fp_mult): Delete.
14916         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
14917         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
14918         * config/sparc/sparc.md (cpu): Add leon3.
14919         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
14920         (swapsi): Likewise.
14921         (atomic_test_and_set): Likewise.
14922         (ldstub): Likewise.
14924 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
14926         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
14927         default for R5900 targets.
14928         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
14929         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
14930         * config/mips/mips.c (mips_option_override): Report an error for
14931         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
14932         for -march=r5900 -mhard-float.
14934 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14936         * df-problems.c (can_move_insns_across): Exit loop once we
14937         find a non-fixed, non-global register.
14939         * ipa-pure-const.c (propagate_nothrow): Exit loop after
14940         setting can_throw.
14942         * omega.c (omega_eliminate_red): Break after setting red_found.
14943         (omega_problem_has_red_equations): Similarly after setting found.
14944         (omega_query_variable): Similarly after setting coupled.
14946 2013-07-22  Marek Polacek  <polacek@redhat.com>
14948         * gimplify.c: Don't include gimple.h twice.
14950 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14952         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
14953         instead of TARGET_THUMB1.
14954         (Pz): New constraint.
14955         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
14956         encodings.
14957         (compare_negsi_si): Likewise.
14958         (compare_addsi2_op0): Likewise.
14959         (compare_addsi2_op1): Likewise.
14960         (addsi3_carryin_<optab>): Likewise.
14961         (addsi3_carryin_alt2_<optab>): Likewise.
14962         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
14963         for arm_restrict_it.
14964         (subsi3_carryin): Likewise.
14965         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
14966         (minmax_arithsi): Disable for arm_restrict_it.
14967         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
14968         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
14969         (satsi_<SAT:code>_shift): Likewise.
14970         (arm_shiftsi3): Add alternative for 16-bit encoding.
14971         (arm32_movhf): Disable for arm_restrict_it.
14972         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
14973         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
14975 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
14977         * config/arm/arm.md (attribute "insn"): Delete.
14978         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
14979         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
14980         (not_shiftsi): Update for attribute change.
14981         (not_shiftsi_compare0): Likewise.
14982         (not_shiftsi_compare0_scratch): Likewise.
14983         (arm_one_cmplsi2): Likewise.
14984         (thumb1_one_cmplsi2): Likewise.
14985         (notsi_compare0): Likewise.
14986         (notsi_compare0_scratch): Likewise.
14987         (thumb1_movdi_insn): Likewise.
14988         (arm_movsi_insn): Likewise.
14989         (movhi_insn_arch4): Likewise.
14990         (movhi_bytes): Likewise.
14991         (arm_movqi_insn): Likewise.
14992         (thumb1_movqi_insn): Likewise.
14993         (arm32_movhf): Likewise.
14994         (thumb1_movhf): Likewise.
14995         (arm_movsf_soft_insn): Likewise.
14996         (thumb1_movsf_insn): Likewise.
14997         (thumb_movdf_insn): Likewise.
14998         (movsicc_insn): Likewise.
14999         (movsfcc_soft_insn): Likewise.
15000         (and_scc): Likewise.
15001         (cond_move): Likewise.
15002         (if_move_not): Likewise.
15003         (if_not_move): Likewise.
15004         (if_shift_move): Likewise.
15005         (if_move_shift): Likewise.
15006         (if_shift_shift): Likewise.
15007         (if_not_arith): Likewise.
15008         (if_arith_not): Likewise.
15009         (cond_move_not): Likewise.
15010         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
15011         (neon_mov<mode>): Likewise.
15012         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
15013         (thumb2_movsi_vfp): Likewise.
15014         (movsf_vfp): Likewise.
15015         (thumb2_movsf_vfp): Likewise.
15016         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15017         change.
15018         (cortexa7_older_only): Likewise.
15019         (cortexa7_younger): Likewise.
15020         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
15021         (1020alu_shift_op): Likewise.
15022         (1020alu_shift_reg_op): Likewise.
15023         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
15024         (alu_shift_op): Likewise.
15025         (alu_shift_reg_op): Likewise.
15026         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
15027         (11_alu_shift_op): Likewise.
15028         (11_alu_shift_reg_op): Likewise.
15029         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
15030         (9_alu_shift_reg_op): Likewise.
15031         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15032         change.
15033         (cortex_a15_alu_shift): Likewise.
15034         (cortex_a15_alu_shift_reg): Likewise.
15035         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
15036         (cortex_a5_alu_shift): Likewise.
15037         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
15038         change.
15039         (cortex_a53_alu_shift): Likewise.
15040         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15041         change.
15042         (cortex_a7_alu_reg): Likewise.
15043         (cortex_a7_alu_shift): Likewise.
15044         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15045         (cortex_a8_alu_shift): Likewise.
15046         (cortex_a8_alu_shift_reg): Likewise.
15047         (cortex_a8_mov): Likewise.
15048         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
15049         (cortex_a9_dp_shift): Likewise.
15050         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
15051         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
15052         (cortex_r4_mov): Likewise.
15053         (cortex_r4_alu_shift): Likewise.
15054         (cortex_r4_alu_shift_reg): Likewise.
15055         * config/arm/fa526.md (526_alu_op): Update for attribute change.
15056         (526_alu_shift_op): Likewise.
15057         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
15058         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
15059         (626te_alu_shift_op): Likewise.
15060         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
15061         (726te_alu_op): Likewise.
15062         (726te_alu_shift_op): Likewise.
15063         (726te_alu_shift_reg_op): Likewise.
15064         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
15065         (mp626_alu_shift_op): Likewise.
15066         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
15067         (pj4_alu_e1_conds): Likewise.
15068         (pj4_alu): Likewise.
15069         (pj4_alu_conds): Likewise.
15070         (pj4_shift): Likewise.
15071         (pj4_shift_conds): Likewise.
15072         (pj4_alu_shift): Likewise.
15073         (pj4_alu_shift_conds): Likewise.
15075 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15077         * config/arm/predicates.md (shiftable_operator_strict_it):
15078         New predicate.
15079         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
15080         Disable cond_exec version for arm_restrict_it.
15081         (thumb2_smaxsi3): Convert to generate cond_exec.
15082         (thumb2_sminsi3): Likewise.
15083         (thumb32_umaxsi3): Likewise.
15084         (thumb2_uminsi3): Likewise.
15085         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
15086         (thumb2_neg_abssi2): Likewise.
15087         (thumb2_mov_scc): Add alternative for 16-bit encoding.
15088         (thumb2_movsicc_insn): Adjust alternatives.
15089         (thumb2_mov_negscc): Disable for arm_restrict_it.
15090         (thumb2_mov_negscc_strict_it): New pattern.
15091         (thumb2_mov_notscc_strict_it): New pattern.
15092         (thumb2_mov_notscc): Disable for arm_restrict_it.
15093         (thumb2_ior_scc): Likewise.
15094         (thumb2_ior_scc_strict_it): New pattern.
15095         (thumb2_cond_move): Adjust for arm_restrict_it.
15096         (thumb2_cond_arith): Disable for arm_restrict_it.
15097         (thumb2_cond_arith_strict_it): New pattern.
15098         (thumb2_cond_sub): Adjust for arm_restrict_it.
15099         (thumb2_movcond): Likewise.
15100         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
15101         (thumb2_zero_extendhisi2_v6): Likewise.
15102         (thumb2_zero_extendqisi2_v6): Likewise.
15103         (orsi_notsi_si): Likewise.
15104         (orsi_not_shiftsi_si): Likewise.
15106 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
15108         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
15109         instruction sequence is 1 byte shorter.
15111 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
15113         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
15114         it is not needed after split.
15116 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
15118         PR target/51784
15119         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
15120         second label for nonlocal goto receivers. Don't output pic base labels
15121         unless we're producing PIC; mark that action unreachable().
15122         (ix86_save_reg): If the function contains a nonlocal label, save the
15123         PIC base reg.
15124         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
15125         * config/darwin.c (emitted_pic_label_num): New GTY.
15126         (update_pic_label_number_if_needed): New.
15127         (machopic_output_function_base_name): Adjust for nonlocal receiver
15128         case.
15129         (machopic_should_output_picbase_label): New.
15130         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
15131         (nonlocal_goto_receiver): New insn and split.
15133 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
15135         * config/aarch64/aarch64-builtins.c
15136         (aarch64_fold_builtin): Fold abs in all modes.
15137         * config/aarch64/aarch64-simd-builtins.def
15138         (abs): Enable for all modes.
15139         * config/aarch64/arm_neon.h
15140         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
15141         (vabs_f64): Add missing intrinsic.
15143 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
15145         * config/aarch64/arm_neon.h (vabs_s64): New function
15147 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
15149         PR target/57516
15150         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
15151         * config/avr/avr.md (adjust_len): Add `round'.
15152         * config/avr/avr-protos.h (avr_out_round): New prototype.
15153         (avr_out_plus): Add `out_label' argument.
15154         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
15155         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
15156         Handle the case where `insn' is just a pattern.
15157         (avr_out_bitop): Handle the case where `insn' is just a pattern.
15158         (avr_out_round): New function.
15159         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
15161 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
15163         * config/microblaze/microblaze.c (microblaze_expand_prologue):
15164         Add check for flag_stack_usage to handle -fstack-usage support
15166 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
15168         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
15169         interaction for new Power8 flags and VSX.
15171 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
15173         PR middle-end/57698
15174         * tree-inline.c (expand_call_inline): Emit errors during
15175         early_inlining only if optimization is not turned on.
15177 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15179         * passes.def: New.
15181         * passes.c (init_optimization_passes): Move the construction of
15182         the pass hierarchy into a new passes.def file.
15184         * Makefile.in (passes.o): Add dependency on passes.def.
15186 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15188         * passes.c (init_optimization_passes): Introduce macros for
15189         constructing the tree of passes (INSERT_PASSES_AFTER,
15190         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
15191         TERMINATE_PASS_LIST).
15193 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
15194             Wei Mi  <wmi@google.com>
15196         PR rtl-optimization/57878
15197         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
15198         top.
15199         (reload_pseudo_compare_func): Check nregs first for reload
15200         pseudos.
15202 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15204         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
15206 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
15208         * read-rtl.c (validate_const_int): Once an invalid character is
15209         seen, quit the loop.
15211         * gengtype.c (write_roots): Similarly once we find the "deletable"
15212         or "if_marked" option.
15214 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15216         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
15217         "xtab" and "sat".  Move value "clz" from here to ...
15218         (attriubte "type"): ... here.
15219         (satsi_<SAT:code>): Delete "insn" attribute.
15220         (satsi_<SAT:code>_shift): Likewise.
15221         (arm_zero_extendqisi2addsi): Likewise.
15222         (arm_extendqisi2addsi): Likewise.
15223         (clzsi2): Update for attribute changes.
15224         (rbitsi2): Likewise.
15225         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
15226         attribute.
15227         (arm_usatsihi): Likewise.
15228         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15230 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15232         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
15233         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
15234         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
15235         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
15236         in alphabetical order.
15237         (attribute "core_cycles"): Update for attribute changes.
15238         (arm_addsi3): Likewise.
15239         (addsi3_compare0): Likewise.
15240         (addsi3_compare0_scratch): Likewise.
15241         (addsi3_compare_op1): Likewise.
15242         (addsi3_compare_op2): Likewise.
15243         (compare_addsi2_op0): Likewise.
15244         (compare_addsi2_op1): Likewise.
15245         (addsi3_carryin_shift_<optab>): Likewise.
15246         (subsi3_carryin_shift): Likewise.
15247         (rsbsi3_carryin_shift): Likewise.
15248         (arm_subsi3_insn): Likewise.
15249         (subsi3_compare0): Likewise.
15250         (subsi3_compare): Likewise.
15251         (arm_andsi3_insn): Likewise.
15252         (thumb1_andsi3_insn): Likewise.
15253         (andsi3_compare0): Likewise.
15254         (andsi3_compare0_scratch): Likewise.
15255         (zeroextractsi_compare0_scratch
15256         (andsi_not_shiftsi_si): Likewise.
15257         (iorsi3_insn): Likewise.
15258         (iorsi3_compare0): Likewise.
15259         (iorsi3_compare0_scratch): Likewise.
15260         (arm_xorsi3): Likewise.
15261         (thumb1_xorsi3_insn): Likewise.
15262         (xorsi3_compare0): Likewise.
15263         (xorsi3_compare0_scratch): Likewise.
15264         (satsi_<SAT:code>_shift): Likewise.
15265         (rrx): Likewise.
15266         (arm_shiftsi3): Likewise.
15267         (shiftsi3_compare0): Likewise.
15268         (not_shiftsi): Likewise.
15269         (not_shiftsi_compare0): Likewise.
15270         (not_shiftsi_compare0_scratch): Likewise.
15271         (arm_one_cmplsi2): Likewise.
15272         (thumb_one_complsi2): Likewise.
15273         (notsi_compare0): Likewise.
15274         (notsi_compare0_scratch): Likewise.
15275         (thumb1_zero_extendhisi2): Likewise.
15276         (arm_zero_extendhisi2): Likewise.
15277         (arm_zero_extendhisi2_v6): Likewise.
15278         (arm_zero_extendhisi2addsi): Likewise.
15279         (thumb1_zero_extendqisi2): Likewise.
15280         (thumb1_zero_extendqisi2_v6): Likewise.
15281         (arm_zero_extendqisi2): Likewise.
15282         (arm_zero_extendqisi2_v6): Likewise.
15283         (arm_zero_extendqisi2addsi): Likewise.
15284         (thumb1_extendhisi2): Likewise.
15285         (arm_extendhisi2): Likewise.
15286         (arm_extendhisi2_v6): Likewise.
15287         (arm_extendqisi): Likewise.
15288         (arm_extendqisi_v6): Likewise.
15289         (arm_extendqisi2addsi): Likewise.
15290         (thumb1_extendqisi2): Likewise.
15291         (thumb1_movdi_insn): Likewise.
15292         (arm_movsi_insn): Likewise.
15293         (movsi_compare0): Likewise.
15294         (movhi_insn_arch4): Likewise.
15295         (movhi_bytes): Likewise.
15296         (arm_movqi_insn): Likewise.
15297         (thumb1_movqi_insn): Likewise.
15298         (arm32_movhf): Likewise.
15299         (thumb1_movhf): Likewise.
15300         (arm_movsf_soft_insn): Likewise.
15301         (thumb1_movsf_insn): Likewise.
15302         (movdf_soft_insn): Likewise.
15303         (thumb_movdf_insn): Likewise.
15304         (arm_cmpsi_insn): Likewise.
15305         (cmpsi_shiftsi): Likewise.
15306         (cmpsi_shiftsi_swp): Likewise.
15307         (arm_cmpsi_negshiftsi_si): Likewise.
15308         (movsicc_insn): Likewise.
15309         (movsfcc_soft_insn): Likewise.
15310         (arith_shiftsi): Likewise.
15311         (arith_shiftsi_compare0
15312         (arith_shiftsi_compare0_scratch
15313         (sub_shiftsi): Likewise.
15314         (sub_shiftsi_compare0
15315         (sub_shiftsi_compare0_scratch
15316         (and_scc): Likewise.
15317         (cond_move): Likewise.
15318         (if_plus_move): Likewise.
15319         (if_move_plus): Likewise.
15320         (if_move_not): Likewise.
15321         (if_not_move): Likewise.
15322         (if_shift_move): Likewise.
15323         (if_move_shift): Likewise.
15324         (if_shift_shift): Likewise.
15325         (if_not_arith): Likewise.
15326         (if_arith_not): Likewise.
15327         (cond_move_not): Likewise.
15328         (thumb1_ashlsi3): Set type attribute.
15329         (thumb1_ashrsi3): Likewise.
15330         (thumb1_lshrsi3): Likewise.
15331         (thumb1_rotrsi3): Likewise.
15332         (shiftsi3_compare0_scratch): Likewise.
15333         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
15334         (neon_mov<mode>): Likewise.
15335         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
15336         attribute changes.
15337         (thumb2_movsi_insn): Likewise.
15338         (thumb2_cmpsi_neg_shiftsi): Likewise.
15339         (thumb2_extendqisi_v6): Likewise.
15340         (thumb2_zero_extendhisi2_v6): Likewise.
15341         (thumb2_zero_extendqisi2_v6): Likewise.
15342         (thumb2_shiftsi3_short): Likewise.
15343         (thumb2_addsi3_compare0_scratch): Likewise.
15344         (orsi_not_shiftsi_si): Likewise.
15345         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
15346         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
15347         changes.
15348         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
15349         (1020alu_shift_op): Likewise.
15350         (1020alu_shift_reg_op): Likewise.
15351         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
15352         (alu_shift_op): Likewise.
15353         (alu_shift_reg_op): Likewise.
15354         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
15355         (11_alu_shift_op): Likewise.
15356         (11_alu_shift_reg_op): Likewise.
15357         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
15358         (9_alu_shift_reg_op): Likewise.
15359         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15360         changes.
15361         (cortex_a15_alu_shift): Likewise.
15362         (cortex_a15_alu_shift_reg): Likewise.
15363         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
15364         changes.
15365         (cortex_a5_alu_shift): Likewise.
15366         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
15367         changes.
15368         (cortex_a53_alu_shift): Likewise.
15369         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15370         changes.
15371         (cortex_a7_alu_reg): Likewise.
15372         (cortex_a7_alu_shift): Likewise.
15373         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
15374         changes.
15375         (cortex_a8_alu_shift): Likewise.
15376         (cortex_a8_alu_shift_reg): Likewise.
15377         (cortex_a8_mov): Likewise.
15378         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
15379         (cortex_a9_dp_shift): Likewise.
15380         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
15381         changes.
15382         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
15383         changes.
15384         (cortex_r4_mov): Likewise.
15385         (cortex_r4_alu_shift): Likewise.
15386         (cortex_r4_alu_shift_reg): Likewise.
15387         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
15388         (526_alu_shift_op): Likewise.
15389         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
15390         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
15391         (626te_alu_shift_op): Likewise.
15392         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
15393         (726te_alu_op): Likewise.
15394         (726te_alu_shift_op): Likewise.
15395         (726te_alu_shift_reg_op): Likewise.
15396         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
15397         (mp626_alu_shift_op): Likewise.
15398         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
15399         (pj4_alu_e1_conds): Likewise.
15400         (pj4_alu): Likewise.
15401         (pj4_alu_conds): Likewise.
15402         (pj4_shift): Likewise.
15403         (pj4_shift_conds): Likewise.
15404         (pj4_alu_shift): Likewise.
15405         (pj4_alu_shift_conds): Likewise.
15406         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15407         changes.
15408         (cortexa7_older_only): Likewise.
15409         (cortexa7_younger): Likewise.
15411 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15413         * ipa-pure-const.c (generate_summary): Rename to...
15414         (pure_const_generate_summary): ... this.
15416 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
15418         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
15420 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
15422         PR target/57909
15423         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
15424         usage in HI mode.
15426 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15428         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
15429         enabled without -march=zEC12.
15430         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
15431         flags to be set.
15433 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15435         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
15436         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
15437         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
15438         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
15439         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
15440         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
15441         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
15442         ISA_HAS_FP4.
15443         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
15444         and ISA_HAS_NMADD3_NMSUB3.
15445         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
15446         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
15447         (nmsub4<mode>, nmsub3<mode>): Likewise.
15448         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
15450 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15452         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
15453         TARGET_MIPS5400 checking.
15455 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
15456             Peter Bergner  <bergner@vnet.ibm.com>
15458         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
15459         registers in the comment.
15460         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
15461         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
15462         rather than FIRST_PSEUDO_REGISTERS.
15464 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
15466         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15467         enable extra ISA flags with TARGET_HTM.
15469 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15471         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
15472         Fix comment typos.
15474 2013-07-15  Cong Hou  <congh@google.com>
15476         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
15477         in compare function for sorting.
15479 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
15481         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
15482         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
15483         * config/rs6000/rs6000.opt: Add -mhtm option.
15484         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
15485         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
15486         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
15487         __HTM__ if the HTM instructions are available.
15488         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
15489         htm_spr_reg_operand): New define_predicates.
15490         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
15491         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
15492         Include htm.md.
15493         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
15494         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
15495         HTM builtin functions.
15496         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
15497         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
15498         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
15499         (rs6000_builtin_mask_calculate): Likewise.
15500         (rs6000_option_override_internal): Likewise.
15501         (bdesc_htm): Add new HTM builtin support.
15502         (htm_spr_num): New function.
15503         (htm_spr_regno): Likewise.
15504         (rs6000_htm_spr_icode): Likewise.
15505         (htm_expand_builtin): Likewise.
15506         (htm_init_builtins): Likewise.
15507         (rs6000_expand_builtin): Add support for HTM builtin functions.
15508         (rs6000_init_builtins): Likewise.
15509         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
15510         option.
15511         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
15512         (TARGET_HTM, MASK_HTM): Define macros.
15513         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
15514         (FIXED_REGISTERS): Likewise.
15515         (CALL_USED_REGISTERS): Likewise.
15516         (CALL_REALLY_USED_REGISTERS): Likewise.
15517         (REG_ALLOC_ORDER): Likewise.
15518         (enum reg_class): Likewise.
15519         (REG_CLASS_NAMES): Likewise.
15520         (REG_CLASS_CONTENTS): Likewise.
15521         (REGISTER_NAMES): Likewise.
15522         (ADDITIONAL_REGISTER_NAMES): Likewise.
15523         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
15524         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
15525         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
15526         * config/rs6000/htm.md: New file.
15527         * config/rs6000/htmintrin.h: New file.
15528         * config/rs6000/htmxlintrin.h: New file.
15530 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15532         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15533         Define SYMBOL_TINY_GOT, update comment.
15534         * config/aarch64/aarch64.c
15535         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
15536         (aarch64_expand_mov_immediate): Likewise.
15537         (aarch64_print_operand): Likewise.
15538         (aarch64_classify_symbol): Likewise.
15539         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
15540         (ldr_got_tiny): Define.
15542 2013-07-13  Tobias Grosser  <tobias@grosser.es>
15544         PR tree-optimization/54094
15545         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
15546           scheduling dimension for the parallelism check from the polyhedral
15547           information in the AST.
15548         * graphite-dependences.c (carries_deps): Do not assume the schedule is
15549           in 2D + 1 form.
15551 2013-07-13  Jason Merrill  <jason@redhat.com>
15553         * print-tree.c (debug_vec_tree): Use debug_raw.
15554         (debug_raw (vec<tree, va_gc> &)): New.
15555         (debug_raw (vec<tree, va_gc> *)): New.
15556         * tree.h: Declare them.
15558 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
15560         * ifcvt.c (ifcvt_after_combine): New static variable.
15561         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
15562         for size.
15563         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
15564         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
15565         rest_of_handle_if_after_reload): Pass new argument for if_convert.
15567 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
15569         * config/mips/mips.c (mips_expand_call): Remove empty statement.
15571 2013-07-12  Michael Matz  <matz@suse.de>
15573         PR middle-end/55771
15574         * convert.c (convert_to_real): Reject non-float inner types.
15576 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
15578         * config/aarch64/aarch64-protos.h
15579         (aarch64_simd_immediate_valid_for_move): Remove.
15580         * config/aarch64/aarch64.c (simd_immediate_info): New member.
15581         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
15582         cases.
15583         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
15585 2013-07-11  Steve Ellcey  <sellcey@mips.com>
15587         * config/mips/mips.c (mips_conditional_register_usage): Do not
15588         use t[0-7] registers in MIPS16 mode when optimizing for size.
15590 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
15592         * config/i386/i386.c (dispatch_function_versions): Fix array
15593         indexing of function_version_info to match actual_versions.
15595 2013-07-11  Teresa Johnson  <tejohnson@google.com>
15597         * vec.h (struct va_gc): Move release out-of-line.
15598         (va_gc::release): Call ggc_free on released vec.
15600 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15602         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
15603         Require GOT register as additional operand in UNSPEC.
15604         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
15605         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15606         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15607         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
15608         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
15609         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
15610         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
15612 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
15614         PR target/57631
15615         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
15616         name seen by assembler/linker rather if available.
15618 2013-07-11  Andreas Schwab  <schwab@suse.de>
15620         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
15622 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
15624         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
15626 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
15628         * doc/tm.texi.in: Move hook documentation to ....
15629         * target.def: ... here.
15631         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
15632         text on @hook line.
15633         * doc/tm.texi: Regenerate.
15635 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
15637         PR c++/57869
15638         * doc/invoke.texi: Document -Wconditionally-supported.
15640 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15642         PR target/57844
15643         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
15644         of my_fp.
15646 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15648         PR target/57506
15649         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
15650         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
15651         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
15652         Remove duplicate devices.
15653         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
15654         * config/avr/t-multilib: Regenerate.
15655         * config/avr/avr-tables.opt: Regenerate.
15656         * doc/avr-mmcu.texi: Regenerate.
15658 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15660         PR target/56987
15661         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
15663 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
15665         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
15666         the cost of MULT when optimizing for size.
15668 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
15670         * config/cr16/cr16-protos.h: Don't include target.h.
15672 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
15674         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
15675         adjust register size for TDmode and TFmode for VSX registers.
15677 2013-07-08  Kai Tietz  <ktietz@redhat.com>
15679         PR target/56892
15680         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
15681         hook_bool_const_tree_true.
15683 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15685         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
15686         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
15687         * config/s390/s390.md: Define FPR*_REGNUM constants.
15688         Fix FPR2_REGNUM constant (18 -> 17).
15689         ("*trunc<BFP:mode><DFP_ALL:mode>2")
15690         ("*trunc<DFP_ALL:mode><BFP:mode>2")
15691         ("trunc<BFP:mode><DFP_ALL:mode>2")
15692         ("trunc<DFP_ALL:mode><BFP:mode>2")
15693         ("*extend<BFP:mode><DFP_ALL:mode>2")
15694         ("*extend<DFP_ALL:mode><BFP:mode>2")
15695         ("extend<BFP:mode><DFP_ALL:mode>2")
15696         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
15697         FPR4_REGNUM.
15699 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
15701         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
15703 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15705         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
15706         and cfun_fpr_bit_p to cfun_fpr_save_p.
15707         (s390_frame_area, s390_register_info, s390_frame_info)
15708         (s390_emit_prologue, s390_emit_epilogue)
15709         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
15710         register numbers.
15711         * config/s390/s390.h: Define *_REGNUM macros for floating point
15712         register numbers.
15714 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
15716         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
15718 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
15720         PR rtl-optimization/57786
15721         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
15722         and break out of the loop when it is set to false.
15724 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
15726         PR target/57819
15727         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
15728         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
15729         (const_int 63)) 0)).
15730         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
15731         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
15732         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
15734         PR rtl-optimization/57829
15735         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
15736         mask bits outside of mode are just sign-extension from mode to HWI.
15738 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
15740         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
15741         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
15742         adjust_address instead of change_address to keep info about alignment.
15743         (emit_strmov): Remove.
15744         (emit_memmov): New function.
15745         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
15746         (expand_movmem_epilogue): Likewise and return updated rtx for
15747         destination.
15748         (expand_constant_movmem_prologue): Likewise and return updated rtx for
15749         destination and source.
15750         (decide_alignment): Refactor, handle vector_loop.
15751         (ix86_expand_movmem): Likewise.
15752         (ix86_expand_setmem): Likewise.
15753         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
15755 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
15757         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
15758         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
15760 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15762         * config/i386/sse.md (sse_movlhps): Change alternative 3
15763         of operand 2 to "m".
15765 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15767         PR target/57807
15768         * config/i386/sse.md (iptr): New mode attribute.
15769         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
15770         (<sse>_vm<plusminus_insn><mode>3): Ditto.
15771         (<sse>_vmmul<mode>3): Ditto.
15772         (<sse>_vmdiv<mode>3): Ditto.
15773         (sse_vmrcpv4sf2): Ditto.
15774         (<sse>_vmsqrt<mode>2): Ditto.
15775         (sse_vmrsqrtv4sf2): Ditto.
15776         (<sse>_vm<code><mode>3): Ditto.
15777         (avx_vmcmp<mode>3): Ditto.
15778         (<sse>_vmmaskcmp<mode>3): Ditto.
15779         (<sse>_comi): Ditto.
15780         (<sse>_ucomi): Ditto.
15781         (*xop_vmfrcz_<mode>): Ditto.
15782         (*fmai_fmadd_<mode>): Ditto.
15783         (*fmai_fmsub_<mode>): Ditto.
15784         (*fmai_fnmadd_<mode>): Ditto.
15785         (*fmai_fnmsub_<mode>): Ditto.
15786         (*fma4i_vmfmadd_<mode>): Ditto.
15787         (*fma4i_vmfmsub_<mode>): Ditto.
15788         (*fma4i_vmfnmadd_<mode>): Ditto.
15789         (*fma4i_vmfnmsub_<mode>): Ditto.
15790         (*xop_vmfrcz_<mode>): Ditto.
15791         (sse_cvtps2pi): Ditto.
15792         (sse_cvttps2pi): Ditto.
15793         (sse_cvtss2si): Ditto.
15794         (sse_cvtss2si_2): Ditto.
15795         (sse_cvtss2siq_2): Ditto.
15796         (sse_cvttss2si): Ditto.
15797         (sse_cvttss2siq): Ditto.
15798         (sse_cvtsd2si): Ditto.
15799         (sse_cvtsd2si_2): Ditto.
15800         (sse_cvtsd2siq_2): Ditto.
15801         (sse_cvttsd2si): Ditto.
15802         (sse_cvttsd2siq): Ditto.
15803         (sse_cvtsd2ss): Ditto.
15804         (sse_cvtss2sd): Ditto.
15805         (avx2_pbroadcast<mode>): Ditto.
15806         (avx2_pbroadcast<mode>_1): Ditto.
15807         (*avx_vperm_broadcast_v4sf): Ditto.
15809         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
15810         (sse_movlhps): Ditto.
15811         (sse_storehps): Ditto.
15812         (sse_loadhps): Ditto.
15813         (sse_storelps): Ditto.
15814         (sse_loadlps): Ditto.
15815         (*vec_concatv4sf): Ditto.
15816         (*vec_interleave_highv2df): Ditto.
15817         (*vec_interleave_lowv2df): Ditto.
15818         (*vec_extractv2df_1_sse): Ditto.
15819         (*vec_extractv2df_0_sse): Ditto.
15820         (sse2_storelpd): Ditto.
15821         (sse2_loadlpd): Ditto.
15822         (sse2_movsd): Ditto.
15823         (*vec_concatv4si): Ditto.
15824         (vec_concatv2di): Ditto.
15826         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
15827         for Intel asm dialect.
15828         (mmx_punpcklwd): Ditto.
15829         (mmx_punpckldq): Ditto.
15831         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
15832         for intel assembler dialect.
15834 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
15836         PR target/29776
15837         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
15838         for BUILT_IN_C{LZ,LRSB}*.
15839         * tree.h (CASE_INT_FN): Add FN##IMAX case.
15840         * tree-vrp.c (extract_range_basic): Handle
15841         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
15842         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
15843         fall thru to code calling set_value*.
15844         * builtins.c (expand_builtin): Remove *IMAX cases.
15845         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
15846         if width is bigger than 2*HWI.
15848 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
15850         PR rtl-optimization/55342
15851         * lra-int.h (lra_subreg_reload_pseudos): New.
15852         * lra.c: Add undoing optional reloads to the block diagram.
15853         (lra_subreg_reload_pseudos): New.
15854         (lra_optional_reload_pseudos): Change comments.
15855         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
15856         lra_optional_reload_pseudos after undo transformations.
15857         * lra-assigns.c (pseudo_prefix_title): New.
15858         (lra_setup_reg_renumber): Use it.
15859         (spill_for): Ditto.  Check subreg reload pseudos too.
15860         (assign_by_spills): Consider subreg reload pseudos too.
15861         * lra-constraints.c (simplify_operand_subreg): Use
15862         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
15863         (curr_insn_transform): Recognize and do optional reloads.
15864         (undo_optional_reloads): New.
15865         (lra_undo_inheritance): Call undo_optional_reloads.
15867 2013-07-05  Thomas Quinot  <quinot@adacore.com>
15869         * tree-complex.c (expand_complex_operations_1): Fix typo.
15871 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
15873         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
15874         (tune_params): New member 'const vec_costs'.
15875         * config/aarch64/aarch64.c (generic_vector_cost): New.
15876         (generic_tunings): New member 'generic_vector_cost'.
15877         (aarch64_builtin_vectorization_cost): New.
15878         (aarch64_add_stmt_cost): New.
15879         (TARGET_VECTORIZE_ADD_STMT_COST): New.
15880         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
15882 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
15884         PR target/57777
15885         * config/i386/predicates.md (vsib_address_operand): Disallow
15886         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
15888 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
15890         PR middle-end/55030
15891         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
15892         expand_builtin_setjmp_receiver.
15893         (expand_label): Adjust, call expand_builtin_setjmp_receiver
15894         with NULL for the label parameter.
15895         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
15896         the frame-pointer.  Adjust comments.
15897         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
15898         only if LABEL is non-NULL.
15900 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
15902         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
15903         (ARM_ABI_AAPCS64): Ditto.
15904         (arm_abi): Ditto.
15905         (ARM_DEFAULT_ABI): Ditto.
15907 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
15909         * config/aarch64/aarch64-builtins.c
15910         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
15911         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
15912         (st1): Likewise.
15913         * config/aarch64/aarch64-simd.md
15914         (aarch64_ld1<VALL:mode>): New.
15915         (aarch64_st1<VALL:mode>): Likewise.
15916         * config/aarch64/arm_neon.h
15917         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
15919 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
15921         * config/i386/i386.c (gate_insert_vzeroupper): Check if
15922         target ISA is AVX.
15923         (ix86_option_override_internal):Turn on all -mavx target flags by
15924         default as they are dependent on AVX anyway.
15926 2013-07-02  Cary Coutant  <ccoutant@google.com>
15928         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
15929         deterministic hash.
15930         (loc_checksum_ordered): Likewise.
15931         (hash_loc_operands): Remove inline keyword.
15933 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
15935         PR tree-optimization/57741
15936         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
15937         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
15938         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
15939         Allow REAL_CST step_exprs if flag_associative_math.
15940         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
15942 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15944         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
15946 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
15948         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
15950 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15952         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
15953         encoding.
15954         (iorsi3_insn): Likewise.
15955         (arm_xorsi3): Likewise.
15957 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
15959         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
15960         to ...
15961         (attribute "type"): ... here, and prefix with "wmmx_".
15962         (attribute "core_cycles"): Update for attribute changes.
15963         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
15964         (tbcstv4hi): Likewise.
15965         (tbcstv2si): Likewise.
15966         (iwmmxt_iordi3): Likewise.
15967         (iwmmxt_xordi3): Likewise.
15968         (iwmmxt_anddi3): Likewise.
15969         (iwmmxt_nanddi3): Likewise.
15970         (iwmmxt_arm_movdi): Likewise.
15971         (iwmmxt_movsi_insn): Likewise.
15972         (mov<mode>_internal): Likewise.
15973         (and<mode>3_iwmmxt): Likewise.
15974         (ior<mode>3_iwmmxt): Likewise.
15975         (xor<mode>3_iwmmxt): Likewise.
15976         (add<mode>3_iwmmxt): Likewise.
15977         (ssaddv8qi3): Likewise.
15978         (ssaddv4hi3): Likewise.
15979         (ssaddv2si3): Likewise.
15980         (usaddv8qi3): Likewise.
15981         (usaddv4hi3): Likewise.
15982         (usaddv2si3): Likewise.
15983         (sub<mode>3_iwmmxt): Likewise.
15984         (sssubv8qi3): Likewise.
15985         (sssubv4hi3): Likewise.
15986         (sssubv2si3): Likewise.
15987         (ussubv8qi3): Likewise.
15988         (ussubv4hi3): Likewise.
15989         (ussubv2si3): Likewise.
15990         (mulv4hi3_iwmmxt): Likewise.
15991         (smulv4hi3_highpart): Likewise.
15992         (umulv4hi3_highpart): Likewise.
15993         (iwmmxt_wmacs): Likewise.
15994         (iwmmxt_wmacsz): Likewise.
15995         (iwmmxt_wmacu): Likewise.
15996         (iwmmxt_wmacuz): Likewise.
15997         (iwmmxt_clrdi): Likewise.
15998         (iwmmxt_clrv8qi): Likewise.
15999         (iwmmxt_clr4hi): Likewise.
16000         (iwmmxt_clr2si): Likewise.
16001         (iwmmxt_uavgrndv8qi3): Likewise.
16002         (iwmmxt_uavgrndv4hi3): Likewise.
16003         (iwmmxt_uavgv8qi3): Likewise.
16004         (iwmmxt_uavgv4hi3): Likewise.
16005         (iwmmxt_tinsrb): Likewise.
16006         (iwmmxt_tinsrh): Likewise.
16007         (iwmmxt_tinsrw): Likewise.
16008         (iwmmxt_textrmub): Likewise.
16009         (iwmmxt_textrmsb): Likewise.
16010         (iwmmxt_textrmuh): Likewise.
16011         (iwmmxt_textrmsh): Likewise.
16012         (iwmmxt_textrmw): Likewise.
16013         (iwmxxt_wshufh): Likewise.
16014         (eqv8qi3): Likewise.
16015         (eqv4hi3): Likewise.
16016         (eqv2si3): Likewise.
16017         (gtuv8qi3): Likewise.
16018         (gtuv4hi3): Likewise.
16019         (gtuv2si3): Likewise.
16020         (gtv8qi3): Likewise.
16021         (gtv4hi3): Likewise.
16022         (gtv2si3): Likewise.
16023         (smax<mode>3_iwmmxt): Likewise.
16024         (umax<mode>3_iwmmxt): Likewise.
16025         (smin<mode>3_iwmmxt): Likewise.
16026         (umin<mode>3_iwmmxt): Likewise.
16027         (iwmmxt_wpackhss): Likewise.
16028         (iwmmxt_wpackwss): Likewise.
16029         (iwmmxt_wpackdss): Likewise.
16030         (iwmmxt_wpackhus): Likewise.
16031         (iwmmxt_wpackwus): Likewise.
16032         (iwmmxt_wpackdus): Likewise.
16033         (iwmmxt_wunpckihb): Likewise.
16034         (iwmmxt_wunpckihh): Likewise.
16035         (iwmmxt_wunpckihw): Likewise.
16036         (iwmmxt_wunpckilb): Likewise.
16037         (iwmmxt_wunpckilh): Likewise.
16038         (iwmmxt_wunpckilw): Likewise.
16039         (iwmmxt_wunpckehub): Likewise.
16040         (iwmmxt_wunpckehuh): Likewise.
16041         (iwmmxt_wunpckehuw): Likewise.
16042         (iwmmxt_wunpckehsb): Likewise.
16043         (iwmmxt_wunpckehsh): Likewise.
16044         (iwmmxt_wunpckehsw): Likewise.
16045         (iwmmxt_wunpckelub): Likewise.
16046         (iwmmxt_wunpckeluh): Likewise.
16047         (iwmmxt_wunpckeluw): Likewise.
16048         (iwmmxt_wunpckelsb): Likewise.
16049         (iwmmxt_wunpckelsh): Likewise.
16050         (iwmmxt_wunpckelsw): Likewise.
16051         (ror<mode>3): Likewise.
16052         (ashr<mode>3_iwmmxt): Likewise.
16053         (lshr<mode>3_iwmmxt): Likewise.
16054         (ashl<mode>3_iwmmxt): Likewise.
16055         (ror<mode>3_di): Likewise.
16056         (ashr<mode>3_di): Likewise.
16057         (lshr<mode>3_di): Likewise.
16058         (ashl<mode>3_di): Likewise.
16059         (iwmmxt_wmadds): Likewise.
16060         (iwmmxt_wmaddu): Likewise.
16061         (iwmmxt_tmia): Likewise.
16062         (iwmmxt_tmiaph): Likewise.
16063         (iwmmxt_tmiabb): Likewise.
16064         (iwmmxt_tmiatb): Likewise.
16065         (iwmmxt_tmiabt): Likewise.
16066         (iwmmxt_tmiatt): Likewise.
16067         (iwmmxt_tmovmskb): Likewise.
16068         (iwmmxt_tmovmskh): Likewise.
16069         (iwmmxt_tmovmskw): Likewise.
16070         (iwmmxt_waccb): Likewise.
16071         (iwmmxt_wacch): Likewise.
16072         (iwmmxt_waccw): Likewise.
16073         (iwmmxt_waligni): Likewise.
16074         (iwmmxt_walignr): Likewise.
16075         (iwmmxt_walignr0): Likewise.
16076         (iwmmxt_walignr1): Likewise.
16077         (iwmmxt_walignr2): Likewise.
16078         (iwmmxt_walignr3): Likewise.
16079         (iwmmxt_wsadb): Likewise.
16080         (iwmmxt_wsadh): Likewise.
16081         (iwmmxt_wsadbz): Likewise.
16082         (iwmmxt_wsadhz): Likewise.
16083         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
16084         (iwmmxt_wabsdiffb): Likewise.
16085         (iwmmxt_wabsdiffh): Likewise.
16086         (iwmmxt_wabsdiffw): Likewise.
16087         (iwmmxt_waddsubhx): Likewise
16088         (iwmmxt_wsubaddhx): Likewise.
16089         (addc<mode>3): Likewise.
16090         (iwmmxt_avg4): Likewise.
16091         (iwmmxt_avg4r): Likewise.
16092         (iwmmxt_wmaddsx): Likewise.
16093         (iwmmxt_wmaddux): Likewise.
16094         (iwmmxt_wmaddsn): Likewise.
16095         (iwmmxt_wmaddun): Likewise.
16096         (iwmmxt_wmulwsm): Likewise.
16097         (iwmmxt_wmulwum): Likewise.
16098         (iwmmxt_wmulsmr): Likewise.
16099         (iwmmxt_wmulumr): Likewise.
16100         (iwmmxt_wmulwsmr): Likewise.
16101         (iwmmxt_wmulwumr): Likewise.
16102         (iwmmxt_wmulwl): Likewise.
16103         (iwmmxt_wqmulm): Likewise.
16104         (iwmmxt_wqmulwm): Likewise.
16105         (iwmmxt_wqmulmr): Likewise.
16106         (iwmmxt_wqmulwmr): Likewise.
16107         (iwmmxt_waddbhusm): Likewise.
16108         (iwmmxt_waddbhusl): Likewise.
16109         (iwmmxt_wqmiabb): Likewise.
16110         (iwmmxt_wqmiabt): Likewise.
16111         (iwmmxt_wqmiatb): Likewise.
16112         (iwmmxt_wqmiatt): Likewise.
16113         (iwmmxt_wqmiabbn): Likewise.
16114         (iwmmxt_wqmiabtn): Likewise.
16115         (iwmmxt_wqmiatbn): Likewise.
16116         (iwmmxt_wqmiattn): Likewise.
16117         (iwmmxt_wmiabb): Likewise.
16118         (iwmmxt_wmiabt): Likewise.
16119         (iwmmxt_wmiatb): Likewise.
16120         (iwmmxt_wmiatt): Likewise.
16121         (iwmmxt_wmiabbn): Likewise.
16122         (iwmmxt_wmiabtn): Likewise.
16123         (iwmmxt_wmiatbn): Likewise.
16124         (iwmmxt_wmiattn): Likewise.
16125         (iwmmxt_wmiawbb): Likewise.
16126         (iwmmxt_wmiawbt): Likewise.
16127         (iwmmxt_wmiawtb): Likewise.
16128         (iwmmxt_wmiawtt): Likewise.
16129         (iwmmxt_wmiawbbn): Likewise.
16130         (iwmmxt_wmiawbtn): Likewise.
16131         (iwmmxt_wmiawtbn): Likewise.
16132         (iwmmxt_wmiawttn): Likewise.
16133         (iwmmxt_wmerge): Likewise.
16134         (iwmmxt_tandc<mode>3): Likewise.
16135         (iwmmxt_torc<mode>3): Likewise.
16136         (iwmmxt_torvsc<mode>3): Likewise.
16137         (iwmmxt_textrc<mode>3): Likewise.
16138         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
16139         (wmmxt_pack): Likewise.
16140         (wmmxt_mult_c1): Likewise.
16141         (wmmxt_mult_c2): Likewise.
16142         (wmmxt_alu_c1): Likewise.
16143         (wmmxt_alu_c2): Likewise.
16144         (wmmxt_alu_c3): Likewise.
16145         (wmmxt_transfer_c1): Likewise.
16146         (wmmxt_transfer_c2): Likewise.
16147         (wmmxt_transfer_c3): Likewise.
16148         (marvell_f_iwmmxt_wstr): Likewise.
16149         (marvell_f_iwmmxt_wldr): Likewise.
16151 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
16153         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
16155 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16157         Revert:
16158         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16159         * lra-constraints.c (need_for_split_p): Check call used hard regs
16160         living through calls.
16162         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
16163         call used regs for call insn.
16165 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
16167         PR target/57736
16168         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
16169         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
16171 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16173         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
16175 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16177         * lra-constraints.c (need_for_split_p): Check call used hard regs
16178         living through calls.
16180 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
16182         PR target/57744
16183         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
16184         to tie with any other modes.  Eliminate Altivec vector mode tests,
16185         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
16186         code, to return 0 if testing MODE2 for a condition, if we've
16187         already tested MODE1 for the same condition.
16189 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16191         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
16192         layout.
16194 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16196         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
16197         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
16199 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16201         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
16202         Define.
16203         (aarch64_symbolic_constant_p): Remove.
16204         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
16205         static.  Fix line length and white space.
16206         (aarch64_symbolic_constant_p): Remove.
16207         * config/aarch64/predicates.md (aarch64_valid_symref):
16208         Use aarch64_classify_symbol_expression.
16210 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16212         * config/arm/constraints.md (Ts): New constraint.
16213         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
16214         16-bit encodings.
16215         (compare_scc): Use "Ts" constraint for operand 0.
16216         (ior_scc_scc): Likewise.
16217         (and_scc_scc): Likewise.
16218         (and_scc_scc_nodom): Likewise.
16219         (ior_scc_scc_cmp): Likewise for operand 7.
16220         (and_scc_scc_cmp): Likewise.
16221         * config/arm/thumb2.md (thumb2_movsi_insn):
16222         Add alternatives for 16-bit encodings.
16223         (thumb2_movhi_insn): Likewise.
16224         (thumb2_movsicc_insn): Likewise.
16225         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
16226         (thumb2_negscc): Use "Ts" constraint.
16227         Move mvn instruction outside cond_exec block.
16228         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
16229         for 16-bit encodings.
16231 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16233         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
16234         encoding.
16235         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
16236         (mulsi3subsi): Likewise.
16237         (mulsidi3adddi): Likewise.
16238         (mulsidi3_v6): Likewise.
16239         (umulsidi3_v6): Likewise.
16240         (umulsidi3adddi_v6): Likewise.
16241         (smulsi3_highpart_v6): Likewise.
16242         (umulsi3_highpart_v6): Likewise.
16243         (mulhisi3tb): Likewise.
16244         (mulhisi3bt): Likewise.
16245         (mulhisi3tt): Likewise.
16246         (maddhisi4): Likewise.
16247         (maddhisi4tb): Likewise.
16248         (maddhisi4tt): Likewise.
16249         (maddhidi4): Likewise.
16250         (maddhidi4tb): Likewise.
16251         (maddhidi4tt): Likewise.
16252         (zeroextractsi_compare0_scratch): Likewise.
16253         (insv_zero): Likewise.
16254         (insv_t2): Likewise.
16255         (anddi_notzesidi_di): Likewise.
16256         (anddi_notsesidi_di): Likewise.
16257         (andsi_notsi_si): Likewise.
16258         (iordi_zesidi_di): Likewise.
16259         (xordi_zesidi_di): Likewise.
16260         (andsi_iorsi3_notsi): Likewise.
16261         (smax_0): Likewise.
16262         (smax_m1): Likewise.
16263         (smin_0): Likewise.
16264         (not_shiftsi): Likewise.
16265         (unaligned_loadsi): Likewise.
16266         (unaligned_loadhis): Likewise.
16267         (unaligned_loadhiu): Likewise.
16268         (unaligned_storesi): Likewise.
16269         (unaligned_storehi): Likewise.
16270         (extv_reg): Likewise.
16271         (extzv_t2): Likewise.
16272         (divsi3): Likewise.
16273         (udivsi3): Likewise.
16274         (arm_zero_extendhisi2addsi): Likewise.
16275         (arm_zero_extendqisi2addsi): Likewise.
16276         (compareqi_eq0): Likewise.
16277         (arm_extendhisi2_v6): Likewise.
16278         (arm_extendqisi2addsi): Likewise.
16279         (arm_movt): Likewise.
16280         (thumb2_ldrd): Likewise.
16281         (thumb2_ldrd_base): Likewise.
16282         (thumb2_ldrd_base_neg): Likewise.
16283         (thumb2_strd): Likewise.
16284         (thumb2_strd_base): Likewise.
16285         (thumb2_strd_base_neg): Likewise.
16286         (arm_negsi2): Add alternative for 16-bit encoding.
16287         (arm_one_cmplsi2): Likewise.
16289 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16291         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
16292         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
16293         (movdfcc): Likewise.
16294         * config/arm/vfp.md (*thumb2_movsf_vfp):
16295         Disable predication for arm_restrict_it.
16296         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
16297         (*thumb2_movdfcc_vfp): Likewise.
16298         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
16299         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
16300         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
16301         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
16302         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
16303         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
16304         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
16305         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
16306         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
16307         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
16308         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
16309         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
16310         Disable predication for arm_restrict_it.
16312 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
16314         * config/i386/bmiintrin.h (_bextr_u32): New.
16315         (_bextr_u64): Ditto.
16317 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
16319         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
16320         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
16321         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
16322         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
16323         * config/mips/n32-elf.h: ...this new file.
16325 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
16327         PR target/57224
16328         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
16329         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
16331 2013-06-27  Catherine Moore  <clm@codesourcery.com>
16333         * config/mips/mips-tables.opt: Regenerate.
16334         * config/mips/mips-cpus.def: Add m14ke and m14kec.
16335         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
16336         * doc/invoke.texi: Add -m14kc.
16338 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16340         PR target/57623
16341         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
16342         constraints of operand 1 and 2.
16344         PR target/57623
16345         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
16346         to match RTL canonicalization.  Swap predicates and
16347         constraints of operand 1 and 2.
16349 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
16351         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
16352         Process OP_INOUT regs for splitting too.
16354 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16356         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
16357         decl before the loop, initialize to NULL.
16358         (vectorizable_load): Initialize ptr_incr to NULL.
16360 2013-06-27  Martin Jambor  <mjambor@suse.cz>
16362         PR lto/57208
16363         * ipa-ref.h (ipa_maybe_record_reference): Declare.
16364         * ipa-ref.c (ipa_maybe_record_reference): New function.
16365         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
16366         * ipa-cp.c (create_specialized_node): Record potential references from
16367         aggvals.
16368         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
16370 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16372         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
16373         parameter 'mode' of type 'enum machine_mode mode'; change to pass
16374         'mode' to force_reg.
16375         (aarch64_add_offset): Update calls to aarch64_force_temporary.
16376         (aarch64_expand_mov_immediate): Likewise.
16378 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16380         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
16381         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
16383 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16385         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
16386         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16387         (struct machine_function): Add tbegin_p.
16388         (s390_canonicalize_comparison): Fold CC mode compares to
16389         conditional jump if possible.
16390         (s390_emit_jump): Return the emitted jump.
16391         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
16392         Handle CCRAWmode compares.
16393         (s390_option_override): Default to -mhtm if available.
16394         (s390_reg_clobbered_rtx): Handle floating point regs as well.
16395         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
16396         FPRs instead of df_regs_ever_live_p.
16397         (s390_optimize_nonescaping_tx): New function.
16398         (s390_init_frame_layout): Extend clobbered_regs array to cover
16399         FPRs as well.
16400         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
16401         (s390_expand_tbegin): New function.
16402         (enum s390_builtin): New enum definition.
16403         (code_for_builtin): New array definition.
16404         (s390_init_builtins): New function.
16405         (s390_expand_builtin): New function.
16406         (TARGET_INIT_BUILTINS): Define.
16407         (TARGET_EXPAND_BUILTIN): Define.
16408         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
16409         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
16410         (s390_alc_comparison): Likewise.
16411         * config/s390/s390-modes.def: Add CCRAWmode.
16412         * config/s390/s390.h (processor_flags): Add PF_TX.
16413         (TARGET_CPU_HTM): Define macro.
16414         (TARGET_HTM): Define macro.
16415         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
16416         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
16417         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16418         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
16419         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
16420         (TBEGIN_MASK, TBEGINC_MASK): New constants.
16421         ("*cc_to_int"): Move up.
16422         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
16423         constants other than 0.
16424         ("*ccraw_to_int"): New insn and splitter definition.
16425         ("tbegin", "tbegin_nofloat", "tbegin_retry")
16426         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
16427         ("tx_assist"): New expander.
16428         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
16429         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
16430         * config/s390/s390.opt: Add -mhtm option.
16431         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
16432         * config/s390/htmxlintrin.h: New file.
16433         * config/s390/htmintrin.h: New file.
16434         * config/s390/s390intrin.h: New file.
16435         * doc/extend.texi: Document htm builtins.
16436         * config.gcc: Add the new header files to extra_headers.
16438 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
16440         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
16441         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
16443 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
16444             Pat Haugen  <pthaugen@us.ibm.com>
16445             Peter Bergner  <bergner@vnet.ibm.com>
16447         * config/rs6000/power8.md: New.
16448         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
16449         setting for power8 entry.
16450         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
16451         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
16452         test for Power4/Power5 only.
16453         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
16454         support.
16455         (force_new_group): Adjust comment.
16456         * config/rs6000/rs6000.md: Include power8.md.
16458 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
16460         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
16461         * config/arm/arm-protos.h (arm_max_conditional_execute): New
16462         declaration.
16463         (tune_params): Update comment.
16464         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
16465         (arm_max_conditional_execute): New function.
16466         (thumb2_final_prescan_insn): Use max_insn_skipped and
16467         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
16469 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
16471         PR tree-optimization/57705
16472         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
16473         SSA_NAME step, provided that it is not defined inside the loop.
16474         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
16475         (get_initial_def_for_induction): Handle SSA_NAME IV step.
16477 2013-06-25  Martin Jambor  <mjambor@suse.cz>
16479         PR middle-end/57670
16480         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
16481         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
16482         calls in the dump.
16483         (ipa_note_param_call): Initialize member_ptr flag.
16484         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
16485         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
16486         (ipa_write_indirect_edge_info): Stream member_ptr flag.
16487         (ipa_read_indirect_edge_info): Likewise.
16489 2013-06-25  Richard Biener  <rguenther@suse.de>
16491         PR middle-end/56977
16492         * passes.c (init_optimization_passes): Move pass_fold_builtins
16493         and pass_dce earlier with -Og.
16495 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
16497         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
16498         <BIT_FIELD_REF>: Remove trailing TAB.
16499         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
16500         remove blank line.
16502 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16504         PR tree-optimization/57358
16505         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
16506         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
16507         (ipa_analyze_params_uses): Generate pessimistic info when true.
16509 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16511         PR tree-optimization/57539
16512         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
16513         global.inlined_to of the new node to it.  All callers changed.
16514         * ipa-inline-transform.c (clone_inlined_nodes): New variable
16515         inlining_into, pass it to cgraph_clone_node.
16516         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
16517         ipa_free_edge_args_substructures.
16518         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
16519         rdesc linked list.  Do not assert rdesc edges have inlined caller.
16520         Assert we have found an rdesc in the rdesc list.
16522 2013-06-24  Richard Biener  <rguenther@suse.de>
16524         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
16525         (pointer_set_lookup): Declare.
16526         (class pointer_map): New template class implementing a
16527         generic pointer to T map.
16528         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
16529         pointer_map<T>::contains, pointer_map<T>::insert,
16530         pointer_map<T>::traverse): New functions.
16531         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
16532         (pointer_set_lookup): New function.
16533         (pointer_set_contains): Use pointer_set_lookup.
16534         (pointer_set_insert): Likewise.
16535         (insert_aux): Remove.
16536         (struct pointer_map_t): Embed a pointer_set_t.
16537         (pointer_map_create): Adjust.
16538         (pointer_map_destroy): Likewise.
16539         (pointer_map_contains): Likewise.
16540         (pointer_map_insert): Likewise.
16541         (pointer_map_traverse): Likewise.
16542         * tree-streamer.h (struct streamer_tree_cache_d): Use a
16543         pointer_map<unsigned> instead of a pointer_map_t.
16544         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
16545         (streamer_tree_cache_lookup): Likewise.
16546         (streamer_tree_cache_create): Likewise.
16547         (streamer_tree_cache_delete): Likewise.
16548         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
16549         pointer_map<unsigned> instead of a pointer_map_t.
16550         (lto_init_tree_ref_encoder): Adjust.
16551         (lto_destroy_tree_ref_encoder): Likewise.
16552         * lto-section-out.c (lto_output_decl_index): Likewise.
16553         (lto_record_function_out_decl_state): Likewise.
16554         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
16556 2013-06-24  Richard Biener  <rguenther@suse.de>
16558         PR tree-optimization/57488
16559         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
16561 2013-06-24  Alan Modra  <amodra@gmail.com>
16563         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
16564         (gen_easy_altivec_constant): Likewise.
16565         * config/rs6000/predicates.md (easy_vector_constant_add_self,
16566         easy_vector_constant_msb): Likewise.
16568 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
16570         PR target/57688
16571         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
16572         add missing return true.
16574 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
16576         PR target/52483
16577         * config/sh/predicates.md (general_extend_operand): Invoke
16578         general_movsrc_operand for memory operands.
16579         (general_movsrc_operand): Allow reg+reg addressing, do not use
16580         general_operand for memory operands.
16582 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
16584         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
16585         when current target options does not apply.
16586         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
16587         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
16588         * config/i386/bmiintrin.h: Pass appropriate target
16589         attributes to header.
16590         * config/i386/mmintrin.h: Ditto.
16591         * config/i386/nmmintrin.h: Ditto.
16592         * config/i386/avx2intrin.h: Ditto.
16593         * config/i386/fxsrintrin.h: Ditto.
16594         * config/i386/tbmintrin.h: Ditto.
16595         * config/i386/xsaveintrin.h: Ditto.
16596         * config/i386/f16cintrin.h: Ditto.
16597         * config/i386/xtestintrin.h: Ditto.
16598         * config/i386/xsaveoptintrin.h: Ditto.
16599         * config/i386/bmi2intrin.h: Ditto.
16600         * config/i386/lzcntintrin.h: Ditto.
16601         * config/i386/smmintrin.h: Ditto.
16602         * config/i386/wmmintrin.h: Ditto.
16603         * config/i386/x86intrin.h: Remove all header include guards.
16604         * config/i386/prfchwintrin.h: Ditto.
16605         * config/i386/pmmintrin.h: Ditto.
16606         * config/i386/tmmintrin.h: Ditto.
16607         * config/i386/xmmintrin.h: Ditto.
16608         * config/i386/popcntintrin.h: Ditto.
16609         * config/i386/rdseedintrin.h: Ditto.
16610         * config/i386/ammintrin.h: Ditto.
16611         * config/i386/emmintrin.h: Ditto.
16612         * config/i386/immintrin.h: Remove all header include guards.
16613         * config/i386/fma4intrin.h: Ditto.
16614         * config/i386/lwpintrin.h: Ditto.
16615         * config/i386/xopintrin.h: Ditto.
16616         * config/i386/ia32intrin.h: Ditto.
16617         * config/i386/avxintrin.h: Ditto.
16618         * config/i386/rtmintrin.h: Ditto.
16619         * config/i386/fmaintrin.h: Ditto.
16620         * config/i386/mm3dnow.h: Ditto.
16622 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
16624         * common/config/i386/i386-common.c: Handle LZCNT.
16626 2013-06-22  Andi Kleen  <ak@linux.intel.com>
16628         * doc/extend.texi: Use __atomic_store_n instead of
16629         __atomic_store in HLE example.
16631 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
16633         * config/sh/sh.c: Remove <cstdlib> workaround.
16635 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16637         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
16639 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16641         * doc/extend.texi: Document that __atomic_clear and
16642           __atomic_test_and_set should only be used with bool.
16644 2013-06-20  Jan Hubicka  <jh@suse.cz>
16646         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
16647         types_same_for_odr.
16648         * tree.c (decls_same_for_odr): New function.
16649         (same_for_edr): New function.
16650         (types_same_for_odr): New function.
16651         (get_binfo_at_offset): Use it.
16652         * tree.h (types_same_for_odr): Declare.
16654 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
16655             Jason Merrill  <jason@redhat.com>
16657         * system.h: Include <cstdlib> as well as <stdlib.h>.
16659 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
16661         PR target/57655
16662         * config/i386/i386.c (construct_container): Report error if
16663         long double is used with disabled x87 float returns.
16665 2013-06-20  Jan Hubicka  <jh@suse.cz>
16667         * lto-cgraph.c (input_symtab): Do not set cgraph state.
16669 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
16671         PR rtl-optimization/57425
16672         PR rtl-optimization/57569
16673         * alias.c (write_dependence_p): Remove parameters mem_mode and
16674         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
16675         Changed all callers.
16676         (canon_anti_dependence): Get comments and semantics in sync.
16677         Add parameter mem_canonicalized.  Changed all callers.
16678         * rtl.h (canon_anti_dependence): Update prototype.
16680 2013-06-20  Richard Biener  <rguenther@suse.de>
16682         * data-streamer-in.c (streamer_read_uhwi): Optimize single
16683         byte case, inline streamer_read_uchar and defer section
16684         overrun check.
16686 2013-06-20  Richard Biener  <rguenther@suse.de>
16688         PR tree-optimization/57584
16689         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
16690         SSA names into the expanded expression that take part in
16691         abnormal coalescing.
16693 2013-06-19  Sharad Singhai  <singhai@google.com>
16695         * gcov.c (print_usage): Handle new option.
16696         (process_args): Ditto.
16697         (get_gcov_intermediate_filename): New function.
16698         (output_intermediate_file): New function.
16699         (output_gcov_file): New function
16700         (generate_results): Handle new option.
16701         (release_function): Relase demangled name.
16702         (read_graph_file): Handle demangled name.
16703         (output_lines): Ditto.
16704         * doc/gcov.texi: Document gcov intermediate format.
16706 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
16708         PR bootstrap/57604
16709         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
16710         (lra_emit_add): Use the functions.  Add comment about Y as an
16711         address segment.
16713 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
16715         PR driver/57652
16716         * collect2.c (collect_atexit): New.
16717         (collect_exit): Delete.
16718         (main): Register collect_atexit with atexit.
16719         (collect_wait): Change collect_exit to exit.
16720         (do_wait): Same.
16721         * collect2.h (collect_exit): Delete.
16722         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
16724 2013-06-19  Wei Mi  <wmi@google.com>
16726         PR rtl-optimization/57518
16727         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
16728         if regno is used in paradoxical subreg.
16729         (update_equiv_regs): Check pdx_subregs[regno] before
16730         set a reg to be equivalent with a mem.
16732 2013-06-19  Matthias Klose  <doko@ubuntu.com>
16734         PR driver/57651
16735         * file-find.h (find_a_file): Add a mode parameter.
16736         * file-find.c (find_a_file): Likewise.
16737         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
16738         with X_OK for the executables.
16739         * collect2.c (main): Call find_a_file with X_OK.
16741 2013-06-19  Steve Ellcey  <sellcey@mips.com>
16743         PR target/56942
16744         * config/mips/mips.md (casesi_internal_mips16_<mode>):
16745         Use NEXT_INSN instead of next_real_insn.
16747 2013-06-19  Jan Hubicka  <jh@suse.cz>
16749         * cgraph.h (const_value_known_p): Replace by ...
16750         (ctor_for_folding): .. this one.
16751         * cgraphunit.c (process_function_and_variable_attributes): Use it.
16752         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
16753         * expr.c (expand_expr_real_1): Likewise.
16754         (string_constant): Likewise.
16755         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
16756         * ipa.c (process_references): Likewise.
16757         (symtab_remove_unreachable_nodes): Likewise.
16758         * ipa-inline-analysis.c (param_change_prob): Likewise.
16759         * gimple-fold.c (canonicalize_constructor_val): Likewise.
16760         (get_base_constructor): Likwise.
16761         * varpool.c (varpool_remove_node): Likewise.
16762         (varpool_remove_initializer): LIkewise.
16763         (dump_varpool_node): LIkwise.
16764         (const_value_known_p): Rewrite to ...
16765         (ctor_for_folding): ... this one.
16767 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
16769         PR driver/57651
16770         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
16771         PERSONALITY in $PATH derived prefixes.
16773 2013-06-19  Jeff Law  <law@redhat.com>
16775         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
16776         in comment.
16778         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
16779         (simplify_bitwise_binary): Use it to simpify certain binary ops on
16780         booleans.
16782 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
16784         * config/arm/vfp.md: Move VFP instruction classification documentation
16785         to ...
16786         * config/arm/arm.md: ... here.  Update instruction classification
16787         documentation.
16789 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
16791         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
16792         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
16793         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
16795 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
16797         PR target/57609
16798         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16799         with NEXT_INSN.  Use tablejump_p to check for jump table data
16800         insns.
16802 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
16804         PR c++/56544
16805         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
16806         that now in C++ the value is correct per the C++ standards.
16808 2013-06-19  Richard Biener  <rguenther@suse.de>
16810         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
16811         for global context.
16813 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16815         Revert:
16816         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16818         PR target/57609
16819         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16820         with next_active_insn.
16822 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
16824         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
16825         functions are inlined during failures to flag an error.
16826         * tree-inline.c (expand_call_inline): Allow the error to be flagged
16827         in early inline pass.
16829 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
16831         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
16832         in comments.
16834 2013-06-18  Julian Brown  <julian@codesourcery.com>
16836         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
16837         Permit virtual register pre-reload if !strict.
16838         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
16839         change.
16840         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
16841         prototype.
16842         * config/arm/neon.md (movmisalign<mode>): Use
16843         neon_perm_struct_or_reg_operand instead of
16844         neon_struct_or_register_operand.
16845         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
16846         neon_permissive_struct_operand instead of neon_struct_operand.
16847         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
16848         neon_vector_mem_operand.
16849         * config/arm/predicates.md (neon_struct_operand): Adjust call to
16850         neon_vector_mem_operand.
16851         (neon_permissive_struct_operand): New.
16852         (neon_struct_or_register_operand): Rename to...
16853         (neon_perm_struct_or_reg_operand): This. Adjust call to
16854         neon_vector_mem_operand.
16856 2013-06-18  Richard Biener  <rguenther@suse.de>
16858         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
16859         * lto-streamer.h: Include pointer-set.h.
16860         (struct lto_decl_slot): Remove.
16861         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
16862         Remove next_index entry.
16863         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16864         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16865         (lto_init_tree_ref_encoder): Adjust.
16866         (lto_destroy_tree_ref_encoder): Likewise.
16867         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16868         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16869         (lto_output_decl_index): Adjust.
16870         (lto_new_out_decl_state): Likewise.
16871         (lto_record_function_out_decl_state): Likewise.
16872         * lto-streamer-out.c (copy_function): Likewise.
16874 2013-06-18  Richard Biener  <rguenther@suse.de>
16876         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
16877         * cgraphunit.c: Include cfgloop.h.
16878         (init_lowered_empty_function): Initialize the loop tree.
16879         (assemble_thunk): Insert new BBs into loops.
16881 2013-06-18  Richard Biener  <rguenther@suse.de>
16883         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
16884         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
16885         the map from cache entry to cache index optional.
16886         (streamer_tree_cache_replace_tree): Adjust accordingly.
16887         (streamer_tree_cache_append): Likewise.
16888         (streamer_tree_cache_delete): Likewise.
16889         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
16890         streamer cache map from cache entry to cache index.
16891         * lto-streamer-out.c (create_output_block): Adjust.
16893 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
16895         * config/arm/arm.md (attribute "insn"): Move multiplication and
16896         division attributes to...
16897         (attribute "type"): ... here.  Remove mult.
16898         (attribute "mul32"): New attribute.
16899         (attribute "mul64"): Add umaal.
16900         (*arm_mulsi3): Update attributes.
16901         (*arm_mulsi3_v6): Likewise.
16902         (*thumb_mulsi3): Likewise.
16903         (*thumb_mulsi3_v6): Likewise.
16904         (*mulsi3_compare0): Likewise.
16905         (*mulsi3_compare0_v6): Likewise.
16906         (*mulsi_compare0_scratch): Likewise.
16907         (*mulsi_compare0_scratch_v6): Likewise.
16908         (*mulsi3addsi): Likewise.
16909         (*mulsi3addsi_v6): Likewise.
16910         (*mulsi3addsi_compare0): Likewise.
16911         (*mulsi3addsi_compare0_v6): Likewise.
16912         (*mulsi3addsi_compare0_scratch): Likewise.
16913         (*mulsi3addsi_compare0_scratch_v6): Likewise.
16914         (*mulsi3subsi): Likewise.
16915         (*mulsidi3adddi): Likewise.
16916         (*mulsi3addsi_v6): Likewise.
16917         (*mulsidi3adddi_v6): Likewise.
16918         (*mulsidi3_nov6): Likewise.
16919         (*mulsidi3_v6): Likewise.
16920         (*umulsidi3_nov6): Likewise.
16921         (*umulsidi3_v6): Likewise.
16922         (*umulsidi3adddi): Likewise.
16923         (*umulsidi3adddi_v6): Likewise.
16924         (*smulsi3_highpart_nov6): Likewise.
16925         (*smulsi3_highpart_v6): Likewise.
16926         (*umulsi3_highpart_nov6): Likewise.
16927         (*umulsi3_highpart_v6): Likewise.
16928         (mulhisi3): Likewise.
16929         (*mulhisi3tb): Likewise.
16930         (*mulhisi3bt): Likewise.
16931         (*mulhisi3tt): Likewise.
16932         (maddhisi4): Likewise.
16933         (*maddhisi4tb): Likewise.
16934         (*maddhisi4tt): Likewise.
16935         (maddhidi4): Likewise.
16936         (*maddhidi4tb): Likewise.
16937         (*maddhidi4tt): Likewise.
16938         (divsi3): Likewise.
16939         (udivsi3): Likewise.
16940         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
16941         (thumb2_mulsi_short_compare0): Likewise.
16942         (thumb2_mulsi_short_compare0_scratch): Likewise.
16943         * config/arm/arm1020e.md (1020mult1): Update attribute change.
16944         (1020mult2): Likewise.
16945         (1020mult3): Likewise.
16946         (1020mult4): Likewise.
16947         (1020mult5): Likewise.
16948         (1020mult6): Likewise.
16949         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
16950         change.
16951         (cortex_a15_mult64): Likewise.
16952         (cortex_a15_sdiv): Likewise.
16953         (cortex_a15_udiv): Likewise.
16954         * config/arm/arm1026ejs.md (mult1): Update attribute change.
16955         (mult2): Likewise.
16956         (mult3): Likewise.
16957         (mult4): Likewise.
16958         (mult5): Likewise.
16959         (mult6): Likewise.
16960         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
16961         (pj4_ir_div): Likewise.
16962         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
16963         (11_mult2): Likewise.
16964         (11_mult3): Likewise.
16965         (11_mult4): Likewise.
16966         (11_mult5): Likewise.
16967         (11_mult6): Likewise.
16968         (11_mult7): Likewise.
16969         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
16970         (cortex_a8_mla): Likewise.
16971         (cortex_a8_mull): Likewise.
16972         (cortex_a8_smulwy): Likewise.
16973         (cortex_a8_smlald): Likewise.
16974         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
16975         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
16976         (cortex_r4_mul_3): Likewise.
16977         (cortex_r4_mla_4): Likewise.
16978         (cortex_r4_mla_3): Likewise.
16979         (cortex_r4_smlald): Likewise.
16980         (cortex_r4_mull): Likewise.
16981         (cortex_r4_sdiv): Likewise.
16982         (cortex_r4_udiv): Likewise.
16983         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
16984         (cortex_a7_idiv): Likewise.
16985         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
16986         (9_mult2): Likewise.
16987         (9_mult3): Likewise.
16988         (9_mult4): Likewise.
16989         (9_mult5): Likewise.
16990         (9_mult6): Likewise.
16991         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
16992         (cortex_a53_sdiv): Likewise.
16993         (cortex_a53_udiv): Likewise.
16994         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
16995         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
16996         (mp626_mult2): Likewise.
16997         (mp626_mult3): Likewise.
16998         (mp626_mult4): Likewise.
16999         * config/arm/fa526.md (526_mult1): Update attribute change.
17000         (526_mult2): Likewise.
17001         * config/arm/arm-generic.md (mult): Update attribute change.
17002         (mult_ldsched_strongarm): Likewise.
17003         (mult_ldsched): Likewise.
17004         (multi_cycle): Likewise.
17005         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
17006         * config/arm/fa606te.md (606te_mult1): Update attribute change.
17007         (606te_mult2): Likewise.
17008         (606te_mult3): Likewise.
17009         (606te_mult4): Likewise.
17010         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
17011         (cortex_a9_mac16): Likewise.
17012         (cortex_a9_multiply): Likewise.
17013         (cortex_a9_mac): Likewise.
17014         (cortex_a9_multiply_long): Likewise.
17015         * config/arm/fa626te.md (626te_mult1): Update attribute change.
17016         (626te_mult2): Likewise.
17017         (626te_mult3): Likewise.
17018         (626te_mult4): Likewise.
17020 2013-06-18  Richard Biener  <rguenther@suse.de>
17022         PR lto/57334
17023         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
17025 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17027         PR target/57609
17028         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
17029         with next_active_insn.
17031 2013-06-18  Alan Modra  <amodra@gmail.com>
17033         * config/rs6000/rs6000.h (enum data_align): New.
17034         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
17035         (DATA_ABI_ALIGNMENT): Define.
17036         (CONSTANT_ALIGNMENT): Correct comment.
17037         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
17038         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
17040 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
17042         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
17043         ATTRIBUTE_UNUSED marking.
17045 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
17047         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
17048         alternative and update.
17049         (aarch64_dup_lanedi): Delete.
17050         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
17051         * config/aarch64/aarch64-simd-builtins.def: Update.
17053 2013-06-17  Richard Biener  <rguenther@suse.de>
17055         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
17056         (lto_input_scc): Declare.
17057         (lto_input_tree_1): Likewise.
17058         (struct lto_stats_d): Add num_tree_bodies_output and
17059         num_pickle_refs_output.
17060         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
17061         (lto_read_tree_1): Split out from ...
17062         (lto_read_tree): ... this.
17063         (lto_input_scc): New function.
17064         (lto_input_tree_1): Split out from ...
17065         (lto_input_tree): ... this.  Handle LTO_tree_scc.
17066         (lto_data_in_create): Create the streamer cache without hashes.
17067         * lto-streamer-out.c (create_output_block): Create the streamer
17068         cache with hashes when not doing WPA.
17069         (lto_write_tree_1): Split out from ...
17070         (lto_write_tree): ... this.
17071         (get_symbol_initial_value): New function.
17072         (lto_output_tree_1): Split out from ...
17073         (lto_output_tree): ... this.  Write trees as series of SCCs
17074         using a DFS walk via DFS_write_tree.
17075         (struct sccs, struct scc_entry): New types.
17076         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
17077         (DFS_write_tree_body): New function.
17078         (DFS_write_tree): Likewise.
17079         (hash_tree): Likewise.
17080         (scc_entry_compare): Likewise.
17081         (hash_scc): Likewise.
17082         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
17083         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
17084         TREE_CHAIN as regular reference.
17085         (streamer_read_integer_cst): Remove.
17086         (streamer_get_pickled_tree): Adjust.
17087         * tree-streamer-out.c (streamer_write_chain): Disable streaming
17088         of DECL_EXTERNALs in BLOCK_VARS for now.
17089         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
17090         reference.
17091         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
17092         Add hash value argument and record that if hashes are recorded
17093         in the cache.
17094         (streamer_tree_cache_insert_1): Adjust.
17095         (streamer_tree_cache_insert): Likewise.
17096         (streamer_tree_cache_insert_at): Rename to ...
17097         (streamer_tree_cache_replace_tree): ... this and adjust.
17098         (streamer_tree_cache_append): Adjust.
17099         (record_common_node): Likewise.
17100         (streamer_tree_cache_create): Add argument whether to
17101         record hash values together with trees.
17102         (streamer_tree_cache_delete): Adjust.
17103         * tree-streamer.h (struct streamer_tree_cache_d): Add
17104         vector of hashes.
17105         (streamer_read_integer_cst): Remove.
17106         (streamer_tree_cache_insert): Adjust.
17107         (streamer_tree_cache_append): Likewise.
17108         (streamer_tree_cache_insert_at): Rename to ...
17109         (streamer_tree_cache_replace_tree): ... this and adjust.
17110         (streamer_tree_cache_create): Add argument whether to record hashes.
17111         (streamer_tree_cache_get): Rename to ...
17112         (streamer_tree_cache_get_tree): ... this.
17113         (streamer_tree_cache_get_hash): New function.
17114         * tree.c (cache_integer_cst): New function.
17115         * tree.h (cache_integer_cst): Declare.
17116         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
17117         * lto-symtab.c (lto_varpool_replace_node): Only release
17118         DECL_INITIAL of non-prevailing decls.
17119         * varpool.c (varpool_remove_initializer): Do not release
17120         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
17122 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17124         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
17125         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
17126         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
17127         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
17128         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
17129         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
17130         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
17131         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
17132         instead of TARGET_64BIT.
17133         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
17134         Require ISA_HAS_<D>DIV.
17136 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
17138         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
17139         (mips*-*-linux*): Move default with_llsc setting to where other
17140         defaults are set.
17141         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
17142         with_arch block.
17143         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
17144         Likewise.  Remove default with_tune setting.  Move default float
17145         setting to its own block.  Handle with_llsc in the same block as above.
17147 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
17149         PR rtl-optimization/57425
17150         PR rtl-optimization/57569
17151         * alias.c (write_dependence_p): Add new parameters mem_mode,
17152         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
17153         Changed all callers.
17154         (canon_anti_dependence): New function.
17155         * cse.c (check_dependence): Use canon_anti_dependence.
17156         * cselib.c (cselib_invalidate_mem): Likewise.
17157         * rtl.h (canon_anti_dependence): Declare.
17159 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17161         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
17162         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
17163         ".set mips3" for 64-bit targets.
17165 2013-06-15  Dehao Chen  <dehao@google.com>
17167         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
17168         * gimple-low.c (gimple_check_call_matching_types): Likewise.
17169         (gimple_check_call_args): Likewise.
17170         * value-prof.c (check_ic_target): Likewise.
17171         * ipa-inline.c (early_inliner): Likewise.
17172         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
17173         * cgraph.c (cgraph_create_edge_1): Likewise.
17174         (cgraph_make_edge_direct): Likewise.
17176 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
17178         PR target/57615
17179         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
17180         rs6000_output_move_128bit to handle emitting quad memory
17181         operations.  Set attribute length to 8 bytes.
17183 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
17185         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
17186         New pattern.
17187         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
17188         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
17189         (aarch64_<su>mlsl<mode>): Likewise.
17191 2013-06-14  Mike Stump  <mikestump@comcast.net>
17193         * Makefile.in (TARGET_H): Add insn-codes.h.
17195 2013-06-14  Alan Modra  <amodra@gmail.com>
17197         PR middle-end/57134
17198         PR middle-end/57586
17199         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
17200         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
17201         bitfield expansion when EXPAND_MEMORY.
17202         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
17204 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
17206         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
17207         test for clearing quad memory on 32-bit later.
17209 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17211         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
17212         (fold_negate_expr): Likewise.
17213         (fold_real_zero_addition_p): Handle vectors.
17214         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
17216 2013-06-14  Alan Modra  <amodra@gmail.com>
17218         * varasm.c (force_const_mem): Revert 2013-06-07 change.
17220 2013-06-13  Jan Hubicka  <jh@suse.cz>
17222         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
17223         Local comdats are not externally visible.
17224         * symtab.c (dump_symtab_base): Dump externally visible.
17225         (verify_symtab_base): Verify back links in the symtab hash.
17227 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17229         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
17230         CONVERT_EXPR as equal nodes.
17232 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17234         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
17236 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17238         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
17239         Generalize to complex and vector.
17240         * tree.c (build_all_ones_cst): New function.
17241         * tree.h (build_all_ones_cst): Declare it.
17243 2013-06-13  Alan Modra  <amodra@gmail.com>
17245         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
17246         * config/rs6000/rs6000.md (signbittf2): New insn.
17247         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
17248         (abstf2_internal, cmptf_internal2): Likewise.
17249         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
17251 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
17252             Pat Haugen  <pthaugen@us.ibm.com>
17253             Peter Bergner  <bergner@vnet.ibm.com>
17255         * config/rs6000/rs6000.c (emit_load_locked): Add support for
17256         power8 byte, half-word, and quad-word atomic instructions.
17257         (emit_store_conditional): Likewise.
17258         (rs6000_expand_atomic_compare_and_swap): Likewise.
17259         (rs6000_expand_atomic_op): Likewise.
17261         * config/rs6000/sync.md (larx): Add new modes for power8.
17262         (stcx): Likewise.
17263         (AINT): New mode iterator to include TImode as well as normal
17264         integer modes on power8.
17265         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
17266         that VSX registers are not considered.  Use AINT mode iterator
17267         instead of INT1 to allow inclusion of quad word atomic operations
17268         on power8.
17269         (load_locked<mode>): Likewise.
17270         (store_conditional<mode>): Likewise.
17271         (atomic_compare_and_swap<mode>): Likewise.
17272         (atomic_exchange<mode>): Likewise.
17273         (atomic_nand<mode>): Likewise.
17274         (atomic_fetch_<fetchop_name><mode>): Likewise.
17275         (atomic_nand_fetch<mode>): Likewise.
17276         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
17277         each type.
17278         (ATOMIC): On power8, add QImode, HImode modes.
17279         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
17280         modes that promote to SImode.
17281         (load_lockedti): Convert TImode arguments to PTImode, so that we
17282         get a guaranteed even/odd register pair.
17283         (load_lockedpti): Likewise.
17284         (store_conditionalti): Likewise.
17285         (store_conditionalpti): Likewise.
17287         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
17288         atomic load/store instructions.
17289         (HSI): Likewise.
17291 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
17293         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
17294         loads.
17295         (insn_count): New attribute, with most cases extracted from...
17296         (length): ...here.  Redefine most cases in terms of insn_count.
17297         (single_insn): Delete.
17298         (can_delay): Use insn_count to check for single instructions.
17299         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
17300         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
17301         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
17302         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
17303         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
17304         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
17305         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
17306         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
17307         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
17308         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
17309         rather than "length".
17310         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
17311         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
17312         Use "insn_count" rather than "length".
17313         * config/mips/mips-dsp.md
17314         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
17315         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
17316         length attributes.
17318 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
17320         PR tree-optimization/57361
17321         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
17323 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
17325         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
17326         to split.
17327         (aarch64_simd_combine<mode>): New instruction expansion.
17328         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
17329         function prototype.
17330         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
17331         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
17333 2013-06-12  Jan Hubicka  <jh@suse.cz>
17335         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
17336         decl has when in streaming stage.
17337         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
17338         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
17340 2013-06-12  Roland Stigge  <stigge@antcom.de>
17342         PR target/57578
17343         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
17345 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17347         PR tree-optimization/57537
17348         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
17349         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
17351 2013-06-12  Richard Biener  <rguenther@suse.de>
17353         * data-streamer.h (streamer_write_char_stream): CSE
17354         obs->current_pointer.
17355         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
17356         streamer_write_char_stream manually and optimize the resulting loop.
17357         (streamer_write_hwi_stream): Likewise.
17359 2013-06-12  Jan Hubicka  <jh@suse.cz>
17361         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
17362         * cgraph.h (varpool_create_empty_node): Declare.
17363         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
17364         duplicated nodes.
17365         * symtab.c (symtab_unregister_node): Be lax about missin entries
17366         in node hash.
17367         (symtab_get_node): Update comment.
17368         * varpool.c (varpool_create_empty_node): Break out from ...
17369         (varpool_node_for_decl): ... here.
17370         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
17372 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
17374         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
17375         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
17376         part.  Use straight-line flow at the end.
17377         <COMPONENT_REF>: Remove superfluous else.
17378         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
17380 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17382         PR target/56564
17383         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
17384         target hook even for !TREE_PUBLIC decls.  If no resolution info
17385         is available, return false for common and external decls.
17387 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17389         * config/rl78/constraints.md (U): New constraint.
17390         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
17391         valloc attribute.
17393 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17395         PR target/57589
17396         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
17397         to allow returning address to AT_PLATFORM name.
17399 2013-06-11  Jan Hubicka  <jh@suse.cz>
17401         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
17402         * cgraph.h (symtab_node_base): Add weakref flag.
17403         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
17404         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
17405         (output_weakrefs): Use weakref flag.
17406         * fold-const.c (simple_operand_p): Handle WEAK.
17407         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
17408         * ipa.c (varpool_externally_visible_p): Drop weakref.
17409         (function_and_variable_visibility): Update comment; fix weakref
17410         sanity checks; do not clear DECL_WEAK on them.
17411         * lto-cgraph.c (lto_output_node): update.
17412         (lto_output_varpool_node): Update.
17413         (input_overwrite_node): Update.
17414         (input_node): Update.
17415         (input_varpool_node): Update.
17416         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
17417         (lto_symtab_merge_symbols): Add sanity check.
17418         (lto_symtab_prevailing_decl): Do not special case weakrefs.
17419         * passes.c (rest_of_decl_compilation): Set static flag, too.
17420         * symtab.c (dump_symtab_base): Dump weakref.
17421         (verify_symtab_base): Sanity check weakrefs.
17422         (symtab_make_decl_local): Remove duplicated code.
17423         (symtab_alias_ultimate_target): Simplify.
17424         * varpool.c (varpool_create_variable_alias): Set weakref flag.
17426 2013-06-11  Tom de Vries  <tom@codesourcery.com>
17428         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
17429         sequence_vect == NULL.
17431 2013-06-11  DJ Delorie  <dj@redhat.com>
17433         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
17434         (rl78_unwind_word_mode): New.
17436 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17438         * final.c (debug_prefix_maps): Make static.
17440 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17442         * function.c (initial_trampoline): Remove stray copy.
17444 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
17446         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
17448 2013-06-11  Martin Jambor  <mjambor@suse.cz>
17450         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
17451         within bounds at the beginning of the function.
17453 2013-06-11  Alan Modra  <amodra@gmail.com>
17455         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
17456         reporting.
17457         (get_named_section): Don't NULL !DECL_P decl.
17459 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
17461         * doc/invoke.texi (core-avx2): Document.
17462         (slm): Likewise.
17463         (atom): Updated with MOVBE.
17465 2013-06-11  Richard Biener  <rguenther@suse.de>
17467         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
17469 2013-06-11  Anton Blanchard  <anton@samba.org>
17471         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
17472         correct shift value in little-endian mode.
17474 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
17476         PR target/56564
17477         * varasm.c (get_variable_align): Move #endif to the right place.
17479 2013-06-10  Cary Coutant  <ccoutant@google.com>
17481         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
17482         for hash so that hash table traversal order is deterministic.
17484 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
17485             Pat Haugen  <pthaugen@us.ibm.com>
17486             Peter Bergner  <bergner@vnet.ibm.com>
17488         * config/rs6000/vector.md (GPR move splitter): Do not split moves
17489         of vectors in GPRS if they are direct moves or quad word load or
17490         store moves.
17492         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
17493         declaration.
17494         (direct_move_p): Likewise.
17495         (quad_load_store_p): Likewise.
17497         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
17498         classes into bins based on the physical register type.
17499         (reg_class_to_reg_type): Likewise.
17500         (IS_STD_REG_TYPE): Likewise.
17501         (IS_FP_VECT_REG_TYPE): Likewise.
17502         (reload_fpr_gpr): Arrays to determine what insn to use if we can
17503         use direct move instructions.
17504         (reload_gpr_vsx): Likewise.
17505         (reload_vsx_gpr): Likewise.
17506         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
17507         information that is a simplification of register classes.  Also
17508         precalculate direct move reload helpers.
17509         (direct_move_p): New function to return true if the operation can
17510         be done as a direct move instruciton.
17511         (quad_load_store_p): New function to return true if the operation
17512         is a quad memory operation.
17513         (rs6000_legitimize_address): If quad memory, only allow register
17514         indirect for TImode addresses.
17515         (rs6000_legitimate_address_p): Likewise.
17516         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
17517         (rs6000_reload_register_type): Likewise.
17518         (register_to_reg_type): Return register type.
17519         (rs6000_secondary_reload_simple_move): New helper function for
17520         secondary reload and secondary memory needed to identify anything
17521         that is a simple move, and does not need reloading.
17522         (rs6000_secondary_reload_direct_move): New helper function for
17523         secondary reload to identify cases that can be done with several
17524         instructions via the direct move instructions.
17525         (rs6000_secondary_reload_move): New helper function for secondary
17526         reload to identify moves between register types that can be done.
17527         (rs6000_secondary_reload): Add support for quad memory operations
17528         and for direct move.
17529         (rs6000_secondary_memory_needed): Likewise.
17530         (rs6000_debug_secondary_memory_needed): Change argument names.
17531         (rs6000_output_move_128bit): New function to return the move to
17532         use for 128-bit moves, including knowing about the various
17533         limitations of quad memory operations.
17535         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
17536         memory operations.  call rs6000_output_move_128bit for the actual
17537         instruciton(s) to generate.
17538         (vsx_movti_64bit): Likewise.
17540         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
17541         (UNSPEC_P8V_MTVSRWZ): Likewise.
17542         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
17543         (UNSPEC_P8V_MTVSRD): Likewise.
17544         (UNSPEC_P8V_XXPERMDI): Likewise.
17545         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
17546         (UNSPEC_FUSION_GPR): Likewise.
17547         (FMOVE128_GPR): New iterator for direct move.
17548         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
17549         (f32_sv): Likewise.
17550         (f32_dm): Likewise.
17551         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
17552         loads and direct move instructions.
17553         (zero_extendsidi2_lfiwzx): Likewise.
17554         (extendsidi2_lfiwax): Likewise.
17555         (extendsidi2_nocell): Likewise.
17556         (floatsi<mode>2_lfiwax): Likewise.
17557         (lfiwax): Likewise.
17558         (floatunssi<mode>2_lfiwzx): Likewise.
17559         (lfiwzx): Likewise.
17560         (fix_trunc<mode>_stfiwx): Likewise.
17561         (fixuns_trunc<mode>_stfiwx): Likewise.
17562         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
17563         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
17564         (parity<mode>2_cmpb): Set length/type attr.
17565         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
17566         for 'mr.' to fast_compare.
17567         (bpermd_<mode>): Change type attr to popcnt.
17568         (p8_fmrgow_<mode>): New insns for power8 direct move support.
17569         (p8_mtvsrwz_1): Likewise.
17570         (p8_mtvsrwz_2): Likewise.
17571         (reload_fpr_from_gpr<mode>): Likewise.
17572         (p8_mtvsrd_1): Likewise.
17573         (p8_mtvsrd_2): Likewise.
17574         (p8_xxpermdi_<mode>): Likewise.
17575         (reload_vsx_from_gpr<mode>): Likewise.
17576         (reload_vsx_from_gprsf): Likewise.
17577         (p8_mfvsrd_3_<mode>): LIkewise.
17578         (reload_gpr_from_vsx<mode>): Likewise.
17579         (reload_gpr_from_vsxsf): Likewise.
17580         (p8_mfvsrd_4_disf): Likewise.
17581         (multi-word GPR splits): Do not split direct moves or quad memory
17582         operations.
17584 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
17586         * tree-into-ssa.c (interesting_blocks): Make static.
17588 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
17590         PR target/56564
17591         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
17592         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
17593         Use DATA_ABI_ALIGNMENT for that case instead if defined.
17594         (get_variable_align): New function.
17595         (get_variable_section, emit_bss, emit_common,
17596         assemble_variable_contents, place_block_symbol): Use
17597         get_variable_align instead of DECL_ALIGN.
17598         (assemble_noswitch_variable): Add align argument, use it
17599         instead of DECL_ALIGN.
17600         (assemble_variable): Adjust caller.  Use get_variable_align
17601         instead of DECL_ALIGN.
17602         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
17603         caller.
17604         (DATA_ABI_ALIGNMENT): Define.
17605         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
17606         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
17607         opt is false, only return the psABI mandated alignment increase.
17608         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
17609         (DATA_ABI_ALIGNMENT): ... this.
17610         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
17611         (DATA_ABI_ALIGNMENT): ... this.
17612         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
17613         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
17614         (DATA_ABI_ALIGNMENT): ... this.
17615         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
17616         * doc/tm.texi: Regenerated.
17618 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
17620         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
17621         cmp_code to construct REG_EQUAL note.
17623 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
17625         PR target/57568
17626         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
17627         that operands[2] doesn't overlap with operands[0].
17629 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
17630             Jan Hubicka  <jh@suse.cz>
17632         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
17633         hack to mark symbols as used.
17635 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
17637         PR rtl-optimization/57559
17638         * lra-constraints.c (process_alt_operands): Don't discourage
17639         memory with known offset for offsetable memory constraint.
17640         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
17642 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
17644         * varasm.c (struct oc_local_state): Reorder fields.
17645         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
17646         and adjust accordingly.
17647         (output_constructor): Reorder initialization code and adjust call to
17648         output_constructor_bitfield.
17650 2013-06-07  Jan Hubicka  <jh@suse.cz>
17652         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
17654 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
17656         * tree-object-size.c (unknown): Make const.
17658 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17660         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
17661         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
17662         for last alternative in the cpu_facility attribute.
17664 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17666         PR target/56315
17667         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
17668         (xordi3): Change operand 2 constraint to arm_xordi_operand.
17669         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
17670         * config/arm/constraints.md (Dg): New constraint.
17671         * config/arm/neon.md (xordi3_neon): Remove.
17672         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
17673         * config/arm/predicates.md (arm_xordi_operand): New predicate.
17675 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17677         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
17678         Clean up alternatives.
17680 2013-06-07  Alan Modra  <amodra@gmail.com>
17682         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
17683         va_list_gpr_size.
17685 2013-06-07  Alan Modra  <amodra@gmail.com>
17687         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
17689 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17691         * config/arm/constraints.md (Df): New constraint.
17692         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
17693         Correct length attribute for last two alternatives.
17695 2013-06-07  Alan Modra  <amodra@gmail.com>
17697         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
17698         override user -mfp-in-toc.
17699         (offsettable_ok_by_alignment): Consider just the current access
17700         rather than the whole object, unless BLKmode.  Handle
17701         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
17702         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
17703         for -mcmodel=medium.
17704         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
17705         override user -mfp-in-toc or -msum-in-toc.  Default to
17706         -mno-fp-in-toc for -mcmodel=medium.
17708 2013-06-06  DJ Delorie  <dj@redhat.com>
17710         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
17711         TARGET_VALID_POINTER_MODE.
17713 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
17714             Pat Haugen  <pthaugen@us.ibm.com>
17715             Peter Bergner  <bergner@vnet.ibm.com>
17717         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
17718         Document new power8 builtins.
17720         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
17721         condition code register, to allow 128-bit logical operations to be
17722         done in the VSX or GPR registers.
17723         (nor<mode>3): Use the canonical form for nor.
17724         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
17725         vclz*, and vpopcnt* vector instructions.
17726         (nand<mode>3): Likewise.
17727         (orc<mode>3): Likewise.
17728         (clz<mode>2): LIkewise.
17729         (popcount<mode>2): Likewise.
17731         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
17732         that only the GPRs are recognized.
17734         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17735         support for new power8 builtins.
17737         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
17738         builtin functions.
17739         (xscvdpspn): Likewise.
17740         (vclz): Likewise.
17741         (vclzb): Likewise.
17742         (vclzh): Likewise.
17743         (vclzw): Likewise.
17744         (vclzd): Likewise.
17745         (vpopcnt): Likewise.
17746         (vpopcntb): Likewise.
17747         (vpopcnth): Likewise.
17748         (vpopcntw): Likewise.
17749         (vpopcntd): Likewise.
17750         (vgbbd): Likewise.
17751         (vmrgew): Likewise.
17752         (vmrgow): Likewise.
17753         (eqv): Likewise.
17754         (eqv_v16qi3): Likewise.
17755         (eqv_v8hi3): Likewise.
17756         (eqv_v4si3): Likewise.
17757         (eqv_v2di3): Likewise.
17758         (eqv_v4sf3): Likewise.
17759         (eqv_v2df3): Likewise.
17760         (nand): Likewise.
17761         (nand_v16qi3): Likewise.
17762         (nand_v8hi3): Likewise.
17763         (nand_v4si3): Likewise.
17764         (nand_v2di3): Likewise.
17765         (nand_v4sf3): Likewise.
17766         (nand_v2df3): Likewise.
17767         (orc): Likewise.
17768         (orc_v16qi3): Likewise.
17769         (orc_v8hi3): Likewise.
17770         (orc_v4si3): Likewise.
17771         (orc_v2di3): Likewise.
17772         (orc_v4sf3): Likewise.
17773         (orc_v2df3): Likewise.
17775         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
17776         allow power8 quad mode in 64-bit.
17777         (rs6000_builtin_vectorized_function): Add support to vectorize
17778         ISA 2.07 count leading zeros, population count builtins.
17779         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
17780         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
17781         (builtin_function_type): Add vgbbd builtin function which takes an
17782         unsigned argument.
17783         (altivec_expand_vec_perm_const): Add support for new power8 merge
17784         instructions.
17786         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
17787         that does not include TImdoe for use with 32-bit.
17788         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
17789         instructions.
17790         (UNSPEC_VSX_CVDPSPN): Likewise.
17791         (vsx_xscvdpspn): Likewise.
17792         (vsx_xscvspdpn): Likewise.
17793         (vsx_xscvdpspn_scalar): Likewise.
17794         (vsx_xscvspdpn_directmove): Likewise.
17795         (vsx_and<mode>3): Split logical operations into 32-bit and
17796         64-bit. Add support to do logical operations on TImode as well as
17797         VSX vector types.  Allow logical operations to be done in either
17798         VSX registers or in general purpose registers in 64-bit mode.  Add
17799         splitters if GPRs were used. For AND, add clobber of CCmode to
17800         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
17801         encoding.
17802         (vsx_and<mode>3_32bit): Likewise.
17803         (vsx_and<mode>3_64bit): Likewise.
17804         (vsx_ior<mode>3): Likewise.
17805         (vsx_ior<mode>3_32bit): Likewise.
17806         (vsx_ior<mode>3_64bit): Likewise.
17807         (vsx_xor<mode>3): Likewise.
17808         (vsx_xor<mode>3_32bit): Likewise.
17809         (vsx_xor<mode>3_64bit): Likewise.
17810         (vsx_one_cmpl<mode>2): Likewise.
17811         (vsx_one_cmpl<mode>2_32bit): Likewise.
17812         (vsx_one_cmpl<mode>2_64bit): Likewise.
17813         (vsx_nor<mode>3): Likewise.
17814         (vsx_nor<mode>3_32bit): Likewise.
17815         (vsx_nor<mode>3_64bit): Likewise.
17816         (vsx_andc<mode>3): Likewise.
17817         (vsx_andc<mode>3_32bit): Likewise.
17818         (vsx_andc<mode>3_64bit): Likewise.
17819         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
17820         and xxlorc instructions.
17821         (vsx_eqv<mode>3_64bit): Likewise.
17822         (vsx_nand<mode>3_32bit): Likewise.
17823         (vsx_nand<mode>3_64bit): Likewise.
17824         (vsx_orc<mode>3_32bit): Likewise.
17825         (vsx_orc<mode>3_64bit): Likewise.
17827         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
17829         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
17830         instruction.
17831         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
17832         (p8_vmrgow): Likewise.
17833         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
17834         GPRs to be split under VSX.
17835         (p8v_clz<mode>2): Add power8 count leading zero support.
17836         (p8v_popcount<mode>2): Add power8 population count support.
17837         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
17838         support.
17840         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
17841         instruction.
17843         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
17844         builtin functions.
17845         (vec_nand): Likewise.
17846         (vec_vclz): Likewise.
17847         (vec_vclzb): Likewise.
17848         (vec_vclzd): Likewise.
17849         (vec_vclzh): Likewise.
17850         (vec_vclzw): Likewise.
17851         (vec_vgbbd): Likewise.
17852         (vec_vmrgew): Likewise.
17853         (vec_vmrgow): Likewise.
17854         (vec_vpopcnt): Likewise.
17855         (vec_vpopcntb): Likewise.
17856         (vec_vpopcntd): Likewise.
17857         (vec_vpopcnth): Likewise.
17858         (vec_vpopcntw): Likewise.
17860 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17862         PR rtl-optimization/57468
17863         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
17864         spilled pseudos.
17866 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17868         PR rtl-optimization/57459
17869         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
17870         type when setting live regs.
17872 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17874         * config/s390/s390.opt (mlra): New option.
17875         * config/s390/s390.c (s390_decompose_address): Check displacement
17876         for all registers for LRA.
17877         (s390_secondary_reload): Don't used secondary reloads for LRA.
17878         (s390_lra_p): New function.
17879         (TARGET_LRA_P): Define.
17880         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
17881         of attribute cpu_facility to zarch for the last alternative.
17882         (*cmpmem_short): Ditto.
17884 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
17886         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
17887         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
17888         (arm_expand_prologue): Likewise.
17890 2013-06-06  Teresa Johnson  <tejohnson@google.com>
17892         PR c++/53743
17893         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
17894         as this is now done by redirect_edge_and_branch_force.
17895         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
17896         barriers, and fix interaction with splitting.
17897         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
17898         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
17899         reflect changes made in the routine.
17900         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
17901         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
17902         since this is called in cfglayout mode, and replace partition fixup
17903         with assert as that is now done by force_nonfallthru_and_redirect.
17904         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
17905         already be marked with region crossing note.
17906         (insert_section_boundary_note): Make non-static, gate on flag
17907         has_bb_partition, rewrite to also check for multiple partitions.
17908         (rest_of_handle_reorder_blocks): Remove call to
17909         insert_section_boundary_note, now done later during free_cfg.
17910         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
17911         * bb-reorder.h (insert_section_boundary_note): Declare.
17912         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
17913         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
17914         invoke insert_section_boundary_note.
17915         (try_redirect_by_replacing_jump): Remove unnecessary
17916         check for region crossing note.
17917         (fixup_partition_crossing): New function.
17918         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
17919         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
17920         in non-cfglayout mode.
17921         (force_nonfallthru_and_redirect): Fixup partition boundaries,
17922         remove old code that tried to do this. Emit barrier correctly
17923         when we are in cfglayout mode.
17924         (last_bb_in_partition): New function.
17925         (rtl_split_edge): Correctly fixup partition boundaries.
17926         (commit_one_edge_insertion): Remove old code that tried to
17927         fixup region crossing edge since this is now handled in
17928         split_block, and set up insertion point correctly since
17929         block may now end in a jump.
17930         (verify_hot_cold_block_grouping): Guard against checking when not in
17931         linearized RTL mode.
17932         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
17933         notes.
17934         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
17935         rtl_verify_flow_info, so not called in cfglayout mode.
17936         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
17937         (fixup_reorder_chain): Remove old code that attempted to fixup region
17938         crossing note as this is now handled in force_nonfallthru_and_redirect.
17939         (duplicate_insn_chain): Don't duplicate switch section notes.
17940         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
17941         note.
17942         * basic-block.h (emit_barrier_after_bb): Declare.
17944 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17946         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
17947         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
17948         arm_usatsihi): Adjust alternatives for arm_restrict_it.
17950 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17952         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
17953         where appropriate.
17954         * config/arm/ldmstm.md: Regenerate.
17956 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17958         * config/arm/sync.md (atomic_loaddi_1):
17959         Disable predication for arm_restrict_it.
17960         (arm_load_exclusive<mode>): Likewise.
17961         (arm_load_exclusivesi): Likewise.
17962         (arm_load_exclusivedi): Likewise.
17963         (arm_load_acquire_exclusive<mode>): Likewise.
17964         (arm_load_acquire_exclusivesi): Likewise.
17965         (arm_load_acquire_exclusivedi): Likewise.
17966         (arm_store_exclusive<mode>): Likewise.
17967         (arm_store_exclusive<mode>): Likewise.
17968         (arm_store_release_exclusivedi): Likewise.
17969         (arm_store_release_exclusive<mode>): Likewise.
17971 2013-06-06  Richard Biener  <rguenther@suse.de>
17973         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
17974         after LTO_null.
17975         (lto_tag_is_tree_code_p): Adjust.
17976         (lto_tag_is_gimple_code_p): Likewise.
17977         (lto_gimple_code_to_tag): Likewise.
17978         (lto_tag_to_gimple_code): Likewise.
17979         (lto_tree_code_to_tag): Likewise.
17980         (lto_tag_to_tree_code): Likewise.
17981         * data-streamer.h (streamer_write_hwi_in_range): Use
17982         uhwi streaming to stream the normalized range.
17983         (streamer_read_hwi_in_range): Likewise.
17985 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17987         * config/arm/arm.md (enabled_for_depr_it): New attribute.
17988         (predicable_short_it): Likewise.
17989         (predicated): Likewise.
17990         (enabled): Handle above.
17991         (define_cond_exec): Set predicated attribute to yes.
17993 2013-06-05  Mike Stump  <mikestump@comcast.net>
17995         * gdbinit.in (__FUNCTION__): Add.
17997 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
17999         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
18000         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
18002 2013-06-05  Jan Hubicka  <jh@suse.cz>
18004         * varasm.c (mark_decl_referenced): Revert the removal until targets
18005         are fixed.
18007 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
18009         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
18010         instead of mark_decl_referenced.
18012 2013-06-05  Jan Hubicka  <jh@suse.cz>
18014         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
18015         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
18016         and symtab_used_from_object_file_p.
18017         (cgraph_make_node_local_1): Clear forced_by_abi.
18018         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
18019         * cgraph.h (symtab_node_base): Add forced_by_abi.
18020         (decide_is_variable_needed): Remove.
18021         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
18022         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
18023         (decide_is_symbol_needed): ... this one; handle symbols in general;
18024         always analyze virtuals; honnor forced_by_abi.
18025         (cgraph_finalize_function): Update.
18026         (varpool_finalize_decl): Update.
18027         (symbol_defined_and_needed): Remove.
18028         (analyze_functions): Update.
18029         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
18030         output_refs, input_overwrite_node): Handle forced_by_abi.
18031         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
18032         (address_taken_from_non_vtable_p): ... this one.
18033         (comdat_can_be_unshared_p_1): New function.
18034         (cgraph_comdat_can_be_unshared_p): Rename to ...
18035         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
18036         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
18037         (function_and_variable_visibility): Clear forced_by_abi as needed.
18038         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
18039         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
18040         * symtab.c (dump_symtab_base): Dump forced_by_abi.
18041         * varpool.c (decide_is_variable_needed): Remove.
18043 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18045         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
18046         (arm_option_override): Override arm_restrict_it where appropriate.
18047         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
18048         * config/arm/arm.opt (mrestrict-it): New command-line option.
18049         * doc/invoke.texi: Document -mrestrict-it.
18051 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18053         * tsan.c (tsan_atomic_table): Make const.
18055 2013-06-05  Richard Biener  <rguenther@suse.de>
18057         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
18058         index associated with the tree we are supposed to replace.
18059         * tree-streamer-out.c (pack_ts_base_value_fields): Output
18060         TREE_ASM_WRITTEN as zero for everything but SSA names.
18062 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18064         * tree-ssa-structalias.c (call_stmt_vars): Make static.
18066 2013-06-04  Jan Hubicka  <jh@suse.cz>
18068         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
18069         (input_node, input_varpool_node): Handle correctly external same
18070         body aliases.
18071         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
18072         nodes at ltrans stage.
18074 2013-06-04  Jan Hubicka  <jh@suse.cz>
18076         * ipa-inline.c (update_caller_keys): Fix availability test.
18077         (update_callee_keys): Likewise.
18078         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
18079         to follow ELF standard.
18081 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
18083         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
18084         (mips64r5900el-*-elf*): New configurations.
18085         * config/mips/mips-cpus.def (r5900): New processor.
18086         * config/mips/mips-tables.opt: Regenerate.
18087         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
18088         (mips_issue_rate): Handle PROCESSOR_R5900.
18089         (mips_reorg_process_insns): Force reorder mode for the R5900.
18090         * config/mips/mips.h (TARGET_MIPS5900): Define.
18091         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
18092         TARGET_MIPS5900.
18093         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
18094         TARGET_MIPS5900.
18095         * config/mips/mips.md (processor): Add r5900.
18096         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
18098 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18100         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
18101         into function to generate MOVI instruction.
18102         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
18103         (aarch64_preferred_simd_mode): Turn into wrapper.
18104         (aarch64_output_scalar_simd_mov_immediate): New function.
18105         * config/aarch64/aarch64-protos.h: Add prototype for above.
18107 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18109         * config/aarch64/aarch64.c (simd_immediate_info): Remove
18110         element_char member.
18111         (sizetochar): Return signed char.
18112         (aarch64_simd_valid_immediate): Remove elchar and other
18113         unnecessary variables.
18114         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
18115         Calculate element_char as required.
18116         * config/aarch64/aarch64-protos.h: Update and move prototype
18117         for aarch64_output_simd_mov_immediate.
18118         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
18119         Update arguments.
18121 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18123         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
18124         information completed by aarch64_simd_valid_immediate.
18125         (aarch64_legitimate_constant_p): Update arguments.
18126         (aarch64_simd_valid_immediate): Work with struct rather than many
18127         pointers.
18128         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
18129         (aarch64_simd_make_constant): Update arguments.
18130         (aarch64_output_simd_mov_immediate): Work with struct rather than
18131         many pointers.  Output immediate directly rather than as operand.
18132         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18133         Update prototype.
18134         * config/aarch64/constraints.md (Dn): Update arguments.
18136 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18138         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
18139         longer static.
18140         (aarch64_simd_immediate_valid_for_move): Remove.
18141         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
18142         (aarch64_simd_make_constant): Update call.
18143         (aarch64_output_simd_mov_immediate): Update call.
18144         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18145         Add prototype.
18146         * config/aarch64/constraints.md (Dn): Update call.
18148 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18150         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
18151         return type to bool for prototype.
18152         (aarch64_legitimate_constant_p): Check for true instead of not -1.
18153         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
18154         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
18156 2013-06-04  Catherine Moore  <clm@codesourcery.com>
18158         * config/mips/mips.opt (meva): New.
18159         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
18160         (ASM_SPEC): Handle -meva.
18161         * doc/invoke.texi (meva):  Document.
18163 2013-06-04  Alan Modra  <amodra@gmail.com>
18165         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
18166         constant output.
18168 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18170         * rtl.def: Add extra fourth optional field to define_cond_exec.
18171         * gensupport.c (process_one_cond_exec): Process attributes from
18172         define_cond_exec.
18173         * doc/md.texi: Document fourth field in define_cond_exec.
18175 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
18177         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
18178         out the processing order as in store_bit_field_1.
18180 2013-06-04  Jan Hubicka  <jh@suse.cz>
18182         PR middle-end/57500
18183         * cgraphunit.c (cgraph_process_same_body_aliases): Create
18184         non-VAR_DECL node if it does not exist yet.
18186 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
18188         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
18189         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
18190         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
18191         target_cpu_default setting.
18193 2013-06-03  Teresa Johnson  <tejohnson@google.com>
18195         * dumpfile.c (opt_info_switch_p): Change -fopt-info
18196         default to -fopt-info=optimized instead of all.
18197         * doc/invoke.texi: Ditto.
18198         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
18199         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
18200         (execute_vect_slp): Emit BB vectorization success under
18201         MSG_OPTIMIZED_LOCATIONS.
18202         * tree-vect-slp.c (vect_slp_transform_bb): Change
18203         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
18204         * tree-vect-loop.c (vect_transform_loop): Ditto.
18206 2013-06-03  Jason Merrill  <jason@redhat.com>
18208         PR c++/57415
18209         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18210         Use TARGET_EXPR for C++.
18212 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
18214         PR rtl-optimization/57268
18215         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
18216         if DEBUG_INSN_P (insn).
18218         Reapply
18219         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18221         PR rtl-optimization/57268
18222         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18223         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18225 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18227         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
18228         (ix86_avoid_lea_for_addr): Likewise.
18229         (exact_dependency_1): Likewise.
18230         (ix86_adjust_cost): Likewise.
18231         (swap_top_of_ready_list): Fix formatting and !reload_completed check
18232         removed.
18233         (do_reorder_for_imul): Fix typo, formatting and
18234         !reload_completed check removed.
18235         (ix86_sched_reorder): Fix typo and formatting.
18236         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
18237         list.
18239 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
18241         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
18243 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
18245         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
18246         <STRING_CST>: Likewise.
18247         <VECTOR_CST>: Likewise.
18249 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
18250             Mikael Morin  <mikael@gcc.gnu.org>
18252         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
18253         * config.in: Regenerated.
18254         * configure: Regenerated.
18256 2013-06-01  Jan Hubicka  <jh@suse.cz>
18258         PR middle-end/57366
18259         * cgraphunit.c (compile): When weakref is not supported,
18260         set up transparent aliases before final output pass.
18261         * varasm.c (assemble_alias): Do not try to do it here.
18263 2013-06-01  Jan Hubicka  <jh@suse.cz>
18265         PR middle-end/57467
18266         * passes.c (for_per_function): Skip unanalyzed functions.
18268 2013-06-01  Jan Hubicka  <jh@suse.cz>
18270         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
18271         (lto_symtab_merge_symbols_1): ... this one.
18272         (lto_symtab_merge_cgraph_nodes): Rename to ...
18273         (lto_symtab_merge_symbols): ... this one; simplify.
18274         * cgraph.c (same_body_aliases_done): Rename to ...
18275         (cpp_implicit_aliases_done): ... this one.
18276         (cgraph_create_function_alias): Update.
18277         (cgraph_same_body_alias): Update.
18278         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
18279         (verify_edge_corresponds_to_fndecl): Simplify.
18280         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
18281         (cgraph_node): Remove same_body_alias.
18282         (varpool_node): Remove alias_of and extra_name_alias.
18283         (same_body_aliases_done): Rename to ..
18284         (cpp_implicit_aliases_done): ... this one.
18285         (symtab_alias_ultimate_target): Add default parameter.
18286         (symtab_resolve_alias): New function.
18287         (fixup_same_cpp_alias_visibility): Declare.
18288         (cgraph_function_node): Add default parameter.
18289         (cgraph_node_asm_name): Likewise.
18290         (cgraph_function_or_thunk_node): Add default parameter; do
18291         not ICE when it is NULL.
18292         (varpool_variable_node): Likewise.
18293         * tree-emutls.c (create_emultls_var): Update.
18294         (ipa_lower_emutls): Update.
18295         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
18296         (cgraph_reset_node): Reset alias info.
18297         (cgraph_finalize_function): Update.
18298         (fixup_same_cpp_alias_visibility): Move to symtab.c.
18299         (analyze_function): Simplify.
18300         (cgraph_process_same_body_aliases): Simplify.
18301         (analyze_functions): Fixup same body aliases.
18302         (handle_alias_pairs): Simplify.
18303         (assemble_thunk): Update.
18304         (assemble_thunks_and_aliases): Update.
18305         (output_weakrefs): Rewrite.
18306         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
18307         (lto_output_varpool_node): Likewise.
18308         (compute_ltrans_boundary): Remve assert.
18309         (get_alias_symbol): New functoin.
18310         (input_node): Rewrite alias handling.
18311         (input_varpool_node): Likewise.
18312         * ipa-pure-const.c (propagate_pure_const): Fix formating.
18313         * ipa.c (process_references): Handle weakrefs correctly.
18314         (symtab_remove_unreachable_nodes): Likewise.
18315         * trans-mem.c (get_cg_data): Update.
18316         (ipa_tm_create_version_alias): Update.
18317         (ipa_tm_execute): Update.
18318         * symtab.c (dump_symtab_base): Dump aliases.
18319         (verify_symtab_base): Verify aliases.
18320         (symtab_node_availability): New function.
18321         (symtab_alias_ultimate_target): Simplify.
18322         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
18323         handle all the fixup cases.
18324         (symtab_resolve_alias): New function.
18325         * passes.c (ipa_write_summaries): Handle weakrefs.
18326         * varpool.c (varpool_analyze_node): Simplify.
18327         (assemble_aliases): Update.
18328         (varpool_create_variable_alias): Simplify.
18329         (varpool_extra_name_alias): Simplify.
18330         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
18331         (lto_symtab_merge_symbols): ... this one.
18333 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
18335         Revert
18336         PR rtl-optimization/57268
18337         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18338         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18340 2013-06-01  Tobias Burnus  <burnus@net-b.de>
18342         Partially reverted:
18343         2013-05-31  Tobias Burnus  <burnus@net-b.de>
18345         PR middle-end/57073
18346         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
18347         further up.
18349 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18351         PR rtl-optimization/57268
18352         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18353         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18355 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18357         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
18358         unordered comparison operators when -fno-trapping-math is in effect
18359         on the e500.
18360         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
18361         and implement unordered comparison operators properly on the e500.
18363 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18365         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
18366         for constant scalar integers.
18367         (simplify_relational_operation_1): Likewise.
18369 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
18371         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
18372         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
18373         Fix comment.
18375 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18376             Igor Zamyatin  <igor.zamyatin@intel.com>
18378         Silvermont (SLM) architecture performance tuning.
18379         * config/i386/i386.h (enum ix86_tune_indices): Add
18380         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
18381         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
18383         * config/i386/i386.c (initial_ix86_tune_features)
18384         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
18385         (ix86_lea_outperforms): Handle Silvermont tuning.
18386         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
18387         call.
18388         (ix86_use_lea_for_mov): Likewise.
18389         (ix86_avoid_lea_for_addr): Likewise.
18390         (ix86_lea_for_add_ok): Likewise.
18391         (exact_dependency_1): New function.
18392         (exact_store_load_dependency): Likewise.
18393         (ix86_adjust_cost): Handle Silvermont tuning.
18394         (do_reoder_for_imul): Likewise.
18395         (swap_top_of_ready_list): New function.
18396         (ix86_sched_reorder): Changed to handle Silvermont tuning.
18398         * config/i386/i386.md (peepholes that split memory operand in fp
18399         converts): New.
18401 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18403         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
18404         Remove un-necessary braces.
18406 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18408         * config/aarch64/aarch64.c (aarch64_classify_symbol):
18409         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
18411 2013-05-31  Tobias Burnus  <burnus@net-b.de>
18413         PR middle-end/57073
18414         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
18416 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18418         PR target/56315
18419         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
18420         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
18421         * config/arm/neon.md (iordi3_neon): Remove.
18422         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
18423         * config/arm/predicates.md (imm_for_neon_logic_operand):
18424         Move to earlier in the file.
18425         (neon_logic_op2): Likewise.
18426         (arm_iordi_operand_neon): New predicate.
18428 2013-05-31  Richard Biener  <rguenther@suse.de>
18430         PR tree-optimization/57478
18431         PR tree-optimization/57453
18432         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
18433         are life as well.
18435 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
18437         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
18438         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
18440 2013-05-30  Tobias Burnus  <burnus@net-b.de>
18441             Thomas Koenig  <tkoenig@gcc.gnu.org>
18443         PR middle-end/57073
18444         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
18445         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
18447 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
18449         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
18451 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
18453         * target.def (register_usage_leveling_p): New hook.
18454         * targhooks.c (default_register_usage_leveling_p): New.
18455         * targhooks.h (default_register_usage_leveling_p): New prototype.
18456         * lra-assigns.c (register_usage_leveling_p): Use the hook.
18457         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
18458         * doc/tm.texi: Update.
18459         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
18461 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
18463         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
18464         (*insv_reg<mode>): New define_insn.
18466 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
18468         PR rtl-optimization/57439
18469         * postreload.c (move2add_valid_value_p): Check that we have
18470         a zero subreg_regno_offset when accessing the register in
18471         the requested mode.
18473 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18474             Igor Zamyatin  <igor.zamyatin@intel.com>
18476         Silvermont (SLM) architecture pipeline model, tuning and
18477         insn selection.
18478         * config.gcc: Add slm config options and target.
18480         * config/i386/slm.md: New.
18482         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
18484         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
18485         PROCESSOR_SLM.
18486         (ix86_target_macros_internal): Likewise.
18488         * config/i386/i386.c (slm_cost): New cost.
18489         (m_SLM): New macro flag.
18490         (initial_ix86_tune_features): Set m_SLM.
18491         (x86_accumulate_outgoing_args): Likewise.
18492         (x86_arch_always_fancy_math_387): Likewise.
18493         (processor_target_table): Add slm cost.
18494         (cpu_names): Add slm cpu name.
18495         (x86_option_override_internal): Set SLM ISA.
18496         (ix86_issue_rate): New case PROCESSOR_SLM.
18497         (ia32_multipass_dfa_lookahead): Likewise.
18498         (fold_builtin_cpu): Add slm.
18500         * config/i386/i386.h (TARGET_SLM): New target macro.
18501         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
18502         (processor_type): Add PROCESSOR_SLM.
18504         * config/i386/i386.md (cpu): Add new value "slm".
18505         (slm.md): Include slm.md.
18507 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
18508             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18510         * config/arm/arm-protos.h: Add and update function protos.
18511         * config/arm/arm.c (use_simple_return_p): New added.
18512         (thumb2_expand_return): Check simple_return flag.
18513         * config/arm/arm.md: Add simple_return and conditional simple_return.
18514         * config/arm/iterators.md: Add iterator for return and simple_return.
18516 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18518         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
18519         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
18520         (arm_emit_vfp_multi_reg_pop): Likewise.
18521         (thumb2_emit_ldrd_pop): Likewise.
18522         (arm_expand_epilogue): Add misc REG_CFA notes.
18523         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
18525 2013-05-29  Lawrence Crowl  <crowl@google.com>
18527         * config/arm/t-arm: Update for below.
18529         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
18530         Change type to hash_table.  Update dependent calls and types.
18532         * config/i386/t-cygming: Update for below.
18534         * config/i386/t-interix: Update for below.
18536         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
18537         Change type to hash_table.  Update dependent calls and types.
18538         (i386_find_on_wrapper_list::wrappers): Likewise.
18540         * config/ia64/t-ia64: Update for below.
18542         * config/ia64/ia64.c (bundle_state_table):
18543         Change type to hash_table.  Update dependent calls and types.
18545         * config/mips/mips.c (mips_reorg_process_insns::htab):
18546         Change type to hash_table.  Update dependent calls and types.
18548         * config/sol2.c (solaris_comdat_htab):
18549         Change type to hash_table.  Update dependent calls and types.
18551         * config/t-sol2: Update for above.
18553 2013-05-29  Teresa Johnson  <tejohnson@google.com>
18555         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
18556         functions are not yet marked as defined.
18558 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
18559             Pat Haugen  <pthaugen@us.ibm.com>
18560             Peter Bergner  <bergner@vnet.ibm.com>
18562         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
18563         instructions.
18564         (VEC_A): Likewise.
18565         (VEC_C): Likewise.
18566         (vrotl<mode>3): Likewise.
18567         (vashl<mode>3): Likewise.
18568         (vlshr<mode>3): Likewise.
18569         (vashr<mode>3): Likewise.
18571         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18572         support for power8 V2DI builtins.
18574         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
18575         power8 V2DI builtins.
18576         (vupkhsw): Likewise.
18577         (vupklsw): Likewise.
18578         (vaddudm): Likewise.
18579         (vminsd): Likewise.
18580         (vmaxsd): Likewise.
18581         (vminud): Likewise.
18582         (vmaxud): Likewise.
18583         (vpkudum): Likewise.
18584         (vpksdss): Likewise.
18585         (vpkudus): Likewise.
18586         (vpksdus): Likewise.
18587         (vrld): Likewise.
18588         (vsld): Likewise.
18589         (vsrd): Likewise.
18590         (vsrad): Likewise.
18591         (vsubudm): Likewise.
18592         (vcmpequd): Likewise.
18593         (vcmpgtsd): Likewise.
18594         (vcmpgtud): Likewise.
18595         (vcmpequd_p): Likewise.
18596         (vcmpgtsd_p): Likewise.
18597         (vcmpgtud_p): Likewise.
18598         (vupkhsw): Likewise.
18599         (vupklsw): Likewise.
18600         (vaddudm): Likewise.
18601         (vmaxsd): Likewise.
18602         (vmaxud): Likewise.
18603         (vminsd): Likewise.
18604         (vminud): Likewise.
18605         (vpksdss): Likewise.
18606         (vpksdus): Likewise.
18607         (vpkudum): Likewise.
18608         (vpkudus): Likewise.
18609         (vrld): Likewise.
18610         (vsld): Likewise.
18611         (vsrad): Likewise.
18612         (vsrd): Likewise.
18613         (vsubudm): Likewise.
18615         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
18616         support for power8 V2DI instructions.
18618         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
18619         power8 V2DI instructions.  Combine pack and unpack insns to use an
18620         iterator for each mode.  Check whether a particular mode supports
18621         Altivec instructions instead of just checking TARGET_ALTIVEC.
18622         (UNSPEC_VPKUWUM): Likewise.
18623         (UNSPEC_VPKSHSS): Likewise.
18624         (UNSPEC_VPKSWSS): Likewise.
18625         (UNSPEC_VPKUHUS): Likewise.
18626         (UNSPEC_VPKSHUS): Likewise.
18627         (UNSPEC_VPKUWUS): Likewise.
18628         (UNSPEC_VPKSWUS): Likewise.
18629         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
18630         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
18631         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
18632         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
18633         (UNSPEC_VUPKHSB): Likewise.
18634         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
18635         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
18636         (UNSPEC_VUPKHSH): Likewise.
18637         (UNSPEC_VUPKLSB): Likewise.
18638         (UNSPEC_VUPKLSH): Likewise.
18639         (VI2): Likewise.
18640         (VI_char): Likewise.
18641         (VI_scalar): Likewise.
18642         (VI_unit): Likewise.
18643         (VP): Likewise.
18644         (VP_small): Likewise.
18645         (VP_small_lc): Likewise.
18646         (VU_char): Likewise.
18647         (add<mode>3): Likewise.
18648         (altivec_vaddcuw): Likewise.
18649         (altivec_vaddu<VI_char>s): Likewise.
18650         (altivec_vadds<VI_char>s): Likewise.
18651         (sub<mode>3): Likewise.
18652         (altivec_vsubcuw): Likewise.
18653         (altivec_vsubu<VI_char>s): Likewise.
18654         (altivec_vsubs<VI_char>s): Likewise.
18655         (altivec_vavgs<VI_char>): Likewise.
18656         (altivec_vcmpbfp): Likewise.
18657         (altivec_eq<mode>): Likewise.
18658         (altivec_gt<mode>): Likewise.
18659         (altivec_gtu<mode>): Likewise.
18660         (umax<mode>3): Likewise.
18661         (smax<mode>3): Likewise.
18662         (umin<mode>3): Likewise.
18663         (smin<mode>3): Likewise.
18664         (altivec_vpkuhum): Likewise.
18665         (altivec_vpkuwum): Likewise.
18666         (altivec_vpkshss): Likewise.
18667         (altivec_vpkswss): Likewise.
18668         (altivec_vpkuhus): Likewise.
18669         (altivec_vpkshus): Likewise.
18670         (altivec_vpkuwus): Likewise.
18671         (altivec_vpkswus): Likewise.
18672         (altivec_vpks<VI_char>ss): Likewise.
18673         (altivec_vpks<VI_char>us): Likewise.
18674         (altivec_vpku<VI_char>us): Likewise.
18675         (altivec_vpku<VI_char>um): Likewise.
18676         (altivec_vrl<VI_char>): Likewise.
18677         (altivec_vsl<VI_char>): Likewise.
18678         (altivec_vsr<VI_char>): Likewise.
18679         (altivec_vsra<VI_char>): Likewise.
18680         (altivec_vsldoi_<mode>): Likewise.
18681         (altivec_vupkhsb): Likewise.
18682         (altivec_vupkhs<VU_char>): Likewise.
18683         (altivec_vupkls<VU_char>): Likewise.
18684         (altivec_vupkhsh): Likewise.
18685         (altivec_vupklsb): Likewise.
18686         (altivec_vupklsh): Likewise.
18687         (altivec_vcmpequ<VI_char>_p): Likewise.
18688         (altivec_vcmpgts<VI_char>_p): Likewise.
18689         (altivec_vcmpgtu<VI_char>_p): Likewise.
18690         (abs<mode>2): Likewise.
18691         (vec_unpacks_hi_v16qi): Likewise.
18692         (vec_unpacks_hi_v8hi): Likewise.
18693         (vec_unpacks_lo_v16qi): Likewise.
18694         (vec_unpacks_hi_<VP_small_lc>): Likewise.
18695         (vec_unpacks_lo_v8hi): Likewise.
18696         (vec_unpacks_lo_<VP_small_lc>): Likewise.
18697         (vec_pack_trunc_v8h): Likewise.
18698         (vec_pack_trunc_v4si): Likewise.
18699         (vec_pack_trunc_<mode>): Likewise.
18701         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
18702         V2DI builtins.
18703         (vec_vmaxsd): Likewise.
18704         (vec_vmaxud): Likewise.
18705         (vec_vminsd): Likewise.
18706         (vec_vminud): Likewise.
18707         (vec_vpksdss): Likewise.
18708         (vec_vpksdus): Likewise.
18709         (vec_vpkudum): Likewise.
18710         (vec_vpkudus): Likewise.
18711         (vec_vrld): Likewise.
18712         (vec_vsld): Likewise.
18713         (vec_vsrad): Likewise.
18714         (vec_vsrd): Likewise.
18715         (vec_vsubudm): Likewise.
18716         (vec_vupkhsw): Likewise.
18717         (vec_vupklsw): Likewise.
18719 2013-05-29  Jan Hubicka  <jh@suse.cz>
18721         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
18722         flags; reorder rest of fields in more consistent way.
18723         (varpool_node): Remove analyzed, finalized and alias.
18724         (cgraph_ndoe): Likewise.
18725         (symtab_alias_ultimate_target): New function.
18726         (cgraph_function_node): Move offline.
18727         (cgraph_reset_node): Declare.
18728         (cgraph_comdat_can_be_unshared_p): Remove.
18729         (varpool_remove_initializer): Declare.
18730         (varpool_first_defined_variable, varpool_next_defined_variable
18731         cgraph_first_defined_function, cgraph_next_defined_function): Update.
18732         (cgraph_function_with_gimple_body_p): Update.
18733         (varpool_all_refs_explicit_p): Update.
18734         (symtab_alias_target): New function.
18735         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
18736         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
18737         (cgraph_function_or_thunk_node): Simplify using
18738         symtab_alias_ultimate_target.
18739         (varpool_variable_node): Likewise.
18740         * cgraph.c (cgraph_create_function_alias): Update.
18741         (cgraph_add_thunk): Update.
18742         (cgraph_remove_node): Update.
18743         (dump_cgraph_node): Do not dump removed flags.
18744         (cgraph_function_body_availability): Update.
18745         (cgraph_propagate_frequency): Update.
18746         (verify_cgraph_node): Check sanity of local flag.
18747         (cgraph_function_node): Move here from cgraph.h; revamp for
18748         cgraph_function_or_thunk_node.
18749         * lto-symtab.c (lto_varpool_replace_node): Update.
18750         (lto_symtab_resolve_can_prevail_p): Update.
18751         (lto_symtab_merge_cgraph_nodes): Update.
18752         * ipa-cp.c (determine_versionability, initialize_node_lattices,
18753         propagate_constants_accross_call, devirtualization_time_bonus,
18754         ipcp_propagate_stage): Update.
18755         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
18756         * ipa-inline-transform.c (clone_inlined_nodes,
18757         preserve_function_body_p): Update.
18758         * ipa-reference.c (propagate): Update.
18759         (write_node_summary_p): Update.
18760         * toplev.c (wrapup_global_declaration_2): Update.
18761         * cgraphunit.c (cgraph_analyze_function): Rename to ...
18762         (analyze_function) ... this one.
18763         (cgraph_process_new_functions): Update.
18764         (cgraph_reset_node): Export.
18765         (cgraph_finalize_function): Update.
18766         (cgraph_add_new_function): Update.
18767         (process_function_and_variable_attributes): Update.
18768         (varpool_finalize_decl): Update.
18769         (symbol_finalized): Remove.
18770         (symbol_finalized_and_needed): Rename to ...
18771         (symbol_defined_and_needed): ... update.
18772         (cgraph_analyze_functions): Update.
18773         (handle_alias_pairs): Update.
18774         (mark_functions_to_output): Update.
18775         (assemble_thunk): Update.
18776         (output_in_order): Update.
18777         (output_weakrefs): Update.
18778         (finalize_compilation_unit): Update.
18779         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
18780         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
18781         input_node, input_varpool_node): Update.
18782         * dbxout.c (dbxout_expand_expr): Update.
18783         * cgraphclones.c (cgraph_clone_node): Update.
18784         (cgraph_copy_node_for_versioning): Update.
18785         (cgraph_materialize_clone): Update.
18786         (cgraph_materialize_all_clones): Update.
18787         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
18788         propagate_pure_const, propagate_nothrow): Update.
18789         * lto-streamer-out.c (lto_output, write_symbol): Update.
18790         * ipa-utils.c (ipa_reverse_postorder): Update.
18791         * ipa-inline.c (can_inline_edge_p): Update.
18792         (update_caller_keys, ipa_inline): Update.
18793         * dwarf2out.c (reference_to_unused,
18794         premark_types_used_by_global_vars_helper): Update.
18795         * tree-eh.c (tree_could_trap_p): Update.
18796         * ipa-split.c (consider_split, execute_split_functions): Update.
18797         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
18798         has_addr_references_p): Update; move ahead in file for better
18799         readability.
18800         (process_references): Simplify.
18801         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
18802         bodies are removed.
18803         (cgraph_comdat_can_be_unshared_p): Make static.
18804         (cgraph_externally_visible_p): Update.
18805         (varpool_externally_visible_p): Update.
18806         (function_and_variable_visibility): Update.
18807         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
18808         ipa_tm_mark_force_output_node): Update.
18809         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
18810         estimate_edge_devirt_benefit, inline_generate_summary,
18811         inline_write_summary): Update.
18812         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
18813         * ipa-prop.c (ipa_compute_jump_functions): Update.
18814         (ipa_print_node_params, ipa_prop_read_section,
18815         ipa_update_after_lto_read, read_replacements_section): Update.
18816         * varasm.c (mark_decl_referenced): Update.
18817         (assemble_alias, dump_tm_clone_pairs): Update.
18818         * tree-inline.c (copy_bb): Update.
18819         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
18820         Update.
18821         * symtab.c (dump_symtab_base): Print new flags.
18822         (verify_symtab_base): Verify new flags.
18823         (symtab_alias_ultimate_target): New function.
18824         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
18825         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
18826         Update.
18827         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
18828         Update.
18829         * i386.c (ix86_get_function_versions_dispatcher,
18830         ix86_generate_version_dispatcher_body): Update.
18831         (fold_builtin_cpu): Use varpool_add_new_variable.
18832         * varpool.c (varpool_remove_initializer): Break out from ...
18833         (varpool_remove_node): ... this one.
18834         (dump_varpool_node, varpool_node_for_asm,
18835         cgraph_variable_initializer_availability, varpool_analyze_node,
18836         varpool_assemble_decl, varpool_remove_unreferenced_decls,
18837         varpool_finalize_named_section_flags, varpool_create_variable_alias):
18838         Update.
18840 2013-05-29  Jan Hubicka  <jh@suse.cz>
18842         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
18844 2013-05-29  Easwaran Raman  <eraman@google.com>
18846         PR tree-optimization/57442
18847         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
18848         when control exits the main loop.
18850 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
18852         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
18853         and RX600.
18854         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
18855         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
18856         * rx/t-rx: Add rx100 under multi library matches option for nofpu
18857         option.
18859 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18861         PR tree-optimization/57441
18862         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
18863         Don't limit size of incr_vec to number of candidates.
18865 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
18867         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
18868         and mips16 directories.
18869         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
18870         (MULTILIB_DIRNAMES): Ditto.
18871         (MULTILIB_EXCEPTIONS): Add new exceptions.
18872         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
18873         (MULTILIB_DIRNAMES): Ditto.
18874         (MULTILIB_EXCEPTIONS): Add new exceptions.
18876 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18877             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18879         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
18880         SYMBOL_TINY_ABSOLUTE.
18881         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
18882         SYMBOL_TINY_ABSOLUTE.
18883         (aarch64_expand_mov_immediate): Likewise.
18884         (aarch64_classify_symbol): Likewise.
18885         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
18886         Permit SYMBOL_TINY_ABSOLUTE.
18887         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
18889 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18890             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18892         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
18893         Refactor if/switch.  Replace gcc_assert with if.
18895 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
18897         * config/i386/i386.c (initial_ix86_tune_features): Enable
18898         FP Reassociation for AMD bdver1 and bdver2.
18900 2013-05-29  Martin Jambor  <mjambor@suse.cz>
18902         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
18903         and IMAGPART_EXPR do not occur within other handled_components.
18905 2013-05-29  Richard Biener  <rguenther@suse.de>
18907         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
18908         access on whether the use is in the BB we currently try to
18909         vectorize.
18910         (vect_bb_vectorization_profitable_p): Pass the BB we currently
18911         vectorize to vect_bb_slp_scalar_cost.
18913 2013-05-29  Richard Biener  <rguenther@suse.de>
18915         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
18916         computing scalar cost offsetted by stmts that are kept live
18917         by scalar uses.
18918         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
18919         for computation of scalar cost.
18921 2013-05-28  Steve Ellcey  <sellcey@mips.com>
18923         * config/mips/mips-cpus.def (mips32r2): Change processor type.
18925 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
18927         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
18928         array notation built-in reduction functions.
18929         * doc/passes.texi (Passes): Added documentation about changes done
18930         for Cilk Plus.
18931         * doc/invoke.texi (C Dialect Options): Added documentation about
18932         the -fcilkplus flag.
18933         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
18934         (BUILTINS_DEF): Depend on cilkplus.def.
18935         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
18936         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
18937         * cilkplus.def: New file.
18939 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
18941         PR rtl-optimization/57439
18942         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
18944 2013-05-28  Easwaran Raman  <eraman@google.com>
18946         PR tree-optimization/57337
18947         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
18948         (find_insert_point): Correctly identify the insertion point
18949         when two statements with the same UID is compared.
18951 2013-05-28  Richard Biener  <rguenther@suse.de>
18953         PR tree-optimization/56787
18954         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
18955         from the list of data references.
18956         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
18957         clobbers.
18958         (vect_analyze_loop_operations): Likewise.
18959         (vect_transform_loop): Remove clobbers.
18961 2013-05-28  Martin Jambor  <mjambor@suse.cz>
18963         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
18964         and REALPART_EXPRs have scalar type.
18966 2013-05-28  Richard Biener  <rguenther@suse.de>
18968         PR tree-optimization/57411
18969         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
18970         virtual operands.
18971         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
18972         virtual operand propagation.
18974 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18976         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
18977         destination register for bmasksi_vis.
18978         (vector_init_bshuffle): Likewise.
18979         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
18981 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
18983         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
18984         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
18985         mode if the instruction isn't available in the original mode.
18986         * config/sparc/sparc.opt (mfix-ut699): New option.
18987         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
18988         (divdf3): Turn into expander.
18989         (divdf3_nofix): New insn.
18990         (divdf3_fix): Likewise.
18991         (divsf3): Disable if -mfix-ut699.
18992         (sqrtdf2): Turn into expander.
18993         (sqrtdf2_nofix): New insn.
18994         (sqrtdf2_fix): Likewise.
18995         (sqrtsf2): Disable if -mfix-ut699.
18997 2013-05-27  Richard Biener  <rguenther@suse.de>
18999         PR middle-end/57412
19000         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
19001         block for the new loop.
19003 2013-05-27  Richard Biener  <rguenther@suse.de>
19005         PR tree-optimization/57343
19006         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
19007         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
19008         (number_of_iterations_cond): Do not build the folded tree.
19010 2013-05-27  Richard Biener  <rguenther@suse.de>
19012         Revert
19013         PR middle-end/57381
19014         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19015         OEP_CONSTANT_ADDRESS_OF retained.
19017         PR tree-optimization/57417
19018         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
19019         for unchanged base.
19020         (set_ssa_val_to): Compare addresses using
19021         get_addr_base_and_unit_offset.
19023 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
19025         PR rtl-optimization/56833
19026         * postreload.c (move2add_record_mode): New function.
19027         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
19028         (move2add_use_add2_insn): Use move2add_record_sym_value.
19029         (move2add_use_add3_insn): Likewise.
19030         (reload_cse_move2add): Use move2add_valid_value_p and
19031         move2add_record_mode.  Invalidate call-clobbered and REG_INC
19032         affected regs by setting reg_mode to VOIDmode.
19033         (move2add_note_store): Don't pretend the inside of a SUBREG is
19034         the actual destination.  Invalidate single/leading registers by
19035         setting reg_mode to VOIDmode.
19036         Use move2add_record_sym_value, move2add_valid_value_p and
19037         move2add_record_mode.
19039 2013-05-27  Richard Biener  <rguenther@suse.de>
19041         PR tree-optimization/57396
19042         * tree-affine.c (double_int_constant_multiple_p): Properly
19043         return false for val == 0 and div != 0.
19045 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19047         * config/mips/mips.h: Use #elif in preprocessor conditions.
19049 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19051         PR target/53916
19052         * config/mips/constraints.md (kl): New constraint.
19053         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
19054         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
19055         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
19056         from LO for MIPS16.
19057         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
19059 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19061         PR target/55777
19062         * config/mips/mips.c (mips_can_inline_p): New function.
19063         (TARGET_CAN_INLINE_P): Define.
19065 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
19067         * sched-int.h (ds_t, dw_t): Make unsigned int.
19068         Fix documentation that describes how all the ds_t bits are used.
19069         Reserve the last bit for delayed-branch scheduling.
19070         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
19071         (BITS_PER_DEP_WEAK): Fix definition and documentation.
19072         (gen_dep_weak_1): Remove prototype.
19073         * sched-deps.c (get_dep_weak_1): Make static.
19074         * target.def (speculate_insn, needs_block_p, gen_spec_check,
19075         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
19076         * doc/tm.texi: Regenerate.
19077         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
19079 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
19081         PR debug/56950
19082         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
19084 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
19085             Sandra Loosemore  <sandra@codesourcery.com>
19087         * config.gcc (powerpc-*): Allow native for with-cpu.
19089 2013-05-24  Jeff Law  <law@redhat.com>
19091         PR tree-optimization/57124
19092         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
19093         conversion feeding a condition if the range has an overflow
19094         if -fstrict-overflow.  Add warnings for when we do make the
19095         transformation.
19097 2013-05-24  Dehao Chen  <dehao@google.com>
19099         * tree-cfg.c (locus_discrim_map): Fix the typo.
19100         (locus_discrim_hasher): Likewise.
19101         (locus_discrim_hasher::hash): Likewise.
19102         (locus_discrim_hasher::equal): Likewise.
19104 2013-05-24  Martin Jambor  <mjambor@suse.cz>
19106         PR tree-optimization/57294
19107         * cgraph.h (ipa_record_stmt_references): Declare.
19108         * cgraphbuild.c (ipa_record_stmt_references): New function.
19109         (build_cgraph_edges): Use ipa_record_stmt_references.
19110         (rebuild_cgraph_edges): Likewise.
19111         (cgraph_rebuild_references): Likewise.
19112         * ipa-prop.c (ipa_modify_call_arguments): Discard references
19113         associated with the old statement and build references from the
19114         newly built statements.
19115         * ipa-ref.c (ipa_remove_stmt_references): New function.
19116         * ipa-ref.h (ipa_remove_stmt_references): Declare.
19118 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
19120         * lra-constraints.c (emit_spill_move): Use smaller mode for
19121         mem-mem moves.
19122         (check_and_process_move): Consider mem-reg moves for secondary
19123         too.
19124         (curr_insn_transform): Don't lose insns emitted before for
19125         secondary memory moves.
19126         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
19127         reg set up in the current insn.
19129 2013-05-24  Dehao Chen  <dehao@google.com>
19131         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
19132         hash function.
19133         (locus_descrim_hasher::equal): Likewise.
19134         (build_gimple_cfg): New discriminator assignment algorithm.
19135         (make_edges): Likewise.
19136         (next_discriminator_for_locus): Likewise.
19137         (same_line_p): Likewise.
19138         (assign_discriminators): Likewise.
19139         (make_cond_expr_edges): Likewise.
19140         (make_gimple_switch_edges): Likewise.
19141         (make_goto_expr_edges): Likewise.
19142         (make_gimple_asm_edges): Likewise.
19144 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
19146         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
19147         X format specifier to only display bottom 16 bits.
19148         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
19149         immediate to match for operand 2, since it will be masked.
19151 2013-05-24  Richard Biener  <rguenther@suse.de>
19153         PR tree-optimization/57287
19154         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
19155         all SSA names that occur in abnormal PHIs.
19157 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
19159         PR tree-ssa/57385
19160         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
19161         that index is not negative.
19163 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
19165         PR rtl-optimization/55177
19166         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
19167         (simplify_byte_swapping_operation): New.
19168         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
19169         (simplify_relational_operation_1): Deal with BSWAP.
19171 2013-05-23  Richard Henderson  <rth@redhat.com>
19173         PR target/56742
19174         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
19175         (ix86_reorg): Call it.
19177 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
19179         PR target/57379
19180         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
19181         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
19182         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
19184 2013-05-23  Christian Bruel  <christian.bruel@st.com>
19186         PR debug/57351
19187         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
19189 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19190             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19192         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
19193         * config/aarch64/constraints.md (Usa): Remove.
19194         * doc/md.texi (AArch64 Usa): Remove.
19196 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19197             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19199         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
19200         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
19201         * config/aarch64/predicates.md (aarch64_const_address): Remove.
19202         (aarch64_mov_operand): Use aarch64_mov_operand_p.
19204 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
19206         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
19207         instruction (AdvSIMD).
19208         * config/aarch64/aarch64-builtins.c
19209         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
19210         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
19212 2013-05-23  Martin Jambor  <mjambor@suse.cz>
19214         PR middle-end/57347
19215         * tree.h (contains_bitfld_component_ref_p): Declare.
19216         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
19217         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
19218         caller.
19219         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
19220         not access a bit-field.  Assert all final offsets are byte-aligned.
19222 2013-05-23  Richard Biener  <rguenther@suse.de>
19224         PR tree-optimization/57380
19225         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
19226         least one invariant or re-used load.
19227         * passes.c (init_optimization_passes): Move pass_phiprop before
19228         pass_forwprop.
19230 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
19232         * config/aarch64/aarch64-simd.md
19233         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
19235 2013-05-23  Richard Biener  <rguenther@suse.de>
19237         PR middle-end/57381
19238         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19239         OEP_CONSTANT_ADDRESS_OF retained.
19241 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
19243         PR middle-end/57344
19244         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
19245         don't lower unit.  Handle unit not being always BITS_PER_WORD.
19247 2013-05-23  Richard Biener  <rguenther@suse.de>
19249         PR rtl-optimization/57341
19250         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
19251         instead of true_dependence.
19253 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
19255         * bb-reorder.c (branch_threshold): Make const.
19256         (exec_threshold): Ditto.
19258 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19259             Pat Haugen  <pthaugen@us.ibm.com>
19260             Peter Bergner  <bergner@vnet.ibm.com>
19262         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
19263         documentation for the power8 crypto builtins.
19265         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
19267         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
19268         macros for defining power8 builtin functions.
19269         (BU_P8V_AV_2): Likewise.
19270         (BU_P8V_AV_P): Likewise.
19271         (BU_P8V_VSX_1): Likewise.
19272         (BU_P8V_OVERLOAD_1): Likewise.
19273         (BU_P8V_OVERLOAD_2): Likewise.
19274         (BU_CRYPTO_1): Likewise.
19275         (BU_CRYPTO_2): Likewise.
19276         (BU_CRYPTO_3): Likewise.
19277         (BU_CRYPTO_OVERLOAD_1): Likewise.
19278         (BU_CRYPTO_OVERLOAD_2): Likewise.
19279         (XSCVSPDP): Fix typo, point to the correct instruction.
19280         (VCIPHER): Add power8 crypto builtins.
19281         (VCIPHERLAST): Likewise.
19282         (VNCIPHER): Likewise.
19283         (VNCIPHERLAST): Likewise.
19284         (VPMSUMB): Likewise.
19285         (VPMSUMH): Likewise.
19286         (VPMSUMW): Likewise.
19287         (VPERMXOR_V2DI): Likewise.
19288         (VPERMXOR_V4SI: Likewise.
19289         (VPERMXOR_V8HI: Likewise.
19290         (VPERMXOR_V16QI: Likewise.
19291         (VSHASIGMAW): Likewise.
19292         (VSHASIGMAD): Likewise.
19293         (VPMSUM): Likewise.
19294         (VPERMXOR): Likewise.
19295         (VSHASIGMA): Likewise.
19297         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
19298         __CRYPTO__ if the crypto instructions are available.
19299         (altivec_overloaded_builtins): Add support for overloaded power8
19300         builtins.
19302         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19303         support for power8 crypto builtins.
19304         (builtin_function_type): Likewise.
19305         (altivec_init_builtins): Add support for builtins that take vector
19306         long long (V2DI) arguments.
19308         * config/rs6000/crypto.md: New file, define power8 crypto
19309         instructions.
19311 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19312             Pat Haugen  <pthaugen@us.ibm.com>
19313             Peter Bergner  <bergner@vnet.ibm.com>
19315         * doc/invoke.texi (Option Summary): Add power8 options.
19316         (RS/6000 and PowerPC Options): Likewise.
19318         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
19319         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
19320         wm, wn, wr documentation.
19322         * config/rs6000/constraints.md (wm): New constraint for VSX
19323         registers if direct move instructions are enabled.
19324         (wn): New constraint for no registers.
19325         (wq): New constraint for quad word even GPR registers.
19326         (wr): New constraint if 64-bit instructions are enabled.
19327         (wv): New constraint if power8 vector instructions are enabled.
19328         (wQ): New constraint for quad word memory locations.
19330         * config/rs6000/predicates.md (const_0_to_15_operand): New
19331         constraint for 0..15 for crypto instructions.
19332         (gpc_reg_operand): If VSX allow registers in VSX registers as well
19333         as GPR and floating point registers.
19334         (int_reg_operand): New predicate to match only GPR registers.
19335         (base_reg_operand): New predicate to match base registers.
19336         (quad_int_reg_operand): New predicate to match even GPR registers
19337         for quad memory operations.
19338         (vsx_reg_or_cint_operand): New predicate to allow vector logical
19339         operations in both GPR and VSX registers.
19340         (quad_memory_operand): New predicate for quad memory operations.
19341         (reg_or_indexed_operand): New predicate for direct move support.
19343         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
19344         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
19345         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
19346         (POWERPC_MASKS): Add power8 options.
19347         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
19348         various options.
19350         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
19351         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
19353         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
19354         (-mpower8-fusion): New power8 options.
19355         (-mpower8-fusion-sign): Likewise.
19356         (-mpower8-vector): Likewise.
19357         (-mcrypto): Likewise.
19358         (-mdirect-move): Likewise.
19359         (-mquad-memory): Likewise.
19361         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
19362         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
19363         registers.
19364         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
19365         (rs6000_debug_vector_unit): Add p8_vector.
19366         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
19367         definitions.  Also print fusion state.
19368         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
19369         (rs6000_builtin_mask_calculate): Add power8 builtin support.
19370         (rs6000_option_override_internal): Add support for power8.
19371         (rs6000_common_init_builtins): Add debugging for skipped builtins
19372         if -mdebug=builtin.
19373         (rs6000_adjust_cost): Add power8 support.
19374         (rs6000_issue_rate): Likewise.
19375         (insn_must_be_first_in_group): Likewise.
19376         (insn_must_be_last_in_group): Likewise.
19377         (force_new_group): Likewise.
19378         (rs6000_register_move_cost): Likewise.
19379         (rs6000_opt_masks): Likewise.
19381         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
19382         power8 capable assembler, default to power7 options.
19383         (TARGET_DIRECT_MOVE): Likewise.
19384         (TARGET_CRYPTO): Likewise.
19385         (TARGET_P8_VECTOR): Likewise.
19386         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
19387         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
19388         (VECTOR_MEM_P8_VECTOR_P): Likewise.
19389         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
19390         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
19391         (TARGET_XSCVDPSPN): Likewise.
19392         (TARGET_XSCVSPDPN): Likewsie.
19393         (TARGET_SYNC_HI_QI): Likewise.
19394         (TARGET_SYNC_TI): Likewise.
19395         (MASK_CRYPTO): Likewise.
19396         (MASK_DIRECT_MOVE): Likewise.
19397         (MASK_P8_FUSION): Likewise.
19398         (MASK_P8_VECTOR): Likewise.
19399         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
19400         temporary used by some of the direct move instructions to get two FP
19401         temporary registers does not force creation of a stack frame.
19402         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
19403         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
19404         that any VSX registers are tieable, even if they are also an
19405         Altivec vector mode.
19406         (r6000_reg_class_enum): Add wm, wr, wv constraints.
19407         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
19408         (RS6000_BTM_CRYPTO): Likewise.
19409         (RS6000_BTM_COMMON): Likewise.
19411         * config/rs6000/rs6000.md (cpu attribute): Add power8.
19412         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
19413         (enum rs6000_vector): Add power8 vector support.
19415 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19417         PR target/19599
19418         PR target/57340
19419         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
19420         (any_sibcall_could_use_r3): this and handle indirect calls.
19421         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
19423 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19425         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
19427 2013-05-22  Richard Biener  <rguenther@suse.de>
19429         PR middle-end/57349
19430         * profile.c (branch_prob): Do not split blocks that are
19431         abnormally receiving from ECF_RETURNS_TWICE functions.
19433 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
19435         * recog.c (offsettable_address_addr_space_p): Fix calculation of
19436         address mode.  Move pointer mode initialization to the same place.
19438 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
19440         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
19441         while it has any effect.
19443 2013-05-21  Easwaran Raman  <eraman@google.com>
19445         PR tree-optimization/57322
19446         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
19447         UID of the statement added to the BB to be 1.
19449 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
19451         PR tree-optimization/57331
19452         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
19453         of conversion from pointer type to integral type with integer.
19455 2013-05-21  Martin Jambor  <mjambor@suse.cz>
19457         PR lto/57289
19458         * ipa-prop.c (ipa_read_node_info): Process param_used and
19459         controlled_uses in the same order as when writing.
19461 2013-05-21  Magnus Granberg  <baldrick@free.fr>
19463         PR plugins/56754
19464         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
19466 2013-05-21  Richard Biener  <rguenther@suse.de>
19468         PR tree-optimization/57318
19469         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
19470         estimate stmts with side-effects as likely eliminated.
19472 2013-05-21  Richard Biener  <rguenther@suse.de>
19474         PR tree-optimization/57330
19475         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
19476         preserve the call stmts fntype.
19478 2013-05-21  Richard Biener  <rguenther@suse.de>
19480         PR tree-optimization/57303
19481         * tree-ssa-sink.c (statement_sink_location): Improve killing
19482         stmt detection and properly handle self-assignments.
19484 2013-05-21  Christian Bruel  <christian.bruel@st.com>
19486         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
19487         spanning registers. LEAF_REG_REMAP is supported only for contiguous
19488         registers. Set register size out of the PARALLEL loop.
19490 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
19492         PR target/56547
19493         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
19494         (*fmasf4, *fmasf4_media): New insns.
19496 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19498         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
19499         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
19500         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
19501         (mips_idiv_insns): Update the comments to say that the returned
19502         instruction counts are in units of BASE_INSN_LENGTH.
19503         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
19504         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
19505         using 2 rather than 4 as the length of indirect MIPS16 and
19506         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
19507         length of a NOP.  Don't divide MIPS16 lengths by 2.
19508         (mips16_split_long_branches): Assume a branch is long if the
19509         length is greater than 4 rather than 8.
19510         * config/mips/mips.md (length): Give MIPS16 lengths directly,
19511         rather than multiplying them by 2.  Multiply instruction counts
19512         by BASE_INSN_LENGTH rather than 4.
19513         (*jump_mips16, tls_get_tp_mips16_<mode>)
19514         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
19516 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19518         * config/mips/mips.md (extended_mips16): Remove branch case.
19519         (length): Remove duplicated extended_mips16 test.
19521 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19523         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
19525 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
19527         * recog.h (Recog_data): Rename to...
19528         (recog_data_d): ...this.
19529         (recog_data): Update accordingly.
19530         * recog.c (recog_data): Likewise.
19531         * reload.c (save_recog_data): Likewise.
19532         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
19533         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
19535 2013-05-17  Julian Brown  <julian@codesourcery.com>
19537         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
19538         found in a REG_EQUAL note, invalidate it.
19540 2013-05-17  Easwaran Raman  <eraman@google.com>
19542         * tree-ssa-reassoc.c (find_insert_point): New function.
19543         (insert_stmt_after): Likewise.
19544         (get_def_stmt): Likewise.
19545         (ensure_ops_are_available): Likewise.
19546         (not_dominated_by): Likewise.
19547         (rewrite_expr_tree): Do not move statements beyond what is
19548         necessary. Remove call to swap_ops_for_binary_stmt...
19549         (reassociate_bb): ... and move it here.
19550         (build_and_add_sum): Assign UIDs for new statements.
19551         (linearize_expr): Likewise.
19552         (do_reassoc): Renumber gimple statement UIDs.
19554 2013-05-17  Jan Hubicka  <jh@suse.cz>
19556         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
19557         weakrefs.
19558         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
19559         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
19560         weakrefs.
19561         (output_weakrefs): Update.
19563 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
19564             Martin Jambor  <mjambor@suse.cz>
19566         PR middle-end/57276
19567         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
19568         value that corresponds to the given aggval is found in values vector.
19570 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
19572         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
19573         sse, sse2, sse3, ssse3 and sse4a flags to options.
19575 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
19577         * gengtype-state.c: (s_expr_writer): New class, to handle
19578         prettifying of output layout of s-expressions.
19579         (state_writer): New class, to write out gtype.state.
19580         (state_written_type_count): Move this variable into member data of
19581         state_writer.
19582         (s_expr_writer::s_expr_writer): New code: constructor for new class
19583         (state_writer::state_writer(): ditto
19584         (s_expr_writer::write_new_line): New function
19585         (s_expr_writer::write_any_indent): ditto
19586         (s_expr_writer::begin_s_expr): ditto
19587         (s_expr_writer::end_s_expr): ditto
19588         (write_state_fileloc): convert to method of state_writer...
19589         (state_writer:: write_state_fileloc): ...and use methods of
19590         s_expr_writer to write indentation into the gtype.state output file
19591         to visually represent the hierarchical structure of the list
19592         structures
19593         (write_state_fields): ditto, renaming to...
19594         (state_writer::write_state_fields)
19595         (write_state_a_string): ditto, renaming to...
19596         (state_writer::write_state_a_string)
19597         (write_state_string_option): ditto, renaming to...
19598         (state_writer::write_state_string_option)
19599         (write_state_type_option): ditto, renaming to...
19600         (state_writer::write_state_type_option)
19601         (write_state_nested_option): ditto, renaming to...
19602         (state_writer::write_state_nested_option)
19603         (write_state_option): ditto, renaming to...
19604         (state_writer::write_state_option)
19605         (write_state_options): ditto, renaming to...
19606         (state_writer::write_state_options)
19607         (write_state_lang_bitmap): ditto, renaming to...
19608         (state_writer::write_state_lang_bitmap)
19609         (write_state_version): ditto, renaming to...
19610         (state_writer::write_state_version)
19611         (write_state_scalar_type): ditto, renaming to...
19612         (state_writer::write_state_scalar_type)
19613         (write_state_string_type): ditto, renaming to...
19614         (state_writer::write_state_string_type)
19615         (write_state_undefined_type): ditto, renaming to...
19616         (state_writer::write_state_undefined_type)
19617         (write_state_struct_union_type): ditto, renaming to...
19618         (state_writer::write_state_struct_union_type)
19619         (write_state_struct_type): ditto, renaming to...
19620         (state_writer::write_state_struct_type)
19621         (write_state_user_struct_type): ditto, renaming to...
19622         (state_writer::write_state_user_struct_type)
19623         (write_state_lang_struct_type): ditto, renaming to...
19624         (state_writer::write_state_lang_struct_type)
19625         (write_state_param_struct_type): ditto, renaming to...
19626         (state_writer::write_state_param_struct_type)
19627         (write_state_pointer_type): ditto, renaming to...
19628         (state_writer::write_state_pointer_type)
19629         (write_state_array_type): ditto, renaming to...
19630         (state_writer::write_state_array_type)
19631         (write_state_gc_used): ditto, renaming to...
19632         (state_writer::write_state_gc_used)
19633         (write_state_common_type_content): ditto, renaming to...
19634         (state_writer::write_state_common_type_content)
19635         (write_state_type): ditto, renaming to...
19636         (state_writer::write_state_type)
19637         (write_state_pair_list): ditto, renaming to...
19638         (state_writer::write_state_pair_list)
19639         (write_state_pair): ditto, renaming to...
19640         (state_writer::write_state_pair)
19641         (write_state_typedefs): ditto, renaming to...
19642         (state_writer::write_state_typedefs)
19643         (write_state_structures): ditto, renaming to...
19644         (state_writer::write_state_structures)
19645         (write_state_param_structs): ditto, renaming to...
19646         (state_writer::write_state_param_structs)
19647         (write_state_variables): ditto, renaming to...
19648         (state_writer::write_state_variables)
19649         (write_state_srcdir): ditto, renaming to...
19650         (state_writer::write_state_srcdir)
19651         (write_state_files_list): ditto, renaming to...
19652         (state_writer::write_state_files_list)
19653         (write_state_languages): ditto, renaming to...
19654         (state_writer::write_state_languages)
19655         (write_state): create a state_writer instance and use it when
19656         writing out the state file
19658 2013-05-17  Mike Stump  <mikestump@comcast.net>
19660         PR rtl-optimization/57304
19661         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
19662         accessing DF_REF_REAL_LOC.
19664 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19666         PR rtl-optimization/57281
19667         PR rtl-optimization/57300
19668         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
19669         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
19670         what the other splitter did if the registers are dead.
19672 2013-05-17  Richard Biener  <rguenther@suse.de>
19674         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
19675         MEM_REF offsets.
19677 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19679         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
19680         linking.
19682 2013-05-17  Marek Polacek  <polacek@redhat.com>
19684         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
19685         length when doing non-zero store of storing '\0' to '\0'.
19687 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19689         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
19690         vect_external_def oprnd1 with loop_vinfo, try to emit
19691         optional cast, negation and and stmts on the loop preheader
19692         edge instead of into the pattern def seq.
19694         PR tree-optimization/57051
19695         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
19696         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
19698 2013-05-16  Nick Clifton  <nickc@redhat.com>
19700         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
19701         (rl78_is_naked_func): New function.
19702         (rl78_expand_prologue): Skip prologue generation for naked functions.
19703         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
19704         * doc/extend.texi (naked): Add RL78 to the list of processors
19705         that supports this attribute.
19707 2013-05-16  Jeff Law  <law@redhat.com>
19709         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
19711 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19713         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
19714         cache parameters using detect_caches_amd also for CYRIX,
19715         NSC and TM2 signatures.
19717 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19718             Dzianis Kahanovich  <mahatma@eu.by>
19720         PR target/45359
19721         PR target/46396
19722         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19723         VIA/Centaur processors and determine their cache parameters
19724         using detect_caches_amd.
19726 2013-05-16  Teresa Johnson  <tejohnson@google.com>
19728         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
19729         (rtl_verify_edges): New function.
19730         (rtl_verify_bb_insns): Ditto.
19731         (rtl_verify_bb_pointers): Ditto.
19732         (rtl_verify_bb_insn_chain): Ditto.
19733         (rtl_verify_fallthru): Ditto.
19734         (rtl_verify_bb_layout): Ditto.
19735         (rtl_verify_flow_info_1): Outline checks into new functions.
19736         (rtl_verify_flow_info): Ditto.
19738 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
19740         * cfghooks.c (copy_bbs): Add update_dominance argument.
19741         * cfghooks.h (copy_bbs): Update prototype.
19742         * tree-cfg.c (gimple_duplicate_sese_region):
19743         Add update_dominance argument.
19744         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
19745         * tree-ssa-loop-ch.c (copy_loop_headers): Update
19746         gimple_duplicate_sese_region call.
19747         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
19748         Update copy_bbs call.
19749         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
19750         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
19752 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19754         * tree-vectorizer.h (NUM_PATTERNS): Increment.
19755         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
19756         vect_recog_rotate_pattern.
19757         (vect_recog_rotate_pattern): New function.
19759 2013-05-16  Jason Merrill  <jason@redhat.com>
19761         * Makefile.in (LLINKER): New variable.
19762         (mostlyclean): Remove link mutex.
19763         * configure.ac: Handle --enable-link-mutex.
19764         * lock-and-run.sh: New script.
19766 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19768         PR target/19599
19769         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
19770         for NULL decl.
19772 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19774         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
19776 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
19778         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
19779         * config/arm/arm.c (next_consecutive_mem): New function.
19780         (gen_movmem_ldrd_strd): Likewise.
19781         * config/arm/arm.md (movmemqi): Update condition and code.
19782         (unaligned_loaddi, unaligned_storedi): New patterns.
19784 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19786         * config.gcc: Obsolete *-*-solaris2.9*.
19787         * doc/install.texi (Specific, *-*-solaris2*): Document it.
19789 2013-05-16  Richard Biener  <rguenther@suse.de>
19791         * passes.c (init_optimization_passes): Move pass_parallelize_loops
19792         earlier, after GRAPHITE transforms and IV canonicalization.
19794 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19796         * omp-low.c (extract_omp_for_data): For collapsed loops,
19797         if at least one of the loops is known at compile time to
19798         iterate zero times, set count to 0.
19799         (expand_omp_regimplify_p): New function.
19800         (expand_omp_for_generic): For collapsed loops, if at least
19801         one of the loops isn't known to iterate at least once,
19802         add runtime check with setting count to 0.
19803         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
19804         For unsigned types if it isn't known at compile time that
19805         the loop will iterate at least once, add runtime check to bypass
19806         the whole loop if initial condition isn't true.
19808 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
19810         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
19812 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
19814         PR middle-end/57286
19815         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
19816         transformations to avoid an infinite loop.
19818 2013-05-16  Marek Polacek  <polacek@redhat.com>
19820         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
19822 2013-05-15  Leif Ekblad  <leif@rdos.net>
19824         * config/i386/i386.c (ix86_decompose_address): Use
19825         DEFAULT_TLS_SEG_REG to access TLS segment register.
19826         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
19827         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
19828         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
19830 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
19832         PR target/57260
19833         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
19834         sibling calls to functions that would normally be lazily bound,
19835         unless $gp is call-clobbered.
19837 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
19839         * config/i386/i386.c (ix86_option_override_internal): Update
19840         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
19841         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
19842         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
19843         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
19844         of TARGET_3DNOW.
19845         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
19847 2013-05-15  Andreas Schwab  <schwab@suse.de>
19849         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
19850         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
19851         third operand.
19853 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19855         * loop-unroll.c (report_unroll_peel): Check decision before
19856         emitting unroll/peel message.
19858 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19860         * function.h (has_bb_partition): New rtl_data flag.
19861         (bb_reorder_complete): Ditto.
19862         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
19863         instead of flag_reorder_blocks_and_partition.
19864         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
19865         with some enhancements.
19866         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
19867         * bb-reorder.c (connect_traces): Check for has_bb_partition
19868         instead of flag_reorder_blocks_and_partition.
19869         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
19870         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
19871         verify_hot_cold_block_grouping.
19872         (partition_hot_cold_basic_blocks): Set has_bb_partition.
19874 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19876         PR target/19599
19877         * config/arm/predicates.md (call_insn_operand): New predicate.
19878         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
19879         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
19880         if insn is not a tail call.
19881         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
19882         registers.
19883         * config/arm/arm.h (enum reg_class): New caller save register class.
19884         (REG_CLASS_NAMES): Likewise.
19885         (REG_CLASS_CONTENTS): Likewise.
19886         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
19887         without decls.
19889 2013-05-15  Richard Biener  <rguenther@suse.de>
19891         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
19892         of MSG_OPTIMIZED_LOCATIONS.
19893         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
19894         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
19895         message.
19896         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
19897         of MSG_OPTIMIZED_LOCATIONS.
19898         (execute_vect_slp): Likewise.
19899         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
19900         (vect_create_cond_for_alias_checks): Likewise.
19901         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
19902         (vect_recog_widen_mult_pattern): Likewise.
19903         (vect_recog_widen_sum_pattern): Likewise.
19904         (vect_recog_over_widening_pattern): Likewise.
19905         (vect_recog_widen_shift_pattern): Likewise.
19906         (vect_recog_vector_vector_shift_pattern): Likewise.
19907         (vect_recog_divmod_pattern): Likewise.
19908         (vect_recog_mixed_size_cond_pattern): Likewise.
19909         (vect_recog_bool_pattern): Likewise.
19910         (vect_pattern_recog_1): Likewise.
19912 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19914         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
19915         non-functions to builtin_unreachable.
19916         * ipa-inline-transform.c (inline_call): Do not assert estimates were
19917         correct when new direct edges were discovered.
19919 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19921         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
19922         header, print symbol order instead of node uid, print more information
19923         about indirect edge targets.
19924         (ipa_make_edge_direct_to_target): Print symbol order instead of node
19925         uids.
19926         (ipa_make_edge_direct_to_target): Likewise.
19927         (remove_described_reference): Likewise.
19928         (propagate_controlled_uses): Likewise.
19929         (ipa_print_node_params): Also print symbol order.
19930         (ipcp_transform_function): Print symbol order instead of node uids.
19931         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
19932         (cgraph_get_create_real_symbol_node): Likewise.
19933         * ipa-cp.c (print_lattice): Likewise.
19934         (print_all_lattices): Likewise.
19935         (determine_versionability): Likewise.
19936         (initialize_node_lattices): Likewise.
19937         (estimate_local_effects): Likewise.
19938         (update_profiling_info): Likewise.
19939         (create_specialized_node): Likewise.
19940         (perhaps_add_new_callers): Likewise.
19941         (decide_about_value): Likewise.
19942         (decide_whether_version_node): Likewise.
19943         (identify_dead_nodes): Likewise.
19944         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
19945         (dump_inline_summary): Likewise.
19946         (estimate_node_size_and_time): Likewise.
19947         (inline_analyze_function): Likewise.
19948         * ipa-inline.c (report_inline_failed_reason): Likewise.
19949         (want_early_inline_function_p): Likewise.
19950         (edge_badness): Likewise.
19951         (update_edge_key): Likewise.
19952         (inline_small_functions): Likewise.  Add dumping of order to two other
19953         dumps.
19954         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
19955         instead of node uids.
19956         (propagate_pure_const): Likewise.
19957         (propagate_pure_const): Likewise.
19958         * ipa-utils.c (dump_cgraph_node_set): Likewise.
19959         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
19960         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
19961         of node uids.
19962         * tree-pretty-print.c (dump_function_header): Likewise.
19963         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
19964         Print symbol order instead of node uids.
19966 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
19968         * config/s390/s390.c (s390_register_move_cost): Don't impose the
19969         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
19971 2013-05-15  Richard Biener  <rguenther@suse.de>
19973         PR tree-optimization/57275
19974         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
19975         return value for fail to do runtime alias checks for gather loads.
19977 2013-05-15  Jan Hubicka  <jh@suse.cz>
19979         PR lto/57038
19980         PR lto/47375
19981         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
19982         weakrefs are not external.
19983         (lto_symtab_merge_decls): Fix thinko when dealing with
19984         non-lto_symtab decls.
19985         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
19986         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
19987         * varpool.c (dump_varpool_node): Dump more flags.
19989 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
19991         * config/i386/i386.c (processor_alias_table): Add instruction
19992         FSGSBASE for AMD bdver3 architecture.
19994 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
19996         * tree.c (warn_deprecated_use): Print file:line using locus color.
19997         * diagnostic.c (diagnostic_report_current_module): Print file:line
19998         and file:line:column using locus color.
20000 2013-05-14  Mike Stump  <mikestump@comcast.net>
20002         * gdbinit.in: Add __null.
20004 2013-05-14  Mike Stump  <mikestump@comcast.net>
20006         * recog.h: Rename struct recog_data to Recog_data.
20007         * recog.c: Likewise.
20008         * reload.c (can_reload_into): Likewise.
20009         * config/picochip/picochip.c: Likewise.
20011 2013-05-14  Mike Stump  <mikestump@comcast.net>
20013         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
20015 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
20017         * resource.h (struct resources): Remove unch_memory member.
20018         (CLEAR_RESOURCE): Don't clear unch_memory.
20019         * resource.c (mark_referenced_resources): Don't set it.
20020         (mark_set_resources): Likewise.
20021         (mark_target_live_regs): Don't clear it.
20022         (init_resource_info): Likewise.
20023         * reorg.c (resource_conflicts_p): Don't compare it.
20024         (redundant_insn): Don't set it.
20026         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
20027         Remove prototypes.
20028         * emit-rtl.c (next_label): Remove unused function.
20029         (skip_consecutive_labels, link_cc0_insns): Move to ...
20030         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
20031         only place where these functions are used, and make them static.
20033 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20035         * fold-const.c (fold_negate_expr): Handle vectors.
20036         (fold_truth_not_expr): Make it static.
20037         (fold_invert_truthvalue): New static function.
20038         (invert_truthvalue_loc): Handle vectors. Do not call
20039         fold_truth_not_expr directly.
20040         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
20041         <TRUTH_NOT_EXPR>: Do not cast to boolean.
20042         (fold_comparison): Handle vector constants.
20043         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
20044         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
20045         * tree.h (fold_truth_not_expr): Remove declaration.
20047 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
20049         * config/aarch64/aarch64-simd.md
20050         (aarch64_vcond_internal<mode>): Rename to...
20051         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
20052         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
20053         float modes. Clarify all iterator modes.
20054         (vcond<mode><mode>): Use new name for vcond expanders.
20055         (vcond<v_cmp_result><mode>): Likewise.
20056         (vcondu<mode><mode>: Likewise.
20057         * config/aarch64/iterators.md (VDQF_COND): New.
20059 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20061         PR bootstrap/57266
20062         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
20063         variable for the shift amount. Check that we shift by non-negative
20064         amounts.
20066 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
20068         PR target/42017
20069         * config/arm/arm.h (EPILOGUE_USES): Only return true
20070         for LR_REGNUM after epilogue_completed.
20072 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
20074         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
20075         is error_mark_node.
20077 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20079         PR target/57261
20080         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
20081         and Solaris 11+/x86 with gld.
20082         * configure: Regenerate.
20084 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20086         * expmed.c (expand_shift_1): Canonicalize rotates by
20087         constant bitsize / 2 to bitsize - 1.
20088         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
20089         case ROTATERT>: Likewise.
20091         Revert:
20092         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20094         * config/i386/i386.md (rotateinv): New code attr.
20095         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20096         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20097         roll $31, %eax, etc.
20099 2013-05-14  Richard Biener  <rguenther@suse.de>
20101         PR middle-end/57235
20102         * tree-eh.c (sink_clobbers): Give up for successors with
20103         multiple predecessors and no virtual uses.
20105 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
20107         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
20108         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
20110 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20112         PR middle-end/57251
20113         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
20114         the case when both op0 and op1 have VOIDmode.
20116 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
20118         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
20119         in multiply-accumulate mode.
20121 2013-05-13  Guozhi Wei  <carrot@google.com>
20123         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
20125 2013-05-13  Kai Tietz  <ktietz@redhat.com>
20127         PR target/56975
20128         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
20129         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
20130         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
20131         * config/i386/i386.c (ix86_option_override_internal): Likewise.
20132         (ix86_expand_prologue): Likewise.
20133         (ix86_expand_split_stack_prologue): Likewise.
20134         (legitimate_pic_address_disp_p): Likewise.
20135         (legitimize_pic_address): Likewise.
20136         (legitimize_tls_address): Likewise.
20137         (legitimize_pe_coff_symbol): Likewise.
20138         (output_pic_addr_const): Likewise.
20139         (construct_plt_address): Likewise.
20140         (ix86_expand_call): Likewise.
20141         (x86_output_mi_thunk): Likewise.
20142         (x86_function_profiler): Likewise.
20144 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
20146         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
20147         similar switch cases.
20148         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
20149         (aarch64_simd_mov_to_<mode>low): Delete.
20150         (aarch64_simd_mov_to_<mode>high): Delete.
20151         (move_lo_quad_<mode>): Add w<-r alternative.
20152         (aarch64_simd_move_hi_quad_<mode>): Likewise.
20153         (aarch64_simd_mov_from_*): Update type attribute.
20154         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
20155         statement.
20157 2013-05-13  Jan Hubicka  <jh@suse.cz>
20159         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
20160         * config/i386/i386.c (ix86_compute_frame_layout,
20161         ix86_expand_epilogue, emit_i387_cw_initialization,
20162         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
20163         ix86_local_alignment): Fix use of size/speed predicates.
20165 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20167         PR tree-optimization/45216
20168         PR tree-optimization/57157
20169         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
20170         the (-Y) & (B - 1) variant if OP is |.
20171         * expmed.c (expand_shift_1): For rotations by const0_rtx just
20172         return shifted.  Use (-op1) & (prec - 1) as other_amount
20173         instead of prec - op1.
20175 2013-05-13  Martin Jambor  <mjambor@suse.cz>
20177         PR middle-end/42371
20178         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
20179         (ipa_constant_data): New type.
20180         (ipa_jump_func): Use ipa_constant_data to hold information about
20181         constant jump functions.
20182         (ipa_get_jf_constant): Adjust to jump function type changes.
20183         (ipa_get_jf_constant_rdesc): New function.
20184         (ipa_param_descriptor): New field controlled_uses.
20185         (ipa_get_controlled_uses): New function.
20186         (ipa_set_controlled_uses): Likewise.
20187         * ipa-ref.h (ipa_find_reference): Declare.
20188         * ipa-prop.c (ipa_cst_ref_desc): New type.
20189         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
20190         changes.
20191         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
20192         New parameter cs.  Adjust all callers.
20193         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
20194         (remove_described_reference): New function.
20195         (jfunc_rdesc_usable): Likewise.
20196         (try_make_edge_direct_simple_call): Decrement controlled use count,
20197         attempt to remove reference if it hits zero.
20198         (combine_controlled_uses_counters): New function.
20199         (propagate_controlled_uses): Likewise.
20200         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
20201         (ipa_edge_duplication_hook): Duplicate reference descriptions.
20202         (ipa_print_node_params): Print described use counter.
20203         (ipa_write_jump_function): Adjust to jump function type changes.
20204         (ipa_read_jump_function): New parameter CS, pass it to
20205         ipa_set_jf_constant.  Adjust caller.
20206         (ipa_write_node_info): Stream controlled use count
20207         (ipa_read_node_info): Likewise.
20208         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
20209         asserting.
20210         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
20211         count.  Remove cloning-added reference if it reaches zero.
20212         * ipa-ref.c (ipa_find_reference): New function.
20214 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
20216         * config/i386/i386.c (processor_target_table): Modified default
20217         alignment values for AMD BD and BT architectures.
20219 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
20221         * tree-vect-generic.c (uniform_vector_p): Move ...
20222         * tree.c (uniform_vector_p): ... here.
20223         * tree.h (uniform_vector_p): Declare it.
20224         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
20225         into a scalar.
20227 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20229         PR tree-optimization/57230
20230         * tree-ssa-strlen.c (handle_char_store): Record length for
20231         array store from STRING_CST.
20233         PR tree-optimization/57230
20234         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
20235         check.
20237 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
20239         * config/epiphany/epiphany.c (epiphany_init): Check size of
20240         NUM_MODES_FOR_MODE_SWITCHING.
20241         (epiphany_expand_prologue):
20242         Remove CONFIG_REGNUM initial value handling code.
20243         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
20244         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
20245         (emit_set_fp_mode, epiphany_mode_after): Likewise.
20246         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
20247         Don't return 1 for FP_MODE_NONE.
20248         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
20249         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
20250         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
20251         * config/epiphany/epiphany.md (save_config): New pattern.
20253 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
20255         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
20257 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
20259         * config/i386/i386.md (memory): Handle sseishft1.
20260         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
20261         (*vec_extractv2di_1): Ditto.
20263 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
20265         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
20266         saved registers.
20268 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
20270         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
20271         Add mthumb/march=armv7-a multilib.
20272         Add mthumb/march=armv7-r multilib.
20273         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
20275 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
20277         * config/v850/t-rtems: Add more multilibs.
20279 2013-05-10  Richard Biener  <rguenther@suse.de>
20281         PR tree-optimization/57214
20282         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
20283         not propagate from SSA names that occur in abnormal PHI nodes.
20285 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
20287         * stor-layout.c (element_precision): New function.
20288         * machmode.h (element_precision): Declare it.
20289         * tree.c (build_minus_one_cst): New function.
20290         (element_precision): Likewise.
20291         * tree.h (build_minus_one_cst): Declare new function.
20292         (element_precision): Likewise.
20293         * fold-const.c (operand_equal_p): Use element_precision.
20294         (fold_binary_loc): Handle vector types.
20295         * convert.c (convert_to_integer): Use element_precision.
20296         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
20297         separately.
20299 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
20301         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
20302         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
20303         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
20304         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20305         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
20306         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
20307         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
20308         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20309         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
20310         (Uuw8): New constraints.
20311         (Usb4): Move into alphabetical order.
20312         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
20313         (sd8_operand, ub8_operand, uw8_operand): New predicates.
20314         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
20315         previously unnamed patterns.
20316         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
20317         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
20318         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
20319         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
20320         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
20321         of specific lengths.
20323 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20325         * config/i386/i386.md (rotateinv): New code attr.
20326         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20327         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20328         roll $31, %eax, etc.
20330         PR tree-optimization/45216
20331         PR tree-optimization/57157
20332         * tree-ssa-forwprop.c (simplify_rotate): New function.
20333         (ssa_forward_propagate_and_combine): Call it.
20335 2013-05-10  Richard Biener  <rguenther@suse.de>
20337         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
20338         disable peeling when we version for aliasing.
20339         (vector_alignment_reachable_p): Honor explicit user alignment.
20340         (vect_supportable_dr_alignment): Likewise.
20341         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
20342         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
20343         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
20344         then peeling to arrange for the cost-model check to come first.
20346 2013-05-10  Alan Modra  <amodra@gmail.com>
20348         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
20349         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
20350         * configure: Regenerate.
20352 2013-05-10  Alan Modra  <amodra@gmail.com>
20354         PR target/55033
20355         * varasm.c (default_elf_select_section): Move !DECL_P check..
20356         (get_named_section): ..to here before calling get_section_name.
20357         Adjust assertion.
20358         (default_section_type_flags): Add DECL_P check.
20359         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
20360         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
20362 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
20364         * config/epiphany/epiphany.c (epiphany_expand_prologue):
20365         When using gen_stack_adjust_str with a register offset, add a
20366         REG_FRAME_RELATED_EXPR note.
20368 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
20370         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
20371         (*vec_extractv4si_zext_mem): Ditto.
20372         (*vec_extractv2di): Add 0->x and x->x alternatives.
20373         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
20374         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
20376 2013-05-09  Jason Merrill  <jason@redhat.com>
20378         N3639 C++1y VLA support
20379         * gimplify.c (gimplify_vla_decl): Don't touch an existing
20380         DECL_VALUE_EXPR.
20382         * tree.c (build_constructor_va): New.
20383         * tree.h: Declare it.
20385 2013-05-09  Martin Jambor  <mjambor@suse.cz>
20387         PR lto/57084
20388         * gimple-fold.c (canonicalize_constructor_val): Call
20389         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
20391 2013-05-09  Jan Hubicka  <jh@suse.cz>
20392             Richard Biener  <rguenther@suse.de>
20394         PR lto/54095
20395         * symtab.c (symtab_make_decl_local): Do not add private names.
20397 2013-05-09  Jan Hubicka  <jh@suse.cz>
20399         PR lto/54095
20400         * symtab.c (insert_to_assembler_name_hash): Handle clones.
20401         (unlink_from_assembler_name_hash): Likewise.
20402         (symtab_prevail_in_asm_name_hash, symtab_register_node,
20403         symtab_unregister_node, symtab_initialize_asm_name_hash,
20404         change_decl_assembler_name): Update.
20406 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
20408         * config/aarch64/aarch64.md: New movtf split.
20409         (*movtf_aarch64): Update.
20410         (aarch64_movdi_tilow): Handle TF modes and rename to
20411         aarch64_movdi_<mode>low.
20412         (aarch64_movdi_tihigh): Handle TF modes and rename to
20413         aarch64_movdi_<mode>high
20414         (aarch64_movtihigh_di): Handle TF modes and rename to
20415         aarch64_mov<mode>high_di
20416         (aarch64_movtilow_di): Handle TF modes and rename to
20417         aarch64_mov<mode>low_di
20418         (aarch64_movtilow_tilow): Remove spurious whitespace.
20419         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
20420         splits.
20421         (aarch64_print_operand): Update.
20423 2013-05-09  Alan Modra  <amodra@gmail.com>
20425         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
20426         powerpc64le.
20427         * configure: Regenerate.
20429 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20431         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
20432         splitter preparation statements.
20433         * config/i386/sse.md (*vec_extract* splitters): Ditto.
20434         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
20435         adjust_address_nv.
20437 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20439         * gimple-ssa-strength-reduction.c (count_candidates): Change
20440         return value to int.
20441         (analyze_candidates_and_replace): Change type of length to int.
20443 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20445         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
20446         (*vec_extract<mode>): Use VI12_128 mode iterator.
20447         (*vec_extract<mode>_mem): Ditto.
20448         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
20449         attribute.
20451 2013-05-08  Diego Novillo  <dnovillo@google.com>
20453         PR bootstrap/54659
20455         Revert:
20456         2012-08-17  Diego Novillo  <dnovillo@google.com>
20458         PR bootstrap/54281
20459         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
20460         * config.in: Regenerate.
20461         * configure: Regenerate.
20462         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
20464 2013-05-08  Jan Hubicka  <jh@suse.cz>
20466         PR lto/54095
20467         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
20468         * cgraph.h (symtab_node_base): Add unique_name.
20469         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
20470         input_overwrite_node, input_varpool_node): Stream unique_name.
20471         * cgraphclones.c (cgraph_create_virtual_clone,
20472         cgraph_function_versioning): Set unique_name.
20473         * ipa.c (function_and_variable_visibility): Set unique_name.
20475 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20477         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
20478         (alloc_cand_and_find_basis): Restrict conditional candidate
20479         processing to CAND_MULTs.
20481 2013-05-08  Jan Hubicka  <jh@suse.cz>
20483         PR lto/54095
20484         lto-symtab.c (lto_symtab_symbol_p): New function.
20485         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
20486         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
20487         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
20488         Skip static symbols.
20490 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
20492         PR tree-optimization/57200
20493         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
20494         Only call inform if the preceding warning_at returns true.
20496 2013-05-07  Han Shen  <shenhan@google.com>
20498         * cfgexpand.c (record_or_union_type_has_array_p): New function.
20499         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
20500         * common.opt (fstack-protector-strong): New option.
20501         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
20502         * doc/invoke.texi (Optimization Options): Document
20503         "-fstack-protector-strong".
20504         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
20506 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20508         * config/mips/mips.c (mips_machine_reorg2): Return 0.
20510 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
20512         * ira.c (update_equiv_regs): Add insn having equiv memory even if
20513         it is not lhs of the insn.
20514         (setup_reg_equiv): Remove insn having equiv memory which it is not
20515         lhs of the insn.
20516         * lra-constraints.c (process_address): Try to improve generation
20517         code for address base + disp.
20518         (lra_constraints): Make correct the code for checking insn setting
20519         up backward equivalence.  Remove insn only if it is in the init
20520         insn list.
20521         * lra-eliminations.c (update_reg_eliminate): Change return value.
20522         (lra_eliminate): Use the result.
20524 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
20526         * config/i386/sse.md (ssescalarnummask): New mode attribute.
20527         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
20528         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
20529         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
20530         register target operands.
20531         (*vec_extractv8hi_sse2): New pattern.
20532         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
20533         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
20534         (*vec_extract<mode>_mem): New insn and split pattern.
20536 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
20538         * config/arm/arm.c (arm_asan_shadow_offset): New function.
20539         (TARGET_ASAN_SHADOW_OFFSET): Define.
20540         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
20541         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
20543 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20545         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
20546         (incr_vec_index): Return -1 if increment not found.
20547         (create_add_on_incoming_edge): Assert if increment not found.
20548         (record_increment): Limit number of increments recorded.
20549         (all_phi_incrs_profitable): Return false if an increment not found.
20550         (replace_profitable_candidates): Don't process increments that were
20551         not recorded.
20552         (analyze_candidates_and_replace): Limit size of incr_vec.
20554 2013-05-07  Richard Biener  <rguenther@suse.de>
20556         * calls.c (special_function_p): setjmp-like functions are leaf.
20557         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
20558         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
20560 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
20562         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
20563         (aarch64_simd_mov<mode>): New expander.
20564         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
20565         (aarch64_simd_mov_to_<mode>high): Likewise.
20566         (aarch64_simd_mov_from_<mode>low): Likewise.
20567         (aarch64_simd_mov_from_<mode>high): Likewise.
20568         (aarch64_dup_lane<mode>): Update.
20569         (aarch64_dup_lanedi): New instruction pattern.
20570         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
20571         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
20573 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20575         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
20576         (replace_mult_candidate): Remove unnecessary argument; remove
20577         unnecessary parameter from call to introduce_cast_before_cand.
20578         (replace_unconditional_candidate): Remove unnecessary parameter
20579         from call to replace_mult_candidate.
20580         (replace_conditional_candidate): Likewise.
20581         (insert_initializers): Use make_temp_ssa_name.
20582         (introduce_cast_before_cand): Remove unnecessary argument; use
20583         make_temp_ssa_name.
20584         (replace_one_candidate): Remove unnecessary argument; remove
20585         unnecessary parameter from calls to introduce_cast_before_cand.
20586         (replace_profitable_candidates): Remove unnecessary parameters
20587         from calls to replace_one_candidate.
20589 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20591         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
20592         phi def as possibly hiding a basis for a CAND_ADD whose operands
20593         have been commuted in the analysis.
20594         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
20596 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20598         * config/aarch64/aarch64.md
20599         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
20600         shift value between 0-4.
20602 2013-05-07  Richard Biener  <rguenther@suse.de>
20604         * double-int.h (rshift): New overload.
20605         * double-int.c (rshift): New function.
20606         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
20607         (create_reference_ops_from_ref): Remove.
20608         (vn_reference_insert): Use shared ops for constructing the
20609         reference and copy it.
20611 2013-05-07  Richard Biener  <rguenther@suse.de>
20613         PR middle-end/57190
20614         * tree-eh.c (sink_clobbers): Properly propagate
20615         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
20617 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
20619         PR tree-optimization/57149
20620         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
20621         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
20622         collect_phi_def_edges, execute_late_warn_uninitialized): Use
20623         uninit_undefined_value_p instead of ssa_undefined_value_p.
20625         PR debug/57184
20626         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
20627         for modifier == EXPAND_INITIALIZER.
20629 2013-05-07  Anton Blanchard  <anton@samba.org>
20631         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
20632         for powerpc64 little endian.
20633         * configure: Regenerate.
20635 2013-05-06  Graham Stott  <grahams@btinternet.com>
20637         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
20638         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
20639         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
20640         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
20642 2013-05-06  Graham Stott  <grahams@btinternet.com>
20644         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
20645         codes which allow non-lvalues.
20647 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
20649         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
20650         components are all 1s.
20651         (integer_minus_onep): New function.
20652         * tree.h (integer_minus_onep): Declare it.
20653         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
20654         integer_minus_onep instead of integer_all_onesp.
20656 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20658         PR target/52933
20659         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
20660         variations of these patterns.
20662 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20664         * config/i386/i386.md (isa): Add x64_sse4 member.
20665         (enabled): Handle x64_sse4.
20666         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
20667         instruction for 64bit SSE4_1 targets.  Update insn attributes.
20668         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
20669         instruction for SSE4_1 targets.  Update insn attributes.
20670         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20671         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
20672         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
20673         const_1 selector.
20674         (*vec_extractv4si): Rename from *sse4_1_pextrd.
20675         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
20676         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
20678 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20680         PR target/57108
20681         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
20683 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
20685         * final.c (do_assembler_dialects): Don't handle curly braces and
20686         vertical bar escaped by % as dialect delimiters.
20687         (output_asm_insn): Print curly braces and vertical bar if escaped
20688         by % and ASSEMBLER_DIALECT defined.
20689         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
20690         * doc/tm.texi: Regenerated.
20692 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20694         * config/mips/mips.c: Include tree-pass.h.
20695         (mips_reorg): Split in pre- and post-dbr_schedule parts.
20696         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
20697         (pass_mips_machine_reorg2): New machine specific pass.
20698         (insert_pass_mips_machine_reorg2): New pass plugin definition.
20699         (mips_option_override): Register the new pass.
20700         * rtl.h (cleanup_barriers): Remove prototype.
20701         (dbr_schedule): Likewise.
20702         * jump.c (cleanup_barriers): Make static.
20703         * reorg.c (dbr_schedule): Likewise.
20705 2013-05-06  Richard Biener  <rguenther@suse.de>
20707         PR tree-optimization/57185
20708         * tree-parloops.c (add_field_for_reduction): Handle anonymous
20709         SSA names properly.
20711 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20713         PR target/57106
20714         * config/i386/i386.c (add_parameter_dependencies): Add dependence
20715         between "first_arg" and "insn", not "last" and "insn".
20717 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20719         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
20720         (find_candidates_in_block): Re-enable slsr_process_phi.
20721         (create_phi_basis): Fix double counting of candidate adjustment.
20723 2013-05-06  Richard Biener  <rguenther@suse.de>
20725         PR middle-end/57147
20726         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
20727         the edge is also fallthru, preserve it and just clear the
20728         abnormal flag.
20729         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
20730         also complex, preserve that and just clear the fallthru flag.
20731         * tree-inline.c (update_ssa_across_abnormal_edges): Also
20732         update virtual operands.
20734 2013-05-06  Alan Modra  <amodra@gmail.com>
20736         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
20737         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
20738         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
20739         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
20740         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20741         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
20743 2013-05-06  Alan Modra  <amodra@gmail.com>
20745         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
20746         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
20747         (DEFAULT_ASM_ENDIAN): Define.
20748         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20749         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
20750         Update -K PIC clause from sysv4.h.
20751         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
20752         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
20754 2013-05-06  Alan Modra  <amodra@gmail.com>
20756         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
20757         twice for little-endian.
20758         (ashrdi3_no_power, ashrdi3): Support little-endian.
20760 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20762         PR target/55303
20763         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
20764         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
20765         related expanders.
20766         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
20767         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
20768         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
20769         New predicates.
20771 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
20772             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20774         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
20775         * config/pa/pa.opt: Make mbig-switch a no-op.
20776         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
20777         (CASE_VECTOR_MODE): Always return SImode.
20778         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
20779         for the !TARGET_BIG_SWITCH case.
20780         * config/pa/pa-linux.h: Likewise.
20781         * config/pa/pa-openbsd.h: Likewise.
20782         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
20783         * config/pa/pa.md (short_jump): Remove define_insn.
20784         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
20785         (casesi0): Remove define_insn.
20786         (type): Remove btable_branch.
20787         (pa_combine_type): Likewise.
20788         (in_nullified_branch_delay): Likewise.
20789         (in_call_delay): Likewise.
20790         (define_delay): Likewise.
20791         (define_insn_reservation "Z3"): Likewise.
20792         (define_insn_reservation "Z4"): Likewise.
20793         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
20794         (pa_adjust_insn_length): Remove adjustment for btable branches.
20795         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
20796         and mno-big-switch
20798 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
20800         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20801         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
20802         Add m->r,x alternatives.
20803         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
20804         splitters using SWI48x mode iterator.
20805         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
20806         TARGET_64BIT.  Add m->x alternative.
20807         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
20808         Add o->x alternative.  Enable for TARGET_SSE.
20809         (sse_storeq): Remove expander.
20810         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
20811         with memory input operand.
20812         (*vec_extractv2di_1 splitter): New.
20813         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
20814         * config/i386/i386.md (ssevecmodelower): New mode attribute.
20816 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20818         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
20819         (INT_LOWPART): Delete.
20820         (extract_MB): Adjust.
20821         (extract_ME): Adjust.
20822         (print_operand): Adjust.
20824 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20826         * config/rs6000/predicates.md (reg_or_add_cint_operand,
20827         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20828         (reg_or_logical_cint_operand, easy_fp_constant,
20829         logical_const_operand): Delete "CONST_DOUBLE" case.
20830         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
20831         "HOST_BITS_PER_WIDE_INT == 64" test.
20832         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
20833         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
20834         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
20835         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
20836         test.
20837         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
20838         CONST_DOUBLE DImode/VOIDmode case.
20839         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
20840         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20841         CONST_DOUBLE VOIDmode case.
20842         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20843         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
20844         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
20845         Delete CONST_DOUBLE case.
20846         (splitters for mov FMOVE64 const_double): Delete
20847         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20848         "HOST_BITS_PER_WIDE_INT >= 64" test.
20849         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
20850         case.
20851         (mov DI const_double): Delete.
20853 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
20855         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
20856         on op shows all bits zero in mode of a lowpart subreg, return zero.
20858 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
20860         PR target/57150
20861         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
20862         to save TFmode registers and DImode to save TImode registers for
20863         caller save operations.
20864         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
20865         mark being partially clobbered since they only use the first
20866         double word.
20868         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
20869         and TDmode only use the upper 64-bits of each VSX register.
20871 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20873         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
20874         (find_candidates_in_block): Disable slsr_process_phi.
20876 2013-05-03  Guozhi Wei  <carrot@google.com>
20878         * coverage.c (coverage_obj_init): Move the construction of gcov
20879         constructor to ...
20880         (build_init_ctor): ... here.
20882 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20884         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
20885         (slsr_cand_d): Redefine def_phi.
20886         (stride_status, phi_adjust_status, count_phis_status): New enums.
20887         (find_phi_def): New.
20888         (find_basis_for_base_expr): New.
20889         (find_basis_for_candidate): Handle hidden bases.
20890         (alloc_cand_and_find_basis): Handle phi candidates.
20891         (slsr_process_phi): New.
20892         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
20893         (create_mul_imm_cand): Likewise.
20894         (create_add_ssa_cand): Exclude phi base candidates.
20895         (create_add_imm_cand): Likewise.
20896         (slsr_process_cast): Likewise.
20897         (slsr_process_copy): Likewise.
20898         (find_candidates_in_block): Handle phi candidates.
20899         (dump_candidate): Likewise.
20900         (unconditional_cands): Delete.
20901         (unconditional_cands_with_known_stride_p): Delete.
20902         (phi_dependent_cand_p): New.
20903         (cand_increment): Handle phi-dependent candidates.
20904         (replace_dependent): Delete.
20905         (replace_mult_candidate): New.
20906         (replace_unconditional_candidate): New.
20907         (incr_vec_index): Move to avoid forward reference.
20908         (create_add_on_incoming_edge): New.
20909         (create_phi_basis): New.
20910         (replace_dependents): Delete.
20911         (replace_conditional_candidate): New.
20912         (phi_add_costs): New.
20913         (replace_uncond_cands_and_profitable_phis): New.
20914         (record_increment): Handle phi adjustments.
20915         (record_phi_increments): New.
20916         (record_increments): Handle phi adjustments.
20917         (phi_incr_cost): New.
20918         (lowest_cost_path): Handle phis.
20919         (total_savings): Likewise.
20920         (analyze_increments): Likewise.
20921         (ncd_with_phi): New.
20922         (ncd_of_cand_and_phis): New.
20923         (nearest_common_dominator_for_cands): Handle phi increments.
20924         (all_phi_incrs_profitable): New.
20925         (replace_profitable_candidates): Handle phi-dependent candidates.
20926         (analyze_candidates_and_replace): Likewise.
20928 2013-05-03  Teresa Johnson  <tejohnson@google.com>
20930         PR bootstrap/57154
20931         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
20932         do not exceed REG_BR_PROB_BASE.
20934 2013-05-03  Jeff Law  <law@redhat.com>
20936         PR tree-optimization/57144
20937         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
20938         operand of the condition will bit into the new type when eliminating
20939         a cast feeding a condition.
20941 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
20943         PR rtl-optimization/57130
20944         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
20945         of COMPARE as in_code to the recursive call if needed.
20947 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
20949         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
20950         (enabled): Handle new members.
20951         * config/i386/sse.md (*vec_concatv2si): Merge from
20952         *vec_concatv2si_sse2 and vec_concatv2si_sse.
20953         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
20955 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
20957         PR tree-optimization/57027
20958         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
20959         for fnms opportunity, check we got the prerequisite kind
20960         of tree / gimple before using accessor functions.
20962 2013-05-03  Richard Biener  <rguenther@suse.de>
20964         * double-int.h (lshift): New overload without precision
20965         and arith argument.
20966         (operator *=, operator +=, operator -=): Move ...
20967         * double-int.c (operator *=, operator +=, operator -=): ... here
20968         and implement more efficiently.
20969         (mul_double_with_sign): Remove.
20970         (lshift_double): Adjust to take unsinged shift argument, push
20971         dispatching code to callers.
20972         (mul_double_wide_with_sign): Add early out for callers that
20973         are not interested in high parts or overflow.
20974         (lshift): New function.
20975         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
20976         dispatch code here.
20977         (lrotate, rrotate): Use logical shifts.
20978         * expr.c (get_inner_reference): Use lshift.
20979         * fixed-value.c (do_fixed_divide): Likewise.
20980         * tree-dfa.c (get_ref_base_and_extent): Likewise.
20981         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
20982         (indirect_refs_may_alias_p): Likewise.
20983         (stmt_kills_ref_p_1): Likewise.
20985 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20987         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
20989 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
20991         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
20992         scalar form of FABD instruction.
20994 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
20996         * lra-constraints.c (process_alt_operands): Add checking alt
20997         number to choose the best alternative.
20999 2013-05-02  Richard Biener  <rguenther@suse.de>
21001         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
21002         bitmap and its handling.
21003         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
21005 2013-05-02  Richard Biener  <rguenther@suse.de>
21007         PR middle-end/57140
21008         * tree-inline.c (copy_loops): Properly handle removed loops.
21009         (copy_cfg_body): Mark destination loops for fixup if source
21010         loops needed fixup.
21012 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
21014         PR target/56732
21015         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
21016         generating simple_return for naked functions.
21018 2013-05-02  Martin Jambor  <mjambor@suse.cz>
21020         PR middle-end/56988
21021         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
21022         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
21023         flags match.
21024         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
21025         ipa_agg_replacement_value structures.
21026         (known_aggs_to_agg_replacement_list): Likewise.
21027         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
21028         (read_agg_replacement_chain): Likewise.
21029         (ipcp_transform_function): Also check that by_ref flags match.
21031 2013-05-02  Richard Biener  <rguenther@suse.de>
21033         * graphds.h (struct graph): Add obstack member.
21034         * graphds.c (new_graph): Initialize obstack and allocate
21035         vertices from it.
21036         (add_edge): Allocate edge from the obstack.
21037         (free_graph): Free the obstack instead of all edges and vertices.
21039 2013-05-02  Teresa Johnson  <tejohnson@google.com>
21041         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
21042         divides.
21043         * cfg.c (update_bb_profile_for_threading): Ditto.
21044         * tree-inline.c (copy_bb): Ditto.
21045         (copy_edges_for_bb): Ditto.
21046         (initialize_cfun): Ditto.
21047         (copy_cfg_body): Ditto.
21048         (expand_call_inline): Ditto.
21049         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
21050         (estimate_node_size_and_time): Ditto.
21051         (inline_merge_summary): Ditto.
21052         * cgraphclones.c (cgraph_clone_edge): Ditto.
21053         (cgraph_clone_node): Ditto.
21054         * sched-rgn.c (compute_dom_prob_ps): Ditto.
21055         (compute_trg_info): Ditto.
21057 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21059         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
21060         S reg when fp attribute set.
21061         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
21063 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21065         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
21066         New pattern.
21067         (*and_one_cmplsi3_compare0_uxtw): Likewise.
21068         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
21069         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
21071 2013-05-02  Richard Biener  <rguenther@suse.de>
21073         * tree-scalar-evolution.c (scev_info_hasher): Remove.
21074         (struct instantiate_cache_entry): New type.
21075         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
21076         (struct instantiate_cache_type): New type.
21077         (set_instantiated_value, get_instantiated_value): Remove.
21078         (get_instantiated_value_entry): New function.
21079         (instantiate_scev_name): Use the new cache and adjust.
21080         (instantiate_scev_poly): Adjust.
21081         (instantiate_scev_binary): Likewise.
21082         (instantiate_array_ref): Likewise.
21083         (instantiate_scev_convert): Likewise.
21084         (instantiate_scev_not): Likewise.
21085         (instantiate_scev_3): Likewise.
21086         (instantiate_scev_2): Likewise.
21087         (instantiate_scev_r): Likewise.
21088         (instantiate_scev): Likewise.
21089         (resolve_mixers): Likewise.
21091 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
21093         PR target/57091
21094         * lra-constraints.c (best_small_class_operands_num): Remove.
21095         (process_alt_operands): Remove small_class_operands_num.  Take
21096         small classes operands into losers and only if the operand is not
21097         matched.  Modify debugging output.
21098         (curr_insn_transform): Remove best_small_class_operands_num.
21099         Print insn name.
21101 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21103         * config/aarch64/aarch64-builtins.c
21104         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
21105         * config/aarch64/aarch64-simd-builtins.def
21106         (reduc_splus_): Add new modes.
21107         (reduc_uplus_): New.
21108         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
21109         (reduc_uplus_v4sf): Likewise.
21110         (reduc_splus_v4sf): Likewise.
21111         (aarch64_addv<mode>): Likewise.
21112         (reduc_uplus_<mode>): Likewise.
21113         (reduc_splus_<mode>): Likewise.
21114         (aarch64_addvv2di): Likewise.
21115         (reduc_uplus_v2di): Likewise.
21116         (reduc_splus_v2di): Likewise.
21117         (aarch64_addvv2si): Likewise.
21118         (reduc_uplus_v2si): Likewise.
21119         (reduc_splus_v2si): Likewise.
21120         (reduc_<sur>plus_<mode>): New.
21121         (reduc_<sur>plus_v2di): Likewise.
21122         (reduc_<sur>plus_v2si): Likewise.
21123         (reduc_<sur>plus_v4sf): Likewise.
21124         (aarch64_addpv4sf): Likewise.
21125         * config/aarch64/arm_neon.h
21126         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
21127         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
21128         add UNSPEC_SADDV, UNSPEC_UADDV.
21129         (SUADDV): New.
21130         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
21132 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21134         * config/aarch64/arm_neon.h
21135         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
21137 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21139         * config/aarch64/aarch64-builtins
21140         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
21142 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21144         * config/aarch64/aarch64-simd-builtins.def
21145         (reduc_smax_): New.
21146         (reduc_smin_): Likewise.
21147         (reduc_umax_): Likewise.
21148         (reduc_umin_): Likewise.
21149         (reduc_smax_nan_): Likewise.
21150         (reduc_smin_nan_): Likewise.
21151         (fmax): Remove.
21152         (fmin): Likewise.
21153         (smax): Update for V2SF, V4SF and V2DF modes.
21154         (smin): Likewise.
21155         (smax_nan): New.
21156         (smin_nan): Likewise.
21157         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
21158         (<su><maxmin><mode>3): ...This, refactor.
21159         (s<maxmin><mode>3): New.
21160         (<maxmin_uns><mode>3): Likewise.
21161         (reduc_<maxmin_uns>_<mode>): Refactor.
21162         (reduc_<maxmin_uns>_v4sf): Likewise.
21163         (reduc_<maxmin_uns>_v2si): Likewise.
21164         (aarch64_<fmaxmin><mode>: Remove.
21165         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
21166         new builtin names.
21167         (vmin<q>_f<32,64>): Likewise.
21168         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21169         (FMAXMIN): New.
21170         (su): Add mappings for smax, smin, umax, umin.
21171         (maxmin): New.
21172         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21173         (FMAXMIN): Rename as...
21174         (FMAXMIN_UNS): ...This.
21175         (maxminv): Remove.
21176         (fmaxminv): Likewise.
21177         (fmaxmin): Likewise.
21178         (maxmin_uns): New.
21179         (maxmin_uns_op): Likewise.
21181 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21183         * config/aarch64/arm_neon.h
21184         (vac<ge, gt><sd>_f<32, 64>): Rename to...
21185         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
21186         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
21188 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21190         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
21191         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
21193 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21195         * config/aarch64/aarch64-simd.md
21196         (vcond<mode>_internal): Handle special cases for constant masks.
21197         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
21198         (vcondu<mode><mode>): Likewise.
21199         (vcond<v_cmp_result><mode>): New.
21201 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21203         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
21204         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
21205         * config/aarch64/aarch64-simd-builtins.def
21206         (cmeq): Update to BUILTIN_VALLDI.
21207         (cmgt): Likewise.
21208         (cmge): Likewise.
21209         (cmle): Likewise.
21210         (cmlt): Likewise.
21211         * config/aarch64/arm_neon.h
21212         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
21213         to builtins or C as appropriate.
21215 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21217         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
21218         (cmgeu): ...This.
21219         (cmhi): Rename to...
21220         (cmgtu): ...This.
21221         * config/aarch64/aarch64-simd.md
21222         (simd_mode): Add SF.
21223         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
21224         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
21225         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
21226         (cstore<mode>_neg): ...This.
21227         * config/aarch64/iterators.md
21228         (VALLF): new.
21229         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
21230         (COMPARISONS): New.
21231         (UCOMPARISONS): Likewise.
21232         (optab): Add missing comparisons.
21233         (n_optab): New.
21234         (cmp_1): Likewise.
21235         (cmp_2): Likewise.
21236         (CMP): Likewise.
21237         (cmp): Remove.
21238         (VCMP_S): Likewise.
21239         (VCMP_U): Likewise.
21240         (V_cmp_result): Add DF, SF modes.
21241         (v_cmp_result): Likewise.
21242         (v): Likewise.
21243         (vmtype): Likewise.
21244         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
21246 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
21248         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
21249         define_insn to define_insn_and_split.
21250         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
21251         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
21252         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
21253         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
21254         (thumb2_negscc): Likewise.
21256 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21258         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
21260 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21262         * config/arm/thumb2.md: Remove trailing whitespaces.
21264 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
21266         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
21267         Use gen_int_mode rather than GEN_INT.
21269 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
21271         * value-prof.c (stream_in_histogram_value): Remove the strayed
21272         debug_gimple_stmt.
21274 2013-04-30  Richard Biener  <rguenther@suse.de>
21276         PR middle-end/57122
21277         * cfghooks.c (split_edge): Properly check for the loop latch edge.
21279 2013-04-30  Richard Biener  <rguenther@suse.de>
21281         PR middle-end/57107
21282         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
21284 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
21286         PR rtl-optimization/56957
21287         PR rtl-optimization/57105
21288         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
21289         variable.  Use just INSN_UID for determining whether an insn
21290         should be only disconnected from the insn stream.
21291         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
21293 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
21295         PR tree-optimization/57104
21296         * tsan.c (instrument_expr): Don't instrument accesses to
21297         DECL_HARD_REGISTER VAR_DECLs.
21299 2013-04-30  Richard Biener  <rguenther@suse.de>
21301         * function.h (loops_for_fn): New inline function.
21302         (set_loops_for_fn): Likewise.
21303         * cfgloop.h (place_new_loop): Add struct function parameter.
21304         (get_loop): Likewise.
21305         (get_loops): Likewise.
21306         (number_of_loops): Likewise.
21307         (fel_next): Adjust.
21308         (fel_init): Likewise.
21309         * cfg.c (get_loop_copy): Adjust.
21310         * cfgloop.c (flow_loops_dump): Likewise.
21311         (record_loop_exits): Likewise.
21312         (verify_loop_structure): Likewise.
21313         * cfgloopanal.c (mark_irreducible_loops): Likewise.
21314         (estimate_reg_pressure_cost): Likewise.
21315         (mark_loop_exit_edges): Likewise.
21316         * cfgloopmanip.c (place_new_loop): Likewise.
21317         (add_loop): Likewise.
21318         (duplicate_loop): Likewise.
21319         * graph.c (draw_cfg_nodes): Likewise.
21320         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
21321         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
21322         (extract_affine_chrec): Likewise.
21323         (build_scop_iteration_domain): Likewise.
21324         * graphite.c (graphite_initialize): Likewise.
21325         * ira-build.c (create_loop_tree_nodes): Likewise.
21326         (more_one_region_p): Likewise.
21327         (rebuild_regno_allocno_maps): Likewise.
21328         (mark_loops_for_removal): Likewise.
21329         (mark_all_loops_for_removal): Likewise.
21330         (remove_unnecessary_regions): Likewise.
21331         (ira_build): Likewise.
21332         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21333         * loop-init.c (fix_loop_structure): Likewise.
21334         (gate_rtl_move_loop_invariants): Likewise.
21335         (gate_rtl_unswitch): Likewise.
21336         (gate_rtl_unroll_and_peel_loops): Likewise.
21337         (rtl_doloop): Likewise.
21338         * lto-streamer-in.c (input_cfg): Likewise.
21339         * lto-streamer-out.c (output_cfg): Likewise.
21340         * modulo-sched.c (sms_schedule): Likewise.
21341         * predict.c (tree_estimate_probability): Likewise.
21342         (tree_estimate_probability_driver): Likewise.
21343         (estimate_loops): Likewise.
21344         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
21345         (move_sese_region_to_fn): Likewise.
21346         (debug_loop_num): Likewise.
21347         * tree-chrec.c (chrec_evaluate): Likewise.
21348         (hide_evolution_in_other_loops_than_loop): Likewise.
21349         (chrec_component_in_loop_num): Likewise.
21350         (reset_evolution_in_loop): Likewise.
21351         (evolution_function_is_invariant_rec_p): Likewise.
21352         * tree-if-conv.c (main_tree_if_conversion): Likewise.
21353         * tree-inline.c (copy_loops): Likewise.
21354         (copy_cfg_body): Likewise.
21355         (tree_function_versioning): Likewise.
21356         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
21357         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
21358         Likewise.
21359         (add_to_evolution_1): Likewise.
21360         (scev_const_prop): Likewise.
21361         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
21362         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
21363         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
21364         (tree_ssa_lim_initialize): Likewise.
21365         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
21366         (verify_loop_closed_ssa): Likewise.
21367         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
21368         (tree_ssa_loop_im): Likewise.
21369         (tree_ssa_loop_unswitch): Likewise.
21370         (tree_vectorize): Likewise.
21371         (check_data_deps): Likewise.
21372         (tree_ssa_loop_ivcanon): Likewise.
21373         (tree_ssa_loop_bounds): Likewise.
21374         (tree_complete_unroll): Likewise.
21375         (tree_complete_unroll_inner): Likewise.
21376         (tree_parallelize_loops): Likewise.
21377         (tree_ssa_loop_prefetch): Likewise.
21378         (tree_ssa_loop_ivopts): Likewise.
21379         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
21380         * tree-vectorizer.c (vectorize_loops): Likewise.
21382 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
21384         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
21385         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
21386         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
21387         with EABI_LINK_SPEC.
21389 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21391         PR target/44578
21392         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
21393         alternative.
21395 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
21397         PR target/57097
21398         * lra-constraints.c (process_alt_operands): Discourage a bit more
21399         using memory for pseudos.  Print cost dump for alternatives.
21400         Modify cost values for conflicts with early clobbers.
21401         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
21403 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21405         PR target/57098
21406         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
21408 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21410         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
21411         from/to S register.
21412         (movdi_aarch64): Support LDR/STR from/to D register.
21414 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21416         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
21417         or higher by default.
21419 2013-04-29  Richard Biener  <rguenther@suse.de>
21421         PR middle-end/57075
21422         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
21423         even if not adding abnormal edges for calls that can make
21424         abnormal gotos.
21426 2013-04-29  Richard Biener  <rguenther@suse.de>
21428         PR middle-end/57103
21429         * tree-cfg.c (move_stmt_op): Fix condition under which to update
21430         TREE_BLOCK.
21431         (move_stmt_r): Remove redundant checking.
21433 2013-04-29  Teresa Johnson  <tejohnson@google.com>
21435         PR bootstrap/57077
21436         * basic-block.h (apply_scale): New function.
21437         (apply_probability): Use apply_scale.
21438         * gimple-streamer-in.c (input_bb): Ditto.
21439         * lto-streamer-in.c (input_cfg): Ditto.
21440         * lto-cgraph.c (merge_profile_summaries): Ditto.
21441         * tree-optimize.c (execute_fixup_cfg): Ditto.
21442         * tree-inline.c (copy_bb): Update comment to use apply_scale.
21443         (copy_edges_for_bb): Ditto.
21444         (copy_cfg_body): Ditto.
21446 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21448         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
21449         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
21450         (tail_merge_optimize): Handle current_loops == NULL.
21452 2013-04-26  Jeff Law  <law@redhat.com>
21454         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
21455         (simplify_cond_using_ranges): Generalize code to simplify
21456         COND_EXPRs where one argument is a constant and the other
21457         is an SSA_NAME created by an integral type conversion.
21459 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21461         * config/arm/arm.md (store_minmaxsi): Use only when
21462         optimize_insn_for_size_p.
21464 2013-04-29  Christian Bruel  <christian.bruel@st.com>
21466         PR target/57108
21467         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
21469 2013-04-29  Richard Biener  <rguenther@suse.de>
21471         PR middle-end/57089
21472         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
21473         loop tree make sure to schedule a fixup for the child as well.
21474         (expand_omp_for_generic): Properly add loops.
21475         (expand_omp_for_static_nochunk): Likewise.
21476         (expand_omp_for_static_chunk): Likewise.
21477         (expand_omp_for): For the degenerate case fixup loops.
21478         (expand_omp_sections): Fix default bb placement in loops.
21479         (expand_omp_atomic_pipeline): Properly add loops.
21481 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21483         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
21485 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21487         * tree-ssa-tail-merge.c: Update header comment.
21489 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21491         * config/aarch64/arm_neon.h
21492         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
21493         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
21494         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
21495         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
21496         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
21497         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
21498         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
21499         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
21501 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21503         * config/aarch64/aarch64-simd.md
21504         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
21505         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
21506         fix_trunc, fixuns_trunc.
21507         (ftrunc<VDQF:mode>2): New.
21508         * config/aarch64/iterators.md (optab): Add fix, fixuns.
21509         (fix_trunc_optab): New.
21511 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21513         * config/aarch64/aarch64-builtins.c
21514         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
21515         iceilf, lround, iroundf.
21517 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21519         PR target/54349
21520         * config/i386/i386.h (enum ix86_tune_indices)
21521         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
21522         New, split from X86_TUNE_INTER_UNIT_MOVES.
21523         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
21524         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
21525         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
21526         (TARGET_INTER_UNIT_MOVES): Remove.
21527         * config/i386/i386.c (initial_ix86_tune_features): Update.
21528         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
21529         (ix86_expand_convert_uns_didf_sse): Use
21530         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21531         (ix86_expand_vector_init_one_nonzero): Ditto.
21532         (ix86_expand_vector_init_interleave): Ditto.
21533         (inline_secondary_memory_needed): Return true for moves from SSE class
21534         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
21535         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21536         * config/i386/constraints.md (Yi, Ym): Depend on
21537         TARGET_INTER_UNIT_MOVES_TO_VEC.
21538         (Yj, Yn): New constraints.
21539         * config/i386/i386.md (*movdi_internal): Change constraints of
21540         operand 1 from Yi to Yj and from Ym to Yn.
21541         (*movsi_internal): Ditto.
21542         (*movdf_internal): Ditto.
21543         (*movsf_internal): Ditto.
21544         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
21545         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21546         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
21547         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
21548         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
21549         * config/i386/sse.md (movdi_to_sse): Ditto.
21550         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
21551         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
21552         TARGET_INTER_UNIT_MOVES.
21553         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
21554         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
21555         instead of TARGET_INTER_UNIT_MOVES.
21556         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
21557         operand 1 from Yi to Yj and from Ym to Yn.
21559 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21561         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
21562         (float_truncate_hi_): Likewise.
21563         (float_extend_lo_): Likewise.
21564         (float_truncate_lo_): Likewise.
21565         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
21566         (aarch64_float_extend_lo_v2df): Likewise.
21567         (vec_unpacks_hi_v4sf): Likewise.
21568         (aarch64_float_truncate_lo_v2sf): Likewise.
21569         (aarch64_float_truncate_hi_v4sf): Likewise.
21570         (vec_pack_trunc_v2df): Likewise.
21571         (vec_pack_trunc_df): Likewise.
21573 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21575         * config/aarch64/aarch64-builtins.c
21576         (aarch64_fold_builtin): Fold float conversions.
21577         * config/aarch64/aarch64-simd-builtins.def
21578         (floatv2si, floatv4si, floatv2di): New.
21579         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
21580         * config/aarch64/aarch64-simd.md
21581         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
21582         * config/aarch64/iterators.md (FLOATUORS): New.
21583         (optab): Add float, floatuns.
21584         (su_optab): Likewise.
21586 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21588         * config/aarch64/aarch64-builtins.c
21589         (aarch64_builtin_vectorized_function): Use new names for
21590         fcvt builtins.
21591         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
21592         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
21593         (fcvtzu): Split as...
21594         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
21595         (fcvtas): Split as...
21596         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
21597         (fcvtau): Split as...
21598         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
21599         (fcvtps): Split as...
21600         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
21601         (fcvtpu): Split as...
21602         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
21603         (fcvtms): Split as...
21604         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
21605         (fcvtmu): Split as...
21606         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
21607         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
21608         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
21609         (lfrintnusf, lfrintnudf): Likewise.
21610         * config/aarch64/aarch64-simd.md
21611         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
21612         define_insn.
21613         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
21614         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
21615         (fcvt_pattern): Likewise.
21617 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21619         * config/aarch64/aarch64-simd.md
21620         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
21621         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
21623 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21625         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
21626         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
21627         (vrnd<a,m,n,p>_f32): Implement using builtins.
21628         (vrnd<i,x><q>_f<32, 64>): New.
21630 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21632         * config/aarch64/aarch64-builtins.c
21633         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
21634         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
21635         (frintz): Rename to...
21636         (btrunc): ...this.
21637         (frintp): Rename to...
21638         (ceil): ...this.
21639         (frintm): Rename to...
21640         (floor): ...this.
21641         (frinti): Rename to...
21642         (nearbyint): ...this.
21643         (frintx): Rename to...
21644         (rint): ...this.
21645         (frinta): Rename to...
21646         (round): ...this.
21647         * config/aarch64/aarch64-simd.md
21648         (aarch64_frint<frint_suffix><mode>): Delete.
21649         (<frint_pattern><mode>2): Convert to insn.
21650         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
21651         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
21652         (frint_pattern): Likewise.
21653         (frint_suffix): Likewise.
21655 2013-04-29  Richard Biener  <rguenther@suse.de>
21657         PR tree-optimization/57081
21658         * loop-init.c: Include tree-flow.h.
21659         (loop_optimizer_finalize): Free number of iteration estimates.
21660         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
21662 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
21664         PR tree-optimization/57083
21665         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
21666         non-singleton shift count range, zero extend low_bound for uns case.
21668         * config/i386/predicates.md (general_vector_operand): New predicate.
21669         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
21670         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
21671         if they aren't nonimmediate operands.  If their original values
21672         satisfy const_vector_equal_evenodd_p, don't shift them.
21673         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
21674         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
21675         if not nonimmediate_operand.
21676         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
21677         instead of register_operand.
21678         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
21680 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
21682         * stor-layout.c (finalize_size_functions): Allocate a structure and
21683         reset cfun before dumping the functions.
21685 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
21687         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
21689         PR target/56866
21690         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
21691         use xop_pmacsdqh if uns_p.
21692         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
21693         the immediate rotate count.
21695 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
21697         * rtl.h (struct rtx_def): Add comment for field jump.
21698         (LRA_SUBREG_P): New macro.
21699         * recog.c (register_operand): Check LRA_SUBREG_P.
21700         * lra.c (lra): Add note at the end of RTL code. Align non-empty
21701         stack frame.
21702         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
21703         (lra_final_code_change): Skip subreg change for operators.
21704         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
21705         if there are no operand changes.
21706         * lra-constraints.c (curr_insn_set): New.
21707         (match_reload): Set LRA_SUBREG_P.
21708         (emit_spill_move): Ditto.
21709         (check_and_process_move): Use curr_insn_set. Process only single
21710         set insns.  Don't initialize sec_mem_p and change_p.
21711         (simplify_operand_subreg): Use LRA_SUBREG_P.
21712         (reg_in_class_p): New function.
21713         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
21714         of #ifdef.  Add code to remove cycling.
21715         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
21716         non-null disp.  Reload inner instead of disp when base and index
21717         are null.  Try to put lo_sum into register.
21718         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
21719         (check_and_process_move): Move code for move cost check to
21720         simple_move_p.  Remove equiv_substitution.
21721         (simple_move_p): New function.
21722         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
21723         curr_insn_set.  Call check_and_process_move only for single set
21724         insns.  Use the new function.  Move call of check_and_process_move
21725         after operand equiv substitution and address process.
21727 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21729         PR go/57045
21730         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
21731         with nonlocal goto receivers or returns twice calls, ignore
21732         unininitialized values from abnormal edges to nl goto receiver
21733         or returns twice call.
21735 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21737         PR tree-optimization/57051
21738         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
21739         and VEC_RSHIFT_EXPR if shift count is a multiple of element
21740         bitsize.
21742 2013-04-26  Richard Biener  <rguenther@suse.de>
21744         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
21745         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
21746         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
21747         (fixup_loop_arrays_after_move): New function.
21748         (move_sese_region_to_fn): Properly outline the loop tree parts
21749         of the SESE region.
21751 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
21753         * config/i386/i386.md (type, unit): Fix long lines.
21755 2013-04-26  Richard Biener  <rguenther@suse.de>
21757         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
21758         (lto-streamer-out.o): Likewise.
21759         * cfgloop.c (init_loops_structure): Export, add struct function
21760         argument and adjust.
21761         (flow_loops_find): Adjust.
21762         * cfgloop.h (enum loop_estimation): Add EST_LAST.
21763         (init_loops_structure): Declare.
21764         * lto-streamer-in.c: Include cfgloop.h.
21765         (input_cfg): Input the loop tree.
21766         * lto-streamer-out.c: Include cfgloop.h.
21767         (output_cfg): Output the loop tree.
21768         (output_struct_function_base): Do not drop PROP_loops.
21770 2013-03-26  Richard Biener  <rguenther@suse.de>
21772         * tree-cfg.c (execute_build_cfg): Build the loop tree.
21773         (pass_build_cfg): Provide PROP_loops.
21774         (move_sese_region_to_fn): Remove loops that are outlined into fn
21775         for now.
21776         * tree-inline.c: Include cfgloop.h.
21777         (initialize_cfun): Do not drop PROP_loops.
21778         (copy_loops): New function.
21779         (copy_cfg_body): Copy loop structure.
21780         (tree_function_versioning): Initialize destination loop tree.
21781         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
21782         (pass_parallelize_loops): Do IL verification.
21783         * loop-init.c (loop_optimizer_init): Fixup loops if required.
21784         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
21785         the CFG make sure we fixup loops as well.
21786         * tree-ssa-tail-merge.c: Include cfgloop.h.
21787         (replace_block_by): When merging loop latches mark loops for fixup.
21788         * lto-streamer-out.c (output_struct_function_base): Drop
21789         PROP_loops for now.
21790         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
21791         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
21792         * ipa-split.c: Include cfgloop.h.
21793         (split_function): Add the new return block to the loop tree root.
21794         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
21795         whether we have removed the forwarder block.
21796         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
21797         * cfgloop.h (place_new_loop): Declare.
21798         * cfgloopmanip.c (place_new_loop): Export.
21799         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
21800         (tree-switch-conversion.o): Likewise.
21801         (tree-complex.o): Likewise.
21802         (tree-inline.o): Likewise.
21803         (tree-ssa-tailmerge.o): Likewise.
21804         (ipa-split.o): Likewise.
21805         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
21806         (tree-ssa-copy.o): Likewise.
21807         * tree-switch-conversion.c: Include cfgloop.h
21808         (process_switch): If we emit a bit-test cascade, schedule loops
21809         for fixup.
21810         * tree-complex.c: Include cfgloop.h.
21811         (expand_complex_div_wide): Properly add new basic-blocks to loops.
21812         * asan.c: Include cfgloop.h.
21813         (create_cond_insert_point): Properly add new basic-blocks to
21814         loops, schedule loop fixup.
21815         * cfgloop.c (verify_loop_structure): Check that looks are not
21816         marked for fixup.
21817         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
21818         to loops.
21819         (expand_omp_for_generic): Likewise.
21820         (expand_omp_sections): Likewise.
21821         (expand_omp_atomic_pipeline): Schedule loops for fixup.
21822         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
21823         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
21824         is initialized, not when loops are present.
21825         * tree-parloops.c (parallelize_loops): Remove checking here.
21826         * passes.c (init_optimization_passes): Schedule a copy-propagation
21827         pass before complete unrolling of inner loops.
21829 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21831         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
21832         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
21833         (colorize_init): Add argument to _WIN32 version.
21834         * toplev.c: Include diagnostic-color.h.
21835         (process_options): Default to -fdiagnostics-color=auto if
21836         GCC_COLORS env var is in the environment.
21837         * common.opt (fdiagnostics-color=): Add Var and Init.
21838         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
21839         env var is in the environment, the default is auto rather than never.
21841         * diagnostic.h (file_name_as_prefix): Add context argument.
21842         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
21843         the string as locus.
21844         * langhooks.c (lhd_print_error_function): Adjust caller.
21846 2013-04-25  Lawrence Crowl  <crowl@google.com>
21848         * var-tracking.c (shared_hash_def::htab):
21849         Change type to hash_table.  Update dependent calls and types.
21851 2013-04-25  Lawrence Crowl  <crowl@google.com>
21853         * Makefile.in: Update as needed below.
21855         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
21856         Move declaration to after the type's method definitons.
21858         * attribs.c (htab_t scoped_attributes::attribute_hash):
21859         Change type to hash_table.  Update dependent calls and types.
21861         * bitmap.c (htab_t bitmap_desc_hash):
21862         Change type to hash_table.  Update dependent calls and types.
21864         * cselib.c (htab_t cselib_hash_table):
21865         Change type to hash_table.  Update dependent calls and types.
21867         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
21868         (hash_string_slot_node): Move implementation into lto-streamer.h
21869         struct string_slot_hasher.
21870         (eq_string_slot_node): Likewise.
21872         * data-streamer-out.c: Update output_block::string_hash_table
21873         dependent calls and types.
21875         * dwarf2cfi.c (htab_t trace_index):
21876         Change type to hash_table.  Update dependent calls and types.
21878         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
21879         Change type to hash_table.  Update dependent calls and types.
21880         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
21881         (htab_t optimize_external_refs::map): Likewise.
21882         (htab_t output_comp_unit::extern_map): Likewise.
21883         (htab_t output_comdat_type_unit::extern_map): Likewise.
21884         (htab_t output_macinfo::macinfo_htab): Likewise.
21885         (htab_t optimize_location_lists::htab): Likewise.
21886         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
21888         * except.c (htab_t ehspec_hash_type):
21889         Change type to hash_table.  Update dependent calls and types.
21890         (assign_filter_values::ttypes): Likewise.
21891         (assign_filter_values::ehspec): Likewise.
21892         (sjlj_assign_call_site_values::ar_hash): Likewise.
21893         (convert_to_eh_region_ranges::ar_hash): Likewise.
21895         * gcse.c (htab_t pre_ldst_table):
21896         Change type to hash_table.  Update dependent calls and types.
21898         * ggc-common.c (htab_t saving_htab):
21899         Change type to hash_table.  Update dependent calls and types.
21900         (htab_t loc_hash): Likewise.
21901         (htab_t ptr_hash): Likewise.
21902         (call_count): Rename ggc_call_count.
21903         (call_alloc): Rename ggc_call_alloc.
21904         (loc_descriptor): Rename make_loc_descriptor.
21905         (add_statistics): Rename ggc_add_statistics.
21907         * ggc-common.c (saving_htab):
21908         Change type to hash_table.  Update dependent calls and types.
21910         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
21911         (push_gimplify_context): Likewise.
21912         (pop_gimplify_context): Likewise.
21913         (struct gimple_temp_hash_elt): Added.
21914         (struct gimplify_hasher): Likewise.
21915         (struct gimplify_ctx.temp_htab):
21916         Change type to hash_table.  Update dependent calls and types.
21918         * gimple-fold.c: Include gimplify-ctx.h.
21920         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
21921         Change type to hash_table.  Update dependent calls and types.
21922         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
21923         avoid potential global name collision.
21925         * gimplify.c: Include gimplify-ctx.h.
21926         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
21927         (htab_t gimplify_ctx::temp_htab):
21928         Update dependent calls and types for new type hash_table.
21929         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
21930         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
21932         * gimplify-ctx.h: New.
21933         (struct gimple_temp_hash_elt): Move from gimplify.c.
21934         (class gimplify_hasher): New.
21935         (struct gimplify_ctx): Move from gimple.h.
21936         (htab_t gimplify_ctx::temp_htab):
21937         Change type to hash_table.  Update dependent calls and types.
21939         * graphite-clast-to-gimple.c: Include graphite-htab.h.
21940         (htab_t ivs_params::newivs_index):
21941         Change type to hash_table.  Update dependent calls and types.
21942         (htab_t ivs_params::params_index): Likewise.
21943         (htab_t print_generated_program::params_index): Likewise.
21944         (htab_t gloog::newivs_index): Likewise.
21945         (htab_t gloog::params_index): Likewise.
21947         * graphite.c: Include graphite-htab.h.
21948         4htab_t graphite_transform_loops::bb_pbb_mapping):
21949         Change type to hash_table.  Update dependent calls and types.
21951         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
21952         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
21953         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
21955         * graphite-dependences.c: Include graphite-htab.h.
21956         (loop_is_parallel_p): Change hash table type of parameter.
21958         * graphite-htab.h: New.
21959         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
21960         (extern find_pbb_via_hash): Move from graphite-poly.h.
21961         (extern loop_is_parallel_p): Move from graphite-poly.h.
21962         (extern get_loop_body_pbbs): Move from graphite-poly.h.
21964         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
21965         (extern loop_is_parallel_p): Move to graphite-htab.h.
21966         (extern get_loop_body_pbbs): Move to graphite-htab.h.
21968         * haifa-sched.c (htab_t delay_htab):
21969         Change type to hash_table.  Update dependent calls and types.
21970         (htab_t delay_htab_i2): Likewise.
21972         * ira-color.c (htab_t allocno_hard_regs_htab):
21973         Change type to hash_table.  Update dependent calls and types.
21975         * ira-costs.c (htab_t cost_classes_htab):
21976         Change type to hash_table.  Update dependent calls and types.
21978         * loop-invariant.c (htab_t merge_identical_invariants::eq):
21979         Change type to hash_table.  Update dependent calls and types.
21981         * loop-iv.c (htab_t bivs):
21982         Change type to hash_table.  Update dependent calls and types.
21984         * loop-unroll.c (htab_t opt_info::insns_to_split):
21985         Change type to hash_table.  Update dependent calls and types.
21986         (htab_t opt_info::insns_with_var_to_expand): Likewise.
21988         * lto-streamer.h (struct string_slot): Move from data-streamer.h
21989         (struct string_slot_hasher): New.
21990         (htab_t output_block::string_hash_table):
21991         Change type to hash_table.  Update dependent calls and types.
21993         * lto-streamer-in.c (freeing_string_slot_hasher): New.
21994         (htab_t file_name_hash_table):
21995         Change type to hash_table.  Update dependent calls and types.
21997         * lto-streamer-out.c: Update output_block::string_hash_table dependent
21998         calls and types.
22000         * lto-streamer.c (htab_t tree_htab):
22001         Change type to hash_table.  Update dependent calls and types.
22003         * omp-low.c: Include gimplify-ctx.h.
22005         * passes.c (htab_t name_to_pass_map):
22006         Change type to hash_table.  Update dependent calls and types.
22007         (pass_traverse): Rename to passes_pass_traverse.
22009         * plugin.c (htab_t event_tab):
22010         Change type to hash_table.  Update dependent calls and types.
22012         * postreload-gcse.c (htab_t expr_table):
22013         Change type to hash_table.  Update dependent calls and types.
22014         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
22016         * sese.c (debug_rename_map_1): Make extern.
22017         (htab_t copy_bb_and_scalar_dependences::rename_map):
22018         Change type to hash_table.  Update dependent calls and types.
22020         * sese.h (extern debug_rename_map): Move to .c file.
22022         * store-motion.c (htab_t store_motion_mems_table):
22023         Change type to hash_table.  Update dependent calls and types.
22025         * trans-mem.c (htab_t tm_new_mem_hash):
22026         Change type to hash_table.  Update dependent calls and types.
22028         * tree-browser.c (htab_t TB_up_ht):
22029         Change type to hash_table.  Update dependent calls and types.
22031         * tree-cfg.c (htab_t discriminator_per_locus):
22032         Change type to hash_table.  Update dependent calls and types.
22034         * tree-complex.c: Include tree-hasher.h
22035         (htab_t complex_variable_components):
22036         Change type to hash_table.  Update dependent calls and types.
22038         * tree-eh.c (htab_t finally_tree):
22039         Change type to hash_table.  Update dependent calls and types.
22041         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
22042         struct int_tree_hasher.
22043         (extern int_tree_map_eq): Likewise.
22044         (uid_decl_map_hash): Removed.
22045         (extern decl_tree_map_eq): Likewise.
22047         * tree-hasher.h: New.
22048         (struct int_tree_hasher): New.
22049         (typedef int_tree_htab_type): New.
22051         * tree-inline.c: Include gimplify-ctx.h.
22053         * tree-mudflap.c: Include gimplify-ctx.h.
22055         * tree-parloops.c: Include tree-hasher.h.
22056         (htab_t eliminate_local_variables_stmt::decl_address):
22057         Change type to hash_table.  Update dependent calls and types.
22058         (htab_t separate_decls_in_region::decl_copies): Likewise.
22060         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
22061         Change type to hash_table.  Update dependent calls and types.
22063         * tree-sra.c (candidates):
22064         Change type to hash_table.  Update dependent calls and types.
22066         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
22067         in tree-flow.h.
22068         (int_tree_map_hash): Likewise.
22070         * tree-ssa-dom.c (htab_t avail_exprs):
22071         Change type to hash_table.  Update dependent calls and types.
22073         * tree-ssa-live.c (var_map_base_init::tree_to_index):
22074         Change type to hash_table.  Update dependent calls and types.
22076         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
22077         Change type to hash_table.  Update dependent calls and types.
22079         * tree-ssa-phiopt.c (seen_ssa_names):
22080         Change type to hash_table.  Update dependent calls and types.
22082         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
22083         Change type to hash_table.  Update dependent calls and types.
22085         * tree-ssa-uncprop.c (equiv):
22086         Change type to hash_table.  Update dependent calls and types.
22088 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
22090         PR rtl-optimization/57003
22091         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
22092         call note_stores with kill_clobbered_value callback again after
22093         killing regs_invalidated_by_call.
22095 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22097         * config/aarch64/aarch64-simd.md
22098         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
22099         (aarch64_simd_bsl<mode>): Likewise.
22100         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
22102 2013-04-25  Marek Polacek  <polacek@redhat.com>
22104         PR tree-optimization/57066
22105         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
22107 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22109         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
22111 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22113         * config/aarch64/aarch64-builtins.c
22114         (aarch64_fold_builtin): New.
22115         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
22116         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
22117         * config/aarch64/aarch64-simd-builtins.def (abs): New.
22118         * config/aarch64/arm_neon.h
22119         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
22121 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22122             Tejas Belagod  <tejas.belagod@arm.com>
22124         * config/aarch64/aarch64-builtins.c
22125         (aarch64_gimple_fold_builtin): New.
22126         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
22127         * config/aarch64/aarch64-simd-builtins.def (addv): New.
22128         * config/aarch64/aarch64-simd.md (addpv4sf): New.
22129         (addvv4sf): Update.
22130         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
22132 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22134         * config/aarch64/aarch64.md
22135         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
22137 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22139         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
22140         (*ngcsi_uxtw): New pattern.
22142 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22143             Julian Brown  <julian@codesourcery.com>
22145         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
22146         (TB_DREG): Add T_V4HF.
22147         (v4hf_UP): New macro.
22148         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22149         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22150         Handle initialisation of V4HF. Adjust initialisation of reinterpret
22151         built-ins.
22152         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22153         (arm_vector_mode_supported_p): Handle V4HF.
22154         (arm_mangle_map): Handle V4HFmode.
22155         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
22156         * config/arm/arm_neon_builtins.def: Add entries for
22157         vcvtv4hfv4sf, vcvtv4sfv4hf.
22158         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
22159         (neon_vcvtv4hfv4sf): Likewise.
22160         * config/arm/neon-gen.ml: Handle half-precision floating point
22161         features.
22162         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
22163         * config/arm/arm_neon.h: Regenerate.
22164         * config/arm/neon.ml (type elts): Add F16.
22165         (type vectype): Add T_float16x4, T_floatHF.
22166         (type vecmode): Add V4HF.
22167         (type features): Add Requires_FP_bit feature.
22168         (elt_width): Handle F16.
22169         (elt_class): Likewise.
22170         (elt_of_class_width): Likewise.
22171         (mode_of_elt): Refactor.
22172         (type_for_elt): Handle F16, fix error messages.
22173         (vectype_size): Handle T_float16x4.
22174         (vcvt_sh): New function.
22175         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
22176         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
22177         (string_of_mode): Handle V4HF.
22178         * doc/arm-neon-intrinsics.texi: Regenerate.
22180 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22182         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
22183         format specifier in 'X' case.
22185 2013-04-25  Alan Modra  <amodra@gmail.com>
22187         PR target/57052
22188         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
22189         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
22190         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
22191         Repeat for many other rotate/shift and mask patterns using subregs.
22192         Name lshiftrt insns.
22193         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
22194         on WORDS_BIG_ENDIAN.
22196 2013-04-25  Alan Modra  <amodra@gmail.com>
22198         * config.gcc: Support little-endian powerpc-linux targets.
22199         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
22200         (LINK_OS_LINUX_SPEC): Define.
22201         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
22202         Preserve MASK_LITTLE_ENDIAN.
22203         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
22204         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
22205         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
22206         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
22207         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
22208         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
22209         Correct fp word order for little-endian.  Don't shift toc entries
22210         smaller than a word for little-endian.
22211         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
22212         (bswapdi2 splits): Correct low-part subreg for little-endian.
22213         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
22214         low/high where such is correct only for be.
22215         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
22216         little-endian for -mcall-aixdesc.
22218 2013-04-25  Alan Modra  <amodra@gmail.com>
22220         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
22221         replace_equiv_address_nv.
22223 2013-04-25  Alan Modra  <amodra@gmail.com>
22225         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
22227 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22229         Revert:
22230         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22231         * rtl.h (struct rtx_def): ...
22233 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22235         PR rtl-optimizations/57046
22236         * lra-constraints (split_reg): Set up lra_risky_transformations_p
22237         for multi-reg splits.
22239 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
22241         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
22243 2013-04-24  Sterling Augustine  <saugustine@google.com>
22245         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
22246         (comp_dir_string, debug_str_dwo_section): New.
22247         (DEBUG_STR_DWO_SECTION): Rename to ...
22248         (DEBUG_DWO_STR_SECTION): ... this.
22249         (DEBUG_NORM_STR_SECTION): Delete.
22250         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
22251         (DEBUG_STR_DWO_SECTION_FLAGS): New.
22252         (find_AT_string): Move most logic to ...
22253         (find_AT_string_in_table): ... here.  New.
22254         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
22255         add_skeleton_AT_string.  Delete logic.
22256         (output_skeleton_debug_sections): Remove call to
22257         add_top_level_skeleton_die_attrs.
22258         (add_comp_dir_attribute): Move logic to comp_dir_string.
22259         (dwarf2out_init): Initialize debug_str_dwo_section.
22260         (output_indirect_string): Call find_string_form.
22261         (output_indirect_strings): Rewrite.
22262         (prune_unused_types): Empty skeleton_debug_str_hash.
22263         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
22264         (dwarf2out_finish):  Call output_indirect_strings.
22266 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22268         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
22270 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22272         * rtl.h (struct rtx_def): Add comment for field jump.
22273         (LRA_SUBREG_P): New macro.
22274         * recog.c (register_operand): Check LRA_SUBREG_P.
22275         * lra.c (lra): Add note at the end of RTL code. Align non-empty
22276         stack frame.
22277         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
22278         (lra_final_code_change): Skip subreg change for operators.
22279         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
22280         if there are no operand changes.
22281         * lra-constraints.c (curr_insn_set): New.
22282         (match_reload): Set LRA_SUBREG_P.
22283         (emit_spill_move): Ditto.
22284         (check_and_process_move): Use curr_insn_set. Process only single
22285         set insns.  Don't initialize sec_mem_p and change_p.
22286         (simplify_operand_subreg): Use LRA_SUBREG_P.
22287         (reg_in_class_p): New function.
22288         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
22289         of #ifdef.  Add code to remove cycling.
22290         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
22291         non-null disp.  Reload inner instead of disp when base and index
22292         are null.  Try to put lo_sum into register.
22293         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
22294         (check_and_process_move): Move code for move cost check to
22295         simple_move_p.  Remove equiv_substitution.
22296         (simple_move_p): New function.
22297         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
22298         curr_insn_set.  Call check_and_process_move only for single set
22299         insns.  Use the new function.  Move call of check_and_process_move
22300         after operand equiv substitution and address process.
22302 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
22304         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
22305         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
22306         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
22308 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22310         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
22312 2013-04-24  Marek Polacek  <polacek@redhat.com>
22314         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
22315         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
22316         (select_loops_exit_conditions): Likewise.
22317         (number_of_iterations_for_all_loops): Likewise.
22318         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
22319         (scev_analysis): Likewise.
22321 2013-04-02  Catherine Moore  <clm@codesourcery.com>
22322             Chao-ying Fu  <fu@mips.com>
22324         * config/mips/micromips.md (jraddiusp): New pattern.
22325         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
22326         instruction if possible.
22328 2013-04-24  Alan Modra  <amodra@gmail.com>
22330         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
22332 2013-04-24  Julian Brown  <julian@codesourcery.com>
22333             Chung-Lin Tang  <cltang@codesourcery.com>
22335         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
22336         dependency behavior in enumeration type DIE generation. Add TODO note
22337         to comments about future DW_FORM_sdata/udata re-work of related code.
22339 2013-04-23  Lawrence Crowl  <crowl@google.com>
22341         * Makefile.in: Update as needed below.
22343         * hash-table.h (class hash_table):
22344         Correct many methods with parameter types compare_type to the correct
22345         value_type.  (Correct code was unlikely to notice the change.)
22346         (hash_table::elements_with_deleted) New.
22347         (class hashtable::iterator): New.
22348         (hashtable::begin()): New.
22349         (hashtable::end()): New.
22350         (FOR_EACH_HASH_TABLE_ELEMENT): New.
22352         * statistics.c (statistics_hashes):
22353         Change type to hash_table.  Update dependent calls and types.
22355         * tree-into-ssa.c (var_infos):
22356         Change type to hash_table.  Update dependent calls and types.
22358         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
22359         Change type to hash_table.  Update dependent calls and types.
22361         * tree-ssa-loop-im.c (struct mem_ref.refs):
22362         Change type to hash_table.  Update dependent calls and types.
22364         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
22365         Change type to hash_table.  Update dependent calls and types.
22367         * tree-ssa-sccvn.c (vn_tables_s::nary):
22368         Change type to hash_table.  Update dependent calls and types.
22369         (vn_tables_s::phis): Likewise.
22370         (vn_tables_s::references): Likewise.
22372         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
22373         (vn_reference_eq): Update parameter and return types.
22375         * tree-ssa-structalias.c (pointer_equiv_class_table):
22376         Change type to hash_table.  Update dependent calls and types.
22377         (location_equiv_class_table): Likewise.
22379         * tree-vect-data-refs.c: Consequential changes for making
22380         peeling a hash_table.
22382         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
22383         (destroy_loop_vec_info): Dependent hash_table update.
22385         * tree-vectorizer.h (peeling_htab):
22386         Change type to hash_table.  Update dependent calls and types.
22388 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
22390         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
22391         to check the register content is equal or not.
22392         * lra-constraints.c (match_reload): Use lra_assign_reg_val
22393         to assign register content record.
22394         * lra-eliminations.c (update_reg_eliminate): Use
22395         lra_update_reg_val_offset to update register content offset.
22396         * lra-int.h (struct lra_reg): Add offset member.
22397         (lra_reg_val_equal_p): New static inline function.
22398         (lra_update_reg_val_offset): New static inline function.
22399         (lra_assign_reg_val): New static inline function.
22400         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
22401         to assign register content record.
22402         (initialize_lra_reg_info_element): Initial offset to zero.
22404 2013-04-23  Catherine Moore  <clm@codesourcery.com>
22406         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
22407         operands.  Record compression.
22409 2013-04-23  Xinliang David Li  <davidxl@google.com>
22411         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
22413 2013-04-23  Richard Biener  <rguenther@suse.de>
22415         PR middle-end/57036
22416         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
22417         parameter, only add abnormal goto edges from the copied body
22418         if the call could perform abnormal gotos.
22419         (copy_cfg_body): Adjust.
22421 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
22423         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
22425 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
22427         * coretypes.h (gimple_stmt_iterator): Add struct to make
22428         compatible with C.
22430 2013-04-23  Richard Biener  <rguenther@suse.de>
22432         PR tree-optimization/57026
22433         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
22434         from SSA names occuring in abnormal PHI nodes.
22436 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22438         * lto/lto.c (print_lto_report_1): Fix LTO report names.
22440 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22442         * lto/lto.c (print_lto_report_1): Declare early.
22443         (read_cgraph_and_symbols): Call print_lto_report_1 early.
22445 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22447         * common.opt (-flto-report-wpa): Add.
22448         * doc/invoke.texi (-flto-report-wpa): Add.
22449         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
22450         (lto_main): dito.
22452 2013-04-22  Xinliang David Li  <davidxl@google.com>
22454         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
22455         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
22456         * Makefile.in: New dependency
22458             David Daney <ddaney.cavm@gmail.com>
22460         * configure.ac (gcc_cv_as_micromips_support):  Use the
22461         --fatal-warnings option.
22462         * configure: Regenerate.
22464 2013-04-22  Marek Polacek  <polacek@redhat.com>
22466         PR sanitizer/56990
22467         * tsan.c (instrument_expr): Don't instrument expression
22468         in case its size is zero.
22470 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
22472         PR target/57032
22473         Revert:
22474         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22476         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22478 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22480         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
22481         (gimple_stmt_iterator): New typedef.
22482         * gimple.h (gimple_stmt_iterator): Rename to...
22483         (gimple_stmt_iterator_d): ... This.
22484         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
22485         trees be valid for GIMPLE and GENERIC.
22486         (TARGET_GIMPLE_FOLD_BUILTIN): New.
22487         * gimple-fold.c (gimple_fold_call): Call target hook
22488         gimple_fold_builtin.
22489         * hooks.c (hook_bool_gsiptr_false): New.
22490         * hooks.h (hook_bool_gsiptr_false): New.
22491         * target.def (fold_stmt): New.
22492         * doc/tm.texi: Regenerate.
22494 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
22496         PR target/57018
22497         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
22498         a set sp if no stack realignment.
22500 2013-04-22  Nick Clifton  <nickc@redhat.com>
22502         * config.gcc (tilegx-linux): Extend extra_objs rather than
22503         overwriting it.
22504         (tilepro-linux): Likewise.
22506 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22508         * config/aarch64/aarch64-builtins.c
22509         (CF): Remove.
22510         (CF0, CF1, CF2, CF3, CF4, CF10): New.
22511         (VAR<1-12>): Add MAP parameter.
22512         (BUILTIN_*): Likewise.
22513         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
22514         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
22515         (aarch64_ushl_n<mode>): Likewise.
22516         (aarch64_sshr_n<mode>): Likewise.
22517         (aarch64_ushr_n<mode>): Likewise.
22518         (aarch64_<maxmin><mode>): Likewise.
22519         (aarch64_sqrt<mode>): Likewise.
22520         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
22521         (vshr<q>_n_*): Likewise.
22523 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22525         * config/aarch64/aarch64-builtins.c
22526         (aarch64_simd_builtin_type_mode): Handle SF types.
22527         (sf_UP): Define.
22528         (BUILTIN_GPF): Define.
22529         (aarch64_init_simd_builtins): Handle SF types.
22530         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
22531         (frecps): Likewise.
22532         (frecpx): Likewise.
22533         * config/aarch64/aarch64-simd.md
22534         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
22535         (aarch64_frecpe<mode>): New.
22536         (aarch64_frecps<mode>): Likewise.
22537         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
22538         (v8type): Add frecp<esx>.
22539         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
22540         (aarch64_frecps<mode>): Likewise.
22541         * config/aarch64/iterators.md (FRECP): New.
22542         (frecp_suffix): Likewise.
22543         * config/aarch64/arm_neon.h
22544         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
22546 2013-04-22  Christian Bruel  <christian.bruel@st.com>
22548         PR target/56995
22549         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
22550         (REG_CLASS_NAMES): Idem.
22551         (REG_CLASS_CONTENTS): Idem.
22552         (REGCLASS_HAS_FP_REG): Idem.
22553         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
22554         (sh_conditional_register_usage): Idem.
22556 2013-04-21  Jeff Law  <law@redhat.com>
22558         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
22559         (ssa_forward_propagate_and_combine): Use it.
22561 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22563         * lra.c: Update the flow chart diagram.
22565 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22567         PR rtl-optimization/56847
22568         * lra-constraints.c (process_alt_operands): Discourage alternative
22569         with non-matche doffsettable memory constraint fro memory with
22570         known offset.
22572 2013-04-19  Richard Biener  <rguenther@suse.de>
22574         PR tree-optimization/56982
22575         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
22576         function.
22577         * gimplify.c (gimplify_call_expr): Notice special calls.
22578         (gimplify_modify_expr): Likewise.
22579         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
22580         abnormal control flow receivers.
22581         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
22582         in the same way as cfun->has_nonlocal_labels.
22583         (gimple_purge_dead_abnormal_call_edges): Likewise.
22584         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
22585         receivers start a basic-block.
22587 2013-04-19  Richard Biener  <rguenther@suse.de>
22589         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
22590         member ...
22591         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
22592         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
22593         (SLP_TREE_LOAD_PERMUTATION): Add.
22594         (vect_transform_slp_perm_load): Adjust prototype.
22595         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
22596         (vect_free_slp_instance): Likewise.
22597         (vect_create_new_slp_node): Likewise.
22598         (vect_supported_slp_permutation_p): Remove.
22599         (vect_slp_rearrange_stmts): Adjust.
22600         (vect_supported_load_permutation_p): Likewise.  Inline
22601         vect_supported_slp_permutation_p here.
22602         (vect_analyze_slp_instance): Compute load permutations per
22603         slp node instead of per instance.
22604         (vect_get_slp_defs): Adjust.
22605         (vect_transform_slp_perm_load): Likewise.
22606         (vect_schedule_slp_instance): Remove redundant code.
22607         (vect_schedule_slp): Remove hack for PR56270, add it ...
22608         * tree-vect-stmts.c (vectorizable_load): ... here, do not
22609         CSE loads for SLP.  Adjust.
22611 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22613         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
22614         spelling in two comments.
22616 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22618         PR target/56797
22619         * config/arm/arm.c (load_multiple_sequence): Require SP
22620         as base register for loads if SP is in the register list.
22622 2013-04-19  Martin Jambor  <mjambor@suse.cz>
22624         PR tree-optimization/56718
22625         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
22626         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
22627         and made public.  Adjusted all callers.
22628         (ipa_intraprocedural_devirtualization): New function.
22629         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
22630         (ipa_intraprocedural_devirtualization): Likewise.
22631         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
22633 2013-04-19  Richard Biener  <rguenther@suse.de>
22635         PR tree-optimization/57000
22636         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
22638 2013-04-19  Terry Guo  <terry.guo@arm.com>
22640         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
22641         Replace with ...
22642         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
22643         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
22644         (cortex_m4_fmacs): Use new reservations.
22645         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
22647 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
22649         PR rtl-optimization/56999
22650         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
22651         related code.
22652         (lra_coalesce): Remove split_origin_bitmap and related code.
22653         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
22654         ranges if necessary.
22656 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
22658         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
22659         New array.
22660         (ix86_expand_call): Remove clobbered_registers array and use
22661         x86_64_ms_sysv_extra_clobbered_registers instead.
22662         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
22663         Declare here.
22664         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
22665         predicate.
22666         * config/i386/i386.md (*call_rex64_ms_sysv): Use
22667         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
22668         (*call_value_rex64_ms_sysv): Ditto.
22670 2013-04-18  Cary Coutant  <ccoutant@google.com>
22672         * dwarf2out.c (output_pubnames): Check die_perennial_p of
22673         parent instead of die_mark.
22675 2013-04-18  Diego Novillo  <dnovillo@google.com>
22677         * gimple.c (create_gimple_tmp): New.
22678         (get_expr_type): New.
22679         (build_assign): New.
22680         (build_type_cast): New.
22681         * gimple.h (enum ssa_mode): Define.
22682         (gimple_seq_set_location): New.
22683         * asan.c (build_check_stmt): Change some gimple_build_* calls
22684         to use build_assign and build_type_cast.
22686 2013-04-18  Richard Biener  <rguenther@suse.de>
22688         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
22689         handle negative step.  Remove redundant checks.
22690         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
22691         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
22692         for negative step and grouped loads fail to vectorize.
22694 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22696         * emit-rtl.c (reset_insn_used_flags): New function.
22697         (reset_all_used_flags): Use it.
22698         (verify_insn_sharing): New function.
22699         (verify_rtl_sharing): Fix verification for SEQUENCEs.
22701 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
22703         PR tree-optimization/56984
22704         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
22705         and (x >> M) >= N don't register any assertion if N << M is the
22706         minimum value.
22708 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22710         * lower-subreg.c (resolve_simple_move): If called self-recursive,
22711         do not delete_insn insns that have not yet been emitted, only
22712         unlink them with remove_insn.
22713         * df-scan.c (df_insn_delete): Revert r197492.
22715 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
22717         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
22718         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
22720 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22722         * config/arm/arm.md (movsicc_insn): Convert define_insn into
22723         define_insn_and_split.
22724         (and_scc,ior_scc,negscc): Likewise.
22725         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
22727 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22729         * config/arm/arm.c (use_return_insn): Return 0 for targets that
22730         can benefit from using a sequence of LDRD instructions in epilogue
22731         instead of a single LDM instruction.
22733 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
22735         PR 45688
22736         * doc/extend.texi: Fix typo.
22738 2013-04-17  Richard Biener  <rguenther@suse.de>
22740         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
22741         (vect_build_slp_tree): ... here.
22742         (vect_build_slp_tree_1): Compute which stmts of the SLP group
22743         match.  Remove special-casing of mismatched complex loads.
22744         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
22745         re-try the match with swapped commutative operands.
22746         (vect_supported_load_permutation_p): Remove special-casing of
22747         mismatched complex loads.
22748         (vect_analyze_slp_instance): Adjust.
22750 2013-04-17  Richard Biener  <rguenther@suse.de>
22752         PR rtl-optimization/56921
22753         * cfgloop.h (struct loop): Add simple_loop_desc member.
22754         (struct niter_desc): Mark with GTY(()).
22755         (simple_loop_desc): Do not use aux field but simple_loop_desc.
22756         * loop-iv.c (get_simple_loop_desc): Likewise.
22757         (free_simple_loop_desc): Likewise.
22759         Revert
22760         2013-04-16  Richard Biener  <rguenther@suse.de>
22762         PR rtl-optimization/56921
22763         * loop-init.c (pass_rtl_move_loop_invariants): Add
22764         TODO_do_not_ggc_collect to todo_flags_finish.
22765         (pass_rtl_unswitch): Same.
22766         (pass_rtl_unroll_and_peel_loops): Same.
22767         (pass_rtl_doloop): Same.
22769 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
22771         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
22772         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
22773         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
22774         references.
22775         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
22776         * tree-streamer.c (record_common_node): Adjust reference in comment.
22778 2013-04-17  Terry Guo  <terry.guo@arm.com>
22780         * config/arm/cortex-m4.md: Add a new bypass.
22782 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22784         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
22785         New pattern.
22786         (*subs_<optab><mode>_multp2): New pattern.
22787         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22788         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22790 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22792         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
22793         (*subs_mul_imm_<mode>): New pattern.
22795 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
22797         PR target/56948
22798         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
22799         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
22800         (vsx_movti_32bit): Change j->wa to O->wa.
22802 2013-04-16  Richard Biener  <rguenther@suse.de>
22804         PR rtl-optimization/56921
22805         * loop-init.c (pass_rtl_move_loop_invariants): Add
22806         TODO_do_not_ggc_collect to todo_flags_finish.
22807         (pass_rtl_unswitch): Same.
22808         (pass_rtl_unroll_and_peel_loops): Same.
22809         (pass_rtl_doloop): Same.
22811 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
22813         * config/arm/arm.c (emit_multi_reg_push): New declaration
22814         for an existing function.
22815         (arm_emit_strd_push): New function.
22816         (arm_expand_prologue): Used here.
22817         (arm_emit_ldrd_pop): New function.
22818         (arm_expand_epilogue): Used here.
22819         (arm_get_frame_offsets): Update condition.
22820         (arm_emit_multi_reg_pop): Add a special case for load of a single
22821         register with writeback.
22823 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
22825         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
22826         description.
22828 2013-04-16  Richard Biener  <rguenther@suse.de>
22830         PR tree-optimization/56756
22831         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
22832         (first_mem_ref_loc): New.
22833         (execute_sm): Place the load temporarily before a previous
22834         access instead of in the latch edge to ensure its SSA dependencies
22835         are defined at points dominating the load.
22837 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
22839         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
22840         correct fix by moving header and footer insn to the footer of
22841         the merged basic block.  Clear BB_END of the merged-away block.
22843         PR middle-end/43631
22844         * emit-rtl.c (make_note_raw): New function.
22845         (link_insn_into_chain): New static inline function.
22846         (add_insn): Use it.
22847         (add_insn_before, add_insn_after): Factor insn chain linking code...
22848         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
22849         using link_insn_into_chain.
22850         (note_outside_basic_block_p): New helper function for emit_note_after
22851         and emit_note_before.
22852         (emit_note_after): Use nobb variant of add_insn_after if the note
22853         should not be contained in a basic block.
22854         (emit_note_before): Use nobb variant of add_insn_before if the note
22855         should not be contained in a basic block.
22856         (emit_note_copy): Use make_note_raw.
22857         (emit_note): Likewise.
22858         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
22859         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
22860         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
22861         the moved barrier the tail of the basic block it follows.
22862         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
22864 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
22866         PR tree-optimization/56962
22867         * gimple-ssa-strength-reduction.c (record_increment): Only set
22868         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
22869         either rhs1 or rhs2 is equal to c->base_expr.
22871 2013-04-15  Richard Biener  <rguenther@suse.de>
22873         PR tree-optimization/56933
22874         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
22875         member.
22876         (GROUP_READ_WRITE_DEPENDENCE): Remove.
22877         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
22878         * tree-vect-data-refs.c (vect_analyze_group_access): Move
22879         dependence check ...
22880         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
22881         ... here.
22882         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
22883         GROUP_READ_WRITE_DEPENDENCE.
22885 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22887         * emit-rtl.c (reset_all_used_flags): New function.
22888         (verify_rtl_sharing): Call reset_all_used_flags before and after
22889         performing the checks.
22891 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22893         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
22894         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
22895         * config/arm/constraints.md (De): New constraint.
22896         * config/arm/neon.md (anddi3_neon): Delete.
22897         (neon_vand<mode>): Expand to standard anddi3 pattern.
22898         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
22899         Move earlier in the file.
22900         (neon_inv_logic_op2): Likewise.
22901         (arm_anddi_operand_neon): New predicate.
22903 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22905         * configure.ac (gcc_cv_ld_as_needed): Set
22906         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
22907         Use -z ignore, -z record on *-*-solaris2*.
22908         (HAVE_LD_AS_NEEDED): Update comment.
22909         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
22910         * configure: Regenerate.
22911         * config.in: Regenerate.
22912         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
22913         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
22914         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
22915         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
22916         equivalents.  Fix markup.
22917         * doc/tm.texi: Regenerate.
22919 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
22921         * config/i386/i386.opt: New option mstack-protector-guard=.
22922         * config/i386/i386-opts.h: Add enum stack_protector_guard.
22923         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
22924         TARGET_SSP_TLS_GUARD.
22925         * config/i386/i386.c (ix86_option_override_internal): Set
22926         ix86_stack_protector_guard.
22927         * config/i386/i386.md (stack_protect_set): Enable for
22928         TARGET_SSP_TLS_GUARD only.
22929         (stack_protect_set_<mode>): Ditto.
22930         (stack_protect_test): Ditto.
22931         (stack_protect_test_<mode>): Ditto.
22932         * doc/invoke.texi (i386 Option): Document.
22934 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
22936         PR target/56890
22937         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
22938         (S_MODES): Set H_MODE bit.
22939         (SF_MODES): Set only S_MODE and SF_MODE bits.
22940         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
22941         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
22942         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
22943         <MODE_FLOAT>: Likewise.
22945 2013-04-15  Joey Ye  <joey.ye@arm.com>
22947         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
22949 2013-04-15  Joey Ye  <joey.ye@arm.com>
22951         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
22952         for real far jump.
22953         (thumb_far_jump_used_p): Count instruction size and set
22954         far_jump_used.
22956 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
22958         * reorg.c (fill_simple_delay_slots): Reindent block of code.
22959         * resource.c (mark_target_live_regs): Reformat conditional block.
22961 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22963         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
22964         notes, they are emitted only just before final.
22965         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
22967 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
22969         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
22970         * cfgrtl.c (delete_insn): Call it here instead.
22971         * lra-spills.c (lra_final_code_change): Use delete_insn.
22972         * haifa-sched.c (sched_remove_insn): Likewise.
22973         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
22974         returning to the nop pool.
22975         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
22976         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
22978 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22980         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
22981         * doc/tm.texi: Regenerated.
22983 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
22985         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
22986         QImode checks.
22988 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
22990         * df-core.c (df_find_def): Compare register numbers.
22991         (df_find_use): Likewise.
22993 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
22995         PR target/56903
22996         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
22997         lra_in_progress for return.
22999 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23001         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
23002         define_insn into define_insn_and_split and emit movsicc patterns.
23004 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23006         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
23008 2013-04-12  Richard Biener  <rguenther@suse.de>
23010         * tree-pass.h (TODO_do_not_ggc_collect): New.
23011         * passes.c (execute_one_ipa_transform_pass): Honor
23012         TODO_do_not_ggc_collect.
23013         (execute_one_pass): Likewise.
23015         Revert
23016         2013-04-10  Richard Biener  <rguenther@suse.de>
23018         * passes.c (init_optimization_passes): Remove reload pass.
23019         * ira.c (do_reload): Merge into ...
23020         (ira): ... this.
23021         (rest_of_handle_reload): Remove.
23022         (pass_reload): Likewise.
23023         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23024         of reload for vzeroupper pass placement.
23026 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
23028         PR tree-optimization/56918
23029         PR tree-optimization/56920
23030         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
23031         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
23032         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
23033         use wide_mul_with_sign method.
23035 2013-04-12  Richard Biener  <rguenther@suse.de>
23037         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
23038         not be considered a gimple constant.
23040 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
23042         * fold-const.c (const_binop): Handle vector shifts by a scalar.
23043         (fold_binary_loc): Call const_binop also for mixed vector-scalar
23044         operations.
23046 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
23047             Jakub Jelinek  <jakub@redhat.com>
23049         * opts.c: Include diagnostic-color.h.
23050         (common_handle_option): Handle OPT_fdiagnostics_color_.
23051         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
23052         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
23053         (diagnostic-color.o): New.
23054         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
23055         (diagnostic_color_rule): New enum.
23056         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
23057         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
23058         the location string.
23059         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
23060         either NULL, or color kind.
23061         * diagnostic-color.c: New file.
23062         * diagnostic-color.h: New file.
23063         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
23064         arguments.
23065         * doc/invoke.texi (-fdiagnostics-color): Document.
23066         * pretty-print.h (pp_show_color): Define.
23067         (struct pretty_print_info): Add show_color field.
23068         * diagnostic.c: Include diagnostic-color.h.
23069         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
23070         macros.  Colorize error:, warning: etc. strings and also the location
23071         string.
23072         (diagnostic_show_locus): Colorize the caret line.
23073         * pretty-print.c: Include diagnostic-color.h.
23074         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
23075         inside of %< %> quotes or quoted through q format modifier.
23077 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23079         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
23081 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23083         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
23084         code in CC_NZ mode.
23085         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
23086         pattern.
23088 2013-04-11  Marek Polacek  <polacek@redhat.com>
23090         PR tree-optimization/48184
23091         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
23093 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
23095         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
23096         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
23097         (skip_simple_arithmetic): Tidy up.
23098         * tree.h (skip_simple_constant_arithmetic): Declare.
23100 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23102         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
23104 2013-04-11  Richard Biener  <rguenther@suse.de>
23106         * tree-vect-loop.c (get_initial_def_for_induction): Properly
23107         generate vector constants.
23109 2013-04-11  Richard Biener  <rguenther@suse.de>
23111         PR tree-optimization/56878
23112         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
23113         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
23114         New function.
23115         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23116         Prefer to align the DR with the most invariant base address.
23118 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23120         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
23121         comment.
23123 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
23125         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
23126         floating-point vector comparisons against 0.
23128 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
23130         PR tree-optimization/56899
23131         * fold-const.c (extract_muldiv_1): Apply distributive law
23132         only if TYPE_OVERFLOW_WRAPS (ctype).
23134 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
23136         PR target/56124
23137         * ira-costs.c (scan_one_insn): Check whether the source rtx of
23138         loading has side effect.
23140 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
23142         * config/sparc/sparc.c: Include tree-pass.h.
23143         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
23144         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
23145         head of file.  Change return type.  Split off gate function.
23146         (sparc_gate_work_around_errata): New function.
23147         (pass_work_around_errata): New pass definition.
23148         (insert_pass_work_around_errata) New pass insert definition to
23149         insert pass_work_around_errata just after delayed-branch scheduling.
23150         (sparc_option_override): Insert the pass.
23151         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
23153 2013-04-10  David S. Miller  <davem@davemloft.net>
23155         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
23156         or -mcpu=hypersparc.
23158         * target.def (cstore_mode): New hook.
23159         * target.h: Include insn-codes.h
23160         * targhooks.c: Likewise.
23161         (default_cstore_mode): New function.
23162         * targhooks.h: Declare it.
23163         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
23164         * doc/tm.texi: Rebuild.
23165         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
23166         target hook, rather than inspecting the insn_data.
23167         * config/sparc/sparc.c (sparc_cstore_mode): New function.
23168         (TARGET_CSTORE_MODE): Redefine.
23169         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
23170         result patterns.
23171         * config/sparc/predicates.md (cstore_result_operand): New special
23172         predicate.
23173         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
23174         Use it for operand 0.
23175         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
23176         (*snesi_special): Likewise.
23177         (*snesi_zero): Likewise.
23178         (*seqsi_zero): Likewise.
23179         (*sltu_insn): Likewise.
23180         (*sgeu_insn): Likewise.
23181         (*seqdi_special): Make operand 0 and comparison operation be of
23182         DImode.
23183         (*snedi_special): Likewise.
23184         (*snedi_special_vis3): Likewise.
23185         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
23186         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
23187         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
23188         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
23189         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
23190         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
23191         (*sltu_extend_sp64): Likewise.
23192         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
23193         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
23194         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
23195         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
23196         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
23198 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
23200         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
23201         (aarch64_start_file): Use the new function.
23203 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23204             Jason Merrill  <jason@redhat.com>
23206         * common.opt: Add -gdwarf.
23207         * opts.c (common_handle_option): Handle it.
23208         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
23210 2013-04-10  Richard Biener  <rguenther@suse.de>
23212         * passes.c (execute_todo): Do not call ggc_collect conditional here.
23213         (execute_one_ipa_transform_pass): But unconditionally here.
23214         (execute_one_pass): And here.
23215         (init_optimization_passes): Remove reload pass.
23216         * tree-pass.h (TODO_ggc_collect): Remove.
23217         (pass_reload): Likewise.
23218         * ira.c (do_reload): Merge into ...
23219         (ira): ... this.
23220         (rest_of_handle_reload): Remove.
23221         (pass_reload): Likewise.
23222         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23223         of reload for vzeroupper pass placement.
23224         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
23225         and todo_flags_finish of all passes.
23227 2013-04-10  Richard Biener  <rguenther@suse.de>
23229         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
23230         first_const_oprnd field, rename first_def_type to first_op_type.
23231         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
23232         (vect_get_and_check_slp_defs): Always use the type of the
23233         operand.  Allow mixed vect_external_def, vect_constant_def types.
23234         (vect_get_constant_vectors): Handle mixed vect_external_def,
23235         vect_constant_def types.
23237 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
23239         PR tree-optimization/55524
23240         * tree-ssa-math-opts.c
23241         (convert_mult_to_fma): Don't use an fms construct
23242         when we don't have an fms operation, but fnma, and it looks
23243         likely that we'll be able to use the latter.
23245 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
23247         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
23248         function.
23249         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
23250         inline fail caused by overwritable functions.
23252 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
23254         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
23255         unnecessary bits in the constant power of two case.
23257 2013-04-10  Richard Biener  <rguenther@suse.de>
23259         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
23260         broken code swapping operands.
23261         (vect_build_slp_tree): Do not compute load permutations here.
23262         (vect_analyze_slp_instance): Compute load permutations here,
23263         after building the SLP tree.
23265 2013-04-09  Christian Bruel  <christian.bruel@st.com>
23267         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
23268         of next/prev_real_insn.
23270 2013-04-09  Jan Hubicka  <jh@suse.cz>
23272         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
23273         Drop aliased parameter.
23274         (function_and_variable_visibility): Do not handle alias pairs.
23275         * cgraph.c (varpool_externally_visible_p): Update prototype.
23276         * varpool.c (varpool_add_new_variable): Update.
23278 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23280         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
23282 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
23284         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
23286         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
23288 2013-04-09  Marek Polacek  <polacek@redhat.com>
23290         PR tree-optimization/48762
23291         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
23293 2013-04-09  Richard Biener  <rguenther@suse.de>
23295         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
23296         dealing with cost.
23297         (vect_build_slp_tree): Likewise.
23298         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
23299         calculating the cost of a SLP instance.
23300         (vect_analyze_slp_instance): Use it from here, after building
23301         the SLP tree.
23303 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23305         PR middle-end/56883
23306         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
23307         expand_omp_for_static_chunk): Use simple_p = true in
23308         force_gimple_operand_gsi calls when assigning to addressable decls.
23310 2013-04-09  Jeff Law  <law@redhat.com>
23312         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
23313         when the boolean was created by converting a wider object which
23314         had a boolean range.
23316 2013-04-09  Richard Biener  <rguenther@suse.de>
23318         * tree-vectorizer.h (slp_void_p): Remove.
23319         (slp_tree): Typedef before _slp_tree declaration.
23320         (struct _slp_tree): Use a vector of slp_tree as children.
23321         (vect_get_place_in_interleaving_chain): Remove.
23322         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
23323         Move ...
23324         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
23325         and make static.
23326         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
23327         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
23328         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
23329         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
23330         Use slp_node instead of slp_void_p and adjust.
23332 2013-04-09  Richard Biener  <rguenther@suse.de>
23334         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
23335         work that is not necessary.
23337 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23339         PR tree-optimization/56854
23340         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
23341         forward into clobber stmts if it would change MEM_REF lhs into
23342         non-MEM_REF.
23344 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
23346         * tree.c (type_hash_lookup, type_hash_add): Make static.
23347         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
23349 2013-04-09  Richard Biener  <rguenther@suse.de>
23351         * tree.h (unsave_expr_now): Remove.
23352         * tree-inline.c (mark_local_for_remap_r): Remove.
23353         (unsave_expr_1): Likewise.
23354         (unsave_r): Likewise.
23355         (unsave_expr_now): Likewise.
23356         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
23357         (propagate_tree_value): Likewise.
23359 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
23361         * doc/rtl.texi (sequence): Rewrite documentation to match the
23362         current use of SEQUENCE rtl objects.
23363         * rtl.def (SEQUENCE): Likewise.
23365         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
23366         Update documentation.
23367         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
23368         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
23370         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
23372 2013-04-08  Teresa Johnson  <tejohnson@google.com>
23374         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
23375         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
23376         methods.
23377         (estimate_edge_size_and_time): Add comment to suggest using rounding
23378         methods.
23379         (estimate_node_size_and_time): Ditto.
23380         (remap_edge_change_prob): Use helper rounding divide methods.
23381         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
23382         (gimple_mod_pow2_value_transform): Ditto.
23383         (gimple_mod_subtract_transform): Ditto.
23384         (gimple_ic_transform): Ditto.
23385         (gimple_stringops_transform): Ditto.
23386         * stmt.c (conditional_probability): Ditto.
23387         (emit_case_dispatch_table): Ditto.
23388         * lto-cgraph.c (merge_profile_summaries): Ditto.
23389         * tree-optimize.c (execute_fixup_cfg): Ditto.
23390         * cfgcleanup.c (try_forward_edges): Ditto.
23391         * cfgloopmanip.c (scale_loop_profile): Ditto.
23392         (loopify): Ditto.
23393         (duplicate_loop_to_header_edge): Ditto.
23394         (lv_adjust_loop_entry_edge): Ditto.
23395         * tree-vect-loop.c (vect_transform_loop): Ditto.
23396         * profile.c (compute_branch_probabilities): Ditto.
23397         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
23398         * lto-streamer-in.c (input_cfg): Ditto.
23399         * gimple-streamer-in.c (input_bb): Ditto.
23400         * ipa-cp.c (update_profiling_info): Ditto.
23401         (update_specialized_profile): Ditto.
23402         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
23403         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
23404         rounding methods.
23405         * sched-rgn.c (compute_dom_prob_ps): Ditto.
23406         (compute_trg_info): Ditto.
23407         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
23408         (purge_dead_edges): Ditto.
23409         * loop-unswitch.c (unswitch_loop): Ditto.
23410         * cgraphclones.c (cgraph_clone_edge): Ditto.
23411         (cgraph_clone_node): Ditto.
23412         * tree-inline.c (copy_bb): Ditto.
23413         (copy_edges_for_bb): Ditto.
23414         (initialize_cfun): Ditto.
23415         (copy_cfg_body): Ditto.
23416         (expand_call_inline): Ditto.
23418 2013-04-08  Kai Tietz  <ktietz@redhat.com>
23420         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
23421         TARGET_CYGWIN64 by TARGET_64BIT.
23423 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23425         * config/epiphany/epiphany.md (GPR_1): New constant.
23426         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
23427         * config/epiphany/epiphany.c (gen_compare_reg):
23428         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
23429         is already in place.
23430         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
23431         Don't require being called during rtl expansion; If y operlaps r0,
23432         return 0.
23433         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
23434         (epiphany_expand_epilogue): Likewise.
23436         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
23437         Don't use CC_FPmode for ORDERED / UNORDERED.
23438         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
23440         * config/epiphany/constraints.md (CnL): New constraint.
23441         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
23442         * config/epiphany/predicates.md (add_operand): Allow 1024.
23444         * config/epiphany/epiphany.md (logical_op): New code iterator.
23445         (op_mnc): New code attribute.
23446         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
23447         (mov_f+1, mov_f+2): New peephole2 patterns.
23449         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
23450         (cstoresi4): Also allow re-use of zero result when doing a NE
23451         comparison to a non-zero operand.
23452         Use (clobber (scratch)) for first insn if the gpr output is not needed.
23454         * config/epiphany/epiphany.md (<insn_opname>v2si3):
23455         Use gen_addsi3_i / gen_subsi3_i.
23457 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23459         PR c++/34949
23460         PR c++/50243
23461         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
23462         contain anything but clobbers, at most one __builtin_stack_restore,
23463         optionally debug stmts and final resx, and if it has at least one
23464         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
23465         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
23466         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
23467         which isn't defaut definition, remove them.
23468         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
23469         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
23470         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
23471         with MEM_REF LHS with SSA_NAME address.
23473 2013-04-08  Jeff Law  <law@redhat.com>
23475         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
23477 2013-04-08  Richard Biener  <rguenther@suse.de>
23479         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
23480         extra newline.
23481         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
23482         determined vector type.
23483         (vect_analyze_data_refs): Likewise.
23484         (vect_get_new_vect_var): Adjust.
23485         (vect_create_destination_var): Preserve SSA name versions.
23486         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
23487         not dump anything here.
23489 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23491         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
23492         Add member lr_slot_known.
23493         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
23494         if necessary.
23495         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
23496         Remove code that sets lr_slot_offset according to what a previous
23497         version of epiphany_emit_save_restore used to do.
23498         (epiphany_emit_save_restore): When doing an lr save or restore,
23499         set/verify lr_slot_known and lr_slot_offset.
23501 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
23503         PR target/54338
23504         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
23505         in ALL_REGS.
23507 2013-04-08  Richard Biener  <rguenther@suse.de>
23509         * alias.c (find_base_term): Fix thinko in previous change.
23511 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23513         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
23514         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
23515         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
23516         if possible to compute val.
23517         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
23518         For QImode integers don't require anything about precision.  Use
23519         const_with_all_bytes_same to find out if the constant doesn't have
23520         repeated bytes in it.
23522 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23524         * config/s390/s390.c (s390_expand_insv): Only accept insertions
23525         within mode size.
23527 2013-04-08  Marek Polacek  <polacek@redhat.com>
23529         PR rtl-optimization/48182
23530         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
23531         value to 1.
23533 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23535         PR target/55487
23536         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
23537         nuses, make sure we have a label.
23539 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23541         PR target/56843
23542         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
23543         (rs6000_emit_swdiv_low_precision): Remove.
23544         (rs6000_emit_swdiv): Rewrite to handle between one and four
23545         iterations of Newton-Raphson generally; modify required number of
23546         iterations for some cases.
23547         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
23549 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
23551         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
23552         set-but-unused variable.
23554         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
23555         basic blocks of released function bodies garbage-collectable.
23557         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
23558         (struct rtl_opt_pass): Add TODO_df_finish.
23560         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
23562 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23564         * config/arm/constraints.md (q): New constraint.
23565         * config/arm/ldrdstrd.md: New file.
23566         * config/arm/arm.md (ldrdstrd.md) New include.
23567         (arm_movdi): Use "q" instead of "r" constraint
23568         for double-word memory access.
23569         (movdf_soft_insn): Likewise.
23570         * config/arm/vfp.md (movdi_vfp): Likewise.
23571         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
23572         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
23573         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
23574         (mem_ok_for_ldrd_strd): Likewise.
23575         (output_move_double): Update assertion.
23577 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23579         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
23581 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23583         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
23584         define_insn_and_split.
23585         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
23587 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23589         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
23590         define_insn_and_split.
23591         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
23592         (shiftsi3_compare): New pattern.
23593         (rrx): New pattern.
23594         * config/arm/unspecs.md (UNSPEC_RRX): New.
23596 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23598         * config/arm/arm.md (negdi_extendsidi): New pattern.
23599         (negdi_zero_extendsidi): Likewise.
23601 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23603         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
23604         define_insn_and_split.
23605         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
23606         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
23608 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23610         * config/arm/arm.md (arm_subdi3): Convert define_insn into
23611         define_insn_and_split.
23612         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
23613         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
23615 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23617         * config/arm/arm.md (subsi3_carryin): New pattern.
23618         (subsi3_carryin_const): Likewise.
23619         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
23620         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
23622 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23624         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
23626 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23628         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
23629         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
23631 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23633         * config/arm/arm.c (arm_expand_builtin): Change fcode
23634         type to unsigned int.
23636 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23638         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
23640 2013-04-04  Ian Lance Taylor  <iant@google.com>
23642         * doc/standards.texi (Standards): The Go frontend supports the Go 1
23643         language standard.
23645 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
23647         PR middle-end/56729
23648         * df-scan.c (df_insn_delete): Disable failing assert.
23650 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23652         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
23653         New function prototype.
23654         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
23655         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
23656         (arm_builtin_vectorized_function): New function.
23658 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23660         * config/arm/arm_neon_builtins.def: New file.
23661         * config/arm/arm.c (neon_builtin_data): Move contents to
23662         arm_neon_builtins.def.
23663         (enum arm_builtins): Include neon builtin definitions.
23664         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
23665         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
23667 2013-04-04  Marek Polacek  <polacek@redhat.com>
23669         PR tree-optimization/48186
23670         * predict.c (maybe_hot_frequency_p): Return false if
23671         HOT_BB_FREQUENCY_FRACTION is 0.
23672         (cgraph_maybe_hot_edge_p): Likewise.
23674 2013-04-04  Richard Biener  <rguenther@suse.de>
23676         PR tree-optimization/56826
23677         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
23678         more accurately.
23680 2013-04-04  Richard Biener  <rguenther@suse.de>
23682         PR tree-optimization/56213
23683         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
23684         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
23686 2013-04-04  Richard Biener  <rguenther@suse.de>
23688         PR tree-optimization/56837
23689         * tree-loop-distribution.c (classify_partition): For non-zero
23690         values require that the value has the same precision as its
23691         mode to be useful as memset value.
23693 2013-04-03  Nick Clifton  <nickc@redhat.com>
23695         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
23696         (fmssf4): Use fmsf.s on E3V5 architectures.
23697         (fnmasf4): Use fnmaf.s on E3V5 architectures.
23698         (fnmssf4): Use fnmsf.s on E3V5 architectures.
23700 2013-04-03  Jeff Law  <law@redhat.com>
23702         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
23703         (lra-eliminations.o): Likewise.
23705 2013-04-03  Teresa Johnson  <tejohnson@google.com>
23707         * gcov-io.c (compute_working_sets): Moved most of body of old
23708         compute_working_sets here from profile.c.
23709         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
23710         (gcov_working_set_t): Moved typedef here from basic-block.h
23711         (compute_working_set): Declare.
23712         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
23713         (get_working_sets): Renamed from compute_working_set,
23714         replace most of body with call to new compute_working_sets.
23715         (get_exec_counts): Replace call to compute_working_sets
23716         to get_working_sets.
23717         * profile.h (get_working_sets): Renamed from compute_working_set.
23718         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
23719         to get_working_sets.
23720         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
23721         * gcov-dump.c (dump_working_sets): New function.
23723 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
23725         * hwint.c (sext_hwi, zext_hwi): New functions.
23726         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
23727         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
23728         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
23729         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
23730         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
23731         (sext_hwi, zext_hwi): New functions.
23733 2013-04-03  Jeff Law  <law@redhat.com>
23735         PR tree-optimization/56799
23736         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
23737         back test for widening conversion erroneously dropped in prior change.
23739 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23741         PR target/56809
23742         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
23743         instead of next_real_insn.
23745 2013-04-03  Marek Polacek  <polacek@redhat.com>
23747         PR sanitizer/55702
23748         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
23750 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23752         PR target/56809
23753         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
23754         next_real_insn.
23755         (thumb1_output_casesi): Likewise.
23756         (thumb2_output_casesi): Likewise.
23758 2013-04-03  Richard Biener  <rguenther@suse.de>
23760         PR tree-optimization/56817
23761         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
23762         Split out ...
23763         (tree_unroll_loops_completely_1): ... new function to manually
23764         walk the loop tree, properly defering outer loops of unrolled
23765         loops to later iterations.
23767 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23769         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
23770         (vectorizable_load): Likewise.
23771         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
23772         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
23774 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23776         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
23777         BIT_FIELD_REF.
23779 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
23781         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
23783 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
23785         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
23787 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23789         PR tree-optimization/56790
23790         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
23791         folding.
23793 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23795         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
23796         Handle VEC_MERGE.
23797         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
23798         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
23799         equal arguments.
23801 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
23803         PR c/19449
23804         * tree.h (force_folding_builtin_constant_p): New decl.
23805         * builtins.c (force_folding_builtin_constant_p): New variable.
23806         (fold_builtin_constant_p): Fold immediately also if
23807         force_folding_builtin_constant_p.
23809 2013-04-03  Richard Biener  <rguenther@suse.de>
23811         PR tree-optimization/56812
23812         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
23813         DRs of the same interleaving chain are independent.
23815 2013-04-02  Jason Merrill  <jason@redhat.com>
23817         * gdbinit.in (pbb): Use debug fn.
23819 2013-04-02  Lawrence Crowl  <crowl@google.com>
23821         * sese.h (struct ivtype_map_elt_s): Remove unused.
23822         (extern debug_ivtype_map): Remove unused.
23823         (extern eq_ivtype_map_elts): Remove unused.
23824         * sese.c (debug_ivtype_map): Removed unused.
23825         (debug_ivtype_map_1): Removed unused.
23826         (debug_ivtype_elt): Remove unused.
23827         (eq_ivtype_map_elts): Remove unused.
23830 2013-04-02  Kai Tietz  <ktietz@redhat.com>
23832         PR target/52790
23833         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
23834         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
23835         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
23836         function.
23837         (legitimize_pe_coff_symbol): Likewise.
23838         (is_imported_p): New helper-function.
23839         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
23840         for Windows x64 targets.
23841         (ix86_expand_prologue): Optimize for pe-coff targets.
23842         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
23843         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
23844         medium/large code-model.
23845         (legitimize_pic_address): Likewise.
23846         (legitimize_tls_address): Likewise.
23847         (ix86_expand_call): Likewise.
23848         (x86_output_mi_thunk): Likewise.
23849         (get_dllimport_decl): Add new beimport argument.
23850         (construct_plt_address): Don't assert for x64 pe-coff targets.
23851         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
23852         targets.
23853         (SYMBOL_FLAG_STUBVAR): New macro.
23854         (SYMBOL_REF_STUBVAR_P): Likewise.
23855         * config/i386/winnt.c (stub_list): New structure.
23856         (stub_head): New local variable.
23857         (i386_pe_record_stub): New function.
23858         (i386_pe_file_end): Emit refptr-stubs.
23860 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
23862         PR rtl-optimization/56745
23863         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
23864         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
23866         PR c++/34949
23867         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
23868         and both of them are MEM_REFs, just compare first argument for
23869         equality and attempt to deal even with differing offsets.
23871         PR c++/34949
23872         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
23873         of gimple_clobber_p to be MEM_REF.
23874         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
23875         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
23876         after gimplification.
23877         * asan.c (get_mem_ref_of_assignment): Don't instrument
23878         gimple_clobber_p stmts.
23879         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
23880         gimple_clobber_p stmt if they have MEM_REF lhs and
23881         are dead because of another gimple_clobber_p stmt.
23882         * tree-ssa-live.c (clear_unused_block_pointer): Treat
23883         gimple_clobber_p stmts like debug stmts.
23884         (remove_unused_locals): Remove clobbers with MEM_REF lhs
23885         that refer to unused VAR_DECLs or uninitialized values.
23886         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
23887         gimple_clobber_p stmts if they refer to removed parameters.
23888         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
23889         formatting.
23891 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
23893         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
23894         using SWI48 mode attribute.
23896 2013-04-02  Wei Mi  <wmi@google.com>
23898         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
23899         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
23900         *<rotate_insn><mode>3_mask in i386.md.
23902 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
23904         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
23906 2013-04-02  Richard Biener  <rguenther@suse.de>
23908         PR tree-optimization/56778
23909         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
23910         Runtime alias tests are not supported for gather loads.
23911         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
23912         stmts referenced from SSA operands before updating SSA form.
23914 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
23915             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23917         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
23918         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
23919         * config/arm/cortex-a53.md: New file.
23920         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
23921         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
23922         * config/arm/arm.c (arm_issue_rate): Likewise.
23923         * config/arm/arm-tune.md: Regenerate
23924         * config/arm/arm-tables.opt: Regenerate.
23925         * config/arm/arm-cores.def: Add cortex-a53.
23927 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
23929         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
23930         non-static link.
23932 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23934         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
23935         scalar load/store operations using B/H registers.
23936         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
23938 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
23940         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
23941         scalar move.
23942         * config/aarch64/aarch64.c
23943         (aarch64_simd_scalar_immediate_valid_for_move): New.
23944         * config/aarch64/aarch64-protos.h
23945         (aarch64_simd_scalar_immediate_valid_for_move): New.
23946         * config/aarch64/constraints.md (Dh, Dq): New.
23947         * config/aarch64/iterators.md (hq): New.
23949 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
23951         * reorg.c (get_branch_condition): Deal with conditional returns.
23952         (fill_simple_delay_slots): Remove dead code dealing with jumps.
23954 2013-04-01  Wei Mi  <wmi@google.com>
23956         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
23957         Truncate operand 2 using %b asm operand modifier.
23958         (*<shift_insn><mode>3_mask): Ditto.
23959         (*<rotate_insn><mode>3_mask): Ditto.
23961 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
23963         PR middle-end/56798
23964         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
23966 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
23968         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
23969         of next_real_insn.
23970         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
23972 2013-03-30  Lawrence Crowl  <crowl@google.com>
23974         * dse.c (clear_alias_sets): Remove never set.
23975         (disqualified_clear_alias_sets): Remove never set.
23976         (clear_alias_mode_pool): Remove never set.
23977         (dse_step0): Remove condition that is never true.
23978         (canon_address): Remove condition that is never true.
23979         (dse_step7): Remove condition that is never true.
23980         (rest_of_handle_dse): Remove condition that is never true.
23981         (rest_of_handle_dse::did_global): Remove never read from above.
23982         (dse_step2_spill): Remove never called from above.
23983         (dse_step5_spill): Remove never called from above.
23985 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
23987         * doc/md.texi (Standard Names) <casesi>: Update documentation for
23988         JUMP_TABLE_DATA changes.
23989         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
23990         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
23991         (Insns) <jump_table_data>: New entry.
23992         * doc/tm.texi: Regenerate.
23994         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
23996         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
23997         for table jump at the end of a basic block using tablejump_p.
23998         * targhooks.c (default_invalid_within_doloop): Likewise.
23999         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
24000         target hook implementation that is identical to the default hook.
24001         (rs6000_invalid_within_doloop): Remove.
24003         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
24004         unused variable from tablejump_p call.
24006         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
24007         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
24008         (INSN_DELETED_P): Likewise.
24009         (emit_jump_table_data): New prototype.
24010         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
24011         after 4th as unused.
24012         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
24013         * sched-vis.c (print_insn): Likewise.
24014         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
24015         insn for compatibility with back ends that use next_active_insn to
24016         identify jump table data.
24017         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
24018         (remove_insn): Likewise.
24019         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
24020         to be emitted.
24021         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
24022         (emit_jump_table_data): New function.
24024         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
24025         basic block, a JUMP_TABLE_DATA never is.
24026         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
24027         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
24028         off from code handling real insns.
24029         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
24030         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
24031         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
24032         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
24033         is not a NONDEBUG_INSN_P.
24034         * ira-costs.c (scan_one_insn): Likewise.
24035         * jump.c (mark_all_labels): Likewise.
24036         (mark_jump_label_1): Likewise.
24037         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24038         * lra.c (get_insn_freq): Expect all insns reaching here to be in
24039         a basic block.
24040         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
24041         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
24042         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
24043         JUMP_TABLE_DATA_P insns.
24044         (calculate_elim_costs_all_insns): Likewise.
24045         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
24046         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
24047         (delete_output_reload): Code style fixups.
24048         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
24049         insn flags on this non-insn.
24050         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
24051         as scheduling barriers, for pre-change compatibility.
24052         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
24053         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
24055         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
24056         redundant JUMP_TABLE_DATA_P test.
24057         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
24058         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24059         (frv_for_each_packet): Likewise.
24060         * config/i386/i386.c (min_insn_size): Likewise.
24061         (ix86_avoid_jump_mispredicts): Likewise.
24062         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24063         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
24064         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24065         (mips16_insn_length): Robustify.
24066         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
24067         (mips16_split_long_branches): Likewise.
24068         * config/pa/pa.c (pa_combine_instructions): Likewise.
24069         * config/rs6000/rs6000.c (get_next_active_insn): Treat
24070         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
24071         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
24072         as contributing to pool range lengths.
24073         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
24074         Remove redundant JUMP_TABLE_DATA_P test.
24075         (sh_loop_align): Likewise.
24076         (split_branches): Likewise.
24077         (sh_insn_length_adjustment): Likewise.
24078         * config/spu/spu.c (get_branch_target): Likewise.
24080 2013-03-29  Jan Hubicka  <jh@suse.cz>
24082         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
24083         gcov streaming; stream hot bb threshold to ltrans.
24084         * predict.c (get_hot_bb_threshold): Break out from ....
24085         (maybe_hot_count_p): ... here.
24086         (set_hot_bb_threshold): New function.
24087         * lto-section-in.c (lto_section_name): Add profile.
24088         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
24089         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
24090         and data-streamer.h
24091         (histogram_entry): New structure.
24092         (histogram, histogram_pool): New global vars.
24093         (histogram_hash): New structure.
24094         (histogram_hash::hash): New method.
24095         (histogram_hash::equal): Likewise.
24096         (account_time_size): New function.
24097         (cmp_counts): New function.
24098         (dump_histogram): New function.
24099         (ipa_profile_generate_summary): New function.
24100         (ipa_profile_write_summary): New function.
24101         (ipa_profile_read_summary): New function.
24102         (ipa_profile): Decide on threshold.
24103         (pass_ipa_profile): Add ipa_profile_write_summary and
24104         ipa_profile_read_summary.
24105         * Makefile.in (ipa.o): Update dependencies.
24106         * lto-streamer.h (LTO_section_ipa_profile): New section.
24108 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24110         * tree.h (VAR_P): New.
24112 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
24114         PR lto/56777
24115         * doc/invoke.texi ([-fwhole-program]): Fix typo.
24117 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
24119         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
24120         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24121         (control_flow_insn_p): Likewise.
24122         * cfgrtl.c (duplicate_insn_chain): Likewise.
24123         * final.c (get_attr_length_1): Likewise.
24124         (shorten_branches): Likewise.
24125         (final_scan_insn): Likewise.
24126         * function.c (instantiate_virtual_regs): Likewise.
24127         * gcse.c (insert_insn_end_basic_block): Likewise.
24128         * ira-costs.c (scan_one_insn): Likewise.
24129         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24130         * lra.c (check_rtl): Likewise.
24131         * reload1.c (elimination_costs_in_insn): Likewise.
24132         * reorg.c (follow_jumps): Likewise.
24134         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
24135         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24136         (thumb_far_jump_used_p): Likewise.
24137         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
24138         (workaround_speculation): Likewise.
24139         (add_sched_insns_for_speculation): Likewise.
24140         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
24141         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24142         (frv_for_each_packet): Likewise.
24143         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
24144         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
24145         (final_emit_insn_group_barriers): Likewise.
24146         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24147         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24148         (mips16_insn_length): Likewise.
24149         * config/pa/pa.c (pa_reorg): Likewise.
24150         (pa_combine_instructions): Likewise.
24151         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
24152         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
24153         (sh_reorg): Likewise.
24154         (split_branches): Likewise.
24155         * config/spu/spu.c (get_branch_target): Likewise.
24157         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
24158         JUMP_TABLE_DATA_P.
24160 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
24162         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
24163         Fix declaration name.
24165 2013-03-28  Lawrence Crowl  <crowl@google.com>
24167         * graphds.h (struct graph.indicies): Remove unused.
24168         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
24169         (SCOP_ORIGINAL_PDDRS): Remove unused.
24170         * sese.h (extern insert_loop_close_phis): Removed unused.
24171         (extern insert_guard_phis): Removed unused.
24172         (extern ivtype_map_elt_info): Removed unused.
24173         (new_ivtype_map_elt): Removed unused.
24174         * sese.c (ivtype_map_elt_info): Removed unused.
24176 2013-03-28  Lawrence Crowl  <crowl@google.com>
24178         * Makefile.in: Add several missing include dependences.
24179         (DUMPFILE_H): New.
24180         (test-dump.o): New.  This object is not added to any executable,
24181         but is present for ad-hoc testing.
24182         * bitmap.c
24183         (debug (const bitmap_head_def &)): New.
24184         (debug (const bitmap_head_def *)): New.
24185         * bitmap.h
24186         (extern debug (const bitmap_head_def &)): New.
24187         (extern debug (const bitmap_head_def *)): New.
24188         * cfg.c
24189         (debug (edge_def &)): New.
24190         (debug (edge_def *)): New.
24191         * cfghooks.c
24192         (debug (basic_block_def &)): New.
24193         (debug (basic_block_def *)): New.
24194         * dumpfile.h
24195         (dump_node (const_tree, int, FILE *)): Correct source file.
24196         * dwarf2out.c
24197         (debug (die_struct &)): New.
24198         (debug (die_struct *)): New.
24199         * dwarf2out.h
24200         (extern debug (die_struct &)): New.
24201         (extern debug (die_struct *)): New.
24202         * gimple-pretty-print.c
24203         (debug (gimple_statement_d &)): New.
24204         (debug (gimple_statement_d *)): New.
24205         * gimple-pretty-print.h
24206         (extern debug (gimple_statement_d &)): New.
24207         (extern debug (gimple_statement_d *)): New.
24208         * ira-build.c
24209         (debug (ira_allocno_copy &)): New.
24210         (debug (ira_allocno_copy *)): New.
24211         (debug (ira_allocno &)): New.
24212         (debug (ira_allocno *)): New.
24213         * ira-int.h
24214         (extern debug (ira_allocno_copy &)): New.
24215         (extern debug (ira_allocno_copy *)): New.
24216         (extern debug (ira_allocno &)): New.
24217         (extern debug (ira_allocno *)): New.
24218         * ira-lives.c
24219         (debug (live_range &)): New.
24220         (debug (live_range *)): New.
24221         * lra-int.h
24222         (debug (lra_live_range &)): New.
24223         (debug (lra_live_range *)): New.
24224         * lra-lives.c
24225         (debug (lra_live_range &)): New.
24226         (debug (lra_live_range *)): New.
24227         * omega.c
24228         (debug (omega_pb_d &)): New.
24229         (debug (omega_pb_d *)): New.
24230         * omega.h
24231         (extern debug (omega_pb_d &)): New.
24232         (extern debug (omega_pb_d *)): New.
24233         * print-rtl.c
24234         (debug (const rtx_def &)): New.
24235         (debug (const rtx_def *)): New.
24236         * print-tree.c
24237         (debug_tree (tree): Move within file.
24238         (debug_raw (const tree_node &)): New.
24239         (debug_raw (const tree_node *)): New.
24240         (dump_tree_via_hooks (const tree_node *, int)): New.
24241         (debug (const tree_node &)): New.
24242         (debug (const tree_node *)): New.
24243         (debug_verbose (const tree_node &)): New.
24244         (debug_verbose (const tree_node *)): New.
24245         (debug_head (const tree_node &)): New.
24246         (debug_head (const tree_node *)): New.
24247         (debug_body (const tree_node &)): New.
24248         (debug_body (const tree_node *)): New.
24249         (debug_vec_tree (tree): Move and reimplement in terms of dump.
24250         (debug (vec<tree, va_gc> &)): New.
24251         (debug (vec<tree, va_gc> *)): New.
24252         * rtl.h
24253         (extern debug (const rtx_def &)): New.
24254         (extern debug (const rtx_def *)): New.
24255         * sbitmap.c
24256         (debug_raw (simple_bitmap_def &)): New.
24257         (debug_raw (simple_bitmap_def *)): New.
24258         (debug (simple_bitmap_def &)): New.
24259         (debug (simple_bitmap_def *)): New.
24260         * sbitmap.h
24261         (extern debug (simple_bitmap_def &)): New.
24262         (extern debug (simple_bitmap_def *)): New.
24263         (extern debug_raw (simple_bitmap_def &)): New.
24264         (extern debug_raw (simple_bitmap_def *)): New.
24265         * sel-sched-dump.c
24266         (debug (vinsn_def &)): New.
24267         (debug (vinsn_def *)): New.
24268         (debug_verbose (vinsn_def &)): New.
24269         (debug_verbose (vinsn_def *)): New.
24270         (debug (expr_def &)): New.
24271         (debug (expr_def *)): New.
24272         (debug_verbose (expr_def &)): New.
24273         (debug_verbose (expr_def *)): New.
24274         (debug (vec<rtx> &)): New.
24275         (debug (vec<rtx> *)): New.
24276         * sel-sched-dump.h
24277         (extern debug (vinsn_def &)): New.
24278         (extern debug (vinsn_def *)): New.
24279         (extern debug_verbose (vinsn_def &)): New.
24280         (extern debug_verbose (vinsn_def *)): New.
24281         (extern debug (expr_def &)): New.
24282         (extern debug (expr_def *)): New.
24283         (extern debug_verbose (expr_def &)): New.
24284         (extern debug_verbose (expr_def *)): New.
24285         (extern debug (vec<rtx> &)): New.
24286         (extern debug (vec<rtx> *)): New.
24287         * sel-sched-ir.h
24288         (_list_iter_cond_expr): Make inline instead of static.
24289         * sreal.c
24290         (debug (sreal &)): New.
24291         (debug (sreal *)): New.
24292         * sreal.h
24293         (extern debug (sreal &)): New.
24294         (extern debug (sreal *)): New.
24295         * tree.h
24296         (extern debug_raw (const tree_node &)): New.
24297         (extern debug_raw (const tree_node *)): New.
24298         (extern debug (const tree_node &)): New.
24299         (extern debug (const tree_node *)): New.
24300         (extern debug_verbose (const tree_node &)): New.
24301         (extern debug_verbose (const tree_node *)): New.
24302         (extern debug_head (const tree_node &)): New.
24303         (extern debug_head (const tree_node *)): New.
24304         (extern debug_body (const tree_node &)): New.
24305         (extern debug_body (const tree_node *)): New.
24306         (extern debug (vec<tree, va_gc> &)): New.
24307         (extern debug (vec<tree, va_gc> *)): New.
24308         * tree-cfg.c
24309         (debug (struct loop &)): New.
24310         (debug (struct loop *)): New.
24311         (debug_verbose (struct loop &)): New.
24312         (debug_verbose (struct loop *)): New.
24313         * tree-dump.c: Add header dependence.
24314         * tree-flow.h
24315         (extern debug (struct loop &)): New.
24316         (extern debug (struct loop *)): New.
24317         (extern debug_verbose (struct loop &)): New.
24318         (extern debug_verbose (struct loop *)): New.
24319         * tree-data-ref.c
24320         (debug (data_reference &)): New.
24321         (debug (data_reference *)): New.
24322         (debug (vec<data_reference_p> &)): New.
24323         (debug (vec<data_reference_p> *)): New.
24324         (debug (vec<ddr_p> &)): New.
24325         (debug (vec<ddr_p> *)): New.
24326         * tree-data-ref.h
24327         (extern debug (data_reference &)): New.
24328         (extern debug (data_reference *)): New.
24329         (extern debug (vec<data_reference_p> &)): New.
24330         (extern debug (vec<data_reference_p> *)): New.
24331         (extern debug (vec<ddr_p> &)): New.
24332         (extern debug (vec<ddr_p> *)): New.
24333         * tree-ssa-alias.c
24334         (debug (pt_solution &)): New.
24335         (debug (pt_solution *)): New.
24336         * tree-ssa-alias.h
24337         (extern debug (pt_solution &)): New.
24338         (extern debug (pt_solution *)): New.
24339         * tree-ssa-alias.c
24340         (debug (_var_map &)): New.
24341         (debug (_var_map *)): New.
24342         (debug (tree_live_info_d &)): New.
24343         (debug (tree_live_info_d *)): New.
24344         * tree-ssa-alias.h
24345         (extern debug (_var_map &)): New.
24346         (extern debug (_var_map *)): New.
24347         (extern debug (tree_live_info_d &)): New.
24348         (extern debug (tree_live_info_d *)): New.
24350 2013-03-28  Jan Hubicka  <jh@suse.cz>
24352         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
24354 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
24356         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
24357         record only when desired or required.
24359 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
24361         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
24362         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
24364 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24366         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
24367         (*andsi3_compare0_uxtw): New pattern.
24368         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
24369         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
24371 2013-03-28  Jan Hubicka  <jh@suse.cz>
24373         * data-streamer-in.c (streamer_read_gcov_count): New function.
24374         * gimple-streamer-out.c: Include value-prof.h.
24375         (output_gimple_stmt): Output histogram.
24376         (output_bb): Use streamer_write_gcov_count.
24377         * value-prof.c: Include data-streamer.h
24378         (dump_histogram_value): Add HIST_TYPE_MAX.
24379         (stream_out_histogram_value): New function.
24380         (stream_in_histogram_value): New function.
24381         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
24382         (stream_out_histogram_value, stream_in_histogram_value): Declare.
24383         * data-streamer-out.c (streamer_write_gcov_count): New function.
24384         (streamer_write_gcov_count_stream): New function.
24385         * lto-cgraph.c (lto_output_edge): Update counter streaming.
24386         (lto_output_node): Likewise.
24387         (input_node, input_edge): Likewise.
24388         * lto-streamer-out.c (output_cfg): Update streaming.
24389         * lto-streamer-in.c (input_cfg): Likewise.
24390         * data-streamer.h (streamer_write_gcov_count,
24391         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
24392         * gimple-streamer-in.c: Include value-prof.h
24393         (input_gimple_stmt): Input histograms.
24394         (input_bb): Update profile streaming.
24396 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24398         * genmodes.c (emit_max_int): New function.
24399         (emit_insn_modes_h): Added call to emit_max_function.
24400         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
24401         Added doc.
24402         * machmode.def: Fixed comment.
24404 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24406         * combine.c (try_combine): Removed useless assert.
24407         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
24409 2013-03-28  Marek Polacek  <polacek@redhat.com>
24410             Richard Biener  <rguenther@suse.de>
24412         PR tree-optimization/56695
24413         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
24414         build signed result of a vector comparison.
24415         * tree-cfg.c (verify_gimple_comparison): Check that a result
24416         of a vector comparison has signed type.
24418 2013-03-28  Richard Biener  <rguenther@suse.de>
24420         PR tree-optimization/37021
24421         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
24422         do not restrict gaps between groups.
24423         * tree-vect-stmts.c (vectorizable_load): Properly account for
24424         a gap between groups.
24426 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
24428         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
24429         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
24430         is not enabled.
24432 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
24434         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
24435         * doc/extend.texi (Named Address Spaces): Ditto.
24436         (Variable Attributes): Ditto.
24438 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24440         * config.build: Add support for cygwin x64 target.
24441         * config.gcc: Likewise.
24442         * config.host: Likewise.
24443         * configure.ac: Likewise
24444         * configure: Regenerated.
24446 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24448         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
24449         * config/i386/t-cygwin-w64: New file.
24450         * config/i386/cygwin-w64.h: New file.
24451         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
24452         and add support for x64-cygwin target.
24453         (CPP_SPEC): Likewise.
24454         (CXX_WRAP_SPEC_LIST): Undefine before define.
24455         (LIBGCJ_SONAME): Use 15 as version.
24457 2013-03-27  Richard Biener  <rguenther@suse.de>
24459         PR tree-optimization/56716
24460         * tree-ssa-structalias.c (perform_var_substitution): Adjust
24461         dumping for ref nodes.
24463 2013-03-27  Martin Jambor  <mjambor@suse.cz>
24465         PR tree-optimization/55334
24466         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
24467         restricted pointers to arrays.
24469 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24471         * Makefile.in (.SUFFIXES): Add .cc.
24472         (.c.o): Apply same recipe for implicit rule .cc.o.
24474 2013-03-27  Richard Biener  <rguenther@suse.de>
24476         PR tree-optimization/37021
24477         * tree-vect-data-refs.c (vect_check_strided_load): Allow
24478         REALPART/IMAGPART_EXPRs around the supported refs.
24479         * tree-ssa-structalias.c (find_func_aliases): Assume that
24480         floating-point values are not used to transfer pointers.
24482 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
24484         * target.def (TARGET_HAS_IFUNC_P): New target hook.
24485         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
24486         * doc/tm.texi: Regenerate.
24487         * targhooks.h (default_has_ifunc_p): New.
24488         * targhooks.c (default_has_ifunc_p): Ditto.
24489         * config/linux-protos.h: New file.
24490         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
24491         hook for linux which disables support of indirect functions in android.
24492         * config/linux-android.c: New file.
24493         * config/t-linux-android.c: Ditto.
24494         * config.gcc: Added new object file linux-android.o.
24495         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
24496         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
24497         * varasm.c (do_assemble_alias): Likewise.
24498         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
24499         doesn't support indirect functions.
24500         * configure: Regenerate.
24502 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
24504         PR target/56102
24505         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
24506         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
24507         mult-word mode.
24509 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24511         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
24513 2013-03-27  Terry Guo  <terry.guo@arm.com>
24515         * config/arm/arm-cores.def: Added core cortex-r7.
24516         * config/arm/arm-tune.md: Regenerated.
24517         * config/arm/arm-tables.opt: Regenerated.
24518         * doc/invoke.texi: Added entry for core cortex-r7.
24520 2013-03-27  Walter Lee  <walt@tilera.com>
24522         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
24523         double-decrement of next_scratch_regno.
24525 2013-03-27  Walter Lee  <walt@tilera.com>
24527         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
24528         input operands.
24529         (insn_v1mulus): Ditto.
24530         (insn_v2muls): Ditto.
24532 2013-03-27  Walter Lee  <walt@tilera.com>
24534         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
24535         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
24537 2013-03-27  Walter Lee  <walt@tilera.com>
24539         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
24540         (*sibcall_value): Ditto.
24542 2013-03-27  Walter Lee  <walt@tilera.com>
24544         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
24545         (insn_mnz_v8qi): ... this ...
24546         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
24547         vector equivalent.
24548         (insn_v<n>mnz): Replaced by ...
24549         (insn_v1mnz): ... this ...
24550         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
24551         equivalent.
24552         (insn_mz_<mode>): Replaced by ...
24553         (insn_mz_v8qi): ... this ...
24554         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
24555         vector equivalent.
24556         (insn_v<n>mz): Replaced by ...
24557         (insn_v1mz): ... this ...
24558         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
24559         equivalent.
24561 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
24563         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
24565 2013-03-26  Roland McGrath  <mcgrathr@google.com>
24567         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
24568         than fprintf with a non-constant, non-format string.
24570 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24572         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
24573         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
24574         operand 0 predicate.
24575         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
24576         attribute.  Use general_x64nomem_operand as operand 1 predicate.
24577         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
24578         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24579         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
24580         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24581         (mov<mode>_insv_1): Remove expander.  Merge insn with
24582         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
24583         Use general_x64nomem_operand as operand 1 predicate.
24584         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
24585         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
24586         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24587         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
24588         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
24589         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
24590         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24591         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
24592         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
24593         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
24594         (general_x64nomem_operand): Ditto.
24596 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
24598         * config/rtems.opt: Add -pthread option.
24600 2013-03-26  Richard Biener  <rguenther@suse.de>
24602         * alias.c (find_base_term): Avoid redundant and not used recursion.
24603         (base_alias_check): Get the initial base term from the caller.
24604         (true_dependence_1): Compute and pass base terms to base_alias_check.
24605         (write_dependence_p): Likewise.
24606         (may_alias_p): Likewise.
24608 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
24610         * config/aarch64/aarch64.c (aarch64_classify_address): Support
24611         PC-relative load in SI modes and above only.
24613 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
24615         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
24616         * config/arm/iwmmxt.md (WCGR0): Update.
24617         (WCGR1, WCGR2, WCGR3): Likewise.
24619 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24621         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
24622         Use x64 and nox64 isa attributes.
24624 2013-03-26  Richard Biener  <rguenther@suse.de>
24626         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
24627         alignment computations and rely on get_object_alignment_1
24628         for the !TYPE_P case.
24629         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
24631 2013-03-26  Walter Lee  <walt@tilera.com>
24633         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
24634         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
24636 2013-03-25  Jeff Law  <law@redhat.com>
24638         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
24639         check for INTEGRAL_TYPE_P that was missing due to checking in
24640         wrong version of prior patch.
24642 2013-03-25  Walter Lee  <walt@tilera.com>
24644         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
24645         TILEGX_INSN_SHUFFLEBYTES1.
24646         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
24647         shufflebytes1.
24648         (tilegx_builtins): Ditto.
24649         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
24651 2013-03-25  Walter Lee  <walt@tilera.com>
24653         * config/tilegx/tilegx.md (floatsisf2): New pattern.
24654         (floatunssisf2): New pattern.
24655         (floatsidf2): New pattern.
24656         (floatunssidf2): New pattern.
24658 2013-03-25  Walter Lee  <walt@tilera.com>
24660         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
24661         tests for constraint J, K, N, P.
24663 2013-03-25  Walter Lee  <walt@tilera.com>
24665         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
24666         Use indirect/pcrel encoding.
24667         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
24668         Ditto.
24670 2013-03-25  Steve Ellcey  <sellcey@mips.com>
24672         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
24673         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
24674         * config/mips/mips.c (mips_option_override): Set IMADD default.
24675         * config/mips/mips.h (PTF_AVOID_IMADD): New.
24676         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
24677         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
24678         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
24679         * doc/invoke.texi (-mimadd/-mno-imadd): New.
24681 2013-03-25  Jeff Law  <law@redhat.com>
24683         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
24684         slightly to avoid creating and folding useless trees.  Simplify
24685         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
24687 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
24689         * config/i386/i386.md (*zero_extendsidi2): Merge with
24690         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
24691         * config/i386/predicates.md (x86_64_zext_operand): Rename from
24692         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
24693         targets.  Clarify comment.
24695 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24697         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
24698         pass-through jump functions differently.
24699         (ipa_read_jump_function): Likewise.  Also use setter functions to set
24700         up jump functions.
24702 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24704         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
24705         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
24706         process it.
24707         (ipa_get_indirect_edge_target): New function.
24708         (devirtualization_time_bonus): New parameter known_aggs, pass it to
24709         ipa_get_indirect_edge_target.  Update all callers.
24710         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
24711         ipa_get_indirect_edge_target_1 instead of calling
24712         ipa_get_indirect_edge_target.
24713         (create_specialized_node): Pass aggvlas to
24714         ipcp_discover_new_direct_edges.
24716 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24718         * config/arm/arm.md (f_sels, f_seld): New types.
24719         (*cmov<mode>): New pattern.
24720         * config/arm/predicates.md (arm_vsel_comparison_operator): New
24721         predicate.
24723 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24725         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
24726         POSIX-printf for mingw-hosted  builds.
24728 2013-03-25  Richard Biener  <rguenther@suse.de>
24730         PR middle-end/56694
24731         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
24732         must-not-throw stmt location.
24734 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24736         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
24737         Emit load-acquire versions when acq is true.
24738         (arm_emit_store_exclusive): Add rel parameter.
24739         Emit store-release versions when rel is true.
24740         (arm_split_compare_and_swap): Use acquire-release instructions
24741         instead.
24742         of barriers when appropriate.
24743         (arm_split_atomic_op): Likewise.
24744         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
24745         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
24746         (VUNSPEC_SLX): Likewise.
24747         (VUNSPEC_LDA): Likewise.
24748         (VUNSPEC_STL): Likewise.
24749         * config/arm/sync.md (atomic_load<mode>): New pattern.
24750         (atomic_store<mode>): Likewise.
24751         (arm_load_acquire_exclusive<mode>): Likewise.
24752         (arm_load_acquire_exclusivesi): Likewise.
24753         (arm_load_acquire_exclusivedi): Likewise.
24754         (arm_store_release_exclusive<mode>): Likewise.
24756 2013-03-25  Catherine Moore  <clm@codesourcery.com>
24758         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
24759         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
24760         * config/mip/predicates.md (lwsp_swsp_operand,
24761         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
24762         sb16_operand, db4_operand, db7_operand, ib3_operand,
24763         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
24764         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
24765         andi16_operand): New predicates.
24766         * config/mips/mips.md (compression): New attribute.
24767         (enabled): New attribute.
24768         (length): Consider compression in computing length.
24769         (shift_compression): New code attribute.
24770         (*add<mode>3): New operands. Record compression.
24771         (sub<mode>3): Likewise.
24772         (one_cmpl<mode>2): Likewise.
24773         (*and<mode>3): Likewise.
24774         (*ior<mode>3): Likewise.
24775         (unnamed pattern for xor): Likewise.
24776         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
24777         (*<optab><mode>3): Likewise.
24778         (*mov<mode>_internal: Likewise.
24779         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
24780         (mips_unsigned_immediate_p): New.
24781         (umips_lwsp_swsp_address_p): New.
24782         (m16_based_address_p): New.
24783         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
24784         (mips_unsigned_immediate_p): New prototype.
24785         (lwsp_swsp_address_p): New prototype.
24786         (m16_based_address_p): New prototype.
24787         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
24788         (mips_signed_immediate_p): New function.
24789         (m16_based_address_p): New function.
24790         (lwsp_swsp_address_p): New function.
24791         (mips_print_operand_punctuation): Recognize short delay slot insns
24792         for microMIPS.add<mode>3"
24794 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24796         PR target/56720
24797         * config/arm/iterators.md (v_cmp_result): New mode attribute.
24798         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
24800 2013-03-25  Richard Biener  <rguenther@suse.de>
24802         PR tree-optimization/56689
24803         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
24804         any edge.
24806 2013-03-25  Richard Biener  <rguenther@suse.de>
24808         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
24809         of bitmap.
24810         (memory_references): Likewise.
24811         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
24812         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
24813         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
24814         (gather_mem_refs_in_loops): Fold into ...
24815         (analyze_memory_references): ... this.  Move initialization
24816         to tree_ssa_lim_initialize.
24817         (fill_always_executed_in): Rename to ...
24818         (fill_always_executed_in_1): ... this.
24819         (fill_always_executed_in): Move contains_call computation to
24820         this new function from ...
24821         (tree_ssa_lim_initialize): ... here.
24822         (tree_ssa_lim): Call fill_always_executed_in.
24824 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24826         * postreload.c (reload_combine): Fix code detecting returns.
24828 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24830         * function.c (emit_use_return_register_into_block): On cc0 targets,
24831         do not emit the sequence between cc0 setter and user.
24833 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24835         * config/i386/predicates.md (local_symbolic_operand): Interpret
24836         dll-imported symbols as none-local.
24838 2013-03-25  Richard Biener  <rguenther@suse.de>
24840         * tree-ssa-loop-im.c (struct depend): Remove.
24841         (struct lim_aux_data): Make depends a vec of gimples.
24842         (free_lim_aux_data): Adjust.
24843         (add_dependency): Likewise.
24844         (set_level): Likewise.
24846 2013-03-25  Richard Biener  <rguenther@suse.de>
24848         PR middle-end/56434
24849         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
24850         the pointer returned by calls with ECF_MALLOC set.
24852 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24854         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
24856 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24858         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
24859         using MMXMODE mode iterator.
24860         (*move<mode>_internal): Merge with *movv2sf_internal and
24861         *movv2sf_internal_rex64 using MMXMODE mode iterator.
24863 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24865         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
24866         (record_last_mem_set_info): Likewise.
24868         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
24869         of XNEWVEC followed by memset.
24870         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
24872 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24874         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
24875         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
24876         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
24877         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
24878         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
24879         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
24880         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
24881         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
24882         BARRIER_P instead of GET_CODE.
24884 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24886         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
24887         inaccuracy in the probing code.
24889         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
24890         (ctrapdi4): Likewise.
24892 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24894         * calls.c (expand_call): Add missing guard to code handling return
24895         of non-BLKmode structures in MSB.
24896         * function.c (expand_function_end): Likewise.
24898 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24900         * combine.c (try_combine): Adjust comment.  Do not add the set of
24901         insn #0 if the destination indirectly is set or dies in insn #2.
24902         Tidy up code to distribute a new note.
24904 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24906         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
24907         also for alternatives 16 and 17.
24909 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24911         * config/i386/sse.md (*mov<mode>_internal): Merge with
24912         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
24913         Emit insn template depending on type attribute.  Use
24914         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24915         movd instead of movq mnemonic for interunit moves.  Rewrite mode
24916         attribute calculation.  Remove unit attribute calculation.
24917         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
24918         Set prefix_data16 attribute for DImode ssemov types.
24919         Use Ym instead of y for SSE-MMX conversion alternatives.
24920         Reorder operand constraints.
24922 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
24924         * df.h (df_insn_delete): Adjust prototype.
24925         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
24926         and let it decide whether mark the basic block dirty.
24927         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
24928         * df-scan.c (df_insn_info_delete): New helper function, split
24929         off from df_insn_delete.
24930         (df_scan_free_bb_info): Use it.
24931         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
24932         Likewise.
24933         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
24934         that the insn is actually an insn and it has a non-NULL basic block.
24935         Do not mark basic block dirty if only deleting a DEBUG_INSN.
24937 2013-03-22  Richard Biener  <rguenther@suse.de>
24939         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
24940         dep_ref members.
24941         (mem_ref_alloc): Do not allocate them.
24942         (refs_independent_p): Do not query or maintain a cache.
24944 2013-03-22  Richard Biener  <rguenther@suse.de>
24946         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
24947         (gather_mem_refs_in_loops): Do not compute it.
24948         (analyze_memory_references): Do not allocate it.
24949         (tree_ssa_lim_finalize): Do not free it.
24950         (for_all_locs_in_loop): Do not query all_refs_in_loop.
24952 2013-03-22  Richard Biener  <rguenther@suse.de>
24954         * is-a.h (as_a): Use gcc_checking_assert.
24956 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
24958         * config/aarch64/aarch64.c (aarch64_print_operand): New
24959         format specifier for printing a constant in hex.
24960         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
24961         format specifier for printing second operand.
24963 2013-03-22  Richard Biener  <rguenther@suse.de>
24965         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
24966         bitmaps.
24967         (gather_mem_refs_in_loops): Perform store accumulation here.
24968         (create_vop_ref_mapping_loop): Remove.
24969         (create_vop_ref_mapping): Likewise.
24970         (analyze_memory_references): Initialize refs_stored_in_loop.
24971         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
24972         (record_indep_loop): Remove.
24973         (record_dep_loop): New function.
24974         (ref_indep_loop_p_1): Adjust to only walk over references
24975         in the loop, not its subloops.
24976         (ref_indep_loop_p): Rename to ...
24977         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
24978         maintaining a more fine-grained cache.
24979         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
24980         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
24982 2013-03-22  Richard Biener  <rguenther@suse.de>
24984         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
24985         (struct mem_ref): Make accesses_in_loop a vec of a vec of
24986         aggregate mem_ref_loc.
24987         (free_mem_ref_locs): Inline into ...
24988         (memref_free): ... this and adjust.
24989         (mem_ref_alloc): Adjust.
24990         (mem_ref_locs_alloc): Remove.
24991         (record_mem_ref_loc): Adjust.
24992         (get_all_locs_in_loop): Rewrite into ...
24993         (for_all_locs_in_loop): ... this iterator.
24994         (rewrite_mem_ref_loc): New functor.
24995         (rewrite_mem_refs): Use for_all_locs_in_loop.
24996         (sm_set_flag_if_changed): New functor.
24997         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
24998         (ref_always_accessed): New functor.
24999         (ref_always_accessed_p): Use for_all_locs_in_loop.
25001 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25003         * tree-pass.h (PROP_gimple_lvec): New.
25004         * passes.c (dump_properties): Handle PROP_gimple_lvec.
25005         (init_optimization_passes): Move pass_lower_vector.
25006         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
25007         PROP_gimple_lvec.
25008         (pass_lower_vector): Provide PROP_gimple_lvec.
25009         (pass_lower_vector_ssa): Likewise.
25010         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
25012 2013-03-21  Mark Wielaard  <mjw@redhat.com>
25014         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
25016 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25018         * config/i386/i386.md (*movdi_internal): Disparage slightly
25019         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
25020         conversion alternatives.
25022 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
25024         PR middle-end/48087
25025         * diagnostic.def (DK_WERROR): New kind.
25026         * diagnostic.h (werrorcount): Define.
25027         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
25028         promoted to DK_ERROR, increment DK_WERROR counter instead of
25029         DK_ERROR counter.
25030         * toplev.c (toplev_main): Call print_ignored_options even if
25031         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
25032         even if just werrorcount is non-zero.
25034         PR debug/55608
25035         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
25036         on failure.
25037         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
25038         (string_cst_pool_decl): New function.
25039         (optimize_one_addr_into_implicit_ptr): New function.
25040         (resolve_addr_in_expr): Optimize DWARF location expression
25041         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
25042         which doesn't live in memory, but has DW_AT_location or
25043         DW_AT_const_value, or refers to a string literal, into
25044         DW_OP_GNU_implicit_pointer.
25045         (optimize_location_into_implicit_ptr): New function.
25046         (resolve_addr): If removing DW_AT_location of a variable because
25047         it was DW_OP_addr of address of the variable, but the variable doesn't
25048         live in memory, try to emit const value attribute for the initializer.
25050 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25052         * tree.h (VECTOR_TYPE_P): New macro.
25053         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
25054         TYPE_MODE): Use it.
25055         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
25056         VEC_COND_EXPR cannot be lvalues.
25057         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
25059 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25061         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
25062         Restrict the transformation to equal modes.
25064 2013-03-21  Richard Biener  <rguenther@suse.de>
25066         PR tree-optimization/39326
25067         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
25068         (MEM_ANALYZABLE): Adjust.
25069         (record_mem_ref_loc): Move bitmap ops ...
25070         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
25071         unanalyzable refs, do not record locations for it.
25072         (analyze_memory_references): Allocate ref zero as shared
25073         unanalyzable ref.
25074         (refs_independent_p): Do not test for unanalyzed mems here.
25075         (ref_indep_loop_p_1): Special-case disambiguation against
25076         the unanalyzed ref.
25077         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
25079 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
25081         * config/arm/arm-protos.h (tune_params): Add
25082         prefer_neon_for_64bits field.
25083         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
25084         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
25085         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
25086         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
25087         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
25088         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
25089         (arm_option_override): Handle -mneon-for-64bits new option.
25090         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
25091         (prefer_neon_for_64bits): Declare new variable.
25092         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
25093         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
25094         (arch_enabled): Handle new arch types. Remove support for onlya8
25095         and nota8.
25096         (one_cmpldi2): Use new arch names.
25097         (zero_extend<mode>di2, extend<mode>di2): Ditto.
25098         * config/arm/arm.opt (mneon-for-64bits): Add option.
25099         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
25100         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
25101         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
25102         of onlya8.
25103         * doc/invoke.texi (-mneon-for-64bits): Document.
25105 2013-03-21  Richard Biener  <rguenther@suse.de>
25107         PR tree-optimization/39326
25108         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
25109         (sort_bbs_in_loop_postorder_cmp): New function.
25110         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
25112 2013-03-21  Richard Biener  <rguenther@suse.de>
25114         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
25115         (vect_insert_into_interleaving_chain): Likewise.
25116         (vect_drs_dependent_in_basic_block): Inline ...
25117         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
25118         split out from ...
25119         (vect_analyze_data_ref_dependence): ... here.  Simplify.
25120         (vect_check_interleaving): Simplify.
25121         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
25122         (vect_slp_analyze_data_ref_dependences): ... this new function.
25123         (dr_group_sort_cmp): New function.
25124         (vect_analyze_data_ref_accesses): Compute data-reference groups
25125         here instead of in vect_analyze_data_ref_dependence.  Use
25126         a more efficient algorithm.
25127         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
25128         vect_slp_analyze_data_ref_dependences.  Call
25129         vect_analyze_data_ref_accesses earlier.
25130         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
25131         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
25132         (vect_slp_analyze_data_ref_dependences): New prototype.
25134 2013-03-21  Richard Biener  <rguenther@suse.de>
25136         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
25137         ref is stored in the loop.
25138         (find_refs_for_sm): Walk only over all stores.
25139         (store_motion_loop): Allocate from lim_bitmap_obstack.
25140         (store_motion): Likewise.
25142 2013-03-21  Richard Biener  <rguenther@suse.de>
25144         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
25145         Update virtual SSA form.
25147 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25149         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
25150         * configure: Regenerate.
25151         * config.in: Regenerate.
25152         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
25153         if !HAVE_LD_EH_FRAME_CIEV3.
25155 2013-03-21  Richard Biener  <rguenther@suse.de>
25157         * tree-cfg.c (verify_expr_no_block): New function.
25158         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
25159         nor DECL_VALUE_EXPR have locations with associated blocks.
25160         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
25161         (clear_unused_block_pointer): Remove code dealing with
25162         blocks in DECL_DEBUG_EXPR locations.
25164 2013-03-21  Richard Biener  <rguenther@suse.de>
25166         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
25167         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
25168         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
25169         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
25170         instead of DECL_DEBUG_EXPR_IS_FROM.
25171         * gimplify.c (gimplify_modify_expr): Likewise.
25172         * tree-cfg.c (verify_expr_location_1): Likewise.
25173         * tree-complex.c (create_one_component_var): Likewise.
25174         * tree-sra.c (create_access_replacement): Likewise.
25175         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
25176         (clear_unused_block_pointer): Likewise.
25177         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
25178         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
25179         * var-tracking.c (var_debug_decl): Likewise.
25180         (track_expr_p): Likewise.
25181         * tree-inline.c (add_local_variables): Likewise.  Set
25182         DECL_HAS_DEBUG_EXPR_P after copying it.
25183         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
25184         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
25186 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25188         PR bootstrap/56656
25189         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
25190         * configure: Regenerate.
25191         * config.in: Regenerate.
25192         * config/i386/i386.md (*movdf_internal): Use
25193         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
25194         movd instead of movq mnemonic for interunit moves.
25195         (*movdi_internal): Ditto.
25197 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25199         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
25200         (abd<mode>_3): New pattern.
25201         (aba<mode>_3): New pattern.
25202         (fabd<mode>_3): New pattern.
25204 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25206         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
25207         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
25208         occurrence of REGISTER_PREFIX as its empty string.
25210 2013-03-20  Jeff Law  <law@redhat.com>
25212         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
25213         addititional equivalences for equality comparisons between an SSA_NAME
25214         and a constant where the SSA_NAME was set from a widening conversion.
25216 2013-03-20  Walter Lee  <walt@tilera.com>
25218         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
25220 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25222         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
25223         depending on type attribute.
25224         (*movti_internal): Ditto.
25225         (*movtf_internal): Ditto.
25226         (*movxf_internal): Ditto.
25227         (*movdf_internal): Ditto.
25228         (*movsf_internal): Ditto.
25230 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25232         * config/i386/i386.md (*movti_internal): Set prefix attribute to
25233         maybe_vex for sselog1 and ssemov types.
25234         (*movdi_internal): Reorder operand constraints.
25235         (*movsi_internal): Ditto.  Set prefix attribute to
25236         maybe_vex for sselog1 and ssemov types.
25237         (*movtf_internal): Set prefix attribute to maybe_vex
25238         for sselog1 and ssemov types.
25239         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
25240         DImode ssemov types.  Reorder operand constraints.
25241         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
25242         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
25243         attribute for SImode ssemov types.  Reorder operand constraints.
25245 2013-03-20  Martin Jambor  <mjambor@suse.cz>
25247         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
25248         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
25250 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
25252         * config/rs6000/predicates.md (indexed_address, update_address_mem
25253         update_indexed_address_mem): New predicates.
25254         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
25255         attribute for load/store instructions.
25256         * config/rs6000/dfp.md (movsd_store): Likewise.
25257         (movsd_load): Likewise.
25258         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
25259         (unnamed HI->DI extend define_insn): Likewise.
25260         (unnamed SI->DI extend define_insn): Likewise.
25261         (unnamed QI->SI extend define_insn): Likewise.
25262         (unnamed QI->HI extend define_insn): Likewise.
25263         (unnamed HI->SI extend define_insn): Likewise.
25264         (unnamed HI->SI extend define_insn): Likewise.
25265         (extendsfdf2_fpr): Likewise.
25266         (movsi_internal1): Likewise.
25267         (movsi_internal1_single): Likewise.
25268         (movhi_internal): Likewise.
25269         (movqi_internal): Likewise.
25270         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
25271         attribute for load/store instructions.
25272         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
25273         instructions.
25274         (mov<mode>_softfloat): Likewise.
25275         (mov<mode>_hardfloat32): Likewise.
25276         (mov<mode>_hardfloat64): Likewise.
25277         (mov<mode>_softfloat64): Likewise.
25278         (movdi_internal32): Likewise.
25279         (movdi_internal64): Likewise.
25280         (probe_stack_<mode>): Likewise.
25282 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
25284         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
25285         floating point, and decimal floating point to reload iterator.
25287         * config/rs6000/constraints.md (wl constraint): New constraints to
25288         return FLOAT_REGS if certain options are used to reduce the number
25289         of separate patterns that exist in the file.
25290         (wx constraint): Likewise.
25291         (wz constraint): Likewise.
25293         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25294         -mdebug=reg, print wg, wl, wx, and wz constraints.
25295         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25296         Initialize the reload functions for 64-bit binary/decimal floating
25297         point types.
25298         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25299         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25300         create the buffer on the stack to overcome not having a 32-bit
25301         load and store.
25302         (rs6000_emit_move): Likewise.
25303         (rs6000_secondary_memory_needed_rtx): Likewise.
25304         (rs6000_alloc_sdmode_stack_slot): Likewise.
25305         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25306         via xxlxor, just like DFmode 0.0.
25308         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
25309         define as 1 if we are running on a power7 or newer.
25310         (enum r6000_reg_class_enum): Add new constraints.
25312         * config/rs6000/dfp.md (movsd): Delete, combine with binary
25313         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25314         with other moves by using conditional constraits (wg).  Use LFIWZX
25315         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
25316         (movsd splitter): Likewise.
25317         (movsd_hardfloat): Likewise.
25318         (movsd_softfloat): Likewise.
25320         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
25321         binary and decimal floating point moves.
25322         (fmove_ok): New attributes to combine binary and decimal floating
25323         point moves, and to combine power6x (mfpgpr) moves along normal
25324         floating moves.
25325         (real_value_to_target): Likewise.
25326         (f32_lr): Likewise.
25327         (f32_lm): Likewise.
25328         (f32_li): Likewise.
25329         (f32_sr): Likewise.
25330         (f32_sm): Likewise.
25331         (f32_si): Likewise.
25332         (movsf): Combine binary and decimal floating point moves.  Combine
25333         power6x (mfpgpr) moves with other moves by using conditional
25334         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
25335         (mov<mode> for SFmode/SDmode); Likewise.
25336         (SFmode/SDmode splitters): Likewise.
25337         (movsf_hardfloat): Likewise.
25338         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
25339         (movsf_softfloat): Likewise.
25340         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
25342         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
25343         wx and wz constraints.
25345         * config/rs6000/constraints.md (wg constraint): New constraint to
25346         return FLOAT_REGS if -mmfpgpr (power6x) was used.
25348         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
25349         constraint.
25351         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25352         -mdebug=reg, print wg, wl, wx, and wz constraints.
25353         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25354         Initialize the reload functions for 64-bit binary/decimal floating
25355         point types.
25356         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25357         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25358         create the buffer on the stack to overcome not having a 32-bit
25359         load and store.
25360         (rs6000_emit_move): Likewise.
25361         (rs6000_secondary_memory_needed_rtx): Likewise.
25362         (rs6000_alloc_sdmode_stack_slot): Likewise.
25363         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25364         via xxlxor, just like DFmode 0.0.
25366         * config/rs6000/dfp.md (movdd): Delete, combine with binary
25367         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25368         with other moves by using conditional constraits (wg).  Use LFIWZX
25369         and STFIWX for loading SDmode on power7.
25370         (movdd splitters): Likewise.
25371         (movdd_hardfloat32): Likewise.
25372         (movdd_softfloat32): Likewise.
25373         (movdd_hardfloat64_mfpgpr): Likewise.
25374         (movdd_hardfloat64): Likewise.
25375         (movdd_softfloat64): Likewise.
25377         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
25378         64-bit binary and decimal floating point moves.
25379         (FMOVE64X): Likewise.
25380         (movdf): Combine 64-bit binary and decimal floating point moves.
25381         Combine power6x (mfpgpr) moves with other moves by using
25382         conditional constraits (wg).
25383         (mov<mode> for DFmode/DDmode): Likewise.
25384         (DFmode/DDmode splitters): Likewise.
25385         (movdf_hardfloat32): Likewise.
25386         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
25387         (movdf_softfloat32): Likewise.
25388         (movdf_hardfloat64_mfpgpr): Likewise.
25389         (movdf_hardfloat64): Likewise.
25390         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
25391         (movdf_softfloat64): Likewise.
25392         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
25393         (reload_<mode>_load): Move to later in the file so they aren't in
25394         the middle of the floating point move insns.
25395         (reload_<mode>_store): Likewise.
25397         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
25398         constraint.
25400         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
25401         constraint if -mdebug=reg.
25402         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
25403         Enable using dd reload support if needed.
25405         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
25406         binary and decimal floating point moves in rs6000.md.
25407         (movtd_internal): Likewise.
25409         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
25410         decimal floating point moves.
25411         (movtf): Likewise.
25412         (movtf_internal): Likewise.
25413         (mov<mode>_internal, TDmode/TFmode): Likewise.
25414         (movtf_softfloat): Likewise.
25415         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
25417         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
25418         movdi_internal64, using wg constraint for move direct operations.
25419         (movdi_internal64): Likewise.
25421         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
25422         MODES_TIEABLE_P for selected modes.  Print the numerical value of
25423         the various virtual registers. Use GPR/FPR first/last values,
25424         instead of hard coding the register numbers.  Print which modes
25425         have reload functions registered.
25426         (rs6000_option_override_internal): If -mdebug=reg, trace the options
25427         settings before/after setting cpu, target and subtarget settings.
25428         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
25429         and for secondary reload failures in rs6000_secondary_reload_inner.
25430         (rs6000_secondary_reload_fail): Likewise.
25431         (rs6000_secondary_reload_inner): Likewise.
25433         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
25434         macros for first/last GPR and FPR registers.
25435         (LAST_GPR_REGNO): Likewise.
25436         (FIRST_FPR_REGNO): Likewise.
25437         (LAST_FPR_REGNO): Likewise.
25439         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
25440         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
25441         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
25442         (vcond<mode><mode>): Likewise.
25443         (vcondu<mode><mode>): Likewise.
25444         (vector_gtu<mode>): Likewise.
25445         (vector_gte<mode>): Likewise.
25446         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
25447         to prevent the compiler from converting DImode operations to TImode.
25448         (ior<mode>3): Likewise.
25449         (and<mode>3): Likewise.
25450         (one_cmpl<mode>2): Likewise.
25451         (nor<mode>3): Likewise.
25452         (andc<mode>3): Likewise.
25454         * config/rs6000/constraints.md (wt constraint): New constraint
25455         that returns VSX_REGS if TImode is allowed in VSX registers.
25457         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
25458         constant under VSX.
25460         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
25461         similar to TImode, but it is restricted to being in the GPRs.
25463         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
25464         TImode to occupy a single VSX register.
25466         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
25467         -mvsx-timode for power7/power8.
25468         (power7 cpu): Likewise.
25469         (power8 cpu): Likewise.
25471         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
25472         sure that TFmode/TDmode take up two registers if they are ever
25473         allowed in the upper VSX registers.
25474         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
25475         registers.
25476         (rs6000_init_hard_regno_mode_ok): Likewise.
25477         (rs6000_debug_reg_global): Add debugging for PTImode and wt
25478         constraint.  Print if LRA is turned on.
25479         (rs6000_option_override_internal): Give an error if -mvsx-timode
25480         and VSX is not enabled.
25481         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
25482         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
25483         to reg+offset addressing.  Use PTImode when checking offset
25484         addresses for validity.
25485         (reg_offset_addressing_ok_p): Likewise.
25486         (rs6000_legitimate_offset_address_p): Likewise.
25487         (rs6000_legitimize_address): Likewise.
25488         (rs6000_legitimize_reload_address): Likewise.
25489         (rs6000_legitimate_address_p): Likewise.
25490         (rs6000_eliminate_indexed_memrefs): Likewise.
25491         (rs6000_emit_move): Likewise.
25492         (rs6000_secondary_reload): Likewise.
25493         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
25494         reloads to fpr registers to continue to use reg+offset addressing,
25495         but 64-bit reloads to altivec registers need reg+reg addressing.
25496         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
25497         it.  Treat LO_SUM like a PLUS operation.
25498         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
25499         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
25500         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
25501         registers to share a register with a smaller sized type, since VSX
25502         puts scalars in the upper 64-bits.
25503         (print_operand): Add support for PTImode.
25504         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
25505         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
25506         registers, but don't have arithmetic support.
25507         (rs6000_memory_move_cost): Add test for VSX.
25508         (rs6000_opt_masks): Add -mvsx-timode.
25510         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
25511         for TImode.
25512         (VSs): Likewise.
25513         (VSr): Use wt constraint for TImode.
25514         (VSv): Drop TImode support.
25515         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
25516         (vsx_movti_64bit): Likewise.
25517         (vsx_movti_32bit): Likewise.
25518         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
25519         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
25520         one '?' on the appropriate output constraint.  Do not allow TImode
25521         logical operations on 32-bit systems.
25522         (vsx_ior<mode>3): Likewise.
25523         (vsx_xor<mode>3): Likewise.
25524         (vsx_one_cmpl<mode>2): Likewise.
25525         (vsx_nor<mode>3): Likewise.
25526         (vsx_andc<mode>3): Likewise.
25527         (vsx_concat_<mode>): Likewise.
25528         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
25530         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
25531         OPTION_MASK_VSX_TIMODE.
25532         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
25533         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
25535         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
25536         (TI2 iterator): New iterator for TImode, PTImode.
25537         (wd mode attribute): Add values for vector types.
25538         (movti_string): Replace TI move operations with operations for TImode
25539         and PTImode.  Add support for TImode being allowed in VSX registers.
25540         (mov<mode>_string, TImode/PTImode): Likewise.
25541         (movti_ppc64): Likewise.
25542         (mov<mode>_ppc64, TImode/PTImode): Likewise.
25543         (TI mode splitters): Likewise.
25545         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
25546         constraint.
25548 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
25550         PR tree-optimization/56355
25551         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
25552         Also handle integers with undefined overflow.
25554 2013-03-20  Catherine Moore  <clm@codesourcery.com>
25555             Maciej W. Rozycki  <macro@codesourcery.com>
25556             Tom de Vries  <tom@codesourcery.com>
25557             Nathan Sidwell  <nathan@codesourcery.com>
25558             Iain Sandoe  <iain@codesourcery.com>
25559             Nathan Froyd  <froydnj@codesourcery.com>
25560             Chao-ying Fu  <fu@mips.com>
25562         * doc/extend.texi: (micromips, nomicromips, nocompression):
25563         Document new function attributes.
25564         * doc/invoke.texi (minterlink-compressed, mmicromips,
25565         m14k, m14ke, m14kec): Document new options.
25566         (minterlink-mips16): Update documentation.
25567         * doc/md.texi (ZC, ZD): Document new constraints.
25568         * configure.ac (gcc_cv_as_micromips): Check if linker
25569         supports the .set micromips directive.
25570         * configure: Regenerate.
25571         * config.in: Regenerate.
25572         * config/mips/mips-tables.opt: Regenerate.
25573         * config/mips/micromips.md: New file.
25574         * constraints.md (ZC, ZD): New constraints.
25575         * config/mips/predicates.md (movep_src_register): New predicate.
25576         (movep_src_operand): New predicate.
25577         (non_volatile_mem_operand): New predicate.
25578         * config/mips/mips.md (multimem): New type.
25579         (length): Differentiate between 17-bit and 18-bit branch offsets.
25580         (MOVEP1, MOVEP2): New mode iterator.
25581         (mov_<load>l): Use ZC constraint.
25582         (mov_<load>r): Likewise.
25583         (mov_<store>l): Likewise.
25584         (mov_<store>r): Likewise.
25585         (*branch_equality<mode>_inverted): Add microMIPS support.
25586         (*branch_equality<mode>): Likewise.
25587         (*jump_absolute): Likewise.
25588         (indirect_jump_<mode>): Likewise.
25589         (tablejump_<mode>): Likewise.
25590         (<optab>_internal): Likewise.
25591         (sibcall_internal): Likewise.
25592         (sibcall_value_internal): Likewise.
25593         (prefetch): Use constraint ZD.
25594         * config/mips/mips.opt (minterlink-compressed): New option.
25595         (minterlink-mips16): Now an alias for minterlink-compressed.
25596         (mmicromips): New option.
25597         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
25598         (compare_and_swap_12): Likewise.
25599         (sync_add<mode>): Likewise.
25600         (sync_<optab>_12): Likewise.
25601         (sync_old_<optab>_12): Likewise.
25602         (sync_new_<optab>_12): Likewise.
25603         (sync_nand_12): Likewise.
25604         (sync_old_nand_12): Likewise.
25605         (sync_new_nand_12): Likewise.
25606         (sync_sub<mode>): Likewise.
25607         (sync_old_add<mode>): Likewise.
25608         (sync_old_sub<mode>): Likewise.
25609         (sync_new_add<mode>): Likewise.
25610         (sync_new_sub<mode>): Likewise.
25611         (sync_<optab><mode>): Likewise.
25612         (sync_old_<optab><mode>): Likewise.
25613         (sync_new_<optab><mode>): Likewise.
25614         (sync_nand<mode>): Likewise.
25615         (sync_old_nand<mode>): Likewise.
25616         (sync_new_nand<mode>): Likewise.
25617         (sync_lock_test_and_set<mode>): Likewise.
25618         (test_and_set_12): Likewise.
25619         (atomic_compare_and_swap<mode>): Likewise.
25620         (atomic_exchange<mode>_llsc): Likewise.
25621         (atomic_fetch_add<mode>_llsc): Likewise.
25622         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
25623         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
25624         (umips_save_restore_pattern_p): Likewise.
25625         (umips_load_store_pair_p): Likewise.
25626         (umips_output_load_store_pair): Likewise.
25627         (umips_movep_target_p): Likewise.
25628         (umips_12bit_offset_address_p): Likewise.
25629         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
25630         (mips_base_mips16): Rename this...
25631         (mips_base_compression_flags): ...to this. Update all uses.
25632         (mips_attribute_table): Add micromips, nomicromips and nocompression.
25633         (mips_mips16_decl_p): Delete.
25634         (mips_nomips16_decl_p): Delete.
25635         (mips_get_compress_on_flags): New function.
25636         (mips_get_compress_off_flags): New function.
25637         (mips_get_compress_mode): New function.
25638         (mips_get_compress_on_name): New function.
25639         (mips_get_compress_off_name): New function.
25640         (mips_insert_attributes): Support multiple compression types.
25641         (mips_merge_decl_attributes): Likewise.
25642         (umips_12bit_offset_address_p): New function.
25643         (mips_start_function_definition): Emit .set micromips directive.
25644         (mips_call_may_need_jalx_p): New function.
25645         (mips_function_ok_for_sibcall): Add microMIPS support.
25646         (mips_print_operand_punctuation): Support short delay slots and
25647         compact jumps.
25648         (umips_swm_mask, umips_swm_encoding): New.
25649         (umips_build_save_restore): New function.
25650         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
25651         (was_mips16_p): Remove.
25652         (old_compression_mode): New.
25653         (mips_set_compression_mode): New function.
25654         (mips_set_current_function): Add microMIPS support.
25655         (mips_option_override): Likewise.
25656         (umips_save_restore_pattern_p): New function.
25657         (umips_output_save_restore): New function.
25658         (umips_load_store_pair_p_1): New function.
25659         (umips_load_store_pair_p): New function.
25660         (umips_output_load_store_pair_1): New function.
25661         (umips_output_load_store_pair): New function.
25662         (umips_movep_target_p) New function.
25663         (mips_prepare_pch_save): Add microMIPS support.
25664         * config/mips/mips.h (TARGET_COMPRESSION): New.
25665         (TARGET_CPU_CPP_BUILTINS): Update macro
25666         to use new compression flags and to support microMIPS.
25667         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
25668         (MIPS_ARCH_FLOAT_SPEC): Likewise.
25669         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
25670         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
25671         (ASM_SPEC): Support mmicromips and mno-micromips.
25672         (M16STORE_REG_P): New macro.
25673         (MIPS_CALL): Support TARGET_MICROMIPS.
25674         (MICROMIPS_J): New macro.
25675         (mips_base_mips16): Rename this...
25676         (mips_base_compression_flags): ...to this.
25677         (UMIPS_12BIT_OFFSET_P): New macro.
25678         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
25679         (MULTILIB_DIRNAMES): Likewise.
25680 2013-03-20  Richard Biener  <rguenther@suse.de>
25682         PR tree-optimization/56661
25683         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
25684         the result does not have to be distinct.
25686 2013-03-20  Richard Biener  <rguenther@suse.de>
25688         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
25689         remap_gimple_op_r.
25691 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
25692             Steven Bosscher  <steven@gcc.gnu.org>
25694         PR rtl-optimization/56605
25695         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
25697 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25699         PR bootstrap/56656
25700         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
25701         that require movd instead of movq.
25703 2013-03-20  Richard Biener  <rguenther@suse.de>
25705         * tree-ssa-structalias.c (struct variable_info): Add pointer
25706         to the first field of an aggregate with sub-vars.  Make
25707         this and the pointer to the next subfield its ID.
25708         (vi_next): New function.
25709         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
25710         storedanything_id, integer_id): Increment by one.
25711         (new_var_info, get_call_vi, lookup_call_clobber_vi,
25712         get_call_clobber_vi): Adjust.
25713         (solution_set_expand): Simplify and speedup.
25714         (solution_set_add): Inline into ...
25715         (set_union_with_increment): ... this.  Adjust accordingly.
25716         (do_sd_constraint): Likewise.
25717         (do_ds_constraint): Likewise.
25718         (do_complex_constraint): Simplify.
25719         (build_pred_graph): Adjust.
25720         (solve_graph): Likewise.  Simplify and speedup.
25721         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
25722         get_constraint_for_component_ref, get_constraint_for_1,
25723         first_vi_for_offset, first_or_preceding_vi_for_offset,
25724         create_function_info_for, create_variable_info_for_1,
25725         create_variable_info_for, intra_create_variable_infos): Adjust.
25726         (init_base_vars): Push NULL for ID zero.
25727         (compute_points_to_sets): Adjust.
25729 2013-03-20  Richard Biener  <rguenther@suse.de>
25731         * cfgloop.c (verify_loop_structure): Streamline and avoid
25732         ICEing on corrupt loop tree.
25733         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
25734         loop tree.
25736 2013-03-20  Richard Biener  <rguenther@suse.de>
25738         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
25739         check whether an SSA update is needed.
25741 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
25743         * config/mips/constraints.md (T): Rename to...
25744         (Yf): ...this.
25745         (U): Rename to...
25746         (Yd): ...this.
25747         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
25748         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
25750 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25752         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
25753         (*subsi3_carryin_uxtw): Likewise.
25755 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25757         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
25758         (*rorsi3_insn_uxtw): Likewise.
25760 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25762         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
25763         (*extrsi5_insn_uxtw): Likewise.
25765 2013-03-19  Richard Biener  <rguenther@suse.de>
25767         PR tree-optimization/56273
25768         * passes.c (init_optimization_passes): Move second VRP after DOM.
25770 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
25772         * config/i386/i386.md (*movti_internal): Merge from
25773         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
25774         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
25775         nox64 isa attributes.
25777 2013-03-18  Richard Biener  <rguenther@suse.de>
25779         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
25780         (unite): Likewise.
25781         (merge_node_constraints): Likewise.
25782         (build_succ_graph): Likewise.
25783         (valid_graph_edge): Inline into single caller.
25784         (unify_nodes): Likewise.  Use bitmap_set_bit return value
25785         and cache varinfo.
25786         (scc_visit): Fix formatting and variable use.
25787         (do_sd_constraint): Use gcc_checking_assert.
25788         (do_ds_constraint): Likewise.
25789         (do_complex_constraint): Likewise.
25790         (condense_visit): Likewise.  Cleanup.
25791         (dump_pred_graph): New function.
25792         (perform_var_substitution): Dump the pred-graph before
25793         variable substitution.
25794         (find_equivalent_node): Use gcc_checking_assert.
25795         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
25797 2013-03-18  Richard Biener  <rguenther@suse.de>
25799         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
25800         Remove cond_expr_stmt_list argument and do not gimplify the
25801         built expression.
25802         (vect_loop_versioning): Adjust.
25803         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
25804         Cleanup to use less temporaries.
25805         (vect_create_data_ref_ptr): Cleanup.
25807 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25809         PR tree-optimization/56635
25810         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
25811         require types_compatible_p types.
25813 2013-03-18  Nick Clifton  <nickc@redhat.com>
25815         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
25816         spurious backslash.
25818         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
25819         Add missing line to comment describing function.
25821 2013-03-18  Richard Biener  <rguenther@suse.de>
25823         PR tree-optimization/56210
25824         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25825         Handle string / character search functions.
25826         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
25828 2013-03-18  Richard Biener  <rguenther@suse.de>
25830         PR middle-end/56483
25831         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
25832         and implement properly.
25833         * gimple.h (gimple_cond_single_var_p): Remove.
25835 2013-03-18  Richard Biener  <rguenther@suse.de>
25837         * tree-data-ref.h (find_data_references_in_loop): Declare.
25838         * tree-data-ref.c (get_references_in_stmt): Use a stack
25839         vector pre-allocated in the callers.
25840         (find_data_references_in_stmt): Adjust.
25841         (graphite_find_data_references_in_stmt): Likewise.
25842         (create_rdg_vertices): Likewise.
25843         (find_data_references_in_loop): Export.
25844         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
25845         Compute dependences here...
25846         (vect_analyze_data_refs): ...not here.  When we encounter
25847         a non-vectorizable data reference in basic-block vectorization
25848         truncate the data reference vector.  Do not bother to
25849         fixup data-dependence information for gather loads.
25850         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
25851         of data references, as reported.
25853 2013-03-18  Richard Biener  <rguenther@suse.de>
25855         PR tree-optimization/3713
25856         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
25857         has_constants and expr.
25858         (stmt_has_constants): Properly valueize SSA names when deciding
25859         whether the stmt has constants.
25861 2013-03-18  Richard Biener  <rguenther@suse.de>
25863         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
25864         whole function when there is nothing to do.
25865         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
25866         * tree-vectorizer.c (vectorize_loops): Update virtual and
25867         loop-closed SSA once.
25868         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
25870 2013-03-18  Richard Biener  <rguenther@suse.de>
25872         PR middle-end/56113
25873         * domwalk.c (bb_postorder): New global static.
25874         (cmp_bb_postorder): New function.
25875         (walk_dominator_tree): Replace scheme imposing an order for
25876         visiting dominator sons by one sorting them at the time they
25877         are pushed on the stack.
25879 2013-03-18  Richard Biener  <rguenther@suse.de>
25881         PR tree-optimization/39326
25882         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
25883         (struct mem_ref): Replace mem member with ao_ref typed member.
25884         (MEM_ANALYZABLE): Adjust.
25885         (memref_eq): Likewise.
25886         (mem_ref_alloc): Likewise.
25887         (gather_mem_refs_stmt): Likewise.
25888         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
25889         (execute_sm_if_changed_flag_set): Adjust.
25890         (execute_sm): Likewise.
25891         (ref_always_accessed_p): Likewise.
25892         (refs_independent_p): Likewise.
25893         (can_sm_ref_p): Likewise.
25895 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25897         PR c/56566
25898         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
25899         return 1 even for !unsignedp.
25901 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25903         * config/i386/i386.md (isa): Add x64 and nox64.
25904         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
25905         (*pushtf): Enable *roF alternative for x64 isa only.
25906         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
25907         mode attribute of integer alternatives to DImode for TARGET_64BIT.
25908         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
25909         (*movtf_internal): Merge from *movtf_internal_rex64 and
25910         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
25911         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
25912         nox64 isa attributes.
25913         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
25914         nox64 isa attributes.
25915         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
25917 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25919         * config/alpha/alpha.c (TARGET_LRA_P): New define.
25921 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
25923         PR target/56640
25924         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
25925         class names.  Remove trailing comma after "ALL_REGS".
25927 2013-03-16  Jan Hubicka  <jh@suse.cz>
25929         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
25930         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
25931         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
25932         of cgraph_get_create_node.
25933         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
25935 2013-03-16  Jason Merrill  <jason@redhat.com>
25937         PR debug/49090
25938         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
25939         with DW_AT_default_value.
25941 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
25943         * BASE-VER: Set to 4.9.0.
25945 2013-03-14  Andi Kleen  <ak@linux.intel.com>
25947         PR target/56619
25948         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
25949         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
25950         Document _x* TSX intrinsics.
25952 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
25953             David Holsgrove  <david.holsgrove@xilinx.com>
25955         * configure.ac: Add MicroBlaze TLS support detection.
25956         * configure: Regenerate.
25957         * config/microblaze/microblaze-protos.h
25958         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
25959         symbol_mentioned_p, label_mentioned_p): Add prototypes.
25960         * config/microblaze/microblaze.c (microblaze_address_type): Add
25961         ADDRESS_TLS and tls_reloc address types.
25962         (microblaze_address_info): Add tls_reloc.
25963         (TARGET_HAVE_TLS): Define.
25964         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
25965          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
25966          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
25967         load_tls_operand,  microblaze_call_tls_get_addr,
25968         microblaze_legitimize_tls_address): New functions.
25969         (microblaze_classify_unspec): Handle UNSPEC_TLS.
25970         (get_base_reg): Use microblaze_tls_symbol_p.
25971         (microblaze_classify_address): Handle TLS.
25972         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
25973         label_mentioned_p and microblaze_tls_referenced_p.
25974         (microblaze_legitimize_address): Handle TLS.
25975         (microblaze_address_insns): Handle ADDRESS_TLS.
25976         (pic_address_needs_scratch): Handle TLS.
25977         (print_operand_address): Handle TLS.
25978         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
25979         (microblaze_expand_move): Handle TLS.
25980         (microblaze_legitimate_constant_p): Check
25981         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
25982         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
25983         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
25984         (PIC_OFFSET_TABLE_REGNUM): Set.
25985         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
25986         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
25987         (addsi3, movsi_internal2, movdf_internal): Update constraints
25988         * config/microblaze/predicates.md (arith_plus_operand): Define
25989         (move_operand): Redefine as move_src_operand,
25990         check microblaze_tls_referenced_p.
25992 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25994         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
25995         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
25997 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
25999         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
26000         CC mode for AND.
26002 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
26004         PR tree-optimization/53265
26005         * common.opt (Waggressive-loop-optimizations): New option.
26006         * tree-ssa-loop-niter.c: Include tree-pass.h.
26007         (do_warn_aggressive_loop_optimizations): New function.
26008         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
26009         if number_of_latch_executions returned constant.
26010         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
26011         early.  If number_of_latch_executions returned constant, set
26012         nb_iterations_upper_bound back to it.
26013         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
26014         field.
26015         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
26016         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
26018         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
26019         (MULTILIB_OSDIRNAMES): Set.
26020         * genmultilib: If defaultosdirname doesn't start with :: , set
26021         defaultosdirname2 instead, clear it and emit two . multilib_raw
26022         entries instead of just one.
26024 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
26026         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26027         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
26028         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26029         (SUBTARGET_OVERRIDE_OPTIONS): New.
26031 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
26033         PR target/49880
26034         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
26035         (musermode): Convert to Var(TARGET_USERMODE).
26036         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
26037         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
26038         * config/sh/sh.c (sh_option_override): Use
26039         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
26040         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
26041         condition.
26042         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
26043         TARGET_SH4.
26044         (udivsi3_i4_single, divsi3_i4_single): Use
26045         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
26047 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
26049         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
26050         default setting.
26052 2013-03-13  Richard Biener  <rguenther@suse.de>
26054         PR tree-optimization/56608
26055         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
26056         calls when vectorizing basic-blocks.
26058 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
26060         PR plugins/45078
26061         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
26062         tm_file.
26064 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
26066         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
26068 2013-03-11  Jan Hubicka  <jh@suse.cz>
26070         PR lto/56557
26071         * lto-streamer-out.c (output_symbol_p): Skip references from
26072         constructors of external variables.
26074 2013-03-11  Jan Hubicka  <jh@suse.cz>
26076         PR middle-end/56571
26077         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
26078         from pseudos.
26079         * emit-rtl.c (verify_rtx_sharing): Likewise.
26080         (copy_insn_1): Likewise.
26081         * rtl.c (copy_rtx): Likewise.
26083 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
26085         PR target/56591
26086         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
26087         output_operand_lossage message.
26089 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
26091         PR target/56470
26092         * arm.c (shift_op): Validate RTL pattern on the fly.
26093         (arm_print_operand, case 'S'): Don't use shift_operator to validate
26094         the RTL.
26096 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26098         PR target/56347
26099         * config/pa/pa.md (call_value): Check for calls to powf and direct to
26100         new call patterns that clobber %fr12.
26101         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
26102         split and postreload patterns.
26103         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
26104         registers %fr12 and %fr12R as call used.
26106 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
26108         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
26109         (canon_address, record_store, replace_read, check_mem_read_rtx,
26110         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
26111         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
26112         rest_of_handle_dse): Likewise.
26114 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
26116         PR middle-end/56524
26117         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
26118         Add base_optabs.
26119         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
26120         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
26121         (save_optabs_if_changed): Replace with...
26122         (init_tree_optimization_optabs): ...this.
26123         * optabs.c (save_optabs_if_changed): Rename to...
26124         (init_tree_optimization_optabs): ...this.  Take the optimization node
26125         as argument.  Do nothing if the base optabs are already correct.
26126         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
26127         to recompute optabs.
26128         * function.h (function): Remove optabs field.
26129         * function.c (invoke_set_current_function_hook): Call
26130         init_tree_optimization_optabs.  Use the result to initialize
26131         this_fn_optabs.
26133 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
26135         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
26136         if GTMA_HAS_NO_INSTRUMENTATION.
26137         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
26138         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
26139         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
26140         * gimple-pretty-print.c (dump_gimple_transaction): Handle
26141         GTMA_HAS_NO_INSTRUMENTATION.
26143 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
26145         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
26146         libasan_preinit.o.
26148 2013-03-08  Marek Polacek  <polacek@redhat.com>
26149             Jakub Jelinek  <jakub@redhat.com>
26151         PR tree-optimization/56478
26152         * predict.c (is_comparison_with_loop_invariant_p): Change the
26153         type of loop_step to tree.
26154         (predict_loops): Adjust.
26155         (predict_iv_comparison): Perform the computations on double_ints.
26157 2013-03-08  Richard Biener  <rguenther@suse.de>
26159         PR tree-optimization/56570
26160         * tree-cfg.c (verify_expr_location_1): Verify locations for
26161         DECL_DEBUG_EXPR.
26162         * tree-sra.c (create_access_replacement): Strip locations
26163         from DECL_DEBUG_EXPRs.
26165 2013-03-08  Richard Biener  <rguenther@suse.de>
26167         * tree-inline.c (expand_call_inline): Do not associate
26168         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
26169         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
26171 2013-03-08  Richard Biener  <rguenther@suse.de>
26173         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
26174         or block changes with -Og.  Fix for location / block encoding
26175         changes and PHI arguments with locations.
26177 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
26179         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
26180         for all counters.
26181         (struct output_info): Likewise.
26182         (register_overhead): Remove bad gcc_assert.
26183         (bitmap_find_bit): If there is only a single bitmap element, do not
26184         count a miss as a search.
26185         (print_statistics): Update for counter type changes.
26186         (dump_bitmap_statistics): Likewise.  Print headers such that they
26187         are properly lined up with the printed counters.
26189 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
26191         PR tree-optimization/56559
26192         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
26193         check that it has only a single use.
26195 2013-03-07  Richard Biener  <rguenther@suse.de>
26197         * doc/invoke.texi (fwhole-program): Discourage use in combination
26198         with -flto.
26200 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
26202         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
26204         PR tree-optimization/56539
26205         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
26206         instead of GSI_CONTINUE_LINKING as last argument to
26207         force_gimple_operand_gsi.  Adjust function comment.
26209         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
26210         aarch64-cores.def.
26212         PR middle-end/56548
26213         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
26214         promoted mode, convert the result back to the original mode.
26216 2013-03-06  Richard Biener  <rguenther@suse.de>
26218         PR middle-end/56294
26219         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
26220         (insert_updated_phi_nodes_compare_uids): New function.
26221         (update_ssa): Sort symbols_to_rename after UID before
26222         traversing it to insert PHI nodes.
26224 2013-03-06  Richard Biener  <rguenther@suse.de>
26226         PR middle-end/50494
26227         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
26228         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
26230         Revert
26231         2013-02-13  Richard Biener  <rguenther@suse.de>
26233         PR lto/50494
26234         * varasm.c (output_constant_def_1): Get the decl representing
26235         the constant as argument.
26236         (output_constant_def): Wrap output_constant_def_1.
26237         (make_decl_rtl): Use output_constant_def_1 with the decl
26238         representing the constant.
26239         (build_constant_desc): Optionally re-use a decl already
26240         representing the constant.
26241         (tree_output_constant_def): Adjust.
26243 2013-03-06  Joey Ye  <joey.ye@arm.com>
26245         PR lto/50293
26246         * gcc.c (convert_white_space): New function.
26247         (main): Handles white space in function name.
26249 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
26251         PR target/56529
26252         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
26253         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
26254         to SH_DIV_CALL_TABLE for TARGET_SH2.
26255         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
26256         list.
26257         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
26258         call-table options.
26260 2013-03-05  Sterling Augustine  <saugustine@google.com>
26261             Cary Coutant  <ccoutant@google.com>
26263         PR debug/55364
26264         * dwarf2out.c (resolve_addr): Don't call
26265         remove_loc_list_addr_table_entries a second time for the same
26266         expression.
26268 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26270         PR debug/56510
26271         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
26272         (avoid_complex_debug_insns): New function.
26273         (expand_debug_locations): Call it.
26275         PR rtl-optimization/56484
26276         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
26277         lifetimes of hard registers on small register class machines.
26279 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26281         * config/microblaze/microblaze-protos.h: Rename
26282         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
26283         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
26284         fast_interrupt.
26285         (microblaze_fast_interrupt_function_p): New function.
26286         (microblaze_is_interrupt_handler): Rename to
26287         microblaze_is_interrupt_variant and add fast_interrupt check.
26288         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
26289         (save_restore_insns): Likewise.
26290         (compute_frame_size): Likewise.
26291         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
26292         (microblaze_globalize_label): Likewise.
26293         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
26294         * config/microblaze/microblaze.md: Use wrapper
26295         microblaze_is_interrupt_variant.
26297 2013-03-05  Kai Tietz  <ktietz@redhat.com>
26299         * sdbout.c (sdbout_one_type): Switch to current function's section
26300         supporting cold/hot.
26302 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26304         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
26305         -mxl-reorder.
26307 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26309         PR middle-end/56461
26310         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
26311         if VALGRIND_GET_VBITS is defined, temporarily make object
26312         memory all defined, and restore previous valgrind addressability
26313         and definability afterwards.  Free this_object at the end.
26315         PR middle-end/56461
26316         * lra.c (lra): Call lra_clear_live_ranges if live_p,
26317         right before calling lra_create_live_ranges, also call it
26318         when clearing live_p.  Only call lra_clear_live_ranges
26319         at the end if live_p.
26321         PR middle-end/56461
26322         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
26324 2013-03-05  Richard Biener  <rguenther@suse.de>
26326         PR tree-optimization/56521
26327         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
26328         value-id.
26330 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
26332         PR c++/55135
26333         * except.h (remove_unreachable_eh_regions): New prototype.
26334         * except.c (remove_eh_handler_splicer): New function, split out
26335         of remove_eh_handler.
26336         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
26337         warning about running it on many EH regions one at a time.
26338         (remove_unreachable_eh_regions_worker): New function, walk the
26339         EH tree in depth-first order and remove non-marked regions.
26340         (remove_unreachable_eh_regions): New function.
26341         * tree-eh.c (mark_reachable_handlers): New function, split out
26342         from remove_unreachable_handlers.
26343         (remove_unreachable_handlers): Use mark_reachable_handlers and
26344         remove_unreachable_eh_regions.
26345         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
26346         and remove_unreachable_eh_regions.
26348 2013-03-05  Richard Biener  <rguenther@suse.de>
26350         PR middle-end/56525
26351         * loop-init.c (fix_loop_structure): Remove loops in two stages,
26352         not freeing them until the end.
26354 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26356         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
26358 2013-03-05  Richard Biener  <rguenther@suse.de>
26360         PR tree-optimization/56270
26361         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
26362         of loads after scheduling an SLP instance.
26364 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26366         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
26367         tic6x.exp.
26368         (check_gcc_parallelize): Run guality.exp as a separate job from
26369         vect.exp with unsorted.exp and $(dg_target_exps) separately from
26370         struct-layout-1.exp with stackalign.exp.
26372         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
26374         PR middle-end/56461
26375         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
26376         load_index sbitmap even if some bit in it isn't set.
26378         PR middle-end/56461
26379         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
26380         (discover_iteration_bound_by_body_walk): Change queues to
26381         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
26382         spelling in comment.  Call safe_push on queues[bound_index] directly.
26383         Release queues[queue_index] in every iteration unconditionally.
26384         Release bounds vector.
26386         PR middle-end/56461
26387         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
26388         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
26389         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
26390         inner_phis vector.
26392 2013-03-05  Richard Biener  <rguenther@suse.de>
26394         PR lto/56515
26395         * tree-inline.c (remap_blocks_to_null): New function.
26396         (expand_call_inline): When expanding a call stmt without
26397         an associated block inline remap all callee blocks to NULL.
26399 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26401         PR rtl-optimization/56494
26402         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
26403         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
26404         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
26406         PR middle-end/56461
26407         * sel-sched-ir.c (free_sched_pools): Release
26408         succs_info_pool.stack[succs_info_pool.max_top] vectors too
26409         if succs_info_pool.max_top isn't -1.
26411         PR bootstrap/56509
26412         * opts.c (opts_obstack, opts_concat): Moved to...
26413         * opts-common.c (opts_obstack, opts_concat): ... here.
26415 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26417         PR middle-end/56461
26418         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
26420 2013-03-04  Martin Jambor  <mjambor@suse.cz>
26422         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
26423         all appropriate places.
26425 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
26427         PR tree-optimization/56424
26428         * ipa-split.c (split_function): Do not set the RSO flag if result is
26429         not by reference and its type is a register type.
26431 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
26433         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
26434         (microblaze_legitimate_pic_operand): Likewise
26435         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
26436         new function microblaze_legitimate_pic_operand
26437         * config/microblaze/microblaze-protos.h
26438         (microblaze_legitimate_pic_operand): Declare.
26440 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26442         * config/microblaze/predicates.md (call_insn_simple_operand):
26443         New predicate for supported rtx code types.
26444         * config/microblaze/microblaze.md (call_internal1): Use
26445         call_insn_simple_operand predicate.
26447 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26449         PR middle-end/56461
26450         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
26451         partitions.ordered_remove.
26453         PR middle-end/56461
26454         * tree-vect-stmts.c (vectorizable_conversion): Don't call
26455         vec_oprnds0.create (1) for modifier == NONE.
26457         PR middle-end/56461
26458         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
26459         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
26460         vec_oprnds1 right before pushing anything to it for
26461         scalar_shift_arg.
26463         PR middle-end/56461
26464         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
26465         set nbbs to 0 instead of having separate code path.
26466         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
26467         instead of false as last argument if returning NULL.
26469 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
26471         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
26472         the attribute is now called "target" instead of "option".
26473         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
26474         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
26475         attribute/pragma name for TARGET_OPTION_VALID_P and
26476         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
26477         * doc/tm.texi: Regenerated.
26479 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
26481         * config/microblaze/microblaze.c:
26482         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
26483         * config/microblaze/microblaze.h: Add -mxl-reorder to
26484         DRIVER_SELF_SPECS.
26485         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
26486         instructions emitted if TARGET_REORDER.
26487         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
26488         or 0 for -m/-mno case, but initialises as 2 to detect default use case
26489         separately.
26491 2013-03-01  Xinliang David Li  <davidxl@google.com>
26493         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
26494         walk length.
26496 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
26498         PR middle-end/56461
26499         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
26500         vector even when returning true.  Fix up function comment formatting.
26502         PR middle-end/56461
26503         * ira-build.c (ira_loop_nodes_count): New variable.
26504         (create_loop_tree_nodes): Initialize it.
26505         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
26507         PR middle-end/56461
26508         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
26509         method on dr_chain and result_chain.
26510         * tree-vect-stmts.c (vectorizable_store): Only call
26511         result_chain.create if j == 0.
26513         PR middle-end/56461
26514         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
26515         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
26516         before overwriting it.
26518 2013-03-01  Tobias Burnus  <burnus@net-b.de>
26520         * doc/extended.texi (C Extensions): Change order in @menu
26521         to match @node.
26522         (Other MIPS Built-in Functions): Move last MIPS entry before
26523         "picoChip Built-in Functions".
26524         (SH Built-in Functions): Move after RX Built-in Functions.
26525         * doc/gcc.texi (Introduction): Change order in @menu
26526         to match @node.
26527         * doc/md.texi (Constraints): Ditto.
26528         * gty.texi (Type Information): Ditto.
26529         (User-provided marking routines for template types): Make
26530         subsection.
26531         * doc/invoke.texi (AArch64 Options): Move before
26532         "Adapteva Epiphany Options".
26534 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
26535             Jakub Jelinek  <jakub@redhat.com>
26537         PR sanitizer/56454
26538         * asan.c (gate_asan): Lookup no_sanitize_address instead of
26539         no_address_safety_analysis attribute.
26540         * doc/extend.texi (no_address_safety_attribute): Rename to
26541         no_sanitize_address attribute, mention no_address_safety_analysis
26542         attribute as deprecated alias.
26544 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
26546         PR middle-end/56461
26547         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
26548         type to vec<vec<tree> > *.
26549         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
26550         to be vec<tree> instead of vec<tree> *, set vec_defs
26551         to vNULL and call vec_defs.create (number_of_vects), adjust other
26552         uses of vec_defs.
26553         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
26554         vectorizable_condition): Adjust vect_get_slp_defs callers.
26556 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26558         * config/aarch64/aarch64.c
26559         (aarch64_float_const_representable): Remove unused variable.
26561 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26563         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
26565 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26567         * config/aarch64/aarch64-builtins.c
26568         (aarch64_init_simd_builtins): Make static.
26570 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26572         * config/aarch64/aarch64.c
26573         (aarch64_simd_make_constant): Make static.
26575 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26577         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
26578         with no initialization to the RHS of debug statements.
26580 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26582         PR tree-optimization/56294
26583         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
26584         Adjust dumping.
26585         (get_access_replacement): Do not call create_access_replacement.
26586         Assert a replacement exists.
26587         (get_repl_default_def_ssa_name): Create the replacement declaration
26588         itself.
26590 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26592         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
26593         final_end_function.
26595 2013-02-28  Marek Polacek  <polacek@redhat.com>
26597         PR rtl-optimization/56466
26598         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
26599         if we're changing a loop.
26600         (peel_loops_completely): Likewise.
26602 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
26604         PR c++/55813
26605         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
26607 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26609         PR target/56445
26610         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
26611         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
26612         INTX_FTYPE_FX, FX_FTYPE_INTX.
26613         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
26615 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26617         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
26618         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
26619         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
26620         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
26621         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
26622         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
26623         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
26624         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
26625         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
26626         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
26627         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
26628         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
26629         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
26630         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
26631         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
26632         (avrxmega6): Increase max flash segments from 5 to 6.
26633         * config/avr/t-multilib: Regenerate.
26634         * config/avr/avr-tables.opt: Regenerate.
26635         * doc/avr-mmcu.texi: Regenerate.
26637 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26639         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
26640         (avr_device_to_arch): Rename to avr_device_to_ld.
26641         (avr_device_to_as): New prototype.
26642         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
26643         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
26644         * config/avr/driver-avr.c (avr_device_to_as): New.
26645         (avr_device_to_arch): Rename to avr_device_to_ld.
26647 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26649         PR middle-end/56461
26650         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
26651         method on dr_chain and result_chain.
26653         PR middle-end/56461
26654         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
26655         pointer_set_destroy on not_executed_last_iteration.
26657         PR middle-end/56461
26658         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
26660         PR middle-end/56461
26661         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
26662         FOR_EACH_DEFINED_FUNCTION when freeing state.
26664         PR middle-end/56461
26665         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
26666         pool_free.
26667         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
26668         overwriting it.
26670         PR middle-end/56461
26671         * ipa-cp.c (decide_whether_version_node): Call vec_free on
26672         known_aggs[i].items and release known_aggs vector.
26674         PR middle-end/56461
26675         * ipa-reference.c (propagate): Free node_info even for alias nodes.
26677 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26679         * config/microblaze/microblaze.c (microblaze_emit_compare):
26680         Use xor for EQ/NE comparisions.
26681         * config/microblaze/microblaze.md (cstoresf4): Add constraints
26682         (cbranchsf4): Adjust operator to comparison_operator.
26684 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26686         PR middle-end/56461
26687         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
26688         vector.
26689         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
26690         vec_safe_push, always update *slot.
26691         (redirect_edge_var_map_clear): Use vec_free.
26692         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
26693         (free_var_map_entry): Use vec_free.
26694         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
26695         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
26697 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
26699         PR middle-end/45472
26700         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
26701         when the may_trap_p bit of the exprs being merged differs.
26702         Reorder tests for speculativeness in the logical and operator.
26704 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26706         * incpath.c (add_standard_paths): Use reconcat instead of concat
26707         where appropriate and avoid leaking memory.
26709         * opts.h: Include obstack.h.
26710         (opts_concat): New prototype.
26711         (opts_obstack): New declaration.
26712         * opts.c (opts_concat): New function.
26713         (opts_obstack): New variable.
26714         (init_options_struct): Call gcc_init_obstack on opts_obstack.
26715         (finish_options): Use opts_concat instead of concat
26716         and XOBNEWVEC instead of XNEWVEC.
26717         * opts-common.c (generate_canonical_option, decode_cmdline_option,
26718         generate_option): Likewise.
26719         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
26720         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
26722         PR target/56455
26723         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
26724         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
26726 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26728         PR middle-end/56461
26729         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
26731 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26733         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
26734         (arm_block_move_unaligned_straight): Likewise.
26735         (arm_adjust_block_mem): Likewise.
26737 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26739         PR target/48901
26740         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
26741         temp, cond and label.
26742         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
26744         PR target/52500
26745         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
26746         * config/c6x/c6x.h (dbx_register_map): Update declaration.
26748         PR target/52501
26749         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
26750         of prologue/epilogue functions.
26752         PR target/52550
26753         * config/tilegx/tilegx.c (tilegx_expand_prologue):
26754         Remove unused variable cfa_offset.
26755         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
26757         PR target/54639
26758         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
26759         type promotion to unsigned.
26761         PR target/54640
26762         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
26763         for HOST_WIDE_INT of 32 bit / same size as int.
26764         (arm_block_move_unaligned_straight): Likewise.
26765         (arm_adjust_block_mem): Likewise.
26767         PR target/54662
26768         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
26769         ALL_CFLAGS.
26771 2013-02-26  Marek Polacek  <polacek@redhat.com>
26773         PR tree-optimization/56426
26774         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
26776 2013-02-26  Richard Biener  <rguenther@suse.de>
26778         PR target/56444
26779         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
26780         unused variable loops.
26782 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26784         PR tree-optimization/56448
26785         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
26786         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
26787         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
26788         later operands of the references, or even first operand for
26789         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
26791         PR tree-optimization/56443
26792         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
26793         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
26794         to type_for_mode langhook.
26796 2013-02-25  Matt Turner  <mattst88@gmail.com>
26798         * doc/invoke.texi: Document r4700.
26800 2013-02-25  Richard Biener  <rguenther@suse.de>
26802         PR tree-optimization/56175
26803         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
26804         split out from ...
26805         (simplify_bitwise_binary): ... here.  Also guard the conversion
26806         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
26808 2013-02-25  Catherine Moore  <clm@codesourcery.com>
26810         Revert:
26811         2013-02-24  Catherine Moore  <clm@codesourcery.com>
26812             Maciej W. Rozycki  <macro@codesourcery.com>
26813             Tom de Vries  <tom@codesourcery.com>
26814             Nathan Sidwell  <nathan@codesourcery.com>
26815             Iain Sandoe  <iain@codesourcery.com>
26816             Nathan Froyd  <froydnj@codesourcery.com>
26817             Chao-ying Fu  <fu@mips.com>
26819         * doc/extend.texi: (micromips, nomicromips, nocompression):
26820         Document new function attributes.
26821         * doc/invoke.texi (minterlink-compressed, mmicromips,
26822         m14k, m14ke, m14kec): Document new options.
26823         (minterlink-mips16): Update documentation.
26824         * doc/md.texi (ZC, ZD): Document new constraints.
26825         * configure.ac (gcc_cv_as_micromips): Check if linker
26826         supports the .set micromips directive.
26827         * configure: Regenerate.
26828         * config.in: Regenerate.
26829         * config/mips/mips-tables.opt: Regenerate.
26830         * config/mips/micromips.md: New file.
26831         * constraints.md (ZC, AD): New constraints.
26832         * config/mips/predicates.md (movep_src_register): New predicate.
26833         (movep_src_operand): New predicate.
26834         (non_volatile_mem_operand): New predicate.
26835         * config/mips/mips.md (multimem): New type.
26836         (length): Differentiate between 17-bit and 18-bit branch offsets.
26837         (MOVEP1, MOVEP2): New mode iterator.
26838         (mov_<load>l): Use ZC constraint.
26839         (mov_<load>r): Likewise.
26840         (mov_<store>l): Likewise.
26841         (mov_<store>r): Likewise.
26842         (*branch_equality<mode>_inverted): Add microMIPS support.
26843         (*branch_equality<mode>): Likewise.
26844         (*jump_absolute): Likewise.
26845         (indirect_jump_<mode>): Likewise.
26846         (tablejump_<mode>): Likewise.
26847         (<optab>_internal): Likewise.
26848         (sibcall_internal): Likewise.
26849         (sibcall_value_internal): Likewise.
26850         (prefetch): Use constraint ZD.
26851         * config/mips/mips.opt (minterlink-compressed): New option.
26852         (minterlink-mips16): Now an alias for minterlink-compressed.
26853         (mmicromips): New option.
26854         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26855         (compare_and_swap_12): Likewise.
26856         (sync_add<mode>): Likewise.
26857         (sync_<optab>_12): Likewise.
26858         (sync_old_<optab>_12): Likewise.
26859         (sync_new_<optab>_12): Likewise.
26860         (sync_nand_12): Likewise.
26861         (sync_old_nand_12): Likewise.
26862         (sync_new_nand_12): Likewise.
26863         (sync_sub<mode>): Likewise.
26864         (sync_old_add<mode>): Likewise.
26865         (sync_old_sub<mode>): Likewise.
26866         (sync_new_add<mode>): Likewise.
26867         (sync_new_sub<mode>): Likewise.
26868         (sync_<optab><mode>): Likewise.
26869         (sync_old_<optab><mode>): Likewise.
26870         (sync_new_<optab><mode>): Likewise.
26871         (sync_nand<mode>): Likewise.
26872         (sync_old_nand<mode>): Likewise.
26873         (sync_new_nand<mode>): Likewise.
26874         (sync_lock_test_and_set<mode>): Likewise.
26875         (test_and_set_12): Likewise.
26876         (atomic_compare_and_swap<mode>): Likewise.
26877         (atomic_exchange<mode>_llsc): Likewise.
26878         (atomic_fetch_add<mode>_llsc): Likewise.
26879         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26880         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26881         (umips_save_restore_pattern_p): Likewise.
26882         (umips_load_store_pair_p): Likewise.
26883         (umips_output_load_store_pair): Likewise.
26884         (umips_movep_target_p): Likewise.
26885         (umips_12bit_offset_address_p): Likewise.
26886         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26887         (mips_base_mips16): Rename this...
26888         (mips_base_compression_flags): ...to this. Update all uses.
26889         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26890         (mips_mips16_decl_p): Delete.
26891         (mips_nomips16_decl_p): Delete.
26892         (mips_get_compress_on_flags): New function.
26893         (mips_get_compress_off_flags): New function.
26894         (mips_get_compress_mode): New function.
26895         (mips_get_compress_on_name): New function.
26896         (mips_get_compress_off_name): New function.
26897         (mips_insert_attributes): Support multiple compression types.
26898         (mips_merge_decl_attributes): Likewise.
26899         (umips_12bit_offset_address_p): New function.
26900         (mips_start_function_definition): Emit .set micromips directive.
26901         (mips_call_may_need_jalx_p): New function.
26902         (mips_function_ok_for_sibcall): Add microMIPS support.
26903         (mips_print_operand_punctuation): Support short delay slots and
26904         compact jumps.
26905         (umips_swm_mask, umips_swm_encoding): New.
26906         (umips_build_save_restore): New function.
26907         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26908         (was_mips16_p): Remove.
26909         (old_compression_mode): New.
26910         (mips_set_compression_mode): New function.
26911         (mips_set_current_function): Add microMIPS support.
26912         (mips_option_override): Likewise.
26913         (umips_save_restore_pattern_p): New function.
26914         (umips_output_save_restore): New function.
26915         (umips_load_store_pair_p_1): New function.
26916         (umips_load_store_pair_p): New function.
26917         (umips_output_load_store_pair_1): New function.
26918         (umips_output_load_store_pair): New function.
26919         (umips_movep_target_p) New function.
26920         (mips_prepare_pch_save): Add microMIPS support.
26921         * config/mips/mips.h (TARGET_COMPRESSION): New.
26922         (TARGET_CPU_CPP_BUILTINS): Update macro
26923         to use new compression flags and to support microMIPS.
26924         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
26925         (MIPS_ARCH_FLOAT_SPEC): Likewise.
26926         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
26927         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
26928         (ASM_SPEC): Support mmicromips and mno-micromips.
26929         (M16STORE_REG_P): New macro.
26930         (MIPS_CALL): Support TARGET_MICROMIPS.
26931         (MICROMIPS_J): New macro.
26932         (mips_base_mips16): Rename this...
26933         (mips_base_compression_flags): ...to this.
26934         (UMIPS_12BIT_OFFSET_P): New macro.
26935         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
26936         (MULTILIB_DIRNAMES): Likewise.
26938 2013-02-25  Tom de Vries  <tom@codesourcery.com>
26940         PR rtl-optimization/56131
26941         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
26942         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
26943         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
26945 2013-02-25  Tobias Burnus  <burnus@net-b.de>
26947         * doc/invoke.texi (-fsanitize=): Move from optimization
26948         to debugging options.
26950 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26952         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
26954 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
26955             Alexander Monakov  <amonakov@ispras.ru>
26957         PR middle-end/56077
26958         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
26959         flush pending lists also on non-jumps.  Adjust comment.
26961 2013-02-24  Catherine Moore  <clm@codesourcery.com>
26962             Maciej W. Rozycki  <macro@codesourcery.com>
26963             Tom de Vries  <tom@codesourcery.com>
26964             Nathan Sidwell  <nathan@codesourcery.com>
26965             Iain Sandoe  <iain@codesourcery.com>
26966             Nathan Froyd  <froydnj@codesourcery.com>
26967             Chao-ying Fu  <fu@mips.com>
26969         * doc/extend.texi: (micromips, nomicromips, nocompression):
26970         Document new function attributes.
26971         * doc/invoke.texi (minterlink-compressed, mmicromips,
26972         m14k, m14ke, m14kec): Document new options.
26973         (minterlink-mips16): Update documentation.
26974         * doc/md.texi (ZC, ZD): Document new constraints.
26975         * configure.ac (gcc_cv_as_micromips): Check if linker
26976         supports the .set micromips directive.
26977         * configure: Regenerate.
26978         * config.in: Regenerate.
26979         * config/mips/mips-tables.opt: Regenerate.
26980         * config/mips/micromips.md: New file.
26981         * constraints.md (ZC, AD): New constraints.
26982         * config/mips/predicates.md (movep_src_register): New predicate.
26983         (movep_src_operand): New predicate.
26984         (non_volatile_mem_operand): New predicate.
26985         * config/mips/mips.md (multimem): New type.
26986         (length): Differentiate between 17-bit and 18-bit branch offsets.
26987         (MOVEP1, MOVEP2): New mode iterator.
26988         (mov_<load>l): Use ZC constraint.
26989         (mov_<load>r): Likewise.
26990         (mov_<store>l): Likewise.
26991         (mov_<store>r): Likewise.
26992         (*branch_equality<mode>_inverted): Add microMIPS support.
26993         (*branch_equality<mode>): Likewise.
26994         (*jump_absolute): Likewise.
26995         (indirect_jump_<mode>): Likewise.
26996         (tablejump_<mode>): Likewise.
26997         (<optab>_internal): Likewise.
26998         (sibcall_internal): Likewise.
26999         (sibcall_value_internal): Likewise.
27000         (prefetch): Use constraint ZD.
27001         * config/mips/mips.opt (minterlink-compressed): New option.
27002         (minterlink-mips16): Now an alias for minterlink-compressed.
27003         (mmicromips): New option.
27004         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
27005         (compare_and_swap_12): Likewise.
27006         (sync_add<mode>): Likewise.
27007         (sync_<optab>_12): Likewise.
27008         (sync_old_<optab>_12): Likewise.
27009         (sync_new_<optab>_12): Likewise.
27010         (sync_nand_12): Likewise.
27011         (sync_old_nand_12): Likewise.
27012         (sync_new_nand_12): Likewise.
27013         (sync_sub<mode>): Likewise.
27014         (sync_old_add<mode>): Likewise.
27015         (sync_old_sub<mode>): Likewise.
27016         (sync_new_add<mode>): Likewise.
27017         (sync_new_sub<mode>): Likewise.
27018         (sync_<optab><mode>): Likewise.
27019         (sync_old_<optab><mode>): Likewise.
27020         (sync_new_<optab><mode>): Likewise.
27021         (sync_nand<mode>): Likewise.
27022         (sync_old_nand<mode>): Likewise.
27023         (sync_new_nand<mode>): Likewise.
27024         (sync_lock_test_and_set<mode>): Likewise.
27025         (test_and_set_12): Likewise.
27026         (atomic_compare_and_swap<mode>): Likewise.
27027         (atomic_exchange<mode>_llsc): Likewise.
27028         (atomic_fetch_add<mode>_llsc): Likewise.
27029         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
27030         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
27031         (umips_save_restore_pattern_p): Likewise.
27032         (umips_load_store_pair_p): Likewise.
27033         (umips_output_load_store_pair): Likewise.
27034         (umips_movep_target_p): Likewise.
27035         (umips_12bit_offset_address_p): Likewise.
27036         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
27037         (mips_base_mips16): Rename this...
27038         (mips_base_compression_flags): ...to this. Update all uses.
27039         (mips_attribute_table): Add micromips, nomicromips and nocompression.
27040         (mips_mips16_decl_p): Delete.
27041         (mips_nomips16_decl_p): Delete.
27042         (mips_get_compress_on_flags): New function.
27043         (mips_get_compress_off_flags): New function.
27044         (mips_get_compress_mode): New function.
27045         (mips_get_compress_on_name): New function.
27046         (mips_get_compress_off_name): New function.
27047         (mips_insert_attributes): Support multiple compression types.
27048         (mips_merge_decl_attributes): Likewise.
27049         (umips_12bit_offset_address_p): New function.
27050         (mips_start_function_definition): Emit .set micromips directive.
27051         (mips_call_may_need_jalx_p): New function.
27052         (mips_function_ok_for_sibcall): Add microMIPS support.
27053         (mips_print_operand_punctuation): Support short delay slots and
27054         compact jumps.
27055         (umips_swm_mask, umips_swm_encoding): New.
27056         (umips_build_save_restore): New function.
27057         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
27058         (was_mips16_p): Remove.
27059         (old_compression_mode): New.
27060         (mips_set_compression_mode): New function.
27061         (mips_set_current_function): Add microMIPS support.
27062         (mips_option_override): Likewise.
27063         (umips_save_restore_pattern_p): New function.
27064         (umips_output_save_restore): New function.
27065         (umips_load_store_pair_p_1): New function.
27066         (umips_load_store_pair_p): New function.
27067         (umips_output_load_store_pair_1): New function.
27068         (umips_output_load_store_pair): New function.
27069         (umips_movep_target_p) New function.
27070         (mips_prepare_pch_save): Add microMIPS support.
27071         * config/mips/mips.h (TARGET_COMPRESSION): New.
27072         (TARGET_CPU_CPP_BUILTINS): Update macro
27073         to use new compression flags and to support microMIPS.
27074         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
27075         (MIPS_ARCH_FLOAT_SPEC): Likewise.
27076         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
27077         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
27078         (ASM_SPEC): Support mmicromips and mno-micromips.
27079         (M16STORE_REG_P): New macro.
27080         (MIPS_CALL): Support TARGET_MICROMIPS.
27081         (MICROMIPS_J): New macro.
27082         (mips_base_mips16): Rename this...
27083         (mips_base_compression_flags): ...to this.
27084         (UMIPS_12BIT_OFFSET_P): New macro.
27085         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
27086         (MULTILIB_DIRNAMES): Likewise.
27088 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
27090         PR target/52555
27091         * target-globals.c (save_target_globals): For init_reg_sets and
27092         target_reinit remporarily set this_fn_optabs to this_target_optabs.
27094 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
27096         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
27097         * config/aarch64/t-aarch64
27098         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
27100 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
27102         PR inline-asm/56148
27103         * lra-constraints.c (process_alt_operands): Reload operand
27104         conflicting with earlier clobber only if no more other conflicting
27105         operands.
27107 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
27109         PR sanitizer/56393
27110         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
27111         if not linking a shared library.
27113 2013-02-22  Seth LaForge  <sethml@google.com>
27115         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
27117 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
27119         * config/arm/arm.md (split for extendsidi): Update condition.
27120         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
27121         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
27122         (qhs_zextenddi_cstr): Likewise.
27124 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27126         PR middle-end/56420
27127         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
27128         avoid signed wrapping.
27129         (expand_mult): Handle properly multiplication by
27130         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
27131         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
27132         in the compiler if coeff is HOST_WIDE_INT_MIN.
27133         (expand_divmod): Don't make ext_op1 static, change it's type to
27134         uhwi.  Avoid undefined behavior in -INTVAL (op1).
27136         PR rtl-optimization/50339
27137         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
27138         field.
27139         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
27140         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
27141         into splitting_ashiftrt field.
27142         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
27143         ASHIFTRT.
27144         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
27145         choices.
27147 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
27149         PR middle-end/56108
27150         * trans-mem.c (execute_tm_mark): Do not expand transactions that
27151         are sure to go irrevocable.
27153 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
27155         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
27156         scalars are valid operands.
27158 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27160         PR tree-optimization/56310
27161         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
27162         only matching indices and non-negative final offsets.
27163         (intersect_aggregates_with_edge): Pass src_idx to
27164         agg_replacements_to_vector.  Pass src_idx insstead of index to
27165         intersect_with_agg_replacements.
27167 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27169         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
27170         instead of hard-wired defaults.
27172 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
27174         * doc/invoke.texi (MIPS Options): Update documentation of the
27175         floating-point multiply-accumulate instruction restrictions.
27177 2013-02-21  Kostya Serebryany  <kcc@google.com>
27179         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
27180         asan_shadow_offset on x86_64 linux.
27182 2013-02-21  Richard Biener  <rguenther@suse.de>
27184         PR tree-optimization/56415
27185         Revert
27186         2013-02-11  Richard Biener  <rguenther@suse.de>
27188         PR tree-optimization/56273
27189         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27190         first VRP run.
27192 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27194         PR bootstrap/56258
27195         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
27196         instead of @itemx.
27198         PR inline-asm/56405
27199         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
27200         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
27202 2013-02-20  Jan Hubicka  <jh@suse.cz>
27204         PR tree-optimization/56265
27205         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
27206         when target is referenced for first time.
27208 2013-02-20  Richard Biener  <rguenther@suse.de>
27210         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
27211         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
27212         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
27213         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
27214         not return anything.
27215         (rename_ssa_copies): Do not remove unused locals.
27216         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
27217         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
27218         * passes.c (execute_function_todo): Do not schedule unused locals
27219         removal if cleanup_tree_cfg did something.
27220         * tree-ssa-live.c (remove_unused_locals): Dump statistics
27221         about the number of removed locals.
27223 2013-02-20  Richard Biener  <rguenther@suse.de>
27225         PR tree-optimization/56398
27226         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
27228 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27230         PR tree-optimization/55334
27231         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
27232         restricted pointers to arrays.
27234 2013-02-20  Richard Biener  <rguenther@suse.de>
27235             Jakub Jelinek  <jakub@redhat.com>
27237         PR tree-optimization/56396
27238         * tree-ssa-ccp.c (n_const_val): New static variable.
27239         (get_value): Return NULL for SSA names we don't have a lattice
27240         entry for.
27241         (ccp_initialize): Initialize n_const_val.
27242         * tree-ssa-copy.c (n_copy_of): New static variable.
27243         (init_copy_prop): Initialize n_copy_of.
27244         (get_value): Return NULL_TREE for SSA names we don't have a
27245         lattice entry for.
27247 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27249         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
27251 2013-02-20  Richard Biener  <rguenther@suse.de>
27253         * genpreds.c (write_lookup_constraint): Do not compare first
27254         letter of the constraint again.
27256 2013-02-20  Richard Biener  <rguenther@suse.de>
27258         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
27259         and ceil_log2.
27260         (get_use_iv_cost): Terminate hashtable walk when coming across
27261         an empty entry.
27263 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
27265         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
27266         reassociation for avx2 targets.
27268 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27270         * config/microblaze/microblaze.c: microblaze_has_clz = 0
27271         Add version check for v8.10.a to enable microblaze_has_clz
27272         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
27273         version and TARGET_PATTERN_COMPARE check
27274         * config/microblaze/microblaze.md: New clzsi2 instruction
27276 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27278         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
27279         function before branching.
27281 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27283         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
27284         DUMP_INSN_RTX_UID.
27285         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
27287 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27289         PR middle-end/55889
27290         * sel-sched.c: Include ira.h.
27291         (implicit_clobber_conflict_p): New function.
27292         (moveup_expr): Use it.
27293         * Makefile.in (sel-sched.o): Depend on ira.h.
27295 2013-02-19  Richard Biener  <rguenther@suse.de>
27297         PR tree-optimization/56384
27298         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
27299         (vn_hash_type): Split out from ...
27300         (vn_hash_constant_with_type): ... here.
27301         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
27302         (vn_phi_eq): Compare types from vn_phi_s structure.
27303         (vn_phi_lookup): Populate vn_phi_s type.
27304         (vn_phi_insert): Likewise.
27306 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
27308         PR tree-optimization/56350
27309         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
27310         if haven't found reduction or nested cycle operand, rather than
27311         asserting we must find it.
27313         PR tree-optimization/56381
27314         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
27315         to fold_build3.
27317 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
27318             Jakub Jelinek  <jakub@redhat.com>
27320         PR target/52555
27321         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
27322         (swap_optab_enable): Same.
27323         (init_all_optabs): Use argument instead of global.
27324         * tree.h (struct tree_optimization_option): New field target_optabs.
27325         * expr.h (init_all_optabs): Add argument to prototype.
27326         (TREE_OPTIMIZATION_OPTABS): New.
27327         (save_optabs_if_changed): Protoize.
27328         * optabs.h: Declare this_fn_optabs.
27329         * optabs.c (save_optabs_if_changed): New.
27330         Declare this_fn_optabs.
27331         (init_optabs): Add argument to init_all_optabs() call.
27332         * function.c (invoke_set_current_function_hook): Handle per
27333         function optabs.
27334         * function.h (struct function): New field optabs.
27335         * config/mips/mips.c (mips_set_mips16_mode): Handle when
27336         optimization_current_node has changed.
27337         * target-globals.h (save_target_globals_default_opts): Protoize.
27338         * target-globals.c (save_target_globals_default_opts): New.
27340 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27342         PR target/56347
27343         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
27344         registers %fr12 and %fr12R as call used.
27346         PR target/56214
27347         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
27348         and HImode, require all displacements to be an integer multiple of
27349         their mode size.
27350         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
27351         only allow QImode and HImode when reload is in progress and strict is
27352         true.  Likewise for symbolic addresses.  Use base14_operand to check
27353         displacements in REG+BASE addresses.
27355 2013-02-18  Richard Biener  <rguenther@suse.de>
27357         PR tree-optimization/56366
27358         * tree-vect-loop.c (get_initial_def_for_induction): Properly
27359         handle sign-conversion of outer-loop initial induction value.
27361 2013-02-18  Richard Biener  <rguenther@suse.de>
27363         PR middle-end/56349
27364         * cfghooks.c (merge_blocks): If we merge a latch into another
27365         block adjust references to it.
27366         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
27367         (verify_loop_structure): Verify that a recorded latch is in fact
27368         a latch.
27370 2013-02-18  Richard Biener  <rguenther@suse.de>
27372         PR tree-optimization/56321
27373         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
27374         order SSA name release and virtual operand unlinking.
27376 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27378         * config/microblaze/microblaze.md (save_stack_block): Define.
27379         (restore_stack_block): Likewise.
27381 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27383         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
27384         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
27385         * config/microblaze/microblaze.c (microblaze_option_override):
27386         Bail out early for PIC modes when target does not support PIC.
27388 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27390         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
27391         Replace with a microblaze version.
27392         (microblaze_trampoline_init): Adapt for microblaze.
27393         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
27394         microblaze.
27396 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
27397             Dodji Seketeli  <dodji@redhat.com>
27399         PR asan/56330
27400         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
27401         (instrument_mem_region_access): Do not forget to always put
27402         instrumentation of the of 'base' and 'base + len' in a "if (len !=
27403         0) statement, even for cases where either 'base' or 'base + len'
27404         are not instrumented -- because they have been previously
27405         instrumented.  Simplify the logic by putting all the statements
27406         instrument 'base + len' inside a sequence, and then insert that
27407         sequence right before the current insertion point.  Then, to
27408         instrument 'base + len', just get an iterator on that statement.
27409         And do not forget to update the pointer to iterator the function
27410         received as argument.
27412 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
27414         PR rtl-optimization/56348
27415         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
27417 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
27419         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
27420         (clean_graph_dump_file): Pass base to start_graph_dump.
27422 2013-02-14  Richard Henderson  <rth@redhat.com>
27424         PR target/55941
27425         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
27427 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
27429         * collect2-aix.h: Define F_LOADONLY.
27431 2013-02-14  Richard Biener  <rguenther@suse.de>
27433         PR lto/50494
27434         * varasm.c (output_constant_def_1): Get the decl representing
27435         the constant as argument.
27436         (output_constant_def): Wrap output_constant_def_1.
27437         (make_decl_rtl): Use output_constant_def_1 with the decl
27438         representing the constant.
27439         (build_constant_desc): Optionally re-use a decl already
27440         representing the constant.
27441         (tree_output_constant_def): Adjust.
27443 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
27445         Fix an asan crash
27446         * asan.c (instrument_builtin_call):  Really put the length of the
27447         second source argument into src1_len.
27449 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27451         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
27452         argument.  If it is false, don't create edge from then_bb to
27453         fallthru_bb.
27454         (insert_if_then_before_iter): Pass true to it.
27455         (build_check_stmt): Pass false to it.
27456         (transform_statements): Flush hash table only on extended basic
27457         block boundaries, rather than at the beginning of every bb.
27458         Don't flush hash table on nonfreeing_call_p calls.
27459         * tree-flow.h (nonfreeing_call_p): New prototype.
27460         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
27462 2013-02-13  David S. Miller  <davem@davemloft.net>
27464         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
27466 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
27468         PR target/56184
27469         * ira.c (max_regno_before_ira): Move from ...
27470         (ira): ... here.
27471         (fix_reg_equiv_init): Use max_regno_before_ira instead of
27472         vec_safe_length.
27474 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27476         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
27478 2013-02-13  Richard Biener  <rguenther@suse.de>
27480         PR lto/56295
27481         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
27482         globals in MEM_REFs.
27484 2013-02-13  Richard Biener  <rguenther@suse.de>
27486         * loop-init.c (loop_optimizer_init): Clear loop state when
27487         re-initializing preserved loops.
27488         * loop-unswitch.c (unswitch_single_loop): Return whether
27489         we unswitched the loop.  Do not verify loop state here.
27490         (unswitch_loops): When we unswitched a loop discover new loops.
27492 2013-02-13  Kostya Serebryany  <kcc@google.com>
27494         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
27495         on x86_64 linux.
27496         * sanitizer.def: Rename __asan_init to __asan_init_v1.
27498 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
27500         Avoid instrumenting duplicated memory access in the same basic block
27501         * Makefile.in (asan.o): Add new dependency on hash-table.h
27502         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
27503         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
27504         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
27505         (free_mem_ref_resources, has_mem_ref_been_instrumented)
27506         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
27507         (get_mem_ref_of_assignment): New functions.
27508         (get_mem_refs_of_builtin_call): Extract from
27509         instrument_builtin_call and tweak a little bit to make it fit with
27510         the new signature.
27511         (instrument_builtin_call): Use the new
27512         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
27513         of is_gimple_builtin_call.
27514         (instrument_derefs, instrument_mem_region_access): Insert the
27515         instrumented memory reference into the hash table.
27516         (maybe_instrument_assignment): Renamed instrument_assignment into
27517         this, and change it to advance the iterator when instrumentation
27518         actually happened and return true in that case.  This makes it
27519         homogeneous with maybe_instrument_assignment, and thus give a
27520         chance to callers to be more 'regular'.
27521         (transform_statements): Clear the memory reference hash table
27522         whenever we enter a new BB, when we cross a function call, or when
27523         we are done transforming statements.  Use
27524         maybe_instrument_assignment instead of instrumentation.  No more
27525         need to special case maybe_instrument_assignment and advance the
27526         iterator after calling it; it's now handled just like
27527         maybe_instrument_call.  Update comment.
27529 2013-02-13  Richard Biener  <rguenther@suse.de>
27531         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
27532         Fix loop discovery code.
27534 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
27536         PR inline-asm/56148
27537         * lra-constraints.c (process_alt_operands): Match early clobber
27538         operand with itself.  Check conflicts with earlyclobber only if
27539         the operand is not reloaded.  Prefer to reload conflicting operand
27540         if earlyclobber and matching operands are the same.
27542 2013-02-12  Richard Biener  <rguenther@suse.de>
27544         PR lto/56297
27545         * lto-streamer-out.c (write_symbol): Do not output symbols
27546         for hard register variables.
27548 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
27550         PR target/54222
27551         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
27552         (umulsidi3_insn, mulsidi3_insn): New insns.
27554 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
27556         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
27557         (struct tune_params): Add vec_costs field.
27558         * config/arm/arm.c (arm_builtin_vectorization_cost)
27559         (arm_add_stmt_cost): New functions.
27560         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
27561         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
27562         (arm_default_vec_cost): New struct of type cpu_vec_costs.
27563         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
27564         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
27565         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
27566         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
27568 2013-02-12  Richard Biener  <rguenther@suse.de>
27570         PR lto/56295
27571         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
27572         decls again if possible.
27574 2013-02-12  Richard Biener  <rguenther@suse.de>
27576         PR middle-end/56288
27577         * tree-ssa.c (verify_ssa_name): Fix check, move
27578         SSA_NAME_IN_FREE_LIST check up.
27580 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
27581             Steven Bosscher   <steven@gcc.gnu.org>
27583         PR rtl-optimization/56151
27584         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
27585         equal to op0 or op1, and last_insn pattern is CODE operation
27586         with MEM dest and one of the operands matches that MEM.
27588 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
27590         * doc/extend.texi: Document Function Multiversioning and "default"
27591         parameter string to target attribute.
27592         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
27593         target attribute parameter is "default".
27594         (ix86_compare_version_priority): Remove checks for target attribute.
27595         (ix86_mangle_function_version_assembler_name): Change error to sorry.
27596         Remove check for target attribute equal to NULL. Add assert.
27597         (ix86_generate_version_dispatcher_body): Change error to sorry.
27599 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
27600             Jack Howarth  <howarth@bromo.med.uc.edu>
27601             Patrick Marlier  <patrick.marlier@gmail.com>
27603         PR libitm/55693
27604         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
27605         define ENDFILE_SPEC as TM_DESTRUCTOR.
27606         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
27608 2013-02-11  Alexander Potapenko  <glider@google.com>
27609             Jack Howarth  <howarth@bromo.med.uc.edu>
27610             Jakub Jelinek  <jakub@redhat.com>
27612         PR sanitizer/55617
27613         * config/darwin.c (cdtor_record): Rename ctor_record.
27614         (sort_cdtor_records): Rename sort_ctor_records.
27615         (finalize_dtors): New routine to sort destructors by
27616         priority before use in assemble_integer.
27617         (machopic_asm_out_destructor): Use finalize_dtors if needed.
27619 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
27621         PR rtl-optimization/56275
27622         * simplify-rtx.c (avoid_constant_pool_reference): Check that
27623         offset is non-negative and less than cmode size before
27624         calling simplify_subreg.
27626 2013-02-11  Richard Biener  <rguenther@suse.de>
27628         PR tree-optimization/56264
27629         * cfgloop.h (fix_loop_structure): Adjust prototype.
27630         * loop-init.c (fix_loop_structure): Return the number of
27631         newly discovered loops.
27632         * tree-cfgcleanup.c (repair_loop_structures): When new loops
27633         are discovered, do a full loop-closed SSA rewrite.
27635 2013-02-11  Richard Biener  <rguenther@suse.de>
27637         PR tree-optimization/56273
27638         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27639         first VRP run.
27640         (check_array_ref): Fix missing newline in dumps.
27641         (search_for_addr_array): Likewise.
27643 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
27645         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
27647 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
27649         PR target/56256
27650         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
27652 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
27654         PR rtl-optimization/56246
27655         * lra-constraints.c (simplify_operand_subreg): Try to reuse
27656         reload pseudo.
27657         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
27658         constraints are satisfied.
27660 2013-02-08  Jeff Law  <law@redhat.com>
27662         PR debug/53948
27663         * emit-rtl.c (reg_is_parm_p): New function.
27664         * regs.h (reg_is_parm_p): New prototype.
27665         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
27666         callee-clobbered registers.
27668 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
27670         PR target/56043
27671         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
27672         If there is no implicit builtin declaration, just return NULL.
27674 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27676         * config/i386/sse.md (FMAMODEM): New mode iterator.
27677         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
27678         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
27680 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27682         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
27683         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
27684         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
27686 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27688         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
27689         (microblaze*-*-elf): Likewise.
27690         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
27691         LINK_SPEC.
27692         * config/microblaze/microblaze-c.c: Add builtin defines for
27693         _LITTLE_ENDIAN and _BIG_ENDIAN.
27694         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
27695         add to TARGET_DEFAULT flags.
27696         Expand ASM_SPEC and LINK_SPEC.
27697         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
27698         * config/microblaze/microblaze.md: Update extendsidi2 and
27699         movdi_internal instructions to use low-order / high-order reg
27700         print_operands.
27701         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
27702         options and inversemask / mask of LITTLE_ENDIAN.
27703         * config/microblaze/t-microblaze: Expand multilib options to
27704         include mlittle-endian (le) and update exceptions patterns.
27706 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
27708         PR rtl-optimization/56195
27709         * lra-constraints.c (get_reload_reg): Don't reuse regs
27710         if they have smaller mode than requested, if they have
27711         wider mode than requested, try to return a SUBREG.
27713         PR tree-optimization/56250
27714         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
27715         if type is unsigned and code isn't MULT_EXPR.
27717 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27719         PR tree-optimization/56064
27720         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
27721         bits according to mode.
27722         * fixed-value.h (fixed_from_double_int)
27723         (const_fixed_from_double_int): Adjust comments.
27725 2013-02-08  Richard Biener  <rguenther@suse.de>
27727         PR lto/56231
27728         * lto-streamer.h (struct data_in): Remove current_file, current_line
27729         and current_col members.
27730         * lto-streamer-out.c (lto_output_location): Stream changed bits
27731         en-block for efficiency.
27732         * lto-streamer-in.c (clear_line_info): Remove.
27733         (lto_input_location): Cache current file, line and column
27734         globally via local statics.  Read changed bits en-block.
27735         (input_function): Do not call clear_line_info.
27736         (lto_read_body): Likewise.
27737         (lto_input_toplevel_asms): Likewise.
27739 2013-02-08  Michael Matz  <matz@suse.de>
27741         PR tree-optimization/52448
27742         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
27743         (nt_call_phase): New static.
27744         (add_or_mark_expr): Only mark accesses with newer phase than any
27745         call seen.
27746         (nonfreeing_call_p): New.
27747         (nt_init_block): Update nt_call_phase, mark blocks as visited.
27748         (nt_fini_block): Keep blocks marked as visited.
27749         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
27751 2013-02-08  Richard Biener  <rguenther@suse.de>
27753         * ira.c (ira): Free broken dominator information.
27755 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27757         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
27759 2013-02-08  Marek Polacek  <polacek@redhat.com>
27761         * cfgloop.c (verify_loop_structure): Add more checking of headers.
27763 2013-02-08  Richard Biener  <rguenther@suse.de>
27765         PR middle-end/56181
27766         * cfgloop.h (flow_loops_find): Adjust.
27767         (bb_loop_header_p): Declare.
27768         * cfgloop.c (bb_loop_header_p): New function split out from ...
27769         (flow_loops_find): ... here.  Adjust function signature,
27770         support incremental loop structure update.
27771         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
27772         * cfgloopmanip.c (fix_loop_structure): Move ...
27773         * loop-init.c (fix_loop_structure): ... here.
27774         (apply_loop_flags): Split out from ...
27775         (loop_optimizer_init): ... here.
27776         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
27777         in incremental mode, only remove dead loops here.
27779 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27781         PR target/54222
27782         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
27783         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
27784         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
27785         (*round<mode>3.libgcc): New insns for fixed-modes.
27786         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
27787         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
27788         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
27789         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
27790         implementations.  Define to __builtin_avr_absFX,
27791         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
27792         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
27793         __builtin_avr_countlsFX, respectively.
27794         * config/avr/avr-c.c (target.h): Include it.
27795         (enum avr_builtin_id): New enum.
27796         (avr_resolve_overloaded_builtin): New static function.
27797         (avr_register_target_pragmas): Use it to set
27798         targetm.resolve_overloaded_builtin.
27799         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
27800         tree nodes used by DEF_BUILTIN.
27801         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
27802         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
27803         <AVR_BUILTIN_xxBITS>: Same.
27805 2013-02-08  Richard Biener  <rguenther@suse.de>
27807         * cfgloop.c (verify_loop_structure): Properly handle
27808         a loop exiting to another loop header.
27809         * ira-int.h (ira_loops): Remove.
27810         * ira.c (ira_loops): Remove.
27811         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
27812         (do_reload): Use loop_optimizer_finalize.
27813         * ira-build.c (create_loop_tree_nodes): Use get_loops and
27814         number_of_loops to access the loop tree.
27815         (more_one_region_p): Likewise.
27816         (finish_loop_tree_nodes): Likewise.
27817         (rebuild_regno_allocno_maps): Likewise.
27818         (mark_loops_for_removal): Likewise.
27819         (mark_all_loops_for_removal): Likewise.
27820         (remove_unnecessary_regions): Likewise.
27821         (ira_build): Likewise.
27822         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
27824 2013-02-08  Richard Biener  <rguenther@suse.de>
27826         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
27827         * ipa-pure-const.c (analyze_function): Avoid calling
27828         mark_irreducible_loops twice.
27829         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
27831 2013-02-07  David S. Miller  <davem@davemloft.net>
27833         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
27834         on 'reg'.
27835         * var-tracking.c (vt_add_function_parameter): Test the presence of
27836         HAVE_window_save properly and do not remap argument registers when
27837         we have a leaf function.
27839 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
27841         PR bootstrap/56227
27842         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
27843         instead of "ll".
27844         * config/i386/i386.c (ix86_print_operand): Ditto.
27846 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27848         * lra-constraints.c (process_alt_operands): Fix recently added comment.
27850 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27852         PR rtl-optimization/56225
27853         * lra-constraints.c (process_alt_operands): Check that reload hard
27854         reg can hold value for strict_low_part.
27856 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27858         PR debug/56154
27859         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
27860         dwarf2out_end_function.
27861         (in_first_function_p, maybe_at_text_label_p,
27862         first_loclabel_num_not_at_text_label): New variables.
27863         (dwarf2out_var_location): In the first function find out
27864         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
27865         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
27866         functions.
27868 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
27870         PR rtl-optimization/56178
27871         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
27872         SUBREG of a register.  Tidy up related block of code.
27873         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
27874         note if the source is a register or a SUBREG of a register.
27876 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27878         PR target/56228
27879         * config/rs6000/rs6000.md (ptrm): New mode attr.
27880         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
27881         call_value_indirect_aix<pttrsize>,
27882         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
27883         m in constraints.
27885 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
27887         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
27888         if -bnortl. Convert to strcmp and strncmp.
27890 2013-02-07  Alan Modra  <amodra@gmail.com>
27892         PR target/54009
27893         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
27894         addresses won't wrap when offsetting.
27895         (rs6000_secondary_reload): Provide secondary reloads needed for
27896         wrapping LO_SUM addresses.
27898 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
27900         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
27901         MACH, just __MACH__.
27903 2013-02-06  Richard Biener  <rguenther@suse.de>
27905         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
27906         instead of calling fix_loop_structure.
27908 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
27910         PR middle-end/56217
27911         * omp-low.c (use_pointer_for_field): Return false if
27912         lower_send_shared_vars doesn't generate any copy-out code.
27914 2013-02-06  Tom de Vries  <tom@codesourcery.com>
27916         PR rtl-optimization/56131
27917         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
27918         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
27919         of the label is NULL.  Add comment.
27921 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
27923         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
27925         PR sanitizer/55374
27926         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
27927         (STATIC_LIBTSAN_LIBS): Likewise.
27928         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
27929         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
27930         is defined, don't add anything else beyond that.
27931         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
27932         (LINK_COMMAND_SPEC): Use them.
27934         PR tree-optimization/56205
27935         * tree-stdarg.c (check_all_va_list_escapes): Return true if
27936         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
27937         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
27939 2013-02-05  Richard Biener  <rguenther@suse.de>
27941         PR tree-optimization/53342
27942         PR tree-optimization/53185
27943         * tree-vectorizer.h (vect_check_strided_load): Remove.
27944         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
27945         not disallow peeling for vectorized strided loads.
27946         (vect_check_strided_load): Make static and simplify.
27947         (vect_analyze_data_refs): Adjust.
27948         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
27949         correctly when vectorizing strided loads.
27951 2013-02-05  Richard Biener  <rguenther@suse.de>
27953         * doc/install.texi: Refer to ISL, not PPL.
27955 2013-02-05  Jan Hubicka  <jh@suse.cz>
27957         PR tree-optimization/55789
27958         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
27960 2013-02-05  Jan Hubicka  <jh@suse.cz>
27962         PR tree-optimization/55789
27963         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
27964         the dead call anyway.
27966 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
27968         PR sanitizer/55374
27969         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
27971 2013-02-04  Alexander Potapenko  <glider@google.com>
27972             Jack Howarth  <howarth@bromo.med.uc.edu>
27973             Jakub Jelinek  <jakub@redhat.com>
27975         PR sanitizer/55617
27976         * config/darwin.c (sort_ctor_records): Stabilized qsort
27977         on constructor priority by using original position.
27978         (finalize_ctors): New routine to sort constructors by
27979         priority before use in assemble_integer.
27980         (machopic_asm_out_constructor): Use finalize_ctors if needed.
27982 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
27984         PR libstdc++/54314
27985         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
27986         about visibility on artificial decls.
27987         * config/sol2.c (solaris_assemble_visibility): Likewise.
27989 2013-02-04  Kai Tietz  <ktietz@redhat.com>
27991         PR target/56186
27992         * config/i386/i386.c (function_value_ms_64): Add additional valtype
27993         argument and improve checking of return-argument types for 16-byte
27994         modes.
27995         (ix86_function_value_1): Add additional valtype argument on call
27996         of function_value_64.
27997         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
27998         handling infunction_value_64 function.
28000 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
28002         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
28004 2013-02-04  Richard Biener  <rguenther@suse.de>
28006         PR tree-optimization/56188
28007         * tree-ssa-structalias.c (label_visit): Consider case with
28008         initially non-empty points-to set.
28009         (perform_var_substitution): Dump node mapping and clean up.
28011 2013-02-04  Richard Guenther  <rguenther@suse.de>
28013         PR lto/56168
28014         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
28015         node prevail as last resort.
28016         (lto_symtab_merge_decls): Remove guard on LTRANS here.
28017         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
28019 2013-02-04  Richard Biener  <rguenther@suse.de>
28021         PR tree-optimization/56113
28022         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
28023         Merge into ...
28024         (equiv_class_lookup_or_add): ... this.
28025         (label_visit): Adjust and fix error in previous patch.
28026         (perform_var_substitution): Adjust.
28028 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
28030         * config/sh/divtab.c: Fix formatting and comments throughout the file.
28031         * config/sh/sh4-300.md: Likewise.
28032         * config/sh/sh4a.md: Likewise.
28033         * config/sh/constraints.md: Likewise.
28034         * config/sh/sh.md: Likewise.
28035         * config/sh/netbsd-elf.h: Likewise.
28036         * config/sh/predicates.md: Likewise.
28037         * config/sh/sh-protos.h: Likewise.
28038         * config/sh/ushmedia.h: Likewise.
28039         * config/sh/linux.h: Likewise.
28040         * config/sh/sh.c: Likewise.
28041         * config/sh/superh.h: Likewise.
28042         * config/sh/elf.h: Likewise.
28043         * config/sh/sh4.md: Likewise.
28044         * config/sh/sh.h: Likewise.
28046 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28048         * config/pa/constraints.md: Adjust unused letters.  Change "T"
28049         constraint to match_test floating_point_store_memory_operand().
28050         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
28051         (base14_operand): New.
28052         (floating_point_store_memory_operand): New.
28053         (integer_store_memory_operand): Revise to use base14_operand and
28054         reg_plus_base_memory_operand.
28055         (move_dest_operand): Allow symbolic_memory_operands.
28056         (symbolic_memory_operand): Check for LO_SOM.
28057         (symbolic_operand): Change default case to break.
28058         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
28059         CONST_DOUBLE values to be reloaded by putting them into memory when
28060         the destination is a floating point register.
28061         (movdf): Remove code to handle CONST_DOUBLE.
28062         (movsf): Likewise.
28063         (reload_indf_r1): New.
28064         (reload_insf_r1): New.
28065         Consistently use "Q" and "T" constraints with integer and floating
28066         point move instructions, respectively.
28067         (movdi): Remove FAIL.
28068         Change predicate for source operand unamed DImode move from
28069         general_operand to move_src_operand.
28070         (umulsidi3): Change predicate for destination operand to
28071         register_operand.
28072         Likewise for similar unamed patterns.
28073         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
28074         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
28075         (hppa_legitimize_address): Simplify mask calculation.
28076         (pa_emit_move_sequence): Revised handling of secondary reloads from
28077         REG+D addresses for floating point loads and stores.  Directly handle
28078         loading CONST0_RTX (mode) to a floating point register.
28079         (pa_secondary_reload): Handle reloading DF and SFmode constant values
28080         to floating point registers.  Don't restrict secondary reloads to
28081         floating point registers to integer modes.  Revise some comments and
28082         cleanup some code.
28083         (TARGET_LEGITIMATE_ADDRESS_P): Define.
28084         (pa_legitimate_address_p): New.
28085         (pa_legitimize_reload_address): New.
28086         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
28087         (STRICT_REG_OK_FOR_BASE_P): New.
28088         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
28089         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
28091 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
28092             Andrew Dixie  <andrewd@gentrack.com>
28094         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
28095         flag set.
28097 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
28099         * expmed.c (extract_bit_field_1): Pass the full width of the
28100         structure to get_best_reg_extraction_insn.
28102 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28104         PR target/54601
28105         * configure.ac (use_cxa_atexit): Add AIX.
28106         * configure: Regenerate.
28108         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
28110 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
28112         PR debug/54793
28113         * final.c (need_profile_function): New variable.
28114         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
28115         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
28116         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
28117         notes, targetm.asm_out.function_prologue doesn't emit anything,
28118         HAVE_prologue and profiler should be emitted before prologue,
28119         set need_profile_function instead of emitting it.
28120         (final_scan_insn): If need_profile_function, emit
28121         profile_function on the first NOTE_INSN_BASIC_BLOCK or
28122         NOTE_INSN_FUNCTION_BEG note.
28124 2013-02-01  Richard Henderson  <rth@redhat.com>
28126         * config/rs6000/rs6000.md (smulditi3): New.
28127         (umulditi3): New.
28129         * config/alpha/alpha.md (umulditi3): New.
28131 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28133         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
28134         (ASM_OUTPUT_ALIGNED_LOCAL): New.
28136 2013-02-01  Richard Biener  <rguenther@suse.de>
28138         PR tree-optimization/56113
28139         * tree-ssa-structalias.c (label_visit): Reduce work for
28140         single-predecessor nodes.
28142 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
28144         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
28145         range isn't testing for zero.
28147 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
28149         PR middle-end/56113
28150         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
28152 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
28153             Nick Clifton  <nickc@redhat.com>
28155         * config/v850/constraints.md (Q): Define as a memory constraint.
28156         * config/v850/predicates.md (label_ref_operand): New predicate.
28157         (e3v5_shift_operand): New predicate.
28158         (ior_operator): New predicate.
28159         * config/v850/t-v850: Add e3v5 multilib.
28160         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
28161         (v850_gen_movdi): Prototype.
28162         * config/v850/v850.c: Add support for e3v5 architecture.
28163         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
28164         TARGET_V850E_UP.
28165         (construct_save_jarl): Add e3v5 long JARL support.
28166         (v850_adjust_insn_length): New function.  Adjust length of call
28167         insns when using e3v5 instructions.
28168         (v850_gen_movdi): New function: Generate instructions to move a
28169         DImode value.
28170         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
28171         (CPP_SPEC): Define __v850e3v5__ as appropriate.
28172         (TARGET_USE_FPU): Enable for e3v5.
28173         (CONST_OK_FOR_W): New macro.
28174         (ADJUST_INSN_LENGTH): Define.
28175         * config/v850/v850.md (UNSPEC_LOOP): Define.
28176         (attr cpu): Add v850e3v5.
28177         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
28178         (movdi): New pattern.
28179         (movdi_internal): New pattern.
28180         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
28181         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
28182         (cstoresf4): Likewise.
28183         (cstoredf4): Likewise.
28184         (insv): New pattern.
28185         (rotlso3_a): New pattern.
28186         (rotlsi3_b): New pattern
28187         (rotlsi3_v850e3v5): New pattern.
28188         (doloop_begin): New pattern.
28189         (fix_loop_counter): New pattern.
28190         (doloop_end): New pattern.
28191         (branch_normal): Add e3v5 long branch support.
28192         (branch_invert): Likewise.
28193         (branch_z_normal): Likewise.
28194         (branch_z_invert): Likewise.
28195         (branch_nz_normal): Likewise.
28196         (branch_nz_invert): Likewise.
28197         (call_internal_short): Add e3v5 register-indirect JARL support.
28198         (call_internal_long): Likewise.
28199         (call_value_internal_short): Likewise.
28200         (call_value_internal_long): Likewise.
28201         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
28202         (mloop): New option.
28203         * config.gcc: Add support for configuring v840e3v5 target.
28204         * doc/invoke.texi: Document new v850 specific command line options.
28206 2013-01-31  Paul Koning  <ni1d@arrl.net>
28208         PR debug/55059
28209         PR debug/54508
28210         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
28211         children if parent is a class.
28212         (prune_unused_types_prune): Don't add DW_AT_declaration.
28214 2013-01-31  Richard Biener  <rguenther@suse.de>
28216         PR tree-optimization/56157
28217         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
28218         match up operand with SLP child.
28220 2013-01-31  Jason Merrill  <jason@redhat.com>
28222         PR debug/54410
28223         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
28224         parameters the first time.
28225         (gen_scheduled_generic_parms_dies): Check completeness here.
28227 2013-01-31  Richard Biener  <rguenther@suse.de>
28229         PR middle-end/53073
28230         * common.opt (faggressive-loop-optimizations): New flag,
28231         enabled by default.
28232         * doc/invoke.texi (faggressive-loop-optimizations): Document.
28233         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
28234         infer_loop_bounds_from_undefined by it.
28236 2013-01-31  Richard Biener  <rguenther@suse.de>
28238         PR tree-optimization/56150
28239         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
28240         visit virtual operands.
28241         (find_uses_to_rename_bb): Likewise.
28243 2013-01-31  Richard Biener  <rguenther@suse.de>
28245         PR tree-optimization/56150
28246         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
28247         mixed store non-store stmts.
28249 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
28251         PR sanitizer/55374
28252         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
28253         LIBASAN_EARLY_SPEC is defined.
28254         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
28255         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
28256         before %o.
28257         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
28259         PR c++/55742
28260         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
28261         invalid args instead of ICEing on it.
28262         (ix86_valid_target_attribute_tree): Return error_mark_node if
28263         ix86_valid_target_attribute_inner_p failed.
28264         (ix86_valid_target_attribute_p): Return false only if
28265         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
28266         target("default") attribute.
28267         (sorted_attr_string): Change argument from const char * to tree,
28268         merge in all target attribute arguments rather than just one.
28269         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
28270         instead of free.  Avoid using strcat.
28271         (ix86_mangle_function_version_assembler_name): Mangle
28272         target("default") as if no target attribute is present.  Adjust
28273         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
28274         instead of xmalloc and XDELETEVEC instead of free.
28275         (ix86_function_versions): Don't return true if one of the decls
28276         doesn't have target attribute.  If they don't and one of the decls
28277         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
28278         sorted_attr_string caller.  Use XDELETEVEC instead of free.
28279         (ix86_supports_function_versions): Remove.
28280         (make_name): Fix up formatting.
28281         (make_dispatcher_decl): Remove resolver_name and its initialization.
28282         Avoid leaking memory.
28283         (is_function_default_version): Return true if there is
28284         target("default") attribute rather than no target attribute at all.
28285         (make_resolver_func): Avoid leaking memory.
28286         (ix86_generate_version_dispatcher_body): Likewise.
28287         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
28288         * target.def (supports_function_versions): Remove.
28289         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
28290         * doc/tm.texi: Regenerated.
28292 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
28294         PR rtl-optimization/56144
28295         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
28296         for values with side effects.
28298 2013-01-30  Richard Biener  <rguenther@suse.de>
28300         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
28301         (sparseset_pop): Likewise.
28302         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
28303         to be able to use quick_push in the worker loop.
28305 2013-01-30  Marek Polacek  <polacek@redhat.com>
28307         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
28309 2013-01-30  Richard Biener  <rguenther@suse.de>
28311         PR lto/56147
28312         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
28314 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
28316         PR tree-optimization/56064
28317         * fixed-value.c (fixed_from_double_int): New function.
28318         * fixed-value.h (fixed_from_double_int): New prototype.
28319         (const_fixed_from_double_int): New static inline function.
28320         * fold-const.c (native_interpret_fixed): New static function.
28321         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
28322         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
28323         (native_encode_fixed): New static function.
28324         (native_encode_expr) <FIXED_CST>: Use it.
28325         (native_interpret_int): Move double_int worker code to...
28326         * double-int.c (double_int::from_buffer): ...this new static method.
28327         * double-int.h (double_int::from_buffer): Prototype it.
28329 2013-01-30  Richard Biener  <rguenther@suse.de>
28331         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
28332         New pointer-map and obstack.
28333         (init_alias_vars): Allocate pointer-map and obstack.
28334         (delete_points_to_sets): Free them.
28335         (find_what_var_points_to): Cache result.
28336         (find_what_p_points_to): Adjust for changed interface of
28337         find_what_var_points_to.
28338         (compute_points_to_sets): Likewise.
28339         (ipa_pta_execute): Likewise.
28341 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
28343         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
28344         * configure: Regenerate.
28345         * config.in: Regenerate.
28346         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
28347         #nobits/#progbits if supported.
28349 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
28351         PR target/56121
28352         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
28353         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
28354         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
28356 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28358         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
28359         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
28361 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28363         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
28364         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
28366 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28368         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
28369         declaration.
28370         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
28371         * config/arm/cortex-a7.md: New bypasses using
28372         arm_mac_accumulator_is_result.
28374 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28376         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
28377         (cortex_a7_neon_mla): Likewise.
28378         (cortex_a7_fpfmad): New reservation.
28379         (cortex_a7_fpmacs): Use ffmas and update required units.
28380         (cortex_a7_fpmuld): Update required units and latency.
28381         (cortex_a7_fpmacd): Likewise.
28382         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
28383         (cortex_a7_neon). Likewise.
28384         (bypass) Update participating units.
28386 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28388         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
28389         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
28390         from fmac to ffma.
28391         * config/arm/vfp11.md (vfp_farith): Use ffmas.
28392         (vfp_fmul): Use ffmad.
28393         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
28394         (cortex_r4_fmacd): Use ffmad.
28395         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
28396         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
28397         (cortex_a9_fmacd): Use ffmad.
28398         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
28399         (cortex_a8_vfp_macd): Use ffmad.
28400         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
28401         (cortex_a5_fpmacd): Use ffmad.
28402         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
28403         (cortex_a15_vfp_macd): Use ffmad.
28404         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
28406 2013-01-29  Jason Merrill  <jason@redhat.com>
28408         PR libstdc++/54314
28409         * varasm.c (default_assemble_visibility): Don't warn about
28410         visibility on artificial decls.
28412 2013-01-29  Richard Biener  <rguenther@suse.de>
28414         PR tree-optimization/56113
28415         * tree-ssa-structalias.c (equiv_class_lookup): Also return
28416         the bitmap leader.
28417         (label_visit): Free duplicate bitmaps and record the leader instead.
28418         (perform_var_substitution): Adjust.
28420 2013-01-29  Richard Biener  <rguenther@suse.de>
28422         PR tree-optimization/55270
28423         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
28424         the CFG, schedule loops for fixup.
28426 2013-01-29  Nick Clifton  <nickc@redhat.com>
28428         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
28429         SP_REG.
28431 2013-01-28  Leif Ekblad  <leif@rdos.net>
28433         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
28434         * config/i386/i386.h (TARGET_RDOS): New macro.
28435         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
28436         * config/i386/i386.c (ix86_option_override_internal): For 64bit
28437         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
28438         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
28439         DEFAULT_LARGE_SECTION_THRESHOLD.
28440         * config/i386/i386.md (R14_REG, R15_REG): New constants.
28441         * config/i386/rdos.h: New file.
28442         * config/i386/rdos64.h: New file.
28444 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
28446         PR other/54814
28447         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
28448         TEST_HARD_REG_BIT.
28450 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28452         PR rtl-optimization/56117
28453         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
28454         call cselib_lookup_from_insn on the MEM before calling
28455         add_insn_mem_dependence.
28457 2013-01-28  Richard Biener  <rguenther@suse.de>
28459         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
28460         to a stmt that didn't have one.
28461         (copy_phis_for_bb): Likewise for PHI arguments.
28462         (copy_debug_stmt): Likewise for debug stmts.
28464 2013-01-28  Richard Biener  <rguenther@suse.de>
28466         PR tree-optimization/56034
28467         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
28468         (partition_builtin_p): Adjust.
28469         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
28470         it is the last partition.
28471         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
28472         up the vertex for the definition.
28473         (classify_partition): Classify whether a partition is a
28474         PKIND_REDUCTION, thus has uses outside of the loop.
28475         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
28476         Merge all PKIND_REDUCTION partitions into the last partition.
28477         (tree_loop_distribution): Seed partitions from reductions as well.
28479 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28481         PR tree-optimization/56125
28482         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
28483         pow(x,c) into sqrt(x) * powi(x, n/2) or
28484         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
28485         optimizing for size.
28486         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
28487         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
28488         integer.
28490         PR tree-optimization/56094
28491         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
28492         to UNKNOWN_LOCATION while gimplifying expr.
28494 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
28496         PR target/56114
28497         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
28498         operand 0 in movabs insn template for -masm=intel asm alternative.
28499         (*movabs<mode>_2): Ditto for operand 1.
28501 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
28503         PR target/54663
28504         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
28505         of microblaze-c.o
28507 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28509         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
28510         tm_file.
28512 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
28514         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
28515         Undef to avoid warning.
28517 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
28519         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
28520         * configure: Regenerate.
28522 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
28524         PR tree-optimization/56098
28525         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
28526         for stmts with volatile ops.
28527         (cond_store_replacement): Don't optimize if assign has volatile ops.
28528         (cond_if_else_store_replacement_1): Don't optimize if either
28529         then_assign or else_assign have volatile ops.
28530         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
28531         volatile ops.
28533 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28535         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
28537 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28539         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
28540         missing ':' in asm example.
28542 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
28544         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
28545         entries into lane and laneq entries.
28546         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
28547         Remove AdvSIMD scalar modes.
28548         (aarch64_sq<r>dmulh_laneq<mode>): New.
28549         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
28550         modes.
28551         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
28552         builtin implementations to relfect changes in RTL in aarch64-simd.md.
28553         * config/aarch64/iterators.md (VCOND): New.
28554         (VCONQ): New.
28556 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28558         PR target/54222
28559         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
28560         Add NULL LIBNAME argument to existing definitions.
28561         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
28562         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
28563         * config/avr/avr.c (DEF_BUILTIN): Same.
28564         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
28565         (avr_expand_builtin): Expand to a vanilla call if a libgcc
28566         implementation is available (DECL_ASSEMBLER_NAME is set).
28567         (avr_fold_absfx): New static function.
28568         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
28569         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
28570         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
28571         AVR_BUILTIN_ABSLLK.
28572         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
28573         (abshk, absk, abslk, absllk): Provide as static inline functions.
28575 2013-01-25  Marek Polacek  <polacek@redhat.com>
28577         PR tree-optimization/56035
28578         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
28580 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
28582         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
28583         (*movtf_internal_rex64): Add (!o,C) alternative
28584         (*movxf_internal_rex64): Ditto.
28585         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
28587 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
28589         * doc/invoke.texi: fix typo.
28590         * doc/objc.texi: fix typo.
28592 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
28594         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
28595         for the first two alternatives.
28597 2013-01-24  Diego Novillo  <dnovillo@google.com>
28599         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
28600         (ggc-zone.o): Remove.
28601         * configure.ac: Remove option --with-gc.
28602         * configure: Re-generate.
28603         * doc/install.texi: Remove documentation for --with-gc.
28604         * gengtype.c (write_enum_defn): Remove.  Update all users.
28605         (write_Types_process_field): Remove generation of gt_e_* argument.
28606         (output_type_enum): Remove.  Update all users.
28607         (write_enum_defn): Remove.  Update all users.
28608         (enum alloc_zone): Remove.  Update all users.
28609         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
28610         * ggc-common.c (ggc_splay_alloc): Remove first argument.
28611         Update all callers.
28612         (struct ptr_data): Remove field TYPE.  Update all users.
28613         (gt_pch_note_object): Remove argument TYPE.  Update all users.
28614         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
28615         Update all users.
28616         * ggc-none.c (ggc_alloc_typed_stat): Remove.
28617         (struct alloc_zone): Remove.
28618         (ggc_internal_alloc_zone_stat): Remove.
28619         (ggc_internal_cleared_alloc_zone_stat): Remove.
28620         * ggc-page.c (ggc_alloc_typed_stat): Remove.
28621         (ggc_pch_count_object): Remove last argument.  Update all users.
28622         (ggc_pch_alloc_object): Remove last argument.  Update all users.
28623         (struct alloc_zone): Remove.
28624         * ggc-zone.c: Remove.
28625         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
28626         (struct alloc_zone): Remove.
28627         (ggc_alloc_typed_stat): Remove.
28628         (ggc_alloc_typed): Remove.
28629         (ggc_splay_alloc): Remove first argument.
28630         (rtl_zone): Remove.  Update all users.
28631         (tree_zone): Remove.  Update all users.
28632         (tree_id_zone): Remove.  Update all users.
28633         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
28634         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
28635         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
28636         * tree-ssanames.c: Remove references to zone allocator in comments.
28638 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
28640         * config/avr/avr.c (avr_out_fract): Make register numbers that
28641         might be outside of source operand signed.
28643 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
28645         * config/i386/constraints.md (Yf): New constraint.
28646         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
28647         of f constraint to conditionaly disable x87 register preferences.
28648         (*movdf_internal): Ditto.
28649         (*movsf_internal): Ditto.
28651 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
28653         PR inline-asm/55934
28654         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
28655         that have operands with impossible constraints.
28656         Add a FIXME for a speed-up opportunity.
28657         * lra-constraints.c (process_alt_operands): Verify that a class
28658         selected from constraints on asms is valid for the operand mode.
28659         (curr_insn_transform): Remove incorrect comment.
28661 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
28663         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
28664         TOC operand is a valid symbol ref in the constant pool.
28666 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28668         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
28670 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
28672         PR target/54222
28673         * config/avr/stdfix.h: New file.
28674         * t-avr (stdfix-gcc.h): New rule to build it.
28675         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
28677 2013-01-23  Kostya Serebryany  <kcc@google.com>
28679         * config/darwin.h: remove dependency on
28680         CoreFoundation (asan on Mac OS).
28682 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
28684         PR target/49069
28685         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
28686         instead of cmpdi_operand for first comparison operand.
28687         Don't assert that comparison operands aren't both constants.
28689 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28691         * doc/install.texi (Downloading the Source): Update references to
28692         downloading separate components.
28694 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28696         * doc/extend.texi (__int128): Improve grammar.
28698 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
28700         PR target/56028
28701         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
28702         alternative to (o,r).
28703         (*movdi_internal_rex64): Remove (!o,n) alternative.
28704         (DImode immediate->memory splitter): Remove.
28705         (DImode immediate->memory peephole2): Remove.
28706         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
28707         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
28708         alternative to (!o,*r).
28709         (*movtf_internal_sse): New pattern.
28710         (*movxf_internal_rex64): New pattern.
28711         (*movxf_internal): Disable for TARGET_64BIT.
28712         (*movdf_internal_rex64): Remove (!o,F) alternative.
28714 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
28716         PR middle-end/56074
28717         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
28718         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
28719         * tree-vect-loop-manip.c (find_loop_location): Also ignore
28720         stmt locations where LOCATION_LOCUS of the stmt location is
28721         UNKNOWN_LOCATION or BUILTINS_LOCATION.
28723         PR target/55686
28724         * config/i386/i386.md (UNSPEC_STOS): New.
28725         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
28726         *strsetqi_1): Add UNSPEC_STOS.
28728 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
28730         PR c++/56067
28731         * doc/invoke.texi: Remove left over -Wsynth example.
28733 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28735         PR tree-optimization/56051
28736         * fold-const.c (fold_binary_loc): Don't fold
28737         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
28738         a narrowing conversion, or widening conversion from signed
28739         to unsigned.
28741 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28743         PR rtl-optimization/56023
28744         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
28745         dependent on debug instruction.
28747 2013-01-21  Martin Jambor  <mjambor@suse.cz>
28749         PR middle-end/56022
28750         * function.c (allocate_struct_function): Call
28751         invoke_set_current_function_hook earlier.
28753 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28755         * reload1.c (init_reload): Only initialize reload_obstack
28756         during the first call.
28758 2013-01-21  Marek Polacek  <polacek@redhat.com>
28760         * cfgloop.c (verify_loop_structure): Fix up grammar.
28762 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
28764         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
28765         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
28767 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28769         PR target/56058
28770         * config/arm/marvell-pj4.md: Update copyright year.
28771         Fix up use of alu to alu_reg and simple_alu_imm.
28773 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28775         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
28777 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
28779         PR target/55433
28780         * lra-constraints.c (curr_insn_transform): Don't reuse original
28781         insn for secondary memory move when memory mode should be different.
28783 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28785         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
28786         atomic_storedi_1): New patterns.
28788 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
28790         btver2 pipeline descriptions.
28791         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
28792         descriptions.
28793         * config/i386/i386.md (btver2_decode): New type attributes.
28794         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
28795         type attributes.
28796         * config/i386/btver2.md: New file describing btver2 pipelines.
28798 2013-01-19  Andrew Pinski  <apinski@cavium.com>
28800         PR tree-optimization/52631
28801         * tree-ssa-sccvn (visit_use): Before looking up the original
28802         statement, try looking up the simplified expression.
28804 2013-01-19  Anthony Green  <green@moxielogic.com>
28806         * config/moxie/moxie.c (moxie_expand_prologue): Set
28807         current_function_static_stack_size.
28809 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28811         PR tree-optimization/56029
28812         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
28813         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
28815 2013-01-18  Sharad Singhai  <singhai@google.com>
28817         PR tree-optimization/55995
28818         * dumpfile.c (dump_loc): Print location only if available.
28819         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
28821 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
28823         PR target/55433
28824         * lra-constraints.c (curr_insn_transform): Reuse original insn for
28825         secondary memory move.
28826         (inherit_reload_reg): Use rclass instead of cl for
28827         check_secondary_memory_needed_p.
28829 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28831         PR middle-end/56015
28832         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
28833         the case where writing real complex part of target modifies op1.
28835 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
28837         * config/aarch64/aarch64-simd.md
28838         (aarch64_vcond_internal<mode>): Handle unordered cases.
28839         * config/aarch64/iterators.md (v_cmp_result): New.
28841 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
28842             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28844         * config/arm/marvell-pj4.md: New file.
28845         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
28846         * config/arm/arm.md (generic_sched): Add marvell_pj4.
28847         (generic_vfp): Likewise.
28848         * config/arm/arm-cores.def: Add marvell-pj4.
28849         * config/arm/arm-tune.md: Regenerate.
28850         * config/arm/arm-tables.opt: Regenerate.
28851         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
28852         * doc/invoke.texi: Document marvell-pj4.
28854 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
28856         * config/aarch64/arm_neon.h: Map scalar types to standard types.
28858 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28860         PR debug/54114
28861         PR debug/54402
28862         PR debug/49888
28863         * var-tracking.c (negative_power_of_two_p): New.
28864         (global_get_addr_cache, local_get_addr_cache): New.
28865         (get_addr_from_global_cache, get_addr_from_local_cache): New.
28866         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
28867         heading comment.
28868         (vt_stack_offset_p): Remove.
28869         (vt_canon_true_dep): Always canonicalize loc's address.
28870         (clobber_overlapping_mems): Make sure we have a MEM.
28871         (local_get_addr_clear_given_value): New.
28872         (val_reset): Clear local cached entries.
28873         (compute_bb_dataflow): Create and release the local cache.
28874         Disable duplicate MEMs clobbering.
28875         (emit_notes_in_bb): Clobber MEMs likewise.
28876         (vt_emit_notes): Create and release the local cache.
28877         (vt_initialize, vt_finalize): Create and release the global
28878         cache, respectively.
28879         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
28881 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28883         PR libmudflap/53359
28884         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
28885         not found in the symtab.
28887 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28889         PR debug/56006
28890         PR rtl-optimization/55547
28891         PR rtl-optimization/53827
28892         PR debug/53671
28893         PR debug/49888
28894         * alias.c (offset_overlap_p): New, factored out of...
28895         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
28896         the conservative special case for symbolic constants.  Don't
28897         adjust zero sizes on alignment.
28899 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
28901         PR rtl-optimization/52573
28902         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
28903         REG_UNUSED for the same register.
28905 2013-01-17  Richard Biener  <rguenther@suse.de>
28906             Marek Polacek  <polacek@redhat.com>
28908         PR rtl-optimization/55833
28909         * loop-unswitch.c (unswitch_loops): Move loop verification...
28910         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
28911         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
28912         Set it to true when we're removing a loop from hierarchy tree in
28913         an irreducible region.
28914         (fix_bb_placements): Adjust caller.
28915         (fix_loop_placements): Likewise.
28917 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
28919         * config/avr/builtins.def (DEF_BUILTIN): Factor out
28920         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
28921         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
28922         Remove ID.  Adjust comments.
28923         * config/avr/avr-c.c (avr_builtin_name): Remove.
28924         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
28925         * config/avr/avr.c (avr_tolower): New static function.
28926         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
28927         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
28928         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
28929         default expansion.
28931 2013-01-17  Jan Hubicka  <jh@suse.cz>
28933         PR tree-optimization/55273
28934         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
28936 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
28938         PR target/55981
28939         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
28940         store through atomic_store<mode>_1.
28941         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
28943 2013-01-17  Martin Jambor  <mjambor@suse.cz>
28945         PR tree-optimizations/55264
28946         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
28947         for virtual methods.
28948         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
28949         virtual methods before inlining is over.
28950         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
28951         virtual functions.
28952         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
28953         non-virtual.
28955 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
28957         PR rtl-optimization/56005
28958         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
28959         pending reads for prefetch.
28961 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
28963         * config/aarch64/aarch64.md
28964         (*cstoresi_neg_uxtw): New pattern.
28965         (*cmovsi_insn_uxtw): New pattern.
28966         (*<optab>si3_uxtw): New pattern.
28967         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
28968         (*<optab>si3_insn_uxtw): New pattern.
28969         (*bswapsi2_uxtw): New pattern.
28971 2013-01-16  Richard Biener  <rguenther@suse.de>
28973         * tree-inline.c (tree_function_versioning): Remove set but
28974         never used variable.
28976 2013-01-16  Richard Biener  <rguenther@suse.de>
28978         PR tree-optimization/55964
28979         * tree-flow.h (rename_variables_in_loop): Remove.
28980         (rename_variables_in_bb): Likewise.
28981         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
28982         (copy_loop_before): Adjust and delete update-ssa status.
28983         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
28984         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
28985         (rename_variables_in_loop): Remove.
28986         (slpeel_update_phis_for_duplicate_loop): Likewise.
28987         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
28988         use available cfg machinery instead of duplicating it.
28989         Update PHI nodes and perform poor-mans SSA update here.
28990         (slpeel_tree_peel_loop_to_edge): Adjust.
28992 2013-01-16  Richard Biener  <rguenther@suse.de>
28994         PR tree-optimization/54767
28995         PR tree-optimization/53465
28996         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
28997         (vrp_visit_phi_node): For PHI arguments coming via backedges
28998         drop all symbolical range information.
28999         (execute_vrp): Compute backedges.
29001 2013-01-16  Richard Biener  <rguenther@suse.de>
29003         * doc/install.texi: Update CLooG and ISL requirements to
29004         0.18.0 and 0.11.1.
29006 2013-01-16  Christian Bruel  <christian.bruel@st.com>
29008         PR target/55301
29009         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
29010         (broken_move): Handle UNSPECV_SP_SWITCH_B.
29011         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
29013 2013-01-16  DJ Delorie  <dj@redhat.com>
29015         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
29016         (UNSPECV_SP_SWITCH_E): New.
29017         (sp_switch_1): Change to an unspec.
29018         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
29019         replace $r15.
29021 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
29023         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
29024         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
29025         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
29026         (expand_mem_thread_fence): Ditto.
29027         (expand_mem_signal_fence): Ditto.
29028         (expand_atomic_load): Ditto.
29029         (expand_atomic_store): Ditto.
29031 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
29033         PR rtl-optimization/55547
29034         PR rtl-optimization/53827
29035         PR debug/53671
29036         PR debug/49888
29037         * alias.c (memrefs_conflict_p): Set sizes to negative after
29038         AND adjustments.
29040 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29042         PR target/55940
29043         * function.c (thread_prologue_and_epilogue_insns): Always
29044         add crtl->drap_reg to set_up_by_prologue.set, even if
29045         stack_realign_drap is false.
29047 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
29049         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
29050         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
29051         *call): Fix indention.
29053 2013-01-15  Tom de Vries  <tom@codesourcery.com>
29055         PR target/55876
29056         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
29057         Update comment.
29059 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
29061         PR rtl-optimization/55153
29062         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
29064 2013-01-15  Martin Jambor  <mjambor@suse.cz>
29066         PR tree-optimization/55920
29067         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
29068         accesses as grp_to_be_debug_replaced.
29070 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29072         PR tree-optimization/55920
29073         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
29074         there is non-useless type conversion needed from debug rhs to lhs,
29075         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
29077 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
29078             Mikael Pettersson  <mikpe@it.uu.se>
29080         PR target/43961
29081         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
29082         Thumb.
29083         (ASM_OUTPUT_CASE_LABEL): Remove.
29084         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
29085         * final.c (shorten_branches): Update alignment of labels before
29086         jump tables if CASE_VECTOR_SHORTEN_MODE.
29088 2013-01-15  Richard Biener  <rguenther@suse.de>
29090         PR bootstrap/55961
29091         * system.h: Do not include gmp.h for building host tools.
29093 2013-01-15  Richard Biener  <rguenther@suse.de>
29095         PR middle-end/55882
29096         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
29097         account for bitpos when computing alignment.
29099 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
29101         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
29102         (ix86_target_macros_internal): Likewise.
29104         * config/i386/i386.c (m_CORE2I7): Removed.
29105         (m_CORE_HASWELL): New macro.
29106         (m_CORE_ALL): Likewise.
29107         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
29108         (initial_ix86_arch_features): Likewise.
29109         (processor_target_table): Initializations for Core avx2.
29110         (cpu_names): New names "core-avx2".
29111         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
29112         PROCESSOR_CORE_HASWELL.
29113         (ix86_issue_rate): New case.
29114         (ia32_multipass_dfa_lookahead): Likewise.
29115         (ix86_sched_init_global): Likewise.
29117         * config/i386/i386.h (TARGET_HASWELL): New macro.
29118         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
29119         (processor_type): New PROCESSOR_HASWELL.
29121 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29123         PR tree-optimization/55955
29124         * tree-vect-loop.c (vectorizable_reduction): Give up early on
29125         *SHIFT_EXPR and *ROTATE_EXPR codes.
29127         PR tree-optimization/48766
29128         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
29129         -ftrapv disable -fwrapv.
29131 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29133         PR target/55974
29134         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
29135         etc. to 1 and not to __flash.
29136         Use LL suffix for __INT24_MAX__ with -mint8.
29137         Use ULL suffix for __UINT24_MAX__ with -mint8.
29139 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29141         * config/avr/avr-arch.h
29142         (struct base_arch_s): Use typedef avr_arch_t instead.
29143         (struct arch_info_s): Use typedef avr_arch_info_t instead.
29144         (struct mcu_type_s): Use typedef avr_mcu_t instead.
29145         * config/avr/avr.c: Same.
29146         * config/avr/avr-devices.c: Same.
29147         * config/avr/driver-avr.c: Same.
29148         * config/avr/gen-avr-mmcu-texi.c: Same.
29149         * config/avr/avr-mcus.def: Adjust comment.
29151 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
29153         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
29154         * config/aarch64/iterators.md (VALLDI): New.
29156 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
29157             Andi Kleen  <ak@linux.intel.com>
29159         PR target/55948
29160         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
29161         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
29162         memmodel flag.
29164 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29166         * config/avr/avr-stdint.h: Remove trailing blanks.
29167         * config/avr/avr-log.h: Same.
29168         * config/avr/avr-arch.h: Same.
29169         * config/avr/avr-devices.c: Same.
29170         * config/avr/avr-dimode.md: Same.
29171         * config/avr/predicates.md: Same.
29172         * config/avr/avr-c.c: Same.  And fix typo.
29174         * config/avr/avr-protos.h: Same.  And:
29175         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
29176         (init_cumulative_args): Rename to avr_init_cumulative_args.
29177         (expand_prologue): Rename to avr_expand_prologue.
29178         (expand_epilogue): Rename to avr_expand_epilogue.
29179         (adjust_insn_length): Rename to avr_adjust_insn_length.
29180         (notice_update_cc): Rename to avr_notice_update_cc.
29181         (final_prescan_insn): Rename to avr_final_prescan_insn.
29182         * config/avr/avr.c: Same.
29183         * config/avr/avr.h: Same.
29184         * config/avr/avr.md: Remove trailing blanks.
29185         (prologue): Use avr_expand_prologue.
29186         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
29188 2013-01-14  Richard Biener  <rguenther@suse.de>
29190         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
29191         verify_location, collect_subblocks): New functions.
29192         (verify_gimple_in_cfg): Verify that locations only reference
29193         BLOCKs in the functions BLOCK tree.
29195 2013-01-14  Richard Biener  <rguenther@suse.de>
29197         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
29198         PHI argument.
29199         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
29200         unshare reference.
29201         (insert_out_of_ssa_copy_on_edge): Likewise.
29202         (rewrite_close_phi_out_of_ssa): Likewise.
29203         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
29204         debug expressions.
29205         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
29206         propagated constants.
29207         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
29208         can not be shared.
29210 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29212         * config/avr/avr-modes.def: Add GPL copyright notice.
29214 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
29216         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
29217         MEMMODEL_MASK to determine memory model.
29218         (atomic_store<mode>): Ditto from operands[2].
29219         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
29221 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
29223         PR fortran/55935
29224         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
29225         (fold_gimple_assign): Don't call unshare_expr here.
29226         (fold_ctor_reference): Call unshare_expr.
29228 2013-01-13  Terry Guo  <terry.guo@arm.com>
29230         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
29231         * doc/fragments.texi: Document MULTILIB_REUSE.
29232         * gcc.c (multilib_reuse): New internal spec.
29233         (set_multilib_dir): Also search multilib from multilib_reuse.
29234         * genmultilib (tmpmultilib3): Refactor code.
29235         (tmpmultilib4): Ditto.
29236         (multilib_reuse): New multilib argument.
29238 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
29240         * Makefile.in: Update copyright.
29242 2013-01-12  Tom de Vries  <tom@codesourcery.com>
29244         PR middle-end/55890
29245         * calls.c (expand_call): Check if arg_nr is valid.
29247 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
29249         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
29250         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
29251         documentation.  Add missing '__' in front of
29252         __builtin_ia32_packssdw256.
29254 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29256         PR target/55719
29257         * config/s390/s390.c (s390_preferred_reload_class): Do not return
29258         NO_REGS for larl operands.
29259         (s390_reload_larl_operand): Use s390_load_address instead of
29260         emit_move_insn.
29262 2013-01-11  Richard Biener  <rguenther@suse.de>
29264         * tree-cfg.c (verify_node_sharing_1): Split out from ...
29265         (verify_node_sharing): ... here.
29266         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
29268 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
29270         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
29271         Substitute TREECHECKING.
29272         * configure: Regenerate.
29273         * Makefile.in (TREECHECKING): New.
29275 2013-01-11  Richard Guenther  <rguenther@suse.de>
29277         PR tree-optimization/44061
29278         * tree-vrp.c (extract_range_basic): Compute zero as
29279         value-range for __builtin_constant_p of function parameters.
29281 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
29283         Update copyright years.
29285 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
29287         PR rtl-optimization/55672
29288         * lra-eliminations.c (mark_not_eliminable): Permit addition with
29289         const to be eliminable.
29291 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
29293         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
29294         * configure: Regenerate.
29296 2013-01-10  Richard Biener  <rguenther@suse.de>
29298         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
29300 2013-01-10  Richard Biener  <rguenther@suse.de>
29302         PR bootstrap/55792
29303         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
29304         locations for virtual PHI arguments.
29305         (rewrite_update_phi_arguments): Likewise.
29307 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
29309         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
29310         on to assembler.
29312 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
29314         PR tree-optimization/55921
29315         * tree-complex.c (expand_complex_asm): New function.
29316         (expand_complex_operations_1): Call it for GIMPLE_ASM.
29318 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29320         PR target/55718
29321         * config/s390/s390.c (s390_symref_operand_p)
29322         (s390_loadrelative_operand_p): Merge the two functions.
29323         (s390_check_qrst_address, print_operand_address): Add parameters
29324         to s390_loadrelative_operand_p invokation.
29325         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
29326         (s390_reload_larl_operand, s390_secondary_reload): Use
29327         s390_loadrelative_operand_p instead of s390_symref_operand_p.
29328         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
29330 2013-01-09  Mike Stump  <mikestump@comcast.net>
29332         * dse.c (record_store): Remove unnecessary assert.
29334 2013-01-09  Jan Hubicka  <jh@suse.cz>
29336         PR tree-optimization/55569
29337         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
29338         * cfgloop.h (scale_loop_profile): Likewise.
29340 2013-01-09  Jan Hubicka  <jh@suse.cz>
29342         PR lto/45375
29343         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
29344         functions.
29345         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
29347 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
29349         PR middle-end/55114
29350         * expr.h (maybe_emit_group_store): Declare.
29351         * expr.c (maybe_emit_group_store): New function.
29352         * builtins.c (expand_builtin_int_roundingfn): Call it.
29353         (expand_builtin_int_roundingfn_2): Likewise.
29355 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
29357         PR rtl-optimization/55829
29358         * lra-constraints.c (match_reload): Add code for absent output.
29359         (curr_insn_transform): Add code for reloads of matched inputs
29360         without output.
29362 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
29364         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
29365         attribute of movddup insn to DF.
29366         (*vec_interleave_lowv2df): Ditto.
29367         (vec_dupv2df): Ditto.
29369 2013-01-09  Jan Hubicka  <jh@suse.cz>
29371         PR tree-optimiation/55875
29372         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
29373         EVERY_ITERATION parameter.
29374         (number_of_iterations_exit): Check if exit is executed every iteration.
29375         (idx_infer_loop_bounds): Similarly here.
29376         (n_of_executions_at_most): Simplify
29377         to only test for cases where statement is dominated by the
29378         particular bound; handle correctly the "postdominance" test.
29379         (scev_probably_wraps_p): Use max loop iterations info
29380         as a global bound first.
29382 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
29383             Nick Clifton  <nickc@redhat.com>
29385         * config/v850/v850.md (cbranchsf4): New pattern.
29386         (cstoresf4): New pattern.
29387         (cbranchdf4): New pattern.
29388         (cstoredf4): New pattern.
29389         (movsicc): Disallow floating point comparisons.
29390         (cmpsf_le_insn): Fix order of operators.
29391         (cmpsf_lt_insn): Likewise.
29392         (cmpsf_eq_insn): Likewise.
29393         (cmpdf_le_insn): Likewise.
29394         (cmpdf_lt_insn): Likewise.
29395         (cmpdf_eq_insn): Likewise.
29396         (cmpsf_ge_insn): Use LE comparison.
29397         (cmpdf_ge_insn): Likewise.
29398         (cmpsf_gt_insn): Use LT comparison.
29399         (cmpdf_gt_insn): Likewise.
29400         (cmpsf_ne_insn): Delete pattern.
29401         (cmpdf_ne_insn): Delete pattern.
29402         * config/v850/v850.c (v850_gen_float_compare): Use
29403         gen_cmpdf_eq_insn for NE comparison.
29404         (v850_float_z_comparison_operator)
29405         (v850_float_nz_comparison_operator): Move from here ...
29406         * config/v850/predicates.md: ... to here.  Move GT and GE
29407         comparisons into v850_float_z_comparison_operator.
29408         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
29409         Delete prototype.
29410         (v850_float_nz_comparison_operator): Likewise.
29412 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29414         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
29415         with calls to gen_insvsi/gen_insvdi.
29417 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
29419         * config/i386/i386.c (initial_ix86_tune_features): Set up
29420         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
29422 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
29423             Jakub Jelinek  <jakub@redhat.com>
29425         PR tree-optimization/48189
29426         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
29427         If nitercst is 0, don't predict the exit edge.
29429 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
29431         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
29432         in asm_fprintf with reg_names.
29433         (aarch64_print_operand_address): Likewise.
29434         (aarch64_return_addr): Likewise.
29435         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
29437 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29439         * config/pa/pa.h (VAL_U6_BITS_P): Define.
29440         (INT_U6_BITS): Likewise.
29441         * config/pa/predicates.md (uint6_operand): New predicate.
29442         (shift5_operand, shift6_operand): Likewise.
29443         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
29444         arith32_operand.
29445         (lshrdi3): Use shift6_operand.
29446         (shrpsi4, shrpdi4): New insn patterns.
29447         (extzv): Delete expander.
29448         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
29449         predicates in unamed zero extract patterns.  Tighten common constraint.
29450         (extv): Delete expander.
29451         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
29452         predicates in unamed sign extract patterns.  Tighten common constraint.
29453         (insv): Delete expander.
29454         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
29455         predicates in unamed insert patterns.  Tighten common constraint.
29456         Change uint32_operand predicate to uint6_operand predicate in unamed
29457         DImode pattern to insert constant values of type 1...1xxxx.
29459 2013-01-04  Jan Hubicka  <jh@suse.cz>
29461         PR tree-optimization/55823
29462         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
29463         issue.
29465 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29466             Uros Bizjak  <ubizjak@gmail.com>
29468         PR rtl-optimization/55845
29469         * df-problems.c (can_move_insns_across): Stop scanning at
29470         volatile_insn_p source instruction or give up if
29471         across_from .. across_to range contains any volatile_insn_p
29472         instructions.
29474 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29476         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
29477         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
29478         Declare.
29479         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
29480         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
29482 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29484         PR fortran/55341
29485         * asan.c (asan_clear_shadow): New function.
29486         (asan_emit_stack_protection): Use it.
29488 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29490         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
29491         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
29492         with tab instead of space.
29494 2013-01-08  Nick Clifton  <nickc@redhat.com>
29496         * config/rl78/rl78.c (rl78_expand_prologue): Always select
29497         register bank 0 at the start of an interrupt handler.
29498         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
29499         MDBH registers.
29501 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29503         * config/aarch64/aarch64-simd.md
29504         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
29505         (aarch64_simd_bsl): Likewise.
29506         (aarch64_vcond_internal<mode>): Likewise.
29507         (vcond<mode><mode>): Likewise.
29508         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
29509         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
29511 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29513         * config/aarch64/aarch64-builtins.c
29514         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
29516 2013-01-08  Martin Jambor  <mjambor@suse.cz>
29518         PR debug/55579
29519         * tree-sra.c (analyze_access_subtree): Return true also after
29520         potentially creating a debug-only replacement.
29522 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29524         PR middle-end/55890
29525         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
29527         PR tree-optimization/54120
29528         * tree-vrp.c (range_fits_type_p): Don't allow
29529         src_precision < precision from signed vr to unsigned_p
29530         if vr->min or vr->max is negative.
29531         (simplify_float_conversion_using_ranges): Test can_float_p
29532         against CODE_FOR_nothing.
29534 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29535             Richard Biener  <rguenther@suse.de>
29537         PR middle-end/55851
29538         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
29539         types instead of just INTEGER_TYPE types.
29541 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
29543         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
29544         TF_SIZE): Define.
29546 2013-01-07  Steve Ellcey  <sellcey@mips.com>
29548         PR target/42661
29549         * config/mips/mips.opt: Change mad to mmad to match documentation.
29551 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29553         PR target/55897
29554         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
29555         .progmemx.data now.
29557 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29559         PR target/55897
29560         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
29561         (avr_addrspace_t): Add .section_name field.
29562         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
29563         array size.
29564         (avr_addrspace): Same.  Initialize .section_name.  Remove last
29565         NULL entry.  Put __memx into .progmemx.data.
29566         (progmem_section_prefix): Remove.
29567         (avr_asm_init_sections): No need to initialize progmem_section.
29568         (avr_asm_named_section): Use avr_addrspace[].section_name to get
29569         section name prefix.
29570         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
29571         retrieve the progmem section.
29572         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
29573         boundary to run over avr_addrspace[].
29574         (avr_register_target_pragmas): Ditto.
29576 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
29578         * varasm.c (output_constant_def_contents): For asan_protect_global
29579         protected strings, adjust DECL_ALIGN if needed, before testing for
29580         anchored symbols.
29581         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
29582         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
29583         normal decls.
29584         (output_object_block): For asan protected decls, emit asan padding
29585         after their contents.
29586         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
29587         (asan_finish_file): Test it here instead.
29589 2013-01-07  Nick Clifton  <nickc@redhat.com>
29590             Matthias Klose  <doko@debian.org>
29591             Doug Kwan  <dougkwan@google.com>
29592             H.J. Lu  <hongjiu.lu@intel.com>
29594         PR driver/55470
29595         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
29597         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
29599         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
29601         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
29603         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
29605 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29607         PR target/54461
29608         * doc/install.texi (Cross-Compiler-Specific Options): Document
29609         --with-avrlibc.
29611 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
29613         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
29614         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
29615         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
29616         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
29617         vqmovun_high_s64): Fix source operand number and update copyright.
29619 2013-01-07  Richard Biener  <rguenther@suse.de>
29621         PR middle-end/55890
29622         * gimple.h (gimple_call_builtin_p): New overload.
29623         * gimple.c (validate_call): New function.
29624         (gimple_call_builtin_p): Likewise.
29625         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29626         Use gimple_call_builtin_p.
29627         (find_func_clobbers): Likewise.
29628         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
29629         (strlen_optimize_stmt): Likewise.
29631 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29633         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
29634         (vld1q_dup_*): Likewise.
29635         (vld1_*): Likewise.
29636         (vld1q_*): Likewise.
29637         (vld1_lane_*): Likewise.
29638         (vld1q_lane_*): Likewise.
29640 2013-01-07  Richard Biener  <rguenther@suse.de>
29642         * lto-streamer.h (LTO_minor_version): Bump to 2.
29644 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29646         * config/aarch64/aarch64-protos.h
29647         (aarch64_const_double_zero_rtx_p): Rename to...
29648         (aarch64_float_const_zero_rtx_p): ...this.
29649         (aarch64_float_const_representable_p): New.
29650         (aarch64_output_simd_mov_immediate): Likewise.
29651         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
29652         move immediate case.
29653         * config/aarch64/aarch64.c
29654         (aarch64_const_double_zero_rtx_p): Rename to...
29655         (aarch64_float_const_zero_rtx_p): ...this.
29656         (aarch64_print_operand): Allow printing of new constants.
29657         (aarch64_valid_floating_const): New.
29658         (aarch64_legitimate_constant_p): Check for valid floating-point
29659         constants.
29660         (aarch64_simd_valid_immediate): Likewise.
29661         (aarch64_vect_float_const_representable_p): New.
29662         (aarch64_float_const_representable_p): Likewise.
29663         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
29664         (aarch64_output_simd_mov_immediate): New.
29665         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
29666         (*movdf_aarch64): Likewise.
29667         * config/aarch64/constraints.md (Ufc): New.
29668         (Y): call aarch64_float_const_zero_rtx.
29669         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
29671 2013-01-07  Richard Biener  <rguenther@suse.de>
29673         PR tree-optimization/55888
29674         PR tree-optimization/55862
29675         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
29676         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
29677         not if it is contained therein.
29679 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29681         * config/avr/t-avr: Typo.
29683 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29685         PR55243
29686         * config/avr/t-avr: Don't automatically rebuild
29687         $(srcdir)/config/avr/t-multilib
29688         $(srcdir)/config/avr/avr-tables.opt
29689         $(srcdir)/doc/avr-mmcu.texi
29690         (avr-mcus): New phony target to build them on request.
29691         (s-avr-mlib, s-avr-mmcu-texi): Remove.
29692         * avr/avr-mcus.def: Adjust comments.
29694 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
29696         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
29698 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29700         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
29702 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29704         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
29706 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
29708         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
29709         to generate profiling.
29710         * config/rs6000/aix64.h (LIB_SPEC): Same.
29712 2013-01-04  Andrew Pinski  <apinski@cavium.com>
29714         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
29715         New function.
29716         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
29718 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
29720         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
29721         unconditionally.
29722         (ix86_expand_move): Ditto.
29723         (ix86_zero_extend_to_Pmode): Ditto.
29724         (ix86_expand_call): Ditto.
29725         (ix86_expand_special_args_builtin): Ditto.
29726         (ix86_expand_builtin): Ditto.
29728 2013-01-04  Richard Biener  <rguenther@suse.de>
29730         PR tree-optimization/55862
29731         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
29732         translating them through PHI nodes.
29734 2013-01-04  Martin Jambor  <mjambor@suse.cz>
29736         PR tree-optimization/55755
29737         * tree-sra.c (sra_modify_assign): Do not check that an access has no
29738         children when trying to avoid producing a VIEW_CONVERT_EXPR.
29740 2013-01-04  Marek Polacek  <polacek@redhat.com>
29742         PR middle-end/55859
29743         * opts.c (default_options_optimization): Clarify error message.
29745 2013-01-04  Richard Biener  <rguenther@suse.de>
29747         PR middle-end/55863
29748         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
29749         reassociation.
29751 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29753         PR target/53789
29754         * config/pa/pa.md (movsi): Revert previous change.
29755         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
29756         references.
29758 2013-01-03  Richard Henderson  <rth@redhat.com>
29760         * config/i386/i386.c (ix86_expand_move): Always assign to op1
29761         after eliminating TLS symbols.
29763 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
29765         PR bootstrap/50167
29766         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
29767         * graphite-poly.c (debug_gmp_value): Likewise.
29769 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
29771         PR target/55712
29772         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
29773         selected code model, define __code_mode_small__, __code_model_medium__,
29774         __code_model_large__, __code_model_32__ or __code_model_kernel__.
29775         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
29776         xchg temporary register with %k.  Declare temporary register as
29777         early clobbered.
29778         [__x86_64__]: For medium and large code models, preserve %rbx register.
29780 2013-01-03  Richard Biener  <rguenther@suse.de>
29782         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
29783         (dump_subscript): Adjust.
29784         (finalize_ddr_dependent): Do not dump redundant info.
29785         (analyze_siv_subscript): Adjust.
29786         (subscript_dependence_tester): Likewise.
29787         (compute_affine_dependence): Likewise.
29789 2013-01-03  Richard Biener  <rguenther@suse.de>
29791         Revert
29792         2013-01-03  Richard Biener  <rguenther@suse.de>
29794         PR tree-optimization/55857
29795         * tree-vect-stmts.c (vectorizable_load): Do not setup
29796         re-alignment for invariant loads.
29798         2013-01-02  Richard Biener  <rguenther@suse.de>
29800         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29801         invariant load do not generate a vector load from the scalar location.
29803 2013-01-03  Richard Biener  <rguenther@suse.de>
29805         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
29806         for not vectorizing.
29807         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
29808         not build INDIRECT_REFs, call get_name once only.
29809         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
29810         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
29812 2013-01-03  Richard Biener  <rguenther@suse.de>
29814         PR tree-optimization/55857
29815         * tree-vect-stmts.c (vectorizable_load): Do not setup
29816         re-alignment for invariant loads.
29818 2013-01-03  Richard Biener  <rguenther@suse.de>
29820         PR lto/55848
29821         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
29822         prefer a built-in decl.
29824 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29826         * gcc.c (process_command): Update copyright notice dates.
29827         * gcov.c (print_version): Likewise.
29828         * gcov-dump.c (print_version): Likewise.
29830         PR rtl-optimization/55838
29831         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
29832         iv0.step, iv1.step and step.
29834 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29835             Marc Glisse  <marc.glisse@inria.fr>
29837         PR tree-optimization/55832
29838         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
29839         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
29840         integer_{one,zero}_node.
29842 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29844         PR debug/54402
29845         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
29846         * var-tracking.c (reverse_op): Don't add reverse ops to
29847         VALUEs that have already
29848         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
29850 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
29852         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
29854 2013-01-02  Teresa Johnson  <tejohnson@google.com>
29856         * dumpfile.c (dump_loc): Print filename with location.
29857         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
29858         new location_t parameter to emit complete unroll message with
29859         new dump framework.
29860         (canonicalize_loop_induction_variables): Compute loops location
29861         and pass to try_unroll_loop_completely.
29862         * loop-unroll.c (report_unroll_peel): New function.
29863         (peel_loops_completely): Use new dump format with location
29864         for main dumpfile message, and invoke report_unroll_peel on success.
29865         (decide_unrolling_and_peeling): Ditto.
29866         (decide_peel_once_rolling): Remove old dumpfile message subsumed
29867         by report_unroll_peel.
29868         (decide_peel_completely): Ditto.
29869         (decide_unroll_constant_iterations): Ditto.
29870         (decide_unroll_runtime_iterations): Ditto.
29871         (decide_peel_simple): Ditto.
29872         (decide_unroll_stupid): Ditto.
29873         * cfgloop.c (get_loop_location): New function.
29874         * cfgloop.h (get_loop_location): Declare.
29876 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29878         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
29879         NULL.
29881 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29883         PR middle-end/55198
29884         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
29885         BLKmode objects when EXPAND_MEMORY is specified.
29887 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29889         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
29890         in loop predicate.
29891         (fold_builtin_cpu): Do not share cpu model decls across statements.
29893 2013-01-02  Jason Merrill  <jason@redhat.com>
29895         PR c++/55804
29896         * tree.c (build_array_type_1): Revert earlier change.
29898 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
29900         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
29901         "cortex-a57".
29902         * config/aarch64/aarch64-tune.md: Re-generate.
29904 2013-01-02  Richard Biener  <rguenther@suse.de>
29906         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29907         invariant load do not generate a vector load from the scalar location.
29909 2013-01-02  Richard Biener  <rguenther@suse.de>
29911         PR bootstrap/55784
29912         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
29913         * configure: Regenerate.
29915 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29917         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
29918         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
29919         (expand_builtin_int_roundingfn_2): Keep the original target around
29920         for the fallback case.
29922 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29924         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
29925         to be clear for sign changes.
29927 2013-01-01  Jan Hubicka  <jh@suse.cz>
29929         * ipa-inline-analysis.c: Fix formatting.
29931 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
29933         PR tree-optimization/55831
29934         * tree-vect-loop.c (get_initial_def_for_induction): Use
29935         gsi_after_labels instead of gsi_start_bb.
29937 Copyright (C) 2013 Free Software Foundation, Inc.
29939 Copying and distribution of this file, with or without modification,
29940 are permitted in any medium without royalty provided the copyright
29941 notice and this notice are preserved.