gcc-defs.exp (dg-additional-files-options): Extend regsub for dg-additional-files...
[official-gcc.git] / gcc / ChangeLog
blob9d45b0ca4fd5bbf008d0d05edbcec87d13990a7e
1 2013-11-06  Richard Biener  <rguenther@suse.de>
3         PR tree-optimization/58653
4         * tree-predcom.c (ref_at_iteration): Rewrite to generate
5         a MEM_REF.
6         (prepare_initializers_chain): Adjust.
8 2013-11-06  Andrew MacLeod  <amacleod@redhat.com>
10         * gimple.h (block_in_transaction): Move to basic-block.h and rename.
11         (gimple_in_transaction): Use bb_in_transaction.
12         * basic-block.h (bb_in_transaction): Relocate here and rename.
13         * tree-ssa-loop-im.c (execute_sm): Use bb_in_transaction.
15 2013-11-06  Richard Biener  <rguenther@suse.de>
17         * tree.c (drop_tree_overflow): New function.
18         * tree.h (drop_tree_overflow): Declare.
19         * gimplify.c (gimplify_expr): Drop TREE_OVERFLOW.
20         * tree-vrp.c (range_int_cst_singleton_p): Use
21         is_overflow_infinity instead of testing TREE_OVERFLOW.
22         (extract_range_from_assert): Likewise.
23         (zero_nonzero_bits_from_vr): Likewise.
24         (extract_range_basic): Likewise.
25         (register_new_assert_for): Use drop_tree_overflow.
26         (vrp_visit_phi_node): Likewise.
28 2013-11-06  Dodji Seketeli  <dodji@redhat.com>
30         preprocessor/58580
31         * input.h (location_get_source_line): Take an additional line_size
32         parameter.
33         * input.c (get_line): New static function definition.
34         (read_line): Take an additional line_length output parameter to be
35         set to the size of the line.  Use the new get_line function do the
36         actual line reading.
37         (location_get_source_line): Take an additional output line_len
38         parameter.  Update the use of read_line to pass it the line_len
39         parameter.
40         * diagnostic.c (adjust_line): Take an additional input parameter
41         for the length of the line, rather than calculating it with
42         strlen.
43         (diagnostic_show_locus): Adjust the use of
44         location_get_source_line and adjust_line with respect to their new
45         signature.  While displaying a line now, do not stop at the first
46         null byte.  Rather, display the zero byte as a space and keep
47         going until we reach the size of the line.
49 2013-11-06  Eric Botcazou  <ebotcazou@adacore.com>
51         * config/i386/i386.c (ix86_expand_prologue): Optimize stack checking for
52         leaf functions without dynamic stack allocation.
53         * config/ia64/ia64.c (ia64_emit_probe_stack_range): Adjust.
54         (ia64_expand_prologue): Likewise.
55         * config/mips/mips.c (mips_expand_prologue): Likewise.
56         * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
57         * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
58         (sparc_flat_expand_prologue): Likewise.
60 2013-11-06  James Greenhalgh  <james.greenhalgh@arm.com>
62         * config/aarch64/arm_neon.h (__ST2_LANE_FUNC): Better model data size.
63         (__ST3_LANE_FUNC): Likewise.
64         (__ST4_LANE_FUNC): Likewise.
66 2013-11-06  Nick Clifton  <nickc@redhat.com>
68         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define the
69         name returned by msp430_mcu_name.
70         (LIB_SPEC): If a -T option has not been specified then set a
71         default, mcu-specific, linker script.
72         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add more mcu names.
73         * config/msp430/msp430.c (msp430x_names): Likewise.
74         Alpha sort the names for ease of comparison.
75         (msp430_mcu_name): New function:  Returns a string suitable for
76         use as a C preprocessor symbol based upon the name of the MCU
77         being targeted.
78         (msp430_option_override): Accept msp430x and msp430xv2 as generic
79         mcu names.
80         * config/msp430/msp430-protos.h (msp430_mcu_name): Prototype.
82         * gcc.c (do_spec_1): Do not insert a space after a %* substitution
83         unless it is the last part of a spec substring.
84         * doc/invoke.texi (Spec Files): Document space insertion
85         behaviour of %*.
87 2013-11-06  Christian Bruel  <christian.bruel@st.com>
89         * config/sh/sh-mem.cc (sh_expand_cmpnstr, sh_expand_cmpstr):
90         Factorize probabilities, Use adjust_address instead of
91         adjust_automodify_address when possible. Enable for optimize.
92         (sh_expand_strlen): New function.
93         * config/sh/sh-protos.h (sh_expand_strlen): Declare.
94         * config/sh/sh.md (strlensi): New pattern.
95         (UNSPEC_BUILTIN_STRLEN): Define.
97 2013-11-06  Jakub Jelinek  <jakub@redhat.com>
99         PR middle-end/58970
100         * expr.c (get_bit_range): Handle *offset == NULL_TREE.
101         (expand_assignment): If *bitpos is negative, set *offset
102         and adjust *bitpos, so that it is not negative.
104 2013-11-06  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
106         * config/i386/bdver3.md : Added two additional decoder units
107         to support issue rate of 4 and remodeled vector unit.
108         * config/i386/i386.c (ix86_issue_rate): Issue rate for BD
109         architectures is set to 4.
110         * config/i386/i386.c (ia32_multipass_dfa_lookahead): DFA
111         lookahead is set to 4 for BD architectures.
113 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
115         * config/rs6000/rs6000.c (rs6000_option_override_internal):
116         Remove restriction against use of VSX instructions when generating
117         code for little endian mode.
119 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
121         * config/rs6000/altivec.md (mulv4si3): Ensure we generate vmulouh
122         for both big and little endian.
123         (mulv8hi3): Swap input operands for merge high and merge low
124         instructions for little endian.
126 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
128         * config/rs6000/altivec.md (vec_widen_umult_even_v16qi): Change
129         define_insn to define_expand that uses even patterns for big
130         endian and odd patterns for little endian.
131         (vec_widen_smult_even_v16qi): Likewise.
132         (vec_widen_umult_even_v8hi): Likewise.
133         (vec_widen_smult_even_v8hi): Likewise.
134         (vec_widen_umult_odd_v16qi): Likewise.
135         (vec_widen_smult_odd_v16qi): Likewise.
136         (vec_widen_umult_odd_v8hi): Likewise.
137         (vec_widen_smult_odd_v8hi): Likewise.
138         (altivec_vmuleub): New define_insn.
139         (altivec_vmuloub): Likewise.
140         (altivec_vmulesb): Likewise.
141         (altivec_vmulosb): Likewise.
142         (altivec_vmuleuh): Likewise.
143         (altivec_vmulouh): Likewise.
144         (altivec_vmulesh): Likewise.
145         (altivec_vmulosh): Likewise.
147 2013-11-05  Mike Stump  <mikestump@comcast.net>
149         * Makefile.in (mostlyclean): Remove c-family objects.
151 2013-11-05  Ian Lance Taylor  <iant@google.com>
153         * config/i386/sync.md (atomic_compare_and_swap<dwi>_doubleword):
154         If possible, add .cfi directives to record change to bx.
155         * config/i386/i386.c (ix86_emit_cfi): New function.
156         * config/i386/i386-protos.h (ix86_emit_cfi): Declare.
158 2013-11-05  Steven Bosscher  <steven@gcc.gnu.org>
160         * rtlanal.c (tablejump_p): Expect a JUMP_TABLE_DATA to always follow
161         immediately after a label for a tablejump pattern.
163         * config/arm/arm.c (is_jump_table): Remove.
164         (create_fix_barrier): Use tablejump_p instead.
165         (arm_reorg): Likewise.
166         (thumb1_output_casesi): Expect JUMP_TABLE_DATA to always be NEXT_INSN.
167         (thumb2_output_casesi): Likewise.
168         * config/aarch64/aarch64.c (aarch64_output_casesi): Likewise.
169         * config/sh/sh.md (casesi_worker_1, casesi_worker_2,
170         casesi_shift_media, casesi_load_media): Likewise.
171         * config/iq2000/iq2000.md: Likewise (in anonymous define_insn).
172         * config/microblaze/microblaze.md: Likewise.
174 2013-11-05  Tobias Burnus  <burnus@net-b.de>
176         * doc/invoke.texi (-Wdate-time): Document.
178 2013-11-05  Richard Sandiford  <rdsandiford@googlemail.com>
180         * double-int.c (lshift_double, rshift_double): Remove
181         SHIFT_COUNT_TRUNCATED handling.
183 2013-11-05  Jeff Law  <law@redhat.com>
185         * Makefile.in (OBJS): Add gimple-ssa-isolate-paths.o
186         * common.opt (-fisolate-erroneous-paths): Add option and
187         documentation.
188         * gimple-ssa-isolate-paths.c: New file.
189         * gimple.c (check_loadstore): New function.
190         (infer_nonnull_range): Moved into gimple.c from tree-vrp.c
191         Verify OP is in the argument list and the argument corresponding
192         to OP is a pointer type.  Use operand_equal_p rather than
193         pointer equality when testing if OP is on the nonnull list.
194         Use check_loadstore rather than count_ptr_derefs.  Handle
195         GIMPLE_RETURN statements.
196         * tree-vrp.c (infer_nonnull_range): Remove.
197         * gimple.h (infer_nonnull_range): Declare.
198         * opts.c (default_options_table): Add OPT_fisolate_erroneous_paths.
199         * passes.def: Add pass_isolate_erroneous_paths.
200         * timevar.def (TV_ISOLATE_ERRONEOUS_PATHS): New timevar.
201         * tree-pass.h (make_pass_isolate_erroneous_paths): Declare.
202         * tree-ssa.c (struct count_ptr_d): Remove.
203         (count_ptr_derefs, count_uses_and_derefs): Remove.
204         * tree-ssa.h (count_uses_and_derefs): Remove.
206 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
208         PR rtl-optimization/58997
209         * loop-iv.c (iv_subreg): For IV_UNKNOWN_EXTEND, expect
210         get_iv_value to be in iv->mode rather than iv->extend_mode.
211         (iv_extend): Likewise.  Otherwise, if iv->extend != extend,
212         use lowpart_subreg on get_iv_value before calling simplify_gen_unary.
213         * loop-unswitch.c (may_unswitch_on): Make sure op[i] is in the right
214         mode.
216 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
218         * gimple.h: Move some prototypes to gimple-expr.h and add to include
219         list.
220         (extract_ops_from_tree, gimple_call_addr_fndecl, is_gimple_reg_type):
221         Move to gimple-expr.h.
222         * gimple-expr.h: New file.  Relocate some prototypes from gimple.h.
223         (types_compatible_p, is_gimple_reg_type, is_gimple_variable,
224         is_gimple_id, virtual_operand_p, is_gimple_addressable,
225         is_gimple_constant, extract_ops_from_tree, gimple_call_addr_fndecl):
226         Relocate here.
227         * gimple.c (extract_ops_from_tree_1, gimple_cond_get_ops_from_tree,
228         gimple_set_body, gimple_body, gimple_has_body_p, is_gimple_lvalue,
229         is_gimple_condexpr, is_gimple_addressable, is_gimple_constant,
230         is_gimple_address, is_gimple_invariant_address,
231         is_gimple_ip_invariant_address, is_gimple_min_invariant,
232         is_gimple_ip_invariant, is_gimple_variable, is_gimple_id,
233         virtual_operand_p, is_gimple_reg, is_gimple_val, is_gimple_asm_val,
234         is_gimple_min_lval, is_gimple_call_addr, is_gimple_mem_ref_addr,
235         gimple_decl_printable_name, useless_type_conversion_p,
236         types_compatible_p, gimple_can_coalesce_p, copy_var_decl): Move to 
237         gimple-expr.[ch].
238         * gimple-expr.c: New File.
239         (useless_type_conversion_p, gimple_set_body, gimple_body,
240         gimple_has_body_p, gimple_decl_printable_name, copy_var_decl,
241         gimple_can_coalesce_p, extract_ops_from_tree_1, 
242         gimple_cond_get_ops_from_tree, is_gimple_lvalue, is_gimple_condexpr,
243         is_gimple_address, is_gimple_invariant_address,
244         is_gimple_ip_invariant_address, is_gimple_min_invariant,
245         is_gimple_ip_invariant, is_gimple_reg, is_gimple_val,
246         is_gimple_asm_val, is_gimple_min_lval, is_gimple_call_addr,
247         is_gimple_mem_ref_addr): Relocate here.
248         * Makefile.in (OBJS): Add gimple-expr.o.
250 2013-11-05  David Malcolm  <dmalcolm@redhat.com>
252         * gengtype-parse.c (struct_field_seq): Support empty structs.
254 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
256         * config/i386/t-rtems (MULTILIB_MATCHES): Fix option typos.
258 2013-11-05  Uros Bizjak  <ubizjak@gmail.com>
260         * config/i386/i386-c.c (ix86_target_macros): Define _SOFT_FLOAT
261         for !TARGET_80387.
262         * config/i386/rtemself.h (TARGET_OS_CPP_BUILTINS): Do not define
263         _SOFT_FLOAT here.
264         (LONG_DOUBLE_TYPE_SIZE): New define.
265         (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Ditto.
267 2013-11-05  Paolo Carlini  <paolo.carlini@oracle.com>
269         PR c++/58724
270         * doc/extend.texi [visibility ("visibility_type")]: Add example
271         about visibility attribute on namespace declaration.
273 2013-11-05  Richard Biener  <rguenther@suse.de>
275         PR ipa/58492
276         * passes.def (all_passes): Start with pass_fixup_cfg again.
278 2013-11-05  Richard Biener  <rguenther@suse.de>
280         PR tree-optimization/58955
281         * tree-loop-distribution.c (pg_add_dependence_edges): Fix
282         edge direction.
284 2013-11-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
286         * config/rs6000/vector.md (vec_pack_sfix_trunc_v2df): Adjust for
287         little endian.
288         (vec_pack_ufix_trunc_v2df): Likewise.
290 2013-11-05  H.J. Lu  <hongjiu.lu@intel.com>
292         PR middle-end/58981
293         * doc/md.texi (@code{movmem@var{m}}): Specify Pmode as mode of
294         pattern, instead of word_mode.
296         * expr.c (emit_block_move_via_movmem): Don't use mode wider than
297         Pmode for size.
298         (set_storage_via_setmem): Likewise.
300 2013-11-05  Andrew MacLeod  <amacleod@redhat.com>
302         * tree-outof-ssa.c (queue_phi_copy_p): Combine phi_ssa_name_p from
303         gimple.h and the rest of the condition in eliminate_build.
304         (eliminate_build): Call new routine.
305         * gimple.h (phi_ssa_name_p): Delete.
307 2013-11-05  Trevor Saunders  <tsaunders@mozilla.com>
309         * vec.c (vec_prefix::calculate_allocation): Don't try to handle the
310         case of no prefix and reserving zero slots, because when that's the
311         case we'll never get here.
312         * vec.h (va_heap::reserve): Don't try and handle
313         vec_prefix::calculate_allocation returning zero because that should
314         never happen.
316 2013-11-05  Richard Biener  <rguenther@suse.de>
318         PR middle-end/58941
319         * tree-dfa.c (get_ref_base_and_extent): Merge common code
320         in MEM_REF and TARGET_MEM_REF handling.  Make sure to
321         process trailing array detection before diving into the
322         view-converted object (and possibly apply some extra offset).
324 2013-11-05  Joseph Myers  <joseph@codesourcery.com>
326         * config/i386/i386.c (ix86_float_exceptions_rounding_supported_p):
327         New function.
328         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Define.
330 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
332         PR tree-optimization/58958
333         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Use
334         get_addr_base_and_unit_offset instead of get_ref_base_and_extent.
336 2013-11-05  Marc Glisse  <marc.glisse@inria.fr>
338         * tree-ssa-alias.h (ranges_overlap_p): Handle negative offsets.
339         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
341 2013-11-05  Jakub Jelinek  <jakub@redhat.com>
343         PR tree-optimization/58984
344         * ipa-prop.c (ipa_load_from_parm_agg_1): Add SIZE_P argument,
345         set *SIZE_P if non-NULL on success.
346         (ipa_load_from_parm_agg, ipa_analyze_indirect_call_uses): Adjust
347         callers.
348         (ipcp_transform_function): Likewise.  Punt if size of access
349         is different from TYPE_SIZE on v->value's type.
351 2013-11-05  Tobias Burnus  <burnus@net-b.de>
353         * doc/invoke.texi (-fopenmp-simd): Document new option.
354         * gimplify.c (gimplify_body): Accept -fopenmp-simd.
355         * omp-low.c (execute_expand_omp, execute_lower_omp): Ditto.
356         * tree.c (attribute_value_equal): Ditto.
358 2013-11-04  Wei Mi  <wmi@google.com>
360         * sched-rgn.c (add_branch_dependences): Keep insns in
361         a SCHED_GROUP at the end of BB to remain their location.
363 2013-11-04  Wei Mi  <wmi@google.com>
365         * config/i386/i386.c (memory_address_length): Extract a part
366         of code to rip_relative_addr_p.
367         (rip_relative_addr_p): New Function.
368         (ix86_macro_fusion_p): Ditto.
369         (ix86_macro_fusion_pair_p): Ditto.
370         * config/i386/i386.h: Add new tune features about macro-fusion.
371         * config/i386/x86-tune.def (DEF_TUNE): Ditto.
372         * doc/tm.texi: Generated.
373         * doc/tm.texi.in: Ditto.
374         * haifa-sched.c (try_group_insn): New Function.
375         (group_insns_for_macro_fusion): Ditto.
376         (sched_init): Call group_insns_for_macro_fusion.
377         * target.def: Add two hooks: macro_fusion_p and
378         macro_fusion_pair_p.
380 2013-11-04  Kostya Serebryany  <kcc@google.com>
382         Update to match the changed asan API.
383         * asan.c (asan_function_start): New function.
384         (asan_emit_stack_protection): Update the string stored in the
385         stack red zone to match new API.  Store the PC of the current
386         function in the red zone.
387         (asan_global_struct): Update the __asan_global definition to match
388         the new API.
389         (asan_add_global): Ditto.
390         * asan.h (asan_function_start): New prototype.
391         * final.c (final_start_function): Call asan_function_start.
392         * sanitizer.def (BUILT_IN_ASAN_INIT): Rename __asan_init_v1
393         to __asan_init_v3.
395 2013-11-04  Wei Mi  <wmi@google.com>
397         * config/i386/i386-c.c (ix86_target_macros_internal): Separate
398         PROCESSOR_COREI7_AVX out from PROCESSOR_COREI7.
399         * config/i386/i386.c (ix86_option_override_internal): Ditto.
400         (ix86_issue_rate): Ditto.
401         (ix86_adjust_cost): Ditto.
402         (ia32_multipass_dfa_lookahead): Ditto.
403         (ix86_sched_init_global): Ditto.
404         (get_builtin_code_for_version): Ditto.
405         * config/i386/i386.h (enum target_cpu_default): Ditto.
406         (enum processor_type): Ditto.
407         * config/i386/x86-tune.def (DEF_TUNE): Ditto.
409 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
411         PR rtl-optimization/58967
412         * config/rs6000/rs6000.c (legitimate_lo_sum_address_p): Remove
413         !lra_in_progress for mode sizes bigger word.
415 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
417         * config/rs6000/altivec.md (vec_widen_umult_hi_v16qi): Swap
418         arguments to merge instruction for little endian.
419         (vec_widen_umult_lo_v16qi): Likewise.
420         (vec_widen_smult_hi_v16qi): Likewise.
421         (vec_widen_smult_lo_v16qi): Likewise.
422         (vec_widen_umult_hi_v8hi): Likewise.
423         (vec_widen_umult_lo_v8hi): Likewise.
424         (vec_widen_smult_hi_v8hi): Likewise.
425         (vec_widen_smult_lo_v8hi): Likewise.
427 2013-11-04  Ian Lance Taylor  <iant@google.com>
429         * builtins.def (ATTR_NOTHROWCALL_LEAF_LIST): Define.
430         * sync-builtins.def: Use ATTR_NOTHROWCALL_LEAF_LIST for all sync
431         builtins that take pointers.
432         * lto-opts.c (lto_write_options): Write -fnon-call-exceptions if set.
433         * lto-wrapper.c (merge_and_complain): Collect OPT_fnon_call_exceptions.
434         (run_gcc): Pass -fnon-call-exceptions.
436 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
438         * optabs.c (expand_vec_perm): Revert one incorrect line from
439         2013-10-31 change.
441         PR tree-optimization/58978
442         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): Don't modify
443         use_stmt by single_imm_use directly.  Only call single_imm_use
444         on SSA_NAMEs.
446 2013-11-04  Vladimir Makarov  <vmakarov@redhat.com>
448         PR rtl-optimization/58968
449         * lra-spills.c (return_regno_p): New function.
450         (lra_final_code_change): Use it.
452 2013-11-04  Joseph Myers  <joseph@codesourcery.com>
454         * doc/cpp.texi (__GCC_IEC_559, __GCC_IEC_559_COMPLEX): Document macros.
455         * target.def (float_exceptions_rounding_supported_p): New hook.
456         * targhooks.c (default_float_exceptions_rounding_supported_p): New
457         function.
458         * targhooks.h (default_float_exceptions_rounding_supported_p): Declare.
459         * doc/tm.texi.in (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P):
460         New @hook.
461         * doc/tm.texi: Regenerate.
462         * config.gcc (powerpc*-*-linux*): Set extra_objs.
463         * config/rs6000/rs6000-linux.c: New file.
464         * config/rs6000/rs6000-protos.h
465         (rs6000_linux_float_exceptions_rounding_supported_p): Declare.
466         * config/rs6000/linux.h
467         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): New macro.
468         * config/rs6000/linux64.h
469         (TARGET_FLOAT_EXCEPTIONS_ROUNDING_SUPPORTED_P): Likewise.
470         * config/rs6000/t-linux (rs6000-linux.o): New rule.
471         * config/rs6000/t-linux64 (rs6000-linux.o): Likewise.
473 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
475         * config/rs6000/vsx.md (*vsx_le_perm_store_<mode> for VSX_D):
476         Replace the define_insn_and_split with a define_insn and two
477         define_splits, with the split after reload re-permuting the source
478         register to its original value.
479         (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
480         (*vsx_le_perm_store_v8hi): Likewise.
481         (*vsx_le_perm_store_v16qi): Likewise.
483 2013-11-04  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
485         * config/rs6000/vector.md (vec_pack_trunc_v2df):  Adjust for
486         little endian.
488 2013-11-04  Jakub Jelinek  <jakub@redhat.com>
490         PR tree-optimization/58946
491         * tree-ssa-reassoc.c (maybe_optimize_range_tests): Update all
492         bbs with bbinfo[idx].op != NULL before all blocks with
493         bbinfo[idx].op == NULL.
495 2013-11-04  Richard Sandiford  <rdsandiford@googlemail.com>
497         * config/avr/avr-log.c (avr_double_int_pop_digit): Delete.
498         (avr_dump_double_int_hex): Likewise.
499         (avr_log_vadump): Remove %D and %X handling.
500         * config/avr/avr.c (avr_double_int_push_digit): Delete.
501         (avr_map_op_t): Change map from double_int to unsigned int.
502         (avr_map_op): Update accordingly.
503         (avr_map, avr_map_metric, avr_has_nibble_0xf, avr_map_decompose)
504         (avr_move_bits, avr_out_insert_bits, avr_fold_builtin): Operate on
505         unsigned ints rather than double_ints.
507 2013-11-03  Marek Polacek  <polacek@redhat.com>
509         Implement -fsanitize=vla-bound.
510         * opts.c (common_handle_option): Handle vla-bound.
511         * sanitizer.def (BUILT_IN_UBSAN_HANDLE_VLA_BOUND_NOT_POSITIVE): Define.
512         * flag-types.h (enum sanitize_code): Add SANITIZE_VLA.
513         * asan.c (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR.
515 2013-11-02  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
517         * config/rs6000/rs6000.c (rs6000_expand_vector_set): Adjust for
518         little endian.
520 2013-11-02  Uros Bizjak  <ubizjak@gmail.com>
522         * config/i386/constraints.md (Ts, Tv): New address constrains.
523         * config/i386/i386.md (*lea<mode>, *<mode>_<bndcheck>): Use Ts
524         constraint for address_no_seg_operand.
525         * config/i386/sse.md (*avx512pf_gatherpf<mode>_mask)
526         (*avx512pf_gatherpf<mode>, *avx512pf_scatterpf<mode>_mask)
527         (*avx512pf_scatterpf<mode>, *avx2_gathersi<mode>)
528         (*avx2_gathersi<mode>_2, *avx2_gatherdi<mode>, *avx2_gatherdi<mode>_2)
529         (*avx2_gatherdi<mode>_3, *avx2_gatherdi<mode>_4)
530         (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2)
531         (*avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2)
532         (*avx512f_scattersi<mode> *avx512f_scatterdi<mode>): Use Tv
533         constraint for vsib_address_operand.
535 2013-11-02  Steven Bosscher  <steven@gcc.gnu.org>
537         * gcse.c (pre_delete): Remove references to regmove from comments.
538         * recog.c: (validate_replace_rtx_1): Likewise.
539         * config/rl78/rl78.c: Likewise.
540         * config/v850/v850.h: Likewise, and remove unused ENABLE_REGMOVE_PASS.
541         * common/config/m32r/m32r-common.c: Don't manipulate OPT_fregmove.
542         * common/config/mmix/mmix-common.c: Likewise.
544 2013-11-01  Trevor Saunders  <tsaunders@mozilla.com>
546         * function.c (reorder_blocks): Convert block_stack to a stack_vec.
547         * gimplify.c (gimplify_compound_lval): Likewise.
548         * graphite-clast-to-gimple.c (gloog): Likewise.
549         * graphite-dependences.c (loop_is_parallel_p): Likewise.
550         * graphite-scop-detection.c (scopdet_basic_block_info): Likewise.
551         (limit_scop); Likewise.
552         (build_scops): Likewise.
553         (dot_scop): Likewise.
554         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
555         (build_scop_drs): Likewise.
556         (insert_stmts): Likewise.
557         (insert_out_of_ssa_copy): Likewise.
558         (remove_phi): Likewise.
559         (rewrite_commutative_reductions_out_of_ssa_close_phi): Likewise.
560         * hw-doloop.c (discover_loop): Likewise.
561         * tree-call-cdce.c (shrink_wrap_one_built_in_call): Likewise.
562         * tree-dfa.c (dump_enumerated_decls): Likewise.
563         * tree-if-conv.c (if_convertable_loop_p): Likewise.
564         * tree-inline.c (tree_function_versioning): Likewise.
565         * tree-loop-distribution.c (build_rdg): Likewise.
566         (rdg_flag_vertex_and_dependent): Likewise.
567         (distribute_loop): Likewise.
568         * tree-parloops.c (loop_parallel_p): Likewise.
569         (eliminate_local_variables): Likewise.
570         (separate_decls_in_region): Likewise.
571         * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
572         * tree-ssa-phiopt.c (cond_if_else_store_replacement): Likewise.
573         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
574         * tree-vect-loop.c (vect_analyze_scaler_cycles_1): Likewise.
575         * tree-vect-patterns.c (vect_pattern_recog): Likewise.
576         * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
577         (vectorizable_condition): Likewise.
579 2013-11-01  Uros Bizjak  <ubizjak@gmail.com>
581         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): Always define as 0/1.
582         * configure: Regenerate.
583         * config/i386/i386.md (*movdi_internal): Change
584         HAVE_AS_IX86_INTERUNIT_MOVQ to runtime check.
585         (*movdf_internal): Ditto.
586         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
587         * config/i386/sse.md (vec_concatv2di): Output interunit movq
588         for HAVE_AS_IX86_INTERUNIT_MOVQ targets.
590 2013-10-31  Robert Suchanek  <Robert.Suchanek@imgtec.com>
592         * lra-spills.c (assign_spill_hard_regs): Remove statement terminator
593         after comment.
595 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
597         Automated part of renaming of symtab_node_base to symtab_node.
599         Patch autogenerated by rename_symtab.py from
600         https://github.com/davidmalcolm/gcc-refactoring-scripts
601         revision 58bb219cc090b2f4516a9297d868c245495ee622
602         with ChangeLog entry fixed up by hand.
604         * cgraph.c (x_cgraph_nodes_queue): Rename symtab_node_base to
605         symtab_node.
606         (cgraph_node_for_asm): Likewise.
607         * cgraph.h (symtab_node_base): Likewise.
608         (cgraph_node): Likewise.
609         (varpool_node): Likewise.
610         (is_a_helper <cgraph_node>::test): Likewise.
611         (is_a_helper <varpool_node>::test): Likewise.
612         (symtab_nodes): Likewise.
613         (symtab_register_node): Likewise.
614         (symtab_unregister_node): Likewise.
615         (symtab_remove_node): Likewise.
616         (symtab_get_node): Likewise.
617         (symtab_node_for_asm): Likewise.
618         (symtab_node_asm_name): Likewise.
619         (symtab_node_name): Likewise.
620         (symtab_insert_node_to_hashtable): Likewise.
621         (symtab_add_to_same_comdat_group): Likewise.
622         (symtab_dissolve_same_comdat_group_list): Likewise.
623         (dump_symtab_node): Likewise.
624         (debug_symtab_node): Likewise.
625         (dump_symtab_base): Likewise.
626         (verify_symtab_node): Likewise.
627         (verify_symtab_base): Likewise.
628         (symtab_used_from_object_file_p): Likewise.
629         (symtab_alias_ultimate_target): Likewise.
630         (symtab_resolve_alias): Likewise.
631         (fixup_same_cpp_alias_visibility): Likewise.
632         (symtab_for_node_and_aliases): Likewise.
633         (symtab_nonoverwritable_alias): Likewise.
634         (availability symtab_node_availability): Likewise.
635         (symtab_semantically_equivalent_p): Likewise.
636         (fixup_same_cpp_alias_visibility): Likewise.
637         (symtab_prevail_in_asm_name_hash): Likewise.
638         (cgraph): Likewise.
639         (varpool): Likewise.
640         (varpool_first_variable): Likewise.
641         (varpool_next_variable): Likewise.
642         (varpool_first_static_initializer): Likewise.
643         (varpool_next_static_initializer): Likewise.
644         (varpool_first_defined_variable): Likewise.
645         (varpool_next_defined_variable): Likewise.
646         (cgraph_first_defined_function): Likewise.
647         (cgraph_next_defined_function): Likewise.
648         (cgraph_first_function): Likewise.
649         (cgraph_next_function): Likewise.
650         (cgraph_first_function_with_gimple_body): Likewise.
651         (cgraph_next_function_with_gimple_body): Likewise.
652         (symtab_alias_target): Likewise.
653         (symtab_real_symbol_p): Likewise.
654         (symtab_can_be_discarded): Likewise.
655         * cgraphbuild.c (mark_address): Likewise.
656         (mark_load): Likewise.
657         (mark_store): Likewise.
658         * cgraphunit.c (decide_is_symbol_needed): Likewise.
659         (first): Likewise.
660         (enqueue_node): Likewise.
661         (referred_to_p): Likewise.
662         (cgraph_process_same_body_aliases): Likewise.
663         (analyze_functions): Likewise.
664         (handle_alias_pairs): Likewise.
665         (output_weakrefs): Likewise.
666         (compile): Likewise.
667         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
668         * ipa-inline-analysis.c (inline_write_summary): Likewise.
669         * ipa-prop.c (remove_described_reference): Likewise.
670         (try_decrement_rdesc_refcount): Likewise.
671         (ipa_edge_duplication_hook): Likewise.
672         * ipa-ref.c (ipa_record_reference): Likewise.
673         (ipa_maybe_record_reference): Likewise.
674         (ipa_clone_ref): Likewise.
675         (ipa_clone_references): Likewise.
676         (ipa_clone_referring): Likewise.
677         (ipa_find_reference): Likewise.
678         (ipa_remove_stmt_references): Likewise.
679         (ipa_clear_stmts_in_references): Likewise.
680         * ipa-ref.h (symtab_node_base): Likewise.
681         (ipa_ref): Likewise.
682         (ipa_record_reference): Likewise.
683         (ipa_maybe_record_reference): Likewise.
684         (ipa_clone_references): Likewise.
685         (ipa_clone_referring): Likewise.
686         (ipa_clone_ref): Likewise.
687         (ipa_find_reference): Likewise.
688         (ipa_remove_stmt_references): Likewise.
689         (ipa_clear_stmts_in_references): Likewise.
690         * ipa-reference.c (ipa_reference_write_optimization_summary):
691         Likewise.
692         * ipa.c (enqueue_node): Likewise.
693         (process_references): Likewise.
694         (walk_polymorphic_call_targets): Likewise.
695         (symtab_remove_unreachable_nodes): Likewise.
696         (address_taken_from_non_vtable_p): Likewise.
697         (comdat_can_be_unshared_p_1): Likewise.
698         (comdat_can_be_unshared_p): Likewise.
699         (can_replace_by_local_alias): Likewise.
700         (function_and_variable_visibility): Likewise.
701         * is-a.h: Likewise (within example in comment).
702         * lto-cgraph.c (input_cgraph_opt_summary): Likewise.
703         (lto_symtab_encoder_encode): Likewise.
704         (lto_symtab_encoder_delete_node): Likewise.
705         (lto_symtab_encoder_in_partition_p): Likewise.
706         (lto_set_symtab_encoder_in_partition): Likewise.
707         (output_refs): Likewise.
708         (compute_ltrans_boundary): Likewise.
709         (output_symtab): Likewise.
710         (input_node): Likewise.
711         (input_ref): Likewise.
712         (input_edge): Likewise.
713         (input_cgraph_1): Likewise.
714         (input_refs): Likewise.
715         (output_cgraph_opt_summary): Likewise.
716         (input_node_opt_summary): Likewise.
717         (input_cgraph_opt_section): Likewise.
718         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
719         Likewise.
720         * lto-streamer-out.c (lto_output): Likewise.
721         (output_symbol_p): Likewise.
722         (produce_symtab): Likewise.
723         * lto-streamer.h (lto_encoder_entry): Likewise.
724         (lto_free_function_in_decl_state_for_node): Likewise.
725         (lto_symtab_encoder_encode): Likewise.
726         (lto_symtab_encoder_delete_node): Likewise.
727         (lto_symtab_encoder_in_partition_p): Likewise.
728         (lto_set_symtab_encoder_in_partition): Likewise.
729         (lto_symtab_encoder_lookup): Likewise.
730         (lsei_node): Likewise.
731         (lto_symtab_encoder_deref): Likewise.
732         * symtab.c (symtab_hash): Likewise.
733         (assembler_name_hash): Likewise.
734         (symtab_nodes): Likewise.
735         (hash_node): Likewise.
736         (eq_node): Likewise.
737         (hash_node_by_assembler_name): Likewise.
738         (eq_assembler_name): Likewise.
739         (insert_to_assembler_name_hash): Likewise.
740         (unlink_from_assembler_name_hash): Likewise.
741         (symtab_prevail_in_asm_name_hash): Likewise.
742         (symtab_register_node): Likewise.
743         (symtab_insert_node_to_hashtable): Likewise.
744         (symtab_unregister_node): Likewise.
745         (symtab_get_node): Likewise.
746         (symtab_remove_node): Likewise.
747         (symtab_initialize_asm_name_hash): Likewise.
748         (symtab_node_for_asm): Likewise.
749         (symtab_add_to_same_comdat_group): Likewise.
750         (symtab_dissolve_same_comdat_group_list): Likewise.
751         (symtab_node_asm_name): Likewise.
752         (symtab_node_name): Likewise.
753         (dump_symtab_base): Likewise.
754         (dump_symtab_node): Likewise.
755         (dump_symtab): Likewise.
756         (debug_symtab_node): Likewise.
757         (verify_symtab_base): Likewise.
758         (verify_symtab_node): Likewise.
759         (verify_symtab): Likewise.
760         (symtab_used_from_object_file_p): Likewise.
761         (symtab_node_availability): Likewise.
762         (symtab_alias_ultimate_target): Likewise.
763         (fixup_same_cpp_alias_visibility): Likewise.
764         (symtab_resolve_alias): Likewise.
765         (symtab_for_node_and_aliases): Likewise.
766         (symtab_for_node_and_aliases): Likewise.
767         (symtab_nonoverwritable_alias_1): Likewise.
768         (symtab_nonoverwritable_alias): Likewise.
769         (symtab_semantically_equivalent_p): Likewise.
770         * value-prof.c (init_node_map): Likewise.
771         * varasm.c (find_decl): Likewise.
772         * varpool.c (varpool_node_for_asm): Likewise.
773         (varpool_remove_unreferenced_decls): Likewise.
775 2013-10-31  David Malcolm  <dmalcolm@redhat.com>
777         Manual part of renaming of symtab_node_base to symtab_node.
779         * ipa-ref.h (symtab_node): Remove typedef to pointer type, as it
780         clashes with the preferred name for the base class.
781         (const_symtab_node): Remove redundant typedef.
783 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
785         * optabs.c (expand_vec_perm): Avoid vector mode punning
786         SUBREGs in SET_DEST.
787         * expmed.c (store_bit_field_1): Likewise.
788         * config/i386/sse.md (movdi_to_sse, vec_pack_sfix_trunc_v2df,
789         vec_pack_sfix_v2df, vec_shl_<mode>, vec_shr_<mode>,
790         vec_interleave_high<mode>, vec_interleave_low<mode>): Likewise.
791         * config/i386/i386.c (ix86_expand_vector_move_misalign,
792         ix86_expand_sse_movcc, ix86_expand_int_vcond, ix86_expand_vec_perm,
793         ix86_expand_sse_unpack, ix86_expand_args_builtin,
794         ix86_expand_vector_init_duplicate, ix86_expand_vector_set,
795         emit_reduc_half, expand_vec_perm_blend, expand_vec_perm_pshufb,
796         expand_vec_perm_interleave2, expand_vec_perm_pshufb2,
797         expand_vec_perm_vpshufb2_vpermq,
798         expand_vec_perm_vpshufb2_vpermq_even_odd, expand_vec_perm_even_odd_1,
799         expand_vec_perm_broadcast_1, expand_vec_perm_vpshufb4_vpermq2,
800         ix86_expand_sse2_mulv4si3, ix86_expand_pinsr): Likewise.
801         (expand_vec_perm_palignr): Likewise.  Modify a copy of *d rather
802         than *d itself.
804 2013-10-31  Uros Bizjak  <ubizjak@gmail.com>
806         * config/i386/i386.c (ix86_expand_sse2_abs): Rename function arguments.
807         Use gcc_unreachable for unhandled modes.  Do not check results of
808         expand_simple_binop.  If not expanded to target, move the result.
810 2013-10-31  Chung-Ju Wu  <jasonwucj@gmail.com>
811             Shiva Chen  <shiva0217@gmail.com>
813         * config.gcc (nds32*-*-*): Add nds32 target.
814         * config/nds32/nds32.c: New file.
815         * config/nds32/nds32.h: New file.
816         * config/nds32/nds32.md: New file.
817         * config/nds32/constants.md: New file.
818         * config/nds32/constraints.md: New file.
819         * config/nds32/iterators.md: New file.
820         * config/nds32/nds32-doubleword.md: New file.
821         * config/nds32/nds32-intrinsic.md: New file.
822         * config/nds32/nds32_intrinsic.h: New file.
823         * config/nds32/nds32-modes.def: New file.
824         * config/nds32/nds32-multiple.md: New file.
825         * config/nds32/nds32.opt: New file.
826         * config/nds32/nds32-opts.h: New file.
827         * config/nds32/nds32-protos.h: New file.
828         * config/nds32/nds32-peephole2.md: New file.
829         * config/nds32/pipelines.md: New file.
830         * config/nds32/predicates.md: New file.
831         * config/nds32/t-mlibs: New file.
832         * common/config/nds32: New directory and files.
834         * doc/invoke.texi (NDS32 options): Document nds32 specific options.
835         * doc/md.texi (NDS32 family): Document nds32 specific constraints.
836         * doc/install.texi (Cross-Compiler-Specific Options): Document
837         --with-nds32-lib for nds32 target.
838         * doc/extend.texi (Function Attributes, Target Builtins): Document
839         nds32 specific attributes.
841 2013-10-31  Vladimir Makarov  <vmakarov@redhat.com>
843         * lra-constraints (process_alt_operands): Use the result
844         elimination register for operand when matching constraints.
846 2013-10-31  Jakub Jelinek  <jakub@redhat.com>
848         * tree-vrp.c (maybe_set_nonzero_bits): New function.
849         (remove_range_assertions): Call it.
851         * tree.c (tree_ctz): New function.
852         * tree.h (tree_ctz): New prototype.
853         * tree-ssanames.h (get_range_info, get_nonzero_bits): Change
854         first argument from tree to const_tree.
855         * tree-ssanames.c (get_range_info, get_nonzero_bits): Likewise.
856         * tree-vectorizer.h (vect_generate_tmps_on_preheader): New prototype.
857         * tree-vect-loop-manip.c (vect_generate_tmps_on_preheader): No longer
858         static.
859         * expr.c (highest_pow2_factor): Reimplemented using tree_ctz.
860         * tree-vect-loop.c (vect_analyze_loop_operations,
861         vect_transform_loop): Don't force scalar loop for bound just because
862         number of iterations is unknown, only do it if it is not known to be
863         a multiple of vectorization_factor.
864         * builtins.c (get_object_alignment_2): Use tree_ctz on offset.
866         * gimple-pretty-print.c (dump_ssaname_info): Print newline also
867         in case of VR_VARYING.  Print get_nonzero_bits if not all ones.
868         * tree-ssanames.h (struct range_info_def): Add nonzero_bits field.
869         (set_nonzero_bits, get_nonzero_bits): New prototypes.
870         * tree-ssa-ccp.c (get_default_value): Use get_range_info to see if
871         a default def isn't partially constant.
872         (ccp_finalize): If after IPA, set_range_info if integral SSA_NAME
873         is known to be partially zero.
874         (evaluate_stmt): If we'd return otherwise VARYING, use get_range_info
875         to see if a default def isn't partially constant.
876         * tree-ssanames.c (set_range_info): Initialize nonzero_bits upon
877         creation of a range, if VR_RANGE, try to improve nonzero_bits from
878         the range.
879         (set_nonzero_bits, get_nonzero_bits): New functions.
881         * tree-cfg.c (assert_unreachable_fallthru_edge_p): New function.
882         * tree-cfg.h (assert_unreachable_fallthru_edge_p): New prototype.
883         * tree-vrp.c (all_imm_uses_in_stmt_or_feed_cond): New function.
884         (remove_range_assertions): If ASSERT_EXPR_VAR has no other immediate
885         uses but in the condition and ASSERT_EXPR and the other successor of
886         the predecessor bb is __builtin_unreachable (), set_range_info of the
887         ASSERT_EXPR_VAR to the range info of the ASSERT_EXPR's lhs.
889 2013-10-31  Martin Jambor  <mjambor@suse.cz>
891         PR rtl-optimization/58934
892         Revert:
893         2013-10-30  Martin Jambor  <mjambor@suse.cz>
894         PR rtl-optimization/10474
895         * ira.c (find_moveable_pseudos): Do not calculate dominance info
896         nor df analysis.
897         (interesting_dest_for_shprep): New function.
898         (split_live_ranges_for_shrink_wrap): Likewise.
899         (ira): Calculate dominance info and df analysis. Call
900         split_live_ranges_for_shrink_wrap.
902 2013-10-31  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
903             Yury Gribov  <y.gribov@samsung.com>
905         PR sanitizer/58543
906         * asan.c (asan_clear_shadow): Allocate a new vreg for temporary
907         shadow pointer to avoid clobbering the main one.
909 2013-10-31  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
911         * lower-subreg.c (resolve_simple_move): Copy REG_INC note.
913 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
915         PR bootstrap/58933
916         * ira-color.c (update_costs_from_copies): Add new parameter.  Use
917         it for calling update_costs_from_allocno.
918         (assign_hard_reg): Call restore_costs_from_copies only for
919         !retry_p.  Pass new argument to update_costs_from_copies.
920         (color_pass): Pass new argument to update_costs_from_copies.
921         (ira_mark_allocation_change): Ditto.
923 2013-10-30  Sharad Singhai  <singhai@google.com>
925         PR middle-end/58134
926         * opts.c (common_handle_option): Remove deprecated option
927         -ftree-vectorizer-verbose.
928         * doc/invoke.texi (Debugging Options): Ditto.
929         * opts-global.c (handle_common_deferred_options): Ditto.
930         (dump_remap_tree_vectorizer_verbose): Delete.
931         * common.opt: Set -ftree-vectorizer-verbose as an ignored option.
933 2013-10-30  DJ Delorie  <dj@redhat.com>
935         * config/rx/rx.c (ADD_RX_BUILTIN0): New macro, used for builtins
936         that take no arguments.
938 2013-10-30  Joern Rennecke  <joern.rennecke@embecosm.com>
940         PR other/58545
941         * reload1.c (update_eliminables_and_spill): New function, broken
942         out of reload.
943         (reload): Use it.  Check for frame size change after frame size
944         alignment, and call update_eliminables_and_spill first if continue-ing.
946 2013-10-30  Cong Hou  <congh@google.com>
948         PR target/58762
949         * config/i386/i386-protos.h (ix86_expand_sse2_abs): New function.
950         * config/i386/i386.c (ix86_expand_sse2_abs): New function.
951         * config/i386/sse.md: Add SSE2 support to abs (8/16/32-bit-int).
953 2013-10-18  Mikael Pettersson  <mikpelinux@gmail.com>
955         PR rtl-optimization/58369
956         * reload1.c (compute_reload_subreg_offset): New function.
957         (choose_reload_regs): Use it to pass endian-correct
958         offset to subreg_regno_offset.
960 2013-10-30  Tobias Burnus  <burnus@net-b.de>
962         PR other/33426
963         * tree-cfg.c (replace_loop_annotate): Replace warning by
964         warning_at.
966 2013-10-30  Jason Merrill  <jason@redhat.com>
968         * configure.ac (loose_warn): Add -Wno-format if
969         --disable-build-format-warnings.
971 2013-10-30  David Malcolm  <dmalcolm@redhat.com>
973         * cgraphunit.c (analyze_functions): Split symtab_node declarations
974         onto multiple lines to make things easier for rename_symtab.py.
976         * symtab.c (symtab_dissolve_same_comdat_group_list): Likewise.
977         (symtab_semantically_equivalent_p): Likewise.
979 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
981         PR target/58784
982         * lra.c (check_rtl): Remove address check before LRA work.
984 2013-10-30  Marc Glisse  <marc.glisse@inria.fr>
986         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for a
987         POINTER_PLUS_EXPR in the defining statement.
989 2013-10-30  Vladimir Makarov  <vmakarov@redhat.com>
991         * regmove.c: Remove.
992         * tree-pass.h (make_pass_regmove): Remove.
993         * timevar.def (TV_REGMOVE): Remove.
994         * passes.def (pass_regmove): Remove.
995         * opts.c (default_options_table): Remove entry for regmove.
996         * doc/passes.texi: Remove regmove pass description.
997         * doc/invoke.texi (-foptimize-register-move, -fregmove): Remove
998         options.
999         (-fdump-rtl-regmove): Ditto.
1000         * common.opt (foptimize-register-move, fregmove): Ignore.
1001         * Makefile.in (OBJS): Remove regmove.o.
1002         * regmove.c: Remove.
1003         * ira-int.h (struct ira_allocno_pref, ira_pref_t): New structure
1004         and type.
1005         (struct ira_allocno) New member allocno_prefs.
1006         (ALLOCNO_PREFS): New macro.
1007         (ira_prefs, ira_prefs_num): New external vars.
1008         (ira_setup_alts, ira_get_dup_out_num, ira_debug_pref): New prototypes.
1009         (ira_debug_prefs, ira_debug_allocno_prefs, ira_create_pref): Ditto.
1010         (ira_add_allocno_pref, ira_remove_pref, ira_remove_allocno_prefs):
1011         Ditto.
1012         (ira_add_allocno_copy_to_list): Remove prototype.
1013         (ira_swap_allocno_copy_ends_if_necessary): Ditto.
1014         (ira_pref_iterator): New type.
1015         (ira_pref_iter_init, ira_pref_iter_cond): New functions.
1016         (FOR_EACH_PREF): New macro.
1017         * ira.c (commutative_constraint_p): Move from ira-conflicts.c.
1018         (ira_get_dup_out_num): Ditto. Rename from get_dup_num.  Modify the
1019         code.
1020         (ira_setup_alts): New function.
1021         (decrease_live_ranges_number): New function.
1022         (ira): Call the above function.
1023         * ira-build.c (ira_prefs, ira_prefs_num): New global vars.
1024         (ira_create_allocno): Initialize allocno prefs.
1025         (pref_pool, pref_vec): New static vars.
1026         (initiate_prefs, find_allocno_pref, ira_create_pref): New
1027         functions.
1028         (add_allocno_pref_to_list, ira_add_allocno_pref, print_pref): Ditto.
1029         (ira_debug_pref, print_prefs, ira_debug_prefs): Ditto.
1030         (print_allocno_prefs, ira_debug_allocno_prefs, finish_pref): Ditto.
1031         (ira_remove_pref, ira_remove_allocno_prefs, finish_prefs): Ditto.
1032         (ira_add_allocno_copy_to_list): Make static.  Rename to
1033         add_allocno_copy_to_list.
1034         (ira_swap_allocno_copy_ends_if_necessary): Make static.  Rename to
1035         swap_allocno_copy_ends_if_necessary.
1036         (remove_unnecessary_allocnos, remove_low_level_allocnos): Call
1037         ira_remove_allocno_prefs.
1038         (ira_flattening): Ditto.
1039         (ira_build): Call initiate_prefs, print_prefs.
1040         (ira_destroy): Call finish_prefs.
1041         * ira-color.c (struct update_cost_record): New.
1042         (struct allocno_color_data): Add new member update_cost_records.
1043         (update_cost_record_pool): New static var.
1044         (init_update_cost_records, get_update_cost_record): New functions.
1045         (free_update_cost_record_list, finish_update_cost_records): Ditto.
1046         (struct update_cost_queue_elem): Add member from.
1047         (initiate_cost_update): Call init_update_cost_records.
1048         (finish_cost_update): Call finish_update_cost_records.
1049         (queue_update_cost, get_next_update_cost): Add new param from.
1050         (Update_allocno_cost, update_costs_from_allocno): New functions.
1051         (update_costs_from_prefs): Ditto.
1052         (update_copy_costs): Rename to update_costs_from_copies.
1053         (restore_costs_from_copies): New function.
1054         (update_conflict_hard_regno_costs): Don't go back.
1055         (assign_hard_reg): Call restore_costs_from_copies.  Add printing
1056         more debug info.
1057         (pop_allocnos): Add priniting more debug info.
1058         (color_allocnos): Remove prefs for conflicting hard regs.
1059         Call update_costs_from_prefs.
1060         * ira-conflicts.c (commutative_constraint_p): Move to ira.c
1061         (get_dup_num): Rename, modify, and move to ira.c
1062         (process_regs_for_copy): Add prefs.
1063         (add_insn_allocno_copies): Put src as first arg of
1064         process_regs_for_copy.  Remove dead code.  Call ira_setup_alts.
1065         * ira-costs.c (record_reg_classes): Modify and move code into
1066         record_operands_costs.
1067         (find_costs_and_classes): Create prefs for the hard reg of small
1068         reg class.
1069         (process_bb_node_for_hard_reg_moves): Add prefs.
1071 2013-10-30  Richard Biener  <rguenther@suse.de>
1073         PR middle-end/57100
1074         * basic-block.h (pre_and_rev_post_order_compute_fn): New function.
1075         * cfganal.c (pre_and_rev_post_order_compute_fn): New function
1076         as worker for ...
1077         (pre_and_rev_post_order_compute): ... which now wraps it.
1078         * graph.c (draw_cfg_nodes_no_loops): Use
1079         pre_and_rev_post_order_compute_fn to avoid ICEing and dependence
1080         on cfun.
1082 2013-10-30  Christian Bruel  <christian.bruel@st.com>
1084         * config/sh/sh-mem.cc (sh_expand_cmpnstr): New function.
1085         (sh_expand_cmpstr): Handle known align and schedule improvements.
1086         * config/sh/sh-protos.h (sh_expand_cmpstrn): Declare.
1087         * config/sh/sh.md (cmpstrnsi): New pattern.
1089 2013-10-30  Martin Jambor  <mjambor@suse.cz>
1091         PR rtl-optimization/10474
1092         * ira.c (find_moveable_pseudos): Do not calculate dominance info
1093         nor df analysis.
1094         (interesting_dest_for_shprep): New function.
1095         (split_live_ranges_for_shrink_wrap): Likewise.
1096         (ira): Calculate dominance info and df analysis. Call
1097         split_live_ranges_for_shrink_wrap.
1099 2013-10-30  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
1101         PR target/58854
1102         * config/arm/arm.c (arm_expand_epilogue_apcs_frame): Emit blockage.
1104 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1106         * tree-core.h (tree_index): Add TI_POINTER_BOUNDS_TYPE.
1107         * tree.h (POINTER_BOUNDS_P): New.
1108         (BOUNDED_TYPE_P): New.
1109         (BOUNDED_P): New.
1110         (pointer_bounds_type_node): New.
1111         * tree.c (build_common_tree_nodes): Initialize
1112         pointer_bounds_type_node.
1113         * gimple.h (gimple_call_get_nobnd_arg_index): New.
1114         (gimple_call_num_nobnd_args): New.
1115         (gimple_call_nobnd_arg): New.
1116         (gimple_return_retbnd): New.
1117         (gimple_return_set_retbnd): New
1118         * gimple.c (gimple_build_return): Increase number of ops
1119         for return statement.
1120         (gimple_call_get_nobnd_arg_index): New.
1121         * gimple-pretty-print.c (dump_gimple_return): Print second op.
1123 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1125         * ipa.c (cgraph_build_static_cdtor_1): Support contructors
1126         with "chkp ctor" and "bnd_legacy" attributes.
1127         * gimplify.c (gimplify_init_constructor): Avoid infinite
1128         loop during gimplification of bounds initializer.
1130 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1132         * c-family/c-common.c (handle_bnd_variable_size_attribute): New.
1133         (handle_bnd_legacy): New.
1134         (c_common_attribute_table): Add bnd_variable_size and bnd_legacy.
1135         * doc/extend.texi: Document bnd_variable_size and bnd_legacy
1136         attributes.
1138 2013-10-29  Ilya Enkovich  <ilya.enkovich@intel.com>
1140         * builtin-types.def (BT_FN_VOID_CONST_PTR): New.
1141         (BT_FN_PTR_CONST_PTR): New.
1142         (BT_FN_CONST_PTR_CONST_PTR): New.
1143         (BT_FN_PTR_CONST_PTR_SIZE): New.
1144         (BT_FN_PTR_CONST_PTR_CONST_PTR): New.
1145         (BT_FN_VOID_PTRPTR_CONST_PTR): New.
1146         (BT_FN_VOID_CONST_PTR_SIZE): New.
1147         (BT_FN_PTR_CONST_PTR_CONST_PTR_SIZE): New.
1148         * chkp-builtins.def: New.
1149         * builtins.def: include chkp-builtins.def.
1150         (DEF_CHKP_BUILTIN): New.
1151         * builtins.c (expand_builtin): Support BUILT_IN_CHKP_INIT_PTR_BOUNDS,
1152         BUILT_IN_CHKP_NULL_PTR_BOUNDS, BUILT_IN_CHKP_COPY_PTR_BOUNDS,
1153         BUILT_IN_CHKP_CHECK_PTR_LBOUNDS, BUILT_IN_CHKP_CHECK_PTR_UBOUNDS,
1154         BUILT_IN_CHKP_CHECK_PTR_BOUNDS, BUILT_IN_CHKP_SET_PTR_BOUNDS,
1155         BUILT_IN_CHKP_NARROW_PTR_BOUNDS, BUILT_IN_CHKP_STORE_PTR_BOUNDS,
1156         BUILT_IN_CHKP_GET_PTR_LBOUND, BUILT_IN_CHKP_GET_PTR_UBOUND,
1157         BUILT_IN_CHKP_BNDMK, BUILT_IN_CHKP_BNDSTX, BUILT_IN_CHKP_BNDCL,
1158         BUILT_IN_CHKP_BNDCU, BUILT_IN_CHKP_BNDLDX, BUILT_IN_CHKP_BNDRET,
1159         BUILT_IN_CHKP_INTERSECT, BUILT_IN_CHKP_ARG_BND, BUILT_IN_CHKP_NARROW,
1160         BUILT_IN_CHKP_EXTRACT_LOWER, BUILT_IN_CHKP_EXTRACT_UPPER.
1161         * common.opt (fcheck-pointer-bounds): New.
1162         * toplev.c (process_options): Check Pointer Bounds Checker is
1163         supported.
1164         * doc/extend.texi: Document Pointer Bounds Checker built-in functions.
1166 2013-10-30  Ilya Enkovich  <ilya.enkovich@intel.com>
1168         * target.def (builtin_chkp_function): New.
1169         (chkp_bound_type): New.
1170         (chkp_bound_mode): New.
1171         (fn_abi_va_list_bounds_size): New.
1172         (load_bounds_for_arg): New.
1173         (store_bounds_for_arg): New.
1174         * targhooks.h (default_load_bounds_for_arg): New.
1175         (default_store_bounds_for_arg): New.
1176         (default_fn_abi_va_list_bounds_size): New.
1177         (default_chkp_bound_type): New.
1178         (default_chkp_bound_mode): New.
1179         (default_builtin_chkp_function): New.
1180         * targhooks.c (default_load_bounds_for_arg): New.
1181         (default_store_bounds_for_arg): New.
1182         (default_fn_abi_va_list_bounds_size): New.
1183         (default_chkp_bound_type): New.
1184         (default_chkp_bound_mode); New.
1185         (default_builtin_chkp_function): New.
1186         * doc/tm.texi.in (TARGET_FN_ABI_VA_LIST_BOUNDS_SIZE): New.
1187         (TARGET_LOAD_BOUNDS_FOR_ARG): New.
1188         (TARGET_STORE_BOUNDS_FOR_ARG): New.
1189         (TARGET_BUILTIN_CHKP_FUNCTION): New.
1190         (TARGET_CHKP_BOUND_TYPE): New.
1191         (TARGET_CHKP_BOUND_MODE): New.
1192         * doc/tm.texi: Regenerated.
1193         * langhooks.h (lang_hooks): Add chkp_supported field.
1194         * langhooks-def.h (LANG_HOOKS_CHKP_SUPPORTED): New.
1195         (LANG_HOOKS_INITIALIZER); Add LANG_HOOKS_CHKP_SUPPORTED.
1197 2013-10-29  Andrew Pinski <apinski@cavium.com>
1199         * tree-ssa-ifcombine.c: Include rtl.h and tm_p.h.
1200         (ifcombine_ifandif): Handle cases where maybe_fold_and_comparisons
1201         fails, combining the branches anyways.
1202         (tree_ssa_ifcombine): Inverse the order of the basic block walk,
1203         increases the number of combinings.
1204         * gimple.h (gsi_start_nondebug_after_labels_bb): New function.
1206 2013-10-29  Mike Stump  <mikestump@comcast.net>
1208         * machmode.def (PARTIAL_INT_MODE): Add precision and name.
1209         * genmodes.c (PARTIAL_INT_MODE): Add precision and name.
1210         (make_vector_mode): Increase namebuf to 16.
1211         (emit_insn_modes_h): When processing BImode, don't
1212         also match partial int modes.
1213         (emit_class_narrowest_mode): Likewise.
1215         * config/bfin/bfin-modes.def: Add precision to PDI.
1216         * config/m32c/m32c-modes.def: Add precision to PSI.
1217         * config/msp430/msp430-modes.def: Add precision to PSI.
1218         * config/rs6000/rs6000-modes.def: Add precision to PTI.
1219         * config/sh/sh-modes.def: Add precision to PSI and PDI.
1221 2013-10-29  Oleg Endo  <olegendo@gcc.gnu.org>
1223         PR target/54236
1224         * config/sh/sh.md (*addc): Rename existing variations to ...
1225         (*addc_r_r_1, *addc_2r_1, *addc_r_1): ... these.
1226         (*addc_r_lsb, *addc_r_r_lsb, *addc_r_lsb_r, *addc_2r_lsb, *addc_r_msb,
1227         *addc_r_r_msb, *addc_2r_msb): New insn_and_split patterns.
1228         * config/sh/sh.c (addsubcosts): Handle some addc special cases.
1230 2013-10-29  Teresa Johnson  <tejohnson@google.com>
1232         PR ipa/58862
1233         * tree-ssa-tail-merge.c (replace_block_by): Tolerate profile
1234         insanities when updating probabilities.
1236 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1238         * gdbhooks.py (CGraphNodePrinter.to_string): Update gdb
1239         prettyprinter for cgraph_node to reflect the conversion of the
1240         symtable types to a C++ class hierarchy: it now *is* a
1241         symtab_node_base, rather than having one (named "symbol").
1243 2013-10-29  Balaji V. Iyer  <balaji.v.iyer@intel.com>
1245         * builtins.c (is_builtin_name): Added a check for __cilkrts_detach and
1246         __cilkrts_pop_frame.  If matched, then return true for built-in
1247         function name.
1248         (expand_builtin): Added BUILT_IN_CILK_DETACH and
1249         BUILT_IN_CILK_POP_FRAME case.
1250         * langhooks-def.h (lhd_install_body_with_frame_cleanup): New prototype.
1251         (lhs_cilk_detect_spawn): Likewise.
1252         (LANG_HOOKS_DECLS): Added LANG_HOOKS_CILKPLUS.
1253         (LANG_HOOKS_CILKPLUS_DETECT_SPAWN_AND_UNWRAP): New #define.
1254         (LANG_HOOKS_CILKPLUS_FRAME_CLEANUP): Likewise.
1255         (LANG_HOOKS_CILKPLUS_GIMPLIFY_SPAWN): Likewise.
1256         (LANG_HOOKS_CILKPLUS): Likewise.
1257         * tree.h (CILK_SPAWN_FN): Likewise.
1258         * builtin.def (DEF_CILK_BUILTIN_STUB): Likewise.
1259         * Makefile.in (C_COMMON_OBJS): Added c-family/cilk.o.
1260         (OBJS): Added cilk-common.o.
1261         (BUILTINS_DEF): Added cilk-builtins.def.
1262         * langhooks.c (lhd_install_body_with_frame_cleanup): New function.
1263         (lhd_cilk_detect_spawn): Likewise.
1264         * langhooks.h (lang_hooks_for_cilkplus): New struct.
1265         (struct lang_hooks): Added new field called "cilkplus."
1266         * cilk-common.c: New file.
1267         * cilk.h: Likewise.
1268         * cilk-builtins.def: Likewise.
1269         * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Added
1270         "__cilk" macro and set it to 200.
1271         * function.h (struct function::cilk_frame_decl): New field.
1272         (struct function::is_cilk_function): Likewise.
1273         (struct function::calls_cilk_spawn): Likewise.
1274         * gimplify.c (gimplify_call_expr): Added a check if the function call
1275         being gimplified is a spawn detach point.  If so, then add pop_frame
1276         and detach function calls.
1277         (gimplify_expr): Added a CILK_SPAWN_STMT and CILK_SYNC_STMT case
1278         for gimplifying _Cilk_spawn and _Cilk_sync statements.
1279         (gimplify_return_expr): Added a check for _Cilk_spawn usage in
1280         function.  If so, added a _Cilk_sync and gimplified it.
1281         (gimplify_modify_expr): Added a check for _Cilk_spawn in MODIFY and
1282         INIT_EXPRs.  If so, then call gimplify_cilk_spawn.
1283         * ipa-inline-analysis (initialize_inline_failed): Prevent inlining of
1284         spawner function.
1285         (can_inline_edge_p): Prevent inling of spawnee function.
1286         * ira.c (ira_setup_eliminable_regset): Force usage of frame pointer
1287         for functions that use Cilk keywords.
1288         * tree-inline.h (struct copy_body_data::remap_var_for_cilk): New field.
1289         * tree-pretty-print.c (dump_generic_node): Added CILK_SPAWN_STMT and
1290         CILK_SYNC_STMT cases.
1291         * tree.def (DEFTREECODE): Added CILK_SPAWN_STMT and CILK_SYNC_STMT
1292         trees.
1293         * generic.texi (CILK_SPAWN_STMT): Added documentation for _Cilk_spawn.
1294         (CILK_SYNC_STMT): Added documentation for _Cilk_sync.
1295         * passes.texi (Cilk Keywords): New section that describes the compiler
1296         code changes for handling Cilk Keywords.
1298 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1300         Patch autogenerated by refactor_symtab.py from
1301         https://github.com/davidmalcolm/gcc-refactoring-scripts
1302         revision 58bb219cc090b2f4516a9297d868c245495ee622
1304         * asan.c (asan_finish_file): Update for conversion of symtab types to
1305         a true class hierarchy.
1306         * cfgexpand.c (estimated_stack_frame_size): Likewise.
1307         * cgraph.c (cgraph_get_body): Likewise.
1308         (cgraph_get_create_real_symbol_node): Likewise.
1309         (verify_cgraph_node): Likewise.
1310         (verify_edge_corresponds_to_fndecl): Likewise.
1311         (verify_edge_count_and_frequency): Likewise.
1312         (cgraph_will_be_removed_from_program_if_no_direct_calls): Likewise.
1313         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
1314         (cgraph_can_remove_if_no_direct_calls_and_refs_p): Likewise.
1315         (cgraph_node_cannot_return): Likewise.
1316         (cgraph_set_pure_flag_1): Likewise.
1317         (cgraph_set_const_flag_1): Likewise.
1318         (cgraph_set_nothrow_flag_1): Likewise.
1319         (cgraph_make_node_local_1): Likewise.
1320         (cgraph_for_node_and_aliases): Likewise.
1321         (cgraph_for_node_thunks_and_aliases): Likewise.
1322         (cgraph_node_can_be_local_p): Likewise.
1323         (cgraph_node_cannot_be_local_p_1): Likewise.
1324         (cgraph_function_body_availability): Likewise.
1325         (dump_cgraph_node): Likewise.
1326         (cgraph_rtl_info): Likewise.
1327         (cgraph_mark_address_taken_node): Likewise.
1328         (cgraph_remove_node): Likewise.
1329         (cgraph_release_function_body): Likewise.
1330         (cgraph_update_edges_for_call_stmt_node): Likewise.
1331         (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
1332         (cgraph_make_edge_direct): Likewise.
1333         (cgraph_resolve_speculation): Likewise.
1334         (cgraph_speculative_call_info): Likewise.
1335         (cgraph_turn_edge_to_speculative): Likewise.
1336         (cgraph_create_edge_1): Likewise.
1337         (cgraph_set_call_stmt): Likewise.
1338         (cgraph_node_for_asm): Likewise.
1339         (cgraph_add_thunk): Likewise.
1340         (cgraph_same_body_alias): Likewise.
1341         (cgraph_create_function_alias): Likewise.
1342         (cgraph_create_node): Likewise.
1343         (cgraph_create_empty_node): Likewise.
1344         (record_function_versions): Likewise.
1345         (used_from_object_file_p): Likewise.
1346         * cgraph.h (symtab_can_be_discarded): Likewise.
1347         (symtab_real_symbol_p): Likewise.
1348         (cgraph_mark_force_output_node): Likewise.
1349         (cgraph_edge_recursive_p): Likewise.
1350         (symtab_alias_target): Likewise.
1351         (varpool_all_refs_explicit_p): Likewise.
1352         (varpool_can_remove_if_no_refs): Likewise.
1353         (cgraph_only_called_directly_or_aliased_p): Likewise.
1354         (cgraph_next_function_with_gimple_body): Likewise.
1355         (cgraph_first_function_with_gimple_body): Likewise.
1356         (cgraph_function_with_gimple_body_p): Likewise.
1357         (cgraph_next_function): Likewise.
1358         (cgraph_first_function): Likewise.
1359         (cgraph_next_defined_function): Likewise.
1360         (cgraph_first_defined_function): Likewise.
1361         (varpool_next_defined_variable): Likewise.
1362         (varpool_first_defined_variable): Likewise.
1363         (varpool_next_static_initializer): Likewise.
1364         (varpool_first_static_initializer): Likewise.
1365         (varpool_next_variable): Likewise.
1366         (varpool_first_variable): Likewise.
1367         (varpool_node_name): Likewise.
1368         (varpool): Likewise.
1369         (cgraph): Likewise.
1370         (is_a_helper <varpool_node>::test): Likewise.
1371         (is_a_helper <cgraph_node>::test): Likewise.
1372         (varpool_variable_node): Likewise.
1373         (cgraph_function_or_thunk_node): Likewise.
1374         (varpool_alias_target): Likewise.
1375         (cgraph_alias_target): Likewise.
1376         (cgraph_node_name): Likewise.
1377         (varpool_node_asm_name): Likewise.
1378         (cgraph_node_asm_name): Likewise.
1379         * cgraphbuild.c (remove_cgraph_callee_edges): Likewise.
1380         (cgraph_rebuild_references): Likewise.
1381         (rebuild_cgraph_edges): Likewise.
1382         (record_eh_tables): Likewise.
1383         (build_cgraph_edges): Likewise.
1384         (mark_store): Likewise.
1385         (mark_load): Likewise.
1386         (mark_address): Likewise.
1387         (record_type_list): Likewise.
1388         (record_reference): Likewise.
1389         * cgraphclones.c (cgraph_materialize_all_clones): Likewise.
1390         (cgraph_materialize_clone): Likewise.
1391         (cgraph_function_versioning): Likewise.
1392         (cgraph_copy_node_for_versioning): Likewise.
1393         (update_call_expr): Likewise.
1394         (cgraph_find_replacement_node): Likewise.
1395         (cgraph_create_virtual_clone): Likewise.
1396         (cgraph_clone_node): Likewise.
1397         * cgraphunit.c (compile): Likewise.
1398         (output_weakrefs): Likewise.
1399         (output_in_order): Likewise.
1400         (expand_function): Likewise.
1401         (assemble_thunks_and_aliases): Likewise.
1402         (expand_thunk): Likewise.
1403         (mark_functions_to_output): Likewise.
1404         (handle_alias_pairs): Likewise.
1405         (analyze_functions): Likewise.
1406         (walk_polymorphic_call_targets): Likewise.
1407         (varpool_finalize_decl): Likewise.
1408         (process_function_and_variable_attributes): Likewise.
1409         (cgraph_process_same_body_aliases): Likewise.
1410         (analyze_function): Likewise.
1411         (cgraph_add_new_function): Likewise.
1412         (cgraph_finalize_function): Likewise.
1413         (referred_to_p): Likewise.
1414         (cgraph_reset_node): Likewise.
1415         (cgraph_process_new_functions): Likewise.
1416         (enqueue_node): Likewise.
1417         (decide_is_symbol_needed): Likewise.
1418         * coverage.c (coverage_compute_profile_id): Likewise.
1419         * dbxout.c (dbxout_expand_expr): Likewise.
1420         * dwarf2out.c (premark_types_used_by_global_vars_helper): Likewise.
1421         (reference_to_unused): Likewise.
1422         * gimple-fold.c (can_refer_decl_in_current_unit_p): Likewise.
1423         * gimplify.c (unvisit_body): Likewise.
1424         (unshare_body): Likewise.
1425         * ipa-cp.c (ipcp_generate_summary): Likewise.
1426         (ipcp_decision_stage): Likewise.
1427         (identify_dead_nodes): Likewise.
1428         (decide_whether_version_node): Likewise.
1429         (decide_about_value): Likewise.
1430         (perhaps_add_new_callers): Likewise.
1431         (create_specialized_node): Likewise.
1432         (update_profiling_info): Likewise.
1433         (ipcp_propagate_stage): Likewise.
1434         (estimate_local_effects): Likewise.
1435         (good_cloning_opportunity_p): Likewise.
1436         (devirtualization_time_bonus): Likewise.
1437         (propagate_constants_accross_call): Likewise.
1438         (initialize_node_lattices): Likewise.
1439         (ipcp_cloning_candidate_p): Likewise.
1440         (determine_versionability): Likewise.
1441         (print_all_lattices): Likewise.
1442         (print_lattice): Likewise.
1443         (ipcp_discover_new_direct_edges): Likewise.
1444         * ipa-devirt.c (ipa_devirt): Likewise.
1445         (likely_target_p): Likewise.
1446         (update_type_inheritance_graph): Likewise.
1447         (possible_polymorphic_call_target_p): Likewise.
1448         (dump_possible_polymorphic_call_targets): Likewise.
1449         (devirt_variable_node_removal_hook): Likewise.
1450         (record_binfo): Likewise.
1451         (maybe_record_node): Likewise.
1452         (build_type_inheritance_graph): Likewise.
1453         * ipa-inline-analysis.c (inline_write_summary): Likewise.
1454         (inline_generate_summary): Likewise.
1455         (inline_analyze_function): Likewise.
1456         (do_estimate_growth): Likewise.
1457         (simple_edge_hints): Likewise.
1458         (estimate_node_size_and_time): Likewise.
1459         (estimate_edge_devirt_benefit): Likewise.
1460         (compute_inline_parameters): Likewise.
1461         (estimate_function_body_sizes): Likewise.
1462         (compute_bb_predicates): Likewise.
1463         (initialize_inline_failed): Likewise.
1464         (dump_inline_summary): Likewise.
1465         (dump_inline_edge_summary): Likewise.
1466         * ipa-inline-transform.c (inline_transform): Likewise.
1467         (preserve_function_body_p): Likewise.
1468         (save_inline_function_body): Likewise.
1469         (inline_call): Likewise.
1470         (clone_inlined_nodes): Likewise.
1471         (can_remove_node_now_p): Likewise.
1472         (can_remove_node_now_p_1): Likewise.
1473         * ipa-inline.c (early_inliner): Likewise.
1474         (early_inline_small_functions): Likewise.
1475         (inline_always_inline_functions): Likewise.
1476         (ipa_inline): Likewise.
1477         (flatten_function): Likewise.
1478         (inline_small_functions): Likewise.
1479         (speculation_useful_p): Likewise.
1480         (recursive_inlining): Likewise.
1481         (update_caller_keys): Likewise.
1482         (reset_edge_caches): Likewise.
1483         (update_edge_key): Likewise.
1484         (edge_badness): Likewise.
1485         (relative_time_benefit): Likewise.
1486         (want_inline_self_recursive_call_p): Likewise.
1487         (want_inline_small_function_p): Likewise.
1488         (want_early_inline_function_p): Likewise.
1489         (num_calls): Likewise.
1490         (can_early_inline_edge_p): Likewise.
1491         (can_inline_edge_p): Likewise.
1492         (report_inline_failed_reason): Likewise.
1493         * ipa-profile.c (ipa_profile): Likewise.
1494         (ipa_propagate_frequency): Likewise.
1495         (ipa_propagate_frequency_1): Likewise.
1496         (ipa_profile_generate_summary): Likewise.
1497         * ipa-prop.c (ipcp_transform_function): Likewise.
1498         (read_replacements_section): Likewise.
1499         (ipa_prop_read_section): Likewise.
1500         (ipa_modify_call_arguments): Likewise.
1501         (ipa_print_node_params): Likewise.
1502         (propagate_controlled_uses): Likewise.
1503         (update_indirect_edges_after_inlining): Likewise.
1504         (remove_described_reference): Likewise.
1505         (ipa_make_edge_direct_to_target): Likewise.
1506         (ipa_analyze_node): Likewise.
1507         (ipa_analyze_params_uses): Likewise.
1508         (ipa_compute_jump_functions): Likewise.
1509         (ipa_get_callee_param_type): Likewise.
1510         (ipa_print_node_jump_functions): Likewise.
1511         (ipa_initialize_node_params): Likewise.
1512         (ipa_populate_param_decls): Likewise.
1513         (ipa_func_spec_opts_forbid_analysis_p): Likewise.
1514         (write_agg_replacement_chain): Likewise.
1515         (ipa_write_node_info): Likewise.
1516         (ipa_edge_duplication_hook): Likewise.
1517         (try_decrement_rdesc_refcount): Likewise.
1518         * ipa-pure-const.c (propagate_nothrow): Likewise.
1519         (propagate_pure_const): Likewise.
1520         (pure_const_read_summary): Likewise.
1521         (pure_const_write_summary): Likewise.
1522         (analyze_function): Likewise.
1523         * ipa-ref-inline.h (ipa_ref_referred_ref_list): Likewise.
1524         (ipa_ref_referring_ref_list): Likewise.
1525         * ipa-ref.c (ipa_clear_stmts_in_references): Likewise.
1526         (ipa_remove_stmt_references): Likewise.
1527         (ipa_find_reference): Likewise.
1528         (ipa_dump_referring): Likewise.
1529         (ipa_dump_references): Likewise.
1530         (ipa_record_reference): Likewise.
1531         * ipa-reference.c (ipa_reference_read_optimization_summary): Likewise.
1532         (ipa_reference_write_optimization_summary): Likewise.
1533         (write_node_summary_p): Likewise.
1534         (propagate): Likewise.
1535         (read_write_all_from_decl): Likewise.
1536         (generate_summary): Likewise.
1537         (analyze_function): Likewise.
1538         (propagate_bits): Likewise.
1539         (ipa_reference_get_not_written_global): Likewise.
1540         (ipa_reference_get_not_read_global): Likewise.
1541         * ipa-split.c (execute_split_functions): Likewise.
1542         (split_function): Likewise.
1543         * ipa-utils.c (ipa_merge_profiles): Likewise.
1544         (dump_cgraph_node_set): Likewise.
1545         (ipa_reverse_postorder): Likewise.
1546         (ipa_edge_within_scc): Likewise.
1547         (ipa_get_nodes_in_cycle): Likewise.
1548         (ipa_free_postorder_info): Likewise.
1549         (ipa_reduced_postorder): Likewise.
1550         (searchc): Likewise.
1551         (recursive_call_p): Likewise.
1552         * ipa.c (ipa_cdtor_merge): Likewise.
1553         (record_cdtor_fn): Likewise.
1554         (function_and_variable_visibility): Likewise.
1555         (varpool_externally_visible_p): Likewise.
1556         (cgraph_externally_visible_p): Likewise.
1557         (comdat_can_be_unshared_p): Likewise.
1558         (comdat_can_be_unshared_p_1): Likewise.
1559         (address_taken_from_non_vtable_p): Likewise.
1560         (ipa_discover_readonly_nonaddressable_vars): Likewise.
1561         (symtab_remove_unreachable_nodes): Likewise.
1562         (walk_polymorphic_call_targets): Likewise.
1563         (process_references): Likewise.
1564         (enqueue_node): Likewise.
1565         (has_addr_references_p): Likewise.
1566         (cgraph_non_local_node_p_1): Likewise.
1567         * is-a.h (varpool_analyze_node): Likewise.
1568         * lto-cgraph.c (input_symtab): Likewise.
1569         (merge_profile_summaries): Likewise.
1570         (input_cgraph_1): Likewise.
1571         (input_edge): Likewise.
1572         (input_varpool_node): Likewise.
1573         (input_node): Likewise.
1574         (input_overwrite_node): Likewise.
1575         (compute_ltrans_boundary): Likewise.
1576         (output_refs): Likewise.
1577         (lto_output_varpool_node): Likewise.
1578         (lto_output_node): Likewise.
1579         (reachable_from_other_partition_p): Likewise.
1580         (referenced_from_other_partition_p): Likewise.
1581         (lto_output_edge): Likewise.
1582         (output_node_opt_summary): Likewise.
1583         (add_node_to): Likewise.
1584         (reachable_from_this_partition_p): Likewise.
1585         (lto_set_symtab_encoder_in_partition): Likewise.
1586         (lto_symtab_encoder_in_partition_p): Likewise.
1587         (lto_set_symtab_encoder_encode_initializer): Likewise.
1588         (lto_symtab_encoder_encode_initializer_p): Likewise.
1589         (lto_set_symtab_encoder_encode_body): Likewise.
1590         (lto_symtab_encoder_encode_body_p): Likewise.
1591         * lto-section-in.c (lto_free_function_in_decl_state_for_node):
1592         Likewise.
1593         * lto-streamer-in.c (lto_read_body): Likewise.
1594         (fixup_call_stmt_edges): Likewise.
1595         (fixup_call_stmt_edges_1): Likewise.
1596         * lto-streamer-out.c (produce_symtab): Likewise.
1597         (output_symbol_p): Likewise.
1598         (write_symbol): Likewise.
1599         (lto_output): Likewise.
1600         (copy_function): Likewise.
1601         (output_function): Likewise.
1602         * passes.c (function_called_by_processed_nodes_p): Likewise.
1603         (ipa_write_optimization_summaries): Likewise.
1604         (ipa_write_summaries): Likewise.
1605         (do_per_function_toporder): Likewise.
1606         (do_per_function): Likewise.
1607         (dump_passes): Likewise.
1608         * symtab.c (symtab_semantically_equivalent_p): Likewise.
1609         (symtab_nonoverwritable_alias): Likewise.
1610         (symtab_nonoverwritable_alias_1): Likewise.
1611         (symtab_for_node_and_aliases): Likewise.
1612         (symtab_resolve_alias): Likewise.
1613         (fixup_same_cpp_alias_visibility): Likewise.
1614         (symtab_alias_ultimate_target): Likewise.
1615         (symtab_used_from_object_file_p): Likewise.
1616         (verify_symtab_base): Likewise.
1617         (dump_symtab_base): Likewise.
1618         (symtab_node_name): Likewise.
1619         (symtab_node_asm_name): Likewise.
1620         (symtab_dissolve_same_comdat_group_list): Likewise.
1621         (symtab_add_to_same_comdat_group): Likewise.
1622         (symtab_unregister_node): Likewise.
1623         (symtab_insert_node_to_hashtable): Likewise.
1624         (symtab_register_node): Likewise.
1625         (unlink_from_assembler_name_hash): Likewise.
1626         (insert_to_assembler_name_hash): Likewise.
1627         (eq_assembler_name): Likewise.
1628         (hash_node_by_assembler_name): Likewise.
1629         (eq_node): Likewise.
1630         (hash_node): Likewise.
1631         * toplev.c (wrapup_global_declaration_2): Likewise.
1632         * trans-mem.c (ipa_tm_execute): Likewise.
1633         (ipa_tm_transform_clone): Likewise.
1634         (ipa_tm_transform_transaction): Likewise.
1635         (ipa_tm_transform_calls_redirect): Likewise.
1636         (ipa_tm_insert_gettmclone_call): Likewise.
1637         (ipa_tm_insert_irr_call): Likewise.
1638         (ipa_tm_create_version): Likewise.
1639         (ipa_tm_create_version_alias): Likewise.
1640         (ipa_tm_mark_forced_by_abi_node): Likewise.
1641         (ipa_tm_mark_force_output_node): Likewise.
1642         (ipa_tm_diagnose_tm_safe): Likewise.
1643         (ipa_tm_mayenterirr_function): Likewise.
1644         (ipa_tm_scan_irr_function): Likewise.
1645         (ipa_tm_note_irrevocable): Likewise.
1646         (ipa_tm_scan_calls_clone): Likewise.
1647         (get_cg_data): Likewise.
1648         * tree-eh.c (tree_could_trap_p): Likewise.
1649         * tree-emutls.c (ipa_lower_emutls): Likewise.
1650         (create_emultls_var): Likewise.
1651         (lower_emutls_function_body): Likewise.
1652         (gen_emutls_addr): Likewise.
1653         (emutls_decl): Likewise.
1654         (new_emutls_decl): Likewise.
1655         * tree-inline.c (tree_function_versioning): Likewise.
1656         (optimize_inline_calls): Likewise.
1657         (expand_call_inline): Likewise.
1658         (estimate_num_insns): Likewise.
1659         (copy_bb): Likewise.
1660         (delete_unreachable_blocks_update_callgraph): Likewise.
1661         * tree-nested.c (gimplify_all_functions): Likewise.
1662         (create_nesting_tree): Likewise.
1663         (check_for_nested_with_variably_modified): Likewise.
1664         * tree-pretty-print.c (dump_function_header): Likewise.
1665         * tree-profile.c (tree_profiling): Likewise.
1666         * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
1667         (modify_function): Likewise.
1668         (convert_callers): Likewise.
1669         (convert_callers_for_node): Likewise.
1670         * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
1671         (associate_varinfo_to_alias): Likewise.
1672         (create_variable_info_for): Likewise.
1673         (get_constraint_for_ssa_var): Likewise.
1674         * tree-vectorizer.c (increase_alignment): Likewise.
1675         * tree.c (find_decls_types_in_var): Likewise.
1676         (find_decls_types_in_node): Likewise.
1677         (free_lang_data_in_decl): Likewise.
1678         * value-prof.c (gimple_ic_transform): Likewise.
1679         (gimple_ic): Likewise.
1680         (check_ic_target): Likewise.
1681         (init_node_map): Likewise.
1682         * varasm.c (decl_binds_to_current_def_p): Likewise.
1683         (default_binds_local_p_1): Likewise.
1684         (dump_tm_clone_pairs): Likewise.
1685         (assemble_alias): Likewise.
1686         (find_decl): Likewise.
1687         (mark_decl_referenced): Likewise.
1688         * varpool.c (varpool_for_node_and_aliases): Likewise.
1689         (varpool_extra_name_alias): Likewise.
1690         (varpool_create_variable_alias): Likewise.
1691         (add_new_static_var): Likewise.
1692         (varpool_finalize_named_section_flags): Likewise.
1693         (varpool_remove_unreferenced_decls): Likewise.
1694         (enqueue_node): Likewise.
1695         (varpool_assemble_decl): Likewise.
1696         (assemble_aliases): Likewise.
1697         (varpool_analyze_node): Likewise.
1698         (cgraph_variable_initializer_availability): Likewise.
1699         (varpool_add_new_variable): Likewise.
1700         (ctor_for_folding): Likewise.
1701         (dump_varpool_node): Likewise.
1702         (varpool_remove_initializer): Likewise.
1703         (varpool_remove_node): Likewise.
1704         (varpool_node_for_decl): Likewise.
1705         (varpool_create_empty_node): Likewise.
1706         * config/i386/i386.c (ix86_generate_version_dispatcher_body): Likewise.
1707         (ix86_get_function_versions_dispatcher): Likewise.
1709 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1711         * cgraph.h (symtab_node_base): Convert to a class;
1712         add GTY((desc ("%h.type"), tag ("SYMTAB_SYMBOL"))), and take
1713         chain_next/prev from symtab_node_def.
1714         (cgraph_node): Inherit from symtab_node; add GTY option
1715         tag ("SYMTAB_FUNCTION").
1716         (varpool_node): Inherit from symtab_node; add GTY option
1717         tag ("SYMTAB_VARIABLE").
1718         (symtab_node_def): Remove.
1719         (is_a_helper <cgraph_node>::test (symtab_node_def *)): Convert to...
1720         (is_a_helper <cgraph_node>::test (symtab_node_base *)): ...this.
1721         (is_a_helper <varpool_node>::test (symtab_node_def *)): Convert to...
1722         (is_a_helper <varpool_node>::test (symtab_node_base *)): ...this.
1724         * ipa-ref.h (symtab_node_def): Drop.
1725         (symtab_node): Change underlying type from symtab_node_def to
1726         symtab_node_base.
1727         (const_symtab_node): Likwise.
1729         * is-a.h: Update examples in comment.
1731         * symtab.c (symtab_hash): Change symtab_node_def to symtab_node_base.
1732         (assembler_name_hash): Likewise.
1734 2013-10-29  Martin Liska  <marxin.liska@gmail.com>
1736         * doc/tree-ssa.texi (gimple_phi_result): Document.
1737         (gimple_phi_num_args, gimple_phi_arg): Likewise.
1738         (gimple_phi_arg_edge, gimple_phi_arg_def): Likewise.
1739         (PHI_RESULT, PHI_NUM_ARGS): Remove.
1740         (PHI_ARG_ELT, PHI_ARG_EDGE, PHI_ARG_DEF): Likewise.
1742 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1744         * expr.h: Revert change and include tree-core.h.
1745         * rtl.h: Revert change and don't include tree-core.h.
1747 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1749         * config/darwin.c: Include gimple.h.
1750         * config/i386/winnt.c: Likewise.
1752 2013-10-29  Marc Glisse  <marc.glisse@inria.fr>
1754         PR tree-optimization/19831
1755         * tree-ssa-alias.c (stmt_kills_ref_p_1): Handle BUILT_IN_FREE.
1757 2013-10-29  Andrew MacLeod  <amacleod@redhat.com>
1759         * tree-outof-ssa.h: Remove include files.
1760         * tree-outof-ssa.c: Add required include files from tree-outof-ssa.h.
1761         * expr.c: Likewise.
1762         * tree-ssa-coalesce.c: Likewise.
1763         * cfgexpand.c: Likewise.
1764         * tree-ssa-ter.c: Likewise.
1765         * ipa-prop.h: Remove gimple.h and tree-core.h from include list.
1766         * lto-streamer.h: Likewise.
1767         * cgraphbuild.c: Add gimple.h to include list.
1768         * data-streamer-in.c: Likewise.
1769         * ipa-cp.c: Likewise.
1770         * tree-streamer.c: Likewise.
1771         * lto-compress.c: Likewise.
1772         * ipa-reference.c: Likewise.
1773         * data-streamer-out.c: Likewise.
1774         * lto-cgraph.c: Likewise.
1775         * cgraphclones.c: Likewise.
1776         * ipa-utils.c: Likewise.
1777         * data-streamer.c: Likewise.
1778         * ipa-split.c: Likewise.
1779         * lto-section-in.c: Likewise.
1780         * tree-streamer-out.c: Likewise.
1781         * ipa-prop.c: Likewise.
1782         * tree-streamer-in.c: Likewise.
1783         * symtab.c: Likewise.
1784         * opts-global.c: Likewise.
1785         * lto-opts.c: Likewise.
1786         * lto-section-out.c: Likewise.
1787         * lto-streamer.c: Likewise.
1788         * rtl.h: Add tree-core.h to include list.
1789         * expr.h: Remove tree-core.h from include list.
1790         * gimple.h: Likewise.
1791         * ipa-utils.h: Likewise.
1792         * streamer-hooks.h: Likewise.
1793         * streamer-hooks.c: Include input.h.
1795 2013-10-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
1797         * config/arm/arm.c (cortexa7_extra_costs): New table.
1798         (arm_cortex_a7_tune): New.
1799         * config/arm/arm-cores.def: Use cortex_a7 tuning for cortex-a7.
1801 2013-10-29  Eric Botcazou  <ebotcazou@adacore.com>
1803         * expr.c (expand_expr_real_1) <MEM_EXPR>: Eliminate small redundancy.
1805 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1807         * doc/gty.texi ("Inheritance and GTY"): Make it clear that
1808         to use autogenerated markers for a class-hierarchy, every class
1809         must have a GTY marker.
1810         * gengtype.h (struct type): Add linked list of subclasses to
1811         the "s" member of the union.
1812         (add_subclass): New decl.
1813         * gengtype-state.c (read_state_struct_type): Set up subclass
1814         linked list.
1815         * gengtype.c (get_ultimate_base_class): New.
1816         (add_subclass): New.
1817         (new_structure): Set up subclass linked list.
1818         (set_gc_used_type): Propagate usage information to subclasses.
1819         (output_mangled_typename): Use get_ultimate_base_class.
1820         (walk_subclasses): Use the subclass linked list, avoiding an
1821         O(N^2) when writing out all types.
1822         (walk_type): Issue an error if the base class is missing a tag,
1823         rather than generating bogus C code.  Add a gcc_unreachable
1824         default case, in case people omit tags from concrete subclasses,
1825         or get the values wrong.
1826         (write_func_for_structure): Issue an error for subclasses for
1827         which the base doesn't have a "desc", since otherwise the
1828         autogenerated routines for the base would silently fail to visit
1829         any subclass fields.
1830         (write_root): Use get_ultimate_base_class, tweaking constness of
1831         tp to match that function's signature.
1833 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1835         * doc/gty.texi (GTY Options): Add note about inheritance to
1836         description of desc and tag.
1837         (Inheritance and GTY): New.
1839 2013-10-29  David Malcolm  <dmalcolm@redhat.com>
1841         * gengtype-parse.c (opts_have): Drop "static" so that
1842         we can use this from gengtype.c.
1843         * gengtype.c (set_gc_used_type): Mark any base class as used;
1844         update field traversal to visit inherited fields.
1845         (output_mangled_typename):  Convert references to classes within
1846         an inheritance hierarchy to reference the ultimate base class,
1847         since only it will have gt_ functions.
1848         (get_string_option): New.
1849         (walk_subclasses): New.
1850         (walk_type): Treat GTY structs that have a "desc" as being the
1851         root of an inheritance hierarchy.  Generate a switch on it
1852         within the marking function which walks all subclasses, adding
1853         cases for them via walk_subclasses.  For subclasses, visit all
1854         fields of the type (including inherited ones).
1855         (write_func_for_structure): Don't write fns for subclasses, only
1856         for the ultimate base class within an inheritance hierarchy.
1857         Subclasses-marking will be handled by the base class marking functions.
1858         (write_types): Likewise.
1859         (write_local_func_for_structure): Likewise.
1860         (USED_BY_TYPED_GC_P): Emit allocators for subclasses that have
1861         a "tag" option (and are thus concrete subclasses).
1862         (write_root): Use the marker function for the ultimate base class.
1863         * gengtype.h (FOR_ALL_INHERITED_FIELDS): New.
1864         (opts_have): Add declaration.
1866 2013-10-28  Vladimir Makarov  <vmakarov@redhat.com>
1868         * lra-spills.c (lra_final_code_change): Remove useless move insns
1869         originated from moves of pseudos.
1871 2013-10-28  Jeff Law  <law@redhat.com>
1873         * lower-subreg.c (resolve_simple_move): Fix comment typo.
1875 2013-10-28  Trevor Saunders  <tsaunders@mozilla.com>
1877         * df-scan.c (df_collection_rec): Adjust.
1878         (copy_defs): New constant.
1879         (copy_uses): Likewise.
1880         (copy_eq_uses): Likewise.
1881         (copy_mw): Likewise.
1882         (copy_all): Likewise.
1883         (df_insn_rescan): Adjust.
1884         (df_notes_rescan): Likewise.
1885         (df_swap_refs): Likewise.
1886         (df_sort_and_compress_refs): Likewise.
1887         (df_sort_and_compress_mws): Likewise.
1888         (df_install_refs): Likewise.
1889         (df_install_mws): Likewise.
1890         (df_refs_add_to_chains): Add flags parameter controlling which vectors
1891         are coppied.
1892         (df_bb_refs_record): Adjust.
1893         (df_record_entry_block_defs): Likewise.
1894         (df_record_exit_block_defs): Likewise.
1895         (df_refs_verify): Likewise.
1896         (df_mws_verify): Likewise.
1897         (df_insn_refs_verify): Likewise.
1898         (df_bb_verify): Likewise.
1899         * ipa-pure-const.c (finish_state): Remove.
1900         (propagate): Adjust.
1901         * tree-data-ref.c tree-ssa-alias.c tree-ssa-loop-ivcanon.c
1902         tree-ssa-threadedge.c tree-vect-loop-manip.c tree-vect-slp.c
1903         var-tracking.c: Adjust.
1904         * vec.c (stack_vecs): Remove.
1905         (register_stack_vec): Likewise.
1906         (stack_vec_register_index): Likewise.
1907         (unregister_stack_vec): Likewise.
1908         * vec.h (struct va_stack): Remove.
1909         (struct vec<T, A, vl_ptr>): Specialize as
1910         struct vec<T, va_heap, vl_ptr> instead since va_heap is the only
1911         allocation strategy compatable with the vl_ptr layout.
1912         (struct vec<T, va_gc, vl_ptr>): Remove because it now gets an empty
1913         specialization anyway.
1914         (class stack_vec): New class.
1915         (vec_stack_alloc): Remove.
1916         (vec<T, va_heap, vl_ptr>::using_auto_storage): New method.
1918 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1919             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1920             Sergey Lega  <sergey.s.lega@intel.com>
1921             Anna Tikhonova  <anna.tikhonova@intel.com>
1922             Ilya Tocar  <ilya.tocar@intel.com>
1923             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1924             Ilya Verbin  <ilya.verbin@intel.com>
1925             Kirill Yukhin  <kirill.yukhin@intel.com>
1926             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1928         * config/i386/i386.md (prefetch): Allow TARGET_AVX512PF.
1929         (*prefetch_avx512pf_<mode>): New.
1930         * config/i386/sse.md (avx512f_vmcmp<mode>3): Ditto.
1931         (avx512f_maskcmp<mode>3): Ditto.
1932         (vashrv16si3): Ditto.
1934 2013-10-28  Alexander Ivchenko  <alexander.ivchenko@intel.com>
1935             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
1936             Sergey Lega  <sergey.s.lega@intel.com>
1937             Anna Tikhonova  <anna.tikhonova@intel.com>
1938             Ilya Tocar  <ilya.tocar@intel.com>
1939             Andrey Turetskiy  <andrey.turetskiy@intel.com>
1940             Ilya Verbin  <ilya.verbin@intel.com>
1941             Kirill Yukhin  <kirill.yukhin@intel.com>
1942             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
1944         * config/i386/i386.md (any_truncate): New.
1945         (trunsuffix): Ditto.
1946         * config/i386/predicates.md (const_8_to_9_operand): New.
1947         (const_10_to_11_operand): Ditto.
1948         (const_12_to_13_operand): Ditto.
1949         (const_14_to_15_operand): Ditto.
1950         (const_16_to_19_operand): Ditto.
1951         (const_20_to_23_operand): Ditto.
1952         (const_24_to_27_operand): Ditto.
1953         (const_28_to_31_operand): Ditto.
1954         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_FIX_NOTRUNC.
1955         (cvtusi2<ssescalarmodesuffix>32): New.
1956         (cvtusi2<ssescalarmodesuffix>64): Ditto.
1957         (ufloatv16siv16sf2): Ditto.
1958         (avx512f_fix_notruncv16sfv16si): Ditto.
1959         (avx512f_ufix_notruncv16sfv16si): Ditto.
1960         (avx512f_vcvtss2usi): Ditto.
1961         (avx512f_vcvtss2usiq): Ditto.
1962         (avx512f_vcvttss2usi): Ditto.
1963         (avx512f_vcvttss2usiq): Ditto.
1964         (avx512f_vcvtsd2usi): Ditto.
1965         (avx512f_vcvtsd2usiq): Ditto.
1966         (avx512f_vcvttsd2usi): Ditto.
1967         (avx512f_vcvttsd2usiq): Ditto.
1968         (ufloatv8siv8df): Ditto.
1969         (avx512f_cvtdq2pd512_2): Ditto.
1970         (avx512f_cvtpd2dq512): Ditto.
1971         (avx512f_ufix_notruncv8dfv8si): Ditto.
1972         (avx512f_cvtpd2ps512): Ditto.
1973         (vec_unpacks_lo_v16sf): Ditto.
1974         (vec_unpacks_hi_v16sf): Ditto.
1975         (vec_unpacks_float_hi_v16si): Ditto.
1976         (vec_unpacks_float_lo_v16si): Ditto.
1977         (avx512f_unpckhps512): Ditto.
1978         (avx512f_unpcklps512): Ditto.
1979         (avx512f_movshdup512): Ditto.
1980         (avx512f_movsldup512): Ditto.
1981         (vec_extract_lo_v32hi): Ditto.
1982         (vec_extract_hi_v32hi): Ditto.
1983         (vec_extract_lo_v64qi): Ditto.
1984         (vec_extract_hi_v64qi): Ditto.
1985         (avx512f_unpckhpd512): Ditto.
1986         (avx512f_movddup512): Ditto.
1987         (avx512f_unpcklpd512): Ditto.
1988         (*avx512f_unpcklpd512): Ditto.
1989         (avx512f_shufps512_1): Ditto.
1990         (avx512f_shufpd512_1): Ditto.
1991         (avx512f_interleave_highv8di): Ditto.
1992         (avx512f_interleave_lowv8di): Ditto.
1993         (PMOV_DST_MODE): Ditto.
1994         (pmov_src_mode): Ditto.
1995         (pmov_src_lower): Ditto.
1996         (pmov_suff): Ditto.
1997         (*avx512f_<code><pmov_src_lower><mode>2): Ditto.
1998         (*avx512f_<code>v8div16qi2): Ditto.
1999         (*avx512f_<code>v8div16qi2_store): Ditto.
2000         (vec_widen_umult_even_v16si): Ditto.
2001         (*vec_widen_umult_even_v16si): Ditto.
2002         (vec_widen_smult_even_v16si): Ditto.
2003         (*vec_widen_smult_even_v16si): Ditto.
2004         (avx512f_interleave_highv16si): Ditto.
2005         (avx512f_interleave_lowv16si): Ditto.
2006         (avx512f_<code>v16qiv16si2): Ditto.
2007         (avx512f_<code>v16hiv16si2): Ditto.
2008         (avx512f_<code>v8qiv8di2): Ditto.
2009         (avx512f_<code>v8hiv8di2): Ditto.
2010         (avx512f_<code>v8siv8di2): Ditto.
2011         (avx512cd_maskb_vec_dupv8di): Ditto.
2012         (avx512cd_maskw_vec_dupv16si): Ditto.
2013         (avx512f_vcvtph2ps512): Ditto.
2014         (avx512f_vcvtps2ph512): Ditto.
2015         (VEC_EXTRACT_MODE): Extened with wider modes.
2016         (VEC_PERM_AVX2): Ditto.
2017         (VEC_PERM_CONST): Ditto.
2019 2013-10-28  Joern Rennecke  <joern.rennecke@embecosm.com>
2021         * config/arc/arc.c (arc_ccfsm_post_advance):
2022         Add comment about TYPE_RETURN.
2024 2013-10-28  Bin Cheng  <bin.cheng@arm.com>
2026         * tree-ssa-loop-ivopts.c (strip_offset_1): Change parameter type.
2027         Count DECL_FIELD_BIT_OFFSET for COMPONENT_REF.
2028         (strip_offset): Convert offset to unsigned number.
2030 2013-10-27  Tom de Vries  <tom@codesourcery.com>
2032         * cfgexpand.c (gimple_expand_cfg): Remove test for parm_birth_insn.
2033         Don't commit insertions after NOTE_INSN_FUNCTION_BEG.
2035 2013-10-27  Oleg Endo  <olegendo@gcc.gnu.org>
2037         * config/sh/sh.c (MSW, LSW): Move and rename macros to...
2038         * config/sh/sh.h (SH_REG_MSW_OFFSET, SH_REG_LSW_OFFSET): ... here.
2039         (TARGET_BIG_ENDIAN): New macro.
2040         * config/sh/sh.md: Use it instead of !TARGET_LITTLE_ENDIAN.
2041         Use SH_REG_MSW_OFFSET and SH_REG_LSW_OFFSET.
2042         * config/sh/sh.c: Likewise.
2043         * config/sh/sh.h: Likewise.
2045 2013-10-27  Hans-Peter Nilsson  <hp@axis.com>
2047         * config/cris/cris.c (cris_emit_trap_for_misalignment): Replace the
2048         removed PRED_MUDFLAP with PRED_NORETURN.  Correct file-path in comment.
2050 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
2052         * config/sh/sh.md (movmemsi): Remove empty constraints and predicates.
2053         Fix formatting.
2054         (cmpstr_t, cmpstrsi): Fix formatting.
2056 2013-10-26  Oleg Endo  <olegendo@gcc.gnu.org>
2058         PR target/52483
2059         * config/sh/predicates.md (general_movdst_operand): Allow reg+reg
2060         addressing, do not use general_operand for memory operands.
2062 2013-10-26  Vladimir Makarov  <vmakarov@redhat.com>
2064         Revert:
2065         2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2066         * lra-spills.c (lra_final_code_change): Remove useless move insns.
2068 2013-10-26  Jeff Law  <law@redhat.com>
2070         * predict.c (PRED_MUDFLAP): Remove.
2071         * targhooks.c (build_va_arg_indirect_ref): Remove mudflap support.
2073         * Makefile.in (C_COMMON_OBJS): Remove tree-mudflap.
2074         (OBJS): Remove tree-nomudflap.o
2075         (GTFILES): Remove tree-mudflap.c
2076         * builtins.c (expand_builtin_alloc): Remove mudflap support.
2077         * gcc.c (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2078         (mfwrap_spec, mflib_spec): Likewise.
2079         (cpp_unique_options, cc1_options, static_specs): Likewise.
2080         * gimplify (gimplify_vla_decl, build_va_arg_indirect_ref): Likewise.
2081         * passes.def: Likewise.
2082         * toplev.c (compile_file, process_options): Likewise.
2083         * tree-inline.c (copy_tree_r): Likewise.
2084         * tree-pass.,h (make_pass_mudflap_1, make_pass_mudflap_2): Likewise.
2085         * varasm.c (make_decl_rtl, make_decl_rtl_for_debug): Likewise.
2086         (build_constant_desc, output_constant_def_contents): Likewise.
2087         (categorize_decl_for_section): Likewise.
2088         * tree-mudflap.c: Removed.
2089         * tree-mudflap.h: Removed.
2090         * tree-nomudflap.c: Removed.
2091         * bfin/uclinux.h (MFWRAP_SPEC): Remove.
2092         * moxie/uclinux.h (MFWRAP_SPEC): Likewise.
2093         * rs6000/aix.h (MFWRAP_SPEC, MFLIB_SPEC): Likewise.
2094         * config/sol2.h (MFLIB_SPEC): Likewise.
2095         * doc/install.texi: Remove mudflap references.
2096         * doc/passes.texi: Similarly.
2097         * doc/sourcebuild.texi: Similarly.
2098         * doc/invoke.texi: Remove mudlfap related options.
2100 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2102         PR rtl-optimization/58759
2103         * lra-constraints.c (lra_constraints): Remove wrong condition to
2104         remove insn setting up an equivalent pseudo.
2106 2013-10-25  Vladimir Makarov  <vmakarov@redhat.com>
2108         * config/rs6000/rs6000-protos.h
2109         (rs6000_secondary_memory_needed_mode): New prototype.
2110         * config/rs6000/rs6000.c: Include ira.h.
2111         (TARGET_LRA_P): Redefine.
2112         (rs6000_legitimate_offset_address_p): Call
2113         legitimate_constant_pool_address_p in strict mode for LRA.
2114         (rs6000_legitimate_address_p): Ditto.
2115         (legitimate_lo_sum_address_p): Add code for LRA.  Use lra_in_progress.
2116         (rs6000_emit_move): Add LRA version of code to generate load/store
2117         of SDmode values.
2118         (rs6000_secondary_memory_needed_mode): New.
2119         (rs6000_alloc_sdmode_stack_slot): Do nothing for LRA.
2120         (rs6000_secondary_reload_class): Return NO_REGS for LRA for
2121         constants, memory, and FP registers.
2122         (rs6000_lra_p): New.
2123         * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): New macro.
2124         * config/rs6000/rs6000.opt (mlra): New option.
2125         * lra-spills.c (lra_final_code_change): Remove useless move insns.
2127 2013-10-25  Yufeng Zhang  <yufeng.zhang@arm.com>
2129         * tree-ssa-math-opts.c (convert_plusminus_to_widen): Call
2130         has_single_use () and not do the conversion if has_single_use ()
2131         returns false for the multiplication result.
2133 2013-10-25  David Malcolm  <dmalcolm@redhat.com>
2135         * tree.h (EXCEPTIONAL_CLASS_P): Rename parameter from "CODE"
2136         to "NODE", since this works on a "tree", not an
2137         "enum tree_code".
2138         (CONSTANT_CLASS_P): Likewise.
2139         (TYPE_P): Likewise.
2140         (DECL_P): Likewise.
2141         (INDIRECT_REF_P): Likewise.
2142         (REFERENCE_CLASS_P): Likewise.
2143         (COMPARISON_CLASS_P): Likewise.
2144         (UNARY_CLASS_P): Likewise.
2145         (BINARY_CLASS_P): Likewise.
2146         (STATEMENT_CLASS_P): Likewise.
2147         (VL_EXP_CLASS_P): Likewise.
2148         (EXPRESSION_CLASS_P): Likewise.
2149         (IS_TYPE_OR_DECL_P): Likewise.
2151 2013-10-25  Marc Glisse  <marc.glisse@inria.fr>
2153         * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Look for an
2154         ADDR_EXPR in the defining statement.
2156 2013-10-25  Richard Biener  <rguenther@suse.de>
2158         PR tree-optimization/58626
2159         * tree-loop-distribution.c (enum rdg_dep_type): Remove
2160         anti_dd, output_dd and input_dd.
2161         (struct rdg_edge): Remove level and relation members.
2162         (RDGE_LEVEL, RDGE_RELATION): Remove.
2163         (dot_rdg_1): Adjust.
2164         (create_rdg_edge_for_ddr): Remove.
2165         (create_rdg_edges_for_scalar): Adjust.
2166         (create_edge_for_control_dependence): Likewise.
2167         (create_rdg_edges): Split into ...
2168         (create_rdg_flow_edges): ... this
2169         (create_rdg_cd_edges): ... and this.
2170         (free_rdg): Adjust.
2171         (build_rdg): Likewise, do not compute data dependences or
2172         add edges for them.
2173         (pg_add_dependence_edges): New function.
2174         (pgcmp): Likewise.
2175         (distribute_loop): First apply all non-dependence based
2176         partition mergings.  Then compute dependences between partitions
2177         and merge and order partitions according to them.
2179 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2181         PR rtl-optimization/58831
2182         * alias.c (init_alias_analysis): At the beginning of each iteration,
2183         set the reg_seen[N] bit if static_reg_base_value[N] is non-null.
2185 2013-10-25  Eric Botcazou  <ebotcazou@adacore.com>
2187         * recog.c (search_ofs): New static variable moved from...
2188         (peep2_find_free_register): ...here.
2189         (peephole2_optimize): Initialize it.
2191 2013-10-25  Tobias Burnus  <burnus@net-b.de>
2193         * doc/invoke.texi (fopenmp): Change supported OpenMP version to 4.0.
2195 2013-10-25  Uros Bizjak  <ubizjak@gmail.com>
2197         * config/i386/i386.h (TARGET_MPX): New define.
2198         (TARGET_MPX_P): Ditto.
2200 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2202         * config/i386/constraints.md (B): New.
2203         (Ti): New.
2204         (Tb): New.
2205         * config/i386/i386-c.c (ix86_target_macros_internal): Add __MPX__.
2206         * config/i386/i386-modes.def (BND32): New.
2207         (BND64): New.
2208         * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): New.
2209         * config/i386/i386.c (isa_opts): Add mmpx.
2210         (regclass_map): Add bound registers.
2211         (dbx_register_map): Likewise.
2212         (dbx64_register_map): Likewise.
2213         (svr4_dbx_register_map): Likewise.
2214         (PTA_MPX): New.
2215         (ix86_option_override_internal): Support MPX ISA.
2216         (ix86_conditional_register_usage): Support bound registers.
2217         (print_reg): Likewise.
2218         (ix86_code_end): Add MPX bnd prefix.
2219         (output_set_got): Likewise.
2220         (ix86_output_call_insn): Likewise.
2221         (ix86_print_operand): Add '!' (MPX bnd) print prefix support.
2222         (ix86_print_operand_punct_valid_p): Likewise.
2223         (ix86_print_operand_address): Support UNSPEC_BNDMK_ADDR and
2224         UNSPEC_BNDMK_ADDR.
2225         (ix86_class_likely_spilled_p): Add bound regs support.
2226         (ix86_hard_regno_mode_ok): Likewise.
2227         (x86_order_regs_for_local_alloc): Likewise.
2228         (ix86_bnd_prefixed_insn_p): New.
2229         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Fix to new value.
2230         (FIXED_REGISTERS): Add bound registers.
2231         (CALL_USED_REGISTERS): Likewise.
2232         (REG_ALLOC_ORDER): Likewise.
2233         (HARD_REGNO_NREGS): Likewise.
2234         (TARGET_MPX): New.
2235         (VALID_BND_REG_MODE): New.
2236         (FIRST_BND_REG): New.
2237         (LAST_BND_REG): New.
2238         (reg_class): Add BND_REGS.
2239         (REG_CLASS_NAMES): Likewise.
2240         (REG_CLASS_CONTENTS): Likewise.
2241         (BND_REGNO_P): New.
2242         (ANY_BND_REG_P): New.
2243         (BNDmode): New.
2244         (HI_REGISTER_NAMES): Add bound registers.
2245         * config/i386/i386.md (UNSPEC_BNDMK): New.
2246         (UNSPEC_BNDMK_ADDR): New.
2247         (UNSPEC_BNDSTX): New.
2248         (UNSPEC_BNDLDX): New.
2249         (UNSPEC_BNDLDX_ADDR): New.
2250         (UNSPEC_BNDCL): New.
2251         (UNSPEC_BNDCU): New.
2252         (UNSPEC_BNDCN): New.
2253         (UNSPEC_MPX_FENCE): New.
2254         (BND0_REG): New.
2255         (BND1_REG): New.
2256         (type): Add mpxmov, mpxmk, mpxchk, mpxld, mpxst.
2257         (length_immediate): Likewise.
2258         (prefix_0f): Likewise.
2259         (memory): Likewise.
2260         (prefix_rep): Check for bnd prefix.
2261         (length_nobnd): New.
2262         (length): Use length_nobnd if specified.
2263         (BND): New.
2264         (bnd_ptr): New.
2265         (BNDCHECK): New.
2266         (bndcheck): New.
2267         (*jcc_1): Add bnd prefix and rename length attr to length_nobnd.
2268         (*jcc_2): Likewise.
2269         (jump): Likewise.
2270         (simple_return_internal): Likewise.
2271         (simple_return_pop_internal): Likewise.
2272         (*indirect_jump): Add MPX bnd prefix.
2273         (*tablejump_1): Likewise.
2274         (simple_return_internal_long): Likewise.
2275         (simple_return_indirect_internal): Likewise.
2276         (<mode>_mk): New.
2277         (*<mode>_mk): New.
2278         (mov<mode>): New.
2279         (*mov<mode>_internal_mpx): New.
2280         (<mode>_<bndcheck>): New.
2281         (*<mode>_<bndcheck>): New.
2282         (<mode>_ldx): New.
2283         (*<mode>_ldx): New.
2284         (<mode>_stx): New.
2285         (*<mode>_stx): New.
2286         * config/i386/predicates.md (lea_address_operand): Rename to...
2287         (address_no_seg_operand): ... this.
2288         (address_mpx_no_base_operand): New.
2289         (address_mpx_no_index_operand): New.
2290         (bnd_mem_operator): New.
2291         * config/i386/i386.opt (mmpx): New.
2292         * doc/invoke.texi: Add documentation for the flags -mmpx, -mno-mpx.
2293         * doc/rtl.texi Add documentation for BND32mode and BND64mode.
2295 2013-10-24  Ilya Enkovich  <ilya.enkovich@intel.com>
2297         * mode-classes.def (MODE_POINTER_BOUNDS): New.
2298         * tree.def (POINTER_BOUNDS_TYPE): New.
2299         * genmodes.c (complete_mode): Support MODE_POINTER_BOUNDS.
2300         (POINTER_BOUNDS_MODE): New.
2301         (make_pointer_bounds_mode): New.
2302         * machmode.h (POINTER_BOUNDS_MODE_P): New.
2303         * stor-layout.c (int_mode_for_mode): Support MODE_POINTER_BOUNDS.
2304         (layout_type): Support POINTER_BOUNDS_TYPE.
2305         * tree-pretty-print.c (dump_generic_node): Support POINTER_BOUNDS_TYPE.
2306         * tree.c (build_int_cst_wide): Support POINTER_BOUNDS_TYPE.
2307         (type_contains_placeholder_1): Likewise.
2308         * tree.h (POINTER_BOUNDS_TYPE_P): New.
2309         * varasm.c (output_constant): Support POINTER_BOUNDS_TYPE.
2310         * doc/rtl.texi (MODE_POINTER_BOUNDS): New.
2312 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2314         * expr.c (expand_expr_real_1): Use mode of memory reference rather than
2315         mode of address computation when calling memory_address_addr_space.
2317 2013-08-24  Richard Henderson  <rth@twiddle.net>
2319         PR rtl/58542
2320         * optabs.c (maybe_emit_atomic_exchange): Use create_input_operand
2321         instead of create_convert_operand_to.
2322         (maybe_emit_sync_lock_test_and_set): Likewise.
2323         (expand_atomic_compare_and_swap): Likewise.
2324         (maybe_emit_compare_and_swap_exchange_loop): Don't convert_modes.
2326 2013-08-24  Sriraman Tallam  <tmsriram@google.com>
2328         * cgraph.c (cgraph_fnver_htab): Move GTY((...)) to be before htab_t.
2329         Change param_is to use the struct and not the pointer to the struct.
2331 2013-10-24  Andrew MacLeod  <amacleod@redhat.com>
2333         * builtins.c (dummy_object, gimplify_va_arg_expr): Move to gimplify.c.
2334         * gimplify.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2335         Move to targhooks.c.
2336         (dummy_object, gimplify_va_arg_expr): Relocate from builtins.c.
2337         * targhooks.c (build_va_arg_indirect_ref, std_gimplify_va_arg_expr):
2338         Relocate from gimplify.c.
2339         * targhooks.h: Add 2 prototypes.
2340         * tree.h: Delete 2 prototypes.
2342 2013-10-24  Igor Shevlyakov  <igor.shevlyakov@gmail.com>
2344         * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p ): Check both
2345         [reg+mult*reg] and [mult*reg] to determine if multiplier is allowed.
2347 2013-10-24  Cong Hou  <congh@google.com>
2349         * convert.c (convert_to_real): Guard those unsafe math function
2350         convertions with flag_unsafe_math_optimizations.  Handle sqrt()
2351         specially.
2353 2013-10-24  Markus Trippelsdorf  <markus@trippelsdorf.de>
2355         PR ipa/58712
2356         * cgraph.c (cgraph_create_edge_1): Add indirect_unknown_callee
2357         as argument.
2358         (cgraph_create_edge): Use the new argument.
2359         (cgraph_create_indirect_edge): Likewise.
2361 2013-10-24  Joern Rennecke  <joern.rennecke@embecosm.com>
2363         * config/arc/arc.c (arc_ccfsm_post_advance): Also handle
2364         TYPE_UNCOND_BRANCH.
2365         (arc_ifcvt) <case 1 and 2>: Check that arc_ccfsm_post_advance
2366         changes statep->state.
2368 2013-10-24  Tobias Burnus  <burnus@net-b.de>
2370         PR other/33426
2371         * tree-cfg.c (replace_loop_annotate): New function.
2372         (execute_build_cfg): Call it.
2373         * gimplify.c (gimple_boolify, gimplify_expr): Handle ANNOTATE_EXPR.
2374         * internal-fn.c (expand_ANNOTATE): New function.
2375         * internal-fn.def (ANNOTATE): Define as new internal function.
2376         * tree-core.h (tree_node_kind): Add annot_expr_ivdep_kind.
2377         * tree-pretty-print.c (dump_generic_node): Handle ANNOTATE_EXPR.
2378         * tree.def (ANNOTATE_EXPR): New DEFTREECODE.
2379         * doc/extend.texi (Pragmas): Document #pragma ivdep.
2380         * doc/generic.texi (Expressions): Document ANNOTATE_EXPR.
2382 2013-10-17  Ian Bolton  <ian.bolton@arm.com>
2383             Marcus Shawcroft  <marcus.shawcroft@arm.com>
2385         * config/aarch64/aarch64.c (aarch64_preferred_reload_class):
2386         Special case reload SP+C into none GENERAL_REGS.
2388 2013-10-24  Michael Matz  <matz@suse.de>
2390         * gengtype.c (is_file_equal): Check that files will be same length.
2392 2013-10-25  Christian Bruel  <christian.bruel@st.com>
2394         * config.gcc (sh-*): Add sh-mem.o to extra_obj.
2395         * config/sh/t-sh (sh-mem.o): New rule.
2396         * config/sh/sh-mem.cc (expand_block_move): Moved here.
2397         (sh_expand_cmpstr): New function.
2398         * config/sh/sh.c (force_into, expand_block_move): Move to sh-mem.c.
2399         * config/sh/sh-protos.h (sh_expand_cmpstr): Declare.
2400         * config/sh/sh.md (cmpstrsi, cmpstr_t): New patterns.
2401         (rotlhi3_8): Rename.
2403 2013-10-24  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2405         * configure.ac (ZW_PROG_COMPILER_DEPENDENCIES): Use CXX instead of CC.
2406         * Makefile.in (CXXDEPMODE): Assign and change users.
2407         (CCDEPMODE): Delete.
2408         * configure: Regenerate.
2410 2013-10-23  David Malcolm  <dmalcolm@redhat.com>
2412         * gengtype-parse.c (require_without_advance): New.
2413         (type): For GTY-marked types that are not GTY((user)), parse any
2414         base classes, requiring them to be single-inheritance, and not
2415         be templates.  For non-GTY-marked types and GTY((user)),
2416         continue to skip over any C++ inheritance specification.
2417         * gengtype-state.c (state_writer::write_state_struct_type):
2418         Write base class of type (if any).
2419         (read_state_struct_type): Read base class of type (if any).
2420         * gengtype.c (new_structure): Add a "base_class" parameter.
2421         (create_optional_field_): Update for new parameter to new_structure.
2422         (adjust_field_rtx_def): Likewise.
2423         (adjust_field_tree_exp): Likewise.
2424         * gengtype.h (struct type): Add "base_class" field to the s
2425         union field.
2426         (new_structure): Add "base" parameter.
2428 2013-10-23  Sriraman Tallam  <tmsriram@google.com>
2430         PR target/57756
2431         * config/i386/i386.c (ix86_option_override_internal):
2432         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...)
2433         (ix86_valid_target_attribute_tree):
2434         Change TARGET_64BIT to TARGET_64BIT_P (opts->...)
2435         Change TARGET_SSE to TARGET_SSE_P (opts->...)
2437 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2439         * tree-ssa-loop.h: Remove include files.
2440         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2441         * cfgloopmanip.c: Move required includes from tree-ssa-loop.h.
2442         * graphite-clast-to-gimple.c: Likewise.
2443         * graphite-scop-detection.c: Likewise.
2444         * graphite-sese-to-poly.c: Likewise.
2445         * ipa-inline-analysis.c: Likewise.
2446         * ipa-pure-const.c: Likewise.
2447         * loop-init.c: Likewise.
2448         * passes.c: Likewise.
2449         * predict.c: Likewise.
2450         * tree-cfg.c: Likewise.
2451         * tree-cfgcleanup.c: Likewise.
2452         * tree-chrec.c: Likewise.
2453         * tree-data-ref.c: Likewise.
2454         * tree-loop-distribution.c: Likewise.
2455         * tree-parloops.c: Likewise.
2456         * tree-predcom.c: Likewise.
2457         * tree-scalar-evolution.c: Likewise.
2458         * tree-ssa-address.c: Likewise.
2459         * tree-ssa.c: Likewise.
2460         * tree-ssa-dce.c: Likewise.
2461         * tree-ssa-loop.c: Likewise.
2462         * tree-ssa-loop-im.c: Likewise.
2463         * tree-ssa-loop-ivcanon.c: Likewise.
2464         * tree-ssa-loop-ivopts.c: Likewise.
2465         * tree-ssa-loop-manip.c: Likewise.
2466         * tree-ssa-loop-niter.c: Likewise.
2467         * tree-ssa-loop-prefetch.c: Likewise.
2468         * tree-ssa-loop-unswitch.c: Likewise.
2469         * tree-ssa-reassoc.c: Likewise.
2470         * tree-vect-data-refs.c: Likewise.
2471         * tree-vect-loop.c: Likewise.
2472         * tree-vect-loop-manip.c: Likewise.
2473         * tree-vectorizer.c: Likewise.
2474         * tree-vect-stmts.c: Likewise.
2475         * tree-vrp.c: Likewise.
2477 2013-10-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2479         * config/rs6000/altivec.md (mulv8hi3): Adjust for little endian.
2481 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2483         PR tree-optimization/58775
2484         PR tree-optimization/58791
2485         * tree-ssa-reassoc.c (reassoc_stmt_dominates_stmt_p): New function.
2486         (insert_stmt_after): Rewritten, don't move the stmt, but really
2487         insert it.
2488         (get_stmt_uid_with_default): Remove.
2489         (build_and_add_sum): Use insert_stmt_after and
2490         reassoc_stmt_dominates_stmt_p.  Fix up uid if bb contains only labels.
2491         (update_range_test): Set uid on stmts added by
2492         force_gimple_operand_gsi.  Don't immediately modify statements
2493         in inter-bb optimization, just update oe->op values.
2494         (optimize_range_tests): Return bool whether any changed have been made.
2495         (update_ops): New function.
2496         (struct inter_bb_range_test_entry): New type.
2497         (maybe_optimize_range_tests): Perform statement changes here.
2498         (not_dominated_by, appears_later_in_bb, get_def_stmt,
2499         ensure_ops_are_available): Remove.
2500         (find_insert_point): Rewritten.
2501         (rewrite_expr_tree): Remove MOVED argument, add CHANGED argument,
2502         return LHS of the (new resp. old) stmt.  Don't call
2503         ensure_ops_are_available, don't reuse SSA_NAMEs, recurse first
2504         instead of last, move new stmt at the right place.
2505         (linearize_expr, repropagate_negates): Don't reuse SSA_NAMEs.
2506         (negate_value): Likewise.  Set uids.
2507         (break_up_subtract_bb): Initialize uids.
2508         (reassociate_bb): Adjust rewrite_expr_tree caller.
2509         (do_reassoc): Don't call renumber_gimple_stmt_uids.
2511 2013-10-23  David Edelsohn  <dje.gcc@gmail.com>
2513         PR target/58838
2514         * config/rs6000/rs6000.md (mulsi3_internal1 and splitter): Add
2515         TARGET_32BIT final condition.
2516         (mulsi3_internal2 and splitter): Same.
2518 2013-10-23  Jeff Law  <law@redhat.com>
2520         * tree-ssa-threadedge.c (thread_across_edge): Do not allow threading
2521         through joiner blocks with abnormal outgoing edges.
2523         * tree-ssa-threadupdate.c (thread_block_1): Renamed from thread_block.
2524         Add parameter JOINERS, to allow/disallow threading through joiner
2525         blocks.
2526         (thread_block): New.  Call thread_block_1.
2527         (mark_threaded_blocks): Remove code to filter out certain cases
2528         of threading through joiner blocks.
2529         (thread_through_all_blocks): Document how we can have a dangling
2530         edge AUX field and clear it.
2532 2013-10-23  Ian Lance Taylor  <iant@google.com>
2534         * doc/invoke.texi (Option Summary): Remove -fno-default-inline.
2535         (C++ Dialect Options): Likewise.
2536         (Optimize Options): Likewise.
2538 2013-10-23  Tom de Vries  <tom@codesourcery.com>
2540         PR tree-optimization/58805
2541         * tree-ssa-tail-merge.c (stmt_local_def): Add gimple_vdef check.
2543 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2545         * tree-vect-patterns.c (vect_recog_divmod_pattern): Optimize
2546         sequence based on get_range_info returned range.
2548 2013-10-23  Andrew MacLeod  <amacleod@redhat.com>
2550         * tree-ssa.h: Remove all #include's
2551         * gengtype.c (open_base_files): Adjust include list for gtype-desc.c.
2552         * alias.c: Move required includes from tree-ssa.h.
2553         * asan.c: Likewise.
2554         * builtins.c: Likewise.
2555         * calls.c: Likewise.
2556         * cfgexpand.c: Likewise.
2557         * cfghooks.c: Likewise.
2558         * cfgloop.c: Likewise.
2559         * cfgloopmanip.c: Likewise.
2560         * cgraph.c: Likewise.
2561         * cgraphbuild.c: Likewise.
2562         * cgraphclones.c: Likewise.
2563         * cgraphunit.c: Likewise.
2564         * dse.c: Likewise.
2565         * except.c: Likewise.
2566         * expr.c: Likewise.
2567         * final.c: Likewise.
2568         * fold-const.c: Likewise.
2569         * ggc-page.c: Likewise.
2570         * gimple-builder.c: Likewise.
2571         * gimple-fold.c: Likewise.
2572         * gimple-iterator.c: Likewise.
2573         * gimple-low.c: Likewise.
2574         * gimple-pretty-print.c: Likewise.
2575         * gimple-ssa-strength-reduction.c: Likewise.
2576         * gimple-streamer-in.c: Likewise.
2577         * gimple-streamer-out.c: Likewise.
2578         * gimplify.c: Likewise.
2579         * graphite-blocking.c: Likewise.
2580         * graphite-clast-to-gimple.c: Likewise.
2581         * graphite-dependences.c: Likewise.
2582         * graphite-interchange.c: Likewise.
2583         * graphite-optimize-isl.c: Likewise.
2584         * graphite-poly.c: Likewise.
2585         * graphite-scop-detection.c: Likewise.
2586         * graphite-sese-to-poly.c: Likewise.
2587         * graphite.c: Likewise.
2588         * ipa-cp.c: Likewise.
2589         * ipa-inline-analysis.c: Likewise.
2590         * ipa-inline-transform.c: Likewise.
2591         * ipa-inline.c: Likewise.
2592         * ipa-prop.c: Likewise.
2593         * ipa-pure-const.c: Likewise.
2594         * ipa-reference.c: Likewise.
2595         * ipa-split.c: Likewise.
2596         * ipa-utils.c: Likewise.
2597         * loop-init.c: Likewise.
2598         * lto-cgraph.c: Likewise.
2599         * lto-section-in.c: Likewise.
2600         * lto-section-out.c: Likewise.
2601         * lto-streamer-in.c: Likewise.
2602         * lto-streamer-out.c: Likewise.
2603         * lto-streamer.c: Likewise.
2604         * omp-low.c: Likewise.
2605         * passes.c: Likewise.
2606         * predict.c: Likewise.
2607         * print-tree.c: Likewise.
2608         * profile.c: Likewise.
2609         * sese.c: Likewise.
2610         * targhooks.c: Likewise.
2611         * tracer.c: Likewise.
2612         * trans-mem.c: Likewise.
2613         * tree-call-cdce.c: Likewise.
2614         * tree-cfg.c: Likewise.
2615         * tree-cfgcleanup.c: Likewise.
2616         * tree-chrec.c: Likewise.
2617         * tree-complex.c: Likewise.
2618         * tree-data-ref.c: Likewise.
2619         * tree-dfa.c: Likewise.
2620         * tree-eh.c: Likewise.
2621         * tree-emutls.c: Likewise.
2622         * tree-if-conv.c: Likewise.
2623         * tree-inline.c: Likewise.
2624         * tree-into-ssa.c: Likewise.
2625         * tree-loop-distribution.c: Likewise.
2626         * tree-mudflap.c: Likewise.
2627         * tree-nested.c: Likewise.
2628         * tree-nrv.c: Likewise.
2629         * tree-object-size.c: Likewise.
2630         * tree-outof-ssa.c: Likewise.
2631         * tree-parloops.c: Likewise.
2632         * tree-phinodes.c: Likewise.
2633         * tree-predcom.c: Likewise.
2634         * tree-pretty-print.c: Likewise.
2635         * tree-profile.c: Likewise.
2636         * tree-scalar-evolution.c: Likewise.
2637         * tree-sra.c: Likewise.
2638         * tree-ssa-address.c: Likewise.
2639         * tree-ssa-alias.c: Likewise.
2640         * tree-ssa-ccp.c: Likewise.
2641         * tree-ssa-coalesce.c: Likewise.
2642         * tree-ssa-copy.c: Likewise.
2643         * tree-ssa-copyrename.c: Likewise.
2644         * tree-ssa-dce.c: Likewise.
2645         * tree-ssa-dom.c: Likewise.
2646         * tree-ssa-dse.c: Likewise.
2647         * tree-ssa-forwprop.c: Likewise.
2648         * tree-ssa-ifcombine.c: Likewise.
2649         * tree-ssa-live.c: Likewise.
2650         * tree-ssa-loop-ch.c: Likewise.
2651         * tree-ssa-loop-im.c: Likewise.
2652         * tree-ssa-loop-ivcanon.c: Likewise.
2653         * tree-ssa-loop-ivopts.c: Likewise.
2654         * tree-ssa-loop-manip.c: Likewise.
2655         * tree-ssa-loop-niter.c: Likewise.
2656         * tree-ssa-loop-prefetch.c: Likewise.
2657         * tree-ssa-loop-unswitch.c: Likewise.
2658         * tree-ssa-loop.c: Likewise.
2659         * tree-ssa-math-opts.c: Likewise.
2660         * tree-ssa-operands.c: Likewise.
2661         * tree-ssa-phiopt.c: Likewise.
2662         * tree-ssa-phiprop.c: Likewise.
2663         * tree-ssa-pre.c: Likewise.
2664         * tree-ssa-propagate.c: Likewise.
2665         * tree-ssa-reassoc.c: Likewise.
2666         * tree-ssa-sccvn.c: Likewise.
2667         * tree-ssa-sink.c: Likewise.
2668         * tree-ssa-strlen.c: Likewise.
2669         * tree-ssa-structalias.c: Likewise.
2670         * tree-ssa-tail-merge.c: Likewise.
2671         * tree-ssa-ter.c: Likewise.
2672         * tree-ssa-threadedge.c: Likewise.
2673         * tree-ssa-threadupdate.c: Likewise.
2674         * tree-ssa-uncprop.c: Likewise.
2675         * tree-ssa-uninit.c: Likewise.
2676         * tree-ssa.c: Likewise.
2677         * tree-ssanames.c: Likewise.
2678         * tree-stdarg.c: Likewise.
2679         * tree-streamer-in.c: Likewise.
2680         * tree-switch-conversion.c: Likewise.
2681         * tree-tailcall.c: Likewise.
2682         * tree-vect-data-refs.c: Likewise.
2683         * tree-vect-generic.c: Likewise.
2684         * tree-vect-loop-manip.c: Likewise.
2685         * tree-vect-loop.c: Likewise.
2686         * tree-vect-patterns.c: Likewise.
2687         * tree-vect-slp.c: Likewise.
2688         * tree-vect-stmts.c: Likewise.
2689         * tree-vectorizer.c: Likewise.
2690         * tree-vrp.c: Likewise.
2691         * tree.c: Likewise.
2692         * tsan.c: Likewise.
2693         * value-prof.c: Likewise.
2694         * var-tracking.c: Likewise.
2695         * varpool.c: Likewise.
2696         * vtable-verify.c: Likewise.
2698 2013-10-23  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2700         * config/tilegx/tilegx.c: Include "tree.h".
2702 2013-10-23  Jakub Jelinek  <jakub@redhat.com>
2704         * gimple-pretty-print.c (dump_ssaname_info): Always print "# " before
2705         the info, not after it.
2706         (gump_gimple_phi): Add COMMENT argument, if true, print "# " after
2707         dump_ssaname_info call.
2708         (pp_gimple_stmt_1): Adjust caller.
2709         (dump_phi_nodes): Likewise.  Don't print "# " here.
2711 2013-10-22  Jan Hubicka  <jh@suse.cz>
2713         * i386.h (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES_EPILOGUES): New
2714         tuning flag.
2715         * x86-tune.def (TARGET_MISALIGNED_MOVE_STRING_PROLOGUES): Define it.
2716         * i386.c (expand_small_movmem_or_setmem): New function.
2717         (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): New
2718         function.
2719         (alg_usable_p): Add support for value ranges; cleanup.
2720         (ix86_expand_set_or_movmem): Add support for misaligned moves.
2722 2013-10-22  Sterling Augustine  <saugustine@google.com>
2724         * doc/invoke.texi: Document -ggnu-pubnames.
2725         * common.opt: Add new option -ggnu-pubnames and modify -gpubnames
2726         logic.
2727         * dwarf2out.c: Include gdb/gdb-index.h.
2728         (DEBUG_PUBNAMES_SECTION, DEBUG_PUBTYPES_SECTION): Handle
2729         debug_generate_pub_sections.
2730         (is_java, output_pubtables, output_pubname): New functions.
2731         (include_pubname_in_output): Handle debug_generate_pub_sections at
2732         level 2.
2733         (size_of_pubnames): Use new local space_for_flags based on
2734         debug_generate_pub_sections.
2735         (output_pubnames): Unify pubnames and pubtypes output logic.
2736         Genericize comments.  Call output_pubname.
2737         (dwarf2out_finish): Move logic to output_pubnames and call it.
2739 2013-10-22  Uros Bizjak  <ubizjak@gmail.com>
2741         PR target/58779
2742         * config/i386/i386.c (put_condition_code) <case GTU, case LEU>:
2743         Remove CCCmode handling.
2744         <case LTU>: Return 'c' suffix for CCCmode.
2745         <case GEU>: Return 'nc' suffix for CCCmode.
2746         (ix86_cc_mode) <case GTU, case LEU>: Do not generate overflow checks.
2747         * config/i386/i386.md (*sub<mode>3_cconly_overflow): Remove.
2748         (*sub<mode>3_cc_overflow): Ditto.
2749         (*subsi3_zext_cc_overflow): Ditto.
2751 2013-10-22  Steve Ellcey  <sellcey@mips.com>
2753         * config/mips/mips.c (mips_rtx_costs):  Fix cost estimate for nor
2754         (AND (NOT OP1) (NOT OP2)).
2756 2013-10-22  Bill Schmidt  <wschmidt@vnet.ibm.com>
2758         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
2759         meaning of merge-high and merge-low masks for little endian; avoid
2760         use of vector-pack masks for little endian for mismatched modes.
2762 2013-10-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
2764         * config/tilepro/tilepro.c: Include "tree.h".
2766 2013-10-22  Andreas Schwab  <schwab@suse.de>
2768         * config/m68k/m68k.c (notice_update_cc): Handle register conflict
2769         with PRE_DEC.
2771 2013-10-22  Paolo Carlini  <paolo.carlini@oracle.com>
2773         * doc/contrib.texi ([Fran@,{c}ois Dumont], [Tim Shen],
2774         [Ed Smith-Rowland]): New entries.
2775         ([Stephen M. Webb]): Update.
2777 2013-10-22  Andrew MacLeod  <amacleod@redhat.com>
2779         * tree-ssa-ter.h: Remove duplicate copy of file contents.
2781 2013-10-21  Marek Polacek  <polacek@redhat.com>
2783         PR middle-end/58809
2784         * fold-const.c (fold_range_test): Return 0 if the type is not
2785         an integral type.
2787 2013-10-21  Richard Sandiford  <rdsandiford@googlemail.com>
2789         * system.h: Move hwint.h include further down.
2790         * hwint.h (sext_hwi, zext_hwi): Define unconditionally.  Add
2791         gcc_checking_asserts.
2792         * hwint.c (sext_hwi, zext_hwi): Delete ENABLE_CHECKING versions.
2794 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2796         Fix volatile issues in optimize_bit_field_compare.
2797         * fold-const.c (optimize_bit_field_compare): Bail out if
2798         lvolatilep or rvolatilep.
2800 2013-10-21  Bernd Edlinger  <bernd.edlinger@hotmail.de>
2802         Fix DECL_BIT_FIELD depencency on flag_strict_volatile_bitfields
2803         and get_inner_reference returning different pmode for non-volatile
2804         bit-field members dependent on flag_strict_volatile_bitfields.
2805         * stor-layout.c (layout_decl): Remove special handling of
2806         flag_strict_volatile_bitfields.
2807         * expr.c (get_inner_reference): Don't use DECL_BIT_FIELD
2808         if flag_strict_volatile_bitfields > 0 and TREE_THIS_VOLATILE.
2810 2013-10-21  Paulo Matos  <pmatos@broadcom.com>
2812         * ipa-inline.c (edge_badness): Cap edge->count at max_count for badness
2813         calculations.
2815 2013-10-21  Jeff Law  <law@redhat.com>
2817         * tree-ssa-threadedge.c (thread_through_normal_block): New
2818         argument VISITED.  Remove VISISTED as a local variable.  When we
2819         have a threadable jump, verify the destination of the jump has not
2820         been visised.
2821         (thread_across_edge): Allocate VISITED bitmap once at function
2822         scope and use it throughout.  Make sure to set appropriate bits in
2823         VISITED for E (start of jump thread path).
2824         * tree-ssa-threadupdate.c (mark_threaded_blocks): Reject threading
2825         through a joiner if any edge on the path has a recorded jump thread.
2827 2013-10-21  Ian Lance Taylor  <iant@google.com>
2829         * doc/invoke.texi (Optimize Options): For -fno-toplevel-reorder,
2830         don't imply that attributes can solve all problems.
2831         (Directory Options): Fix typo.
2833 2013-10-21  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
2835         * config/arm/arm.c (cortexa9_extra_costs): Update mult costs for
2836         extend and extend_add.
2838 2013-10-21  Richard Biener  <rguenther@suse.de>
2840         PR tree-optimization/58794
2841         * fold-const.c (operand_equal_p): Compare FIELD_DECL operand
2842         of COMPONENT_REFs with OEP_CONSTANT_ADDRESS_OF left in place.
2844 2013-10-21  Richard Biener  <rguenther@suse.de>
2846         PR middle-end/58742
2847         * fold-const.c (fold_binary_loc): Fold ((T) (X /[ex] C)) * C
2848         to (T) X for sign-changing conversions (or no conversion).
2850 2013-10-20  Uros Bizjak  <ubizjak@gmail.com>
2852         * config/i386/i386.md (kxnor<mode>): Add FLAGS_REG clobber.
2854 2013-10-20  Jan Hubicka  <jh@suse.cz>
2856         * config/i386/i386-tune.def: Add comment; organize into categories
2858 2013-10-21  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
2860         * config/i386/i386.c (expand_set_or_movmem_via_loop): Add issetmem
2861         argument.  Update function comment.
2862         (expand_set_or_movmem_via_rep): New function combining
2863         expand_movmem_via_rep_mov and expand_setmem_via_rep_stos.
2864         (expand_movmem_via_rep_mov): Remove.
2865         expand_setmem_via_rep_stos): Remove.
2866         (expand_movmem_epilogue): Update calls correspondingly.
2867         (expand_setmem_epilogue_via_loop): Likewise.
2868         (emit_memset): New.
2869         (expand_setmem_epilogue): Add VEC_VALUE argument, refactor.
2870         (expand_set_or_movmem_prologue): New function combining
2871         expand_movmem_prologue and expand_setmem_prologue.
2872         (expand_movmem_prologue): Remove.
2873         (expand_setmem_prologue): Remove.
2874         (expand_set_or_movmem_constant_prologue): New function combining
2875         expand_constant_movmem_prologue and expand_constant_setmem_prologue.
2876         (expand_constant_movmem_prologue): Remove.
2877         (expand_constant_setmem_prologue): Remove.
2878         (promote_duplicated_reg): Allow vector-const0 value.
2879         (ix86_expand_set_or_movmem): New function combining ix86_expand_movmem
2880         and ix86_expand_setmem.
2881         (ix86_expand_movmem): Call ix86_expand_set_or_movmem.
2882         (ix86_expand_setmem): Call ix86_expand_set_or_movmem.
2884 2013-10-21  Diego Novillo  <dnovillo@google.com>
2886         * asan.c: Include tree.h
2887         * bb-reorder.c: Likewise.
2888         * cfgcleanup.c: Likewise.
2889         * cfgloopmanip.c: Likewise.
2890         * data-streamer-in.c: Likewise.
2891         * data-streamer-out.c: Likewise.
2892         * data-streamer.c: Likewise.
2893         * dwarf2cfi.c: Likewise.
2894         * graphite-blocking.c: Likewise.
2895         * graphite-clast-to-gimple.c: Likewise.
2896         * graphite-dependences.c: Likewise.
2897         * graphite-interchange.c: Likewise.
2898         * graphite-optimize-isl.c: Likewise.
2899         * graphite-poly.c: Likewise.
2900         * graphite-scop-detection.c: Likewise.
2901         * graphite-sese-to-poly.c: Likewise.
2902         * graphite.c: Likewise.
2903         * ipa-devirt.c: Likewise.
2904         * ipa-profile.c: Likewise.
2905         * ipa.c: Likewise.
2906         * ira.c: Likewise.
2907         * loop-init.c: Likewise.
2908         * loop-unroll.c: Likewise.
2909         * lower-subreg.c: Likewise.
2910         * lto/lto-object.c: Likewise.
2911         * recog.c: Likewise.
2912         * reginfo.c: Likewise.
2913         * tree-loop-distribution.c: Likewise.
2914         * tree-parloops.c: Likewise.
2915         * tree-ssa-strlen.c: Likewise.
2916         * tree-streamer.c: Likewise.
2917         * value-prof.c: Likewise.
2918         * target-globals.c: Likewise.
2919         * expr.h: Include tree-core.h instead of tree.h.
2920         * gimple.h: Likewise.
2921         * ipa-prop.h: Likewise.
2922         * ipa-utils.h: Likewise.
2923         * lto-streamer.h: Likewise.
2924         * streamer-hooks.h: Likewise.
2925         * ipa-reference.h: Include cgraph.h instead of tree.h.
2926         * cgraph.h: Include basic-block.h instead of tree.h.
2927         * tree-streamer.h: Do not include tree.h.
2928         * genattrtab.c (write_header): Generate inclusion of
2929         tree.h.
2930         * genautomata.c (main): Likewise.
2931         * genemit.c: Likewise.
2932         * genopinit.c: Likewise.
2933         * genoutput.c (output_prologue): Likewise.
2934         * genpeep.c: Likewise.
2936 2013-10-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2938         * config/rs6000/altivec.md (vec_unpacku_hi_v16qi): Adjust for
2939         little endian.
2940         (vec_unpacku_hi_v8hi): Likewise.
2941         (vec_unpacku_lo_v16qi): Likewise.
2942         (vec_unpacku_lo_v8hi): Likewise.
2944 2013-10-20  Jan Hubicka  <jh@suse.cz>
2946         * config/i386/x86-tune.def (X86_TUNE_SLOW_IMUL_IMM32_MEM,
2947         X86_TUNE_SLOW_IMUL_IMM8): Keep enabled only for K8 and AMDFAM10.
2948         (X86_TUNE_USE_VECTOR_FP_CONVERTS): Disable for generic.
2950 2013-10-20  Richard Sandiford  <rdsandiford@googlemail.com>
2952         * config/mips/mips.h (ISA_HAS_WSBH): Define.
2953         * config/mips/mips.md (UNSPEC_WSBH, UNSPEC_DSBH, UNSPEC_DSHD): New
2954         constants.
2955         (bswaphi2, bswapsi2, bswapdi2, wsbh, dsbh, dshd): New patterns.
2957 2013-10-19  John David Anglin  <danglin@gcc.gnu.org>
2959         PR target/58603
2960         * system.h: Undef m_slot.
2962 2013-10-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
2964         * config/rs6000/rs6000.c (vspltis_constant): Make sure we check
2965         all elements for both endian flavors.
2967 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2969         PR target/58792
2970         * config/i386/i386.c (ix86_function_value_regno): Add DX_REG,
2971         ST1_REG and XMM1_REG for 32bit and 64bit targets.  Also add DI_REG
2972         and SI_REG for 64bit SYSV ABI targets.
2974 2013-10-19  Uros Bizjak  <ubizjak@gmail.com>
2976         * mode-switching.c (create_pre_exit): Rename maybe_builtin_apply
2977         to multi_reg_return.  Clarify that we are skipping USEs of multiple
2978         return registers.  Use bool type where appropriate.
2980 2013-10-18  Jan Hubicka  <jh@suse.cz>
2982         * config/i386/i386.h (ACCUMULATE_OUTGOING_ARGS): Disable accumulation
2983         for cold functions.
2984         * x86-tune.def (X86_TUNE_USE_LEAVE): Update comment.
2985         (X86_TUNE_PUSH_MEMORY): Likewise.
2986         (X86_TUNE_AVX256_UNALIGNED_LOAD_OPTIMAL,
2987         X86_TUNE_AVX256_UNALIGNED_STORE_OPTIMAL): New.
2988         (X86_TUNE_ACCUMULATE_OUTGOING_ARGS, X86_TUNE_ALWAYS_FANCY_MATH_387):
2989         New.
2990         * i386.c (x86_accumulate_outgoing_args, x86_arch_always_fancy_math_387,
2991         x86_avx256_split_unaligned_load, x86_avx256_split_unaligned_store):
2992         Remove.
2993         (ix86_option_override_internal): Update to use tune features instead
2994         of variables.
2996 2013-10-18  Cong Hou  <congh@google.com>
2998         PR tree-optimization/58508
2999         * tree-vect-loop-manip.c (vect_loop_versioning): Hoist loop invariant
3000         statement that contains data refs with zero-step.
3002 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3004         * tree-ssa.h: Don't include gimple-low.h, tree-ssa-address.h,
3005         sbitmap.h, tree-ssa-threadedge.h, tree-ssa-dom.h and tree-cfgcleanup.h.
3006         * gimple-low.c (gimple_check_call_arg,
3007         gimple_check_call_matching_types): Move to cgraph.c.
3008         * gimple-low.h: Remove prototype.
3009         * cgraph.c: (gimple_check_call_arg, gimple_check_call_matching_types):
3010         Relocate from gimple-low.c.
3011         * cgraph.h: Add prototype.  Don't include basic-block.h.
3012         * gimplify.c: Add gimple-low to include list.
3013         * omp-low.c: Add gimple-low and tree-cfgcleanup.h to include list.
3014         * tree-eh.c: Add gimple-low to include list.
3015         * tree-nested.c: Likewise.
3016         * cfgexpand.c: Add tree-ssa-address.h to include list.
3017         * expr.c: Likewise.
3018         * gimple-fold.c: Likewise.
3019         * gimple-ssa-strength-reduction.c: Likewise.
3020         * trans-mem.c: Likewise.
3021         * tree-mudflap.c: Likewise.
3022         * tree-ssa-loop-ivopts.c: Likewise.
3023         * tree-ssa-dom.c: Include tree-ssa-threadedge.h and tree-ssa-dom.h.
3024         (degenerate_phi_result): Move to tree-phinodes.c.
3025         * tree-ssa-loop-ch.c: Include tree-ssa-threadedge.h.
3026         * tree-ssa-threadedge.c: Likewise.
3027         * tree-vrp.c: Likewise.
3028         * tree-phinodes.c (degenerate_phi_result): Relocate here.
3029         * tree-ssa-dom.h (degenerate_phi_result): Remove Prototype.
3030         * tree-phinodes.h (degenerate_phi_result): Add prototype.
3031         * tree-ssa-copy.c: Include tree-ssa-dom.h.
3032         * tree-ssa-forwprop.c: Likewise.
3033         * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing,
3034         pass_data_cleanup_cfg_post_optimizing,
3035         make_pass_cleanup_cfg_post_optimizing): Relocate from tree-optimize.c.
3036         * tree-optimize.c: Delete File.
3037         * graphite.c: Include tree-cfgcleanup.h.
3038         * passes.c: Likewise.
3039         * tree-cfg.c: Likewise.
3040         * tree-profile.c: Likewise.
3041         * tree-ssa-dse.c: Likewise.
3042         * tree-ssa-loop-ivcanon.c: Likewise.
3043         * tree-switch-conversion.c: Don't include tree-ssa-operands.h.
3044         * tree-outof-ssa.c: Include sbitmap.h.
3045         * tree-ssa-live.c: Likewise.
3046         * tree-ssa-propagate.c: Likewise.
3047         * tree-ssa-structalias.c: Likewise.
3048         * tree-stdarg.c: Likewise.
3049         * Makefile.in (OBJS): Delete tree-optimize.o.
3050         * basic-block.h (gcov_type, gcov_type_unsigned): Move to coretypes.h.
3051         * coretypes.h (gcov_type, gcov_type_unsigned): Relocate here.
3052         * varasm.c: Include basic-block.h.
3053         * cfgloop.h: Include function.h instead of basic-block.h
3054         (bb_loop_depth): Move to cfgloop.c.
3055         * cfgloop.c (bb_loop_depth): Relocate from cfgloop.h.
3057 2013-10-18  Teresa Johnson  <tejohnson@google.com>
3059         * predict.c (probably_never_executed): Compare frequency-based
3060         count to number of training runs.
3061         * params.def (UNLIKELY_BB_COUNT_FRACTION): New parameter.
3063 2013-10-18  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3065         * config/arm/arm.c (cortexa9_extra_costs): New table.
3066         (arm_cortex_a9_tune): Use cortexa9_extra_costs.
3068 2013-10-18  Jeff Law  <law@redhat.com>
3070         * tree-ssa-threadupdate.c: Do not include "tm.h" or "tm_p.h".
3072         * tree-ssa-threadupdate.c: Include "dbgcnt.h".
3073         (register_jump_thread): Add "registered_jump_thread" debug
3074         counter support.
3075         * dbgcnt.def (registered_jump_thread): New debug counter.
3077 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3079         * config/rs6000/rs6000.c: Include cgraph.h.
3081 2013-10-18  Teresa Johnson  <tejohnson@google.com>
3083         * tree-ssa-tail-merge.c (replace_block_by): Update edge
3084         weights during merging.
3086 2013-10-18  Andrew MacLeod  <amacleod@redhat.com>
3088         * tree-cfg.h: Rename from tree-flow.h.  Remove #includes.
3089         * tree-ssa.h: Relocate required #includes from tree-cfg.h.
3090         * tree-ssa-operands.h: Remove prototype.
3091         * tree-ssa-operands.c (virtual_operand_p): Move to gimple.c.
3092         * gimple.c (virtual_operand_p): Relocate from gimple.c.
3093         * gimple.h: Add prototype.
3094         * gimple-ssa.h: Include tree-ssa-operands.h.
3095         * tree-dump.c: Add tree-cfg.h to include list.
3096         * tree-ssa-alias.c: Add ipa-reference.h to include list.
3097         * config/alpha/alpha.c: Include gimple-ssa.h instead of tree-flow.h.
3098         * config/i386/i386.c: Don't include tree-flow.h.
3099         * config/rs6000/rs6000.c: Likewise.
3101 2013-10-18  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
3103         * config/frv/frv.c (frv_init_cumulative_args): Fix wrong cast.
3105 2013-10-18  Richard Biener  <rguenther@suse.de>
3107         * stor-layout.c (layout_type): Do not change TYPE_PRECISION
3108         or TYPE_UNSIGNED of integral types.
3109         (set_min_and_max_values_for_integral_type): Leave TYPE_MIN/MAX_VALUE
3110         NULL_TREE for zero-precision integral types.
3112 2013-10-18  James Greenhalgh  <james.greenhalgh@arm.com>
3114         * config/aarch64/arm_neon.h
3115         (vcvt<ds>_n_<fsu><32,64>_<fsu><32,64>): Correct argument types.
3117 2013-10-17  Sriraman Tallam  <tmsriram@google.com>
3119         PR target/57756
3120         * opth-gen.awk: Define target_flags_explicit.
3122 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3124         * config/rs6000/rs6000.c (enum rs6000_reload_reg_type): Add new
3125         fields to the reg_addr array that describes the valid addressing
3126         mode for any register, general purpose registers, floating point
3127         registers, and Altivec registers.
3128         (FIRST_RELOAD_REG_CLASS): Likewise.
3129         (LAST_RELOAD_REG_CLASS): Likewise.
3130         (struct reload_reg_map_type): Likewise.
3131         (reload_reg_map_type): Likewise.
3132         (RELOAD_REG_VALID): Likewise.
3133         (RELOAD_REG_MULTIPLE): Likewise.
3134         (RELOAD_REG_INDEXED): Likewise.
3135         (RELOAD_REG_OFFSET): Likewise.
3136         (RELOAD_REG_PRE_INCDEC): Likewise.
3137         (RELOAD_REG_PRE_MODIFY): Likewise.
3138         (reg_addr): Likewise.
3139         (mode_supports_pre_incdec_p): New helper functions to say whether
3140         a given mode supports PRE_INC, PRE_DEC, and PRE_MODIFY.
3141         (mode_supports_pre_modify_p): Likewise.
3142         (rs6000_debug_vector_unit): Rearrange the -mdebug=reg output to
3143         print the valid address mode bits for each mode.
3144         (rs6000_debug_print_mode): Likewise.
3145         (rs6000_debug_reg_global): Likewise.
3146         (rs6000_setup_reg_addr_masks): New function to set up the address
3147         mask bits for each type.
3148         (rs6000_init_hard_regno_mode_ok): Use memset to clear arrays.
3149         Call rs6000_setup_reg_addr_masks to set up the address mask bits.
3150         (rs6000_legitimate_address_p): Use mode_supports_pre_incdec_p and
3151         mode_supports_pre_modify_p to determine if PRE_INC, PRE_DEC, and
3152         PRE_MODIFY are supported.
3153         (rs6000_output_move_128bit): Change to use {src,dest}_vmx_p for altivec
3154         registers, instead of {src,dest}_av_p.
3155         (rs6000_print_options_internal): Tweak the debug output slightly.
3157 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3159         * config/i386/sse.md (*vec_widen_smult_even_v8si): Remove
3160         isa attribute.
3162 2013-10-17  Andrew MacLeod  <amacleod@redhat.com>
3164         * tree-flow.h (struct omp_region): Move to omp-low.c.
3165         Remove omp_ prototypes and variables.
3166         * gimple.h (omp_reduction_init): Move prototype to omp-low.h.
3167         (copy_var_decl): Relocate prototype from tree-flow.h.
3168         * gimple.c (copy_var_decl): Relocate from omp-low.c.
3169         * tree.h: Move prototype to omp-low.h.
3170         * omp-low.h: New File.  Relocate prototypes here.
3171         * omp-low.c (struct omp_region): Make local here.
3172         (root_omp_region): Make static.
3173         (copy_var_decl) Move to gimple.c.
3174         (new_omp_region): Make static.
3175         (make_gimple_omp_edges): New.  Refactored from tree-cfg.c make_edges.
3176         * tree-cfg.c: Include omp-low.h.
3177         (make_edges): Factor out OMP specific bits to make_gimple_omp_edges.
3178         * gimplify.c: Include omp-low.h.
3179         * tree-parloops.c: Likewise.
3181 2013-10-17  Uros Bizjak  <ubizjak@gmail.com>
3183         * config/i386/i386.c (ix86_fixup_binary_operands): When both source
3184         operands are in memory, prefer to force non-matched operand 1 to
3185         the register.
3187 2013-10-17  Michael Meissner  <meissner@linux.vnet.ibm.com>
3189         PR target/58673
3190         * config/rs6000/rs6000.c (rs6000_legitimate_address_p): Only
3191         restrict TImode addresses to single indirect registers if both
3192         -mquad-memory and -mvsx-timode are used.
3193         (rs6000_output_move_128bit): Use quad_load_store_p to determine if
3194         we should emit load/store quad.  Remove using %y for quad memory
3195         addresses.
3197         * config/rs6000/rs6000.md (mov<mode>_ppc64, TI/PTImode): Add
3198         constraints to allow load/store quad on machines where TImode is
3199         not allowed in VSX registers.  Use 'n' instead of 'F' constraint
3200         for TImode to load integer constants.
3202 2013-10-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
3204         * config/aarch64/aarch64.c (aarch64_print_operand): Handle 'c'.
3206 2013-10-17  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3208         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Adjust
3209         handling of STACK_REG.
3211 2013-10-17  Richard Biener  <rguenther@suse.de>
3213         PR tree-optimization/58143
3214         * tree-ssa-loop-im.c (arith_code_with_undefined_signed_overflow):
3215         New function.
3216         (rewrite_to_defined_overflow): Likewise.
3217         (move_computations_dom_walker::before_dom): Rewrite stmts
3218         with undefined signed overflow that are not always executed
3219         into unsigned arithmetic.
3221 2013-10-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
3223         PR target/57756
3224         * config/rs6000/rs6000.opt (rs6000_isa_flags_explicit): Move the
3225         explicit isa flag to be an options variable, instead of using
3226         global_options_set.  Remove define from rs6000.h.
3227         * config/rs6000/rs6000.h (rs6000_isa_flags_explicit): Likewise.
3229         * config/rs6000/rs6000.c (rs6000_option_override_internal):
3230         Initialize rs6000_isa_flags_explicit.
3231         (rs6000_function_specific_save): Add gcc_options* parameter, so
3232         that the powerpc builds after the 2013-10-15 changes.
3233         (rs6000_function_specific_restore): Likewise.
3235 2013-10-16  DJ Delorie  <dj@redhat.com>
3237         * config/rl78/rl78.c (rl78_alloc_address_registers_macax): Verify
3238         op is a REG before checking REGNO.
3239         (rl78_alloc_physical_registers): Verify pattern is a SET before
3240         checking SET_SRC.
3242 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3244         * config/rs6000/vector.md (vec_unpacks_hi_v4sf): Correct for
3245         endianness.
3246         (vec_unpacks_lo_v4sf): Likewise.
3247         (vec_unpacks_float_hi_v4si): Likewise.
3248         (vec_unpacks_float_lo_v4si): Likewise.
3249         (vec_unpacku_float_hi_v4si): Likewise.
3250         (vec_unpacku_float_lo_v4si): Likewise.
3252 2013-10-16  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
3254         * config/rs6000/vsx.md (vsx_concat_<mode>): Adjust output for LE.
3255         (vsx_concat_v2sf): Likewise.
3257 2013-10-16  James Greenhalgh  <james.greenhalgh@arm.com>
3259         * config/aarch64/aarch64.md
3260         (*mov<mode>_aarch64): Fix output template for DUP (element) Scalar.
3262 2013-10-16  Andrew MacLeod  <amacleod@redhat.com>
3264         PR tree-optimization/58697
3265         * cfgloop.c (get_estimated_loop_iterations_int): Rename from
3266         estimated_loop_iterations_int.
3267         (max_stmt_executions_int): Call get_max_loop_iterations_int.
3268         (get_max_loop_iterations_int): New.  HWINT version of
3269         get_max_loop_iterations.
3270         * cfgloop.h: Add prototypes.
3271         * loop-iv.c (find_simple_exit): call get_estimated_loop_iterations_int.
3272         * loop-unroll.c (decide_peel_once_rolling): Call
3273         get_estimated_loop_iterations_int.
3274         * tree-ssa-loop-niter.c (estimated_loop_iterations_int): Add back.
3275         * tree-ssa-loop-niter.h: Tweak prototypes.
3277 2013-10-16  David Malcolm  <dmalcolm@redhat.com>
3279         * gengtype-parse.c (struct_field_seq): Ignore access-control
3280         keywords ("public:" etc).
3282 2013-10-16  Marcus Shawcroft  <marcus.shawcroft@arm.com>
3284         * config/aarch64/aarch64.c (aarch64_regno_regclass): Classify
3285         FRAME_POINTER_REGNUM and ARG_POINTER_REGNUM as POINTER_REGS.
3287 2013-10-16  Yvan Roux  <yvan.roux@linaro.org>
3289         * config/arm/arm.opt (mlra): New option.
3290         * config/arm/arm.c (arm_lra_p): New function.
3291         (TARGET_LRA_P): Define.
3293 2013-10-16  Paulo Matos  <pmatos@broadcom.com>
3295         * tree-core.h (tree_code_name): Remove.
3296         * tree.h (get_tree_code_name): New prototype.
3297         * tree.c (tree_code_name): Make static.
3298         (get_tree_code_name): New function.
3299         (dump_tree_statistics, tree_check_failed, tree_not_check_failed,
3300         tree_class_check_failed, tree_range_check_failed,
3301         tree_not_class_check_failed, omp_clause_check_failed,
3302         tree_contains_struct_check_failed, tree_operand_check_failed): Use new
3303         wrapper get_tree_code_name instead of calling tree_code_name directly.
3304         * tree-vrp.c (dump_asserts_for): Likewise.
3305         * tree-dump.c (dequeue_and_dump): Likewise.
3306         * tree-pretty-print.c (do_niy, dump_generic_node): Likewise.
3307         * tree-pretty-print.h (pp_unsupported_tree): Likewise.
3308         * lto-streamer-out.c (lto_write_tree, DFS_write_tree): Likewise.
3309         * tree-ssa-dom.c (print_expr_hash_elt): Likewise.
3310         * gimple-pretty-print.c (dump_unary_rhs, dump_binary_rhs,
3311         dump_ternary_rhs, dump_gimple_assign, dump_gimple_cond,
3312         dump_gimple_omp_for): Likewise.
3313         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3314         * tree-ssa-pre.c (print_pre_expr): Likewise.
3315         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3316         * print-tree.c (print_node_brief, print_node): Likewise.
3317         * gimple.c (gimple_check_failed): Likewise.
3318         * lto-streamer.c (lto_tag_name, print_lto_report): Likewise.
3319         * config/frv/frv.c (frv_init_cumulative_args): Likewise.
3320         * config/mep/mep.c (mep_validate_vliw): Likewise.
3321         * config/iq2000/iq2000.c (init_cumulative_args): Likewise.
3322         * config/rs6000/rs6000.c (init_cumulative_args): Likewise.
3324 2013-10-16 Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
3326         * config/i386/i386.c (ix86_option_override_internal): Enable FMA4
3327         for AMD bdver3.
3329 2013-10-16  Hans-Peter Nilsson  <hp@axis.com>
3331         * config/cris/t-elfmulti (MULTILIB_OPTIONS, MULTILIB_DIRNAMES)
3332         (MULTILIB_MATCHES): Add multilib for -march=v8.
3334 2013-10-15 Sriraman Tallam  <tmsriram@google.com>
3336         PR target/57756
3337         * optc-save-gen.awk: Add extra parameter to the save and restore
3338         target calls.
3339         * opth-gen.awk: Generate new TARGET_* macros  to accept a parameter.
3340         * tree.c (build_optimization_node): New parameter.  Add extra parameter
3341         to call to cl_optimization_save.
3342         (build_target_option_node): New parameter. Add extra parameter
3343         to call to cl_target_option_save.
3344         * tree.h (build_optimization_node): New parameter.
3345         (build_target_option_node): New parameter.
3346         * c-family/c-common.c (handle_optimize_attribute): Fix calls to
3347         build_optimization_node and build_target_option_node.
3348         * c-family/c-pragma.c (handle_pragma_optimize): Ditto.
3349         (handle_pragma_push_options): Ditto.
3350         * toplev.c (process_options): Ditto.
3351         * opts.c (init_options_struct): Check for opts_set non-null.
3352         * target.def (target_option.save): New parameter.
3353         (target_option.restore): New parameter.
3354         * tm.texi: Generate.
3355         * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
3356         (ix86_pragma_target_parse): Ditto.
3357         * config/i386/i386-protos.h (ix86_valid_target_attribute_tree): New
3358         parameters.
3359         * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix calls
3360         to  build_optimization_node and build_target_option_node.
3361         (rs6000_valid_attribute_p): Ditto.
3362         (rs6000_pragma_target_parse): Ditto.
3363         * config/i386/i386.opt (x_ix86_target_flags_explicit): New TargetSave
3364         data.
3365         * config/i386/i386.h:
3366         TARGET_64BIT_P: New Macro
3367         TARGET_MMX_P: New Macro.
3368         TARGET_3DNOW_P: New Macro.
3369         TARGET_3DNOW_A_P: New Macro.
3370         TARGET_SSE_P: New Macro.
3371         TARGET_SSE2_P: New Macro.
3372         TARGET_SSE3_P: New Macro.
3373         TARGET_SSSE3_P: New Macro.
3374         TARGET_SSE4_1_P: New Macro.
3375         TARGET_SSE4_2_P: New Macro.
3376         TARGET_AVX_P: New Macro.
3377         TARGET_AVX2_P: New Macro.
3378         TARGET_AVX512F_P: New Macro.
3379         TARGET_AVX512PF_P: New Macro.
3380         TARGET_AVX512ER_P: New Macro.
3381         TARGET_AVX512CD_P: New Macro.
3382         TARGET_FMA_P: New Macro.
3383         TARGET_SSE4A_P: New Macro.
3384         TARGET_FMA4_P: New Macro.
3385         TARGET_XOP_P: New Macro.
3386         TARGET_LWP_P: New Macro.
3387         TARGET_ABM_P: New Macro.
3388         TARGET_BMI_P: New Macro.
3389         TARGET_BMI2_P: New Macro.
3390         TARGET_LZCNT_P: New Macro.
3391         TARGET_TBM_P: New Macro.
3392         TARGET_POPCNT_P: New Macro.
3393         TARGET_SAHF_P: New Macro.
3394         TARGET_MOVBE_P: New Macro.
3395         TARGET_CRC32_P: New Macro.
3396         TARGET_AES_P: New Macro.
3397         TARGET_PCLMUL_P: New Macro.
3398         TARGET_CMPXCHG16B_P: New Macro.
3399         TARGET_FSGSBASE_P: New Macro.
3400         TARGET_RDRND_P: New Macro.
3401         TARGET_F16C_P: New Macro.
3402         TARGET_RTM_P: New Macro.
3403         TARGET_HLE_P: New Macro.
3404         TARGET_RDSEED_P: New Macro.
3405         TARGET_PRFCHW_P: New Macro.
3406         TARGET_ADX_P: New Macro.
3407         TARGET_FXSR_P: New Macro.
3408         TARGET_XSAVE_P: New Macro.
3409         TARGET_XSAVEOPT_P: New Macro.
3410         TARGET_LP64_P: New Macro.
3411         TARGET_X32_P: New Macro.
3412         TARGET_FPMATH_DEFAULT_P: New Macro.
3413         TARGET_FLOAT_RETURNS_IN_80387_P: New Macro.
3414         * config/i386/i386.c (ix86_option_override_internal): New parameters.
3415         opts and opts_set.
3416         Change ix86_tune_string to access opts->x_ix86_tune_string.
3417         Change ix86_isa_flags to access opts->x_ix86_isa_flags.
3418         Change ix86_arch_string to access opts->x_ix86_arch_string.
3419         Change ix86_stringop_alg to access opts->x_ix86_stringop_alg.
3420         Change ix86_pmode to access opts->x_ix86_pmode.
3421         Change ix86_abi to access opts->x_ix86_abi.
3422         Change ix86_cmodel to access opts->x_ix86_cmodel.
3423         Change ix86_asm_dialect to access opts->x_ix86_asm_dialect.
3424         Change ix86_isa_flags_explicit to access
3425         opts->x_ix86_isa_flags_explicit.
3426         Change ix86_dump_tunes to access opts->x_ix86_dump_tunes.
3427         Change ix86_regparm to access opts->x_ix86_regparm.
3428         Change ix86_branch_cost to access opts->x_ix86_branch_cost.
3429         Change ix86_preferred_stack_boundary_arg to access
3430         opts->x_ix86_preferred_stack_boundary_arg.
3431         Change ix86_force_align_arg_pointer to access
3432         opts->x_ix86_force_align_arg_pointer.
3433         Change ix86_incoming_stack_boundar_arg to access
3434         opts->x_ix86_incoming_stack_boundar_arg.
3435         Change ix86_fpmath to access opts->x_ix86_fpmath.
3436         Change ix86_veclibabi_type to access opts->x_ix86_veclibabi_type.
3437         Change ix86_recip_name to access opts->x_ix86_recip_name.
3438         Change ix86_stack_protector_guard to access
3439         opts->x_ix86_stack_protector_guard.
3440         Change ix86_tune_memcpy_strategy to access
3441         opts->x_ix86_tune_memcpy_strategy.
3442         Change ix86_tune_memset_strategy to access
3443         opts->x_ix86_tune_memset_strategy.
3444         Change global_options to access opts.
3445         Change global_options_set to access opts_set.
3446         Change TARGET_64BIT to TARGET_64BIT_P (opts->...).
3447         Change TARGET_MMX to TARGET_MMX_P (opts->...).
3448         Change TARGET_3DNOW to TARGET_3DNOW_P (opts->...).
3449         Change TARGET_3DNOW_A to TARGET_3DNOW_A_P (opts->...).
3450         Change TARGET_SSE to TARGET_SSE_P (opts->...).
3451         Change TARGET_SSE2 to TARGET_SSE2_P (opts->...).
3452         Change TARGET_SSE3 to TARGET_SSE3_P (opts->...).
3453         Change TARGET_SSSE3 to TARGET_SSSE3_P (opts->...).
3454         Change TARGET_SSE4_1 to TARGET_SSE4_1_P (opts->...).
3455         Change TARGET_SSE4_2 to TARGET_SSE4_2_P (opts->...).
3456         Change TARGET_AVX to TARGET_AVX_P (opts->...).
3457         Change TARGET_AVX2 to TARGET_AVX2_P (opts->...).
3458         Change TARGET_AVX512F to TARGET_AVX512F_P (opts->...).
3459         Change TARGET_AVX512PF to TARGET_AVX512PF_P (opts->...).
3460         Change TARGET_AVX512ER to TARGET_AVX512ER_P (opts->...).
3461         Change TARGET_AVX512CD to TARGET_AVX512CD_P (opts->...).
3462         Change TARGET_FMA to TARGET_FMA_P (opts->...).
3463         Change TARGET_SSE4A to TARGET_SSE4A_P (opts->...).
3464         Change TARGET_FMA4 to TARGET_FMA4_P (opts->...).
3465         Change TARGET_XOP to TARGET_XOP_P (opts->...).
3466         Change TARGET_LWP to TARGET_LWP_P (opts->...).
3467         Change TARGET_ABM to TARGET_ABM_P (opts->...).
3468         Change TARGET_BMI to TARGET_BMI_P (opts->...).
3469         Change TARGET_BMI2 to TARGET_BMI2_P (opts->...).
3470         Change TARGET_LZCNT to TARGET_LZCNT_P (opts->...).
3471         Change TARGET_TBM to TARGET_TBM_P (opts->...).
3472         Change TARGET_POPCNT to TARGET_POPCNT_P (opts->...).
3473         Change TARGET_SAHF to TARGET_SAHF_P (opts->...).
3474         Change TARGET_MOVBE to TARGET_MOVBE_P (opts->...).
3475         Change TARGET_CRC32 to TARGET_CRC32_P (opts->...).
3476         Change TARGET_AES to TARGET_AES_P (opts->...).
3477         Change TARGET_PCLMUL to TARGET_PCLMUL_P (opts->...).
3478         Change TARGET_CMPXCHG16B to TARGET_CMPXCHG16B_P (opts->...).
3479         Change TARGET_FSGSBASE to TARGET_FSGSBASE_P (opts->...).
3480         Change TARGET_RDRND to TARGET_RDRND_P (opts->...).
3481         Change TARGET_F16C to TARGET_F16C_P (opts->...).
3482         Change TARGET_RTM to TARGET_RTM_P (opts->...).
3483         Change TARGET_HLE to TARGET_HLE_P (opts->...).
3484         Change TARGET_RDSEED to TARGET_RDSEED_P (opts->...).
3485         Change TARGET_PRFCHW to TARGET_PRFCHW_P (opts->...).
3486         Change TARGET_ADX to TARGET_ADX_P (opts->...).
3487         Change TARGET_FXSR to TARGET_FXSR_P (opts->...).
3488         Change TARGET_XSAVE to TARGET_XSAVE_P (opts->...).
3489         Change TARGET_XSAVEOPT to TARGET_XSAVEOPT_P (opts->...).
3490         Change TARGET_LP64 to TARGET_LP64_P (opts->...).
3491         Change TARGET_X32 to TARGET_X32_P (opts->...).
3492         Change TARGET_FPMATH_DEFAULT to TARGET_FPMATH_DEFAULT_P (opts->...).
3493         Change TARGET_FLOAT_RETURNS_IN_80387 to
3494         TARGET_FLOAT_RETURNS_IN_80387_P (opts->...).
3495         (ix86_function_specific_save): New parameter. Use opts-> fields
3496         to replace global fields.
3497         (ix86_function_specific_restore): Ditto.
3498         (ix86_valid_target_attribute_inner_p): New parameters.
3499         Fix recursive call.
3500         Fix call to ix86_handle_option and set_option.
3501         (ix86_valid_target_attribute_tree): New parameters.
3502         Change global_options to access opts.
3503         Change global_options_set to access opts_set.
3504         Fix call to ix86_valid_target_attribute_inner_p.
3505         Change ix86_tune_string to access opts->x_ix86_tune_string.
3506         Change ix86_arch_string to access opts->x_ix86_arch_string.
3507         Change ix86_fpmath to access opts->x_ix86_fpmath
3508         Fix call to ix86_option_override_internal.
3509         Fix call to ix86_add_new_builtins.
3510         Fix calls to build_optimization_node and build_target_option_node.
3511         (ix86_valid_target_attribute_p): Remove access to global_options.
3512         Use new gcc_options structure func_options.
3513         Fix call to ix86_valid_target_attribute_tree.
3514         Fix call to  build_optimization_node.
3515         (get_builtin_code_for_version): Fix call to
3516         ix86_valid_target_attribute_tree.
3518 2013-10-15  David Malcolm  <dmalcolm@redhat.com>
3520         * Makefile.in (PICFLAG): New.
3521         (enable_host_shared): New.
3522         (INTERNAL_CFLAGS): Use PICFLAG.
3523         (LIBIBERTY): Use pic build of libiberty.a if configured with
3524         --enable-host-shared.
3525         * configure.ac: Add --enable-host-shared, setting up new
3526         PICFLAG variable.
3527         * configure: Regenerate.
3528         * doc/install.texi (--enable-shared): Add note contrasting it with ...
3529         (--enable-host-shared): New option.
3531 2013-10-15  Richard Biener  <rguenther@suse.de>
3533         * tree-tailcall.c (find_tail_calls): Don't use tail-call recursion
3534         for built-in functions.
3536 2013-10-15  Zhenqiang Chen  <zhenqiang.chen@arm.com>
3538         * tree-ssa-reassoc.c: Include rtl.h and tm_p.h.
3539         (optimize_range_tests_1): New function,
3540         extracted from optimize_range_tests.
3541         (optimize_range_tests_xor): Similarly.
3542         (optimize_range_tests_diff): New function.
3543         (optimize_range_tests): Use optimize_range_tests_1.
3545 2013-10-15  Cong Hou  <congh@google.com>
3547         * tree-vect-loop.c (vect_is_simple_reduction_1): Relax the
3548         requirement of the reduction pattern so that one operand of the
3549         reduction operation can come from outside of the loop.
3551 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3553         * config/arm/neon-schedgen.ml: Remove.
3554         * config/arm/cortex-a9-neon.md: Remove comment regarding
3555         neon-schedgen.ml.
3557 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3559         * config/arm/types: Remove old neon types.
3561 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3563         * config/arm/cortex-a7.md
3564         (cortex_a7_neon_type): New.
3565         (cortex_a7_neon_mul): Update for new types.
3566         (cortex_a7_neon_mla): Likewise.
3567         (cortex_a7_neon): Likewise.
3569 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3571         * config/arm/cortex-a15-neon.md
3572         (cortex_a15_neon_type): New,
3574         (cortex_a15_neon_int_1): Remove.
3575         (cortex_a15_neon_int_2): Likewise.
3576         (cortex_a15_neon_int_3): Likewise.
3577         (cortex_a15_neon_int_4): Likewise.
3578         (cortex_a15_neon_int_5): Likewise.
3579         (cortex_a15_neon_vqneg_vqabs): Likewise.
3580         (cortex_a15_neon_vmov): Likewise.
3581         (cortex_a15_neon_vaba): Likewise.
3582         (cortex_a15_neon_vaba_qqq): Likewise.
3583         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3584         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3585         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3586         Likewise.
3587         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3588         (cortex_a15_neon_mla_qqq_8_16): Likewise.
3589         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar): Likewise.
3590         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3591         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3592         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
3593         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3594         (cortex_a15_neon_shift_1): Likewise.
3595         (cortex_a15_neon_shift_2): Likewise.
3596         (cortex_a15_neon_shift_3): Likewise.
3597         (cortex_a15_neon_vshl_ddd): Likewise.
3598         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3599         (cortex_a15_neon_vsra_vrsra): Likewise.
3600         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3601         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3602         (cortex_a15_neon_bp_3cycle): Likewise.
3603         (cortex_a15_neon_ldm_2): Likewise.
3604         (cortex_a15_neon_stm_2): Likewise.
3605         (cortex_a15_neon_mcr): Likewise.
3606         (cortex_a15_neon_mrc): Likewise.
3607         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
3608         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
3609         (cortex_a15_neon_fp_vmul_ddd): Likewise.
3610         (cortex_a15_neon_fp_vmul_qqd): Likewise.
3611         (cortex_a15_neon_fp_vmla_ddd): Likewise.
3612         (cortex_a15_neon_fp_vmla_qqq): Likewise.
3613         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
3614         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
3615         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3616         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3617         (cortex_a15_neon_bp_simple): Likewise.
3618         (cortex_a15_neon_bp_2cycle): Likewise.
3619         (cortex_a15_neon_bp_3cycle): Likewise.
3620         (cortex_a15_neon_vld1_1_2_regs): Likewise.
3621         (cortex_a15_neon_vld1_3_4_regs): Likewise.
3622         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3623         (cortex_a15_neon_vld2_4_regs): Likewise.
3624         (cortex_a15_neon_vld3_vld4): Likewise.
3625         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3626         (cortex_a15_neon_vst1_3_4_regs): Likewise.
3627         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Rename to...
3628         (cortex_a15_neon_vst2_4_regs_vst3): ...This, update for new attributes.
3629         (cortex_a15_neon_vst3_vst4): Rename to...
3630         (cortex_a15_neon_vst4): This, update for new attributes.
3631         (cortex_a15_neon_vld1_vld2_lane): Update for new attributes.
3632         (cortex_a15_neon_vld3_vld4_lane): Likewise.
3633         (cortex_a15_neon_vst1_vst2_lane): Likewise.
3634         (cortex_a15_neon_vst3_vst4_lane): Likewise.
3635         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
3636         (cortex_a15_neon_ldm_2): Likewise.
3637         (cortex_a15_neon_stm_2): Likewise.
3638         (cortex_a15_neon_mcr): Likewise.
3639         (cortex_a15_neon_mcr_2_mcrr): Likewise.
3640         (cortex_a15_neon_mrc): Likewise.
3641         (cortex_a15_neon_mrrc): Likewise.
3643         (cortex_a15_neon_abd): New.
3644         (cortex_a15_neon_abd_q): Likewise.
3645         (cortex_a15_neon_aba): Likewise.
3646         (cortex_a15_neon_aba_q): Likewise.
3647         (cortex_a15_neon_acc): Likewise.
3648         (cortex_a15_neon_acc_q): Likewise.
3649         (cortex_a15_neon_arith_basic): Likewise.
3650         (cortex_a15_neon_arith_complex): Likewise.
3651         (cortex_a15_neon_multiply): Likewise.
3652         (cortex_a15_neon_multiply_q): Likewise.
3653         (cortex_a15_neon_mla): Likewise.
3654         (cortex_a15_neon_mla_q): Likewise.
3655         (cortex_a15_neon_sat_mla_long): Likewise.
3656         (cortex_a15_neon_shift_acc): Likewise.
3657         (cortex_a15_neon_shift_imm_basic): Likewise.
3658         (cortex_a15_neon_shift_imm_complex): Likewise.
3659         (cortex_a15_neon_shift_reg_basic): Likewise.
3660         (cortex_a15_neon_shift_reg_basic_q): Likewise.
3661         (cortex_a15_neon_shift_reg_complex): Likewise.
3662         (cortex_a15_neon_shift_reg_complex_q): Likewise.
3663         (cortex_a15_neon_fp_negabs): Likewise
3664         (cortex_a15_neon_fp_arith): Likewise
3665         (cortex_a15_neon_fp_arith_q): Likewise
3666         (cortex_a15_neon_fp_cvt_int): Likewise
3667         (cortex_a15_neon_fp_cvt_int_q): Likewise
3668         (cortex_a15_neon_fp_cvt_16): Likewise
3669         (cortex_a15_neon_fp_mul): Likewise
3670         (cortex_a15_neon_fp_mul_q): Likewise
3671         (cortex_a15_neon_fp_mla): Likewise
3672         (cortex_a15_neon_fp_mla_q): Likewise
3673         (cortex_a15_neon_fp_recps_rsqrte): Likewise.
3674         (cortex_a15_neon_fp_recps_rsqrte_q): Likewise.
3675         (cortex_a15_neon_bitops): Likewise.
3676         (cortex_a15_neon_bitops_q): Likewise.
3677         (cortex_a15_neon_from_gp): Likewise.
3678         (cortex_a15_neon_from_gp_q): Likewise.
3679         (cortex_a15_neon_tbl3_tbl4): Likewise.
3680         (cortex_a15_neon_zip_q): Likewise.
3681         (cortex_a15_neon_to_gp): Likewise.
3682         (cortex_a15_neon_load_a): Likewise.
3683         (cortex_a15_neon_load_b): Likewise.
3684         (cortex_a15_neon_load_c): Likewise.
3685         (cortex_a15_neon_load_d): Likewise.
3686         (cortex_a15_neon_load_e): Likewise.
3687         (cortex_a15_neon_load_f): Likewise.
3688         (cortex_a15_neon_store_a): Likewise.
3689         (cortex_a15_neon_store_b): Likewise.
3690         (cortex_a15_neon_store_c): Likewise.
3691         (cortex_a15_neon_store_d): Likewise.
3692         (cortex_a15_neon_store_e): Likewise.
3693         (cortex_a15_neon_store_f): Likewise.
3694         (cortex_a15_neon_store_g): Likewise.
3695         (cortex_a15_neon_store_h): Likewise.
3696         (cortex_a15_vfp_to_from_gp): Likewise.
3698 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3700         * config/arm/cortex-a9-neon.md (cortex_a9_neon_type): New.
3702         (cortex_a9_neon_vshl_ddd): Remove.
3703         (cortex_a9_neon_vst3_vst4): Likewise.
3704         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3706         (cortex_a9_neon_bit_ops_q): New.
3708         (cortex_a9_neon_int_1): Use cortex_a8_neon_type.
3709         (cortex_a9_neon_int_2): Likewise.
3710         (cortex_a9_neon_int_3): Likewise.
3711         (cortex_a9_neon_int_4): Likewise.
3712         (cortex_a9_neon_int_5): Likewise.
3713         (cortex_a9_neon_vqneg_vqabs): Likewise.
3714         (cortex_a9_neon_vmov): Likewise.
3715         (cortex_a9_neon_vaba): Likewise.
3716         (cortex_a9_neon_vaba_qqq): Likewise.
3717         (cortex_a9_neon_shift_1): Likewise.
3718         (cortex_a9_neon_shift_2): Likewise.
3719         (cortex_a9_neon_shift_3): Likewise.
3720         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3721         (cortex_a9_neon_vsra_vrsra): Likewise.
3722         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3723         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3724         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3725         Likewise.
3726         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3727         (cortex_a9_neon_mla_qqq_8_16): Likewise.
3728         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3729         Likewise.
3730         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3731         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3732         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
3733         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3734         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
3735         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
3736         (cortex_a9_neon_fp_vsum): Likewise.
3737         (cortex_a9_neon_fp_vmul_ddd): Likewise.
3738         (cortex_a9_neon_fp_vmul_qqd): Likewise.
3739         (cortex_a9_neon_fp_vmla_ddd): Likewise.
3740         (cortex_a9_neon_fp_vmla_qqq): Likewise.
3741         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
3742         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
3743         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3744         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3745         (cortex_a9_neon_bp_simple): Likewise.
3746         (cortex_a9_neon_bp_2cycle): Likewise.
3747         (cortex_a9_neon_bp_3cycle): Likewise.
3748         (cortex_a9_neon_ldr): Likewise.
3749         (cortex_a9_neon_str): Likewise.
3750         (cortex_a9_neon_vld1_1_2_regs): Likewise.
3751         (cortex_a9_neon_vld1_3_4_regs): Likewise.
3752         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3753         (cortex_a9_neon_vld2_4_regs): Likewise.
3754         (cortex_a9_neon_vld3_vld4): Likewise.
3755         (cortex_a9_neon_vld1_vld2_lane): Likewise.
3756         (cortex_a9_neon_vld3_vld4_lane): Likewise.
3757         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
3758         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3759         (cortex_a9_neon_vst1_3_4_regs): Likewise.
3760         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
3761         (cortex_a9_neon_vst1_vst2_lane): Likewise.
3762         (cortex_a9_neon_vst3_vst4_lane): Likewise.
3763         (cortex_a9_neon_mcr): Likewise.
3764         (cortex_a9_neon_mcr_2_mcrr): Likewise.
3765         (cortex_a9_neon_mrc): Likewise.
3766         (cortex_a9_neon_mrrc): Likewise.
3768 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3770         * config/arm/cortex-a8-neon.md (cortex_a8_neon_type): New.
3772         (cortex_a8_neon_vshl_ddd): Remove.
3773         (cortex_a8_neon_vst3_vst4): Likewise.
3774         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
3776         (cortex_a8_neon_bit_ops_q): New.
3778         (cortex_a8_neon_int_1): Use cortex_a8_neon_type.
3779         (cortex_a8_neon_int_2): Likewise..
3780         (cortex_a8_neon_int_3): Likewise.
3781         (cortex_a8_neon_int_5): Likewise.
3782         (cortex_a8_neon_vqneg_vqabs): Likewise.
3783         (cortex_a8_neon_int_4): Likewise.
3784         (cortex_a8_neon_vaba): Likewise.
3785         (cortex_a8_neon_vaba_qqq): Likewise.
3786         (cortex_a8_neon_shift_1): Likewise.
3787         (cortex_a8_neon_shift_2): Likewise.
3788         (cortex_a8_neon_shift_3): Likewise.
3789         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
3790         (cortex_a8_neon_vsra_vrsra): Likewise.
3791         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3792         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
3793         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_long_scalar):
3794         Likewise.
3795         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
3796         (cortex_a8_neon_mla_qqq_8_16): Likewise.
3797         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_long_scalar_qdd_64_32_long):
3798         Likewise.
3799         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
3800         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
3801         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
3802         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
3803         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
3804         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
3805         (cortex_a8_neon_fp_vsum): Likewise.
3806         (cortex_a8_neon_fp_vmul_ddd): Likewise.
3807         (cortex_a8_neon_fp_vmul_qqd): Likewise.
3808         (cortex_a8_neon_fp_vmla_ddd): Likewise.
3809         (cortex_a8_neon_fp_vmla_qqq): Likewise.
3810         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
3811         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
3812         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
3813         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
3814         (cortex_a8_neon_bp_simple): Likewise.
3815         (cortex_a8_neon_bp_2cycle): Likewise.
3816         (cortex_a8_neon_bp_3cycle): Likewise.
3817         (cortex_a8_neon_ldr): Likewise.
3818         (cortex_a8_neon_str): Likewise.
3819         (cortex_a8_neon_vld1_1_2_regs): Likewise.
3820         (cortex_a8_neon_vld1_3_4_regs): Likewise.
3821         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
3822         (cortex_a8_neon_vld2_4_regs): Likewise.
3823         (cortex_a8_neon_vld3_vld4): Likewise.
3824         (cortex_a8_neon_vld1_vld2_lane): Likewise.
3825         (cortex_a8_neon_vld3_vld4_lane): Likewise.
3826         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
3827         (cortex_a8_neon_vst1_3_4_regs): Likewise.
3828         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
3829         (cortex_a8_neon_vst1_vst2_lane): Likewise.
3830         (cortex_a8_neon_vst3_vst4_lane): Likewise.
3831         (cortex_a8_neon_mcr): Likewise.
3832         (cortex_a8_neon_mcr_2_mcrr): Likewise.
3833         (cortex_a8_neon_mrc): Likewise.
3834         (cortex_a8_neon_mrrc): Likewise.
3836 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
3838         * config/aarch64/iterators.md (Vetype): Add SF and DF modes.
3839         (fp): New.
3840         * config/aarch64/aarch64-simd.md (neon_type): Remove.
3841         (aarch64_simd_dup<mode>): Add "type" attribute.
3842         (aarch64_dup_lane<mode>): Likewise.
3843         (aarch64_dup_lane_<vswap_width_name><mode>): Likewise.
3844         (*aarch64_simd_mov<mode>): Likewise.
3845         (aarch64_simd_mov_from_<mode>low): Likewise.
3846         (aarch64_simd_mov_from_<mode>high): Likewise.
3847         (orn<mode>3): Likewise.
3848         (bic<mode>3): Likewise.
3849         (add<mode>3): Likewise.
3850         (sub<mode>3): Likewise.
3851         (mul<mode>3): Likewise.
3852         (*aarch64_mul3_elt<mode>): Likewise.
3853         (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
3854         (*aarch64_mul3_elt_to_128df): Likewise.
3855         (*aarch64_mul3_elt_to_64v2df): Likewise.
3856         (neg<mode>2): Likewise.
3857         (abs<mode>2): Likewise.
3858         (abd<mode>_3): Likewise.
3859         (aba<mode>_3): Likewise.
3860         (fabd<mode>_3): Likewise.
3861         (*fabd_scalar<mode>3): Likewise.
3862         (and<mode>3): Likewise.
3863         (ior<mode>3): Likewise.
3864         (xor<mode>3): Likewise.
3865         (one_cmpl<mode>2): Likewise.
3866         (aarch64_simd_vec_set<mode>): Likewise.
3867         (aarch64_simd_lshr<mode>): Likewise.
3868         (aarch64_simd_ashr<mode>): Likewise.
3869         (aarch64_simd_imm_shl<mode>): Likewise.
3870         (aarch64_simd_reg_sshl<mode): Likewise.
3871         (aarch64_simd_reg_shl<mode>_unsigned): Likewise.
3872         (aarch64_simd_reg_shl<mode>_signed): Likewise.
3873         (aarch64_simd_vec_setv2di): Likewise.
3874         (aarch64_simd_vec_set<mode>): Likewise.
3875         (aarch64_mla<mode>): Likewise.
3876         (*aarch64_mla_elt<mode>): Likewise.
3877         (*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
3878         (aarch64_mls<mode>): Likewise.
3879         (*aarch64_mls_elt<mode>): Likewise.
3880         (*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
3881         (<su><maxmin><mode>3): Likewise.
3882         (move_lo_quad_<mode>): Likewise.
3883         (aarch64_simd_move_hi_quad_<mode>): Likewise.
3884         (aarch64_simd_vec_pack_trunc_<mode>): Likewise.
3885         (vec_pack_trunc_<mode>): Likewise.
3886         (aarch64_simd_vec_unpack<su>_lo_<mode>): Likewise.
3887         (aarch64_simd_vec_unpack<su>_hi_<mode>): Likewise.
3888         (*aarch64_<su>mlal_lo<mode>): Likewise.
3889         (*aarch64_<su>mlal_hi<mode>): Likewise.
3890         (*aarch64_<su>mlsl_lo<mode>): Likewise.
3891         (*aarch64_<su>mlsl_hi<mode>): Likewise.
3892         (*aarch64_<su>mlal<mode>): Likewise.
3893         (*aarch64_<su>mlsl<mode>): Likewise.
3894         (aarch64_simd_vec_<su>mult_lo_<mode>): Likewise.
3895         (aarch64_simd_vec_<su>mult_hi_<mode>): Likewise.
3896         (add<mode>3): Likewise.
3897         (sub<mode>3): Likewise.
3898         (mul<mode>3): Likewise.
3899         (div<mode>3): Likewise.
3900         (neg<mode>2): Likewise.
3901         (abs<mode>2): Likewise.
3902         (fma<mode>4): Likewise.
3903         (*aarch64_fma4_elt<mode>): Likewise.
3904         (*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
3905         (*aarch64_fma4_elt_to_128df): Likewise.
3906         (*aarch64_fma4_elt_to_64v2df): Likewise.
3907         (fnma<mode>4): Likewise.
3908         (*aarch64_fnma4_elt<mode>): Likewise.
3909         (*aarch64_fnma4_elt_<vswap_width_name><mode>
3910         (*aarch64_fnma4_elt_to_128df): Likewise.
3911         (*aarch64_fnma4_elt_to_64v2df): Likewise.
3912         (<frint_pattern><mode>2): Likewise.
3913         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): Likewise.
3914         (<optab><fcvt_target><VDQF:VDQF:mode>2): Likewise.
3915         (vec_unpacks_lo_v4sf): Likewise.
3916         (aarch64_float_extend_lo_v2df): Likewise.
3917         (vec_unpacks_hi_v4sf): Likewise.
3918         (aarch64_float_truncate_lo_v2sf): Likewise.
3919         (aarch64_float_truncate_hi_v4sf): Likewise.
3920         (aarch64_vmls<mode>): Likewise.
3921         (<su><maxmin><mode>3): Likewise.
3922         (<maxmin_uns><mode>3): Likewise.
3923         (reduc_<sur>plus_<mode>): Likewise.
3924         (reduc_<sur>plus_v2di): Likewise.
3925         (reduc_<sur>plus_v2si): Likewise.
3926         (reduc_<sur>plus_<mode>): Likewise.
3927         (aarch64_addpv4sf): Likewise.
3928         (clz<mode>2): Likewise.
3929         (reduc_<maxmin_uns>_<mode>): Likewise.
3930         (reduc_<maxmin_uns>_v2di): Likewise.
3931         (reduc_<maxmin_uns>_v2si): Likewise.
3932         (reduc_<maxmin_uns>_<mode>): Likewise.
3933         (reduc_<maxmin_uns>_v4sf): Likewise.
3934         (aarch64_simd_bsl<mode>_internal): Likewise.
3935         (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
3936         (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
3937         (aarch64_get_lane<mode>): Likewise.
3938         (*aarch64_combinez<mode>): Likewise.
3939         (aarch64_combine<mode>): Likewise.
3940         (aarch64_simd_combine<mode>): Likewise.
3941         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): Likewise.
3942         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): Likewise.
3943         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>): Likewise.
3944         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Likewise.
3945         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Likewise.
3946         (aarch64_<sur>h<addsub><mode>): Likewise.
3947         (aarch64_<sur><addsub>hn<mode>): Likewise.
3948         (aarch64_<sur><addsub>hn2<mode>): Likewise.
3949         (aarch64_pmul<mode>): Likewise.
3950         (aarch64_<su_optab><optab><mode>): Likewise.
3951         (aarch64_<sur>qadd<mode>): Likewise.
3952         (aarch64_sqmovun<mode>): Likewise.
3953         (aarch64_<sur>qmovn<mode>): Likewise.
3954         (aarch64_s<optab><mode>): Likewise.
3955         (aarch64_sq<r>dmulh<mode>): Likewise.
3956         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3957         (aarch64_sq<r>dmulh_laneq<mode>): Likewise.
3958         (aarch64_sq<r>dmulh_lane<mode>): Likewise.
3959         (aarch64_sqdml<SBINQOPS:as>l<mode>): Likewise.
3960         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3961         (aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
3962         (aarch64_sqdml<SBINQOPS:as>l_n<mode>): Likewise.
3963         (aarch64_sqdml<SBINQOPS:as>l2<mode>_internal): Likewise.
3964         (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
3965         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
3966         (aarch64_sqdmull<mode>): Likewise.
3967         (aarch64_sqdmull_lane<mode>_internal): Likewise.
3968         (aarch64_sqdmull_n<mode>): Likewise.
3969         (aarch64_sqdmull2<mode>_internal): Likewise.
3970         (aarch64_sqdmull2_lane<mode>_internal): Likewise.
3971         (aarch64_sqdmull2_n<mode>_internal): Likewise.
3972         (aarch64_<sur>shl<mode>): Likewise.
3973         (aarch64_<sur>q<r>shl<mode>
3974         (aarch64_<sur>shll_n<mode>): Likewise.
3975         (aarch64_<sur>shll2_n<mode>): Likewise.
3976         (aarch64_<sur>shr_n<mode>): Likewise.
3977         (aarch64_<sur>sra_n<mode>): Likewise.
3978         (aarch64_<sur>s<lr>i_n<mode>): Likewise.
3979         (aarch64_<sur>qshl<u>_n<mode>): Likewise.
3980         (aarch64_<sur>q<r>shr<u>n_n<mode>): Likewise.
3981         (aarch64_cm<optab><mode>): Likewise.
3982         (aarch64_cm<optab>di): Likewise.
3983         (aarch64_cm<optab><mode>): Likewise.
3984         (aarch64_cm<optab>di): Likewise.
3985         (aarch64_cmtst<mode>): Likewise.
3986         (aarch64_cmtstdi): Likewise.
3987         (aarch64_cm<optab><mode>): Likewise.
3988         (*aarch64_fac<optab><mode>): Likewise.
3989         (aarch64_addp<mode>): Likewise.
3990         (aarch64_addpdi): Likewise.
3991         (sqrt<mode>2): Likewise.
3992         (vec_load_lanesoi<mode>): Likewise.
3993         (vec_store_lanesoi<mode>): Likewise.
3994         (vec_load_lanesci<mode>): Likewise.
3995         (vec_store_lanesci<mode>): Likewise.
3996         (vec_load_lanesxi<mode>): Likewise.
3997         (vec_store_lanesxi<mode>): Likewise.
3998         (*aarch64_mov<mode>): Likewise.
3999         (aarch64_ld2<mode>_dreg): Likewise.
4000         (aarch64_ld2<mode>_dreg): Likewise.
4001         (aarch64_ld3<mode>_dreg): Likewise.
4002         (aarch64_ld3<mode>_dreg): Likewise.
4003         (aarch64_ld4<mode>_dreg): Likewise.
4004         (aarch64_ld4<mode>_dreg): Likewise.
4005         (aarch64_tbl1<mode>): Likewise.
4006         (aarch64_tbl2v16qi): Likewise.
4007         (aarch64_combinev16qi): Likewise.
4008         (aarch64_<PERMUTE:perm_insn><PERMUTE:perm_hilo><mode>): Likewise.
4009         (aarch64_st2<mode>_dreg): Likewise.
4010         (aarch64_st2<mode>_dreg): Likewise.
4011         (aarch64_st3<mode>_dreg): Likewise.
4012         (aarch64_st3<mode>_dreg): Likewise.
4013         (aarch64_st4<mode>_dreg): Likewise.
4014         (aarch64_st4<mode>_dreg): Likewise.
4015         (*aarch64_simd_ld1r<mode>): Likewise.
4016         (aarch64_frecpe<mode>): Likewise.
4017         (aarch64_frecp<FRECP:frecp_suffix><mode>): Likewise.
4018         (aarch64_frecps<mode>): Likewise.
4020 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4022         * config/arm/iterators.md (V_elem_ch): New.
4023         (q): Likewise.
4024         (VQH_type): Likewise.
4025         * config/arm/arm.md (is_neon_type): New.
4026         (conds): Use is_neon_type.
4027         (anddi3_insn): Update type attribute.
4028         (xordi3_insn): Likewise.
4029         (one_cmpldi2): Likewise.
4030         * config/arm/vfp.md (movhf_vfp_neon): Update type attribute.
4031         * config/arm/neon.md (neon_mov): Update type attribute.
4032         (*movmisalign<mode>_neon_store): Likewise.
4033         (*movmisalign<mode>_neon_load): Likewise.
4034         (vec_set<mode>_internal): Likewise.
4035         (vec_set<mode>_internal): Likewise.
4036         (vec_setv2di_internal): Likewise.
4037         (vec_extract<mode>): Likewise.
4038         (vec_extract<mode>): Likewise.
4039         (vec_extractv2di): Likewise.
4040         (*add<mode>3_neon): Likewise.
4041         (adddi3_neon): Likewise.
4042         (*sub<mode>3_neon): Likewise.
4043         (subdi3_neon): Likewise.
4044         (fma<VCVTF:mode>4): Likewise.
4045         (fma<VCVTF:mode>4_intrinsic): Likewise.
4046         (*fmsub<VCVTF:mode>4): Likewise.
4047         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
4048         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
4049         (ior<mode>3): Likewise.
4050         (and<mode>3): Likewise.
4051         (orn<mode>3_neon): Likewise.
4052         (orndi3_neon): Likewise.
4053         (bic<mode>3_neon): Likewise.
4054         (bicdi3_neon): Likewise.
4055         (xor<mode>3): Likewise.
4056         (one_cmpl<mode>2): Likewise.
4057         (abs<mode>2): Likewise.
4058         (neg<mode>2): Likewise.
4059         (negdi2_neon): Likewise.
4060         (*umin<mode>3_neon): Likewise.
4061         (*umax<mode>3_neon): Likewise.
4062         (*smin<mode>3_neon): Likewise.
4063         (*smax<mode>3_neon): Likewise.
4064         (vashl<mode>3): Likewise.
4065         (vashr<mode>3_imm): Likewise.
4066         (vlshr<mode>3_imm): Likewise.
4067         (ashl<mode>3_signed): Likewise.
4068         (ashl<mode>3_unsigned): Likewise.
4069         (neon_load_count): Likewise.
4070         (ashldi3_neon_noclobber): Likewise.
4071         (ashldi3_neon): Likewise.
4072         (signed_shift_di3_neon): Likewise.
4073         (unsigned_shift_di3_neon): Likewise.
4074         (ashrdi3_neon_imm_noclobber): Likewise.
4075         (lshrdi3_neon_imm_noclobber): Likewise.
4076         (<shift>di3_neon): Likewise.
4077         (widen_ssum<mode>3): Likewise.
4078         (widen_usum<mode>3): Likewise.
4079         (quad_halves_<code>v4si): Likewise.
4080         (quad_halves_<code>v4sf): Likewise.
4081         (quad_halves_<code>v8hi): Likewise.
4082         (quad_halves_<code>v16qi): Likewise.
4083         (reduc_splus_v2di): Likewise.
4084         (neon_vpadd_internal<mode>): Likewise.
4085         (neon_vpsmin<mode>): Likewise.
4086         (neon_vpsmax<mode>): Likewise.
4087         (neon_vpumin<mode>): Likewise.
4088         (neon_vpumax<mode>): Likewise.
4089         (*ss_add<mode>_neon): Likewise.
4090         (*us_add<mode>_neon): Likewise.
4091         (*ss_sub<mode>_neon): Likewise.
4092         (*us_sub<mode>_neon): Likewise.
4093         (neon_vadd<mode>_unspec): Likewise.
4094         (neon_vaddl<mode>): Likewise.
4095         (neon_vaddw<mode>): Likewise.
4096         (neon_vhadd<mode>): Likewise.
4097         (neon_vqadd<mode>): Likewise.
4098         (neon_vaddhn<mode>): Likewise.
4099         (neon_vmul<mode>): Likewise.
4100         (neon_vfms<VCVTF:mode>): Likewise.
4101         (neon_vmlal<mode>): Likewise.
4102         (neon_vmls<mode>): Likewise.
4103         (neon_vmlsl<mode>): Likewise.
4104         (neon_vqdmulh<mode>): Likewise.
4105         (neon_vqdmlal<mode>): Likewise.
4106         (neon_vqdmlsl<mode>): Likewise.
4107         (neon_vmull<mode>): Likewise.
4108         (neon_vqdmull<mode>): Likewise.
4109         (neon_vsub<mode>_unspec): Likewise.
4110         (neon_vsubl<mode>): Likewise.
4111         (neon_vsubw<mode>): Likewise.
4112         (neon_vqsub<mode>): Likewise.
4113         (neon_vhsub<mode>): Likewise.
4114         (neon_vsubhn<mode>): Likewise.
4115         (neon_vceq<mode>): Likewise.
4116         (neon_vcge<mode>): Likewise.
4117         (neon_vcgeu<mode>): Likewise.
4118         (neon_vcgt<mode>): Likewise.
4119         (neon_vcgtu<mode>): Likewise.
4120         (neon_vcle<mode>): Likewise.
4121         (neon_vclt<mode>): Likewise.
4122         (neon_vcage<mode>): Likewise.
4123         (neon_vcagt<mode>): Likewise.
4124         (neon_vtst<mode>): Likewise.
4125         (neon_vabd<mode>): Likewise.
4126         (neon_vabdl<mode>): Likewise.
4127         (neon_vaba<mode>): Likewise.
4128         (neon_vabal<mode>): Likewise.
4129         (neon_vmax<mode>): Likewise.
4130         (neon_vmin<mode>): Likewise.
4131         (neon_vpaddl<mode>): Likewise.
4132         (neon_vpadal<mode>): Likewise.
4133         (neon_vpmax<mode>): Likewise.
4134         (neon_vpmin<mode>): Likewise.
4135         (neon_vrecps<mode>): Likewise.
4136         (neon_vrsqrts<mode>): Likewise.
4137         (neon_vqabs<mode>): Likewise.
4138         (neon_vqneg<mode>): Likewise.
4139         (neon_vcls<mode>): Likewise.
4140         (clz<mode>2): Likewise.
4141         (popcount<mode>2): Likewise.
4142         (neon_vrecpe<mode>): Likewise.
4143         (neon_vrsqrte<mode>): Likewise.
4144         (neon_vget_lane<mode>_sext_internal): Likewise.
4145         (neon_vget_lane<mode>_zext_internal): Likewise.
4146         (neon_vdup_n<mode>): Likewise.
4147         (neon_vdup_n<mode>): Likewise.
4148         (neon_vdup_nv2di): Likewise.
4149         (neon_vdup_lane<mode>_interal): Likewise.
4150         (*neon_vswp<mode>): Likewise.
4151         (neon_vcombine<mode>): Likewise.
4152         (float<mode><V_cvtto>2): Likewise.
4153         (floatuns<mode><V_cvtto>2): Likewise.
4154         (fix_trunc<mode><V_cvtto>2): Likewise.
4155         (fixuns_trunc<mode><V_cvtto>2
4156         (neon_vcvt<mode>): Likewise.
4157         (neon_vcvt<mode>): Likewise.
4158         (neon_vcvtv4sfv4hf): Likewise.
4159         (neon_vcvtv4hfv4sf): Likewise.
4160         (neon_vcvt_n<mode>): Likewise.
4161         (neon_vcvt_n<mode>): Likewise.
4162         (neon_vmovn<mode>): Likewise.
4163         (neon_vqmovn<mode>): Likewise.
4164         (neon_vqmovun<mode>): Likewise.
4165         (neon_vmovl<mode>): Likewise.
4166         (neon_vmul_lane<mode>): Likewise.
4167         (neon_vmul_lane<mode>): Likewise.
4168         (neon_vmull_lane<mode>): Likewise.
4169         (neon_vqdmull_lane<mode>): Likewise.
4170         (neon_vqdmulh_lane<mode>): Likewise.
4171         (neon_vqdmulh_lane<mode>): Likewise.
4172         (neon_vmla_lane<mode>): Likewise.
4173         (neon_vmla_lane<mode>): Likewise.
4174         (neon_vmlal_lane<mode>): Likewise.
4175         (neon_vqdmlal_lane<mode>): Likewise.
4176         (neon_vmls_lane<mode>): Likewise.
4177         (neon_vmls_lane<mode>): Likewise.
4178         (neon_vmlsl_lane<mode>): Likewise.
4179         (neon_vqdmlsl_lane<mode>): Likewise.
4180         (neon_vext<mode>): Likewise.
4181         (neon_vrev64<mode>): Likewise.
4182         (neon_vrev32<mode>): Likewise.
4183         (neon_vrev16<mode>): Likewise.
4184         (neon_vbsl<mode>_internal): Likewise.
4185         (neon_vshl<mode>): Likewise.
4186         (neon_vqshl<mode>): Likewise.
4187         (neon_vshr_n<mode>): Likewise.
4188         (neon_vshrn_n<mode>): Likewise.
4189         (neon_vqshrn_n<mode>): Likewise.
4190         (neon_vqshrun_n<mode>): Likewise.
4191         (neon_vshl_n<mode>): Likewise.
4192         (neon_vqshl_n<mode>): Likewise.
4193         (neon_vqshlu_n<mode>): Likewise.
4194         (neon_vshll_n<mode>): Likewise.
4195         (neon_vsra_n<mode>): Likewise.
4196         (neon_vsri_n<mode>): Likewise.
4197         (neon_vsli_n<mode>): Likewise.
4198         (neon_vtbl1v8qi): Likewise.
4199         (neon_vtbl2v8qi): Likewise.
4200         (neon_vtbl3v8qi): Likewise.
4201         (neon_vtbl4v8qi): Likewise.
4202         (neon_vtbl1v16qi): Likewise.
4203         (neon_vtbl2v16qi): Likewise.
4204         (neon_vcombinev16qi): Likewise.
4205         (neon_vtbx1v8qi): Likewise.
4206         (neon_vtbx2v8qi): Likewise.
4207         (neon_vtbx3v8qi): Likewise.
4208         (neon_vtbx4v8qi): Likewise.
4209         (*neon_vtrn<mode>_insn): Likewise.
4210         (*neon_vzip<mode>_insn): Likewise.
4211         (*neon_vuzp<mode>_insn): Likewise.
4212         (neon_vld1<mode>): Likewise.
4213         (neon_vld1_lane<mode>): Likewise.
4214         (neon_vld1_lane<mode>): Likewise.
4215         (neon_vld1_dup<mode>): Likewise.
4216         (neon_vld1_dup<mode>): Likewise.
4217         (neon_vld1_dupv2di): Likewise.
4218         (neon_vst1<mode>): Likewise.
4219         (neon_vst1_lane<mode>): Likewise.
4220         (neon_vst1_lane<mode>): Likewise.
4221         (neon_vld2<mode>): Likewise.
4222         (neon_vld2<mode>): Likewise.
4223         (neon_vld2_lane<mode>): Likewise.
4224         (neon_vld2_lane<mode>): Likewise.
4225         (neon_vld2_dup<mode>): Likewise.
4226         (neon_vst2<mode>): Likewise.
4227         (neon_vst2<mode>): Likewise.
4228         (neon_vst2_lane<mode>): Likewise.
4229         (neon_vst2_lane<mode>): Likewise.
4230         (neon_vld3<mode>): Likewise.
4231         (neon_vld3qa<mode>): Likewise.
4232         (neon_vld3qb<mode>): Likewise.
4233         (neon_vld3_lane<mode>): Likewise.
4234         (neon_vld3_lane<mode>): Likewise.
4235         (neon_vld3_dup<mode>): Likewise.
4236         (neon_vst3<mode>): Likewise.
4237         (neon_vst3qa<mode>): Likewise.
4238         (neon_vst3qb<mode>): Likewise.
4239         (neon_vst3_lane<mode>): Likewise.
4240         (neon_vst3_lane<mode>): Likewise.
4241         (neon_vld4<mode>): Likewise.
4242         (neon_vld4qa<mode>): Likewise.
4243         (neon_vld4qb<mode>): Likewise.
4244         (neon_vld4_lane<mode>): Likewise.
4245         (neon_vld4_lane<mode>): Likewise.
4246         (neon_vld4_dup<mode>): Likewise.
4247         (neon_vst4<mode>): Likewise.
4248         (neon_vst4qa<mode>): Likewise.
4249         (neon_vst4qb<mode>): Likewise.
4250         (neon_vst4_lane<mode>): Likewise.
4251         (neon_vst4_lane<mode>): Likewise.
4252         (neon_vec_unpack<US>_lo_<mode>): Likewise.
4253         (neon_vec_unpack<US>_hi_<mode>): Likewise.
4254         (neon_vec_<US>mult_lo_<mode>): Likewise.
4255         (neon_vec_<US>mult_hi_<mode>): Likewise.
4256         (neon_vec_<US>shiftl_<mode>): Likewise.
4257         (neon_unpack<US>_<mode>): Likewise.
4258         (neon_vec_<US>mult_<mode>): Likewise.
4259         (vec_pack_trunc_<mode>): Likewise.
4260         (neon_vec_pack_trunc_<mode>): Likewise.
4261         (neon_vabd<mode>_2): Likewise.
4262         (neon_vabd<mode>_3): Likewise.
4264 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4266         * config/aarch64/aarch64.md (movtf_aarch64): Update type attribute.
4267         (load_pair): Update type attribute.
4268         (store_pair): Update type attribute.
4269         * config/aarch64/iterators.md (q): New.
4271 2013-10-15  James Greenhalgh  <james.greenhalgh@arm.com>
4273         * config/arm/types.md: Add new types for Neon insns.
4275 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4276             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4277             Sergey Lega  <sergey.s.lega@intel.com>
4278             Anna Tikhonova  <anna.tikhonova@intel.com>
4279             Ilya Tocar  <ilya.tocar@intel.com>
4280             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4281             Ilya Verbin  <ilya.verbin@intel.com>
4282             Kirill Yukhin  <kirill.yukhin@intel.com>
4283             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4285         * config/i386/sse.md (unspec): Add UNSPEC_RCP14, UNSPEC_RSQRT14,
4286         UNSPEC_FIXUPIMM, UNSPEC_SCALEF, UNSPEC_GETEXP, UNSPEC_GETMANT,
4287         UNSPEC_EXP2, UNSPEC_RCP28, UNSPEC_RSQRT28.
4288         (rcp14<mode>): New.
4289         (srcp14<mode>): Ditto.
4290         (rsqrt14<mode>): Ditto.
4291         (rsqrt14<mode>): Ditto.
4292         (avx512f_vmscalef<mode>): Ditto.
4293         (avx512f_scalef<mode>): Ditto.
4294         (avx512f_getexp<mode>): Ditto.
4295         (avx512f_sgetexp<mode>): Ditto.
4296         (avx512f_fixupimm<mode>): Ditto.
4297         (avx512f_sfixupimm<mode>): Ditto.
4298         (avx512f_rndscale<mode>): Ditto.
4299         (*avx512er_exp2<mode>): Ditto.
4300         (*avx512er_rcp28<mode>): Ditto.
4301         (avx512er_rsqrt28<mode>): Ditto.
4302         (avx512f_getmant<mode>): Ditto.
4303         (avx512f_getmant<mode>): Ditto.
4304         (avx512f_rndscale<mode>): Fix formatting.
4306 2013-10-15  Martin Jambor  <mjambor@suse.cz>
4308         * ipa-utils.h (ipa_edge_within_scc): Declare.
4309         * ipa-cp.c (edge_within_scc): Moved...
4310         * ipa-utils.c (ipa_edge_within_scc): ...here.  Updated all callers.
4312 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4313             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4314             Sergey Lega  <sergey.s.lega@intel.com>
4315             Anna Tikhonova  <anna.tikhonova@intel.com>
4316             Ilya Tocar  <ilya.tocar@intel.com>
4317             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4318             Ilya Verbin  <ilya.verbin@intel.com>
4319             Kirill Yukhin  <kirill.yukhin@intel.com>
4320             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4322         * config/i386/predicates.md (const_8_to_15_operand): New.
4323         (const_16_to_31_operand): Ditto.
4324         * config/i386/sse.md (V8FI): New.
4325         (V16FI): Ditto.
4326         (reduc_splus_v8df): Ditto.
4327         (reduc_splus_v16sf): Ditto.
4328         (avx512f_vextract<shuffletype>32x4_1): Ditto.
4329         (vec_extract_hi_<mode>): Ditto.
4330         (avx512f_vinsert<shuffletype>32x4_1): Ditto.
4331         (vec_set_lo_<mode>): Ditto.
4332         (vec_set_hi_<mode>): Ditto.
4333         (avx512f_shuf_<shuffletype>64x2_1): Ditto.
4334         (avx512f_shuf_<shuffletype>32x4_1): Ditto.
4335         (avx512f_pshufd_1): Ditto.
4336         (avx512f_broadcast<mode>): Ditto.
4337         (avx512f_broadcast<mode>): Ditto.
4338         (define_split): Split vec_extract_lo into move.
4339         (ssequartermode): Ditto.
4340         (ssedoublemode): Extened with wider modes.
4341         (vec_extract_lo_<mode>): Ditto.
4343 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4344             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4345             Sergey Lega  <sergey.s.lega@intel.com>
4346             Anna Tikhonova  <anna.tikhonova@intel.com>
4347             Ilya Tocar  <ilya.tocar@intel.com>
4348             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4349             Ilya Verbin  <ilya.verbin@intel.com>
4350             Kirill Yukhin  <kirill.yukhin@intel.com>
4351             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4353         * config/i386/predicates.md (register_or_constm1_operand): New.
4354         * config/i386/sse.md (unspec): Add UNSPEC_UNSIGNED_PCMP, UNSPEC_TESTM,
4355         UNSPEC_TESTNM, UNSPEC_VTERNLOG, UNSPEC_ALIGN, UNSPEC_CONFLICT,
4356         UNSPEC_MASKED_EQ, UNSPEC_MASKED_GT, UNSPEC_GATHER_PREFETCH,
4357         UNSPEC_SCATTER_PREFETCH
4358         (VI48_512): New.
4359         (avx512f_ucmp<mode>3): Ditto.
4360         (avx512f_vternlog<mode>): Ditto.
4361         (avx512f_align<mode>): Ditto.
4362         (<shift_insn><mode>3): Ditto.
4363         (avx512f_<rotate>v<mode>): Ditto.
4364         (avx512f_<rotate><mode>): Ditto.
4365         (avx512f_eq<mode>3): Ditto.
4366         (avx512f_eq<mode>3_1): Ditto.
4367         (avx512f_gt<mode>3): Ditto.
4368         (avx512f_testm<mode>3): Ditto.
4369         (avx512f_testnm<mode>3): Ditto.
4370         (avx512pf_gatherpf<mode>): Ditto.
4371         (*avx512pf_gatherpf<mode>_mask): Ditto.
4372         (*avx512pf_gatherpf<mode>): Ditto.
4373         (avx512pf_scatterpf<mode>): Ditto.
4374         (*avx512pf_scatterpf<mode>_mask): Ditto.
4375         (*avx512pf_scatterpf<mode>): Ditto.
4376         (avx512f_vec_dup_gpr<mode>): Ditto.
4377         (clz<mode>2): Ditto.
4378         (conflict<mode>): Ditto.
4379         (REDUC_SMINMAX_MODE): Extened with wider modes.
4380         (reduc_<code>_<mode>): Ditto.
4381         (vlshr<mode>3): Ditto.
4382         (vashl<mode>3): Ditto.
4384 2013-10-15  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4385             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4386             Sergey Lega  <sergey.s.lega@intel.com>
4387             Anna Tikhonova  <anna.tikhonova@intel.com>
4388             Ilya Tocar  <ilya.tocar@intel.com>
4389             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4390             Ilya Verbin  <ilya.verbin@intel.com>
4391             Kirill Yukhin  <kirill.yukhin@intel.com>
4392             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4394         * config/i386/sse.md (unspec): Added UNSPEC_VPERMI2, UNSPEC_VPERMT2,
4395         UNSPEC_SCATTER.
4396         (VI48F_512): New.
4397         (avx512fmaskmode): Ditto.
4398         (bcstscalarsuff): Ditto.
4399         (avx512f_blendm<mode>): Ditto.
4400         (cmp_imm_predicate): Ditto.
4401         (avx512f_cmp<mode>3): Ditto.
4402         (avx512f_vec_dup<mode>): Ditto.
4403         (avx512f_vec_dup_mem<mode>): Ditto.
4404         (avx512f_vpermi2var<mode>3): Ditto.
4405         (avx512f_vpermt2var<mode>3): Ditto.
4406         (vec_init<mode>): Ditto.
4407         (avx512f_gathersi<mode>): Ditto.
4408         (*avx512f_gathersi<mode>): Ditto.
4409         (*avx512f_gathersi<mode>_2): Ditto.
4410         (avx512f_gatherdi<mode>): Ditto.
4411         (*avx512f_gatherdi<mode>): Ditto.
4412         (*avx512f_gatherdi<mode>_2): Ditto.
4413         (avx512f_scattersi<mode>): Ditto.
4414         (*avx512f_scattersi<mode>): Ditto.
4415         (avx512f_scatterdi<mode>): Ditto.
4416         (*avx512f_scatterdi<mode>): Ditto.
4417         (sseintprefix): Extened with wider modes.
4418         (VEC_GATHER_IDXSI): Ditto.
4419         (VEC_GATHER_IDXDI): Ditto.
4420         (VEC_GATHER_SRCDI): Ditto.
4422 2013-10-15  Matthew Gretton-Dann  <matthew.gretton-dann@arm.com>
4423             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
4425         * config/arm/t-aprofile: New file.
4426         * config.gcc: Handle --with-multilib-list option.
4428 2013-10-15  Bernd Schmidt  <bernds@codesourcery.com>
4430         * reload1.c (reloads_unique_chain_p): Ensure that r1 is
4431         the input for r2.
4433 2013-10-15  Richard Biener  <rguenther@suse.de>
4435         * tree-loop-distribution.c (build_empty_rdg): Inline into
4436         single user.
4437         (rdg_flag_vertex): Inline into single user.
4438         (rdg_flag_vertex_and_dependent): Likewise.
4439         (build_rdg_partition_for_vertex): Remove processed bitmap.
4440         (rdg_build_partitions): Simplify.
4442 2013-10-15  Richard Biener  <rguenther@suse.de>
4444         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1):
4445         Restructure forwarding through conversions and copies to
4446         avoid performing copy-propagation the wrong way.  Adjust
4447         recursion invocations.
4448         (forward_propagate_addr_expr): Add argument stating if we
4449         are recursing from a single-use.
4450         (ssa_forward_propagate_and_combine): Adjust.
4452 2013-10-14  David Malcolm  <dmalcolm@redhat.com>
4454         * dumpfile.h (gcc::dump_manager): New class, to hold state
4455         relating to dumpfile management.
4456         (get_dump_file_name): Remove in favor of method of dump_manager.
4457         (dump_initialized_p): Likewise.
4458         (dump_start): Likewise.
4459         (dump_finish): Likewise.
4460         (dump_switch_p): Likewise.
4461         (dump_register): Likewise.
4462         (get_dump_file_info): Likewise.
4463         * context.c (gcc::context::context): Construct the dump_manager
4464         instance.
4465         * context.h (gcc::context::get_dumps): New.
4466         (gcc::context::m_dumps): New.
4467         * coverage.c (coverage_init): Port to dump_manager API.
4468         * dumpfile.c (extra_dump_files): Convert to field of gcc::dump_manager.
4469         (extra_dump_files_in_use): Likewise.
4470         (extra_dump_files_alloced): Likewise.
4471         (gcc::dump_manager::dump_manager): New.
4472         (dump_register): Convert to...
4473         (gcc::dump_manager::dump_register): ...method, replacing
4474         function-static next_dump with m_next_dump field.
4475         (get_dump_file_info): Convert to...
4476         (gcc::dump_manager::get_dump_file_info): ...method.
4477         (get_dump_file_name): Convert to...
4478         (gcc::dump_manager::get_dump_file_name): ...method.
4479         (dump_start): Convert to...
4480         (gcc::dump_manager::dump_start): ...method.
4481         (dump_finish): Convert to...
4482         (gcc::dump_manager::dump_finish): ...method.
4483         (dump_begin): Replace body with...
4484         (gcc::dump_manager::dump_begin): ...new method.
4485         (dump_phase_enabled_p): Convert to...
4486         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4487         (dump_phase_enabled_p): Convert to...
4488         (gcc::dump_manager::dump_phase_enabled_p): ...method.
4489         (dump_initialized_p):  Convert to...
4490         (gcc::dump_manager::dump_initialized_p): ...method.
4491         (dump_flag_name): Replace body with...
4492         (gcc::dump_manager::dump_flag_name): ...new method.
4493         (dump_enable_all): Convert to...
4494         (gcc::dump_manager::dump_enable_all): ...new method.
4495         (opt_info_enable_passes): Convert to...
4496         (gcc::dump_manager::opt_info_enable_passes): ...new method.
4497         (dump_switch_p_1): Convert to...
4498         (gcc::dump_manager::dump_switch_p_1): ...new method.
4499         (dump_switch_p):  Convert to...
4500         (gcc::dump_manager::dump_switch_p): ...new method.
4501         (opt_info_switch_p): Port to dump_manager API.
4502         (enable_rtl_dump_file): Likewise.
4503         * opts-global.c (handle_common_deferred_options): Port to new
4504         dump_manager API.
4505         * passes.c (pass_manager::finish_optimization_passes): Likewise.
4506         (pass_manager::register_one_dump_file): Likewise.
4507         (pass_manager::register_pass): Likewise.
4508         (pass_init_dump_file): Likewise.
4509         (pass_fini_dump_file): Likewise.
4510         * statistics.c (statistics_early_init): Likewise.
4512 2013-10-14  Richard Biener  <rguenther@suse.de>
4514         * gimple.c (gimple_canonical_types, canonical_type_hash_cache,
4515         iterative_hash_canonical_type, gimple_canonical_type_hash,
4516         gimple_canonical_types_compatible_p, gimple_canonical_type_eq,
4517         gimple_register_canonical_type, print_gimple_types_stats,
4518         free_gimple_type_tables): Move to lto/lto.c
4519         (gt-gimple.h): Do not include.
4520         * gimple.h (gimple_register_canonical_type,
4521         print_gimple_types_stats, free_gimple_type_tables): Remove.
4522         * Makefile.in (GTFILES): Remove gimple.c.
4524 2013-10-14  Travis Snoozy  <quandary@remstate.com>
4526         PR target/58716
4527         * config/msp430/msp430.c (msp430_option_override): Correct thinko
4528         scanning for msp430x targets.
4530 2013-10-14  Eric Botcazou  <ebotcazou@adacore.com>
4532         PR bootstrap/58509
4533         * config/sparc/sparc-protos.h (widen_mem_for_ldd_peep): Declare.
4534         (registers_ok_for_ldd_peep): Move around.
4535         * config/sparc/sparc.c (widen_mem_for_ldd_peep): New.
4536         * config/sparc/sparc.md (widening peepholes): Use it.
4538 2013-10-14  Richard Biener  <rguenther@suse.de>
4540         PR middle-end/58712
4541         PR middle-end/55358
4542         * gimple.c (iterative_hash_canonical_type): Make sure to
4543         record the hash into the correct hashtable slot.
4545 2013-10-13  Eric Botcazou  <ebotcazou@adacore.com>
4547         PR rtl-optimization/58662
4548         * combine.c (try_combine): Take into account death nodes on I2 when
4549         splitting a PARALLEL of two independent SETs.  Fix dump message.
4551 2013-10-12  Oleg Endo  <olegendo@gcc.gnu.org>
4553         PR target/51244
4554         * config/sh/sh_treg_combine.cc: New SH specific RTL pass.
4555         * config.gcc (SH extra_objs): Add sh_ifcvt.o.
4556         * config/sh/t-sh (sh_treg_combine.o): New entry.
4557         * config/sh/sh.c (sh_fixed_condition_code_regs): New function that
4558         implements the target hook TARGET_FIXED_CONDITION_CODE_REGS.
4559         (register_sh_passes): New function.  Register sh_treg_combine pass.
4560         (sh_option_override): Invoke it.
4561         (sh_canonicalize_comparison): Handle op0_preserve_value.
4562         * sh.md (*cbranch_t"): Do not try to optimize missed test and branch
4563         opportunities.  Canonicalize branch condition.
4564         (nott): Allow only if pseudos can be created for non-SH2A.
4566 2013-10-12  H.J. Lu  <hongjiu.lu@intel.com>
4568         PR target/58690
4569         * config/i386/i386.c (ix86_copy_addr_to_reg): New function.
4570         (ix86_expand_movmem): Replace copy_addr_to_reg with
4571         ix86_copy_addr_to_reg.
4572         (ix86_expand_setmem): Likewise.
4574 2013-10-12  Alexander Monakov  <amonakov@ispras.ru>
4576         * config/i386/i386.c (ix86_expand_sse_compare_and_jump): Use mode
4577         provided by ix86_fp_compare_mode instead of CCFPUmode.
4579 2013-10-12  James Greenhalgh  <james.greenhalgh@arm.com>
4581         * config/aarch64/arm_neon.h
4582         (vtbx<1,3>_<psu>8): Fix register constriants.
4584 2013-10-11  Jeff Law  <law@redhat.com>
4586         PR tree-optimization/58640
4587         * tree-ssa-threadupdate.c (mark_threaded_blocks): Truncate jump
4588         threading paths that cross over two loop entry points.
4590 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4592         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): Generalize to
4593         handle vector float as well.
4594         (*vsx_le_perm_load_v4si): Likewise.
4595         (*vsx_le_perm_store_v2di): Likewise.
4596         (*vsx_le_perm_store_v4si): Likewise.
4598 2013-10-11  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
4600         * config/rs6000/vector.md (vec_realign_load<mode>): Generate vperm
4601         directly to circumvent subtract from splat{31} workaround.
4602         * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_le): New
4603         prototype.
4604         * config/rs6000/rs6000.c (altivec_expand_vec_perm_le): New.
4605         * config/rs6000/altivec.md (define_c_enum "unspec"): Add
4606         UNSPEC_VPERM_X and UNSPEC_VPERM_UNS_X.
4607         (altivec_vperm_<mode>): Convert to define_insn_and_split to
4608         separate big and little endian logic.
4609         (*altivec_vperm_<mode>_internal): New define_insn.
4610         (altivec_vperm_<mode>_uns): Convert to define_insn_and_split to
4611         separate big and little endian logic.
4612         (*altivec_vperm_<mode>_uns_internal): New define_insn.
4613         (vec_permv16qi): Add little endian logic.
4615 2013-10-11  Marc Glisse  <marc.glisse@inria.fr>
4617         * doc/extend.texi (returns_nonnull): Remove arguments.
4619 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4620             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4621             Sergey Lega  <sergey.s.lega@intel.com>
4622             Anna Tikhonova  <anna.tikhonova@intel.com>
4623             Ilya Tocar  <ilya.tocar@intel.com>
4624             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4625             Ilya Verbin  <ilya.verbin@intel.com>
4626             Kirill Yukhin  <kirill.yukhin@intel.com>
4627             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4629         * config/i386/sse.md (VI48F_256_512): New.
4630         (avx2_permvar<mode>): Change to ...
4631         (<avx2_avx512f>_permvar<mode>): This.
4633 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4634             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4635             Sergey Lega  <sergey.s.lega@intel.com>
4636             Anna Tikhonova  <anna.tikhonova@intel.com>
4637             Ilya Tocar  <ilya.tocar@intel.com>
4638             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4639             Ilya Verbin  <ilya.verbin@intel.com>
4640             Kirill Yukhin  <kirill.yukhin@intel.com>
4641             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4643         * config/i386/i386.c (bdesc_args): Change corresponding pattern for
4644         __builtin_ia32_cvtps2dq, __builtin_ia32_cvtps2dq256.
4645         * config/i386/sse.md (VI4_AVX): New.
4646         (sf2simodelower): Ditto.
4647         (sse2_cvtps2dq): Change to ...
4648         (<sse2_avx_avx512f>_fix_notrunc<sf2simodelower><mode>): This.
4650 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4651             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4652             Sergey Lega  <sergey.s.lega@intel.com>
4653             Anna Tikhonova  <anna.tikhonova@intel.com>
4654             Ilya Tocar  <ilya.tocar@intel.com>
4655             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4656             Ilya Verbin  <ilya.verbin@intel.com>
4657             Kirill Yukhin  <kirill.yukhin@intel.com>
4658             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4660         * config/i386/sse.md (V_512): New.
4661         (VI_512): Ditto.
4662         (vcond<V_512:mode><VF_512:mode>): Ditto.
4663         (vcond<V_512:mode><VI_512:mode>): Ditto.
4664         (vcondu<V_512:mode><VI_512:mode>): Ditto.
4666 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4667             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4668             Sergey Lega  <sergey.s.lega@intel.com>
4669             Anna Tikhonova  <anna.tikhonova@intel.com>
4670             Ilya Tocar  <ilya.tocar@intel.com>
4671             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4672             Ilya Verbin  <ilya.verbin@intel.com>
4673             Kirill Yukhin  <kirill.yukhin@intel.com>
4674             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4676         * config/i386/i386.c (ix86_rtx_costs): Enable fma for TARGET_AVX512F.
4677         * config/i386/sse.md (FMAMODEM): Changed modes and conditions.
4678         (FMAMODE): Ditto.
4679         (fma<mode>4): Removed condition.
4680         (fms<mode>4): Ditto.
4681         (fnma<mode>4): Ditto.
4682         (fnms<mode>4): Ditto.
4683         (fma4i_fmadd_<mode>): Ditto.
4684         (*fma_fmadd_<mode>): Ditto.
4685         (*fma_fmsub_<mode>): Ditto.
4686         (*fma_fnmadd_<mode>): Ditto.
4687         (*fma_fnmsub_<mode>): Ditto.
4688         (fmaddsub_<mode>): Allow for TARGET_AVX512F.
4689         (*fma_fmaddsub_<mode>): Ditto.
4690         (*fma_fmsubadd_<mode>): Ditto.
4691         (*fmai_fmadd_<mode>): Ditto.
4692         (*fmai_fmsub_<mode>): Ditto.
4693         (*fmai_fnmadd_<mode>): Ditto.
4694         (*fmai_fnmsub_<mode>): Ditto.
4696 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4697             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4698             Sergey Lega  <sergey.s.lega@intel.com>
4699             Anna Tikhonova  <anna.tikhonova@intel.com>
4700             Ilya Tocar  <ilya.tocar@intel.com>
4701             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4702             Ilya Verbin  <ilya.verbin@intel.com>
4703             Kirill Yukhin  <kirill.yukhin@intel.com>
4704             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4706         * config/i386/sse.md (VI248_AVX2_8_AVX512F): New.
4707         (VI124_256): Changed to ...
4708         (VI124_256_48_512): This.
4709         (ssepackmode): Extended with wider modes.
4710         (<code><mode>3): Changed iterator.
4711         (*avx2_<code><mode>3): Ditto.
4712         (vec_pack_trunc_<mode>): Ditto.
4714 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4715             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4716             Sergey Lega  <sergey.s.lega@intel.com>
4717             Anna Tikhonova  <anna.tikhonova@intel.com>
4718             Ilya Tocar  <ilya.tocar@intel.com>
4719             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4720             Ilya Verbin  <ilya.verbin@intel.com>
4721             Kirill Yukhin  <kirill.yukhin@intel.com>
4722             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4724         * config/i386/sse.md (VI124_AVX2_48_AVX512F): New.
4725         (VI8F_256_512): Ditto.
4726         (abs<mode>2): Changed iterator.
4727         (avx2_perm<mode>): Changed to ...
4728         (<avx2_avx512f>_perm<mode>): This.
4729         (avx2_perm<mode>_1): Changed to ...
4730         (<avx2_avx512f>_perm<mode>_1): This.
4732 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4733             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4734             Sergey Lega  <sergey.s.lega@intel.com>
4735             Anna Tikhonova  <anna.tikhonova@intel.com>
4736             Ilya Tocar  <ilya.tocar@intel.com>
4737             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4738             Ilya Verbin  <ilya.verbin@intel.com>
4739             Kirill Yukhin  <kirill.yukhin@intel.com>
4740             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4742         * config/i386/sse.md (VI48_AVX512F): New.
4743         (VI48_AVX2): Changed to ...
4744         (VI48_AVX2_48_AVX512F): This.
4745         (avx2_ashrv<mode>): Changed to ...
4746         (<avx2_avx512f>_ashrv<mode>): This.
4747         (avx2_<shift_insn>v<mode>): Changed to ...
4748         (<avx2_avx512f>_<shift_insn>v<mode>): This.
4750 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4751             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4752             Sergey Lega  <sergey.s.lega@intel.com>
4753             Anna Tikhonova  <anna.tikhonova@intel.com>
4754             Ilya Tocar  <ilya.tocar@intel.com>
4755             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4756             Ilya Verbin  <ilya.verbin@intel.com>
4757             Kirill Yukhin  <kirill.yukhin@intel.com>
4758             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4760         * config/i386/sse.md (VI4_AVX512F): New.
4761         (VI8_AVX2_AVX512F): Ditto.
4762         (mul<mode>3): Extended with wider modes.
4763         (*<sse4_1_avx2>_mul<mode>3): Ditto.
4764         (mul<mode>3): Ditto.
4765         (vec_widen_<s>mult_odd_<mode>): Ditto.
4767 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4768             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4769             Sergey Lega  <sergey.s.lega@intel.com>
4770             Anna Tikhonova  <anna.tikhonova@intel.com>
4771             Ilya Tocar  <ilya.tocar@intel.com>
4772             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4773             Ilya Verbin  <ilya.verbin@intel.com>
4774             Kirill Yukhin  <kirill.yukhin@intel.com>
4775             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4777         * config/i386/sse.md (VI2_AVX512F): New.
4778         (VI124_AVX512F): Ditto.
4779         (sseunpackmode): Extended with wider modes.
4780         (sseunpackfltmode): Ditto.
4781         (vec_unpacks_float_hi_<mode>): Ditto.
4782         (vec_unpacks_float_lo_<mode>): Ditto.
4783         (vec_unpacku_float_hi_<mode>): Ditto.
4784         (vec_unpacku_float_lo_<mode>): Ditto.
4785         (vec_unpacks_lo_<mode>): Ditto.
4786         (vec_unpacks_hi_<mode>): Ditto.
4787         (vec_unpacku_lo_<mode>): Ditto.
4788         (vec_unpacku_hi_<mode>): Ditto.
4790 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4791             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4792             Sergey Lega  <sergey.s.lega@intel.com>
4793             Anna Tikhonova  <anna.tikhonova@intel.com>
4794             Ilya Tocar  <ilya.tocar@intel.com>
4795             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4796             Ilya Verbin  <ilya.verbin@intel.com>
4797             Kirill Yukhin  <kirill.yukhin@intel.com>
4798             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4800         * config/i386/i386.md (multdiv): New.
4801         (multdiv_mnemonic): Ditto.
4802         * config/i386/sse.md (<sse>_vmmul<mode>3): Changed to...
4803         (<sse>_vm<multdiv_mnemonic><mode>3): This.
4804         (<sse>_vmdiv<mode>3): Removed.
4806 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4807             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4808             Sergey Lega  <sergey.s.lega@intel.com>
4809             Anna Tikhonova  <anna.tikhonova@intel.com>
4810             Ilya Tocar  <ilya.tocar@intel.com>
4811             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4812             Ilya Verbin  <ilya.verbin@intel.com>
4813             Kirill Yukhin  <kirill.yukhin@intel.com>
4814             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4816         * config/i386/sse.md (V): Extended with wider modes.
4817         (VF2): Ditto.
4818         (ssehalfvecmode): Ditto.
4819         (i128): Ditto.
4820         (ssepackfltmode): Ditto.
4821         (avx_vec_concat<mode>): Ditto.
4822         (V_256_512): New iterator.
4823         (VF2_512_256): Ditto.
4824         (si2dfmode): New attribute.
4825         (si2dfmodelower): Ditto.
4826         (sf2dfmode): Ditto.
4827         (concat_tg_mode): Ditto.
4828         (floatv4siv4df2): Changed to ...
4829         (float<si2dfmodelower><mode>2): This.
4830         (avx_cvtps2pd256): Changed to ...
4831         (<sse2_avx_avx512f>_cvtps2pd<avxsizesuffix>): This.
4832         (vec_pack_trunc_v4df): Changed to ...
4833         (vec_pack_trunc_<mode>): This.
4834         (avx_vpermil<mode>): Changed to ...
4835         (<sse2_avx_avx512f>_vpermil<mode>): This.
4836         (<fixsuffix>fix_truncv8dfv8si2): New.
4837         (vec_pack_sfix_trunc_v8df): Ditto.
4838         (avx512f_rndscale<mode>): Ditto.
4839         (avx512f_roundpd512): Ditto.
4840         (vec_pack_ufix_trunc_<mode>): Updated iterator.
4842 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4843             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4844             Sergey Lega  <sergey.s.lega@intel.com>
4845             Anna Tikhonova  <anna.tikhonova@intel.com>
4846             Ilya Tocar  <ilya.tocar@intel.com>
4847             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4848             Ilya Verbin  <ilya.verbin@intel.com>
4849             Kirill Yukhin  <kirill.yukhin@intel.com>
4850             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4852         * config/i386/i386.md (any_fix): New iterator.
4853         (fixsuffix): New attribute.
4854         * config/i386/sse.md (VF1): Extened with wider modes.
4855         (VI): Ditto.
4856         (VI_AVX2): Ditto.
4857         (VI8): Ditto.
4858         (sseintvecmodelower): Ditto.
4859         (ssescalarmode): Ditto.
4860         (ssescalarnum): Ditto.
4861         (VF1_128_256): New.
4862         (ssexmmmode): Ditto.
4863         (<fixsuffix>fix_truncv16sfv16si2): Ditto.
4864         (<sse>_rcp<mode>2): Change iterator.
4865         (rsqrt<mode>2): Ditto.
4866         (<sse>_rsqrt<mode>2): Ditto.
4867         (avx2_vec_dup<mode>): Ditto.
4868         (<sse4_1>_round<ssemodesuffix>_sfix<avxsizesuffix>): Ditto.
4869         (round<mode>2_sfix): Ditto.
4870         (avx2_pbroadcast<mode>): Ditto.
4871         (*andnot<mode>3): Handle XI mode.
4872         (*<code><mode>3): Ditto.
4873         (AVXTOSSEMODE): Removed.
4874         (avx_vpermil<mode>): Changed to ...
4875         (<sse2_avx_avx512f>_vpermil<mode>): This.
4877 2013-10-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
4878             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
4879             Sergey Lega  <sergey.s.lega@intel.com>
4880             Anna Tikhonova  <anna.tikhonova@intel.com>
4881             Ilya Tocar  <ilya.tocar@intel.com>
4882             Andrey Turetskiy  <andrey.turetskiy@intel.com>
4883             Ilya Verbin  <ilya.verbin@intel.com>
4884             Kirill Yukhin  <kirill.yukhin@intel.com>
4885             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
4887         * config/i386/sse.md (<sse2>_movnt<mode>): Update constraint to "v".
4888         (<sse>_comi): Ditto.
4889         (<sse>_ucomi): Ditto.
4890         (sse_cvtss2siq_2): Ditto.
4891         (sse2_cvtsd2si): Ditto.
4892         (sse2_cvtsd2siq): Ditto.
4893         (sse2_cvttsd2si): Ditto.
4894         (sse2_cvttsd2siq): Ditto.
4895         (<shift_insn><mode>3): Ditto.
4896         (sse2_cvtsi2sdq): Update constraint and prefix.
4897         (sse_cvtsi2ss): Update prefix.
4898         (sse_cvtsi2ssq): Ditto.
4900 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4902         * tree-vrp.c (infer_nonnull_range): Use is_gimple_call,
4903         ignore internal calls.
4905 2013-10-11  Richard Biener  <rguenther@suse.de>
4907         * tree-pretty-print.c (dump_generic_node): Allow to dump both (D)
4908         and (ab) for SSA_NAMEs.  Mark INTEGER_CSTs with (OVF) if
4909         TREE_OVERFLOW is set.
4911 2013-10-11  Thomas Schwinge  <thomas@codesourcery.com>
4913         * tree.h (OMP_CLAUSE_CODE): Remove duplicate definition.
4915         * gimple.c: GIMPLE statements have subcodes, not sub-codes.
4916         * gimple.h: Likewise.
4918         * doc/generic.texi (OpenMP): OMP_CLAUSE_* are subcodes, not sub-codes.
4920         * doc/generic.texi (Adding new DECL node types): Explain *_CHECK
4921         macros.
4923         * doc/gimple.texi (is_gimple_omp): Move into the correct section.
4925         * acinclude.m4 (gcc_GAS_FLAGS): Add more gcc_cv_as_flags overrides.
4926         * configure: Regenerate.
4928 2013-10-11  Jakub Jelinek  <jakub@redhat.com>
4930         * tree-pretty-print.c (dump_omp_clause): Handle OMP_CLAUSE__LOOPTEMP_
4931         and new OpenMP 4.0 clauses, handle UDR OMP_CLAUSE_REDUCTION,
4932         formatting fixes, use pp_colon instead of pp_character (..., ':'),
4933         similarly pp_right_paren.
4934         (dump_generic_node): Handle OMP_DISTRIBUTE, OMP_TEAMS,
4935         OMP_TARGET_DATA, OMP_TARGET, OMP_TARGET_UPDATE, OMP_TASKGROUP,
4936         allow OMP_FOR_INIT to be NULL, handle OMP_ATOMIC_SEQ_CST.
4937         * tree.c (omp_clause_num_ops, omp_clause_code_name): Add OpenMP 4.0
4938         clauses.
4939         (omp_declare_simd_clauses_equal,
4940         omp_remove_redundant_declare_simd_attrs): New functions.
4941         (attribute_value_equal): Use omp_declare_simd_clauses_equal.
4942         (walk_tree_1): Handle new OpenMP 4.0 clauses.
4943         * tree.h (OMP_LOOP_CHECK): Define.
4944         (OMP_FOR_BODY, OMP_FOR_CLAUSES, OMP_FOR_INIT, OMP_FOR_COND,
4945         OMP_FOR_INCR, OMP_FOR_PRE_BODY): Use it.
4946         (OMP_TASKGROUP_BODY, OMP_TEAMS_BODY, OMP_TEAMS_CLAUSES,
4947         OMP_TARGET_DATA_BODY, OMP_TARGET_DATA_CLAUSES, OMP_TARGET_BODY,
4948         OMP_TARGET_CLAUSES, OMP_TARGET_UPDATE_CLAUSES, OMP_CLAUSE_SIZE,
4949         OMP_ATOMIC_SEQ_CST, OMP_CLAUSE_DEPEND_KIND, OMP_CLAUSE_MAP_KIND,
4950         OMP_CLAUSE_MAP_ZERO_BIAS_ARRAY_SECTION, OMP_CLAUSE_PROC_BIND_KIND,
4951         OMP_CLAUSE_REDUCTION_OMP_ORIG_REF, OMP_CLAUSE_ALIGNED_ALIGNMENT,
4952         OMP_CLAUSE_NUM_TEAMS_EXPR, OMP_CLAUSE_THREAD_LIMIT_EXPR,
4953         OMP_CLAUSE_DEVICE_ID, OMP_CLAUSE_DIST_SCHEDULE_CHUNK_EXPR,
4954         OMP_CLAUSE_SIMDLEN_EXPR): Define.
4955         (OMP_CLAUSE_DECL): Change range up to OMP_CLAUSE__LOOPTEMP_.
4956         (omp_remove_redundant_declare_simd_attrs): New prototype.
4957         * gimple.def (GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET,
4958         GIMPLE_OMP_TEAMS): New codes.
4959         (GIMPLE_OMP_RETURN): Use GSS_OMP_ATOMIC_STORE instead of GSS_BASE.
4960         * omp-low.c (struct omp_context): Add cancel_label and cancellable
4961         fields.
4962         (target_nesting_level): New variable.
4963         (extract_omp_for_data): Handle GF_OMP_FOR_KIND_DISTRIBUTE and
4964         OMP_CLAUSE_DIST_SCHEDULE.  Don't fallback to library implementation
4965         for collapse > 1 static schedule unless ordered.
4966         (get_ws_args_for): Add par_stmt argument.  Handle combined loops.
4967         (determine_parallel_type): Adjust get_ws_args_for caller.
4968         (install_var_field): Handle mask & 4 for double indirection.
4969         (scan_sharing_clauses): Ignore shared clause on teams construct.
4970         Handle OMP_CLAUSE__LOOPTEMP_ and new OpenMP 4.0 clauses.
4971         (create_omp_child_function): If inside target or declare target
4972         constructs, set "omp declare target" attribute on the child function.
4973         (find_combined_for): New function.
4974         (scan_omp_parallel): Handle combined loops.
4975         (scan_omp_target, scan_omp_teams): New functions.
4976         (check_omp_nesting_restrictions): Check new OpenMP 4.0 nesting
4977         restrictions and set ctx->cancellable for cancellable constructs.
4978         (scan_omp_1_stmt): Call check_omp_nesting_restrictions also on
4979         selected builtin calls.  Handle GIMPLE_OMP_TASKGROUP,
4980         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS.
4981         (build_omp_barrier): Add lhs argument, return gimple rather than tree.
4982         (omp_clause_aligned_alignment): New function.
4983         (lower_rec_simd_input_clauses): Only call SET_DECL_VALUE_EXPR on decls.
4984         (lower_rec_input_clauses): Add FD argument.  Ignore shared clauses
4985         on teams constructs.  Handle user defined reductions and new
4986         OpenMP 4.0 clauses.
4987         (lower_reduction_clauses): Don't set placeholder to address of ref
4988         if it has already the right type.
4989         (lower_send_clauses): Handle OMP_CLAUSE__LOOPTEMP_.
4990         (expand_parallel_call): Use the new non-_start suffixed builtins,
4991         handle OMP_CLAUSE_PROC_BIND, don't call the outlined function
4992         and GOMP_parallel_end after the call.
4993         (expand_task_call): Handle OMP_CLAUSE_DEPEND.
4994         (expand_omp_for_init_counts): Handle combined loops.
4995         (expand_omp_for_init_vars): Add inner_stmt argument, handle combined
4996         loops.
4997         (expand_omp_for_generic): Likewise.  Use GOMP_loop_end_cancel at the
4998         end of cancellable loops.
4999         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
5000         Likewise.  Handle collapse > 1 loops.
5001         (expand_omp_simd): Handle combined loops.
5002         (expand_omp_for): Add inner_stmt argument, adjust callers of
5003         expand_omp_for* functions, use expand_omp_for_static*chunk even
5004         for collapse > 1 unless ordered.
5005         (expand_omp_sections): Use GOMP_sections_end_cancel at the end
5006         of cancellable sections.
5007         (expand_omp_single): Remove need_barrier variable, just rely on
5008         gimple_omp_return_nowait_p.  Adjust build_omp_barrier caller.
5009         (expand_omp_synch): Allow GIMPLE_OMP_TASKGROUP and GIMPLE_OMP_TEAMS.
5010         (expand_omp_atomic_load, expand_omp_atomic_store,
5011         expand_omp_atomic_fetch_op): Handle gimple_omp_atomic_seq_cst_p.
5012         (expand_omp_target): New function.
5013         (expand_omp): Handle combined loops.  Handle GIMPLE_OMP_TASKGROUP,
5014         GIMPLE_OMP_TEAMS, GIMPLE_OMP_TARGET.
5015         (build_omp_regions_1): Immediately close region for
5016         GF_OMP_TARGET_KIND_UPDATE.
5017         (maybe_add_implicit_barrier_cancel): New function.
5018         (lower_omp_sections): Adjust lower_rec_input_clauses caller.  Handle
5019         cancellation.
5020         (lower_omp_single): Likewise.  Add clobber after the barrier.
5021         (lower_omp_taskgroup): New function.
5022         (lower_omp_for): Handle combined loops.  Adjust
5023         lower_rec_input_clauses caller.  Handle cancellation.
5024         (lower_depend_clauses): New function.
5025         (lower_omp_taskreg): Lower depend clauses.  Adjust
5026         lower_rec_input_clauses caller.  Add clobber after the call.  Handle
5027         cancellation.
5028         (lower_omp_target, lower_omp_teams): New functions.
5029         (lower_omp_1): Handle cancellation.  Handle GIMPLE_OMP_TASKGROUP,
5030         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GOMP_barrier, GOMP_cancel
5031         and GOMP_cancellation_point calls.
5032         (lower_omp): Fold stmts inside of target region.
5033         (diagnose_sb_1, diagnose_sb_2): Handle GIMPLE_OMP_TASKGROUP,
5034         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
5035         * builtin-types.def (DEF_FUNCTION_TYPE_8): Document.
5036         (BT_FN_VOID_OMPFN_PTR_UINT,
5037         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG,
5038         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG,
5039         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT): Remove.
5040         (BT_FN_VOID_OMPFN_PTR_UINT_UINT_UINT,
5041         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_UINT,
5042         BT_FN_VOID_OMPFN_PTR_UINT_LONG_LONG_LONG_LONG_UINT,
5043         BT_FN_BOOL_INT, BT_FN_BOOL_INT_BOOL, BT_FN_VOID_UINT_UINT,
5044         BT_FN_VOID_INT_PTR_SIZE_PTR_PTR_PTR,
5045         BT_FN_VOID_INT_OMPFN_PTR_SIZE_PTR_PTR_PTR,
5046         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR): New.
5047         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1,
5048         call_may_clobber_ref_p_1): Handle BUILT_IN_GOMP_BARRIER_CANCEL,
5049         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_LOOP_END_CANCEL,
5050         BUILT_IN_GOMP_SECTIONS_END_CANCEL.  Don't handle
5051         BUILT_IN_GOMP_PARALLEL_END.
5052         * gimple-low.c (lower_stmt): Handle GIMPLE_OMP_TASKGROUP,
5053         GIMPLE_OMP_TARGET and GIMPLE_OMP_TEAMS.
5054         * gimple-pretty-print.c (dump_gimple_omp_for): Handle
5055         GF_OMP_FOR_KIND_DISTRIBUTE.
5056         (dump_gimple_omp_target, dump_gimple_omp_teams): New functions.
5057         (dump_gimple_omp_block): Handle GIMPLE_OMP_TASKGROUP.
5058         (dump_gimple_omp_return): Print lhs if it has any.
5059         (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Handle
5060         gimple_omp_atomic_seq_cst_p.
5061         (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP, GIMPLE_OMP_TARGET
5062         and GIMPLE_OMP_TEAMS.
5063         * langhooks.c (lhd_omp_mappable_type): New function.
5064         * tree-vectorizer.c (struct simd_array_to_simduid): Fix up comment.
5065         * langhooks.h (struct lang_hooks_for_types): Add omp_mappable_type
5066         hook.
5067         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP,
5068         GOVD_ALIGNED and GOVD_MAP_TO_ONLY.
5069         (enum omp_region_type): Add ORT_TEAMS, ORT_TARGET_DATA and ORT_TARGET.
5070         (struct gimplify_omp_ctx): Add combined_loop field.
5071         (gimplify_call_expr, gimplify_modify_expr): Don't call fold_stmt
5072         on stmts inside of target region.
5073         (is_gimple_stmt): Return true for OMP_DISTRIBUTE and OMP_TASKGROUP.
5074         (omp_firstprivatize_variable): Handle GOVD_MAP, GOVD_ALIGNED,
5075         ORT_TARGET and ORT_TARGET_DATA.
5076         (omp_add_variable): Avoid checks on readding var for GOVD_ALIGNED.
5077         Handle GOVD_MAP.
5078         (omp_notice_threadprivate_variable): Complain about threadprivate
5079         variables in target region.
5080         (omp_notice_variable): Complain about vars with non-mappable type
5081         in target region.  Handle ORT_TEAMS, ORT_TARGET and ORT_TARGET_DATA.
5082         (omp_check_private): Ignore ORT_TARGET* regions.
5083         (gimplify_scan_omp_clauses, gimplify_adjust_omp_clauses_1,
5084         gimplify_adjust_omp_clauses): Handle new OpenMP 4.0 clauses.
5085         (find_combined_omp_for): New function.
5086         (gimplify_omp_for): Handle gimplification of combined loops.
5087         (gimplify_omp_workshare): Gimplify also OMP_TARGET, OMP_TARGET_DATA,
5088         OMP_TEAMS.
5089         (gimplify_omp_target_update): New function.
5090         (gimplify_omp_atomic): Handle OMP_ATOMIC_SEQ_CST.
5091         (gimplify_expr): Handle OMP_DISTRIBUTE, OMP_TARGET, OMP_TARGET_DATA,
5092         OMP_TARGET_UPDATE, OMP_TEAMS, OMP_TASKGROUP.
5093         (gimplify_body): If fndecl has "omp declare target" attribute, add
5094         implicit ORT_TARGET context around it.
5095         * tree.def (OMP_DISTRIBUTE, OMP_TEAMS, OMP_TARGET_DATA, OMP_TARGET,
5096         OMP_TASKGROUP, OMP_TARGET_UPDATE): New tree codes.
5097         * tree-nested.c (convert_nonlocal_reference_stmt,
5098         convert_local_reference_stmt, convert_gimple_call): Handle
5099         GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5100         * omp-builtins.def (BUILT_IN_GOMP_TASK): Use
5101         BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT_PTR
5102         instead of BT_FN_VOID_OMPFN_PTR_OMPCPYFN_LONG_LONG_BOOL_UINT.
5103         (BUILT_IN_GOMP_TARGET, BUILT_IN_GOMP_TARGET_DATA,
5104         BUILT_IN_GOMP_TARGET_END_DATA, BUILT_IN_GOMP_TARGET_UPDATE,
5105         BUILT_IN_GOMP_TEAMS, BUILT_IN_BARRIER_CANCEL,
5106         BUILT_IN_GOMP_LOOP_END_CANCEL,
5107         BUILT_IN_GOMP_SECTIONS_END_CANCEL, BUILT_IN_OMP_GET_TEAM_NUM,
5108         BUILT_IN_OMP_GET_NUM_TEAMS, BUILT_IN_GOMP_TASKGROUP_START,
5109         BUILT_IN_GOMP_TASKGROUP_END, BUILT_IN_GOMP_PARALLEL_LOOP_STATIC,
5110         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC,
5111         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED,
5112         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME, BUILT_IN_GOMP_PARALLEL,
5113         BUILT_IN_GOMP_PARALLEL_SECTIONS, BUILT_IN_GOMP_CANCEL,
5114         BUILT_IN_GOMP_CANCELLATION_POINT): New built-ins.
5115         (BUILT_IN_GOMP_PARALLEL_LOOP_STATIC_START,
5116         BUILT_IN_GOMP_PARALLEL_LOOP_DYNAMIC_START,
5117         BUILT_IN_GOMP_PARALLEL_LOOP_GUIDED_START,
5118         BUILT_IN_GOMP_PARALLEL_LOOP_RUNTIME_START,
5119         BUILT_IN_GOMP_PARALLEL_START, BUILT_IN_GOMP_PARALLEL_END,
5120         BUILT_IN_GOMP_PARALLEL_SECTIONS_START): Remove.
5121         * tree-inline.c (remap_gimple_stmt, estimate_num_insns):
5122         Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5123         * gimple.c (gimple_build_omp_taskgroup, gimple_build_omp_target,
5124         gimple_build_omp_teams): New functions.
5125         (walk_gimple_op): Handle GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS and
5126         GIMPLE_OMP_TASKGROUP.  Walk optional lhs on GIMPLE_OMP_RETURN.
5127         (walk_gimple_stmt, gimple_copy): Handle GIMPLE_OMP_TARGET,
5128         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5129         * gimple.h (enum gf_mask): GF_OMP_FOR_KIND_DISTRIBUTE,
5130         GF_OMP_FOR_COMBINED, GF_OMP_FOR_COMBINED_INTO,
5131         GF_OMP_TARGET_KIND_MASK, GF_OMP_TARGET_KIND_REGION,
5132         GF_OMP_TARGET_KIND_DATA, GF_OMP_TARGET_KIND_UPDATE,
5133         GF_OMP_ATOMIC_SEQ_CST): New.
5134         (gimple_build_omp_taskgroup, gimple_build_omp_target,
5135         gimple_build_omp_teams): New prototypes.
5136         (gimple_has_substatements): Handle GIMPLE_OMP_TARGET,
5137         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5138         (gimple_omp_subcode): Use GIMPLE_OMP_TEAMS instead of
5139         GIMPLE_OMP_SINGLE as end of range.
5140         (gimple_omp_return_set_lhs, gimple_omp_return_lhs,
5141         gimple_omp_return_lhs_ptr, gimple_omp_atomic_seq_cst_p,
5142         gimple_omp_atomic_set_seq_cst, gimple_omp_for_combined_p,
5143         gimple_omp_for_set_combined_p, gimple_omp_for_combined_into_p,
5144         gimple_omp_for_set_combined_into_p, gimple_omp_target_clauses,
5145         gimple_omp_target_clauses_ptr, gimple_omp_target_set_clauses,
5146         gimple_omp_target_kind, gimple_omp_target_set_kind,
5147         gimple_omp_target_child_fn, gimple_omp_target_child_fn_ptr,
5148         gimple_omp_target_set_child_fn, gimple_omp_target_data_arg,
5149         gimple_omp_target_data_arg_ptr, gimple_omp_target_set_data_arg,
5150         gimple_omp_teams_clauses, gimple_omp_teams_clauses_ptr,
5151         gimple_omp_teams_set_clauses): New inlines.
5152         (CASE_GIMPLE_OMP): Add GIMPLE_OMP_TARGET, GIMPLE_OMP_TEAMS
5153         and GIMPLE_OMP_TASKGROUP.
5154         * tree-core.h (enum omp_clause_code): Add new OpenMP 4.0 clause codes.
5155         (enum omp_clause_depend_kind, enum omp_clause_map_kind,
5156         enum omp_clause_proc_bind_kind): New.
5157         (union omp_clause_subcode): Add depend_kind, map_kind and
5158         proc_bind_kind fields.
5159         * tree-cfg.c (make_edges): Handle GIMPLE_OMP_TARGET,
5160         GIMPLE_OMP_TEAMS and GIMPLE_OMP_TASKGROUP.
5161         * langhooks-def.h (lhd_omp_mappable_type): New prototype.
5162         (LANG_HOOKS_OMP_MAPPABLE_TYPE): Define.
5163         (LANG_HOOKS_FOR_TYPES_INITIALIZER): Add it.
5165 2013-10-10  Teresa Johnson  <tejohnson@google.com>
5167         * predict.c (tree_estimate_probability): Add new parameter
5168         for estimate_bb_frequencies.
5169         (estimate_bb_frequencies): Add new parameter to force estimation.
5170         (rebuild_frequencies): When max frequency in function is small,
5171         recompute counts from frequencies.
5172         * predict.h (estimate_bb_frequencies): New parameter.
5174 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5176         * ipa-inline.c (ipa_inline): Fix leak of "order" when
5177         optimizations are disabled.
5179 2013-10-10  David Malcolm  <dmalcolm@redhat.com>
5181         * coverage.c (coverage_finish): Fix leak of da_file_name.
5183 2013-10-10  Jan Hubicka  <jh@suse.cz>
5185         * config/i386/x86-tune.def: Enable X86_TUNE_SSE_TYPELESS_STORES
5186         for generic, enable X86_TUNE_SSE_LOAD0_BY_PXOR for Bulldozer,
5187         Bobcat and generic.
5189 2013-10-10  Jakub Jelinek  <jakub@redhat.com>
5191         PR middle-end/58670
5192         * stmt.c (expand_asm_operands): Add FALLTHRU_BB argument,
5193         if any labels are in FALLTHRU_BB, use a special label emitted
5194         immediately after the asm goto insn rather than label_rtx
5195         of the LABEL_DECL.
5196         (expand_asm_stmt): Adjust caller.
5197         * cfgrtl.c (commit_one_edge_insertion): Force splitting of
5198         edge if the last insn in predecessor is a jump with single successor,
5199         but it isn't simplejump_p.
5201 2013-10-10  Richard Biener  <rguenther@suse.de>
5203         PR tree-optimization/58656
5204         * tree-ssa-pre.c (phi_translate): Do not cache failed translations.
5206 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5208         * gimplify.c: Include expr.h and tm_p.h for targets with special
5209         va-arg padding requirements.
5211 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5213         * tree-flow.h: Move some prototypes to gimple.h.
5214         (gimple_fold_indirect_ref): Move prototype to gimple-fold.h.
5215         * gimple.h: Relocate some prototypes from tree-flow.h
5216         * builtins.c (std_gimplify_va_arg_expr, build_va_arg_indirect_ref):
5217         Move to gimplify.c.
5218         * gimplify.c (gimple_fold_indirect_ref): Move to gimple-fold.c.
5219         (build_va_arg_indirect_ref): Relocate and make static.
5220         (std_gimplify_va_arg_expr): Relocate here.
5221         * gimple-fold.c (gimple_fold_indirect_ref): Relocate here.
5222         * gimple-fold.h (gimple_fold_indirect_ref): Add prototype.
5224 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5226         * doc/md.texi: Document the mnemonic attribute.
5228 2013-10-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5230         PR target/57377
5231         * gensupport.c (gen_mnemonic_attr): Handle (set (attr x) y) and
5232         (set_attr_alternative x ...) when searching for user defined
5233         mnemonic attribute.
5235 2013-10-10  Andrew MacLeod  <amacleod@redhat.com>
5237         * config/aplha/alpha.c: Add gimple-ssa.h to include list.
5239 2013-10-09  Easwaran Raman <eraman@google.com>
5241         * params.def (PARAM_MIN_SIZE_FOR_STACK_SHARING): New param...
5242         * cfgexpand.c (defer_stack_allocation): ...use here
5243         * doc/invoke.texi: Add documentation for min-size-for-stack-sharing.
5245 2013-10-09  Zhenqiang Chen  <zhenqiang.chen@arm.com>
5247         * tree-ssa-phiopts.c (rhs_is_fed_for_value_replacement): New function.
5248         (operand_equal_for_value_replacement): New function, extracted from
5249         value_replacement and enhanced to catch more cases.
5250         (value_replacement): Use operand_equal_for_value_replacement.
5252 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5254         * loop-doloop.c (doloop_modify, doloop_optimize): Use
5255         get_max_loop_iterations.
5257 2013-10-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5259         * config/arm/aarch-common.c (arm_early_load_addr_dep):
5260         Place comment above function.
5262 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5264         * tree-flow.h: Remove all remaining prototypes, enums and structs that
5265         are not related to tree-cfg.c.
5266         * tree-ssa-address.h: New file.  Relocate prototypes.
5267         * tree-ssa-address.c: (struct mem_address): Relocate from tree-flow.h.
5268         (addr_for_mem_ref): New.  Combine call to get_address_description and
5269         return addr_for_mem_ref.
5270         * expr.c (expand_expr_real_1): Use new addr_for_mem_ref routine.
5271         * tree-ssa-live.h: Adjust prototypes.
5272         * passes.c: Include tree-ssa-live.h.
5273         * gimple-pretty-print.h (gimple_dump_bb): Add prototype.
5274         * graphite.c (graphite_transform_loops): Make static.
5275         (graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5276         make_pass_graphite, pass_data_graphite_transforms,
5277         make_pass_graphite_transforms): Relocate here from tree-ssa-loop.c.
5278         * ipa-pure-const.c (warn_function_noreturn): Make static.
5279         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5280         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5281         Relocate from tree-cfg.c
5282         * tree-cfg.c (tree_node_can_be_shared, gimple_empty_block_p): Make
5283         static.
5284         (execute_warn_function_noreturn, gate_warn_function_noreturn,
5285         class pass_warn_function_noreturn, make_pass_warn_function_noreturn):
5286         Move to ipa-pure-const.c.
5287         (execute_fixup_cfg, class pass_fixup_cfg, make_pass_fixup_cfg):
5288         Relocate from tree-optimize.c.
5289         * tree-optimize.c (execute_fixup_cfg, class pass_fixup_cfg,
5290         make_pass_fixup_cfg): Move to tree-cfg.c.
5291         * tree-chrec.h: (enum ev_direction): Relocate here from tree-flow.h.
5292         Relocate some prototypes.
5293         * tree-data-ref.h (tree_check_data_deps) Add prototype.
5294         * tree-dump.c (dump_function_to_file): Remove prototype.
5295         Add tree-flow.h to the include file.
5296         * tree-dump.h: Remove prototype.
5297         * tree-parloops.h: New File.  Add prototypes.
5298         * tree-parloops.c (gate_tree_parallelize_loops, tree_parallelize_loops,
5299         pass_data_parallelize_loops,  make_pass_parallelize_loops): Relocate
5300         from tree-ssa-loop.c.
5301         * tree-predcom.c (run_tree_predictive_commoning,
5302         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom):
5303         Relocate here from tree-ssa-loop.c.
5304         * tree-ssa-dom.c (tree_ssa_dominator_optimize) Don't call
5305         ssa_name_values.release ().
5306         * tree-ssa-threadedge.h: New File.  Relocate prototypes here.
5307         (ssa_name_values): Relocate from tree-flow.h.
5308         * tree-ssa.h: Include tree-ssa-threadedge.h and tree-ssa-address.h.
5309         * tree-ssa-loop.c (run_tree_predictive_commoning,
5310         gate_tree_predictive_commoning, pass_data_predcom, make_pass_predcom,
5311         graphite_transforms, gate_graphite_transforms, pass_data_graphite,
5312         make_pass_graphite, pass_data_graphite_transforms,
5313         make_pass_graphite_transforms, gate_tree_parallelize_loops,
5314         tree_parallelize_loops, pass_data_parallelize_loops,
5315         make_pass_parallelize_loops): Move to other files.
5316         * tree-vectorizer.h (lpeel_tree_duplicate_loop_to_edge_cfg): Prototype
5317         moved here.
5318         * tree.h: Remove prototypes from tree-address.c.
5320 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5322         * tree-flow.h (tm_restart_node, gimple_df): Move to gimple-ssa.h.
5323         (struct int_tree_map): Move to tree-hasher.h
5324         (SCALE, LABEL, PERCENT): Move to gimple.h
5325         * tree-flow-inline.h: Delete.  Move functions to other files.
5326         (unmodifiable_var_p, ref_contains_array_ref): Unused, so delete.
5327         * gimple-ssa.h (tm_restart_node, gimple_df): Relocate from tree-flow.h.
5328         (gimple_in_ssa_p, gimple_vop): Relocate from tree-flow-inline.h
5329         * gimple.h (imple_stmt_max_uid, set_gimple_stmt_max_uid,
5330         inc_gimple_stmt_max_uid, get_lineno): Relocate from tree-flow-inline.h.
5331         (SCALE, LABEL, PERCENT): Relocate from tree-flow.h
5332         * tree-hasher.h: Don't include tree-flow.h.
5333         (struct int_tree_map): Relocate from tree-flow.h.
5334         * tree-sra.c (contains_view_convert_expr_p): Relocate from
5335         tree-flow-inline.h and make static.
5336         * tree-ssa-alias.h (ranges_overlap_p): Relocate from
5337         tree-flow-inline.h.
5338         * tree-ssa-operands.c (gimple_ssa_operands): Relocate from
5339         tree-flow-inline.h and make static.
5340         * tree.h (is_global_var, may_be_aliased): Relocate from
5341         tree-flow-inline.h.
5342         * Makefile.in (GTFILES): Remove tree-flow.h and add gimple-ssa.h.
5343         * value-prof.c: No longer include tree-flow-inline.h.
5344         * tree-switch-conversion.c: No longer include tree-flow-inline.h.
5346 2013-10-09  Andrew MacLeod  <amacleod@redhat.com>
5348         * tree-flow.h: Move some protoypes.  Include new tree-ssa-loop.h.
5349         (struct affine_iv, struct tree_niter_desc): Move to tree-ssa-loop.h.
5350         (enum move_pos): Move to tree-ssa-loop-im.h
5351         * cfgloop.h: Move some prototypes.
5352         (gcov_type_to_double_int): relocate from tree-ssa-loop.niter.c.
5353         * tree-flow-inline.h (loop_containing_stmt): Move to tree-ssa-loop.h.
5354         * tree-ssa-loop.h: New File.  Include other tree-ssa-loop-*.h files.
5355         (struct affine_iv, struct tree_niter_desc): Relocate from tree-flow.h.
5356         (loop_containing_stmt): Relocate from tree-flow-inline.h.
5357         * tree-ssa-loop-ch.c: (do_while_loop_p): Make static.
5358         * tree-ssa-loop-im.c (for_each_index): Move to tree-ssa-loop.c.
5359         (enum move_pos): Relocate here.
5360         (lsm_tmp_name_add, gen_lsm_tmp_name, get_lsm_tmp_name): Move to
5361         tree-ssa-loop.c.
5362         (execute_sm_if_changed_flag_set): Change get_lsm_tmp_name call.
5363         (tree_ssa_loop_im, gate_tree_ssa_loop_im, pass_data_lim,
5364         make_pass_lim): Relocate here from tree-ssa-loop.c.
5365         * tree-ssa-loop-ivcanon.c (tree_num_loop_insns): Move to
5366         tree-ssa-loop.c.
5367         (loop_edge_to_cancel, unloop_loops): Make static.
5368         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5369         make_pass_iv_canon): Relocate from tree-ssa-loop.c.
5370         (tree_complete_unroll, gate_tree_complete_unroll,
5371         pass_data_complete_unroll, make_pass_complete_unroll): Relocate here.
5372         (tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5373         pass_data_complete_unrolli, make_pass_complete_unrolli): Relocate here.
5374         * tree-ssa-loop-ivopts.c: Remove local prototypes.
5375         (stmt_invariant_in_loop_p): Remove unused function.
5376         * tree-ssa-loop-ivopts.h: New file.  Add prototypes.
5377         * tree-ssa-loop-manip.h: New file.  Add prototypes.
5378         * tree-ssa-loop-niter.c (record_niter_bound): Move to cfgloop.c.
5379         (gcov_type_to_double_int): Move to cfgloop.h.
5380         (double_int_cmp, bound_index,
5381         estimate_numbers_of_iterations_loop): Make static.
5382         (estimated_loop_iterations): Factor out get_estimated_loop_iterations.
5383         (max_loop_iterations): Factor out get_max_loop_iterations.
5384         (estimated_loop_iterations_int, max_stmt_executions_int): Move to
5385         cfgloop.c.
5386         * tree-ssa-loop-niter.h: New file.  Add prototypes.
5387         * tree-ssa-loop-prefetch.c (tree_ssa_loop_prefetch,
5388         gate_tree_ssa_loop_prefetch, pass_data_loop_prefetch,
5389         make_pass_loop_prefetch): Relocate from tree-ssa-loop.c.
5390         * tree-ssa-loop-unswitch.c (tree_ssa_loop_unswitch,
5391         gate_tree_ssa_loop_unswitch, pass_data_tree_unswitch,
5392         make_pass_tree_unswitch): Relocate from tree-ssa-loop.c.
5393         * tree-ssa-loop.c (tree_ssa_loop_im, gate_tree_ssa_loop_im,
5394         pass_data_lim, make_pass_lim): Move to tree-ssa-loop-im.c.
5395         (tree_ssa_loop_unswitch, gate_tree_ssa_loop_unswitch,
5396         pass_data_tree_unswitch, make_pass_tree_unswitch): Move.
5397         (tree_ssa_loop_ivcanon, gate_tree_ssa_loop_ivcanon, pass_data_iv_canon,
5398         make_pass_iv_canon, tree_complete_unroll, gate_tree_complete_unroll,
5399         pass_data_complete_unroll, make_pass_complete_unroll,
5400         tree_complete_unroll_inner, gate_tree_complete_unroll_inner,
5401         pass_data_complete_unrolli, make_pass_complete_unrolli): Move to
5402         tree-ssa-loop-ivcanon.c.
5403         (tree_ssa_loop_prefetch, gate_tree_ssa_loop_prefetch,
5404         pass_data_loop_prefetch, make_pass_loop_prefetch): Move to
5405         tree-ssa-loop-prefetch.c.
5406         (for_each_index, lsm_tmp_name_add, gen_lsm_tmp_name): Relocate from
5407         tree-ssa-loop-im.c.
5408         (get_lsm_tmp_name): Relocate and add suffix parameter.
5409         (tree_num_loop_insns): Relocate from tree-ssa-ivcanon.c.
5410         * tree-scalar-evolution.h (simple_iv): Don't use affive_iv typedef.
5411         * cfgloop.c (record_niter_bound, estimated_loop_iterations_int,
5412         max_stmt_executions_int): Move from tree-ssa-loop-niter.c.
5413         (get_estimated_loop_iterations): Factor out accessor from
5414         estimated_loop_iterations in tree-ssa-loop-niter.c.
5415         (get_max_loop_iterations): Factor out accessor from
5416         _max_loop_iterations in tree-ssa-niter.c.
5417         * loop-unroll.c (decide_unroll_constant_iterations,
5418         decide_unroll_runtime_iterations, decide_peel_simple,
5419         decide_unroll_stupid): Use new get_* accessors.
5421 2013-10-09  Marc Glisse  <marc.glisse@inria.fr>
5423         PR tree-optimization/20318
5424         * doc/extend.texi (returns_nonnull): New function attribute.
5425         * fold-const.c (tree_expr_nonzero_warnv_p): Look for returns_nonnull
5426         attribute.
5427         * tree-vrp.c (gimple_stmt_nonzero_warnv_p): Likewise.
5428         (stmt_interesting_for_vrp): Accept all GIMPLE_CALL.
5430 2013-10-09  Eric Botcazou  <ebotcazou@adacore.com>
5432         PR middle-end/58570
5433         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): Return
5434         false if both components are bitfields.
5436 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5438         * config/aarch64/arm_neon.h (vclz_s8, vclz_s16, vclz_s32)
5439         (vclzq_s8, vclzq_s16, vclzq_s32, vclz_u8, vclz_u16, vclz_u32)
5440         (vclzq_u8, vclzq_u16, vclzq_u32): Replace ASM with C.
5441         * config/aarch64/aarch64.h
5442         (CLZ_DEFINED_VALUE_AT_ZERO): Macro fixed for clz.
5443         * config/aarch64/aarch64-simd-builtins.def
5444         (VAR1 (UNOP, clz, 0, v4si)): Replaced with iterator.
5446 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5448         * config/aarch64/arm_neon.h (vadd_f64, vsub_f64): Implementation added.
5450 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5452         * config/aarch64/arm_neon.h (vdiv_f64): Added.
5454 2013-10-09  Alex Velenko  <Alex.Velenko@arm.com>
5456         * config/aarch64/arm_neon.h (vneg_f32): Asm replaced with C.
5457         (vneg_f64): New intrinsic.
5458         (vneg_s8): Asm replaced with C.
5459         (vneg_s16): Likewise.
5460         (vneg_s32): Likewise.
5461         (vneg_s64): New intrinsic.
5462         (vnegq_f32): Asm replaced with C.
5463         (vnegq_f64): Likewise.
5464         (vnegq_s8): Likewise.
5465         (vnegq_s16): Likewise.
5466         (vnegq_s32): Likewise.
5467         (vnegq_s64): Likewise.
5469 2013-10-09  Renlin Li  <Renlin.Li@arm.com>
5471         * config/arm/arm.c (arm_output_mi_thunk): Use plus_constant.
5473 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5475         * config/s390/s390.c (s390_register_info_stdarg_fpr): Remove
5476         packed stack special handling.
5477         (s390_frame_info, s390_emit_prologue, s390_emit_epilogue): Switch
5478         back to fixed stack slots for FPRs saved due to stdarg.
5480 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5482         * config/s390/s390.c (s390_frame_info): Restructure function.
5484 2013-10-09  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5486         * config/s390/s390.c (struct s390_frame_layout): New field
5487         gpr_save_slots.
5488         (cfun_save_arg_fprs_p, cfun_gpr_save_slot): New macros.
5489         (s390_reg_clobbered_rtx, s390_regs_ever_clobbered): Change type of
5490         regs_ever_clobbered to char*.
5491         (s390_regs_ever_clobbered): Check crtl->saves_all_registers instead
5492         of cfun->has_nonlocal_label.  Ignore frame related restore INSNs.
5493         (s390_register_info): Enable FPR save slots.  Move/Copy some
5494         functionality into ...
5495         (s390_register_info_gprtofpr, s390_register_info_stdarg_fpr)
5496         (s390_register_info_stdarg_gpr, s390_optimize_register_info): New
5497         function.
5498         (s390_frame_info): Do gpr slot allocation here now.  stdarg does
5499         not imply a stack frame.
5500         (s390_init_frame_layout): Remove variable clobbered_regs.
5501         (s390_update_register_info): Remove function.
5502         (s390_hard_regno_rename_ok): Call-saved regs without a save slot
5503         cannot be used for register renaming.
5504         (s390_hard_regno_scratch_ok): New function.
5505         (TARGET_HARD_REGNO_SCRATCH_OK): Define target hook.
5506         (s390_initial_elimination_offset): Change offset calculation of
5507         the return address pointer.
5508         (save_gprs): Deal with only r6 being saved from the call-saved regs.
5509         (restore_gprs): Set frame related flag.
5510         (s390_save_gprs_to_fprs, s390_restore_gprs_from_fprs): New functions.
5511         (s390_emit_prologue): Call s390_register_info instead of
5512         s390_update_frame_layout.  Call s390_save_gprs_to_fprs.
5513         (s390_emit_epilogue): Call s390_restore_gprs_from_fprs.
5514         (s390_optimize_prologue): Call s390_optimize_register_info.
5515         Try to remove also FPR slot save/restore INSNs.  Remove frame
5516         related flags from restore INSNs.
5518 2013-10-08  DJ Delorie  <dj@redhat.com>
5520         * config/rl78/rl78-expand.md (movqi): use operands[] not operandN.
5521         (movhi): Likewise.
5523         * config/rl78/rl78.c (rl78_print_operand_1): Change %c to %C to
5524         avoid conflict with the MI use of %c.
5525         * config/rl78/rl78-real.md: change %c to %C throughout.
5526         * config/rl78/rl78-virt.md: Likewise.
5528 2013-10-08  Jan Hubicka  <jh@suse.cz>
5530         * config/i386/i386.c (ix86_option_override_internal): Switch
5531         to SSE math for -ffast-math when target ISA supports SSE2.
5533 2013-10-08  Andrew MacLeod  <amacleod@redhat.com>
5535         * tree-flow.h: Remove some prototypes.
5536         * tree.h: Remove some protypes, add a couple.
5537         * tree.c (using_eh_for_cleanups_flag, using_eh_for_cleanups,
5538         using_eh_for_cleanups_p): Add interface routines for front ends.
5539         * tree-eh.h: New file.  Add protoptyes.
5540         * tree-eh.c (using_eh_for_cleanups_p, using_eh_for_cleanups): Delete.
5541         (add_stmt_to_eh_lp_fn): Make static.
5542         (lower_try_finally): Use new using_eh_for_cleanups_p.
5543         * emit-rtl.c: Include tree-eh.h.
5544         * gimple.h: Include tree-eh.h.
5546 2013-10-08  Marc Glisse  <marc.glisse@inria.fr>
5548         PR tree-optimization/58480
5549         * tree-vrp.c (infer_nonnull_range): New function.
5550         (infer_value_range): Call infer_nonnull_range.
5552 2013-10-08  Dehao Chen  <dehao@google.com>
5554         PR tree-optimization/58619
5555         * tree-inline.c (copy_phis_for_bb): Combine location data
5556         only if non-null.
5558 2013-10-08  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
5560         PR target/58423
5561         * config/arm/arm.c (arm_emit_ldrd_pop): Attach
5562         RTX_FRAME_RELATED_P on INSN.
5564 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5566         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const_le): New.
5567         (altivec_expand_vec_perm_const): Call it.
5569 2013-10-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
5571         * config/rs6000/vector.md (mov<mode>): Emit permuted move
5572         sequences for LE VSX loads and stores at expand time.
5573         * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_move): New
5574         prototype.
5575         * config/rs6000/rs6000.c (rs6000_const_vec): New.
5576         (rs6000_gen_le_vsx_permute): New.
5577         (rs6000_gen_le_vsx_load): New.
5578         (rs6000_gen_le_vsx_store): New.
5579         (rs6000_gen_le_vsx_move): New.
5580         * config/rs6000/vsx.md (*vsx_le_perm_load_v2di): New.
5581         (*vsx_le_perm_load_v4si): New.
5582         (*vsx_le_perm_load_v8hi): New.
5583         (*vsx_le_perm_load_v16qi): New.
5584         (*vsx_le_perm_store_v2di): New.
5585         (*vsx_le_perm_store_v4si): New.
5586         (*vsx_le_perm_store_v8hi): New.
5587         (*vsx_le_perm_store_v16qi): New.
5588         (*vsx_xxpermdi2_le_<mode>): New.
5589         (*vsx_xxpermdi4_le_<mode>): New.
5590         (*vsx_xxpermdi8_le_V8HI): New.
5591         (*vsx_xxpermdi16_le_V16QI): New.
5592         (*vsx_lxvd2x2_le_<mode>): New.
5593         (*vsx_lxvd2x4_le_<mode>): New.
5594         (*vsx_lxvd2x8_le_V8HI): New.
5595         (*vsx_lxvd2x16_le_V16QI): New.
5596         (*vsx_stxvd2x2_le_<mode>): New.
5597         (*vsx_stxvd2x4_le_<mode>): New.
5598         (*vsx_stxvd2x8_le_V8HI): New.
5599         (*vsx_stxvd2x16_le_V16QI): New.
5601 2013-10-07  Renlin Li  <Renlin.Li@arm.com>
5603         * config/arm/arm-cores.def (cortex-a53): Use cortex tuning.
5605 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5607         * config/s390/s390.c (s390_register_info): Make the call-saved FPR
5608         loop to work also for 31bit ABI.
5609         Save the stack pointer for frame_size > 0.
5611 2013-10-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
5613         * config/s390/s390.md ("tbegin", "tbegin_nofloat", "tbegin_retry")
5614         ("tbegin_retry_nofloat", "tend", "tabort", "tx_assist"): Remove
5615         constraint letters from expanders.
5616         ("tbegin_retry", "tbegin_retry_nofloat"): Change predicate of the
5617         retry count to general_operand.
5618         ("tabort"): Give operand 0 a mode.
5619         ("tabort_1"): Add mode and constraint letter for operand 0.
5620         * doc/extend.texi: Fix protoype of __builtin_non_tx_store.
5622 2013-10-04  Jeff Law  <law@redhat.com>
5624         * tree-ssa-threadedge.c: Fix some trailing whitespace problems.
5626         * tree-ssa-threadedge.c (thread_through_normal_block): Broken
5627         out of ...
5628         (thread_across_edge): Here.  Call it.
5630 2013-10-04  Cary Coutant  <ccoutant@google.com>
5632         * dwarf2out.c (dw_sra_loc_expr): Release addr_table entries when
5633         discarding a location list expression (or a piece of one).
5635 2013-10-03  Jan Hubicka  <jh@suse.cz>
5637         * config/i386/i386.c (ix86_issue_rate): Pentium4, Nocona has issue
5638         rate of 2.  Core2, Corei7 and Haswell has issue rate of 4.
5639         (ix86_adjust_cost): Remove Atom case; fix core2/corei7/Haswell case.
5641 2013-10-03  Jan Hubicka  <jh@suse.cz>
5643         * config/i386/i386.c (ix86_option_override_internal): Do not enable
5644         accumulate-outgoing-args when producing unwind info.
5646 2013-10-03  Wei Mi  <wmi@google.com>
5648         * lra-constraints.c (insert_move_for_subreg): New function
5649         extracted from simplify_operand_subreg.
5650         (simplify_operand_subreg): Add reload for paradoxical subreg.
5652 2013-10-03  Rong Xu  <xur@google.com>
5654         * ipa-inline-analysis.c (find_foldable_builtin_expect): Find
5655         the candidate of builtin_expect such that we should fix the
5656         size/time estimation.
5657         (estimate_function_body_sizes): Do the acutally size/time fix-up
5658         for builtin_expect.
5660 2013-10-03  Rong Xu  <xur@google.com>
5662         * predict.c (tree_predict_by_opcode): Get the probability
5663         for builtin_expect from param builtin_expect_probability.
5664         * params.def (BUILTIN_EXPECT_PROBABILITY): New parameter.
5665         * predict.def (PRED_BUILTIN_EXPECT_RELAXED): Fix comments.
5666         * doc/invoke.texi: Add documentation for builtin-expect-probability.
5668 2013-10-03  Marc Glisse  <marc.glisse@inria.fr>
5670         PR c++/19476
5671         * common.opt (fcheck-new): Moved from c.opt. Make it 'Common'.
5672         * calls.c (alloca_call_p): Use get_callee_fndecl.
5673         * fold-const.c (tree_expr_nonzero_warnv_p): Handle operator new.
5674         * tree-vrp.c (gimple_stmt_nonzero_warnv_p, stmt_interesting_for_vrp):
5675         Likewise.
5676         (vrp_visit_stmt): Remove duplicated code.
5678 2013-10-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
5680         * config/rs6000/rs6000-builtin.def (XSRDPIM): Use floatdf2,
5681         ceildf2, btruncdf2, instead of vsx_* name.
5683         * config/rs6000/vsx.md (vsx_add<mode>3): Change arithmetic
5684         iterators to only do V2DF and V4SF here.  Move the DF code to
5685         rs6000.md where it is combined with SF mode.  Replace <VSv> with
5686         just 'v' since only vector operations are handled with these insns
5687         after moving the DF support to rs6000.md.
5688         (vsx_sub<mode>3): Likewise.
5689         (vsx_mul<mode>3): Likewise.
5690         (vsx_div<mode>3): Likewise.
5691         (vsx_fre<mode>2): Likewise.
5692         (vsx_neg<mode>2): Likewise.
5693         (vsx_abs<mode>2): Likewise.
5694         (vsx_nabs<mode>2): Likewise.
5695         (vsx_smax<mode>3): Likewise.
5696         (vsx_smin<mode>3): Likewise.
5697         (vsx_sqrt<mode>2): Likewise.
5698         (vsx_rsqrte<mode>2): Likewise.
5699         (vsx_fms<mode>4): Likewise.
5700         (vsx_nfma<mode>4): Likewise.
5701         (vsx_copysign<mode>3): Likewise.
5702         (vsx_btrunc<mode>2): Likewise.
5703         (vsx_floor<mode>2): Likewise.
5704         (vsx_ceil<mode>2): Likewise.
5705         (vsx_smaxsf3): Delete scalar ops that were moved to rs6000.md.
5706         (vsx_sminsf3): Likewise.
5707         (vsx_fmadf4): Likewise.
5708         (vsx_fmsdf4): Likewise.
5709         (vsx_nfmadf4): Likewise.
5710         (vsx_nfmsdf4): Likewise.
5711         (vsx_cmpdf_internal1): Likewise.
5713         * config/rs6000/rs6000.h (TARGET_SF_SPE): Define macros to make it
5714         simpler to select whether a target has SPE or traditional floating
5715         point support in iterators.
5716         (TARGET_DF_SPE): Likewise.
5717         (TARGET_SF_FPR): Likewise.
5718         (TARGET_DF_FPR): Likewise.
5719         (TARGET_SF_INSN): Macros to say whether floating point support
5720         exists for a given operation for expanders.
5721         (TARGET_DF_INSN): Likewise.
5723         * config/rs6000/rs6000.c (Ftrad): New mode attributes to allow
5724         combining of SF/DF mode operations, using both traditional and VSX
5725         registers.
5726         (Fvsx): Likewise.
5727         (Ff): Likewise.
5728         (Fv): Likewise.
5729         (Fs): Likewise.
5730         (Ffre): Likewise.
5731         (FFRE): Likewise.
5732         (abs<mode>2): Combine SF/DF modes using traditional floating point
5733         instructions.  Add support for using the upper DF registers with
5734         VSX support, and SF registers with power8-vector support.  Update
5735         expanders for operations supported by both the SPE and traditional
5736         floating point units.
5737         (abs<mode>2_fpr): Likewise.
5738         (nabs<mode>2): Likewise.
5739         (nabs<mode>2_fpr): Likewise.
5740         (neg<mode>2): Likewise.
5741         (neg<mode>2_fpr): Likewise.
5742         (add<mode>3): Likewise.
5743         (add<mode>3_fpr): Likewise.
5744         (sub<mode>3): Likewise.
5745         (sub<mode>3_fpr): Likewise.
5746         (mul<mode>3): Likewise.
5747         (mul<mode>3_fpr): Likewise.
5748         (div<mode>3): Likewise.
5749         (div<mode>3_fpr): Likewise.
5750         (sqrt<mode>3): Likewise.
5751         (sqrt<mode>3_fpr): Likewise.
5752         (fre<Fs>): Likewise.
5753         (rsqrt<mode>2): Likewise.
5754         (cmp<mode>_fpr): Likewise.
5755         (smax<mode>3): Likewise.
5756         (smin<mode>3): Likewise.
5757         (smax<mode>3_vsx): Likewise.
5758         (smin<mode>3_vsx): Likewise.
5759         (negsf2): Delete SF operations that are merged with DF.
5760         (abssf2): Likewise.
5761         (addsf3): Likewise.
5762         (subsf3): Likewise.
5763         (mulsf3): Likewise.
5764         (divsf3): Likewise.
5765         (fres): Likewise.
5766         (fmasf4_fpr): Likewise.
5767         (fmssf4_fpr): Likewise.
5768         (nfmasf4_fpr): Likewise.
5769         (nfmssf4_fpr): Likewise.
5770         (sqrtsf2): Likewise.
5771         (rsqrtsf_internal1): Likewise.
5772         (smaxsf3): Likewise.
5773         (sminsf3): Likewise.
5774         (cmpsf_internal1): Likewise.
5775         (copysign<mode>3_fcpsgn): Add VSX/power8-vector support.
5776         (negdf2): Delete DF operations that are merged with SF.
5777         (absdf2): Likewise.
5778         (nabsdf2): Likewise.
5779         (adddf3): Likewise.
5780         (subdf3): Likewise.
5781         (muldf3): Likewise.
5782         (divdf3): Likewise.
5783         (fred): Likewise.
5784         (rsqrtdf_internal1): Likewise.
5785         (fmadf4_fpr): Likewise.
5786         (fmsdf4_fpr): Likewise.
5787         (nfmadf4_fpr): Likewise.
5788         (nfmsdf4_fpr): Likewise.
5789         (sqrtdf2): Likewise.
5790         (smaxdf3): Likewise.
5791         (smindf3): Likewise.
5792         (cmpdf_internal1): Likewise.
5793         (lrint<mode>di2): Use TARGET_<MODE>_FPR macro.
5794         (btrunc<mode>2): Delete separate expander, and combine with the
5795         insn and add VSX instruction support.  Use TARGET_<MODE>_FPR.
5796         (btrunc<mode>2_fpr): Likewise.
5797         (ceil<mode>2): Likewise.
5798         (ceil<mode>2_fpr): Likewise.
5799         (floor<mode>2): Likewise.
5800         (floor<mode>2_fpr): Likewise.
5801         (fma<mode>4_fpr): Combine SF and DF fused multiply/add support.
5802         Add support for using the upper registers with VSX and
5803         power8-vector.  Move insns to be closer to the define_expands. On
5804         VSX systems, prefer the traditional form of FMA over the VSX
5805         version, since the traditional form allows the target not to
5806         overlap with the inputs.
5807         (fms<mode>4_fpr): Likewise.
5808         (nfma<mode>4_fpr): Likewise.
5809         (nfms<mode>4_fpr): Likewise.
5811 2013-10-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
5812             Richard Earnshaw  <richard.earnshaw@arm.com>
5814         * config/arm/aarch-common-protos.h (struct alu_cost_table): New.
5815         (struct mult_cost_table): Likewise.
5816         (struct mem_cost_table): Likewise.
5817         (struct fp_cost_table): Likewise.
5818         (struct vector_cost_table): Likewise.
5819         (cpu_cost_table): Likewise.
5820         * config/arm/arm.opt (mold-rts-costs): New option.
5821         (mnew-generic-costs): Likewise.
5822         * config/arm/arm.c (generic_extra_costs): New table.
5823         (cortexa15_extra_costs): Likewise.
5824         (arm_slowmul_tune): Use NULL as new costs.
5825         (arm_fastmul_tune): Likewise.
5826         (arm_strongarm_tune): Likewise.
5827         (arm_xscale_tune): Likewise.
5828         (arm_9e_tune): Likewise.
5829         (arm_v6t2_tune): Likewise.
5830         (arm_cortex_a5_tune): Likewise.
5831         (arm_cortex_a9_tune): Likewise.
5832         (arm_v6m_tune): Likewise.
5833         (arm_fa726te_tune): Likewise.
5834         (arm_cortex_a15_tune): Use cortex15_extra_costs.
5835         (arm_cortex_tune): Use generict_extra_costs.
5836         (shifter_op_p): New function.
5837         (arm_unspec_cost): Likewise.
5838         (LIBCALL_COST): Define.
5839         (arm_new_rtx_costs): New function.
5840         (arm_rtx_costs): Use arm_new_rtx_costs when core-specific
5841         table is available. Use old costs otherwise unless mnew-generic-costs
5842         is specified.
5843         * config/arm/arm-protos.h (tune_params): Add insn_extra_cost field.
5844         (cpu_cost_table): Declare.
5846 2013-10-03  Marcus Shawcroft  <marcus.shawcroft@arm.com>
5848         PR target/58460
5849         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>)
5850         (*subs_mul_imm_<mode>)
5851         (*add_<shift>_<mode>, *add_<shift>_si_uxtw,*add_mul_imm_<mode>)
5852         (*sub_<shift>_<mode>)
5853         (*sub_<shift>_si_uxtw,*sub_mul_imm_<mode>, *sub_mul_imm_si_uxtw):
5854         Remove k constraint.
5856 2013-10-03  Ian Bolton  <ian.bolton@arm.com>
5858         * config/aarch64/aarch64.c (aarch64_secondary_reload): Remove legacy
5859         code.
5860         * config/aarch64/aarch64.md (reload_sp_immediate): Likewise.
5862 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5864         * predict.c (probably_never_executed): New function.
5865         (probably_never_executed_bb_p): Invoke probably_never_executed.
5866         (probably_never_executed_edge_p): Ditto.
5867         * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5868         Treat profile insanities conservatively.
5870 2013-10-02  John David Anglin  <danglin@gcc.gnu.org>
5872         * config.gcc (hppa*64*-*-linux*): Don't add pa/t-linux to tmake_file.
5874 2013-10-02  Vladimir Makarov  <vmakarov@redhat.com>
5876         * lra-constraints.c (process_alt_operand): Calculate scratch_p and
5877         use it.  Use smaller increase for scratch.  Don't increase reject
5878         for early clobber scratch.
5879         * lra-eliminations.c (eliminate_regs_in_insn): Remove all insns
5880         setting eliminated regs except setting fp from hfp.
5881         (lra_eliminate): Check lra_insn_recog_data on NULL.
5883 2013-10-02  Michael Meissner  <meissner@linux.vnet.ibm.com>
5885         PR target/58587
5886         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Turn off
5887         setting -mvsx-timode by default until the underlying problem is fixed.
5888         (RS6000_CPU, power7 defaults): Likewise.
5890 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5892         * config/x-linux (host-linux.o): Remove header dependencies.
5893         Use $(COMPILE) and $(POSTCOMPILE).
5894         * config/t-linux-android (linux-android.o): Ditto.
5896 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5898         * Makefile.in (expmed.o-warn): Remove.
5900 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5902         * graphite-scop-detection.c: Include tree-ssa-propagate,h.
5903         * graphite-sese-to-poly.c: Include tree-ssa-propagate.h.
5905 2013-10-02  Teresa Johnson  <tejohnson@google.com>
5907         * dojump.c (do_jump_1): Divide probability between
5908         both conditions of a TRUTH_ANDIF_EXPR/TRUTH_ORIF_EXPR.
5910 2013-10-02  Tom Tromey  <tromey@redhat.com>
5912         * Makefile.in (DRIVER_DEFINES): Use $(if), not $(and).
5914 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5916         * tree-flow.h: Remove some prototypes.
5917         * tree-ssa-dce.c (mark_virtual_operand_for_renaming,
5918         mark_virtual_phi_result_for_renaming): Move to tree-into-ssa.c.
5919         * tree-into-ssa.c (mark_virtual_operand_for_renaming,
5920         mark_virtual_phi_result_for_renaming): Relocate here.
5921         * tree-into-ssa.h: Add prototypes.
5922         * tree-ssa-phiopt.c: (tree_ssa_phiopt_worker) Use
5923         single_pred_before_succ_order.
5924         (blocks_in_phiopt_order): Rename and move to cfganal.c.
5925         (nonfreeing_call_p) Move to gimple.c.
5926         * cfganal.c (single_pred_before_succ_order): Move and renamed from
5927         tree-ssa-phiopt.c.
5928         * basic-block.h (single_pred_before_succ_order): Add prototype.
5929         * gimple.c (nonfreeing_call_p): Relocate here.
5930         * gimple.h: Add prototype.
5931         * tree-ssa-ifcombine.c: Include tree-ssa-phiopt.h.
5932         * tree-ssa-dom.h: New file.  Relocate prototypes here.
5933         * tree-ssa.h: Include tree-ssa-dom.h.
5935 2013-10-02  Uros Bizjak  <ubizjak@gmail.com>
5937         * config/i386/x-i386 (driver-i386.o): Remove header dependencies.
5938         Use $(COMPILE) and $(POSTCOMPILE).
5940         * config/alpha/x-alpha (driver-alpha.o): Ditto.
5942 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5944         * tree-flow.h: Remove some prototypes.
5945         * gimple-fold.h: Add prototypes from gimple.h and tree-flow.h.
5946         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5947         * tree-ssa-copy.c (may_propagate*, propagate_value, replace_exp,
5948         propagate_tree_value*): Move from here to...
5949         * tree-ssa-propagate.c (may_propagate*, propagate_value, replace_exp,
5950         propagate_tree_value*): Relocate here.
5951         * tree-ssa-propagate.h: Relocate prototypes from tree-flow.h.
5952         * gimple.h: Include gimple-fold.h, move prototypes into gimple-fold.h.
5953         * gimple-fold.c: Remove gimple-fold.h from include list.
5954         * tree-vrp.c: Remove gimple-fold.h from include list.
5955         * tree-ssa-sccvn.c: Remove gimple-fold.h from include list.
5956         * tree-ssa-ccp.c: Remove gimple-fold.h from include list.
5957         * tree-scalar-evolution.c: Add tree-ssa-propagate.h to include list.
5958         * tree-ssa-pre.c: Add tree-ssa-propagate.h to include list.
5959         * sese.c: Add tree-ssa-propagate.h to include list.
5961 2013-10-02  Richard Biener  <rguenther@suse.de>
5963         * tree-loop-distribution.c: Include tree-vectorizer.h for
5964         find_loop_location.
5965         (enum partition_kind): Remove PKIND_REDUCTION.
5966         (struct partition_s): Remove has_writes member, add reduction_p member.
5967         (partition_alloc): Adjust.
5968         (partition_builtin_p): Likewise.
5969         (partition_has_writes): Remove.
5970         (partition_reduction_p): New function.
5971         (partition_merge_into): Likewise.
5972         (generate_code_for_partition): Commonize builtin partition
5973         handling tail.
5974         (rdg_cannot_recompute_vertex_p): Remove.
5975         (already_processed_vertex_p): Likewise.
5976         (rdg_flag_vertex): Do not set has_writes.
5977         (classify_partition): Adjust.
5978         (rdg_build_partitions): Do not set has_writes, treat all
5979         partitions as useful.
5980         (distribute_loop): Record number of library calls generated.  Adjust.
5981         (tree_loop_distribution): Report number of loops and library
5982         calls generated as opt-info.
5984 2013-10-02  Andrew MacLeod  <amacleod@redhat.com>
5986         * tree-flow.h: Include new .h files.  Move prototypes.
5987         * tree-cfgcleanup.h: New file.  Add prototypes from tree-flow.h.
5988         * tree-dfa.h: New File.  Add prototypes from tree-flow.h.
5989         (get_addr_base_and_unit_offset_1) Move from tree-flow-inline.h.
5990         * tree-pretty-print.h: Add prototypes from tree-flow.h.
5991         * tree-into-ssa.h: New File.  Add prototypes from tree-flow.h.
5992         ({debug|dump}*): Move debugging prototypes out of tree-into-ssa.c.
5993         * tree-into-ssa.c ({debug|dump}*): Move prototypes to header file.
5994         * tree.h (get_ref_base_and_extent): Move prototype out.
5995         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Move to
5996         tree-dfa.h.
5997         * gimple-low.h: New File.  Add prototypes from tree-flow.h.
5998         * gimple-low.c (try_catch_may_fallthru, block_may_fallthru): Move to...
5999         * tree.c (try_catch_may_fallthru, block_may_fallthru): Here.
6000         * tree-scalar-evolution.c: Include tree.h.
6001         * sese.c: Include tree.h.
6002         * dumpfile.c: Move gimple-pretty-print.h include after tree.h.
6003         * dwarf2out.c: Include tree-dfa.h.
6004         * tree-chrec.c: Include tree.h.
6005         * tree-data-ref.c: Include tree.h.
6007 2013-10-02  Yufeng Zhang  <yufeng.zhang@arm.com>
6009         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref):
6010         Fix whitespace.
6012 2013-10-02  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
6014         * config/t-sol2 (sol2-c.o): Remove header dependencies.
6015         Use $(COMPILE) and $(POSTCOMPILE).
6016         (sol2-cxx.o): Likewise.
6017         (sol2-stubs.o): Likewise.
6018         (sol2.o): Likewise.
6019         * config/x-solaris (host-solaris.o): Likewise.
6021         * config/sparc/t-sparc (sparc.o): Remove.
6022         (sparc-c.o): Remove header dependencies.
6023         Use $(COMPILE) and $(POSTCOMPILE).
6024         * config/sparc/x-sparc: Likewise.
6026 2013-10-02  Joern Rennecke  <joern.rennecke@embecosm.com>
6028         * config/arc/arc-opts.h: Add 2013 to Copyright years.
6029         * config/arc/arc700.md: Likewise.
6030         * config/arc/arc-modes.def: Likewise.
6031         * config/arc/arc-simd.h: Likewise.
6032         * config/arc/t-arc-uClibc: Likewise.
6033         * config/arc/t-arc-newlib: Likewise.
6035 2013-10-02  Renlin Li  <renlin.li@arm.com>
6037         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6038         plus_constant.
6039         (aarch64_expand_epilogue): Likewise.
6041 2013-10-02  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
6042             Yufeng Zhang  <yufeng.zhang@arm.com>
6044         * gimple-ssa-strength-reduction.c (legal_cast_p_1): Forward
6045         declaration.
6046         (backtrace_base_for_ref): Call get_unwidened with 'base_in' if
6047         'base_in' represent a conversion and legal_cast_p_1 holds; set
6048         'base_in' with the returned value from get_unwidened.
6050 2013-10-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
6052         * config/arm/arm.c (arm_legitimize_reload_address): Explain why
6053         plus_constant is not used.
6055 2013-10-01  Wei Mi  <wmi@google.com>
6057         * config/i386/x86-tune.def (DEF_TUNE): Remove m_CORE_ALL.
6058         * config/i386/i386.md: Add define_peephole2 to
6059         break partial reg stall for cvtss2sd/cvtsd2ss.
6061 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6063         * config/arc/arc.c (pass_arc_ifcvt::clone):
6064         Update for ctxt_ -> m_ctxt change.
6066 2013-10-01  Jeff Law  <law@redhat.com>
6068         * tree-ssa-threadupdate.c (struct redirection_data): Delete
6069         outgoing_edge and intermediate_edge fields.  Instead store the path.
6070         (redirection_data::hash): Hash on the last edge's destination index.
6071         (redirection_data::equal): Check the entire thread path.
6072         (lookup_redirectio_data): Corresponding changes.
6073         (create_edge_and_update_destination_phis): Likewise.
6074         (thread_single_edge): Likewise.
6076 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6077             Diego Novillo <dnovillo@google.com>
6079         * config/arc/simdext.md (UNSPEC_ARC_SIMD_VLD32WH): Delete.
6080         (UNSPEC_ARC_SIMD_VLD32WL): Likewise.
6081         (vld32wh_insn, vld32wl_insn): Delete commented-out old
6082         versions of these patterns.
6084         * doc/extend.texi (long_call/medium_call/short_call): Typo fix.
6085         (__builtin_arc_aligned): Likewise.
6087         * config/arc/arc.md: Expand adc_0 comment stating the intended
6088         purpose and why it isn't ready.
6089         Replace commented out call_value_via_label_mixed with a
6090         plain comment about bl_s.
6092         * config/arc/arc.c (stdio.h): Don't include directly.
6093         (arc_expand_epilogue): Remove [0]: Remove fp_restored_p.
6094         Remove if (1) condition.
6095         (arc_encode_section_info): Fix comment.
6097 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6099         * config/arc/arc.c (arc_conditional_register_usage):
6100         Use ARC_FIRST_SIMD_VR_REG / ARC_LAST_SIMD_VR_REG.
6101         Also set reg_alloc_order for DMA config regs.
6103 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6104             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6106         * doc/install.texi (--with-cpu): Mention ARC.
6107         (arc-*-elf32): New paragraph.
6108         (arc-linux-uclibc): Likewise.
6109         * doc/md.texi (Machine Constraints): Add ARC part.
6110         * doc/invoke.texi: (menu): Add ARC Options.
6111         (Machine Dependent Options) <ARC Options>: Add synopsis.
6112         (node ARC Options): Add.
6113         * doc/extend.texi (long_call / short_call attribute): Add ARC.
6114         (ARC Built-in Functions): New section defining
6115         generic ARC built-in functions.
6116         (ARC SIMD Built-in Functions): New section defining SIMD specific
6117         built-in functions.
6118         (Declaring Attributes of Functions): Extended
6119         description of short_call and long_call attributes for ARC and
6120         added index entries.
6122 2013-10-01  Saurabh Verma  <saurabh.verma@codito.com>
6123             Ramana Radhakrishnan  <ramana.radhakrishnan@codito.com>
6124             Joern Rennecke  <joern.rennecke@embecosm.com>
6125             Muhammad Khurram Riaz  <khurram.riaz@arc.com>
6126             Brendan Kehoe  <brendan@zen.org>
6127             Michael Eager  <eager@eagercon.com>
6128             Simon Cook  <simon.cook@embecosm.com>
6129             Jeremy Bennett  <jeremy.bennett@embecosm.com>
6131         * config/arc, common/config/arc: New directories.
6133 2013-10-01  Joern Rennecke  <joern.rennecke@embecosm.com>
6134             Brendan Kehoe  <brendan@zen.org>
6135             Simon Cook  <simon.cook@embecosm.com>
6137         * config.gcc (arc*-*-elf*, arc*-*-linux-uclibc*): New configurations.
6139 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6141         * tree-ssa-live.h (coalesce_ssa_name): Move Prototype to...
6142         * tree-ssa-coalesce.h: New. Move prototype to here.
6143         * tree-outof-ssa.h: Include tree-ssa-coalesce.h.
6144         * tree-ssa-coalesce.c: Include tree-outof-ssa.h.
6145         (gimple_can_coalesce_p): Move to...
6146         * gimple.c (gimple_can_coalesce_p): Here.
6148 2013-10-01  Andrew MacLeod  <amacleod@redhat.com>
6150         * tree-into-ssa.c (enum need_phi_state): Relocate from tree-flow.h.
6151         (dump_decl_set): Move to gimple.c.
6152         * gimple.h: Don't include tree-ssa-operands.h.
6153         (dump_decl_set): Add prototype.
6154         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6155         Move to gimple-ssa.h.
6156         (phi_ssa_name_p, phi_nodes, phi_nodes_ptr, gimple_phi_arg_def,
6157         gimple_phi_arg_def_ptr, gimple_phi_arg_edge, gimple_phi_arg_location,
6158         gimple_phi_arg_location_from_edge, gimple_phi_arg_set_location,
6159         gimple_phi_arg_has_location): Relocate from tree-flow-inline.h
6160         * gimple.c (walk_stmt_load_store_ops): Use gimple_phi_arg_def rather
6161         than PHI_ARG_DEF.
6162         (dump_decl_set): Relocate here.
6163         * gimple-ssa.h: New file.
6164         (gimple_vuse_op, gimple_vdef_op, update_stmt, update_stmt_if_modified):
6165         Relocate from gimple.h.
6166         * tree-cfg.c (has_zero_uses_1, single_imm_use_1): Move to...
6167         * tree-ssa-operands.c (swap_ssa_operands): Rename from
6168         swap_tree_operands and remove non-ssa path.
6169         (has_zero_uses_1, single_imm_use_1): Relocate from tree-cfg.c.
6170         * tree-ssa-reassoc.c (linearize_expr_tree, repropagate_negates): Use
6171         swap_ssa_operands.
6172         * tree-vect-loop.c (destroy_loop_vec_info, vect_is_slp_reduction,
6173         vect_is_simple_reduction_1): Use swap_ssa_operands.
6174         * tree-flow.h: Move various prototypes to tree-phinodes.h.
6175         (enum need_phi_state): Move to tree-into-ssa.c.
6176         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6177         BREAK_FROM_IMM_USE_STMT): Move to ssa-iterators.h.
6178         (swap_tree_operands): Rename and move prototype to tree-ssa-operands.h.
6179         * tree-flow-inline.h (delink_imm_use, link_imm_use_to_list,
6180         link_imm_use, set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6181         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6182         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6183         num_imm_uses): Move to ssa-iterators.h.
6184         (get_use_from_ptr, get_def_from_ptr): Move to tree-ssa-operands.h
6185         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Move to
6186         tree-phinodes.h.
6187         (op_iter_done, op_iter_next_def, op_iter_next_tree,
6188         clear_and_done_ssa_iter, op_iter_init, op_iter_init_use,
6189         op_iter_init_def, op_iter_init_tree, single_ssa_tree_operand,
6190         single_ssa_use_operand, single_ssa_def_operand, zero_ssa_operands,
6191         num_ssa_operands, delink_stmt_imm_use, single_phi_def,
6192         op_iter_init_phiuse, op_iter_init_phidef, end_imm_use_stmt_p,
6193         end_imm_use_stmt_traverse, move_use_after_head, link_use_stmts_after,
6194         first_imm_use_stmt, next_imm_use_stmt, first_imm_use_on_stmt,
6195         end_imm_use_on_stmt_p, next_imm_use_on_stmt): Move to ssa-iterators.h.
6196         (gimple_phi_arg_def, gimple_phi_arg_def_ptr, gimple_phi_arg_edge,
6197         gimple_phi_arg_location, gimple_phi_arg_location_from_edge,
6198         gimple_phi_arg_set_location, gimple_phi_arg_has_location, phi_nodes,
6199         phi_nodes_ptr, phi_ssa_name_p): Move to gimple.h.
6200         (set_phi_nodes): Move to tree-phinodes.h.
6201         * tree-ssa-operands.h (enum ssa_op_iter_type,
6202         struct ssa_operand_iterator_d, SSA_OP*, FOR_EACH_SSA*, SINGLE_SSA*,
6203         ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS): Move to ssa-iterators.h.
6204         (dump_decl_set): Remove prototype.
6205         (get_use_from_ptr, get_def_from_ptr): Relocate from tree-flow.h.
6206         * tree-phinodes.h: New file.  Move some prototypes from tree-flow.h.
6207         (set_phi_nodes): Relocate from tree-flow-inline.h.
6208         (gimple_phi_arg_imm_use_ptr, phi_arg_index_from_use): Relocate from
6209         tree-flow-inline.h
6210         * tree-ssa.h: Add tree-phinodes.h, gimple-ssa.h, ssa-iterators.h to
6211         include list.  Temporarily add gimple.h to include list.
6212         * ssa-iterators.h: New file.
6213         (struct immediate_use_iterator_d, FOR_EACH_IMM_*,
6214         BREAK_FROM_IMM_USE_STMT): Relocate from tree-flow.h.
6215         (enum ssa_op_iter_type, struct ssa_operand_iterator_d, SSA_OP*,
6216         FOR_EACH_SSA*, SINGLE_SSA*, ZERO_SSA_OPERANDS, NUM_SSA_OPERANDS):
6217         Relocate from tree-ssa-operands.h.
6218         (delink_imm_use, link_imm_use_to_list, link_imm_use,
6219         set_ssa_use_from_ptr, link_imm_use_stmt, relink_imm_use,
6220         relink_imm_use_stmt, end_readonly_imm_use_p, first_readonly_imm_use,
6221         next_readonly_imm_use, has_zero_uses, has_single_use, single_imm_use,
6222         num_imm_uses, get_use_from_ptr, get_def_from_ptr,
6223         phi_arg_index_from_use, op_iter_done, op_iter_next_def,
6224         op_iter_next_tree, clear_and_done_ssa_iter, op_iter_init,
6225         op_iter_init_use, op_iter_init_def, op_iter_init_tree,
6226         single_ssa_tree_operand, single_ssa_use_operand, single_ssa_def_operand,
6227         zero_ssa_operands, num_ssa_operands, delink_stmt_imm_use,
6228         single_phi_def, op_iter_init_phiuse, op_iter_init_phidef,
6229         end_imm_use_stmt_p, end_imm_use_stmt_traverse, move_use_after_head,
6230         link_use_stmts_after, first_imm_use_stmt, next_imm_use_stmt,
6231         first_imm_use_on_stmt, end_imm_use_on_stmt_p, next_imm_use_on_stmt):
6232         Relocate from tree-flow-inline.h.
6233         * tree-outof-ssa.h: Change _SSAEXPAND_H macro to GCC_TREE_OUTOF_SSA_H.
6235 2013-10-01  Vidya Praveen  <vidyapraveen@arm.com>
6237         * aarch64-simd.md
6238         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l2<mode>_internal): Rename to ...
6239         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal): ... this;
6240         Insert '\t' to output template.
6241         (aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_lo_internal): New.
6242         (aarch64_saddl2<mode>, aarch64_uaddl2<mode>): Modify to call
6243         gen_aarch64_<ANY_EXTEND:su><ADDSUB:optab>l<mode>_hi_internal() instead.
6244         (aarch64_ssubl2<mode>, aarch64_usubl2<mode>): Ditto.
6246 2013-10-01  Uros Bizjak  <ubizjak@gmail.com>
6248         * doc/install.texi (Host/target specific installation notes for GCC):
6249         Put @anchor before @heading.
6250         * doc/gcc.texi (titlepage): Use @uref and http:// prefix for website.
6251         Use @email for email addresses.
6253 2013-10-01  Jeff Law  <law@redhat.com>
6255         * tree-ssa-threadedge.c (thread_across_edge): Make path a pointer to
6256         a vec.  Only delete the path if we create one without successfully
6257         registering a jump thread.
6258         * tree-ssa-threadupdate.h (register_jump_thread): Pass in path vector
6259         as a pointer.
6260         * tree-ssa-threadupdate.c (threaded_edges): Remove.  No longer used
6261         (paths): New vector of jump threading paths.
6262         (THREAD_TARGET, THREAD_TARGET2): Remove accessor macros.
6263         (THREAD_PATH): New accessor macro for the entire thread path.
6264         (lookup_redirection_data): Get intermediate and final outgoing edge
6265         from the thread path.
6266         (create_edge_and_update_destination_phis): Copy the threading path.
6267         (ssa_fix_duplicate_block_edges): Get edges and block types from the
6268         jump threading path.
6269         (ssa_redirect_edges): Get edges and block types from the jump threading
6270         path.  Free the path vector.
6271         (thread_block): Get edges from the jump threading path.  Look at the
6272         entire path to see if we thread to a loop exit.  If we cancel a jump
6273         thread request, then free the path vector.
6274         (thread_single_edge): Get edges and block types from the jump threading
6275         path.  Free the path vector.
6276         (thread_through_loop_header): Get edges and block types from the jump
6277         threading path.  Free the path vector.
6278         (mark_threaded_blocks): Iterate over the vector of paths and store
6279         the path on the appropriate edge.  Get edges and block types from the
6280         jump threading path.
6281         (mark_threaded_blocks): Get edges and block types from the jump
6282         threading path.  Free the path vector.
6283         (thread_through_all_blocks): Use the vector of paths rather than
6284         a vector of 3-edge sets.
6285         (register_jump_thread): Accept pointer to a path vector rather
6286         than the path vector itself.  Store the path vector for later use.
6287         Simplify.
6289 2013-10-01  Jakub Jelinek  <jakub@redhat.com>
6290             Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
6292         PR target/58574
6293         * config/s390/s390.c (s390_split_branches): Modify check for table
6294         jump insns.
6295         (s390_chunkify_start): Rearrange table jump insn check in order to
6296         deal with compare and branch insns correctly.
6298 2013-10-01  Kugan Vivekanandarajah  <kuganv@linaro.org>
6300         PR target/58578
6301         Revert
6302         2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
6303         * config/arm/arm.md (arm_ashldi3_1bit):  define_insn into
6304         define_insn_and_split.
6305         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
6306         (shiftsi3_compare): New pattern.
6307         (rrx): New pattern.
6308         * config/arm/unspecs.md (UNSPEC_RRX): New.
6310 2013-10-01  Alan Modra  <amodra@gmail.com>
6312         * stmt.c (expand_asm_operands): Revert part of 2013-09-24 special
6313         casing inout operands.
6315 2013-10-01  Richard Biener  <rguenther@suse.de>
6317         PR tree-optimization/58553
6318         * tree-loop-distribution.c (struct partition_s): Add niter member.
6319         (classify_partition): Populate niter member for the partition
6320         and properly identify whether the relevant store happens before
6321         or after the loop exit.
6322         (generate_memset_builtin): Use niter member from the partition.
6323         (generate_memcpy_builtin): Likewise.
6325 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6327         * vec.h (vec_prefix, vec): Prefix member names with "m_".
6328         * vec.c (vec_prefix::calculate_allocation): Update accordingly.
6330 2013-09-30  Richard Sandiford  <rdsandiford@googlemail.com>
6332         * basic-block.h (edge_list): Prefix member names with "m_".
6333         * context.h (context): Likewise.
6334         * domwalk.h (dom_walker): Likewise.
6335         * gengtype-state.c (s_expr_writer, state_writer): Likewise.
6336         * graphite-sese-to-poly.c (sese_dom_walker): Likewise.
6337         * hash-table.h (hash_table): Likewise.
6338         * machmode.h (bit_field_mode_iterator): Likewise.
6339         * pass_manager.h (pass_list): Likewise.
6340         * tree-into-ssa.c (mark_def_dom_walker): Likewise.
6341         * tree-pass.h (pass_data): Likewise.
6342         * tree-ssa-dom.c (dom_opt_dom_walker): Likewise.
6343         * tree-ssa-phiopt.c (nontrapping_dom_walker): Likewise,
6344         * tree-ssa-uncprop.c (uncprop_dom_walker): Likewise.
6345         * asan.c (pass_data_asan): Update accordingly.
6346         * cfganal.c (control_dependences::find_control_dependence): Likewise.
6347         (control_dependences::control_dependences): Likewise.
6348         (control_dependences::~control_dependences): Likewise.
6349         (control_dependences::~control_dependences): Likewise.
6350         (control_dependences::get_edges_dependent_on): Likewise.
6351         * cgraphbuild.c (pass_data_rebuild_cgraph_edges::clone): Likewise.
6352         (pass_data_remove_cgraph_callee_edges::clone): Likewise.
6353         * context.c (gcc::context::context): Likewise.
6354         * cprop.c (pass_rtl_cprop::clone): Likewise.
6355         * domwalk.c (dom_walker::walk): Likewise.
6356         * ipa-inline-analysis.c (pass_inline_parameters::clone): Likewise.
6357         * ipa-pure-const.c (pass_local_pure_const::clone): Likewise.
6358         * mode-switching.c (pass_mode_switching::clone): Likewise.
6359         * passes.c (opt_pass::opt_pass): Likewise.
6360         (pass_manager::pass_manager): Likewise.
6361         * predict.c (pass_strip_predict_hints::clone): Likewise.
6362         * recog.c (pass_data pass_data_peephole2::clone): Likewise.
6363         (pass_split_all_insns::clone): Likewise.
6364         * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6365         Likewise.
6366         (bit_field_mode_iterator::next_mode): Likewise.
6367         (bit_field_mode_iterator::prefer_smaller_modes): Likewise.
6368         * tree-cfg.c (pass_split_crit_edges::clone): Likewise.
6369         * tree-cfgcleanup.c (pass_merge_phi::clone): Likewise.
6370         * tree-complex.c (pass_lower_complex::clone): Likewise.
6371         * tree-eh.c (pass_cleanup_eh::clone): Likewise.
6372         * tree-object-size.c (pass_object_sizes::clone): Likewise.
6373         * tree-optimize.c (pass_fixup_cfg::clone): Likewise.
6374         * tree-ssa-ccp.c (pass_data_ccp::clone): Likewise.
6375         (pass_fold_builtins::clone): Likewise.
6376         * tree-ssa-copy.c (pass_data_copy_prop::clone): Likewise.
6377         * tree-ssa-copyrename.c (pass_rename_ssa_copies::clone): Likewise.
6378         * tree-ssa-dce.c (pass_dce::clone, pass_dce_loop::clone): Likewise.
6379         (pass_cd_dce::clone): Likewise.
6380         * tree-ssa-dom.c (pass_dominator::clone): Likewise.
6381         (pass_phi_only_cprop::clone): Likewise.
6382         * tree-ssa-dse.c (pass_dse::clone): Likewise.
6383         * tree-ssa-forwprop.c (pass_forwprop::clone): Likewise.
6384         * tree-ssa-loop.c (pass_lim::clone): Likewise.
6385         * tree-ssa-phiopt.c (pass_phiopt::clone): Likewise.
6386         * tree-ssa-pre.c (pass_fre::clone): Likewise.
6387         * tree-ssa-reassoc.c (pass_reassoc::clone): Likewise.
6388         * tree-ssa-uninit.c (pass_late_warn_uninitialized::clone): Likewise.
6389         * tree-tailcall.c (pass_tail_recursion::clone): Likewise.
6390         * tree-vect-generic.c (pass_lower_vector_ssa::clone): Likewise.
6391         * tree-vrp.c (pass_vrp::clone): Likewise.
6392         * tsan.c (pass_tsan::clone): Likewise.
6394 2013-09-30  Jakub Jelinek  <jakub@redhat.com>
6396         PR middle-end/58564
6397         * fold-const.c (tree_unary_nonnegative_warnv_p): Use
6398         INTEGRAL_TYPE_P (t) instead of TREE_CODE (t) == INTEGER_TYPE.
6400         PR middle-end/58564
6401         * fold-const.c (fold_ternary_loc): For A < 0 : <sign bit of A> : 0
6402         optimization, punt if sign_bit_p looked through any zero extension.
6404 2013-09-30  Teresa Johnson  <tejohnson@google.com>
6406         * tree-ssa-threadupdate.c (ssa_fix_duplicate_block_edges):
6407         Update redirected out edge count in joiner case.
6408         (ssa_redirect_edges): Common the joiner and non-joiner cases
6409         so that joiner case gets profile updates.
6411 2013-09-30  Richard Biener  <rguenther@suse.de>
6413         PR tree-optimization/58554
6414         * tree-loop-distribution.c (classify_partition): Require
6415         unconditionally executed stores for memcpy and memset recognition.
6416         (tree_loop_distribution): Calculate dominance info.
6418 2013-09-30  Venkataramanan Kumar  <venkataramanan.kumar@linaro.org>
6420         * config/aarch64/aarch64.h (MCOUNT_NAME): Define.
6421         (NO_PROFILE_COUNTERS): Likewise.
6422         (PROFILE_HOOK): Likewise.
6423         (FUNCTION_PROFILER): Likewise.
6424         * config/aarch64/aarch64.c (aarch64_function_profiler): Remove.
6426 2013-09-30  Iain Sandoe  <iain@codesourcery.com>
6428         * config/rs6000/darwin.md (load_macho_picbase_si): Wrap machopic
6429         calls and defines in TARGET_MACHO conditional.
6430         (load_macho_picbase_di): Likewise.
6431         (reload_macho_picbase): Likewise.
6432         (reload_macho_picbase_si): Likewise.
6433         (reload_macho_picbase_di): Likewise.
6434         (nonlocal_goto_receiver): Likewise.
6436 2013-09-30  Nick Clifton  <nickc@redhat.com>
6438         * config/msp430/msp430.c (msp430x_names): New array.  Lists MCUs
6439         that use the MSP430X ISA.
6440         (msp430_option_override): Scan -mmcu command line option for any
6441         MCU name that supports the MSP430X ISA.
6442         * config/msp430/t-msp430 (MULTILIB_MATCHES): Add matches for known
6443         -mmcu options which enable the MSP430X ISA.
6445 2013-09-30  Richard Biener  <rguenther@suse.de>
6447         PR middle-end/58532
6448         * tree-cfg.c (make_abnormal_goto_edges): Skip debug statements
6449         before looking for setjmp-like calls.
6451 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6453         PR target/10901
6454         * config/darwin-protos.h (machopic_get_function_picbase): New.
6455         * config/darwin.c (machopic_get_function_picbase): New.
6456         * config/rs6000/darwin.md (load_macho_picbase_si): Update picbase
6457         label for a new func.  (load_macho_picbase_di): Likewise.
6458         (reload_macho_picbase): New expand.
6459         (reload_macho_picbase_si): New insn.
6460         (reload_macho_picbase_di): New insn.
6461         (nonlocal_goto_receiver): New define and split.
6462         * config/rs6000/rs6000.md (unspec enum): Add UNSPEC_RELD_MPIC.
6463         (unspecv enum): Add UNSPECV_NLGR.
6465 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6467         * config/rs6000/rs6000.c (rs6000_init_dwarf_reg_sizes_extra): Ensure
6468         that altivec registers are correctly sized on Darwin.
6470 2013-09-29  Iain Sandoe  <iain@codesourcery.com>
6472         * config/t-darwin (darwin.o, darwin-c.o, darwin-f.o,
6473         darwin-driver.o): Use COMPILE and POSTCOMPILE.
6474         * config/x-darwin (host-darwin.o): Likewise.
6475         * config/i386/x-darwin (host-i386-darwin.o): Likewise.
6476         * config/rs6000/x-darwin (host-ppc-darwin.o): Likewise.
6477         * config/rs6000/x-darwin64 (host-ppc64-darwin.o): Likewise.
6479 2013-09-29  Uros Bizjak  <ubizjak@gmail.com>
6481         * doc/invoke.texi: Fix usage of @tie{} command.
6483 2013-09-29  Eric Botcazou  <ebotcazou@adacore.com>
6485         * config/sparc/sync.md: Add peephole for consecutive memory barriers.
6487 2013-09-28  Jan Hubicka  <jh@suse.cz>
6489         * config/i386/x86-tune.def: Add documentation for each of the options;
6490         add whitespace.
6492 2013-09-28  Jan Hubicka  <jh@suse.cz>
6494         * x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL): Enable for
6495         generic.
6496         (X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Likewise.
6497         (X86_TUNE_FOUR_JUMP_LIMIT): Drop for generic and buldozer.
6498         (X86_TUNE_PAD_RETURNS): Drop for buldozer chips.
6499         (X86_TUNE_AVOID_VECTOR_DECODE): Drop for generic.
6500         (X86_TUNE_REASSOC_FP_TO_PARALLEL): Enable for generic.
6502 2013-09-28  Richard Sandiford  <rdsandiford@googlemail.com>
6504         * alloc-pool.c, asan.c, auto-inc-dec.c, basic-block.h, bb-reorder.c,
6505         bitmap.c, bitmap.h, bt-load.c, builtins.c, calls.c, cfgcleanup.c,
6506         cfgexpand.c, cfghooks.c, cfgloop.c, cfgloopmanip.c, cfgrtl.c, cgraph.c,
6507         cgraph.h, cgraphbuild.c, cgraphclones.c, cgraphunit.c, collect2.c,
6508         combine-stack-adj.c, combine.c, compare-elim.c, context.c, context.h,
6509         cprop.c, cse.c, cselib.c, dbxout.c, dce.c, defaults.h, df-core.c,
6510         df-problems.c, df-scan.c, df.h, diagnostic.c, double-int.c, dse.c,
6511         dumpfile.c, dwarf2asm.c, dwarf2cfi.c, dwarf2out.c, emit-rtl.c,
6512         errors.c, except.c, expmed.c, expr.c, file-find.c, final.c,
6513         fixed-value.c, fold-const.c, function.c, fwprop.c, gcc-ar.c, gcc.c,
6514         gcov-io.c, gcov-io.h, gcov.c, gcse.c, genattr-common.c, genattr.c,
6515         genattrtab.c, genautomata.c, genconfig.c, genemit.c, genextract.c,
6516         genflags.c, gengenrtl.c, gengtype-state.c, gengtype.c, genmodes.c,
6517         genopinit.c, genoutput.c, genpeep.c, genpreds.c, genrecog.c,
6518         gensupport.c, ggc-common.c, ggc-page.c, gimple-fold.c, gimple-low.c,
6519         gimple-pretty-print.c, gimple-ssa-strength-reduction.c, gimple.c,
6520         gimple.h, godump.c, graphite-clast-to-gimple.c,
6521         graphite-optimize-isl.c, graphite-poly.h, graphite-sese-to-poly.c,
6522         graphite.c, haifa-sched.c, hash-table.c, hash-table.h, hwint.c,
6523         hwint.h, ifcvt.c, incpath.c, init-regs.c, input.h, intl.c, intl.h,
6524         ipa-cp.c, ipa-devirt.c, ipa-inline-analysis.c, ipa-inline.c,
6525         ipa-profile.c, ipa-pure-const.c, ipa-reference.c, ipa-split.c,
6526         ipa-utils.c, ipa.c, ira-build.c, ira.c, jump.c, loop-doloop.c,
6527         loop-init.c, loop-invariant.c, loop-iv.c, lower-subreg.c, lto-cgraph.c,
6528         lto-streamer-in.c, lto-streamer-out.c, lto-wrapper.c, mcf.c,
6529         mode-switching.c, modulo-sched.c, omp-low.c, optabs.c, opts.c,
6530         pass_manager.h, passes.c, plugin.c, postreload-gcse.c, postreload.c,
6531         predict.c, prefix.c, pretty-print.c, print-rtl.c, print-tree.c,
6532         profile.c, read-md.c, real.c, real.h, recog.c, ree.c, reg-stack.c,
6533         regcprop.c, reginfo.c, regmove.c, regrename.c, regs.h, regstat.c,
6534         reload1.c, reorg.c, rtl.c, rtl.h, rtlanal.c, sbitmap.c, sched-rgn.c,
6535         sdbout.c, sel-sched-ir.c, sel-sched.c, sparseset.c, stack-ptr-mod.c,
6536         statistics.c, stmt.c, stor-layout.c, store-motion.c, streamer-hooks.h,
6537         system.h, target-hooks-macros.h, targhooks.c, targhooks.h, toplev.c,
6538         tracer.c, trans-mem.c, tree-browser.c, tree-call-cdce.c, tree-cfg.c,
6539         tree-cfgcleanup.c, tree-complex.c, tree-data-ref.c, tree-data-ref.h,
6540         tree-eh.c, tree-emutls.c, tree-flow.h, tree-if-conv.c, tree-into-ssa.c,
6541         tree-iterator.c, tree-loop-distribution.c, tree-mudflap.c,
6542         tree-nested.c, tree-nomudflap.c, tree-nrv.c, tree-object-size.c,
6543         tree-optimize.c, tree-pass.h, tree-pretty-print.c, tree-profile.c,
6544         tree-scalar-evolution.c, tree-sra.c, tree-ssa-ccp.c,
6545         tree-ssa-coalesce.c, tree-ssa-copy.c, tree-ssa-copyrename.c,
6546         tree-ssa-dce.c, tree-ssa-dom.c, tree-ssa-dse.c, tree-ssa-forwprop.c,
6547         tree-ssa-ifcombine.c, tree-ssa-live.c, tree-ssa-loop-ch.c,
6548         tree-ssa-loop-im.c, tree-ssa-loop-ivopts.c, tree-ssa-loop-prefetch.c,
6549         tree-ssa-loop.c, tree-ssa-math-opts.c, tree-ssa-operands.c,
6550         tree-ssa-phiopt.c, tree-ssa-phiprop.c, tree-ssa-pre.c,
6551         tree-ssa-reassoc.c, tree-ssa-sink.c, tree-ssa-strlen.c,
6552         tree-ssa-structalias.c, tree-ssa-threadedge.c, tree-ssa-threadupdate.c,
6553         tree-ssa-uncprop.c, tree-ssa-uninit.c, tree-ssa.c, tree-ssanames.c,
6554         tree-stdarg.c, tree-switch-conversion.c, tree-tailcall.c,
6555         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
6556         tree-vect-stmts.c, tree-vectorizer.c, tree-vectorizer.h, tree-vrp.c,
6557         tree.c, tree.h, tsan.c, tsystem.h, value-prof.c, var-tracking.c,
6558         varasm.c, vec.h, vmsdbgout.c, vtable-verify.c, web.c: Add missing
6559         whitespace before "(".
6561 2013-09-28  Sandra Loosemore  <sandra@codesourcery.com>
6563         * expr.h (extract_bit_field): Remove packedp parameter.
6564         * expmed.c (extract_fixed_bit_field): Remove packedp parameter
6565         from forward declaration.
6566         (store_split_bit_field): Remove packedp arg from calls to
6567         extract_fixed_bit_field.
6568         (extract_bit_field_1): Remove packedp parameter and packedp
6569         argument from recursive calls and calls to extract_fixed_bit_field.
6570         (extract_bit_field): Remove packedp parameter and corresponding
6571         arg to extract_bit_field_1.
6572         (extract_fixed_bit_field): Remove packedp parameter.  Remove code
6573         to issue warnings.
6574         (extract_split_bit_field): Remove packedp arg from call to
6575         extract_fixed_bit_field.
6576         * expr.c (emit_group_load_1): Adjust calls to extract_bit_field.
6577         (copy_blkmode_from_reg): Likewise.
6578         (copy_blkmode_to_reg): Likewise.
6579         (read_complex_part): Likewise.
6580         (store_field): Likewise.
6581         (expand_expr_real_1): Likewise.
6582         * calls.c (store_unaligned_arguments_into_pseudos): Adjust call
6583         to extract_bit_field.
6584         * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Adjust
6585         call to extract_bit_field.
6586         * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Adjust
6587         call to extract_bit_field.
6588         * doc/invoke.texi (Code Gen Options): Remove mention of warnings
6589         and special packedp behavior from -fstrict-volatile-bitfields
6590         documentation.
6592 2013-09-27  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
6594         * lra-eliminations.c (init_elim_table): Guard value_p.
6596 2013-09-27  Michael Meissner  <meissner@linux.vnet.ibm.com>
6598         * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Allow
6599         DFmode, DImode, and SFmode in the upper VSX registers based on the
6600         -mupper-regs-{df,sf} flags.  Fix wu constraint to be ALTIVEC_REGS
6601         if -mpower8-vector.  Combine -mvsx-timode handling with the rest
6602         of the VSX register handling.
6604         * config/rs6000/rs6000.md (f32_lv): Use %x0 for VSX regsters.
6605         (f32_sv): Likewise.
6606         (zero_extendsidi2_lfiwzx): Add support for loading into the
6607         Altivec registers with -mpower8-vector.  Use wu/wv constraints to
6608         only do VSX memory options on Altivec registers.
6609         (extendsidi2_lfiwax): Likewise.
6610         (extendsfdf2_fpr): Likewise.
6611         (mov<mode>_hardfloat, SF/SD modes): Likewise.
6612         (mov<mode>_hardfloat32, DF/DD modes): Likewise.
6613         (mov<mode>_hardfloat64, DF/DD modes): Likewise.
6614         (movdi_internal64): Likewise.
6616 2013-09-27  Xinliang David Li  <davidxl@google.com>
6618         * opts.c (finish_options): Adjust parameters
6619         according to vect cost model.
6620         (common_handle_option): Set dynamic vect cost
6621         model for FDO.
6622         targhooks.c (default_add_stmt_cost): Compute stmt cost
6623         unconditionally.
6624         * tree-vect-loop.c (vect_estimate_min_profitable_iters):
6625         Use helper function.
6626         * tree-vectorizer.h (unlimited_cost_model): New function.
6627         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use helper function.
6628         * tree-vect-data-refs.c (vect_peeling_hash_insert): Use helper
6629         function.
6630         (vect_enhance_data_refs_alignment): Ditto.
6631         * flag-types.h: New enum.
6632         * common/config/i386/i386-common.c (ix86_option_init_struct):
6633         No need to initialize vect_cost_model flag.
6634         * config/i386/i386.c (ix86_add_stmt_cost): Compute stmt cost
6635         unconditionally.
6637 2013-09-27  Diego Novillo  <dnovillo@google.com>
6639         * gimple.h (enum ssa_mode): Remove.
6641 2013-09-27  Paulo Matos  <pmatos@broadcom.com>
6643         * cfgloop.h (number_of_loops): Fix typo in check for null.
6645 2013-09-27  Jakub Jelinek  <jakub@redhat.com>
6647         PR middle-end/58551
6648         * tree-cfg.c (move_sese_region_to_fn): Also move loops that
6649         are children of outermost saved_cfun's loop, and set it up to
6650         be moved to dest_cfun's outermost loop.  Fix up num_nodes adjustments
6651         if loop != loop0 and SESE region contains bbs that belong to loop0.
6653 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6655         * rtlanal.c (must_be_base_p, must_be_index_p): Delete.
6656         (binary_scale_code_p, get_base_term, get_index_term): New functions.
6657         (set_address_segment, set_address_base, set_address_index)
6658         (set_address_disp): Accept the argument unconditionally.
6659         (baseness): Remove must_be_base_p and must_be_index_p checks.
6660         (decompose_normal_address): Classify as much as possible in the
6661         main loop.
6663 2013-09-27  Richard Sandiford  <rdsandiford@googlemail.com>
6665         * cse.c (count_reg_usage): Handle INT_LIST.
6666         * lra-eliminations.c (lra_eliminate_regs_1): Likewise.
6667         * reginfo.c (reg_scan_mark_refs): Likewise.
6668         * reload1.c (eliminate_regs_1): Likewise.
6670 2013-09-27  Iain Sandoe  <iain@codesourcery.com>
6672         PR middle-end/58547
6673         * rtlanal.c (lsb_bitfield_op_p): Make both parts of the comparison
6674         signed.
6676 2013-09-27  Richard Biener  <rguenther@suse.de>
6678         PR tree-optimization/58459
6679         * tree-ssa-forwprop.c (forward_propagate_addr_expr): Remove
6680         restriction not propagating into loops.
6682 2013-09-26  Florian Weimer  <fw@deneb.enyo.de>
6684         * tree-ssa.h (walk_use_def_chains_fn, walk_use_def_chains): Delete.
6685         * tree-ssa.c (walk_use_def_chains_1, walk_use_def_chains): Delete.
6686         * doc/tree-ssa.texi (Walking use-def chains): Delete.
6688 2013-09-26  Richard Biener  <rguenther@suse.de>
6690         * tree-into-ssa.c (rewrite_into_ssa): Make more SSA names to anonymous.
6692 2013-09-26  Richard Biener  <rguenther@suse.de>
6694         * alias.h (component_uses_parent_alias_set): Rename to ...
6695         (component_uses_parent_alias_set_from): ... this.
6696         * alias.c (component_uses_parent_alias_set): Rename to ...
6697         (component_uses_parent_alias_set_from): ... this and return
6698         the desired parent.
6699         (reference_alias_ptr_type_1): Use the result from
6700         component_uses_parent_alias_set_from instead of stripping
6701         components one at a time.
6702         * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
6704 2013-09-26  Andrew MacLeod  <amacleod@redhat.com>
6706         * tree-ssa-live.h (find_replaceable_exprs, dump_replaceable_exprs):
6707         Move prototypes to...
6708         * tree-ssa-ter.h: New File.  Move prototypes here.
6709         * tree-flow.h (stmt_is_replaceable_p): Remove prototype.
6710         * tree-outof-ssa.h: New. Rename ssaexpand.h, include tree-ssa-ter.h.
6711         * tree-outof-ssa.c (ssa_is_replaceable_p): New.  Refactor common bits
6712         from is_replaceable_p.
6713         * tree-ssa-ter.c (is_replaceable_p, stmt_is_replaceable_p): Delete.
6714         (ter_is_replaceable_p): New.  Use new refactored ssa_is_replaceable_p.
6715         (process_replaceable): Use ter_is_replaceable_p.
6716         (find_replaceable_in_bb): Use ter_is_replaceable_p.
6717         * expr.c (stmt_is_replaceable_p): Relocate from tree-ssa-ter.c.  Use
6718         newly refactored ssa_is_replaceable_p.
6719         * cfgexpand.c: Include tree-outof-ssa.h.
6720         * ssaexpand.h: Delete.
6722 2013-09-26  Andrew MacLeod <amacleod@redhat.com>
6724         * gimple.c (gimple_replace_lhs): Move to tree-ssa.c and rename.
6725         (struct count_ptr_d, count_ptr_derefs, count_uses_and_derefs): Move to
6726         tree-ssa.c
6727         (create_gimple_tmp): Delete.
6728         (get_expr_type, build_assign, build_type_cast): Move to...
6729         * gimple-builder.c: New File.
6730         (get_expr_type): Relocate from gimple.c.
6731         (build_assign, build_type_cast): Change to only create ssanames.
6732         * gimple.h: Move prototypes to...
6733         * gimple-builder.h: New File. Here.
6734         * tree-ssa.h: And here.
6735         * tree-ssa.c (struct count_ptr_d, count_ptr_derefs,
6736         count_uses_and_derefs): Relocate from gimple.c.
6737         (gimple_replace_ssa_lhs): Renamed gimple_replace_ssa from gimple.c
6738         * tree-ssa-reassoc.c (repropagate_negates): Use gimple_replace_ssa_lhs.
6739         * tree-ssa-math-opts (execute_cse_reciprocals): Use
6740         gimple_replace_ssa_lhs.
6741         * asan.c: Include gimple-builder.h.
6742         * Makefile.in: Add gimple-builder.o.
6744 2013-09-26  Richard Biener  <rguenther@suse.de>
6746         * tree-ssa-live.c (var_map_base_init): Handle SSA names with
6747         DECL_IGNORED_P base VAR_DECLs like anonymous SSA names.
6748         (loe_visit_block): Use gcc_checking_assert.
6749         * tree-ssa-coalesce.c (create_outofssa_var_map): Use
6750         gimple_assign_ssa_name_copy_p.
6751         (gimple_can_coalesce_p): Adjust according to the var_map_base_init
6752         change.
6754 2013-09-26  David Edelsohn  <dje.gcc@gmail.com>
6756         * config/rs6000/t-rs6000 (rs6000.o): Remove.
6757         (rs6000-c.o): Use COMPILE and POSTCOMPILE.
6759 2013-09-26  Richard Biener  <rguenther@suse.de>
6761         PR tree-optimization/58539
6762         * tree-vect-loop.c (vect_create_epilog_for_reduction): Honor
6763         the fact that debug statements are not taking part in loop-closed
6764         SSA construction.
6766 2013-09-26  Nick Clifton  <nickc@redhat.com>
6768         * config/msp430/msp430.c (msp430_expand_epilogue): Fix compile
6769         time warning message.
6770         (msp430_print_operand_raw): Delete unused letter parameter.
6771         (TARGET_PRINT_OPERAND_ADDRESS): Define.
6772         (msp430_print_operand_address): New function.
6773         (msp430_print_operand): Move address printing code from here to
6774         new function.
6775         * config/msp430/msp430.md (movsipsi2): Add comment in generated
6776         assembler.
6777         (zero_extendpsisi2): Likewise.
6778         (extendpsisi2): New pattern.
6779         (andneghi3): New pattern.
6781 2013-09-26  Yvan Roux  <yvan.roux@linaro.org>
6783         * config/aarch64/aarch64.opt (mlra): New option.
6784         * config/aarch64/aarch64.c (aarch64_lra_p): New function.
6785         (TARGET_LRA_P): Define.
6787 2013-09-26  Eric Botcazou  <ebotcazou@adacore.com>
6789         * expr.c (expand_assignment): Remove obsolete comment.
6791 2013-09-25  Jeff Law  <law@redhat.com>
6793         * tree-flow.h (thread_through_all_blocks): Prototype moved into
6794         tree-ssa-threadupdate.h.
6795         (register_jump_thread): Similarly.
6796         * tree-ssa-threadupdate.h: New header file.
6797         * tree-ssa-dom.c: Include tree-ssa-threadupdate.h.
6798         * tree-vrp.c: Likewise.
6799         * tree-ssa-threadedge.c: Include tree-ssa-threadupdate.h.
6800         (thread_around_empty_blocks): Change type of path vector argument to
6801         an edge,type pair from just an edge.  Initialize both elements when
6802         appending to a jump threading path.  Tweak references to elements
6803         appropriately.
6804         (thread_across_edge): Similarly.  Release memory for the elements
6805         as needed.
6806         * tree-ssa-threadupdate.c: Include tree-ssa-threadupdate.h.
6807         (dump_jump_thread_path): New function broken out from
6808         register_jump_thread.
6809         (register_jump_thread): Use dump_jump_thread_path.  Change type of
6810         path vector entries.  Search the path for NULL edges and dump
6811         the path if one is found.  Tweak the conversion of path to 3-edge
6812         form to use the block copy type information embedded in the path.
6814 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6816         * lra.c (update_inc_notes): Remove all REG_DEAD and REG_UNUSED notes.
6818 2013-09-25  Yvan Roux  <yvan.roux@linaro.org>
6819             Vladimir Makarov  <vmakarov@redhat.com>
6821         * rtlanal.c (lsb_bitfield_op_p): New predicate for bitfield operations
6822         from the least significant bit.
6823         (strip_address_mutations): Add bitfield operations handling.
6824         (must_be_index_p): Add shifting and rotate operations handling.
6825         (set_address_base): Use must_be_base_p predicate.
6826         (set_address_index): Use must_be_index_p predicate.
6828 2013-09-25  Alexander Ivchenko  <alexander.ivchenko@intel.com>
6829             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
6830             Sergey Lega  <sergey.s.lega@intel.com>
6831             Anna Tikhonova  <anna.tikhonova@intel.com>
6832             Ilya Tocar  <ilya.tocar@intel.com>
6833             Andrey Turetskiy  <andrey.turetskiy@intel.com>
6834             Ilya Verbin  <ilya.verbin@intel.com>
6835             Kirill Yukhin  <kirill.yukhin@intel.com>
6836             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
6838         * config/i386/i386.c (ix86_avx256_split_vector_move_misalign):
6839         Use new names.
6840         (ix86_expand_vector_move_misalign): Support new unaligned load and
6841         stores and use new names.
6842         (CODE_FOR_sse2_storedqu): Rename to ...
6843         (CODE_FOR_sse2_storedquv16qi): ... this.
6844         (CODE_FOR_sse2_loaddqu): Rename to ...
6845         (CODE_FOR_sse2_loaddquv16qi): ... this.
6846         (CODE_FOR_avx_loaddqu256): Rename to ...
6847         (CODE_FOR_avx_loaddquv32qi): ... this.
6848         (CODE_FOR_avx_storedqu256): Rename to ...
6849         (CODE_FOR_avx_storedquv32qi): ... this.
6850         * config/i386/i386.md (fpint_logic): New.
6851         * config/i386/sse.md (VMOVE): Extend for AVX512.
6852         (VF): Ditto.
6853         (VF_128_256): New.
6854         (VF_512): Ditto.
6855         (VI_UNALIGNED_LOADSTORE): Ditto.
6856         (sse2_avx_avx512f): Ditto.
6857         (sse2_avx2): Extend for AVX512.
6858         (sse4_1_avx2): Ditto.
6859         (avx2_avx512f): New.
6860         (sse): Extend for AVX512.
6861         (sse2): Ditto.
6862         (sse4_1): Ditto.
6863         (avxsizesuffix): Ditto.
6864         (sseintvecmode): Ditto.
6865         (ssePSmode): Ditto.
6866         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Ditto.
6867         (<sse>_storeu<ssemodesuffix><avxsizesuffix>): Ditto.
6868         (<sse2>_loaddqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6869         (<sse2_avx_avx512f>_loaddqu<mode>): ... this.
6870         (<sse2>_storedqu<avxsizesuffix>): Extend for AVX512 and rename to ...
6871         (<sse2_avx_avx512f>_storedqu<mode): ... this.
6872         (<sse>_movnt<mode>): Replace constraint "x" with "v".
6873         (STORENT_MODE): Extend for AVX512.
6874         (*absneg<mode>2): Replace constraint "x" with "v".
6875         (*mul<mode>3): Ditto.
6876         (*ieee_smin<mode>3): Ditto.
6877         (*ieee_smax<mode>3): Ditto.
6878         (avx_cmp<mode>3): Replace VF with VF_128_256.
6879         (*<sse>_maskcmp<mode>3_comm): Ditto.
6880         (<sse>_maskcmp<mode>3): Ditto.
6881         (<sse>_andnot<mode>3): Extend for AVX512.
6882         (<code><mode>3, anylogic): Replace VF with VF_128_256.
6883         (<code><mode>3, fpint_logic): New.
6884         (*<code><mode>3): Extend for AVX512.
6885         (avx512flogicsuff): New.
6886         (avx512f_<logic><mode>): Ditto.
6887         (<sse>_movmsk<ssemodesuffix><avxsizesuffix>): Replace VF with
6888         VF_128_256.
6889         (<sse4_1>_blend<ssemodesuffix><avxsizesuffix>): Ditto.
6890         (<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>): Ditto.
6891         (<sse4_1>_dp<ssemodesuffix><avxsizesuffix>): Ditto.
6892         (avx_vtest<ssemodesuffix><avxsizesuffix>): Ditto.
6893         (<sse4_1>_round<ssemodesuffix><avxsizesuffix>): Ditto.
6894         (xop_vpermil2<mode>3): Ditto.
6895         (*avx_vpermilp<mode>): Extend for AVX512 and rename to ...
6896         (*<sse2_avx_avx512f>_vpermilp<mode>): ... this.
6897         (avx_vpermilvar<mode>3): Extend for AVX512 and rename to ...
6898         (<sse2_avx_avx512f>_vpermilvar<mode>3): ... this.
6900 2013-09-25  Tom Tromey  <tromey@redhat.com>
6902         * Makefile.in (PARTITION_H, LTO_SYMTAB_H, COMMON_TARGET_DEF_H)
6903         (RTL_ERROR_H, TRANS_MEM_H, COVERAGE_H, DEMANGLE_H, ALIAS_H)
6904         (SCHED_INT_H, SEL_SCHED_IR_H, SEL_SCHED_DUMP_H, VALTRACK_H, DDG_H)
6905         (GGC_INTERNAL_H, DECNUM_H, BACKTRACE_H, MKDEPS_H, TREE_HASHER_H)
6906         (TREE_SSA_LIVE_H, SSAEXPAND_H, DWARF2OUT_H, SCEV_H, OMEGA_H)
6907         (TREE_DATA_REF_H, IRA_INT_H, LRA_INT_H, DBGCNT_H, DATA_STREAMER_H)
6908         (GIMPLE_STREAMER_H, TREE_STREAMER_H, STREAMER_HOOKS_H)
6909         (TREE_VECTORIZER_H, IPA_INLINE_H, GSTAB_H, LIBFUNCS_H)
6910         (GRAPHITE_HTAB_H): Remove.
6912 2013-09-25  Tom Tromey  <tromey@redhat.com>
6914         * config/mcore/t-mcore (CROSS_FLOAT_H): Remove.
6916 2013-09-25  Tom Tromey  <tromey@redhat.com>
6918         * config/t-glibc (glibc-c.o): Use COMPILE and POSTCOMPILE.
6920 2013-09-25  Tom Tromey  <tromey@redhat.com>
6922         * config/i386/t-i386 (i386.o): Remove.
6923         (i386-c.o): Use COMPILE and POSTCOMPILE.
6925 2013-09-25  Tom Tromey  <tromey@redhat.com>
6927         * Makefile.in ($(out_object_file)): Use COMPILE and POSTCOMPILE.
6929 2013-09-25  Tom Tromey  <tromey@redhat.com>
6931         * Makefile.in (graph.o, sbitmap.o, sparseset.o, gcc-ar.o)
6932         (gcc-ranlib.o, gcc-nm.o, collect2.o, collect2-aix.o, tlink.o)
6933         (lto-wrapper.o, default-c.o, attribs.o, incpath.o, prefix.o)
6934         (gcc.o, options.o, options-save.o, version.o, gtype-desc.o)
6935         (trans-mem.o, ggc-common.o, ggc-page.o, ggc-none.o, stringpool.o)
6936         (convert.o, double-int.o, lto-compress.o, data-streamer-in.o)
6937         (data-streamer-out.o, data-streamer.o, gimple-streamer-in.o)
6938         (gimple-streamer-out.o, tree-streamer.o, tree-streamer-in.o)
6939         (tree-streamer-out.o, streamer-hooks.o, lto-cgraph.o)
6940         (lto-streamer-in.o, lto-streamer-out.o, lto-section-in.o)
6941         (lto-section-out.o, lto-opts.o, lto-streamer.o, langhooks.o)
6942         (test-dump.o, tree.o, tree-dump.o, tree-inline.o, print-tree.o)
6943         (stor-layout.o, asan.o, tsan.o, ubsan.o, tree-ssa-tail-merge.o)
6944         (tree-ssa-structalias.o, tree-ssa-uninit.o, tree-ssa.o)
6945         (tree-into-ssa.o, tree-ssa-ter.o, tree-ssa-coalesce.o)
6946         (tree-outof-ssa.o, tree-ssa-dse.o, tree-ssa-forwprop.o)
6947         (tree-ssa-phiprop.o, tree-ssa-ifcombine.o, tree-ssa-phiopt.o)
6948         (tree-nrv.o, tree-ssa-copy.o, tree-ssa-propagate.o)
6949         (tree-ssa-dom.o, tree-ssa-uncprop.o, tree-ssa-threadedge.o)
6950         (tree-ssa-threadupdate.o, tree-ssanames.o, tree-phinodes.o)
6951         (domwalk.o, tree-ssa-live.o, tree-ssa-copyrename.o)
6952         (tree-ssa-pre.o, tree-ssa-sccvn.o)
6953         (gimple-ssa-strength-reduction.o, tree-vrp.o, tree-cfg.o)
6954         (tree-cfgcleanup.o, tree-tailcall.o, tree-ssa-sink.o)
6955         (tree-nested.o, tree-if-conv.o, tree-iterator.o, tree-dfa.o)
6956         (tree-ssa-operands.o, tree-eh.o, tree-ssa-loop.o)
6957         (tree-ssa-loop-unswitch.o, tree-ssa-address.o)
6958         (tree-ssa-loop-niter.o, tree-ssa-loop-ivcanon.o)
6959         (tree-ssa-loop-ch.o, tree-ssa-loop-prefetch.o, tree-predcom.o)
6960         (tree-ssa-loop-ivopts.o, tree-affine.o, tree-ssa-loop-manip.o)
6961         (tree-ssa-loop-im.o, tree-ssa-math-opts.o, tree-ssa-alias.o)
6962         (tree-ssa-reassoc.o, tree-optimize.o, gimplify.o)
6963         (gimple-iterator.o, gimple-fold.o, gimple-low.o, omp-low.o)
6964         (tree-browser.o, omega.o, tree-chrec.o, tree-scalar-evolution.o)
6965         (tree-data-ref.o, sese.o, graphite.o, graphite-blocking.o)
6966         (graphite-clast-to-gimple.o, graphite-dependences.o)
6967         (graphite-interchange.o, graphite-poly.o)
6968         (graphite-scop-detection.o, graphite-sese-to-poly.o)
6969         (graphite-optimize-isl.o, tree-vect-loop.o)
6970         (tree-vect-loop-manip.o, tree-vect-patterns.o, tree-vect-slp.o)
6971         (tree-vect-stmts.o, tree-vect-data-refs.o, tree-vectorizer.o)
6972         (vtable-verify.o, tree-loop-distribution.o, tree-parloops.o)
6973         (tree-stdarg.o, tree-object-size.o, internal-fn.o, gimple.o)
6974         (gimple-pretty-print.o, tree-mudflap.o, tree-nomudflap.o)
6975         (tree-pretty-print.o, tree-diagnostic.o, fold-const.o)
6976         (diagnostic.o, diagnostic-color.o, opts.o, opts-global.o)
6977         (opts-common.o, targhooks.o, common/common-targhooks.o, input.o)
6978         (toplev.o, hwint.o, passes.o, plugin.o, main.o, host-default.o)
6979         (rtl-error.o, rtl.o, print-rtl.o, rtlanal.o, varasm.o, function.o)
6980         (statistics.o, stmt.o, except.o, expr.o, dojump.o, builtins.o)
6981         (calls.o, expmed.o, explow.o, optabs.o, dbxout.o, debug.o)
6982         (sdbout.o, dwarf2out.o, dwarf2cfi.o, dwarf2asm.o, vmsdbgout.o)
6983         (xcoffout.o, godump.o, emit-rtl.o, real.o, realmpfr.o, dfp.o)
6984         (fixed-value.o, jump.o, simplify-rtx.o, symtab.o, cgraph.o)
6985         (cgraphunit.o, cgraphclones.o, cgraphbuild.o, varpool.o, ipa.o)
6986         (ipa-profile.o, ipa-devirt.o, ipa-prop.o, ipa-ref.o, ipa-cp.o)
6987         (ipa-split.o, ipa-inline.o, ipa-inline-analysis.o)
6988         (ipa-inline-transform.o, ipa-utils.o, ipa-reference.o)
6989         (ipa-pure-const.o, coverage.o, cselib.o, cse.o, dce.o, dumpfile.o)
6990         (dse.o, fwprop.o, web.o, ree.o, cprop.o, gcse.o, store-motion.o)
6991         (resource.o, lcm.o, mode-switching.o, tree-ssa-dce.o)
6992         (tree-call-cdce.o, tree-ssa-ccp.o, tree-ssa-strlen.o, tree-sra.o)
6993         (tree-switch-conversion.o, tree-complex.o, tree-emutls.o)
6994         (tree-vect-generic.o, df-core.o, df-problems.o, df-scan.o)
6995         (regstat.o, valtrack.o, var-tracking.o, profile.o, mcf.o)
6996         (tree-profile.o, value-prof.o, loop-doloop.o, alloc-pool.o)
6997         (auto-inc-dec.o, cfg.o, cfghooks.o, cfgexpand.o, cfgrtl.o)
6998         (cfganal.o, cfgbuild.o, cfgcleanup.o, cfgloop.o, cfgloopanal.o)
6999         (graphds.o, loop-iv.o, loop-invariant.o, cfgloopmanip.o)
7000         (loop-init.o, loop-unswitch.o, loop-unroll.o, dominance.o)
7001         (et-forest.o, combine.o, reginfo.o, bitmap.o, vec.o, hash-table.o)
7002         (reload.o, reload1.o, rtlhooks.o, postreload.o, postreload-gcse.o)
7003         (caller-save.o, bt-load.o, reorg.o, alias.o, stack-ptr-mod.o)
7004         (init-regs.o, ira-build.o, ira-costs.o, ira-conflicts.o)
7005         (ira-color.o, ira-emit.o, ira-lives.o, ira.o, lra.o)
7006         (lra-assigns.o, lra-coalesce.o, lra-constraints.o)
7007         (lra-eliminations.o, lra-lives.o, lra-spills.o, regmove.o)
7008         (combine-stack-adj.o, compare-elim.o, ddg.o, modulo-sched.o)
7009         (haifa-sched.o, sched-deps.o, sched-rgn.o, sched-ebb.o)
7010         (sched-vis.o, sel-sched.o, sel-sched-dump.o, sel-sched-ir.o)
7011         (final.o, recog.o, reg-stack.o, sreal.o, predict.o, lists.o)
7012         (bb-reorder.o, tracer.o, timevar.o, regcprop.o, regrename.o)
7013         (ifcvt.o, params.o, pointer-set.o, hooks.o, pretty-print.o)
7014         (errors.o, dbgcnt.o, lower-subreg.o, target-globals.o)
7015         (hw-doloop.o, file-find.o, context.o, $(common_out_object_file))
7016         (insn-attrtab.o, insn-automata.o, insn-dfatab.o, insn-emit.o)
7017         (insn-enums.o, insn-extract.o, insn-latencytab.o, insn-modes.o)
7018         (insn-opinit.o, insn-output.o, insn-peep.o, insn-preds.o)
7019         (insn-recog.o, intl.o, cppbuiltin.o, cppdefault.o, gcov.o)
7020         (gcov-dump.o): Remove.
7021         (default-c.o): Use COMPILE and POSTCOMPILE.
7022         (CFLAGS-gcc.o): New variable.
7023         ($(common_out_object_file)): Use COMPILE and POSTCOMPILE.
7025 2013-09-25  Tom Tromey  <tromey@redhat.com>
7027         * Makefile.in (c-family/cppspec.o, c-family/c-common.o)
7028         (c-family/c-cppbuiltin.o, c-family/c-dump.o, c-family/c-format.o)
7029         (c-family/c-gimplify.o, c-family/c-lex.o, c-family/c-omp.o)
7030         (c-family/c-opts.o, c-family/c-pch.o, c-family/c-ppoutput.o)
7031         (c-family/c-pragma.o, c-family/c-pretty-print.o)
7032         (c-family/c-semantics.o, c-family/c-ada-spec.o)
7033         (c-family/array-notation-common.o, c-family/stub-objc.o)
7034         (c-family/c-ubsan.o): Remove.
7036 2013-09-25  Tom Tromey  <tromey@redhat.com>
7038         * Makefile.in (C_TREE_H): Reference c/c-tree.h.
7040 2013-09-25  Tom Tromey  <tromey@redhat.com>
7042         * Makefile.in (DRIVER_DEFINES): Use $(and), not shell code,
7043         to add -DENABLE_SHARED_LIBGCC.
7044         (gcc.o): Don't use subshell.
7046 2013-09-25  Tom Tromey  <tromey@redhat.com>
7048         * Makefile.in (OUTPUT_OPTION): Define as "-o $@".
7049         * configure.ac: Don't invoke AM_PROG_CC_C_O.
7050         (NO_MINUS_C_MINUS_O, OUTPUT_OPTION): Don't subst.
7051         * configure, config.in: Rebuild.
7053 2013-09-25  Tom Tromey  <tromey@redhat.com>
7055         * Makefile.in (CCDEPMODE, DEPDIR, depcomp, COMPILE.base)
7056         (COMPILE, POSTCOMPILE): New variables.
7057         (.cc.o .c.o): Use COMPILE, POSTCOMPILE.
7058         (DEPFILES): New variable.
7059         Include ".Po" files.
7060         * configure.ac: Add checks for dependency checking.
7061         * configure, aclocal.m4: Regenerate.
7063 2013-09-25  Tom Tromey  <tromey@redhat.com>
7065         * Makefile.in (ALL_HOST_BACKEND_OBJS): Add lto-wrapper.o.
7066         ($(ALL_HOST_OBJS)): Move order-only dependency to end of file.
7068 2013-09-25  Tom Tromey  <tromey@redhat.com>
7070         * Makefile.in (generated_files): Add options.h,
7071         target-hooks-def.h, insn-opinit.h,
7072         common/common-target-hooks-def.h, pass-instances.def,
7073         c-family/c-target-hooks-def.h.
7075 2013-09-25  Jeff Law  <law@redhat.com>
7077         * tree-ssa-threadedge.c (thread_across_edge): Use foo.last () rather
7078         than foo[foo.length () - 1] to access last member in a vec.
7079         * tree-ssa-threadupdate.c (register_jump_thread): Similarly.
7081 2013-09-25  Richard Biener  <rguenther@suse.de>
7083         PR middle-end/58521
7084         * tree.c (iterative_hash_expr): Remove MEM_REF special handling.
7086 2013-09-25  Jan Hubicka  <jh@suse.cz>
7088         * cgraph.c (cgraph_resolve_speculation): Use semantical equivalency
7089         test.
7091 2013-09-25  Marek Polacek  <polacek@redhat.com>
7093         PR sanitizer/58420
7094         * ubsan.c (ubsan_type_descriptor): Handle IDENTIFIER_NODEs
7095         when determining the type name.
7097 2013-09-24  Oleg Endo  <olegendo@gcc.gnu.org>
7099         * config/sh/sh.md: Fix formatting.
7101 2013-09-24  Xinliang David Li  <davidxl@google.com>
7103         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Check
7104         max peel iterations parameter.
7105         * param.def: New parameter.
7106         * doc/invoke.texi: Document New parameter.
7108 2013-09-24  Christophe Lyon  <christophe.lyon@linaro.org>
7110         * gimple-pretty-print.c: Various whitespace tweaks.
7111         * tree-core.h: Likewise.
7112         * tree-pretty-print.c: Likewise.
7113         * tree-ssa-alias.c: Likewise.
7114         * tree-ssa-copy.c: Likewise.
7115         * tree-ssanames.c: Likewise.
7116         * tree-ssanames.h: Likewise.
7117         * tree-vrp.c: Likewise.
7119 2013-09-24  Alan Modra  <amodra@gmail.com>
7121         PR middle-end/57134
7122         PR middle-end/57586
7123         * stmt.c (expand_asm_operands): Call expand_expr with EXPAND_MEMORY
7124         for output operands that disallow regs.  Don't use EXPAND_WRITE on
7125         inout operands.
7127 2013-09-24  Richard Biener  <rguenther@suse.de>
7129         PR middle-end/58513
7130         * tree.c (reference_alias_ptr_type): Move ...
7131         * alias.c (reference_alias_ptr_type): ... here and implement
7132         in terms of the new reference_alias_ptr_type_1.
7133         (ref_all_alias_ptr_type_p): New helper.
7134         (get_deref_alias_set_1): Drop flag_strict_aliasing here,
7135         use ref_all_alias_ptr_type_p.
7136         (get_deref_alias_set): Add flag_strict_aliasing check here.
7137         (reference_alias_ptr_type_1): New function, split out from ...
7138         (get_alias_set): ... here.
7139         (alias_ptr_types_compatible_p): New function.
7140         * alias.h (reference_alias_ptr_type): Declare.
7141         (alias_ptr_types_compatible_p): Likewise.
7142         * tree.h (reference_alias_ptr_type): Remove.
7143         * fold-const.c (operand_equal_p): Use alias_ptr_types_compatible_p
7144         to compare MEM_REF alias types.
7146 2013-09-24  Richard Biener  <rguenther@suse.de>
7148         * tree-vrp.c (vrp_finalize): Check for SSA name presence.
7150 2013-09-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
7152         * config/rs6000/rs6000.c (rs6000_vector_reload): Delete, combine
7153         reload helper function arrays into a single array reg_addr.
7154         (reload_fpr_gpr): Likewise.
7155         (reload_gpr_vsx): Likewise.
7156         (reload_vsx_gpr): Likewise.
7157         (struct rs6000_reg_addr): Likewise.
7158         (reg_addr): Likewise.
7159         (rs6000_debug_reg_global): Change rs6000_vector_reload,
7160         reload_fpr_gpr, reload_gpr_vsx, reload_vsx_gpr uses to reg_addr.
7161         (rs6000_init_hard_regno_mode_ok): Likewise.
7162         (rs6000_secondary_reload_direct_move): Likewise.
7163         (rs6000_secondary_reload): Likewise.
7165         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add new
7166         constraints: wu, ww, and wy.  Repurpose wv constraint added during
7167         power8 changes.  Put wg constraint in alphabetical order.
7169         * config/rs6000/rs6000.opt (-mvsx-scalar-float): New debug switch
7170         for future work to add ISA 2.07 VSX single precision support.
7171         (-mvsx-scalar-double): Change default from -1 to 1, update
7172         documentation comment.
7173         (-mvsx-scalar-memory): Rename debug switch to -mupper-regs-df.
7174         (-mupper-regs-df): New debug switch to control whether DF values
7175         can go in the traditional Altivec registers.
7176         (-mupper-regs-sf): New debug switch to control whether SF values
7177         can go in the traditional Altivec registers.
7179         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print wu, ww,
7180         and wy constraints.
7181         (rs6000_init_hard_regno_mode_ok): Use ssize_t instead of int for
7182         loop variables.  Rename -mvsx-scalar-memory to -mupper-regs-df.
7183         Add new constraints, wu/ww/wy.  Repurpose wv constraint.
7184         (rs6000_debug_legitimate_address_p): Print if we are running
7185         before, during, or after reload.
7186         (rs6000_secondary_reload): Add a comment.
7187         (rs6000_opt_masks): Add -mupper-regs-df, -mupper-regs-sf.
7189         * config/rs6000/constraints.md (wa constraint): Sort w<x>
7190         constraints.  Update documentation string.
7191         (wd constraint): Likewise.
7192         (wf constraint): Likewise.
7193         (wg constraint): Likewise.
7194         (wn constraint): Likewise.
7195         (ws constraint): Likewise.
7196         (wt constraint): Likewise.
7197         (wx constraint): Likewise.
7198         (wz constraint): Likewise.
7199         (wu constraint): New constraint for ISA 2.07 SFmode scalar
7200         instructions.
7201         (ww constraint): Likewise.
7202         (wy constraint): Likewise.
7203         (wv constraint): Repurpose ISA 2.07 constraint that we did not use
7204         in the previous submissions.
7205         * doc/md.texi (PowerPC and IBM RS6000): Likewise.
7207 2013-09-23  Richard Sandiford  <rdsandiford@googlemail.com>
7209         * doc/rtl.texi (REG_NOTES): Say that int_list can also be used.
7210         (REG_BR_PROB): Say that the probability is stored in an int_list.
7211         * reg-notes.def: Update commentary to mention INT_LIST.
7212         * rtl.def (EXPR_LIST, INSN_LIST): Capitalize comments.
7213         (INT_LIST): New rtx.
7214         * rtl.h (add_int_reg_note, add_shallow_copy_of_reg_note): Declare.
7215         * rtlanal.c (int_reg_note_p): New function.
7216         (alloc_reg_note): Assert that the note does not have an int argument.
7217         (add_int_reg_note, add_shallow_copy_of_reg_note): New functions.
7218         * combine.c (distribute_notes): Use add_shallow_copy_of_rtx.
7219         * cse.c (cse_process_notes_1): Expect REG_EQUAL to be an EXPR_LIST
7220         rather than an INSN_LIST.  Handle INT_LIST.
7221         * ifcvt.c (cond_exec_process_insns): Take the probability as an int
7222         rather than an rtx.  Use gen_rtx_INT_LIST to create a REG_BR_PROB note.
7223         (cond_exec_process_if_block): Use XINT to extract REG_BR_PROB values.
7224         Manipulate them as ints rather than rtxes.
7225         * reg-stack.c (subst_asm_stack_regs): Only handle EXPR_LIST notes.
7226         * regmove.c (copy_src_to_dest): Likewise.
7227         * sched-vis.c (print_insn_with_notes): Handle INT_LIST.
7229         * config/i386/winnt.c (i386_pe_seh_unwind_emit): Sink pat assignment
7230         into the cases that need it.
7231         * config/arm/arm.c (arm_unwind_emit): Likewise.
7233         * asan.c (asan_clear_shadow): Use add_int_reg_note for REG_BR_PROB.
7234         * emit-rtl.c (try_split, emit_copy_of_insn_after): Likewise.
7235         * loop-doloop.c (add_test, doloop_modify): Likewise.
7236         * loop-unswitch.c (compare_and_jump_seq): Likewise.
7237         * optabs.c (emit_cmp_and_jump_insn_1): Likewise.
7238         * predict.c (combine_predictions_for_insn): Likewise.
7239         * print-rtl.c (print_rtx): Handle INT_LIST.
7240         * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Likewise.
7241         * config/alpha/alpha.c (emit_unlikely_jump): Likewise.
7242         * config/arm/arm.c (emit_unlikely_jump): Likewise.
7243         * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise.
7244         (ix86_split_fp_branch, predict_jump): Likewise.
7245         * config/rs6000/rs6000.c (emit_unlikely_jump): Likewise.
7246         * config/sh/sh.c (expand_cbranchsi4): Likewise.
7247         * config/spu/spu.c (ea_load_store_inline): Likewise.
7249         * cfgbuild.c (compute_outgoing_frequencies): Use XINT to access the
7250         value of a REG_BR_PROB note.
7251         * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
7252         (update_br_prob_note, rtl_verify_edges, purge_dead_edges): Likewise.
7253         * emit-rtl.c (try_split): Likewise.
7254         * predict.c (br_prob_note_reliable_p): Likewise.
7255         (invert_br_probabilities, combine_predictions_for_insn): Likewise.
7256         * reorg.c (mostly_true_jump): Likewise.
7257         * config/bfin/bfin.c (cbranch_predicted_taken_p): Likewise.
7258         * config/frv/frv.c (frv_print_operand_jump_hint): Likewise.
7259         * config/i386/i386.c (ix86_print_operand): Likewise.
7260         * config/ia64/ia64.c (ia64_print_operand): Likewise.
7261         * config/mmix/mmix.c (mmix_print_operand): Likewise.
7262         * config/rs6000/rs6000.c (output_cbranch): Likewise.
7263         * config/s390/s390.c (s390_expand_tbegin): Likewise.
7264         * config/sh/sh.c (sh_print_operand, sh_adjust_cost): Likewise.
7265         * config/sparc/sparc.c (output_cbranch): Likewise.
7266         * config/spu/spu.c (get_branch_target): Likewise.
7267         * config/tilegx/tilegx.c (cbranch_predicted_p): Likewise.
7268         * config/tilepro/tilepro.c (cbranch_predicted_p): Likewise.
7270 2013-09-23  Jan Hubicka  <jh@suse.cz>
7272         * ipa-cp.c (ipa_get_indirect_edge_target_1): Add sanity check
7273         for ipa-devirt.
7274         * ipa-utils.h (possible_polymorphic_call_target_p): New function.
7275         * ipa-devirt.c (possible_polymorphic_call_target_p): Be tolerant
7276         of external calls
7277         * gimple-fold.c: Include ipa-utils.h and gimple-pretty-print.h
7278         (gimple_fold_call): Dump inconsistent devirtualizations; add
7279         sanity check for type based devirtualizations.
7280         * ipa-prop.c: Include ipa-utils.h
7281         (ipa_intraprocedural_devirtualization): Add sanity check.
7282         (try_make_edge_direct_virtual_call): Likewise.
7284 2013-09-23  Eric Botcazou  <ebotcazou@adacore.com>
7286         * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Recurse on copy
7287         assignment statements.
7289 2013-09-23  Kugan Vivekanandarajah  <kuganv@linaro.org>
7291         * gimple-pretty-print.c (dump_ssaname_info): New function.
7292         (dump_gimple_phi): Call it.
7293         (pp_gimple_stmt_1): Likewise.
7294         * tree-core.h (tree_ssa_name): New union ssa_name_info_type field.
7295         (range_info_def): Declare.
7296         * tree-pretty-print.c (pp_double_int): New function.
7297         (dump_generic_node): Call it.
7298         * tree-pretty-print.h (pp_double_int): Declare.
7299         * tree-ssa-alias.c (dump_alias_info): Check pointer type.
7300         * tree-ssanames.h (range_info_def): New structure.
7301         (value_range_type): Move definition here.
7302         (set_range_info, value_range_type, duplicate_ssa_name_range_info):
7303         Declare.
7304         * tree-ssanames.c (make_ssa_name_fn): Check pointer type at
7305         initialization.
7306         (set_range_info): New function.
7307         (get_range_info): Likewise.
7308         (duplicate_ssa_name_range_info): Likewise.
7309         (duplicate_ssa_name_fn): Check pointer type and call
7310         duplicate_ssa_name_range_info.
7311         * tree-ssa-copy.c (fini_copy_prop): Likewise.
7312         * tree-vrp.c (value_range_type): Remove definition, now in
7313         tree-ssanames.h.
7314         (vrp_finalize): Call set_range_info to update value range of SSA_NAMEs.
7315         * tree.h (SSA_NAME_PTR_INFO): Macro changed to access via union.
7316         (SSA_NAME_RANGE_INFO): New macro.
7318 2013-09-23  Richard Biener  <rguenther@suse.de>
7320         PR tree-optimization/58464
7321         * tree-ssa-pre.c (phi_trans_lookup): Remove.
7322         (phi_trans_add): Change to add conditionally on being not present.
7323         (phi_translate_1): Remove recursion detection here.
7324         (phi_translate): Pre-seed the cache with NULL to catch
7325         recursion here in a more generic way.
7326         (bitmap_find_leader): Adjust comment.
7327         (get_representative_for): Dump value-numbers.
7328         (create_expression_by_pieces): Likewise.
7329         (insert_into_preds_of_block): Likewise.
7331 2013-09-23  Christian Bruel  <christian.bruel@st.com>
7333         PR target/58475
7334         * config/sh/sh.md (movsf_ie): Allow fpul_operand.
7335         * config/sh/predicate.md (arith_reg_operand): Disallow FPUL_REG.
7337 2013-09-23  James Greenhalgh  <james.greenhalgh@arm.com>
7339         Revert r202780:
7340         2013-09-20  Renlin Li  <renlin.li@arm.com>
7342         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7343         plus_constant.
7344         (aarch64_expand_epilogue): Likewise.
7345         (aarch64_legitimize_reload_address): Likewise.
7347 2013-09-22  Eric Botcazou  <ebotcazou@adacore.com>
7349         * gimplify.c (gimplify_asm_expr): Reset the TREE_CHAIN of clobbers to
7350         NULL_TREE before pushing them onto the vector.  Likewise for labels.
7352 2013-09-21  Eric Botcazou  <ebotcazou@adacore.com>
7354         * config/ia64/predicates.md (ia64_cbranch_operator): Accept unordered
7355         comparison operators when -fno-trapping-math is in effect.
7356         * config/ia64/ia64.c (ia64_expand_compare): Add support for unordered
7357         comparison operators in TFmode and assert that unsupported operators
7358         cannot reach here.
7359         (ia64_print_operand): Likewise.
7361 2013-09-21  Jan Hubicka  <jh@suse.cz>
7363         * x86-tune.def (partial_reg_stall): Disable for CoreI7 and newer.
7364         (sse_typeless_stores): Enable for core
7365         (sse_load0_by_pxor): Likewise.
7366         (four_jump_limit): Disable for core.
7367         (pad_returns): Likewise.
7368         (avoid_vector_decode): Likewise.
7369         (fuse_cmp_and_branch): Enable for cores.
7370         * i386.c (x86_accumulate_outgoing_args): Disable for cores.
7372 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7374         PR middle-end/56791
7375         * config/pa/pa.c (pa_option_override): Disable auto increment and
7376         decrement instructions until reload is completed.
7378         * config/pa/pa-linux.h (TARGET_OS_CPP_BUILTINS): Define
7379         __GCC_HAVE_SYNC_COMPARE_AND_SWAP_1, __GCC_HAVE_SYNC_COMPARE_AND_SWAP_2,
7380         and __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4.
7382 2013-09-20  DJ Delorie  <dj@redhat.com>
7383             Nick Clifton  <nickc@redhat.com>
7385         * config/rl78/rl78.c: Various whitespace and comment tweaks.
7386         (need_to_save): Save bank 0 on interrupts.
7387         (characterize_address): Strip far address wrappers.
7388         (rl78_as_legitimate_address): Likewise.
7389         (transcode_memory_rtx): Likewise.
7390         (rl78_peep_movhi_p): Disable this peephole after devirt.
7391         (rl78_propogate_register_origins): Forget all origins when a
7392         CLOBBER is seen.
7393         * config/rl78/rl78-virt.md: Various whitespace tweaks.
7394         * config/rl78/rl78-real.md: Various whitespace tweaks.  Additional
7395         peephole2's.
7396         * config/rl78/rl78.md (sel_rb): Disable for G10 just in case.
7397         * config/rl78/rl78-expand.md (movqi): Check for subregs of consts.
7398         * config/rl78/rl78.h (LINK_SPEC): Pass -gc-sections unless
7399         relocating.
7400         * config/rl78/constraints.md: Various whitespace and paren tweaks.
7402 2013-09-20  John David Anglin  <danglin@gcc.gnu.org>
7404         * config/pa/pa.md: In "scc" insn patterns, change output template to
7405         handle const0_rtx in reg_or_0_operand operands.
7407 2013-09-20  Martin Husemann  <martin@NetBSD.org>
7409         PR target/56875
7410         * config/vax/vax.c (vax_output_int_move): Use D format specifier.
7411         * config/vax/vax.md (ashldi3, <unnamed>): Ditto.
7413 2013-09-20  Richard Biener  <rguenther@suse.de>
7415         PR middle-end/58484
7416         * tree-scalar-evolution.c (struct scev_info_str): Shrink by
7417         remembering SSA name version and block index.
7418         (new_scev_info_str): Adjust.
7419         (hash_scev_info): Likewise.  Also hash the block index.
7420         (eq_scev_info): Adjust.
7421         (find_var_scev_info): Likewise.
7422         (struct instantiate_cache_entry): Remove.
7423         (struct instantiate_cache_type): Use a htab to map name, block
7424         to chrec.
7425         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7426         (get_instantiated_value_entry): Likewise.
7427         (hash_idx_scev_info, eq_idx_scev_info): New functions.
7428         (instantiate_scev_name): Adjust.
7430 2013-09-20  Jeff Law  <law@redhat.com>
7432         * tree-ssa-dom.c (record_temporary_equivalences): Add comment.
7434 2013-09-20  Yufeng Zhang  <yufeng.zhang@arm.com>
7436         * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
7437         Call aarch64_simd_expand_args to update op[argc].
7439 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7441         * plugin.c (parse_plugin_arg_opt): Accept equal sign inside
7442         plugin argument.
7444 2013-09-20  Basile Starynkevitch  <basile@starynkevitch.net>
7446         * gengtype.c (file_rules): Added rule for *.cc files.
7447         (get_output_file_with_visibility): Give fatal message when no
7448         rules found.
7450 2013-09-20  Renlin Li  <renlin.li@arm.com>
7452         * config/aarch64/aarch64.c (aarch64_expand_prologue): Use plus_constant.
7453         (aarch64_expand_epilogue): Likewise.
7454         (aarch64_legitimize_reload_address): Likewise.
7456 2013-09-20  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7458         PR middle-end/57748
7459         * expr.c (expand_assignment): Remove misalignp code path.
7461 2013-09-20  Marek Polacek  <polacek@redhat.com>
7463         PR sanitizer/58413
7464         * ubsan.c (get_ubsan_type_info_for_type): Use TYPE_SIZE instead of
7465         TYPE_PRECISION.  Add asserts.
7467 2013-09-20  Richard Biener  <rguenther@suse.de>
7469         PR tree-optimization/58453
7470         * tree-loop-distribution.c (distribute_loop): Apply the cost
7471         model for -ftree-loop-distribute-patterns, too.
7473 2013-09-20  Richard Biener  <rguenther@suse.de>
7475         PR middle-end/58473
7476         * tree-chrec.h (build_polynomial_chrec): Use gcc_checking_assert,
7477         make type comparison less strict.
7479 2013-09-20  Alan Modra  <amodra@gmail.com>
7481         * configure: Regenerate.
7482         * aclocal.m4: Regenerate.
7484 2013-09-20  Marek Polacek  <polacek@redhat.com>
7486         PR other/58467
7487         * doc/extend.texi: Document that attribute used is meant to be used
7488         on variables with static storage duration.
7490 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7492         PR tree-optimization/58472
7493         * tree-vect-stmts.c (vectorizable_store, vectorizable_load): For
7494         simd_lane_access set inv_p = false.
7495         * omp-low.c (lower_rec_input_clauses): Set TREE_NO_WARNING on
7496         the simduid magic VAR_DECL.
7498 2013-09-19  Jan Hubicka  <jh@suse.cz>
7500         * i386.c (generic_memcpy, generic_memset): Fix 32bit template.
7502 2013-09-17  Jeff Law  <law@redhat.com>
7504         * tree-ssa-dom.c (record_temporary_equivalences): New function
7505         split out of dom_opt_dom_walker::after_dom_children.
7506         (dom_opt_dom_walker::thread_across_edge): Move common code
7507         in here from dom_opt_dom_walker::after_dom_children.
7508         (dom_opt_dom_walker::after_dom_children): Corresponding simplifictions.
7510 2013-09-19  Jan Hubicka  <jh@suse.cz>
7512         * i386.h (TARGET_GENERIC32, TARGET_GENERIC64): Remove.
7513         (TARGET_GENERIC): Use PROCESOR_GENERIC
7514         (enum processor_type): Unify generic32 and 64.
7515         * i386.md (cpu): Likewise.
7516         * x86-tune.def (use_leave): Enable for generic32.
7517         (avoid_vector_decode, slow_imul_imm32_mem, slow_imul_imm8): Likewise.
7518         * athlon.md: Change generic64 to generic in all occurences.
7519         * i386-c.c (ix86_target_macros_internal): Unify generic64 and 32.
7520         (ix86_target_macros_internal): Likewise.
7521         * driver-i386.c (host_detect_local_cpu): Likewise.
7522         * i386.c (generic64_memcpy, generic64_memset, generic64_cost): Rename
7523         to ..
7524         (generic_memcpy, generic_memset, generic_cost): This one.
7525         (generic32_memcpy, generic32_memset, generic32_cost): Remove.
7526         (m_GENERIC32, m_GENERIC64): Remove.
7527         (m_GENERIC): Turn into one flag.
7528         (processor_target): Unify generic tunnings.
7529         (ix86_option_override_internal): Replace generic32/64 by generic.
7530         (ix86_issue_rate): Likewise.
7531         (ix86_adjust_cost): Likewise.
7533 2013-09-19  Jan Hubicka  <jh@suse.cz>
7535         * cgraph.c (cgraph_create_edge_1): Avoid uninitialized read
7536         of speculative flag.
7538 2013-09-19  Jakub Jelinek  <jakub@redhat.com>
7540         * omp-low.c (expand_omp_sections): Always pass len - 1 to
7541         GOMP_sections_start, even if !exit_reachable.
7543 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7545         * lra-constraints.c (need_for_all_save_p): Use macro
7546         HARD_REGNO_CALL_PART_CLOBBERED.
7547         * lra-lives.c (check_pseudos_live_through_calls): Use the macro to
7548         set up pseudo conflict hard regs.
7550 2013-09-18  Michael Meissner  <meissner@linux.vnet.ibm.com>
7552         PR target/58452
7553         * config/rs6000/paired.md (movmisalignv2sf): Fix to allow memory
7554         operands.
7556 2013-09-18  Vladimir Makarov  <vmakarov@redhat.com>
7558         PR rtl-optimization/58438
7559         * lra.c (lra): Clear lra_optional_reload_pseudos in upper loop.
7560         * lra-constraints.c (undo_optional_reloads): Keep optional reloads
7561         from previous subpasses.
7563 2013-09-18  Richard Earnshaw  <rearnsha@arm.com>
7565         * arm.c (arm_get_frame_offsets): Validate architecture supports
7566         LDRD/STRD before accepting the tuning preference.
7567         (arm_expand_prologue): Likewise.
7568         (arm_expand_epilogue): Likewise.
7570 2013-09-18  Richard Biener  <rguenther@suse.de>
7572         PR tree-optimization/58417
7573         * tree-chrec.c (chrec_fold_plus_1): Assert that we do not
7574         have chrecs with symbols defined in the loop as operands.
7575         (chrec_fold_multiply): Likewise.
7576         * tree-scalar-evolution.c (interpret_rhs_expr): Instantiate
7577         parameters before folding binary operations.
7578         (struct instantiate_cache_entry_hasher): Remove.
7579         (struct instantiate_cache_type): Use a pointer-map.
7580         (instantiate_cache_type::instantiate_cache_type): New function.
7581         (instantiate_cache_type::get): Likewise.
7582         (instantiate_cache_type::set): Likewise.
7583         (instantiate_cache_type::~instantiate_cache_type): Adjust.
7584         (get_instantiated_value_entry): Likewise.
7585         (global_cache): New global.
7586         (instantiate_scev_r, instantiate_scev_poly, instantiate_scev_binary,
7587         instantiate_array_ref, instantiate_scev_convert, instantiate_scev_3,
7588         instantiate_scev_2, instantiate_scev_1): Do not pass along cache.
7589         (instantiate_scev_name): Adjust.
7590         (instantiate_scev): Construct global instead of local cache.
7591         (resolve_mixers): Likewise.
7593 2013-09-18  Daniel Morris  <danielm@ecoscentric.com>
7594             Paolo Carlini  <paolo.carlini@oracle.com>
7596         PR c++/58458
7597         * doc/implement-cxx.texi: Fix references to the C++ standards.
7599 2013-09-18  Jakub Jelinek  <jakub@redhat.com>
7601         * omp-low.c (copy_var_decl): Copy DECL_ATTRIBUTES.
7602         * tree-vect-data-refs.c (vect_analyze_data_refs): For
7603         simd_lane_access drs, update also DR_ALIGNED_TO.
7605 2013-09-18  Marek Polacek  <polacek@redhat.com>
7607         PR sanitizer/58411
7608         * doc/extend.texi: Document no_sanitize_undefined attribute.
7609         * builtins.c (fold_builtin_0): Don't sanitize function if it has the
7610         no_sanitize_undefined attribute.
7612 2013-09-18  Nick Clifton  <nickc@redhat.com>
7614         * config/msp430/msp430.h (ASM_SPEC): Pass -md on to the assembler.
7615         (ASM_DECLARE_FUNCTION_NAME): Define.
7617 2013-09-17  Trevor Saunders  <tsaunders@mozilla.com>
7619         * compare-elim.c (find_comparison_dom_walker): New class
7620         (find_comparisons_in_bb): Rename to
7621         find_comparison_dom_walker::before_dom_children
7622         (find_comparisons): Adjust
7623         * domwalk.c (walk_dominator_tree): Rename to dom_walker::walk, and
7624         adjust.
7625         (init_walk_dominator_tree, fini_walk_dominator_tree): Remove
7626         * domwalk.h (dom_walk_data): Convert it To a class dom_walker.
7627         (init_walk_dominator_tree): Remove declaration.
7628         (fini_walk_dominator_tree): Remove declaration.
7629         * fwprop.c (single_def_use_dom_walker): New class
7630         (single_def_use_enter_block): Convert to
7631         single_def_use_dom_walker::before_dom_children.
7632         (single_def_use_leave_block): Convert to
7633         single_def_use_dom_walker::after_dom_children.
7634         (build_single_def_use_links): Adjust.
7635         * gimple-ssa-strength-reduction.c (find_candidates_dom_walker): New
7636         class.
7637         (find_candidates_in_block): Convert to
7638         find_candidates_dom_walker::before_dom_children.
7639         (execute_strength_reduction): Adjust.
7640         * graphite-sese-to-poly.c (struct bsc, build_sese_conditions): Remove.
7641         (sese_dom_walker): New class.
7642         (sese_dom_walker::sese_dom_walker): New constructor.
7643         (sese_dom_walker::~sese_dom_walker): New destructor.
7644         (build_sese_conditions_before): Convert to
7645         sese_dom_walker::before_dom_children.
7646         (build_sese_conditions_after): Convert to
7647         sese_dom_walker::after_dom_children.
7648         (build_poly_scop): Adjust
7649         * tree-into-ssa.c (rewrite_dom_walker): New class
7650         (rewrite_enter_block): Convert to
7651         rewrite_dom_walker::before_dom_children.
7652         (rewrite_leave_block): Convert to
7653         rewrite_dom_walker::after_dom_children.
7654         (rewrite_update_dom_walker): New class.
7655         (rewrite_update_enter_block): Convert to
7656         rewrite_update_dom_walker::before_dom_children.
7657         (rewrite_update_leave_block): Convert to
7658         rewrite_update_dom_walker::after_dom_children.
7659         (rewrite_blocks, rewrite_into_ssa): Adjust.
7660         (mark_def_dom_walker): New class.
7661         (mark_def_dom_walker::mark_def_dom_walker): New constructor.
7662         (mark_def_dom_walker::~mark_def_dom_walker): New destructor.
7663         (mark_def_sites_blocks): Convert to
7664         mark_def_dom_walker::before_dom_children.
7665         (mark_def_site_blocks): Remove.
7666         * tree-ssa-dom.c (dom_opt_dom_walker): New class.
7667         (tree_ssa_dominator_optimize): Adjust.
7668         (dom_thread_across_edge): Convert to method
7669         dom_opt_dom_walker::thread_across_edge.
7670         (dom_opt_enter_block): Convert to member function
7671         dom_opt_dom_walker::before_dom_children.
7672         (dom_opt_leave_block): Convert to member function
7673         dom_opt_dom_walker::after_dom_children.
7674         * tree-ssa-dse.c (dse_dom_walker): New class.
7675         (dse_enter_block): Convert to member function
7676         dse_dom_walker::before_dom_children.
7677         (tree_ssa_dse): Adjust.
7678         * tree-ssa-loop-im.c (invariantness_dom_walker): New class.
7679         (determine_invariantness_stmt): Convert to method
7680         invariantness_dom_walker::before_dom_children.
7681         (determine_invariantness): Remove
7682         (move_computations_dom_walker): New class.
7683         (move_computations_stmt): Convert to method
7684         move_computations_dom_walker::before_dom_children.
7685         (move_computations, tree_ssa_lim): Adjust.
7686         * tree-ssa-phiopt.c (nontrapping_dom_walker): New class.
7687         (nt_init_block): Convert to method
7688         notrappping_dom_walker::before_dom_children.
7689         (nt_fini_block): Convert to method
7690         method nontrapping_dom_walker::after_dom_children.
7691         (get_non_trapping): Adjust.
7692         * tree-ssa-pre.c (eliminate_dom_walker): New class.
7693         (eliminate_bb): Convert to method
7694         eliminate_dom_walker::before_dom_children.
7695         (eliminate_leave_block): Convert to method
7696         eliminate_dom_walker::after_dom_children.
7697         (eliminate): Adjust.
7698         * tree-ssa-strlen.c (strlen_dom_walker): New class.
7699         (strlen_enter_block): Convert to method
7700         strlen_dom_walker::before_dom_children.
7701         (strlen_leave_block): Convert to method
7702         method strlen_dom_walker::after_dom_children.
7703         (tree_ssa_strlen): Adjust.
7704         * tree-ssa-uncprop.c (uncprop_dom_walker): New class.
7705         (tree_ssa_uncprop): Adjust.
7706         (uncprop_leave_block): Convert to method
7707         uncprop_dom_walker::after_dom_children.
7708         (uncprop_leave_block): Convert to method
7709         uncprop_dom_walker::before_dom_children.
7711 2013-09-18  Bin Cheng  <bin.cheng@arm.com>
7713         * config/arm/arm.c (thumb1_reorg): Search for flag setting insn before
7714         branch in same basic block.  Check both src and dest of the move insn.
7716 2013-09-17  Nick Clifton  <nickc@redhat.com>
7718         * config/rl78/rl78-real.md (bf): New pattern.
7719         (bt): New pattern.
7720         * config/rl78/rl78.c (rl78_print_operand_1): Handle %B.
7721         (rl78_print_operand): Do not put a # before a %B.
7722         * config/rl78/rl78.opt: Tweak doc strings.
7724 2013-09-17  DJ Delorie  <dj@redhat.com>
7726         * config/rl78/constraints.md (Wcv): Allow up to $r31.
7727         * config/rl78/rl78.c (rl78_asm_file_start: Likewise.
7728         (rl78_option_override): Likewise, if -mallregs.
7729         (is_virtual_register): Likewise.
7730         * config/rl78/rl78.h (reg_class): Extend VREGS to $r31.
7731         (REGNO_OK_FOR_BASE_P): Likewise.
7732         * config/rl78/rl78.opt (-mallregs): New.
7734 2013-09-17  Nick Clifton  <nickc@redhat.com>
7736         * config/rl78/rl78.c (need_to_save): Change return type to bool.
7737         For interrupt functions: save all call clobbered registers if the
7738         interrupt handler is not a leaf function.
7739         (rl78_expand_prologue): Always recompute the frame information.
7740         For interrupt functions: only select bank 0 if one of the bank 0
7741         registers is going to be psuhed.
7743 2013-09-17  DJ Delorie  <dj@redhat.com>
7745         * config/rl78/constraints.md: For each W* constraint, rename to C*
7746         and create a W* constraint that checks for an optional ES: prefix
7747         pattern also.
7748         * config/rl78/rl78.md (UNS_ES_ADDR): New.
7749         (es_addr): New.  Used to wrap far addresses.
7750         * config/rl78/rl78-protos.h (rl78_es_addr): New.
7751         (rl78_es_base): New.
7752         * config/rl78/rl78.c (rl78_as_legitimate_address): Accept "unspec"
7753         wrapped far addresses.
7754         (rl78_print_operand_1): Unwrap far addresses before processing.
7755         (rl78_lo16): Wrap far addresses in unspecs.
7756         (rl78_es_addr): New.
7757         (rl78_es_base): New.
7758         (insn_ok_now): Check for not-yet-wrapped far addresses.
7759         (transcode_memory_rtx): Properly re-wrap far addresses.
7761 2013-09-17  Sebastian Huber  <sebastian.huber@embedded-brains.de>
7763         * config/sparc/t-rtems: Add leon3 multilibs.
7765 2013-09-17  Cong Hou  <congh@google.com>
7767         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Fix a bug
7768         when checking the dot production pattern. The type of rhs operand
7769         of multiply is now checked correctly.
7771 2013-09-17  Jeff Law  <law@redhat.com>
7773         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
7774         edge implied equivalences into successor phis.
7775         * tree-ssa-threadupdate.c (phi_args_equal_on_edges): Moved into
7776         here from tree-ssa-threadedge.c.
7777         (mark_threaded_blocks): When threading through a joiner, if both
7778         successors of the joiner's clone reach the same block, verify the
7779         PHI arguments are equal.  If not, cancel the jump threading request.
7780         * tree-ssa-threadedge.c (phi_args_equal_on_edges): Moved into
7781         tree-ssa-threadupdate.c
7782         (thread_across_edge): Don't check PHI argument equality when
7783         threading through joiner block here.
7785 2013-09-17  Andrew MacLeod <amacleod@redhat.com>
7787         * tree-flow.h (ssa_undefined_value_p): Remove prototype.
7788         * tree-ssa.c (ssa_undefined_value_p): Move pass independent parts here.
7789         (warn_uninit, warn_uninitialized_vars,
7790         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7791         Move to tree-ssa-uninit.c.
7792         * tree-ssa-uninit.c (ssa_undefined_value_p): Move to tree-ssa.c.
7793         (has_undefined_value_p): New.  Pass dependant parts of
7794         ssa_undefined_value_p.
7795         (uninit_undefined_value_p): Use has_undefined_value_p.
7796         (warn_uninit, warn_uninitialized_vars,
7797         execute_early_warn_uninitialized, make_pass_early_warn_uninitialized):
7798         Move from tree-ssa.c.
7799         * tree-ssa.h: Adjust prototypes.
7801 2013-09-17  Jan Hubicka  <jh@suse.cz>
7803         PR middle-end/58332
7804         * cif-code.def (FUNCTION_NOT_OPTIMIZED): New CIF code.
7805         * ipa-inline.c (can_inline_edge_p): Do not downgrade
7806         FUNCTION_NOT_OPTIMIZED.
7807         * ipa-inline-analysis.c (compute_inline_parameters): Function
7808         not optimized is not inlinable unless it is alwaysinline.
7809         (inline_analyze_function): Force calls in not optimized
7810         function not inlinable.
7812 2013-09-17  Jan Hubicka  <jh@suse.cz>
7814         PR middle-end/58329
7815         * ipa-devirt.c (ipa_devirt): Be ready for symtab_nonoverwritable_alias
7816         to return NULL.
7817         * ipa.c (function_and_variable_visibility): Likewise.
7818         * ipa-profile.c (ipa_profile): Likewise.
7820 2013-09-17  Bernd Edlinger  <bernd.edlinger@hotmail.de>
7822         PR ipa/58398
7823         * cgraph.c (cgraph_function_body_availability): Check for ifunc
7824         attribute, and don't inline the resolver in this case.
7826 2013-09-17  Teresa Johnson  <tejohnson@google.com>
7828         * coverage.c (get_coverage_counts): Add missing newline.
7830 2013-09-17  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
7832         PR tree-optimization/58088
7833         * fold-const.c (mask_with_trailing_zeros): New function.
7834         (fold_binary_loc): Make sure we don't recurse infinitely
7835         when the X in (X & C1) | C2 is a tree of the form (Y * K1) & K2.
7836         Use mask_with_trailing_zeros where appropriate.
7838 2013-09-17  Yuri Rumyantsev  <ysrumyan@gmail.com>
7840         * config/i386/i386.c (distance_agu_use_in_bb) : Proper initialization
7841         of 'prev' var to get better distance estimation.
7843 2013-09-17  Eric Botcazou  <ebotcazou@adacore.com>
7845         * tree-inline.h (struct copy_body_data): Add transform_parameter.
7846         * tree-inline.c (is_parameter_of): New predicate.
7847         (remap_gimple_op_r): Do not propagate TREE_THIS_NOTRAP on MEM_REF if
7848         a parameter has been remapped.
7849         (copy_tree_body_r): Likewise on INDIRECT_REF and MEM_REF.
7850         (optimize_inline_calls): Initialize transform_parameter.
7851         (copy_gimple_seq_and_replace_locals): Likewise.
7852         (tree_function_versioning): Likewise.
7853         (maybe_inline_call_in_expr): Likewise.
7855 2013-09-17  Nick Clifton  <nickc@redhat.com>
7857         * config/msp430/msp430-protos.h: Add prototypes for new functions.
7858         * config/msp430/msp430.c (msp430_preserve_reg_p): Add support for
7859         interrupt handlers.
7860         (is_attr_func): New function.
7861         (msp430_is_interrupt_func): New function.
7862         (is_naked_func): New function.
7863         (is_reentrant_func): New function.
7864         (is_critical_func): New function.
7865         (msp430_start_function): Add annotations for function attributes.
7866         (msp430_attr): New function.
7867         (msp430_attribute_table): New.
7868         (msp430_function_section): New function.
7869         (TARGET_ASM_FUNCTION_SECTION): Define.
7870         (msp430_builtin): New enum.
7871         (msp430_init_builtins): New function.
7872         (msp430_builtin_devl): New function.
7873         (msp430_expand_builtin): New function.
7874         (TARGET_INIT_BUILTINS): Define.
7875         (TARGET_EXPAND_BUILTINS): Define.
7876         (TARGET_BUILTIN_DECL): Define.
7877         (msp430_expand_prologue): Add support for naked, interrupt,
7878         critical and reentrant functions.
7879         (msp430_expand_epilogue): Likewise.
7880         (msp430_print_operand): Handle 'O' character.
7881         * config/msp430/msp430.h (TARGET_CPU_CPP_BUILTINS): Define
7882         NO_TRAMPOLINES.
7883         * config/msp430/msp430.md (unspec): Add UNS_DINT, UNS_EINT,
7884         UNS_PUSH_INTR, UNS_POP_INTR, UNS_BIC_SR, UNS_BIS_SR.
7885         (pushm): Use a 'n' rather than an 'i' constraint.
7886         (msp_return): Add generation of the interrupt return instruction.
7887         (disable_interrupts): New pattern.
7888         (enable_interrupts): New pattern.
7889         (push_intr_state): New pattern.
7890         (pop_intr_state): New pattern.
7891         (bic_SR): New pattern.
7892         (bis_SR): New pattern.
7893         * doc/extend.texi: Document MSP430 function attributes and builtin
7894         functions.
7896 2013-09-17  Richard Biener  <rguenther@suse.de>
7898         PR tree-optimization/58432
7899         * tree-loop-distribution.c (tree_loop_distribution): Also
7900         scan PHIs for outside loop uses and seed a partition from them.
7902 2013-09-17  Bin Cheng  <bin.cheng@arm.com>
7904         * gimple-ssa-strength-reduction.c (backtrace_base_for_ref): New.
7905         (restructure_reference): Call backtrace_base_for_ref.
7907 2013-09-17  Alan Modra  <amodra@gmail.com>
7909         PR target/57589
7910         * config/rs6000/driver-rs6000.c (elf_platform): Revert 2013-06-11
7911         patch.
7913 2013-09-16  DJ Delorie  <dj@redhat.com>
7915         * config/rl78/rl78.c (rl78_asm_file_start): Specify alternate
7916         vregs location for RL78/G10.
7917         (rl78_expand_prologue): Avoid SEL on G10.
7918         (rl78_expand_epilogue): Likewise.
7919         (rl78_peep_movhi_p): Can't move a constant to memory in HImode.
7920         * config/rl78/rl78.h (TARGET_CPU_CPP_BUILTINS): Define
7921         __RL78_G10__ when appropriate.
7922         (ASM_SPEC): Pass -mg10 along to the assembler.
7923         * config/rl78/rl78.md (sel_rb): Disable for G10.
7924         * config/rl78/rl78.opt: Add -mg10 option.
7925         * config/rl78/t-rl78: Add -mg10 multilib.
7927 2013-09-16  Xinliang David Li  <davidxl@google.com>
7929         * tree-if-conv.c (main_tree_if_conversion): Check new flag.
7930         * omp-low.c (omp_max_vf): Ditto.
7931         (expand_omp_simd): Ditto.
7932         * tree-vectorizer.c (vectorize_loops): Ditto.
7933         (gate_vect_slp): Ditto.
7934         (gate_increase_alignment): Ditto.
7935         * tree-ssa-pre.c (inhibit_phi_insertion): Ditto.
7936         * tree-ssa-loop.c (gate_tree_vectorize): Ditto.
7937         (gate_tree_vectorize): Name change.
7938         (tree_vectorize): Ditto.
7939         (pass_vectorize::gate): Call new function.
7940         (pass_vectorize::execute): Ditto.
7941         * opts.c: O3 default setting change.
7942         (finish_options): Check new flag.
7943         * doc/invoke.texi: Document new flags.
7944         * common.opt: New flags.
7946 2013-09-16  Andreas Schwab  <schwab@linux-m68k.org>
7948         * doc/tm.texi.in (Cond Exec Macros): Remove node.
7949         (Condition Code): Don't reference it.
7950         * doc/tm.texi: Regenerate.
7952 2013-09-16  Vladimir Makarov  <vmakarov@redhat.com>
7954         PR middle-end/58418
7955         * lra-constraints.c (undo_optional_reloads): Consider all optional
7956         reload even if it did not get a hard reg.
7958 2013-09-16  Teresa Johnson  <tejohnson@google.com>
7960         * dumpfile.c (dump_loc): Remove newline emission.
7961         * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Add newline
7962         emission to dump_printf_loc calls where missing.
7963         (vect_mark_for_runtime_alias_test): Ditto.
7964         (vect_analyze_data_ref_dependence): Ditto.
7965         (vect_analyze_data_ref_dependences): Ditto.
7966         (vect_slp_analyze_data_ref_dependence): Ditto.
7967         (vect_slp_analyze_data_ref_dependences): Ditto.
7968         (vect_compute_data_ref_alignment): Ditto.
7969         (vect_update_misalignment_for_peel): Ditto.
7970         (vect_verify_datarefs_alignment): Ditto.
7971         (vector_alignment_reachable_p): Ditto.
7972         (vect_get_data_access_cost): Ditto.
7973         (vect_enhance_data_refs_alignment): Ditto.
7974         (vect_find_same_alignment_drs): Ditto.
7975         (vect_analyze_data_refs_alignment): Ditto.
7976         (vect_analyze_group_access): Ditto.
7977         (vect_analyze_data_ref_access): Ditto.
7978         (vect_analyze_data_ref_accesses): Ditto.
7979         (vect_prune_runtime_alias_test_list): Ditto.
7980         (vect_analyze_data_refs): Ditto.
7981         (vect_create_addr_base_for_vector_ref): Ditto.
7982         (vect_create_data_ref_ptr): Ditto.
7983         (vect_grouped_store_supported): Ditto.
7984         (vect_grouped_load_supported): Ditto.
7985         * value-prof.c (check_counter): Ditto.
7986         (check_ic_target): Ditto.
7987         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Ditto.
7988         (vect_recog_widen_mult_pattern): Ditto.
7989         (vect_recog_widen_sum_pattern): Ditto.
7990         (vect_recog_over_widening_pattern): Ditto.
7991         (vect_recog_widen_shift_pattern): Ditto.
7992         (vect_recog_rotate_pattern): Ditto.
7993         (vect_recog_vector_vector_shift_pattern): Ditto.
7994         (vect_recog_divmod_pattern): Ditto.
7995         (vect_recog_mixed_size_cond_pattern): Ditto.
7996         (vect_recog_bool_pattern): Ditto.
7997         (vect_pattern_recog_1): Ditto.
7998         (vect_pattern_recog): Ditto.
7999         * tree-vect-loop.c (vect_determine_vectorization_factor): Ditto.
8000         (vect_is_simple_iv_evolution): Ditto.
8001         (vect_analyze_scalar_cycles_1): Ditto.
8002         (vect_get_loop_niters): Ditto.
8003         (vect_analyze_loop_1): Ditto.
8004         (vect_analyze_loop_form): Ditto.
8005         (vect_analyze_loop_operations): Ditto.
8006         (vect_analyze_loop_2): Ditto.
8007         (vect_analyze_loop): Ditto.
8008         (report_vect_op): Ditto.
8009         (vect_is_slp_reduction): Ditto.
8010         (vect_is_simple_reduction_1): Ditto.
8011         (vect_get_known_peeling_cost): Ditto.
8012         (vect_estimate_min_profitable_iters): Ditto.
8013         (vect_model_reduction_cost): Ditto.
8014         (vect_model_induction_cost): Ditto.
8015         (get_initial_def_for_induction): Ditto.
8016         (vect_create_epilog_for_reduction): Ditto.
8017         (vectorizable_reduction): Ditto.
8018         (vectorizable_induction): Ditto.
8019         (vectorizable_live_operation): Ditto.
8020         (vect_loop_kill_debug_uses): Ditto.
8021         (vect_transform_loop): Ditto.
8022         * tree-vect-stmts.c (vect_mark_relevant): Ditto.
8023         (vect_stmt_relevant_p): Ditto.
8024         (process_use): Ditto.
8025         (vect_mark_stmts_to_be_vectorized): Ditto.
8026         (vect_model_simple_cost): Ditto.
8027         (vect_model_promotion_demotion_cost): Ditto.
8028         (vect_model_store_cost): Ditto.
8029         (vect_get_store_cost): Ditto.
8030         (vect_model_load_cost): Ditto.
8031         (vect_get_load_cost): Ditto.
8032         (vect_init_vector_1): Ditto.
8033         (vect_get_vec_def_for_operand): Ditto.
8034         (vect_finish_stmt_generation): Ditto.
8035         (vectorizable_call): Ditto.
8036         (vectorizable_conversion): Ditto.
8037         (vectorizable_assignment): Ditto.
8038         (vectorizable_shift): Ditto.
8039         (vectorizable_operation): Ditto.
8040         (vectorizable_store): Ditto.
8041         (vectorizable_load): Ditto.
8042         (vectorizable_condition): Ditto.
8043         (vect_analyze_stmt): Ditto.
8044         (vect_transform_stmt): Ditto.
8045         (vect_is_simple_use): Ditto.
8046         * tree-vect-loop-manip.c (slpeel_make_loop_iterate_ntimes): Ditto.
8047         (vect_can_advance_ivs_p): Ditto.
8048         (vect_update_ivs_after_vectorizer): Ditto.
8049         (vect_do_peeling_for_loop_bound): Ditto.
8050         (vect_gen_niters_for_prolog_loop): Ditto.
8051         (vect_update_inits_of_drs): Ditto.
8052         (vect_create_cond_for_alias_checks): Ditto.
8053         * tree-vect-slp.c (vect_get_and_check_slp_defs): Ditto.
8054         (vect_build_slp_tree_1): Ditto.
8055         (vect_supported_load_permutation_p): Ditto.
8056         (vect_analyze_slp_instance): Ditto.
8057         (vect_analyze_slp): Ditto.
8058         (vect_make_slp_decision): Ditto.
8059         (vect_detect_hybrid_slp): Ditto.
8060         (vect_bb_vectorization_profitable_p): Ditto.
8061         (vect_slp_analyze_bb_1): Ditto.
8062         (vect_update_slp_costs_according_to_vf): Ditto.
8063         (vect_get_mask_element): Ditto.
8064         (vect_transform_slp_perm_load): Ditto.
8065         (vect_schedule_slp_instance): Ditto.
8066         (vect_schedule_slp): Ditto.
8067         (vect_slp_transform_bb): Ditto.
8068         * profile.c (read_profile_edge_counts): Ditto.
8069         (compute_branch_probabilities): Ditto.
8070         * coverage.c (get_coverage_counts): Ditto.
8072 2013-09-16  Diego Novillo  <dnovillo@google.com>
8074         * tree-core.h: Add missing comment lines from refactoring of tree.h.
8076 2013-09-16  Jan Hubicka  <jh@suse.cz>
8078         * gimple-fold.c (can_refer_decl_in_current_unit_p): Do not accept
8079         abstract functions; for static functions check the presence of body.
8081 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8083         * config/aarch64/aarch64-simd-builtins.def (fma): New.
8084         * config/aarch64/aarch64-simd.md
8085         (aarch64_mla_elt<mode>): New.
8086         (aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
8087         (aarch64_mls_elt<mode>): Likewise.
8088         (aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
8089         (aarch64_fma4_elt<mode>): Likewise.
8090         (aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
8091         (aarch64_fma4_elt_to_128v2df): Likewise.
8092         (aarch64_fma4_elt_to_64df): Likewise.
8093         (fnma<mode>4): Likewise.
8094         (aarch64_fnma4_elt<mode>): Likewise.
8095         (aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8096         (aarch64_fnma4_elt_to_128v2df): Likewise.
8097         (aarch64_fnma4_elt_to_64df): Likewise.
8098         * config/aarch64/iterators.md (VDQSF): New.
8099         * config/aarch64/arm_neon.h
8100         (vfm<as><sdq>_lane<q>_f<32, 64>): Convert to C implementation.
8101         (vml<sa><q>_lane<q>_<fsu><16, 32, 64>): Likewise.
8103 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8105         * config/aarch64/aarch64-simd.md (aarch64_mul3_elt<mode>): New.
8106         (aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8107         (aarch64_mul3_elt_to_128df): Likewise.
8108         (aarch64_mul3_elt_to_64v2df): Likewise.
8109         * config/aarch64/iterators.md (VEL): Also handle DFmode.
8110         (VMUL): New.
8111         (VMUL_CHANGE_NLANES) Likewise.
8112         (h_con): Likewise.
8113         (f): Likewise.
8114         * config/aarch64/arm_neon.h
8115         (vmul<q>_lane<q>_<suf><16,32,64>): Convert to C implementation.
8117 2013-09-16  James Greenhalgh  <james.greenhalgh@arm.com>
8119         * config/aarch64/arm_neon.h
8120         (vcvtx_high_f32_f64): Fix parameters.
8122 2013-09-16  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8123             Uros Bizjak  <ubizjak@gmail.com>
8125         * config/alpha.c: Include tree-ssanames.h.
8127 2013-09-16  Richard Biener  <rguenther@suse.de>
8129         * tree-loop-distribution.c (enum rdg_dep_type): Add control_dd.
8130         (dot_rdg_1): Handle control_dd.
8131         (create_edge_for_control_dependence): New function.
8132         (create_rdg_edges): Add control dependences if asked for.
8133         (build_rdg): Likewise.
8134         (generate_loops_for_partition): If there are not necessary
8135         control stmts remove all their dependencies.
8136         (collect_condition_stmts, rdg_flag_loop_exits): Remove.
8137         (distribute_loop): Pass on control dependences.
8138         (tree_loop_distribution): Compute control dependences and remove
8139         restriction on number of loop nodes.
8141 2013-09-16  Jakub Jelinek  <jakub@redhat.com>
8143         * ipa-prop.c (ipa_compute_jump_functions_for_edge): Return early
8144         for internal calls.
8146 2013-09-16  Richard Sandiford  <rdsandiford@googlemail.com>
8148         * cse.c (try_const_anchors): Punt on CC modes.
8150 2013-09-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
8152         * config/vax/constraints.md (T): Add missing CONSTANT_P check.
8154 2013-09-14  John David Anglin  <danglin@gcc.gnu.org>
8156         PR target/58382
8157         * config/pa/pa.c (pa_expand_prologue): Change mode in gen_rtx_POST_INC
8158         calls to word_mode.
8160 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8162         PR target/48094
8163         * config/darwin.c (darwin_objc2_section): Note if ObjC Metadata is
8164         seen.
8165         (darwin_objc1_section): Likewise.
8166         (darwin_file_end): Emit Image Info section when required.
8168 2013-09-14  Jan Hubicka  <jh@suse.cz>
8170         * tree-into-ssa.c (gate_into_ssa): New.
8171         (pass_data_build_ssa): Use it.
8172         * cgraph.h (expand_thunk): Update prototype.
8173         * cgraphunit.c (analyze_function): Expand thunks early.
8174         (expand_thunk): Fix DECL_CONTEXT of reust_decl;
8175         build proper cgraph; set in_ssa_p; clear bogus TREE_ASM_WRITTEN;
8176         set lowered flag; do not add new function.
8177         (assemble_thunks_and_aliases): Update.
8178         * tree-ssa.c (gate_init_datastructures): New gate.
8179         (pass_data_init_datastructures): Use it.
8181 2013-09-14  Iain Sandoe  <iain@codesourcery.com>
8183         PR target/58269
8184         * config/i386/i386.c (ix86_function_arg_regno_p): Make Darwin use the
8185         xmm register set described in the psABI.
8187 2013-09-13  Evgeny Gavrin <e.gavrin@samsung.com>
8189         * dwarf2out.c (should_emit_struct_debug): Add check
8190         for type_decl variable is not NULL.
8192 2013-09-13  Jacek Caban  <jacek@codeweavers.com>
8194         * config.gcc: Use new winnt-c.c target hooks
8195         * config/t-winnt: New file
8196         * config/winnt-c.c: New file
8197         * doc/tm.texi.in: Document new hook
8198         * doc/tm.texi: Regenerated
8200 2013-09-13  Jan Hubicka  <jh@suse.cz>
8202         PR middle-end/58094
8203         * ipa-inline.c (check_callers): New function.
8204         (check_caller_edge): Remove.
8205         (want_inline_function_to_all_callers_p): Also permit alises that are
8206         called dirrectly.
8207         (inline_to_all_callers): Terminate the walk when devirtualization
8208         introduce new calls.
8210 2013-09-13  Jan Hubicka  <jh@suse.cz>
8212         * ipa-inline-analysis.c (struct growth_data): Add node.
8213         (do_estimate_growth_1): Fix detection of recursion.
8215 2013-09-13  Jakub Jelinek  <jakub@redhat.com>
8217         PR tree-optimization/58392
8218         * tree-cfg.c (move_sese_region_to_fn): Rename loop variable
8219         to avoid shadowing of outer loop variable.  If
8220         saved_cfun->has_simduid_loops or saved_cfun->has_force_vect_loops,
8221         replace_by_duplicate_decl simduid of loops that have it set and
8222         set dest_cfun->has_simduid_loops and/or
8223         dest_cfun->has_force_vect_loops.
8224         * omp-low.c (build_outer_var_ref): Call maybe_lookup_decl_in_outer_ctx
8225         instead of maybe_lookup_decl.
8226         * tree-inline.c (copy_loops): Change blocks_to_copy argument to id.
8227         Use id->blocks_to_copy instead of blocks_to_copy.  Adjust recursive
8228         call.  Copy over force_vect and copy and remap simduid.  Set
8229         cfun->has_simduid_loops and/or cfun->has_force_vect_loops.
8230         (copy_cfg_body): Remove blocks_to_copy argument.  Use
8231         id->blocks_to_copy instead of blocks_to_copy.  Adjust copy_loops
8232         caller.  Don't set cfun->has_simduid_loops and/or
8233         cfun->has_force_vect_loops here.
8234         (copy_body): Remove blocks_to_copy argument.  Adjust copy_cfg_body
8235         caller.
8236         (expand_call_inline, tree_function_versioning): Adjust copy_body
8237         callers.
8239 2013-09-13  Martin Jambor  <mjambor@suse.cz>
8241         PR bootstrap/58388
8242         * ipa-prop.c (try_make_edge_direct_simple_call): Be less strict in
8243         the assert if the edge was a speculative one.
8245 2013-09-13  Richard Biener  <rguenther@suse.de>
8247         * tree-data-ref.h (known_dependences_p): Move here ...
8248         * tree-loop-distribution.c (known_dependences_p): ... from here.
8249         (dump_rdg_component, debug_rdg_component): Remove.
8250         (dump_rdg): Adjust.
8251         (generate_loops_for_partition): Use gimple_uid instead of
8252         relying on matching stmt visit order.
8253         (rdg_build_partitions): Take starting stmt vector.
8254         (ldist_gen): Merge into ...
8255         (distribute_loop): ... this function.  Do not compute starting
8256         vertices vector.
8257         * tree-cfg.c (gimple_duplicate_bb): Copy UID for PHIs.
8259 2013-09-13  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8261         * config/arm/arm.md (arm_cmpsi_insn): Split rI alternative.
8262         Set type attribute correctly. Set predicable_short_it attribute.
8263         (cmpsi_shiftsi): Remove %? from output template.
8265 2013-09-13  Richard Biener  <rguenther@suse.de>
8267         * tree-loop-distribution.c (struct rdg_component,
8268         rdg_defs_used_in_other_loops_p, free_rdg_components,
8269         rdg_build_components): Remove.
8270         (stmts_from_loop): Do not record virtual PHIs.
8271         (generate_loops_for_partition): Skip virtual PHIs.
8272         (build_rdg_partition_for_component): Rename to ...
8273         (build_rdg_partition_for_vertex): ... this and adjust.
8274         (rdg_build_partitions): Take a vector of starting vertices
8275         instead of components.  Remove unnecessary leftover handling.
8276         (ldist_gen): Do not build components or record other stores.
8277         (distribute_loop): Do not distribute loops containing stmts
8278         with side-effects.
8280 2013-09-13  Christian Bruel  <christian.bruel@st.com>
8282         PR target/58314
8283         * config/sh/sh.md (mov<mode>_reg_reg): Allow memory reloads.
8285 2013-09-13  Kai Tietz  <ktietz@redhat.com>
8287         * config.gcc: Separate cases for mingw and cygwin targets,
8288         and add 64-bit cygwin target case.
8290         * config/i386/winnt-cxx.c (i386_pe_type_dllexport_p): Don't
8291         dll-export inline-functions.
8292         * config/i386/winnt.c (i386_pe_determine_dllexport_p): Likewise.
8295 2013-09-13  Jeff Law  <law@redhat.com>
8297         PR middle-end/58387
8298         Revert:
8299         2013-09-06  Jeff Law  <law@redhat.com>
8301         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8302         edge implied equivalences into successor phis.
8304 2013-09-12  DJ Delorie  <dj@redhat.com>
8306         * config/rl78/rl78-virt.md: Change from | to \; for asm line
8307         separators.
8309 2013-09-12  Brooks Moses  <bmoses@google.com>
8311         PR driver/42955
8312         * Makefile.in: Do not install driver binaries in $(target)/bin.
8314 2013-09-12  DJ Delorie  <dj@redhat.com>
8316         * config/rl78/rl78.opt (mrelax): New.
8317         * config/rl78/rl78.h (ASM_SPEC): New, pass on -mrelax to gas.
8318         * config/rl78/rl78.h (LINK_SPEC): New, pass on -mrelax to ld.
8320         * config/rl78/rl78.c (rl78_expand_prologue): Use AX to copy
8321         between SP and FP.
8322         (rl78_expand_epilogue): Likewise.
8324 2013-09-12  Vladimir Makarov  <vmakarov@redhat.com>
8326         PR middle-end/58335
8327         * lra-eliminations.c (remove_reg_equal_offset_note): New.
8328         (eliminate_regs_in_insn): Rewrite frame pointer to hard frame
8329         pointer elimination with using remove_reg_equal_offset_note.
8331 2013-09-12  DJ Delorie  <dj@redhat.com>
8333         * config/msp430/: New port.
8334         * config.gcc (msp430): Added.
8335         * doc/invoke.texi: Document MSP430 options.
8336         * doc/install.texi: Document msp430-elf
8337         * doc/md.texi: Document msp430-elf
8338         * doc/contrib.texi: Document msp430-elf
8340         * cfgexpand.c (expand_debug_expr): Avoid sign-extending SImode to
8341         PSImode.
8343 2013-09-12  Martin Jambor  <mjambor@suse.cz>
8345         PR ipa/58389
8346         * ipa-prop.c (remove_described_reference): Give up if the edge in the
8347         reference descriptor is NULL.
8348         (ipa_edge_removal_hook): If owning a reference descriptor, set its
8349         edge to NULL.
8351 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8353         * tree-flow.h (FREE_SSANAMES): Move to tree-ssanames.c
8354         (SSANAMES, MODIFIED_NORETURN_CALLS, DEFAULT_DEFS, ptr_info_def,
8355         num_ssa_names, ssa_name): Move to tree-ssanames.h + prototypes.
8356         * tree-flow-inline.h (make_ssa_name, copy_ssa_name, duplicate_ssa_name,
8357         make_temp_ssa_name): move to tree-ssanames.h
8358         * tree-ssa-alias.h: Move prototype.
8359         * tree-ssa.h: Include tree-ssanames.h.
8360         * tree-ssanames.c (FREE_SSANAMES): Move to here.
8361         * tree-ssanames.h: New.  Move items from tree-flow*.h
8362         * Makefile.in (tree-ssanames.h): Add to tree-ssanames.o and GTFILES.
8364 2013-09-12  Richard Biener  <rguenther@suse.de>
8366         PR tree-optimization/58404
8367         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Also
8368         propagate non-invariant addresses into dereferences wrapped
8369         in component references.
8371 2013-09-12  Richard Biener  <rguenther@suse.de>
8373         PR tree-optimization/58402
8374         * passes.def: Move pass_late_warn_uninitialized later.
8376 2013-09-12  Andrew MacLeod  <amacleod@redhat.com>
8378         * tree-ssa.h: New.  Move content from tree-flow.h and
8379         tree-flow-inline.h.
8380         * tree-flow.h (_edge_var_map, edge_var_map_vector): Move to tree-ssa.h.
8381         Move prototypes belonging to tree-ssa.c.
8382         * tree-flow-inline.h (redirect_edge_var_map_def,
8383         redirect_edge_var_map_result, redirect_edge_var_map_location): Move to
8384         tree-ssa.h.
8385         * gimple.h: Adjust prototypes.
8386         * tree-ssa.c (useless_type_conversion_p, types_compatible_p): Move
8387         to...
8388         * gimple.c (useless_type_conversion_p, types_compatible_p): Here.
8389         * tree.h: Move prototype to tree-ssa.h.
8390         * gengtype.c (open_base_files): Replace tree-flow.h with tree-ssa.h.
8391         * Makefile.in: (TREE_SSA_H, TREE_FLOW_H): Adjust dependencies.
8392         * alias.c, asan.c, builtins.c, calls.c, cfgexpand.c, cfghooks.c,
8393         cfgloop.c, cfgloopmanip.c, cgraph.c, cgraphbuild.c, cgraphclones.c,
8394         cgraphunit.c, dse.c, except.c, expr.c, final.c, fold-const.c,
8395         ggc-page.c, gimple-fold.c, gimple-iterator.c, gimple-low.c,
8396         gimple-pretty-print.c, gimple-ssa-strength-reduction.c,
8397         gimple-streamer-in.c, gimple-streamer-out.c, gimple.c, gimplify.c,
8398         graphite-blocking.c, graphite-clast-to-gimple.c,
8399         graphite-dependences.c, graphite-interchange.c,
8400         graphite-optimize-isl.c, graphite-poly.c, graphite-scop-detection.c,
8401         graphite-sese-to-poly.c, graphite.c, ipa-cp.c, ipa-inline-analysis.c,
8402         ipa-inline-transform.c, ipa-inline.c, ipa-prop.c, ipa-pure-const.c,
8403         ipa-reference.c, ipa-split.c, ipa-utils.c,
8404         loop-init.c, lto-cgraph.c, lto-section-in.c, lto-section-out.c,
8405         lto-streamer-in.c, lto-streamer-out.c, lto-streamer.c, omp-low.c,
8406         passes.c, predict.c, print-tree.c, profile.c, sese.c, targhooks.c,
8407         tracer.c, trans-mem.c, tree-call-cdce.c, tree-cfg.c, tree-cfgcleanup.c,
8408         tree-chrec.c, tree-complex.c, tree-data-ref.c, tree-dfa.c, tree-eh.c,
8409         tree-emutls.c, tree-if-conv.c, tree-inline.c, tree-into-ssa.c,
8410         tree-loop-distribution.c, tree-mudflap.c, tree-nested.c, tree-nrv.c,
8411         tree-object-size.c, tree-optimize.c, tree-outof-ssa.c, tree-parloops.c,
8412         tree-phinodes.c, tree-predcom.c, tree-pretty-print.c, tree-profile.c,
8413         tree-scalar-evolution.c, tree-sra.c, tree-ssa*.c, tree-stdarg.c,
8414         tree-streamer-in.c, tree-switch-conversion.c, tree-tailcall.c,
8415         tree-vect-data-refs.c, tree-vect-generic.c, tree-vect-loop-manip.c,
8416         tree-vect-loop.c, tree-vect-patterns.c, tree-vect-slp.c,
8417         tree-vect-stmts.c, tree-vectorizer.c, tree-vrp.c, tsan.c,
8418         value-prof.c, var-tracking.c,
8419         varpool.c, vtable-verify.c: Replace tree-flow.h with tree-ssa.h
8421 2013-09-12  Richard Biener  <rguenther@suse.de>
8423         PR tree-optimization/58396
8424         * tree-loop-distribution.c (create_rdg_edges): Free unused DDRs.
8425         (build_rdg): Take a loop-nest parameter, fix memleaks.
8426         (distribute_loop): Compute loop-nest here and pass it to build_rdg.
8428 2013-09-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
8430         * config/i386/x86-tune.def: Turn on X86_TUNE_AVOID_MEM_OPND_FOR_CMOVE
8431         for SLM.
8433 2013-09-12  Cameron McInally  <cameron.mcinally@nyu.edu>
8435         * doc/extend.texi: Fix errors in x86 FMA builtin naming.
8436         The FMA instruction names should have a 'v' prefix.
8438 2013-09-12  Richard Biener  <rguenther@suse.de>
8440         * tree-loop-distribution.c (dot_rdg_1): Make graph prettier.
8441         (dot_rdg): Use popen instead of system in optional code.
8442         (remaining_stmts, upstream_mem_writes): Remove global bitmaps.
8443         (already_processed_vertex_p): Adjust.
8444         (has_anti_or_output_dependence, predecessor_has_mem_write,
8445         mark_nodes_having_upstream_mem_writes, has_upstream_mem_writes,
8446         rdg_flag_uses): Remove.
8447         (rdg_flag_vertex): Simplify.
8448         (rdg_flag_vertex_and_dependent): Rely on a correct RDG and
8449         remove recursion.
8450         (build_rdg_partition_for_component): Process the first vertex
8451         of a component only.
8452         (ldist_gen): Do not compute remaining_stmts or upstream_mem_writes.
8454 2013-09-12  Alan Modra  <amodra@gmail.com>
8456         * config/rs6000/rs6000.c (toc_relative_expr_p): Use add_cint_operand.
8458 2013-09-11  DJ Delorie  <dj@redhat.com>
8459             Nick Clifton  <nickc@redhat.com>
8461         * config/rl78/predicates.md (rl78_cmp_operator_signed): New.
8462         (rl78_stack_based_mem): New.
8463         * config/rl78/constraints.md (Iv08): New.
8464         (Iv16): New.
8465         (Iv24): New.
8466         (Is09): New.
8467         (Is17): New.
8468         (Is25): New.
8469         (ISsi): New.
8470         (IShi): New.
8471         (ISqi): New.
8472         * config/rl78/rl78-expand.md (movqi): Reject more SUBREG operands.
8473         (movhi): Likewise.
8474         (movsi): Change from expand to insn-and-split.
8475         (ashrsi3): Clobber AX.
8476         (lshrsi3): New.
8477         (ashlsi3): New.
8478         (cbranchsi4): New.
8479         * config/rl78/rl78.md (CC_REG): Fix.
8480         (addsi3): Allow memory and immediate operands.
8481         (addsi3_internal): Split into...
8482         (addsi3_internal_virt): ...new, and ...
8483         (addsi3_internal_real): ...new.
8484         (subsi): New.
8485         (subsi3_internal_virt): New.
8486         (subsi3_internal_real): New.
8487         (mulsi3): Add memory operand.
8488         (mulsi3_rl78): Likewise.
8489         (mulsi3_g13): Likewise.
8490         * config/rl78/rl78-real.md (cbranchqi4_real_signed): New.
8491         (cbranchqi4_real): Add more constraint options.
8492         (cbranchhi4_real): Expand pattern.
8493         (cbranchhi4_real_signed): New.
8494         (cbranchhi4_real_inverted): New.
8495         (cbranchsi4_real_lt): New.
8496         (cbranchsi4_real_ge): New.
8497         (cbranchsi4_real_signed): New.
8498         (cbranchsi4_real): New.
8499         (peephole2): New.
8500         * config/rl78/rl78-virt.md (ashrsi3_virt): Add custom cases for
8501         constant shifts.
8502         (lshrsi3_virt): Likewise.
8503         (ashlsi3_virt): Likewise.
8504         (cbranchqi4_virt_signed): New.
8505         (cbranchhi4_virt_signed): New.
8506         (cbranchsi4_virt): New.
8507         * config/rl78/rl78.c: Whitespace fixes throughout.
8508         (move_elim_pass): New.
8509         (pass_data_rl78_move_elim): New.
8510         (pass_rl78_move_elim): New.
8511         (make_pass_rl78_move_elim): New.
8512         (rl78_devirt_info): Run devirt earlier.
8513         (rl78_move_elim_info): New.
8514         (rl78_asm_file_start): Register it.
8515         (rl78_split_movsi): New.
8516         (rl78_as_legitimate_address): Allow virtual base registers when
8517         appropriate.
8518         (rl78_addr_space_convert): Remove spurious debug stuff.
8519         (rl78_print_operand_1): Add z,s,S,r,E modifiers.
8520         (rl78_print_operand): More cases for not printing '#'.
8521         (rl78_expand_compare): Remove most of the logic.
8522         (content_memory): New.
8523         (clear_content_memory): New.
8524         (get_content_index): New.
8525         (get_content_name): New.
8526         (display_content_memory): New.
8527         (update_content): New.
8528         (record_content): New.
8529         (already_contains): New.
8530         (insn_ok_now): Re-recog insns with virtual registers.
8531         (add_postponed_content_update): New.
8532         (process_postponed_content_update): New.
8533         (gen_and_emit_move): New.
8534         (transcode_memory_rtx): Record new location content.
8535         Use gen_and_emit_move.
8536         (force_into_acc): New.
8537         (move_to_acc): Use gen_and_emit_move.
8538         (move_from_acc): Likewise.
8539         (move_acc_to_reg): Likewise.
8540         (move_to_x): Likewise.
8541         (move_to_hl): Likewise.
8542         (move_to_de): Likewise.
8543         (rl78_alloc_physical_registers_op1): Record location content.
8544         (has_constraint): New.
8545         (rl78_alloc_physical_registers_op2): Record location content.
8546         Optimize use of HL.
8547         (rl78_alloc_physical_registers_ro1): Likewise.
8548         (rl78_alloc_physical_registers_cmp): Likewise.
8549         (rl78_alloc_physical_registers_umul): Likewise.
8550         (rl78_alloc_address_registers_macax): New.
8551         (rl78_alloc_physical_registers): Initialize and set location
8552         content memory as needed.
8553         (rl78_reorg): Make sure split2 is called.
8554         (rl78_rtx_costs): New.
8556 2013-09-11  Richard Sandiford  <rdsandiford@googlemail.com>
8558         * simplify-rtx.c (simplify_unary_operation_1): Use simplify_gen_binary
8559         for (not (neg ...)) and (neg (not ...)) cases.
8561 2013-09-11  Richard Biener  <rguenther@suse.de>
8563         PR middle-end/58377
8564         * passes.def: Split critical edges before late uninit warning passes.
8565         * tree-cfg.c (pass_split_crit_edges): Implement clone method.
8567 2013-09-11  Jakub Jelinek  <jakub@redhat.com>
8569         PR tree-optimization/58385
8570         * fold-const.c (build_range_check): If both low and high are NULL,
8571         use omit_one_operand_loc to preserve exp side-effects.
8573 2013-09-11  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8575         * config/arm/arm.md (arm_shiftsi3): New alternative l/l/M.
8577 2013-09-11  Richard Biener  <rguenther@suse.de>
8579         * tree-data-ref.c (dump_rdg_vertex, debug_rdg_vertex,
8580         dump_rdg_component, debug_rdg_component, dump_rdg, debug_rdg,
8581         dot_rdg_1, dot_rdg, rdg_vertex_for_stmt, create_rdg_edge_for_ddr,
8582         create_rdg_edges_for_scalar, create_rdg_edges, create_rdg_vertices,
8583         stmts_from_loop, known_dependences_p, build_empty_rdg,
8584         build_rdg, free_rdg, rdg_defs_used_in_other_loops_p): Move ...
8585         * tree-loop-distribution.c: ... here.
8586         * tree-data-ref.h (struct rdg_vertex, RDGV_STMT, RDGV_DATAREFS,
8587         RDGV_HAS_MEM_WRITE, RDGV_HAS_MEM_READS, RDG_STMT, RDG_DATAREFS,
8588         RDG_MEM_WRITE_STMT, RDG_MEM_READS_STMT, enum rdg_dep_type,
8589         struct rdg_edge, RDGE_TYPE, RDGE_LEVEL, RDGE_RELATION): Move ...
8590         * tree-loop-distribution.c: ... here.
8591         * tree-loop-distribution.c: Include gimple-pretty-print.h.
8592         (struct partition_s): Add loops member.
8593         (partition_alloc, partition_free, rdg_flag_uses, rdg_flag_vertex,
8594         rdg_flag_vertex_and_dependent, rdg_flag_loop_exits,
8595         build_rdg_partition_for_component, rdg_build_partitions): Adjust.
8597 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8598             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
8599             Sergey Lega  <sergey.s.lega@intel.com>
8600             Anna Tikhonova  <anna.tikhonova@intel.com>
8601             Ilya Tocar  <ilya.tocar@intel.com>
8602             Andrey Turetskiy  <andrey.turetskiy@intel.com>
8603             Ilya Verbin  <ilya.verbin@intel.com>
8604             Kirill Yukhin  <kirill.yukhin@intel.com>
8605             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
8607         * config/i386/constraints.md (k): New.
8608         (Yk): Ditto.
8609         * config/i386/i386.c (const regclass_map): Add new mask registers.
8610         (dbx_register_map): Ditto.
8611         (dbx64_register_map): Ditto.
8612         (svr4_dbx_register_map): Ditto.
8613         (ix86_conditional_register_usage): Squash mask registers if AVX512F is
8614         disabled.
8615         (ix86_preferred_reload_class): Disable constants for mask registers.
8616         (ix86_secondary_reload): Do spill of mask register using 32-bit insn.
8617         (ix86_hard_regno_mode_ok): Support new mask registers.
8618         (x86_order_regs_for_local_alloc): Ditto.
8619         * config/i386/i386.h (FIRST_PSEUDO_REGISTER): Update.
8620         (FIXED_REGISTERS): Add new mask registers.
8621         (CALL_USED_REGISTERS): Ditto.
8622         (REG_ALLOC_ORDER): Ditto.
8623         (VALID_MASK_REG_MODE): New.
8624         (FIRST_MASK_REG): Ditto.
8625         (LAST_MASK_REG): Ditto.
8626         (reg_class): Add MASK_EVEX_REGS, MASK_REGS.
8627         (MAYBE_MASK_CLASS_P): New.
8628         (REG_CLASS_NAMES): Add MASK_EVEX_REGS, MASK_REGS.
8629         (REG_CLASS_CONTENTS): Ditto.
8630         (MASK_REGNO_P): New.
8631         (ANY_MASK_REG_P): Ditto.
8632         (HI_REGISTER_NAMES): Add new mask registers.
8633         * config/i386/i386.md (MASK0_REG, MASK1_REG, MASK2_REG, MASK3_REG,
8634         MASK4_REG, MASK5_REG, MASK6_REG, MASK7_REG): Constants for new
8635         mask registers.
8636         (attribute "type"): Add mskmov, msklog.
8637         (attribute "length_immediate"): Support them.
8638         (attribute "memory"): Ditto.
8639         (attribute "prefix_0f"): Ditto.
8640         (*movhi_internal): Support new mask registers.
8641         (*movqi_internal): Ditto.
8642         (define_split): Split out clobber pattern is a logic
8643         insn on mask registers.
8644         (*k<logic><mode>): New.
8645         (*andhi_1): Extend to support mask regs.
8646         (*andqi_1): Extend to support mask regs.
8647         (kandn<mode>): New.
8648         (define_split): Split and-not to and and not if operands
8649         are not mask regs.
8650         (*<code><mode>_1): Separate HI mode to new pattern...
8651         (*<code>hi_1): This.
8652         (*<code>qi_1): Extend to support mask regs.
8653         (kxnor<mode>): New.
8654         (kortestzhi): Ditto.
8655         (kortestchi): Ditto.
8656         (kunpckhi): Ditto.
8657         (*one_cmpl<mode>2_1): Remove HImode and handle it...
8658         (*one_cmplhi2_1): ...Here, now with mask registers support.
8659         (*one_cmplqi2_1): Support new mask registers.
8660         (HI/QImode arithmetics splitter): Don't split if mask registers
8661         are used.
8662         (HI/QImode not splitter): Ditto.
8663         * config/i386/predicated.md (mask_reg_operand): New.
8664         (general_reg_operand): Ditto.
8666 2013-09-11  Alexander Ivchenko  <alexander.ivchenko@intel.com>
8668         * doc/invoke.texi: Document fxsr, xsave and xsaveopt options.
8669         * doc/extend.texi: Document fxsr, xsave and xsaveopt builtins.
8671 2013-09-10  Jeff Law  <law@redhat.com>
8673         PR tree-optimization/58380
8674         * tree-ssa-threadupdate.c (thread_block): Recognize another case
8675         of threading through a buried loop header.
8677         * tree-ssa-threadedge.c (thread_around_empty_blocks): Correct
8678         return value for single successor case.
8680 2013-09-10  Jan Hubicka  <jh@suse.cz>
8682         * ipa-devirt.c (ipa_devirt): Enable with LTO.
8684 2013-09-10  Richard Earnshaw  <rearnsha@arm.com>
8686         PR target/58361
8687         * arm/vfp.md (combine_vcvt_f32_<FCVTI32typename>): Fix pattern to
8688         support conditional execution.
8689         (combine_vcvt_f64_<FCVTI32typename>): Likewise.
8691 2013-09-10  Vladimir Makarov  <vmakarov@redhat.com>
8693         * lra.c (lra): Clear lra_optional_reload_pseudos before every
8694         constraint pass.
8695         * lra-constraints.c (curr_insn_transform): Switch on optional reloads.
8696         Check destination too to check move insn.
8697         (undo_optional_reloads): Add check that the original peudo did not
8698         changed its allocation and the optional reload was inherited on last
8699         inheritance pass.  Break loop after deciding to keep optional reload.
8700         (lra_undo_inheritance): Add check that inherited pseudo still in
8701         memory.
8703 2013-09-10  James Greenhalgh  <james.greenhalgh@arm.com>
8705         * config/aarch64/aarch64.md (generic_sched): New.
8706         * config/aarch64/aarch64-generic.md (load): Make conditional
8707         on generic_sched attribute.
8708         (nonload): Likewise.
8710 2013-09-10  Jan Hubicka  <jh@suse.cz>
8712         * lto-cgraph.c: Include ipa-utils.h.
8713         (compute_ltrans_boundary): Also add possible targets into the boundary.
8715 2013-09-10  Jan Hubicka  <jh@suse.cz>
8717         * gimple-fold.c (gimple_get_virt_method_for_binfo): Pass real
8718         VAR_DECL of vtable rather than full expression.
8720 2013-09-10  Jan Hubicka  <jh@suse.cz>
8721             Paolo Carlini  <paolo.carlini@oracle.com>
8723         * cgraphunit.c (analyze_functions): Save input_location, set it
8724         to UNKNOWN_LOCATION and restore it at the end.
8726 2013-09-10  Martin Jambor  <mjambor@suse.cz>
8728         * ipa-cp.c (propagate_constants_topo): Do not ignore SCC
8729         represented by a thunk.
8731 2013-09-10  Jeff Law  <law@redhat.com>
8733         PR tree-optimization/58343
8734         * tree-ssa-threadupdate.c (thread_block): Identify and disable
8735         jump threading requests through loop headers buried in the middle
8736         of a jump threading path.
8738         * tree-ssa-threadedge.c (thread_around_empty_blocks): Fix thinko
8739         in return value/type.
8741 2013-09-10  Jakub Jelinek  <jakub@redhat.com>
8743         PR rtl-optimization/58365
8744         * cfgcleanup.c (merge_memattrs): Also clear MEM_READONLY_P
8745         resp. MEM_NOTRAP_P if they differ, or set MEM_VOLATILE_P if
8746         it differs.
8748 2013-09-10  Richard Biener  <rguenther@suse.de>
8750         * tree-data-ref.h (build_rdg): Drop all parameters but loop.
8751         * tree-data-ref.c (create_rdg_vertices): Collect all data
8752         references, signal failure to the caller, use data-ref API.
8753         (build_rdg): Compute data references only once.  Maintain lifetime
8754         of data references and data dependences from within RDG.
8755         (free_rdg): Free dependence relations.
8756         * tree-loop-distribution.c (rdg_flag_uses): Drop weird code
8757         inventing extra dependences.
8758         (distribute_loop): Update for RDG API changes.
8760 2013-09-10  Kai Tietz  <ktietz@redhat.com>
8762         * doc/invoke.texi (fms-extensions): Document changed
8763         behavior for ms-abi targets.
8764         * config/i386/i386.c (ix86_option_override_internal):
8765         Set default value of option -fms-extension for ms-abi targets.
8767 2013-09-10  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
8769         * config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
8771 2013-09-10  Alan Modra  <amodra@gmail.com>
8773         PR target/58330
8774         * config/rs6000/rs6000.md (bswapdi2_64bit): Disable for volatile mems.
8776 2013-09-10  Alan Modra  <amodra@gmail.com>
8778         * config/rs6000/predicates.md (add_cint_operand): New.
8779         (reg_or_add_cint_operand, small_toc_ref): Use add_cint_operand.
8780         * config/rs6000/rs6000.md (largetoc_high_plus): Restrict offset
8781         using add_cint_operand.
8782         (largetoc_high_plus_aix): Likewise.
8784 2013-09-09  Jakub Jelinek  <jakub@redhat.com>
8786         PR tree-optimization/58364
8787         * tree-ssa-reassoc.c (init_range_entry): For BIT_NOT_EXPR on
8788         BOOLEAN_TYPE, only invert in_p and continue with arg0 if
8789         the current range can't be an unconditional true or false.
8791 2013-09-09  James Greenhalgh  <james.greenhalgh@arm.com>
8793         * config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
8795 2013-09-09  Uros Bizjak  <ubizjak@gmail.com>
8797         * ipa-prop.c (ipa_modify_call_arguments): Initialize deref_align.
8799 2013-09-09  Paolo Carlini  <paolo.carlini@oracle.com>
8801         PR c++/43452
8802         * doc/invoke.texi (-Wdelete-incomplete): Document it.
8804 2013-09-09  Ian Bolton  <ian.bolton@arm.com>
8806         * config/aarch64/aarch64.c (aarch64_preferred_reload_class): Return
8807         NO_REGS for immediate that can't be moved directly into FP_REGS.
8809 2013-09-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
8811         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
8812         comparison with negated operand.
8813         * config/aarch64/aarch64.md (compare_neg<mode>): Match canonical
8814         RTL form.
8816 2013-09-09  Richard Biener  <rguenther@suse.de>
8818         PR middle-end/58326
8819         * cfgloopmanip.c (fix_bb_placements): When fixing the placement
8820         of a subloop record all its block as affecting loop-closed SSA form.
8822 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8824         * expmed.c (lshift_value): Take an unsigned HOST_WIDE_INT instead
8825         of an rtx/bitpos pair.
8826         (store_fixed_bit_field): Update accordingly.
8828 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8830         * asan.c (asan_emit_stack_protection): Use gen_int_mode instead of
8831         GEN_INT.
8832         * builtins.c (expand_errno_check): Likewise.
8833         * dwarf2cfi.c (init_return_column_size): Likewise.
8834         * except.c (sjlj_mark_call_sites): Likewise.
8835         * expr.c (move_by_pieces_1, store_by_pieces_2): Likewise.
8836         * lra-constraints.c (emit_inc): Likewise.
8837         * ree.c (combine_set_extension): Likewise.
8838         * regmove.c (fixup_match_2): Likewise.
8839         * reload1.c (inc_for_reload): Likewise.
8841 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8843         * combine.c (simplify_set, expand_field_assignment, extract_left_shift)
8844         (force_to_mode, simplify_shift_const_1, simplify_comparison):
8845         Use gen_int_mode with the mode of the associated simplify_* call.
8846         * explow.c (probe_stack_range, anti_adjust_stack_and_probe): Likewise.
8847         * expmed.c (expand_shift_1): Likewise.
8848         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8849         * loop-iv.c (iv_number_of_iterations): Likewise.
8850         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8851         * simplify-rtx.c (simplify_binary_operation_1): Likewise.
8853 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8855         * asan.c (asan_clear_shadow): Use gen_int_mode with the mode
8856         of the associated expand_* call.
8857         (asan_emit_stack_protection): Likewise.
8858         * builtins.c (round_trampoline_addr): Likewise.
8859         * explow.c (allocate_dynamic_stack_space, probe_stack_range): Likewise.
8860         * expmed.c (expand_smod_pow2, expand_sdiv_pow2, expand_divmod)
8861         (emit_store_flag): Likewise.
8862         * expr.c (emit_move_resolve_push, push_block, emit_single_push_insn_1)
8863         (emit_push_insn, optimize_bitfield_assignment_op, expand_expr_real_1):
8864         Likewise.
8865         * function.c (instantiate_virtual_regs_in_insn): Likewise.
8866         * ifcvt.c (noce_try_store_flag_constants): Likewise.
8867         * loop-unroll.c (unroll_loop_runtime_iterations): Likewise.
8868         * modulo-sched.c (generate_prolog_epilog): Likewise.
8869         * optabs.c (expand_binop, widen_leading, expand_doubleword_clz)
8870         (expand_ctz, expand_ffs, expand_unop): Likewise.
8872 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8874         * alias.c (addr_side_effect_eval): Use gen_int_mode with the mode
8875         of the associated gen_rtx_* call.
8876         * caller-save.c (init_caller_save): Likewise.
8877         * combine.c (find_split_point, make_extraction): Likewise.
8878         (make_compound_operation): Likewise.
8879         * dwarf2out.c (mem_loc_descriptor): Likewise.
8880         * explow.c (plus_constant, probe_stack_range): Likewise.
8881         * expmed.c (expand_mult_const): Likewise.
8882         * expr.c (emit_single_push_insn_1, do_tablejump): Likewise.
8883         * reload1.c (init_reload): Likewise.
8884         * valtrack.c (cleanup_auto_inc_dec): Likewise.
8885         * var-tracking.c (adjust_mems): Likewise.
8886         * modulo-sched.c (sms_schedule): Likewise, but use gen_rtx_GT
8887         rather than gen_rtx_fmt_ee.
8889 2013-09-09  Jan Hubicka  <jh@suse.cz>
8891         PR middle-end/58294
8892         * value-prof.c (gimple_ic): Copy also abnormal edges.
8894 2013-09-09  Richard Sandiford  <rdsandiford@googlemail.com>
8896         * asan.c (asan_shadow_cst): Use gen_int_mode.
8898 2013-09-08  Jan Hubicka  <jh@suse.cz>
8900         * ipa-profile.c: Add toplevel comment.
8901         (ipa_propagate_frequency_1): Be more conservative when profile is read.
8902         (contains_hot_call_p): New function.
8903         (ipa_propagate_frequency): Set frequencies based on counts when
8904         profile is read.
8905         * predict.c (compute_function_frequency): Use PROFILE_READ gueard for
8906         profile; do not tamper with profile after inlining if it is read.
8908 2013-09-08  Jan Hubicka  <jh@suse.cz>
8910         * ipa-prop.c (try_make_edge_direct_simple_call): Do not special case
8911         speculative edges.
8913 2013-09-08  Jan Hubicka  <jh@suse.cz>
8915         * ipa.c (walk_polymorphic_call_targets): Fix redirection before IPA
8916         summary generation.
8918 2013-09-08  Jeff Law  <law@redhat.com>
8920         PR bootstrap/58340
8921         * tree-ssa-threadedge.c (thread_across_edge): Fix initialization
8922         of 'found'.
8924 2013-09-08  Andi Kleen  <ak@linux.intel.com>
8926         * tree-inline.c (estimate_num_insns): Limit asm cost to 1000.
8928 2013-09-08  Jan Hubicka  <jh@suse.cz>
8930         * ipa.c (walk_polymorphic_call_targets): Fix inliner summary update.
8932 2013-09-08  Richard Sandiford  <rdsandiford@googlemail.com>
8934         * ira.c (update_equiv_regs): Only call set_paradoxical_subreg
8935         for non-debug insns.
8936         * lra.c (new_insn_reg): Take the containing insn as a parameter.
8937         Only modify lra_reg_info[].biggest_mode if it's non-debug insn.
8938         (collect_non_operand_hard_regs, add_regs_to_insn_regno_info): Update
8939         accordingly.
8941 2013-09-08  Jan Hubicka  <jh@suse.cz>
8943         * cgraphunit.c (walk_polymorphic_call_targets): Permit 0 possible
8944         targets and devirtualize to BUILT_IN_UNREACHABLE.
8945         * timevar.def (TV_IPA_UNREACHABLE): New timevar.
8946         * ipa.c (walk_polymorphic_call_targets): New function.
8947         (symtab_remove_unreachable_nodes): Use it; do not keep all virtual
8948         functions; use the new timevar.
8949         * ipa-devirt.c (maybe_record_node): Do not insert static nodes that
8950         was removed from the program.
8951         (record_binfo): If BINFO corresponds to an anonymous namespace, we may
8952         not consider it in the walk when its vtable is dead.
8953         (possible_polymorphic_call_targets_1): Pass anonymous flag to
8954         record_binfo.
8955         (devirt_variable_node_removal_hook): New function.
8956         (possible_polymorphic_call_targets): Also register
8957         devirt_variable_node_removal_hook.
8958         (ipa_devirt): Do not do non-speculative devirtualization.
8959         (gate_ipa_devirt): One execute if devirtualizing speculatively.
8961 2013-09-08  Jan Hubicka  <jh@suse.cz>
8963         * cgraph.h (varpool_node_hook, varpool_node_hook_list,
8964         varpool_add_node_removal_hook, varpool_add_variable_insertion_hook,
8965         varpool_remove_variable_insertion_hook): Declare.
8966         * varpool.c (varpool_node_hook_list): New structure.
8967         (first_varpool_node_removal_hook,
8968         first_varpool_variable_insertion_hook): New variables.
8969         (varpool_add_node_removal_hook, varpool_remove_node_removal_hook,
8970         varpool_call_node_removal_hooks, varpool_add_variable_insertion_hook,
8971         varpool_remove_variable_insertion_hook,
8972         varpool_call_variable_insertion_hooks): New functions.
8973         (varpool_remove_node): Use it.
8975 2013-09-08  Paolo Carlini  <paolo.carlini@oracle.com>
8977         PR c++/54941
8978         * diagnostic.c (diagnostic_build_prefix): When s.file is
8979         "<built-in>" don't output line and column numbers.
8981 2013-09-06  Jan Hubicka  <jh@suse.cz>
8983         * cgraphunit.c (expand_thunk): Get body before touching arguments.
8984         * lto-streamer-out.c: Stream thunks, too.
8985         * lto-streamer-in.c (input_function): Pop cfun here
8986         (lto_read_body): Instead of here.
8988 2013-09-06  Caroline Tice  <cmtice@google.com>
8990         * doc/install.texi: Add documentation for the --enable-vtable-verify
8991         and the --disable-libvtv configure options.
8993 2013-09-06  Jeff Law  <law@redhat.com>
8995         * tree-ssa-dom.c (cprop_into_successor_phis): Also propagate
8996         edge implied equivalences into successor phis.
8998 2013-09-06  Joern Rennecke  <joern.rennecke@embecosm.com>
9000         * resource.c (mark_referenced_resources): Handle COND_EXEC.
9002 2013-09-06  Claudiu Zissulescu  <claziss@synopsys.com>
9004         * resource.c (mark_target_live_regs): Compute resources taking
9005         into account if a call is predicated or not.
9007 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9009         * toplev.c (output_stack_usage): Be prepared for suffixes created by
9010         the compiler in the function names.
9012 2013-09-06  Jan Hubicka  <jh@suse.cz>
9014         PR middle-end/58094
9015         * ipa-inline.c (has_caller_p): New function.
9016         (want_inline_function_to_all_callers_p): Use it.
9017         (sum_callers, inline_to_all_callers): Break out from ...
9018         (ipa_inline): ... here.
9020 2013-09-06  Jan Hubicka  <jh@suse.cz>
9022         * config/i386/i386.c (ix86_hard_regno_mode_ok): AVX modes are valid
9023         only when AVX is enabled.
9025 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9027         * config/aarch64/aarch64.md
9028         (*movtf_aarch64): Use neon_<ls>dm_2 as type where v8type
9029         is fpsimd_<load/store>2.
9030         (load_pair<mode>): Likewise.
9031         (store_pair<mode>): Likewise.
9033 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9035         * config/arm/types.md (type): Add "mrs" type.
9036         * config/aarch64/aarch64.md
9037         (aarch64_load_tp_hard): Make type "mrs".
9038         * config/arm/arm.md
9039         (load_tp_hard): Make type "mrs".
9040         * config/arm/cortex-a15.md: Update with new attributes.
9041         * config/arm/cortex-a5.md: Update with new attributes.
9042         * config/arm/cortex-a53.md: Update with new attributes.
9043         * config/arm/cortex-a7.md: Update with new attributes.
9044         * config/arm/cortex-a8.md: Update with new attributes.
9045         * config/arm/cortex-a9.md: Update with new attributes.
9046         * config/arm/cortex-m4.md: Update with new attributes.
9047         * config/arm/cortex-r4.md: Update with new attributes.
9048         * config/arm/fa526.md: Update with new attributes.
9049         * config/arm/fa606te.md: Update with new attributes.
9050         * config/arm/fa626te.md: Update with new attributes.
9051         * config/arm/fa726te.md: Update with new attributes.
9053 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9055         * config/aarch64/aarch64.md
9056         (*movti_aarch64): Use "multiple" for type where v8type is "move2".
9057         (*movtf_aarch64): Likewise.
9058         * config/arm/arm.md
9059         (thumb1_movdi_insn): Use "multiple" for type where more than one
9060         instruction is used for a move.
9061         (*arm32_movhf): Likewise.
9062         (*thumb_movdf_insn): Likewise.
9064 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9066         * config/arm/types.md (type): Rename fcpys to fmov.
9067         * config/arm/vfp.md
9068         (*arm_movsi_vfp): Rename type fcpys as fmov.
9069         (*thumb2_movsi_vfp): Likewise
9070         (*movhf_vfp_neon): Likewise
9071         (*movhf_vfp): Likewise
9072         (*movsf_vfp): Likewise
9073         (*thumb2_movsf_vfp): Likewise
9074         (*movsfcc_vfp): Likewise
9075         (*thumb2_movsfcc_vfp): Likewise
9076         * config/aarch64/aarch64-simd.md
9077         (move_lo_quad_<mode>): Replace type mov_reg with fmovs.
9078         * config/aarch64/aarch64.md
9079         (*movsi_aarch64): Replace type mov_reg with fmovs.
9080         (*movdi_aarch64): Likewise
9081         (*movsf_aarch64): Likewise
9082         (*movdf_aarch64): Likewise
9083         * config/arm/arm.c
9084         (cortexa7_older_only): Rename TYPE_FCPYS to TYPE_FMOV.
9085         * config/arm/iwmmxt.md
9086         (*iwmmxt_movsi_insn): Rename type fcpys as fmov.
9087         * config/arm/arm1020e.md: Update with new attributes.
9088         * config/arm/cortex-a15-neon.md: Update with new attributes.
9089         * config/arm/cortex-a5.md: Update with new attributes.
9090         * config/arm/cortex-a53.md: Update with new attributes.
9091         * config/arm/cortex-a7.md: Update with new attributes.
9092         * config/arm/cortex-a8-neon.md: Update with new attributes.
9093         * config/arm/cortex-a9.md: Update with new attributes.
9094         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9095         * config/arm/cortex-r4f.md: Update with new attributes.
9096         * config/arm/marvell-pj4.md: Update with new attributes.
9097         * config/arm/vfp11.md: Update with new attributes.
9099 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9101         * config/aarch64/aarch64.md
9102         (*madd<mode>): Fix type attribute.
9103         (*maddsi_uxtw): Likewise.
9104         (*msub<mode>): Likewise.
9105         (*msubsi_uxtw): Likewise.
9106         (<su_optab>maddsidi4): Likewise.
9107         (<su_optab>msubsidi4): Likewise.
9109 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9111         * config/arm/types.md: Split fdiv<sd> as fsqrt<sd>, fdiv<sd>.
9112         * config/arm/arm.md (core_cycles): Remove fdiv.
9113         * config/arm/vfp.md:
9114         (*sqrtsf2_vfp): Update for attribute changes.
9115         (*sqrtdf2_vfp): Likewise.
9116         * config/aarch64/aarch64.md:
9117         (sqrt<mode>2): Update for attribute changes.
9118         * config/arm/arm1020e.md: Update with new attributes.
9119         * config/arm/cortex-a15-neon.md: Update with new attributes.
9120         * config/arm/cortex-a5.md: Update with new attributes.
9121         * config/arm/cortex-a53.md: Update with new attributes.
9122         * config/arm/cortex-a7.md: Update with new attributes.
9123         * config/arm/cortex-a8-neon.md: Update with new attributes.
9124         * config/arm/cortex-a9.md: Update with new attributes.
9125         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9126         * config/arm/cortex-r4f.md: Update with new attributes.
9127         * config/arm/marvell-pj4.md: Update with new attributes.
9128         * config/arm/vfp11.md: Update with new attributes.
9130 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9132         * config/arm/types.md
9133         (type): Split f_cvt as f_cvt, f_cvtf2i, f_cvti2f.
9134         * config/aarch64/aarch64.md
9135         (l<fcvt_pattern><su_optab><GPF:mode><GPI:mode>2): Update with
9136         new attributes.
9137         (fix_trunc<GPF:mode><GPI:mode>2): Likewise.
9138         (fixuns_trunc<GPF:mode><GPI:mode>2): Likewise.
9139         (float<GPI:mode><GPF:mode>2): Likewise.
9140         * config/arm/vfp.md
9141         (*truncsisf2_vfp): Update with new attributes.
9142         (*truncsidf2_vfp): Likewise.
9143         (fixuns_truncsfsi2): Likewise.
9144         (fixuns_truncdfsi2): Likewise.
9145         (*floatsisf2_vfp): Likewise.
9146         (*floatsidf2_vfp): Likewise.
9147         (floatunssisf2): Likewise.
9148         (floatunssidf2): Likewise.
9149         (*combine_vcvt_f32_<FCVTI32typename>): Likewise.
9150         (*combine_vcvt_f64_<FCVTI32typename>): Likewise.
9151         * config/arm/arm1020e.md: Update with new attributes.
9152         * config/arm/cortex-a15-neon.md: Update with new attributes.
9153         * config/arm/cortex-a5.md: Update with new attributes.
9154         * config/arm/cortex-a53.md: Update with new attributes.
9155         * config/arm/cortex-a7.md: Update with new attributes.
9156         * config/arm/cortex-a8-neon.md: Update with new attributes.
9157         * config/arm/cortex-a9.md: Update with new attributes.
9158         * config/arm/cortex-m4-fpu.md: Update with new attributes.
9159         * config/arm/cortex-r4f.md: Update with new attributes.
9160         * config/arm/marvell-pj4.md: Update with new attributes.
9161         * config/arm/vfp11.md: Update with new attributes.
9163 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9165         * config/aarch64/arm_neon.h
9166         (vqtbl<1,2,3,4><q>_s8): Fix control vector parameter type.
9167         (vqtbx<1,2,3,4><q>_s8): Likewise.
9169 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9171         * config/arm/types.md: Add "no_insn", "multiple" and "untyped" types.
9172         * config/arm/arm-fixed.md: Add type attribute to all insn patterns.
9173         (add<mode>3): Add type attribute.
9174         (add<mode>3): Likewise.
9175         (usadd<mode>3): Likewise.
9176         (ssadd<mode>3): Likewise.
9177         (sub<mode>3): Likewise.
9178         (sub<mode>3): Likewise.
9179         (ussub<mode>3): Likewise.
9180         (sssub<mode>3): Likewise.
9181         (ssmulsa3): Likewise.
9182         (usmulusa3): Likewise.
9183         (arm_usatsihi): Likewise.
9184         * config/arm/vfp.md
9185         (*movdi_vfp): Add types for all instructions.
9186         (*movdi_vfp_cortexa8): Likewise.
9187         (*movhf_vfp_neon): Likewise.
9188         (*movhf_vfp): Likewise.
9189         (*movdf_vfp): Likewise.
9190         (*thumb2_movdf_vfp): Likewise.
9191         (*thumb2_movdfcc_vfp): Likewise.
9192         * config/arm/arm.md: Add type attribute to all insn patterns.
9193         (*thumb1_adddi3): Add type attribute.
9194         (*arm_adddi3): Likewise.
9195         (*adddi_sesidi_di): Likewise.
9196         (*adddi_zesidi_di): Likewise.
9197         (*thumb1_addsi3): Likewise.
9198         (addsi3_compare0): Likewise.
9199         (*addsi3_compare0_scratch): Likewise.
9200         (*compare_negsi_si): Likewise.
9201         (cmpsi2_addneg): Likewise.
9202         (*addsi3_carryin_<optab>): Likewise.
9203         (*addsi3_carryin_alt2_<optab>): Likewise.
9204         (*addsi3_carryin_clobercc_<optab>): Likewise.
9205         (*subsi3_carryin): Likewise.
9206         (*subsi3_carryin_const): Likewise.
9207         (*subsi3_carryin_compare): Likewise.
9208         (*subsi3_carryin_compare_const): Likewise.
9209         (*arm_subdi3): Likewise.
9210         (*thumb_subdi3): Likewise.
9211         (*subdi_di_zesidi): Likewise.
9212         (*subdi_di_sesidi): Likewise.
9213         (*subdi_zesidi_di): Likewise.
9214         (*subdi_sesidi_di): Likewise.
9215         (*subdi_zesidi_ze): Likewise.
9216         (thumb1_subsi3_insn): Likewise.
9217         (*arm_subsi3_insn): Likewise.
9218         (*anddi3_insn): Likewise.
9219         (*anddi_zesidi_di): Likewise.
9220         (*anddi_sesdi_di): Likewise.
9221         (*ne_zeroextracts): Likewise.
9222         (*ne_zeroextracts): Likewise.
9223         (*ite_ne_zeroextr): Likewise.
9224         (*ite_ne_zeroextr): Likewise.
9225         (*anddi_notdi_di): Likewise.
9226         (*anddi_notzesidi): Likewise.
9227         (*anddi_notsesidi): Likewise.
9228         (andsi_notsi_si): Likewise.
9229         (thumb1_bicsi3): Likewise.
9230         (*iordi3_insn): Likewise.
9231         (*iordi_zesidi_di): Likewise.
9232         (*iordi_sesidi_di): Likewise.
9233         (*thumb1_iorsi3_insn): Likewise.
9234         (*xordi3_insn): Likewise.
9235         (*xordi_zesidi_di): Likewise.
9236         (*xordi_sesidi_di): Likewise.
9237         (*arm_xorsi3): Likewise.
9238         (*andsi_iorsi3_no): Likewise.
9239         (*smax_0): Likewise.
9240         (*smax_m1): Likewise.
9241         (*arm_smax_insn): Likewise.
9242         (*smin_0): Likewise.
9243         (*arm_smin_insn): Likewise.
9244         (*arm_umaxsi3): Likewise.
9245         (*arm_uminsi3): Likewise.
9246         (*minmax_arithsi): Likewise.
9247         (*minmax_arithsi_): Likewise.
9248         (*satsi_<SAT:code>): Likewise.
9249         (arm_ashldi3_1bit): Likewise.
9250         (arm_ashrdi3_1bit): Likewise.
9251         (arm_lshrdi3_1bit): Likewise.
9252         (*arm_negdi2): Likewise.
9253         (*thumb1_negdi2): Likewise.
9254         (*arm_negsi2): Likewise.
9255         (*thumb1_negsi2): Likewise.
9256         (*negdi_extendsid): Likewise.
9257         (*negdi_zero_extend): Likewise.
9258         (*arm_abssi2): Likewise.
9259         (*thumb1_abssi2): Likewise.
9260         (*arm_neg_abssi2): Likewise.
9261         (*thumb1_neg_abss): Likewise.
9262         (one_cmpldi2): Likewise.
9263         (extend<mode>di2): Likewise.
9264         (*compareqi_eq0): Likewise.
9265         (*arm_extendhisi2addsi): Likewise.
9266         (*arm_movdi): Likewise.
9267         (*thumb1_movdi_insn): Likewise.
9268         (*arm_movt): Likewise.
9269         (*thumb1_movsi_insn): Likewise.
9270         (pic_add_dot_plus_four): Likewise.
9271         (pic_add_dot_plus_eight): Likewise.
9272         (tls_load_dot_plus_eight): Likewise.
9273         (*thumb1_movhi_insn): Likewise.
9274         (*thumb1_movsf_insn): Likewise.
9275         (*movdf_soft_insn): Likewise.
9276         (*thumb_movdf_insn): Likewise.
9277         (cbranchsi4_insn): Likewise.
9278         (cbranchsi4_scratch): Likewise.
9279         (*negated_cbranchsi4): Likewise.
9280         (*tbit_cbranch): Likewise.
9281         (*tlobits_cbranch): Likewise.
9282         (*tstsi3_cbranch): Likewise.
9283         (*cbranchne_decr1): Likewise.
9284         (*addsi3_cbranch): Likewise.
9285         (*addsi3_cbranch_scratch): Likewise.
9286         (*arm_cmpdi_insn): Likewise.
9287         (*arm_cmpdi_unsig): Likewise.
9288         (*arm_cmpdi_zero): Likewise.
9289         (*thumb_cmpdi_zero): Likewise.
9290         (*deleted_compare): Likewise.
9291         (*mov_scc): Likewise.
9292         (*mov_negscc): Likewise.
9293         (*mov_notscc): Likewise.
9294         (*cstoresi_eq0_thumb1_insn): Likewise.
9295         (cstoresi_nltu_thumb1): Likewise.
9296         (cstoresi_ltu_thu): Likewise.
9297         (thumb1_addsi3_addgeu): Likewise.
9298         (*arm_jump): Likewise.
9299         (*thumb_jump): Likewise.
9300         (*check_arch2): Likewise.
9301         (arm_casesi_internal): Likewise.
9302         (thumb1_casesi_dispatch): Likewise.
9303         (*arm_indirect_jump): Likewise.
9304         (*thumb1_indirect_jump): Likewise.
9305         (nop): Likewise.
9306         (*and_scc): Likewise.
9307         (*ior_scc): Likewise.
9308         (*compare_scc): Likewise.
9309         (*cond_move): Likewise.
9310         (*cond_arith): Likewise.
9311         (*cond_sub): Likewise.
9312         (*cmp_ite0): Likewise.
9313         (*cmp_ite1): Likewise.
9314         (*cmp_and): Likewise.
9315         (*cmp_ior): Likewise.
9316         (*ior_scc_scc): Likewise.
9317         (*ior_scc_scc_cmp): Likewise.
9318         (*and_scc_scc): Likewise.
9319         (*and_scc_scc_cmp): Likewise.
9320         (*and_scc_scc_nod): Likewise.
9321         (*negscc): Likewise.
9322         (movcond_addsi): Likewise.
9323         (movcond): Likewise.
9324         (*ifcompare_plus_move): Likewise.
9325         (*if_plus_move): Likewise.
9326         (*ifcompare_move_plus): Likewise.
9327         (*if_move_plus): Likewise.
9328         (*ifcompare_arith_arith): Likewise.
9329         (*if_arith_arith): Likewise.
9330         (*ifcompare_arith_move): Likewise.
9331         (*if_arith_move): Likewise.
9332         (*ifcompare_move_arith): Likewise.
9333         (*if_move_arith): Likewise.
9334         (*ifcompare_move_not): Likewise.
9335         (*if_move_not): Likewise.
9336         (*ifcompare_not_move): Likewise.
9337         (*if_not_move): Likewise.
9338         (*ifcompare_shift_move): Likewise.
9339         (*if_shift_move): Likewise.
9340         (*ifcompare_move_shift): Likewise.
9341         (*if_move_shift): Likewise.
9342         (*ifcompare_shift_shift): Likewise.
9343         (*ifcompare_not_arith): Likewise.
9344         (*ifcompare_arith_not): Likewise.
9345         (*if_arith_not): Likewise.
9346         (*ifcompare_neg_move): Likewise.
9347         (*if_neg_move): Likewise.
9348         (*ifcompare_move_neg): Likewise.
9349         (*if_move_neg): Likewise.
9350         (prologue_thumb1_interwork): Likewise.
9351         (*cond_move_not): Likewise.
9352         (*sign_extract_onebit): Likewise.
9353         (*not_signextract_onebit): Likewise.
9354         (stack_tie): Likewise.
9355         (align_4): Likewise.
9356         (align_8): Likewise.
9357         (consttable_end): Likewise.
9358         (consttable_1): Likewise.
9359         (consttable_2): Likewise.
9360         (consttable_4): Likewise.
9361         (consttable_8): Likewise.
9362         (consttable_16): Likewise.
9363         (*thumb1_tablejump): Likewise.
9364         (prefetch): Likewise.
9365         (force_register_use): Likewise.
9366         (thumb_eh_return): Likewise.
9367         (load_tp_hard): Likewise.
9368         (load_tp_soft): Likewise.
9369         (tlscall): Likewise.
9370         (*arm_movtas_ze): Likewise.
9371         (*arm_rev): Likewise.
9372         (*arm_revsh): Likewise.
9373         (*arm_rev16): Likewise.
9374         * config/arm/thumb2.md
9375         (*thumb2_smaxsi3): Likewise.
9376         (*thumb2_sminsi3): Likewise.
9377         (*thumb32_umaxsi3): Likewise.
9378         (*thumb2_uminsi3): Likewise.
9379         (*thumb2_negdi2): Likewise.
9380         (*thumb2_abssi2): Likewise.
9381         (*thumb2_neg_abss): Likewise.
9382         (*thumb2_movsi_insn): Likewise.
9383         (tls_load_dot_plus_four): Likewise.
9384         (*thumb2_movhi_insn): Likewise.
9385         (*thumb2_mov_scc): Likewise.
9386         (*thumb2_mov_negs): Likewise.
9387         (*thumb2_mov_negs): Likewise.
9388         (*thumb2_mov_nots): Likewise.
9389         (*thumb2_mov_nots): Likewise.
9390         (*thumb2_movsicc_): Likewise.
9391         (*thumb2_movsfcc_soft_insn): Likewise.
9392         (*thumb2_indirect_jump): Likewise.
9393         (*thumb2_and_scc): Likewise.
9394         (*thumb2_ior_scc): Likewise.
9395         (*thumb2_ior_scc_strict_it): Likewise.
9396         (*thumb2_cond_move): Likewise.
9397         (*thumb2_cond_arith): Likewise.
9398         (*thumb2_cond_ari): Likewise.
9399         (*thumb2_cond_sub): Likewise.
9400         (*thumb2_negscc): Likewise.
9401         (*thumb2_movcond): Likewise.
9402         (thumb2_casesi_internal): Likewise.
9403         (thumb2_casesi_internal_pic): Likewise.
9404         (*thumb2_alusi3_short): Likewise.
9405         (*thumb2_mov<mode>_shortim): Likewise.
9406         (*thumb2_addsi_short): Likewise.
9407         (*thumb2_subsi_short): Likewise.
9408         (thumb2_addsi3_compare0): Likewise.
9409         (*thumb2_cbz): Likewise.
9410         (*thumb2_cbnz): Likewise.
9411         (*thumb2_one_cmplsi2_short): Likewise.
9412         (*thumb2_negsi2_short): Likewise.
9413         (*orsi_notsi_si): Likewise.
9414         * config/arm/arm1020e.md: Update with new attributes.
9415         * config/arm/arm1026ejs.md: Update with new attributes.
9416         * config/arm/arm1136jfs.md: Update with new attributes.
9417         * config/arm/arm926ejs.md: Update with new attributes.
9418         * config/arm/cortex-a15.md: Update with new attributes.
9419         * config/arm/cortex-a5.md: Update with new attributes.
9420         * config/arm/cortex-a53.md: Update with new attributes.
9421         * config/arm/cortex-a7.md: Update with new attributes.
9422         * config/arm/cortex-a8.md: Update with new attributes.
9423         * config/arm/cortex-a9.md: Update with new attributes.
9424         * config/arm/cortex-m4.md: Update with new attributes.
9425         * config/arm/cortex-r4.md: Update with new attributes.
9426         * config/arm/fa526.md: Update with new attributes.
9427         * config/arm/fa606te.md: Update with new attributes.
9428         * config/arm/fa626te.md: Update with new attributes.
9429         * config/arm/fa726te.md: Update with new attributes.
9431 2013-09-06  James Greenhalgh  <james.greenhalgh@arm.com>
9433         * config/aarch64/aarch64-simd.md
9434         (aarch64_sqdml<SBINQOPS:as>l_n<mode>_internal): Use
9435         <vwx> iterator to ensure correct register choice.
9436         (aarch64_sqdml<SBINQOPS:as>l2_n<mode>_internal): Likewise.
9437         (aarch64_sqdmull_n<mode>): Likewise.
9438         (aarch64_sqdmull2_n<mode>_internal): Likewise.
9439         * config/aarch64/arm_neon.h
9440         (vml<as><q>_lane<q>_<su>16): Use 'x' constraint for element vector.
9441         (vml<as><q>_n_<su>16): Likewise.
9442         (vml<as>l_high_lane<q>_<su>16): Likewise.
9443         (vml<as>l_high_n_<su>16): Likewise.
9444         (vml<as>l_lane<q>_<su>16): Likewise.
9445         (vml<as>l_n_<su>16): Likewise.
9446         (vmul<q>_lane<q>_<su>16): Likewise.
9447         (vmul<q>_n_<su>16): Likewise.
9448         (vmull_lane<q>_<su>16): Likewise.
9449         (vmull_n_<su>16): Likewise.
9450         (vmull_high_lane<q>_<su>16): Likewise.
9451         (vmull_high_n_<su>16): Likewise.
9452         (vqrdmulh<q>_n_s16): Likewise.
9454 2013-09-06  Tejas Belagod  <tejas.belagod@arm.com>
9456         * config/aarch64/arm_neon.h: Fix all vdup<bhsd_lane<q> intrinsics to
9457         have the correct lane parameter.
9459 2013-09-06  Richard Biener <rguenther@suse.de>
9461         * cfganal.c (control_dependences::~control_dependences):
9462         Properly free all of the vector.
9464 2013-09-06  Kirill Yukhin  <kirill.yukhin@intel.com>
9466         PR target/58269
9467         * config/i386/i386.c (ix86_conditional_register_usage):
9468         Proper initialize extended SSE registers.
9470 2013-09-06  Jan Hubicka  <jh@suse.cz>
9472         PR tree-optimization/58311
9473         * ipa-devirt.c (gate_ipa_devirt): Only execute when optimizing.
9475 2013-09-06  Jan Hubicka  <jh@suse.cz>
9477         * Makefile.in (tree-sra.o): Update dependencies.
9478         * tree-sra.c: Include ipa-utils.h
9479         (scan_function): Use recursive_call_p.
9480         (has_caller_p): New function.
9481         (cgraph_for_node_and_aliases): Count also callers of aliases.
9483 2013-09-06  Jan Hubicka  <jh@suse.cz>
9485         PR middle-end/58094
9486         * cgraph.h (symtab_semantically_equivalent_p): Declare.
9487         * tree-tailcall.c: Include ipa-utils.h.
9488         (find_tail_calls): Use it.
9489         * ipa-pure-const.c (check_call): Likewise.
9490         * ipa-utils.c (recursive_call_p): New function.
9491         * ipa-utils.h (recursive_call_p): Dclare.
9492         * symtab.c (symtab_nonoverwritable_alias): Fix formatting.
9493         (symtab_semantically_equivalent_p): New function.
9494         * Makefile.in (tree-tailcall.o): Update dependencies.
9496 2013-09-06  Eric Botcazou  <ebotcazou@adacore.com>
9498         * ipa-split.c (split_function): Set DECL_NO_INLINE_WARNING_P on the
9499         non-inlinable part.
9501 2013-09-06  Richard Biener  <rguenther@suse.de>
9503         * lto-streamer.h (lto_global_var_decls): Remove.
9504         * Makefile.in (OBJS): Remove lto-symtab.o.
9505         (lto-symtab.o): Remove.
9506         (GTFILES): Remove lto-symtab.c
9507         * lto-symtab.c: Move to lto/
9509 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9511         * config/s390/s390.md (UNSPEC_FPINT_FLOOR, UNSPEC_FPINT_BTRUNC)
9512         (UNSPEC_FPINT_ROUND, UNSPEC_FPINT_CEIL, UNSPEC_FPINT_NEARBYINT)
9513         (UNSPEC_FPINT_RINT): New constant definitions.
9514         (FPINT, fpint_name, fpint_roundingmode): New integer iterator
9515         definition with 2 attributes.
9516         ("<FPINT:fpint_name><BFP:mode>2", "rint<BFP:mode>2")
9517         ("<FPINT:fpint_name><DFP:mode>2", "rint<DFP:mode>2"): New pattern
9518         definitions.
9520 2013-09-06  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
9522         * config/s390/s390.md: Add "bcr_flush" value to mnemonic attribute.
9523         ("mem_thread_fence_1"): Use bcr 14,0 for z196 and later.
9524         Set the mnemonic attribute to "bcr_flush".  Set the "z196prop"
9525         attribute to "z196_alone".
9526         * config/s390/2827.md: Add "bcr_flush" to "ooo_groupalone" and
9527         "zEC12_simple".
9529 2013-09-06  Richard Biener  <rguenther@suse.de>
9531         * basic-block.h (class control_dependences): New.
9532         * tree-ssa-dce.c (control_dependence_map): Remove.
9533         (cd): New global.
9534         (EXECUTE_IF_CONTROL_DEPENDENT): Remove.
9535         (set_control_dependence_map_bit, clear_control_dependence_bitmap,
9536         find_pdom, find_control_dependence, find_all_control_dependences):
9537         Move to cfganal.c.
9538         (mark_control_dependent_edges_necessary,
9539         find_obviously_necessary_stmts, propagate_necessity, tree_dce_init,
9540         tree_dce_done, perform_tree_ssa_dce): Adjust.
9541         * cfganal.c (set_control_dependence_map_bit,
9542         clear_control_dependence_bitmap, find_pdom, find_control_dependence,
9543         find_all_control_dependences): Move from tree-ssa-dce.c and
9544         implement as methods of control_dependences class.
9545         (control_dependences::control_dependences): New.
9546         (control_dependences::~control_dependences): Likewise.
9547         (control_dependences::get_edges_dependent_on): Likewise.
9548         (control_dependences::get_edge): Likewise.
9550 2013-09-04  Jan Hubicka  <jh@suse.cz>
9552         * tree.c (types_same_for_odr): Drop overactive check.
9553         * ipa-devirt.c (hash_type_name): Likewise.
9555 2013-09-04  Jan Hubicka  <jh@suse.cz>
9557         * cgraphunit.c (walk_polymorphic_call_targets): Break out from ...
9558         (analyze_functions): ... here.
9560 2013-09-04  Jan Hubicka  <jh@suse.cz>
9562         PR middle-end/58201
9563         * cgraphunit.c (analyze_functions): Clear AUX fields
9564         after processing; initialize assembler name has.
9566 2013-09-05  Jeff Law  <law@redhat.com>
9568         * tree-ssa-threadedge.c (thread_around_empty_blocks): Renamed
9569         from thread_around_empty_block.  Record threading path into PATH.
9570         Recurse if threading through the initial block is successful.
9571         (thread_across_edge): Corresponding changes to slightly simplify.
9573 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9575         * config/aarch64/aarch64.md
9576         (type): Remove frecpe, frecps, frecpx.
9577         (aarch64_frecp<FRECP:frecp_suffix><mode>): Move to aarch64-simd.md,
9578         fix to be a TARGET_SIMD instruction.
9579         (aarch64_frecps): Remove.
9580         * config/aarch64/aarch64-simd.md
9581         (aarch64_frecp<FRECP:frecp_suffix><mode>): New, moved from aarch64.md
9582         (aarch64_frecps<mode>): Handle all float/vector of float modes.
9584 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9585             Sofiane Naci  <sofiane.naci@arm.com>
9587         * config/arm/types.md (define_attr "type"): Expand "arlo_imm"
9588         into "adr", "alu_imm", "alus_imm", "logic_imm", "logics_imm".
9589         Expand "arlo_reg" into "adc_reg", "adc_imm", "adcs_reg", "adcs_imm",
9590         "alu_ext", "alu_reg", "alus_ext", "alus_reg", "bfm", "csel",
9591         "logic_reg", "logics_reg", "rev".  Expand "arlo_shift" into
9592         "alu_shift_imm", "alus_shift_imm", "logic_shift_imm",
9593         "logics_shift_imm".  Expand "arlo_shift_reg" into "alu_shift_reg",
9594         "alus_shift_reg", "logic_shift_reg", "logics_shift_reg".  Expand "clz"
9595         into "clz, "rbit".  Rename "shift" to "shift_imm".
9596         * config/arm/arm.md (define_attr "core_cycles"): Update for attribute
9597         changes.  Update for attribute changes all occurrences of arlo_* and
9598         shift* types.
9599         * config/arm/arm-fixed.md: Update for attribute changes
9600         all occurrences of arlo_* types.
9601         * config/arm/thumb2.md: Update for attribute changes all occurrences
9602         of arlo_* types.
9603         * config/arm/arm.c (xscale_sched_adjust_cost):  (rtx insn, rtx
9604         (cortexa7_older_only): Likewise.
9605         (cortexa7_younger):  Likewise.
9606         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
9607         (1020alu_shift_op): Likewise.
9608         (1020alu_shift_reg_op): Likewise.
9609         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
9610         (alu_shift_op): Likewise.
9611         (alu_shift_reg_op): Likewise.
9612         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
9613         (11_alu_shift_op): Likewise.
9614         (11_alu_shift_reg_op): Likewise.
9615         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
9616         (9_alu_shift_reg_op): Likewise.
9617         * config/arm/cortex-a15.md (cortex_a15_alu): Update for
9618         attribute changes.
9619         (cortex_a15_alu_shift): Likewise.
9620         (cortex_a15_alu_shift_reg): Likewise.
9621         * config/arm/cortex-a5.md (cortex_a5_alu): Update for
9622         attribute changes.
9623         (cortex_a5_alu_shift): Likewise.
9624         * config/arm/cortex-a53.md (cortex_a53_alu): Update for
9625         attribute changes.
9626         (cortex_a53_alu_shift): Likewise.
9627         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for
9628         attribute changes.
9629         (cortex_a7_alu_reg): Likewise.
9630         (cortex_a7_alu_shift): Likewise.
9631         * config/arm/cortex-a8.md (cortex_a8_alu): Update for
9632         attribute changes.
9633         (cortex_a8_alu_shift): Likewise.
9634         (cortex_a8_alu_shift_reg): Likewise.
9635         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
9636         (cortex_a9_dp_shift): Likewise.
9637         * config/arm/cortex-m4.md (cortex_m4_alu): Update for
9638         attribute changes.
9639         * config/arm/cortex-r4.md
9640         (cortex_r4_alu): Update for attribute changes.
9641         (cortex_r4_mov): Likewise.
9642         (cortex_r4_alu_shift_reg): Likewise.
9643         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
9644         (526_alu_shift_op): Likewise.
9645         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
9646         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
9647         (626te_alu_shift_op): Likewise.
9648         * config/arm/fa726te.md (726te_alu_op): Update for attribute changes.
9649         (726te_alu_shift_op): Likewise.
9650         (726te_alu_shift_reg_op): Likewise.
9651         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
9652         (mp626_alu_shift_op): Likewise.
9653         * config/arm/marvell-pj4.md (pj4_alu): Update for attribute changes.
9654         (pj4_alu_conds): Likewise.
9655         (pj4_shift): Likewise.
9656         (pj4_shift_conds): Likewise.
9657         (pj4_alu_shift): Likewise.
9658         (pj4_alu_shift_conds): Likewise.
9659         * config/aarch64/aarch64.md: Update for attribute change
9660         all occurrences of arlo_* and shift* types.
9662 2013-09-05  Mike Stump  <mikestump@comcast.net>
9664         * tree.h: Move documentation for tree_function_decl to tree-core.h
9665         with the declaration.
9667 2013-09-05  Peter Bergner  <bergner@vnet.ibm.com>
9669         PR target/58139
9670         * reginfo.c (choose_hard_reg_mode): Scan through all mode classes
9671         looking for widest mode.
9673 2013-09-05  Eric Botcazou  <ebotcazou@adacore.com>
9675         * config.gcc (*-*-vxworks*): Do not override an existing extra_objs.
9677 2013-09-05  Richard Biener  <rguenther@suse.de>
9679         PR tree-optimization/58137
9680         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size):
9681         Do not create vectors of pointers.
9682         * tree-vect-loop.c (get_initial_def_for_induction): Use proper
9683         types for the components of the vector initializer.
9684         * tree-cfg.c (verify_gimple_assign_binary): Remove special-casing
9685         allowing pointer vectors with PLUS_EXPR/MINUS_EXPR.
9687 2013-09-05  Martin Jambor  <mjambor@suse.cz>
9689         * ipa-prop.c (remove_described_reference): Accept missing references,
9690         return false if that hppens, otherwise return true.
9691         (cgraph_node_for_jfunc): New function.
9692         (try_decrement_rdesc_refcount): Likewise.
9693         (try_make_edge_direct_simple_call): Use them.
9694         (ipa_edge_removal_hook): Remove references from rdescs.
9695         (ipa_edge_duplication_hook): Clone rdescs and their references
9696         when the new edge has the same caller as the old one.
9697         * cgraph.c (cgraph_resolve_speculation): Remove speculative
9698         reference before removing any edges.
9700 2013-09-05  Richard Earnshaw  <rearnsha@arm.com>
9702         * arm.c (thumb2_emit_strd_push): Rewrite to use pre-decrement on
9703         initial store.
9704         * thumb2.md (thumb2_storewb_parisi): New pattern.
9706 2013-09-05  Yufeng Zhang  <yufeng.zhang@arm.com>
9708         * config/aarch64/aarch64-option-extensions.def: Add
9709         AARCH64_OPT_EXTENSION of 'crc'.
9710         * config/aarch64/aarch64.h (AARCH64_FL_CRC): New define.
9711         (AARCH64_ISA_CRC): Ditto.
9712         * doc/invoke.texi (-march and -mcpu feature modifiers): Add
9713         description of the CRC extension.
9715 2013-09-05  Alexander Ivchenko  <alexander.ivchenko@intel.com>
9717         * config/rs6000/linux64.h: Define OPTION_BIONIC and OPTION_UCLIBC.
9718         * config/rs6000/linux.h: Ditto.
9719         * alpha/linux.h: Ditto.
9720         * config/bfin/uclinux.h: Define TARGET_LIBC_HAS_FUNCTION as
9721         no_c99_libc_has_function.
9722         * config/c6x/uclinux-elf.h: Ditto.
9723         * config/lm32/uclinux-elf.h: Ditto.
9724         * config/m68k/uclinux.h: Ditto.
9725         * config/moxie/uclinux.h: Ditto.
9726         * config.gcc (bfin*-linux-uclibc*): Add t-linux-android to tmake_file.
9727         (crisv32-*-linux*, cris-*-linux*): Ditto.
9728         * config/bfin/bfin.c: Include "tm_p.h".
9730 2013-09-05  Richard Biener  <rguenther@suse.de>
9732         * tree-vect-loop.c (vect_analyze_loop_operations): Properly
9733         check for a definition without a basic-block.
9735 2013-09-05  James Greenhalgh  <james.greenhalgh@arm.com>
9736             Sofiane Naci <sofiane.naci@arm.com>
9738         * config/aarch64/aarch64.md
9739         (*movti_aarch64): Rename r_2_f and f_2_r.
9740         (*movsf_aarch64): Likewise.
9741         (*movdf_aarch64): Likewise.
9742         (*movtf_aarch64): Likewise.
9743         (aarch64_movdi_<mode>low): Likewise.
9744         (aarch64_movdi_<mode>high): Likewise.
9745         (aarch64_mov<mode>high_di): Likewise.
9746         (aarch64_mov<mode>low_di): Likewise.
9747         (aarch64_movtilow_tilow): Likewise.
9748         * config/arm/arm.md (attribute "neon_type"): Delete.  Move attribute
9749         values to config/arm/types.md
9750         (attribute "conds"): Update for attribute change.
9751         (anddi3_insn): Likewise.
9752         (iordi3_insn): Likewise.
9753         (xordi3_insn): Likewise.
9754         (one_cmpldi2): Likewise.
9755         * config/arm/types.md (type): Add Neon types.
9756         * config/arm/neon.md (neon_mov<mode>): Remove "neon_type" attribute,
9757         use "type" attribute.
9758         (movmisalign<mode>_neon_store): Likewise.
9759         (movmisalign<mode>_neon_load): Likewise.
9760         (vec_set<mode>_internal): Likewise.
9761         (vec_setv2di_internal): Likewise.
9762         (vec_extract<mode>): Likewise.
9763         (vec_extractv2di): Likewise.
9764         (add<mode>3_neon): Likewise.
9765         (adddi3_neon): Likewise.
9766         (sub<mode>3_neon): Likewise.
9767         (subdi3_neon): Likewise.
9768         (mul<mode>3_neon): Likewise.
9769         (mul<mode>3add<mode>_neon): Likewise.
9770         (mul<mode>3neg<mode>add<mode>_neon): Likewise.
9771         (fma<VCVTF:mode>4)): Likewise.
9772         (fma<VCVTF:mode>4_intrinsic): Likewise.
9773         (fmsub<VCVTF:mode>4)): Likewise.
9774         (fmsub<VCVTF:mode>4_intrinsic): Likewise.
9775         (neon_vrint<NEON_VRINT:nvrint_variant><VCVTF:mode>): Likewise.
9776         (ior<mode>3): Likewise.
9777         (and<mode>3): Likewise.
9778         (anddi3_neon): Likewise.
9779         (orn<mode>3_neon): Likewise.
9780         (orndi3_neon): Likewise.
9781         (bic<mode>3_neon): Likewise.
9782         (bicdi3_neon): Likewise.
9783         (xor<mode>3): Likewise.
9784         (one_cmpl<mode>2): Likewise.
9785         (abs<mode>2): Likewise.
9786         (neg<mode>2): Likewise.
9787         (umin<mode>3_neon): Likewise.
9788         (umax<mode>3_neon): Likewise.
9789         (smin<mode>3_neon): Likewise.
9790         (smax<mode>3_neon): Likewise.
9791         (vashl<mode>3): Likewise.
9792         (vashr<mode>3_imm): Likewise.
9793         (vlshr<mode>3_imm): Likewise.
9794         (ashl<mode>3_signed): Likewise.
9795         (ashl<mode>3_unsigned): Likewise.
9796         (neon_load_count): Likewise.
9797         (ashldi3_neon_noclobber): Likewise.
9798         (signed_shift_di3_neon): Likewise.
9799         (unsigned_shift_di3_neon): Likewise.
9800         (ashrdi3_neon_imm_noclobber): Likewise.
9801         (lshrdi3_neon_imm_noclobber): Likewise.
9802         (widen_ssum<mode>3): Likewise.
9803         (widen_usum<mode>3): Likewise.
9804         (quad_halves_<code>v4si): Likewise.
9805         (quad_halves_<code>v4sf): Likewise.
9806         (quad_halves_<code>v8hi): Likewise.
9807         (quad_halves_<code>v16qi): Likewise.
9808         (reduc_splus_v2di): Likewise.
9809         (neon_vpadd_internal<mode>): Likewise.
9810         (neon_vpsmin<mode>): Likewise.
9811         (neon_vpsmax<mode>): Likewise.
9812         (neon_vpumin<mode>): Likewise.
9813         (neon_vpumax<mode>): Likewise.
9814         (ss_add<mode>_neon): Likewise.
9815         (us_add<mode>_neon): Likewise.
9816         (ss_sub<mode>_neon): Likewise.
9817         (us_sub<mode>_neon): Likewise.
9818         (neon_vadd<mode>_unspec): Likewise.
9819         (neon_vaddl<mode>): Likewise.
9820         (neon_vaddw<mode>): Likewise.
9821         (neon_vhadd<mode>): Likewise.
9822         (neon_vqadd<mode>): Likewise.
9823         (neon_vaddhn<mode>): Likewise.
9824         (neon_vmul<mode>): Likewise.
9825         (neon_vmla<mode>): Likewise.
9826         (neon_vmlal<mode>): Likewise.
9827         (neon_vmls<mode>): Likewise.
9828         (neon_vmlsl<mode>): Likewise.
9829         (neon_vqdmulh<mode>): Likewise.
9830         (neon_vqdmlal<mode>): Likewise.
9831         (neon_vqdmlsl<mode>): Likewise.
9832         (neon_vmull<mode>): Likewise.
9833         (neon_vqdmull<mode>): Likewise.
9834         (neon_vsub<mode>_unspec): Likewise.
9835         (neon_vsubl<mode>): Likewise.
9836         (neon_vsubw<mode>): Likewise.
9837         (neon_vqsub<mode>): Likewise.
9838         (neon_vhsub<mode>): Likewise.
9839         (neon_vsubhn<mode>): Likewise.
9840         (neon_vceq<mode>): Likewise.
9841         (neon_vcge<mode>): Likewise.
9842         (neon_vcgeu<mode>): Likewise.
9843         (neon_vcgt<mode>): Likewise.
9844         (neon_vcgtu<mode>): Likewise.
9845         (neon_vcle<mode>): Likewise.
9846         (neon_vclt<mode>): Likewise.
9847         (neon_vcage<mode>): Likewise.
9848         (neon_vcagt<mode>): Likewise.
9849         (neon_vtst<mode>): Likewise.
9850         (neon_vabd<mode>): Likewise.
9851         (neon_vabdl<mode>): Likewise.
9852         (neon_vaba<mode>): Likewise.
9853         (neon_vabal<mode>): Likewise.
9854         (neon_vmax<mode>): Likewise.
9855         (neon_vmin<mode>): Likewise.
9856         (neon_vpaddl<mode>): Likewise.
9857         (neon_vpadal<mode>): Likewise.
9858         (neon_vpmax<mode>): Likewise.
9859         (neon_vpmin<mode>): Likewise.
9860         (neon_vrecps<mode>): Likewise.
9861         (neon_vrsqrts<mode>): Likewise.
9862         (neon_vqabs<mode>): Likewise.
9863         (neon_vqneg<mode>): Likewise.
9864         (neon_vcls<mode>): Likewise.
9865         (clz<mode>2): Likewise.
9866         (popcount<mode>2): Likewise.
9867         (neon_vrecpe): Likewise.
9868         (neon_vrsqrte): Likewise.
9869         (neon_vget_lane<mode>_sext_internal): Likewise.
9870         (neon_vget_lane<mode>_zext_internal): Likewise.
9871         (neon_vdup_n<mode>): Likewise.
9872         (neon_vdup_nv2di): Likewise.
9873         (neon_vdpu_lane<mode>_internal): Likewise.
9874         (neon_vswp<mode>): Likewise.
9875         (float<mode><V_cvtto>2): Likewise.
9876         (floatuns<mode><V_cvtto>2): Likewise.
9877         (fix_trunc<mode><V_cvtto>)2): Likewise
9878         (fixuns_trunc<mode><V_cvtto)2): Likewise.
9879         (neon_vcvt<mode>): Likewise.
9880         (neon_vcvtv4sfv4hf): Likewise.
9881         (neon_vcvtv4hfv4sf): Likewise.
9882         (neon_vcvt_n<mode>): Likewise.
9883         (neon_vmovn<mode>): Likewise.
9884         (neon_vqmovn<mode>): Likewise.
9885         (neon_vqmovun<mode>): Likewise.
9886         (neon_vmovl<mode>): Likewise.
9887         (neon_vmul_lane<mode>): Likewise.
9888         (neon_vmull_lane<mode>): Likewise.
9889         (neon_vqdmull_lane<mode>): Likewise.
9890         (neon_vqdmulh_lane<mode>): Likewise.
9891         (neon_vmla_lane<mode>): Likewise.
9892         (neon_vmlal_lane<mode>): Likewise.
9893         (neon_vqdmlal_lane<mode>): Likewise.
9894         (neon_vmls_lane<mode>): Likewise.
9895         (neon_vmlsl_lane<mode>): Likewise.
9896         (neon_vqdmlsl_lane<mode>): Likewise.
9897         (neon_vext<mode>): Likewise.
9898         (neon_vrev64<mode>): Likewise.
9899         (neon_vrev32<mode>): Likewise.
9900         (neon_vrev16<mode>): Likewise.
9901         (neon_vbsl<mode>_internal): Likewise.
9902         (neon_vshl<mode>): Likewise.
9903         (neon_vqshl<mode>): Likewise.
9904         (neon_vshr_n<mode>): Likewise.
9905         (neon_vshrn_n<mode>): Likewise.
9906         (neon_vqshrn_n<mode>): Likewise.
9907         (neon_vqshrun_n<mode>): Likewise.
9908         (neon_vshl_n<mode>): Likewise.
9909         (neon_vqshl_n<mode>): Likewise.
9910         (neon_vqshlu_n<mode>): Likewise.
9911         (neon_vshll_n<mode>): Likewise.
9912         (neon_vsra_n<mode>): Likewise.
9913         (neon_vsri_n<mode>): Likewise.
9914         (neon_vsli_n<mode>): Likewise.
9915         (neon_vtbl1v8qi): Likewise.
9916         (neon_vtbl2v8qi): Likewise.
9917         (neon_vtbl3v8qi): Likewise.
9918         (neon_vtbl4v8qi): Likewise.
9919         (neon_vtbx1v8qi): Likewise.
9920         (neon_vtbx2v8qi): Likewise.
9921         (neon_vtbx3v8qi): Likewise.
9922         (neon_vtbx4v8qi): Likewise.
9923         (neon_vtrn<mode>_internal): Likewise.
9924         (neon_vzip<mode>_internal): Likewise.
9925         (neon_vuzp<mode>_internal): Likewise.
9926         (neon_vld1<mode>): Likewise.
9927         (neon_vld1_lane<mode>): Likewise.
9928         (neon_vld1_dup<mode>): Likewise.
9929         (neon_vld1_dupv2di): Likewise.
9930         (neon_vst1<mode>): Likewise.
9931         (neon_vst1_lane<mode>): Likewise.
9932         (neon_vld2<mode>): Likewise.
9933         (neon_vld2_lane<mode>): Likewise.
9934         (neon_vld2_dup<mode>): Likewise.
9935         (neon_vst2<mode>): Likewise.
9936         (neon_vst2_lane<mode>): Likewise.
9937         (neon_vld3<mode>): Likewise.
9938         (neon_vld3qa<mode>): Likewise.
9939         (neon_vld3qb<mode>): Likewise.
9940         (neon_vld3_lane<mode>): Likewise.
9941         (neon_vld3_dup<mode>): Likewise.
9942         (neon_vst3<mode>): Likewise.
9943         (neon_vst3qa<mode>): Likewise.
9944         (neon_vst3qb<mode>): Likewise.
9945         (neon_vst3_lane<mode>): Likewise.
9946         (neon_vld4<mode>): Likewise.
9947         (neon_vld4qa<mode>): Likewise.
9948         (neon_vld4qb<mode>): Likewise.
9949         (neon_vld4_lane<mode>): Likewise.
9950         (neon_vld4_dup<mode>): Likewise.
9951         (neon_vst4<mode>): Likewise.
9952         (neon_vst4qa<mode>): Likewise.
9953         (neon_vst4qb<mode>): Likewise.
9954         (neon_vst4_lane<mode>): Likewise.
9955         (neon_vec_unpack<US>_lo_<mode>): Likewise.
9956         (neon_vec_unpack<US>_hi_<mode>): Likewise.
9957         (neon_vec_<US>mult_lo_<mode>): Likewise.
9958         (neon_vec_<US>mult_hi_<mode>): Likewise.
9959         (neon_vec_<US>shiftl_<mode>): Likewise.
9960         (neon_unpack<US>_<mode>): Likewise.
9961         (neon_vec_<US>mult_<mode>): Likewise.
9962         (vec_pack_trunc_<mode>): Likewise.
9963         (neon_vec_pack_trunk_<mode>): Likewise.
9964         (neon_vabd<mode>_2): Likewise.
9965         (neon_vabd<mode>_3): Likewise.
9966         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
9967         (thumb2_movsi_vfp): Likewise.
9968         (movdi_vfp): Likewise.
9969         (movdi_vfp_cortexa8): Likewise.
9970         (movhf_vfp_neon): Likewise.
9971         (movhf_vfp): Likewiwse.
9972         (movsf_vfp): Likewiwse.
9973         (thumb2_movsf_vfp): Likewiwse.
9974         (movdf_vfp): Likewise.
9975         (thumb2_movdf_vfp): Likewise.
9976         (movsfcc_vfp): Likewise.
9977         (thumb2_movsfcc_vfp): Likewise.
9978         (movdfcc_vfp): Likewise.
9979         (thumb2_movdfcc_vfp): Likewise.
9980         * config/arm/arm.c (cortexa7_older_only): Update for attribute change.
9981         * config/arm/arm1020e.md (v10_c2v): Update for attribute change.
9982         (v10_v2c): Likewise.
9983         * config/arm/cortex-a15-neon.md (cortex_a15_neon_int_1): Update for
9984         attribute change.
9985         (cortex_a15_neon_int_2): Likewise.
9986         (cortex_a15_neon_int_3): Likewise.
9987         (cortex_a15_neon_int_4): Likewise.
9988         (cortex_a15_neon_int_5): Likewise.
9989         (cortex_a15_neon_vqneg_vqabs): Likewise.
9990         (cortex_a15_neon_vmov): Likewise.
9991         (cortex_a15_neon_vaba): Likewise.
9992         (cortex_a15_neon_vaba_qqq): Likewise.
9993         (cortex_a15_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9994         (cortex_a15_neon_mul_qqq_8_16_32_ddd_32): Likewise.
9995         (cortex_a15_neon_mul_qdd_64_32_long_qqd_16_ddd_32_\
9996         scalar_64_32_long_scalar): Likewise.
9997         (cortex_a15_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
9998         (cortex_a15_neon_mla_qqq_8_16): Likewise.
9999         (cortex_a15_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10000         lotype_qdd_64_32_long): Likewise.
10001         (cortex_a15_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10002         (cortex_a15_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10003         (cortex_a15_neon_mul_qqd_32_scalar): Likewise.
10004         (cortex_a15_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10005         (cortex_a15_neon_shift_1): Likewise.
10006         (cortex_a15_neon_shift_2): Likewise.
10007         (cortex_a15_neon_shift_3): Likewise.
10008         (cortex_a15_neon_vshl_ddd): Likewise.
10009         (cortex_a15_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10010         (cortex_a15_neon_vsra_vrsra): Likewise.
10011         (cortex_a15_neon_fp_vadd_ddd_vabs_dd): Likewise.
10012         (cortex_a15_neon_fp_vadd_qqq_vabs_qq): Likewise.
10013         (cortex_a15_neon_fp_vmul_ddd): Likewise.
10014         (cortex_a15_neon_fp_vmul_qqd): Likewise.
10015         (cortex_a15_neon_fp_vmla_ddd): Likewise.
10016         (cortex_a15_neon_fp_vmla_qqq): Likewise.
10017         (cortex_a15_neon_fp_vmla_ddd_scalar): Likewise.
10018         (cortex_a15_neon_fp_vmla_qqq_scalar): Likewise.
10019         (cortex_a15_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10020         (cortex_a15_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10021         (cortex_a15_neon_bp_simple): Likewise.
10022         (cortex_a15_neon_bp_2cycle): Likewise.
10023         (cortex_a15_neon_bp_3cycle): Likewise.
10024         (cortex_a15_neon_vld1_1_2_regs): Likewise.
10025         (cortex_a15_neon_vld1_3_4_regs): Likewise.
10026         (cortex_a15_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10027         (cortex_a15_neon_vld2_4_regs): Likewise.
10028         (cortex_a15_neon_vld3_vld4): Likewise.
10029         (cortex_a15_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10030         (cortex_a15_neon_vst1_3_4_regs): Likewise.
10031         (cortex_a15_neon_vst2_4_regs_vst3_vst4): Likewise.
10032         (cortex_a15_neon_vst3_vst4): Likewise.
10033         (cortex_a15_neon_vld1_vld2_lane): Likewise.
10034         (cortex_a15_neon_vld3_vld4_lane" 10
10035         (cortex_a15_neon_vst1_vst2_lane): Likewise.
10036         (cortex_a15_neon_vst3_vst4_lane): Likewise.
10037         (cortex_a15_neon_vld3_vld4_all_lanes): Likewise.
10038         (cortex_a15_neon_ldm_2): Likewise.0
10039         (cortex_a15_neon_stm_2): Likewise.
10040         (cortex_a15_neon_mcr): Likewise.
10041         (cortex_a15_neon_mcr_2_mcrr): Likewise.
10042         (cortex_a15_neon_mrc): Likewise.
10043         (cortex_a15_neon_mrrc): Likewise.
10044         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
10045         change.
10046         (cortex_a15_alu_shift): Likewise.
10047         (cortex_a15_alu_shift_reg): Likewise.
10048         (cortex_a15_mult32): Likewise.
10049         (cortex_a15_mult64): Likewise.
10050         (cortex_a15_block): Likewise.
10051         (cortex_a15_branch): Likewise.
10052         (cortex_a15_load1): Likewise.
10053         (cortex_a15_load3): Likewise.
10054         (cortex_a15_store1): Likewise.
10055         (cortex_a15_store3): Likewise.
10056         (cortex_a15_call): Likewise.
10057         * config/arm/cortex-a5.md (cortex_a5_r2f): Update for attribute change.
10058         (cortex_a5_f2r): Likewise.
10059         * config/arm/cortex-a53.md (cortex_a53_r2f): Update for attribute
10060         change.
10061         (cortex_a53_f2r): Likewise.
10062         * config/arm/cortex-a7.md
10063         (cortex_a7_branch): Update for attribute change.
10064         (cortex_a7_call): Likewise.
10065         (cortex_a7_alu_imm): Likewise.
10066         (cortex_a7_alu_reg): Likewise.
10067         (cortex_a7_alu_shift): Likewise.
10068         (cortex_a7_mul): Likewise.
10069         (cortex_a7_load1): Likewise.
10070         (cortex_a7_store1): Likewise.
10071         (cortex_a7_load2): Likewise.
10072         (cortex_a7_store2): Likewise.
10073         (cortex_a7_load3): Likewise.
10074         (cortex_a7_store3): Likewise.
10075         (cortex_a7_load4): Likewise.
10076         (cortex_a7_store4): Likewise.
10077         (cortex_a7_fpalu): Likewise.
10078         (cortex_a7_fconst): Likewise.
10079         (cortex_a7_fpmuls): Likewise.
10080         (cortex_a7_neon_mul): Likewise.
10081         (cortex_a7_fpmacs): Likewise.
10082         (cortex_a7_neon_mla: Likewise.
10083         (cortex_a7_fpmuld: Likewise.
10084         (cortex_a7_fpmacd: Likewise.
10085         (cortex_a7_fpfmad: Likewise.
10086         (cortex_a7_fdivs: Likewise.
10087         (cortex_a7_fdivd: Likewise.
10088         (cortex_a7_r2f: Likewise.
10089         (cortex_a7_f2r: Likewise.
10090         (cortex_a7_f_flags: Likewise.
10091         (cortex_a7_f_loads: Likewise.
10092         (cortex_a7_f_loadd: Likewise.
10093         (cortex_a7_f_stores: Likewise.
10094         (cortex_a7_f_stored: Likewise.
10095         (cortex_a7_neon): Likewise.
10096         * config/arm/cortex-a8-neon.md
10097         (cortex_a8_neon_mrc): Update for attribute change.
10098         (cortex_a8_neon_mrrc): Likewise.
10099         (cortex_a8_neon_int_1): Likewise.
10100         (cortex_a8_neon_int_2): Likewise.
10101         (cortex_a8_neon_int_3): Likewise.
10102         (cortex_a8_neon_int_4): Likewise.
10103         (cortex_a8_neon_int_5): Likewise.
10104         (cortex_a8_neon_vqneg_vqabs): Likewise.
10105         (cortex_a8_neon_vmov): Likewise.
10106         (cortex_a8_neon_vaba): Likewise.
10107         (cortex_a8_neon_vaba_qqq): Likewise.
10108         (cortex_a8_neon_vsma): Likewise.
10109         (cortex_a8_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10110         (cortex_a8_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10111         (cortex_a8_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10112         long_scalar): Likewise.
10113         (cortex_a8_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10114         (cortex_a8_neon_mla_qqq_8_16): Likewise.
10115         (cortex_a8_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10116         long_scalar_qdd_64_32_long): Likewise.
10117         (cortex_a8_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10118         (cortex_a8_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10119         (cortex_a8_neon_mul_qqd_32_scalar): Likewise.
10120         (cortex_a8_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10121         (cortex_a8_neon_shift_1): Likewise.
10122         (cortex_a8_neon_shift_2): Likewise.
10123         (cortex_a8_neon_shift_3): Likewise.
10124         (cortex_a8_neon_vshl_ddd): Likewise.
10125         (cortex_a8_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10126         (cortex_a8_neon_vsra_vrsra): Likewise.
10127         (cortex_a8_neon_fp_vadd_ddd_vabs_dd): Likewise.
10128         (cortex_a8_neon_fp_vadd_qqq_vabs_qq): Likewise.
10129         (cortex_a8_neon_fp_vsum): Likewise.
10130         (cortex_a8_neon_fp_vmul_ddd): Likewise.
10131         (cortex_a8_neon_fp_vmul_qqd): Likewise.
10132         (cortex_a8_neon_fp_vmla_ddd): Likewise.
10133         (cortex_a8_neon_fp_vmla_qqq): Likewise.
10134         (cortex_a8_neon_fp_vmla_ddd_scalar): Likewise.
10135         (cortex_a8_neon_fp_vmla_qqq_scalar): Likewise.
10136         (cortex_a8_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10137         (cortex_a8_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10138         (cortex_a8_neon_bp_simple): Likewise.
10139         (cortex_a8_neon_bp_2cycle): Likewise.
10140         (cortex_a8_neon_bp_3cycle): Likewise.
10141         (cortex_a8_neon_ldr): Likewise.
10142         (cortex_a8_neon_str): Likewise.
10143         (cortex_a8_neon_vld1_1_2_regs): Likewise.
10144         (cortex_a8_neon_vld1_3_4_regs): Likewise.
10145         (cortex_a8_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10146         (cortex_a8_neon_vld2_4_regs): Likewise.
10147         (cortex_a8_neon_vld3_vld4): Likewise.
10148         (cortex_a8_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10149         (cortex_a8_neon_vst1_3_4_regs): Likewise.
10150         (cortex_a8_neon_vst2_4_regs_vst3_vst4): Likewise.
10151         (cortex_a8_neon_vst3_vst4): Likewise.
10152         (cortex_a8_neon_vld1_vld2_lane): Likewise.
10153         (cortex_a8_neon_vld3_vld4_lane): Likewise.
10154         (cortex_a8_neon_vst1_vst2_lane): Likewise.
10155         (cortex_a8_neon_vst3_vst4_lane): Likewise.
10156         (cortex_a8_neon_vld3_vld4_all_lanes): Likewise.
10157         (cortex_a8_neon_mcr): Likewise.
10158         (cortex_a8_neon_mcr_2_mcrr): Likewise.
10159         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
10160         * config/arm/cortex-a9-neon.md (ca9_neon_mrc): Update for attribute
10161         change.
10162         (ca9_neon_mrrc): Likewise.
10163         (cortex_a9_neon_int_1): Likewise.
10164         (cortex_a9_neon_int_2): Likewise.
10165         (cortex_a9_neon_int_3): Likewise.
10166         (cortex_a9_neon_int_4): Likewise.
10167         (cortex_a9_neon_int_5): Likewise.
10168         (cortex_a9_neon_vqneg_vqabs): Likewise.
10169         (cortex_a9_neon_vmov): Likewise.
10170         (cortex_a9_neon_vaba): Likewise.
10171         (cortex_a9_neon_vaba_qqq): Likewise.
10172         (cortex_a9_neon_vsma): Likewise.
10173         (cortex_a9_neon_mul_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10174         (cortex_a9_neon_mul_qqq_8_16_32_ddd_32): Likewise.
10175         (cortex_a9_neon_mul_qdd_64_32_long_qqd_16_ddd_32_scalar_64_32_\
10176         long_scalar): Likewise.
10177         (cortex_a9_neon_mla_ddd_8_16_qdd_16_8_long_32_16_long): Likewise.
10178         (cortex_a9_neon_mla_qqq_8_16): Likewise.
10179         (cortex_a9_neon_mla_ddd_32_qqd_16_ddd_32_scalar_qdd_64_32_\
10180         long_scalar_qdd_64_32_long): Likewise.
10181         (cortex_a9_neon_mla_qqq_32_qqd_32_scalar): Likewise.
10182         (cortex_a9_neon_mul_ddd_16_scalar_32_16_long_scalar): Likewise.
10183         (cortex_a9_neon_mul_qqd_32_scalar): Likewise.
10184         (cortex_a9_neon_mla_ddd_16_scalar_qdd_32_16_long_scalar): Likewise.
10185         (cortex_a9_neon_shift_1): Likewise.
10186         (cortex_a9_neon_shift_2): Likewise.
10187         (cortex_a9_neon_shift_3): Likewise.
10188         (cortex_a9_neon_vshl_ddd): Likewise.
10189         (cortex_a9_neon_vqshl_vrshl_vqrshl_qqq): Likewise.
10190         (cortex_a9_neon_vsra_vrsra): Likewise.
10191         (cortex_a9_neon_fp_vadd_ddd_vabs_dd): Likewise.
10192         (cortex_a9_neon_fp_vadd_qqq_vabs_qq): Likewise.
10193         (cortex_a9_neon_fp_vsum): Likewise.
10194         (cortex_a9_neon_fp_vmul_ddd): Likewise.
10195         (cortex_a9_neon_fp_vmul_qqd): Likewise.
10196         (cortex_a9_neon_fp_vmla_ddd): Likewise.
10197         (cortex_a9_neon_fp_vmla_qqq): Likewise.
10198         (cortex_a9_neon_fp_vmla_ddd_scalar): Likewise.
10199         (cortex_a9_neon_fp_vmla_qqq_scalar): Likewise.
10200         (cortex_a9_neon_fp_vrecps_vrsqrts_ddd): Likewise.
10201         (cortex_a9_neon_fp_vrecps_vrsqrts_qqq): Likewise.
10202         (cortex_a9_neon_bp_simple): Likewise.
10203         (cortex_a9_neon_bp_2cycle): Likewise.
10204         (cortex_a9_neon_bp_3cycle): Likewise.
10205         (cortex_a9_neon_ldr): Likewise.
10206         (cortex_a9_neon_str): Likewise.
10207         (cortex_a9_neon_vld1_1_2_regs): Likewise.
10208         (cortex_a9_neon_vld1_3_4_regs): Likewise.
10209         (cortex_a9_neon_vld2_2_regs_vld1_vld2_all_lanes): Likewise.
10210         (cortex_a9_neon_vld2_4_regs): Likewise.
10211         (cortex_a9_neon_vld3_vld4): Likewise.
10212         (cortex_a9_neon_vst1_1_2_regs_vst2_2_regs): Likewise.
10213         (cortex_a9_neon_vst1_3_4_regs): Likewise.
10214         (cortex_a9_neon_vst2_4_regs_vst3_vst4): Likewise.
10215         (cortex_a9_neon_vst3_vst4): Likewise.
10216         (cortex_a9_neon_vld1_vld2_lane): Likewise.
10217         (cortex_a9_neon_vld3_vld4_lane): Likewise.
10218         (cortex_a9_neon_vst1_vst2_lane): Likewise.
10219         (cortex_a9_neon_vst3_vst4_lane): Likewise.
10220         (cortex_a9_neon_vld3_vld4_all_lanes): Likewise.
10221         (cortex_a9_neon_mcr): Likewise.
10222         (cortex_a9_neon_mcr_2_mcrr): Likewise.
10223         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
10224         (cortex_a9_fps): Likewise.
10225         * config/arm/cortex-m4-fpu.md (cortex_m4_vmov_2): Update for attribute
10226         change.
10227         (cortex_m4_fmuls): Likewise.
10228         * config/arm/cortex-r4f.md (cortex_r4_mcr): Update for attribute
10229         change.
10230         (cortex_r4_mrc): Likewise.
10231         * config/arm/iterators.md: Update comment referring to neon_type.
10232         * config/arm/iwmmxt.md (iwmmxt_arm_movdi): Update for attribute change.
10233         (iwmmxt_movsi_insn): Likewise.
10234         * config/arm/marvell-pj4.md (pj4_vfp_to_core): Update for
10235         attribute change.
10236         (pj4_core_to_vfp): Likewise.
10237         * config/arm/neon-schedgen.ml (emit_insn_reservations): Update for
10238         attribute change.
10239         * config/arm/vfp11.md (vfp_fload): Update for attribute change.
10240         (vfp_fstore): Likewise.
10241         * doc/md.texi: Change references to neon_type to refer to type.
10243 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10245         * tree.h (DECL_BUILT_IN): Fix typo in comment.
10247 2013-09-04  David Edelsohn  <dje.gcc@gmail.com>
10249         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Only emit
10250         lglobl if not weak.
10252 2013-09-04  Easwaran Raman  <eraman@google.com>
10254         PR middle-end/57370
10255         * tree-ssa-reassoc.c (get_stmt_uid_with_default): New function,
10256         (build_and_add_sum): Use it.
10257         (appears_later_in_bb): Simplify code.
10259 2013-09-04  Teresa Johnson  <tejohnson@google.com>
10261         * dumpfile.c (dump_finish): Don't close stderr/stdout.
10263 2013-09-04  James Greenhalgh  <james.greenhalgh@arm.com>
10265         * config/aarch64/arm_neon.h (vaddvq_<su>64): Fix return types.
10267 2013-09-04  Jan Hubicka  <jh@suse.cz>
10269         * Makefile.in (ipa-devirt.o): Add dependency on diagnostic.h
10270         * ipa-devirt.c: Include diganostic.h
10271         (odr_type_d): Add types and types_set.
10272         (hash_type_name): Work for types with vtables during LTO.
10273         (odr_hasher::remove): Fix comment; destroy types_set.
10274         (add_type_duplicate): New function,
10275         (get_odr_type): Use it.
10276         (dump_type_inheritance_graph): Dump type duplicates.
10277         * ipa.c (symtab_remove_unreachable_nodes): Build type inheritance
10278         graph.
10279         * tree.c (types_same_for_odr): Give exact answers on types with
10280         virtual tables.
10282 2013-09-04  Dodji Seketeli  <dodji@redhat.com>
10284         * tree.h (DECL_BUILT_IN, DECL_IS_BUILTIN): Add more comments
10285         explaining their differences.
10287 2013-09-04  Sandeep Kumar Singh<Sandeep.Singh2@kpitcummins.com>
10289         * config/rx/rx.h: Add option -mcpu for target variants RX100 and RX200.
10291 2013-09-03  Jeff Law  <law@redhat.com>
10293         * tree-ssa-threadedge.c (thread_across_edge): Record entire path
10294         when not threading through a joiner block.  Pass joiner/no joiner
10295         state to register_jump_thread.
10296         * tree-ssa-threadupdate.c (register_jump_thread): Get joiner/no joiner
10297         state from argument rather than implying on path length.
10298         Dump the entire jump thread path into debugging dump.
10299         * tree-flow.h (register_jump_thread): Update prototype.
10301 2013-08-29  Xinliang David Li  <davidxl@google.com>
10303         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10304         Remove a trivial gcc_assert.
10306 2013-08-29  Xinliang David Li  <davidxl@google.com>
10308         * tree-vect-slp.c (destroy_bb_vec_info): Data ref cleanup.
10309         * tree-vect-loop.c (destroy_bb_vec_info): Ditto.
10310         * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
10311         Delay base decl alignment adjustment.
10312         * tree-vectorizer.c (vect_destroy_datarefs): New function.
10313         * tree-vectorizer.h: New data structure.
10314         (set_dr_misalignment): New function.
10315         (dr_misalignment): Ditto.
10316         * tree-vect-stmts.c (vectorizable_store): Ensure alignment.
10317         (vectorizable_load): Ditto.
10318         (ensure_base_align): New function.
10319         (vectorize_loops): Add dbg_cnt support.
10320         (execute_vect_slp): Ditto.
10321         * dbgcnt.def: New debug counter.
10322         * Makefile: New dependency.
10324 2013-09-03  Meador Inge  <meadori@codesourcery.com>
10326         Revert:
10328         2013-08-30  Meador Inge  <meadori@codesourcery.com>
10330         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10332 2013-09-03  David Edelsohn  <dje.gcc@gmail.com>
10334         * config/rs6000/rs6000.h (ASM_OUTPUT_DEF_FROM_DECLS): Emit lglobl for
10335         function descriptor.
10337 2013-09-03  Richard Biener  <rguenther@suse.de>
10339         * tree-affine.c (add_elt_to_tree): Fix association issue,
10340         avoid useless converts and make sure to always return a
10341         properly typed result.
10343 2013-09-03  Richard Biener  <rguenther@suse.de>
10345         PR middle-end/57656
10346         * fold-const.c (negate_expr_p): Fix division case.
10347         (negate_expr): Likewise.
10349 2013-09-03  Richard Biener  <rguenther@suse.de>
10351         PR lto/58285
10352         * tree-streamer-out.c: Include tm.h.
10353         * Makefile.in (tree-streamer-out.o): Depend on $(TM_H).
10355 2013-09-03  Jan Hubicka  <jh@suse.cz>
10357         * tree-profile.c (tree_profiling): Cleanup CFG when done.
10359 2013-09-03  Alan Modra  <amodra@gmail.com>
10361         * config.gcc (powerpc*-*-linux*): Add support for little-endian
10362         multilibs to big-endian target and vice versa.
10363         * config/rs6000/t-linux64: Use := assignment on all vars.
10364         (MULTILIB_EXTRA_OPTS): Remove fPIC.
10365         (MULTILIB_OSDIRNAMES): Specify using mapping from multilib_options.
10366         * config/rs6000/t-linux64le: New file.
10367         * config/rs6000/t-linux64bele: New file.
10368         * config/rs6000/t-linux64lebe: New file.
10370 2013-09-02  Jan Hubicka  <jh@suse.cz>
10372         * ipa-inline-transform.c (inline_transform): Do not
10373         optimize_inline_calls when not optimizing.
10375 2013-09-02  Jan Hubicka  <jh@suse.cz>
10377         * lto-symtab.c (lto_symtab_merge_symbols): Add comments; merge
10378         duplicated nodes for assembler names.
10379         * symtab.c (symtab_unregister_node): Do not attempt to unlink
10380         hard registers from assembler name hash.
10382 2013-09-02  Jan Hubicka  <jh@suse.cz>
10384         * ipa-split.c (execute_split_functions): Split externally visible
10385         functions called once.
10387 2013-09-02  Martin Jambor  <mjambor@suse.cz>
10389         PR ipa/58106
10390         * ipa-prop.c (ipa_edge_duplication_hook): Always put new rdesc to the
10391         linked list.  When finding the correct duplicate, also consider also
10392         the caller in additon to its inlined_to node.
10394 2013-09-02  James Greenhalgh  <james.greenhalgh@arm.com>
10396         * config/aarch64/aarch64-simd-builtins.def
10397         (dup_lane_scalar): Remove.
10398         * config/aarch64/aarch64-simd.md
10399         (aarch64_simd_dup): Add 'w->w' alternative.
10400         (aarch64_dup_lane<mode>): Allow for VALL.
10401         (aarch64_dup_lane_scalar<mode>): Remove.
10402         (aarch64_dup_lane_<vswap_width_name><mode>): New.
10403         (aarch64_get_lane_signed<mode>): Add w->w altenative.
10404         (aarch64_get_lane_unsigned<mode>): Likewise.
10405         (aarch64_get_lane<mode>): Likewise.
10406         * config/aarch64/aarch64.c (aarch64_evpc_dup): New.
10407         (aarch64_expand_vec_perm_const_1): Use aarch64_evpc_dup.
10408         * config/aarch64/iterators.md (VSWAP_WIDTH): New.
10409         (VCON): Change container of V2SF.
10410         (vswap_width_name): Likewise.
10411         * config/aarch64/arm_neon.h
10412         (__aarch64_vdup_lane_any): New.
10413         (__aarch64_vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10414         (vdup<q>_n_<psuf><8,16,32,64>): Convert to C implementation.
10415         (vdup<q>_lane<q>_<fpsu><8,16,32,64>): Likewise.
10417 2013-09-02  Eric Botcazou  <ebotcazou@adacore.com>
10419         PR middle-end/56382
10420         * expr.c (emit_move_complex): Do not move complex FP values as parts if
10421         the source or the destination is a single hard register.
10423 2013-09-02  Richard Biener  <rguenther@suse.de>
10425         PR middle-end/57511
10426         * tree-scalar-evolution.c (instantiate_scev_name): Allow
10427         non-linear SCEVs.
10429 2013-09-02  Richard Biener  <rguenther@suse.de>
10431         * tree-affine.c (add_elt_to_tree): Avoid converting all pointer
10432         arithmetic to sizetype.
10434 2013-09-02  Bin Cheng  <bin.cheng@arm.com>
10436         * tree-ssa-loop-ivopts.c (set_autoinc_for_original_candidates):
10437         Find auto-increment use both before and after candidate.
10439 2013-09-02  Marek Polacek  <polacek@redhat.com>
10441         * Makefile.in (ubsan.o): Add $(TM_P_H) dependency.
10443 2013-09-01  Jan Hubicka  <jh@suse.cz>
10445         * Makefile.in: Add ipa-profile.o
10446         (ipa.o, ipa-devrit.o, ipa-inline-analysis.o): Adjust dependencies.
10447         * cgraph.c (struct cgraph_propagate_frequency_data,
10448         cgraph_propagate_frequency_1, cgraph_propagate_frequency): Move to
10449         ipa-profile.c; replace cgraph_ by ipa_ prefix.
10450         * cgraph.h (cgraph_propagate_frequency): Remove.
10451         * ipa-inline-analysis.c: Include ipa-utils.h;
10452         drop duplicated cfgloop.h.
10453         (inline_update_callee_summaries): Update.
10454         * ipa-profile.c: New file.
10455         * ipa-utils.h (ipa_propagate_frequency): Declare.
10456         * ipa.c: Do not include pointer-set.h, hash-table.h, lto-streamer.h,
10457         data-streamer.h, value-prof.h.
10458         (symtab_remove_unreachable_nodes): Update profile.
10459         (struct histogram_entry, histogram, histogram_pool, histogram_hash,
10460         account_time_size, cmp_counts, dump_histogram,
10461         ipa_profile_generate_summary, ipa_profile_write_summary,
10462         ipa_profile_read_summary, ipa_profile, gate_ipa_profile,
10463         pass_data_ipa_profile, pass_ipa_profile, make_pass_ipa_profile):
10464         Move to ipa-profile.c.
10466 2013-09-01  John David Anglin  <danglin@gcc.gnu.org>
10468         * config/pa/pa.md: Allow "const 0" operand 1 in "scc" insns.
10470 2013-09-01  Jan Hubicka  <jh@suse.cz>
10472         * common.opt (fdevirtualize-speculatively): New function.
10473         * invoke.texi (fdevirtualize-speculatively): Document.
10474         * ipa-devirt.c: Include ipa-inline.h
10475         (likely_target_p): New function.
10476         (ipa_devirt): New function.
10477         (gate_ipa_devirt): New function.
10478         (pass_data_ipa_devirt): New static var.
10479         (pass_ipa_devirt): Likewise.
10480         (make_pass_ipa_devirt): New function.
10481         * opts.c (default_options): Add OPT_fdevirtualize_speculatively.
10482         (common_handle_option): Disable devirtualization when
10483         value range profiling is available.
10484         * passes.def (pass_ipa_devirt): Add.
10485         * timever.def (TV_IPA_DEVIRT): New timevar.
10486         * tree-pass.h (make_pass_ipa_devirt):
10488 2013-09-01  Iain Sandoe  <iain@codesourcery.com>
10490         * config/darwin.h (LINK_COMMAND_SPEC_A): Revise sanitizer specs to
10491         include sanitize(undefined).
10493 2013-08-31  Diego Novillo  <dnovillo@google.com>
10495         * Makefile.in (TREE_CORE_H): Define.
10496         (TREE_H): Use.
10497         (GTFILES): Add tree-core.h.
10498         * builtins.c (built_in_class_names): Use BUILT_IN_LAST to
10499         size the array.
10500         * tree-core.h: New file.
10501         Move all data structures, enum, typedefs, global
10502         declarations and constants from ...
10503         * tree.h: ... here.
10505 2013-08-31  Jan Hubicka  <jh@suse.cz>
10507         * bulitins.c (expand_builtin): Do not early exit for gcov
10508         instrumented functions.
10510 2013-08-31  Marek Polacek  <polacek@redhat.com>
10512         * ubsan.c: Include tm_p.h.
10514 2013-08-31  Jan Hubicka  <jh@suse.cz>
10516         * gimple-streamer-in.c (input_gimple_stmt): Silence parameter unused
10517         warning.
10519         * cgraph.c (cgraph_get_body): Update call of lto_input_function_body.
10520         * gimple-streamer-in.c (input_gimple_stmt): Move sanity check to ...
10521         * tree-cfg.c (verify_gimple_label): ... here.
10522         * ipa-utils.c: Include lto-streamer.h, ipa-inline.h
10523         (ipa_merge_profiles): New function.
10524         * lto-streamer-in.c (lto_read_body): Take node instead of fn_decl.
10525         (lto_input_function_body): Likewise.
10526         * ipa-utils.h (ipa_merge_profiles): Declare.
10527         * lto-streamer.h (lto_input_function_body): Update prototype.
10528         (emit_label_in_global_context_p): Remove.
10529         * lto-symtab.c: Include ipa-utils.h
10530         (lto_cgraph_replace_node): Use ipa_merge_profiles.
10532 2013-08-31  Jan Hubicka  <jh@suse.cz>
10534         * cgraph.c (cgraph_speculative_call_info): Fix ref lookup
10536 2013-08-31  Jan Hubicka  <jh@suse.cz>
10538         * basic-block.h (apply_scale): Make scale parmeter gcov_type.
10540 2013-08-31  Uros Bizjak  <ubizjak@gmail.com>
10542         * config/alpha/alpha.c (alpha_emit_conditional_move): Update
10543         "cmp" RTX before signed_comparison_operator check to account
10544         for "code" changes.
10546 2013-08-30  Jan Hubicka  <jh@suse.cz>
10548         * ipa-prop.c (ipa_set_jf_known_type): Check that we add only records.
10549         (detect_type_change_1): Rename to ...
10550         (detect_type_change): ... this one; early return on non-polymorphic
10551         types.
10552         (detect_type_change_ssa): Add comp_type parameter; update
10553         use of detect_type_change.
10554         (compute_complex_assign_jump_func): Add param_type parameter;
10555         update use of detect_type_change_ssa.
10556         (compute_complex_ancestor_jump_func): Likewise.
10557         (ipa_get_callee_param_type): New function.
10558         (ipa_compute_jump_functions_for_edge): Compute parameter type;
10559         update calls to the jump function computation functions.
10561 2013-08-30  Teresa Johnson  <tejohnson@google.com>
10562             Steven Bosscher  <steven@gcc.gnu.org>
10564         * cfgrtl.c (fixup_new_cold_bb): New routine.
10565         (commit_edge_insertions): Invoke fixup_partitions.
10566         (find_partition_fixes): New routine.
10567         (fixup_partitions): Ditto.
10568         (verify_hot_cold_block_grouping): Update comments.
10569         (rtl_verify_edges): Invoke find_partition_fixes.
10570         (rtl_verify_bb_pointers): Update comments.
10571         (rtl_verify_bb_layout): Ditto.
10572         * basic-block.h (probably_never_executed_edge_p): Declare.
10573         (fixup_partitions): Ditto.
10574         * cfgcleanup.c (try_optimize_cfg): Invoke fixup_partitions.
10575         * bb-reorder.c (sanitize_hot_paths): New function.
10576         (find_rarely_executed_basic_blocks_and_crossing_edges): Invoke
10577         sanitize_hot_paths.
10578         * predict.c (probably_never_executed_edge_p): New routine.
10579         * cfg.c (check_bb_profile): Add partition insanity warnings.
10581 2013-08-30  Meador Inge  <meadori@codesourcery.com>
10583         * tree-vrp.c (check_array_ref): Bail out on zero-length arrays.
10585 2013-08-30  Marek Polacek  <polacek@redhat.com>
10587         * Makefile.in (ubsan.o): Add.
10588         (c-family/c-ubsan.o): Add.
10589         (builtins.o): Add ubsan.h dependency.
10590         * ubsan.h: New file.
10591         * ubsan.c: New file.
10592         * common.opt: Add -fsanitize=undefined option.
10593         (flag_sanitize): Add variable.
10594         (fsanitize=): Add option.  Add Driver.
10595         (fsanitize=thread): Remove option.
10596         (fsanitize=address): Likewise.
10597         (static-libubsan): New option.
10598         * doc/invoke.texi: Document the new flag and -static-libubsan.
10599         * sanitizer.def (DEF_SANITIZER_BUILTIN): Define.
10600         (BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE): Define.
10601         * builtin-attrs.def (ATTR_COLD): Define.
10602         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10603         * builtins.def (BUILT_IN_UBSAN_HANDLE_DIVREM_OVERFLOW,
10604         BUILT_IN_UBSAN_HANDLE_SHIFT_OUT_OF_BOUNDS): Define.
10605         * flag-types.h (sanitize_code): New enum.
10606         * opts.c (common_handle_option): Parse command line arguments
10607         of -fsanitize=.  Add -fsanitize=unreachable option.
10608         * varasm.c (get_variable_section): Adjust.
10609         (assemble_noswitch_variable): Likewise.
10610         (assemble_variable): Likewise.
10611         (output_constant_def_contents): Likewise.
10612         (categorize_decl_for_section): Likewise.
10613         (place_block_symbol): Likewise.
10614         (output_object_block): Likewise.
10615         * builtins.def: Likewise.
10616         * toplev.c (compile_file): Likewise.
10617         (process_options): Likewise.
10618         * cppbuiltin.c: Likewise.
10619         * tsan.c (tsan_pass): Likewise.
10620         (tsan_gate): Likewise.
10621         (tsan_gate_O0): Likewise.
10622         * cfgexpand.c (partition_stack_vars): Likewise.
10623         (expand_stack_vars): Likewise.
10624         (defer_stack_allocation): Likewise.
10625         (expand_used_vars): Likewise.
10626         * cfgcleanup.c (old_insns_match_p): Likewise.
10627         * asan.c (asan_finish_file): Likewise.
10628         (asan_instrument): Likewise.
10629         (gate_asan): Likewise.
10630         (initialize_sanitizer_builtins): Build BT_FN_VOID_PTR_PTR_PTR.
10631         (ATTR_COLD_NOTHROW_LEAF_LIST): Define.
10632         (asan_global_struct): Use pointer_sized_int_node instead
10633         calling build_nonstandard_integer_type.
10634         (initialize_sanitizer_builtins): Likewise.
10635         (asan_finish_file): Likewise.
10636         * gcc.c: Document %{%:function(args):X}.
10637         (static_spec_functions): Add sanitize.
10638         (handle_spec_function): Add retval_nonnull argument and if non-NULL,
10639         store funcval != NULL there.
10640         (do_spec_1): Adjust handle_spec_function caller.
10641         (handle_braces): Allow %:function(args) as condition.
10642         (sanitize_spec_function): New function.
10643         (ADD_STATIC_LIBUBSAN_LIBS): Define.
10644         (LIBUBSAN_SPEC): Likewise.
10645         (LIBUBSAN_EARLY_SPEC): Likewise.
10646         (SANITIZER_SPEC): Handle libubsan.
10647         (SANITIZER_EARLY_SPEC): Likewise.
10648         * config/darwin.h (LINK_COMMAND_SPEC_A): Use %:sanitize(address)
10649         instead of fsanitize=address.
10650         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Use %:sanitize(address)
10651         instead of fsanitize=address*.
10652         * builtins.c: Include ubsan.h.
10653         (fold_builtin_0): Instrument __builtin_unreachable.
10654         * config/rs6000/rs6000.h (FRAME_GROWS_DOWNWARD): Use flag_sanitize
10655         instead of flag_asan.
10656         * tree.h (enum tree_index): Add TI_POINTER_SIZED_TYPE.
10657         (pointer_sized_int_node): Define.
10658         * tree.c (build_common_tree_nodes): Initialize pointer_sized_int_node.
10660 2013-08-30  Mike Stump  <mikestump@comcast.net>
10662         * doc/install.texi (Prerequisites): Note regression in Tcl 8.6
10663         with RE patterns.
10665 2013-08-29  Jan Hubicka  <jh@suse.cz>
10667         * cgraph.c (cgraph_function_body_availability): Handle weakref
10668         correctly.
10669         * passes.def: Remove pass_fixup_cfg.
10670         * ipa-inline.c (ipa_inline): When not optimizing, do not inline;
10671         track when we need to remove functions.
10672         (gate_ipa_inline): Execute inlining always; add comment why.
10673         (pass_data_ipa_inline): Remove TODO_remove_functions.
10674         * ipa-inline-analysis.c (inline_generate_summary): When not optimizing
10675         do not produce summaries.
10676         * symtab.c (change_decl_assembler_name): Handle renaming of weakrefs.
10677         (symtab_nonoverwritable_alias): Assert we are not called on weakref.
10678         * varpool.c (cgraph_variable_initializer_availability): Fix weakrefs,
10679         constant pool and vtable.
10681 2013-08-30  Tejas Belagod  <tejas.belagod@arm.com>
10683         * config/aarch64/arm_neon.h (__AARCH64_UINT64_C, __AARCH64_INT64_C):
10684         New arm_neon.h's internal macros to specify 64-bit constants.
10685         Avoid using stdint.h's macros.
10687 2013-08-30  Joern Rennecke  <joern.rennecke@embecosm.com>
10689         * recog.c (verify_changes): Verify that changes[i].old is non-zero
10690         before applying REG_P.
10692 2013-08-30  Jakub Jelinek  <jakub@redhat.com>
10694         PR tree-optimization/58277
10695         * tree-ssa-strlen.c (strlen_enter_block): If do_invalidate gave up
10696         after seeing too many stmts with vdef in between dombb and current
10697         bb, invalidate everything.
10699 2013-08-30  Richard Biener  <rguenther@suse.de>
10701         * fold-const.c (fold_single_bit_test): Fix overflow test.
10703 2013-08-30  Eric Botcazou  <ebotcazou@adacore.com>
10705         * function.c (assign_parm_setup_reg): For a parameter passed by pointer
10706         and which can live in a register, always retrieve the value on entry.
10707         * var-tracking.c (add_stores): Treat the copy on entry for a parameter
10708         passed by invisible reference specially.
10709         (emit_notes_in_bb) <MO_VAL_USE>: Emit notes before the instruction.
10710         (vt_add_function_parameter): Correctly deal with a parameter passed by
10711         invisible reference.
10713 2013-08-30  Jan Hubicka  <jh@suse.cz>
10715         * tree.c (set_call_expr_flags): Fix handling of TM_PURE.
10717 2013-08-30  Richard Biener  <rguenther@suse.de>
10719         PR tree-optimization/58228
10720         * tree-vect-data-refs.c (vect_analyze_data_ref_access): Do not
10721         allow invariant loads in nested loop vectorization.
10723 2013-08-30  Richard Biener  <rguenther@suse.de>
10725         PR tree-optimization/58223
10726         * tree-loop-distribution.c (has_anti_dependence): Rename to ...
10727         (has_anti_or_output_dependence): ... this and adjust to also
10728         look for output dependences.
10729         (mark_nodes_having_upstream_mem_writes): Adjust.
10730         (rdg_flag_uses): Likewise.
10732 2013-08-30  Richard Biener  <rguenther@suse.de>
10734         PR tree-optimization/58010
10735         * tree-vect-loop.c (vect_create_epilog_for_reduction): Remove
10736         assert that we have a loop-closed PHI.
10738 2013-08-29  Jan Hubicka  <jh@suse.cz>
10740         * lto-symtab.c (lto_cgraph_replace_node): Free decl_in_state.
10741         * cgraph.c (cgraph_release_function_body): Free decl_in_state.
10742         * lto-section-in.c (lto_free_function_in_decl_state): New function.
10743         (lto_free_function_in_decl_state_for_node): New function.
10745 2013-08-29  Xinliang David Li  <davidxl@google.com>
10747         * loop-unroll.c (report_unroll_peel): Minor message change.
10748         * tree-vect-loop-manip.c (vect_do_peeling_for_alignment):
10749         Emit alignment peeling message with default -fopt-info.
10750         (vect_loop_versioning): Emit loop version info message.
10751         * tree-vectorizer.c (vectorize_loops): Minor message change.
10752         (execute_vect_slp): Ditto.
10754 2013-08-29  Eric Botcazou  <ebotcazou@adacore.com>
10756         * cgraphclones.c (cgraph_create_virtual_clone): Compute the DECL_NAME
10757         of the clone from the DECL_NAME of the original function.
10759 2013-08-29  Oleg Endo  <olegendo@gcc.gnu.org>
10761         * passes.c (register_pass): Add overload.
10762         * tree-pass.h (register_pass): Forward declare it.  Add comment.
10764 2013-08-29  Jan Hubicka  <jh@suse.cz>
10766         * lto-streamer-out.c (hash_tree): Stream DECL_FINAL_P,
10767         DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P and TYPE_FINAL_P.
10768         * lto-streamer-in.c (unpack_ts_decl_with_vis_value_fields): Stream
10769         DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10770         (unpack_ts_type_common_value_fields): Stream TYPE_FINAL_P.
10771         * tree-streamer-out.c (pack_ts_decl_with_vis_value_fields):
10772         Add DECL_FINAL_P, DECL_CXX_CONSTRUCTOR_P and DECL_CXX_DESTRUCTOR_P.
10773         (pack_ts_type_common_value_fields): Add TYPE_FINAL_P.
10775 2013-08-29  Teresa Johnson  <tejohnson@google.com>
10777         * dumpfile.c (dump_loc): Output column number.
10778         * dumpfile.h (OPTGROUP_OTHER): Add and enable under OPTGROUP_ALL.
10779         * doc/invoke.texi: Document optall -fopt-info flag.
10780         * profile.c (read_profile_edge_counts): Use new dump framework.
10781         (compute_branch_probabilities): Ditto.
10782         * passes.c (pass_manager::register_one_dump_file): Use OPTGROUP_OTHER
10783         when pass not in any opt group.
10784         * pass_manager.h (pass_manager::get_pass_profile): New method.
10785         * value-prof.c (check_counter): Use new dump framework.
10786         (check_ic_target): Ditto.
10787         * coverage.c (get_coverage_counts): Ditto.
10788         (coverage_init): Setup new dump framework.
10790 2013-08-29  Richard Biener  <rguenther@suse.de>
10792         PR tree-optimization/58246
10793         * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Properly
10794         handle the dominance check inside a basic-block.
10796 2013-08-29  Richard Biener  <rguenther@suse.de>
10798         PR middle-end/57287
10799         * tree-ssa-copy.c (may_propagate_copy): Allow propagating
10800         of default defs that appear in abnormal PHI nodes.
10802 2013-08-29  Richard Biener  <rguenther@suse.de>
10804         PR tree-optimization/57685
10805         * tree-vrp.c (register_edge_assert_for_1): Recurse only for
10806         single-use operands to avoid exponential complexity.
10808 2013-08-28  Dehao Chen  <dehao@google.com>
10810         * ipa-inline.c (edge_badness): Fix integer underflow.
10812 2013-08-28  Uros Bizjak  <ubizjak@gmail.com>
10814         * gtm-builtins.def (_ITM_free): Declare leaf.
10816 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10818         PR target/58067
10819         * config/i386/i386.md (*tls_global_dynamic_64_largepic): New insn.
10820         (*tls_local_dynamic_base_64_largepic): Likewise.
10821         (tls_global_dynamic_64_<mode>, tls_local_dynamic_base_64_<mode>):
10822         Remove predicate from call operand.
10823         * config/i386/i386.c (ix86_tls_get_addr): For -mcmodel=large -fpic
10824         return sum of pic_offset_table_rtx and UNSPEC_PLTOFF of the symbol.
10826 2013-08-28  Jeff Law  <law@redhat.com>
10828         * tree-ssa-threadedge.c (thread_around_empty_block): Remove
10829         checks for the number of predecessors and successors allowed.
10830         * tree-ssa-threadupdate.c (mark_threaded_blocks): Ignore requests
10831         which require copying a joiner block if there is a request which
10832         is a subpath that requires no joiner block copying.
10834 2013-08-28  Jan Hubicka  <jh@suse.cz>
10836         * lto-streamer-out.c (DFS_write_tree_body): Drop
10837         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX and BINFO_VPTR_INDEX.
10838         (hash_tree): Do not hash DECL_DEFER_OUTPUT, BINFO_INHERITANCE_CHAIN,
10839         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX, DECL_IN_TEXT_SECTION.
10840         * tree-streamer-in.c (unpack_ts_decl_common_value_fields):
10841         Do not read DECL_ERROR_ISSUED.
10842         (unpack_ts_decl_with_vis_value_fields): Do not read
10843         DECL_DEFER_OUTPUT.
10844         (lto_input_ts_binfo_tree_pointers): Do not read
10845         BINFO_INHERITANCE_CHAIN, BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX
10846         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Do not
10847         write DECL_ERROR_ISSUED..
10848         (pack_ts_decl_with_vis_value_fields): Do not write
10849         DECL_DEFER_OUTPUT.
10850         (write_ts_binfo_tree_pointers): Do not read BINFO_INHERITANCE_CHAIN,
10851         BINFO_SUBVTT_INDEX, BINFO_VPTR_INDEX.
10852         * print-tree.c (print_node): Do not print DECL_ERROR_ISSUED.
10853         * tree.h (tree_decl_common): Update comment.
10854         (DECL_ERROR_ISSUED): Remove.
10856 2013-08-28  Jakub Jelinek  <jakub@redhat.com>
10858         PR middle-end/58257
10859         * omp-low.c (copy_var_decl): Copy over TREE_NO_WARNING flag.
10861 2013-08-28  Jan Hubicka  <jh@suse.cz>
10863         * builtins.def (free): Declare leaf.
10865 2013-08-27  David Malcolm  <dmalcolm@redhat.com>
10867         * gdbhooks.py: New.
10868         * configure.ac (gdbinit.in): Add import of gcc/gdbhooks.py.
10869         * configure: Regenerate.
10871 2013-08-27  Martin Jambor  <mjambor@suse.cz>
10873         * ipa-prop.h (ipa_pass_through_data): New field type_preserved.
10874         (ipa_ancestor_jf_data): Likewise.
10875         (ipa_get_jf_pass_through_agg_preserved): Fix comment typo.
10876         (ipa_get_jf_pass_through_type_preserved): New function.
10877         (ipa_get_jf_ancestor_agg_preserved): Fix comment typo.
10878         (ipa_get_jf_ancestor_type_preserved): New function.
10879         * ipa-cp.c (ipa_get_jf_pass_through_result): Honor type_preserved flag.
10880         (ipa_get_jf_ancestor_result): Likewise.
10881         (propagate_vals_accross_pass_through): Use
10882         ipa_get_jf_pass_through_result to do all the value mappings.
10883         * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Dump the
10884         type_preserved flag.
10885         (ipa_set_jf_cst_copy): New function.
10886         (ipa_set_jf_simple_pass_through): Set the type_preserved flag.
10887         (ipa_set_jf_arith_pass_through): Likewise.
10888         (ipa_set_ancestor_jf): Likewise.
10889         (compute_complex_assign_jump_func): Set type_preserved instead of
10890         punting.
10891         (ipa_compute_jump_functions_for_edge): Likewise.
10892         (combine_known_type_and_ancestor_jfs): Honor type_preserved.
10893         (update_jump_functions_after_inlining): Update type_preserved.
10894         Explicitely create jump functions when combining one with pass_through.
10895         (ipa_write_jump_function): Stream the type_preserved flags.
10896         (ipa_read_jump_function): Likewise.
10898 2013-08-27  Jakub Jelinek  <jakub@redhat.com>
10899             Aldy Hernandez  <aldyh@redhat.com>
10901         * Makefile.in (omp-low.o): Depend on $(TARGET_H).
10902         * cfgloop.h (struct loop): Add safelen, force_vect, simduid.
10903         * function.h (struct function): Add has_force_vect_loops and
10904         has_simduid_loops.
10905         * gimple-pretty-print.c (dump_gimple_omp_for): Handle GF_OMP_FOR_KIND*.
10906         * gimple.c (gimple_build_omp_critical): Add KIND argument and
10907         handle it.
10908         * gimple.def: Update CLAUSES comments.
10909         * gimple.h (enum gf_mask): Add GF_OMP_FOR_KIND_{FOR,SIMD}.
10910         (gimple_build_omp_for): Add argument to prototype.
10911         (gimple_omp_for_kind): New.
10912         (gimple_omp_for_set_kind): New.
10913         * gimplify.c (enum gimplify_omp_var_data): Add GOVD_LINEAR to
10914         GOVD_DATA_SHARE_CLASS.
10915         (enum omp_region_type): Add ORT_SIMD.
10916         (gimple_add_tmp_var): Handle ORT_SIMD.
10917         (gimplify_var_or_parm_decl): Same.
10918         (is_gimple_stmt): Same.
10919         (omp_firstprivatize_variable): Same.
10920         (omp_add_variable): Only use splay_tree_insert if lookup failed.
10921         (omp_notice_variable): Handle ORT_SIMD.
10922         (omp_is_private): Add SIMD argument and handle it as well as ORT_SIMD.
10923         (omp_check_private): Handle ORT_SIMD.
10924         (gimplify_scan_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10925         OMP_CLAUSE_SAFELEN.
10926         (gimplify_adjust_omp_clauses_1): Handle GOVD_LINEAR.
10927         Handle OMP_CLAUSE_LASTPRIVATE.
10928         (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_LINEAR and
10929         OMP_CLAUSE_SAFELEN.
10930         (gimplify_omp_for): Handle OMP_SIMD and OMP_CLAUSE_LINEAR.
10931         (gimplify_expr): Handle OMP_SIMD.
10932         * internal-fn.c (expand_GOMP_SIMD_LANE): New.
10933         (expand_GOMP_SIMD_VF): New.
10934         (expand_GOMP_SIMD_LAST_LANE): New.
10935         * internal-fn.def (GOMP_SIMD_LANE): New.
10936         (GOMP_SIMD_VF): New.
10937         (GOMP_SIMD_LAST_LANE): New.
10938         * omp-low.c: Include target.h.
10939         (extract_omp_for_data): Handle OMP_SIMD, OMP_CLAUSE_LINEAR,
10940         OMP_CLAUSE_SAFELEN.
10941         (check_omp_nesting_restrictions): Same.
10942         (omp_max_vf): New.
10943         (lower_rec_simd_input_clauses): New.
10944         (lower_rec_input_clauses): Handle OMP_SIMD, GF_OMP_FOR_KIND_SIMD,
10945         OMP_CLAUSE_LINEAR.
10946         (lower_lastprivate_clauses): Handle OMP_CLAUSE_LINEAR,
10947         GF_OMP_FOR_KIND_SIMD, OMP_SIMD.
10948         (expand_omp_build_assign): New.
10949         (expand_omp_for_init_counts): New.
10950         (expand_omp_for_init_vars): New.
10951         (extract_omp_for_update_vars): New.
10952         (expand_omp_for_generic): Use expand_omp_for_{init,update}_vars
10953         and rewrite accordingly.
10954         (expand_omp_simd): New.
10955         (expand_omp_for): Use expand_omp_simd.
10956         (lower_omp_for_lastprivate): Unshare vinit when appropriate.
10957         (lower_omp_for): Do not lower the body.
10958         * tree-data-ref (get_references_in_stmt): Allow IFN_GOMP_SIMD_LANE
10959         in their own loops.
10960         * tree-flow.h (find_omp_clause): Remove prototype.
10961         * tree-if-conv.c (main_tree_if_conversion): Run if doing if conversion,
10962         forcing vectorization of the loop, or if flag_tree_vectorize.
10963         (gate_tree_if_conversion): Similarly.
10964         * tree-inline.c (remap_gimple_stmt): Pass for kind argument to
10965         gimple_build_omp_for.
10966         (copy_cfg_body): set has_force_vect_loops and has_simduid_loops.
10967         * tree-parloops (create_parallel_loop): Pass kind argument to
10968         gimple_build_omp_for.
10969         * tree-pretty-print.c (dump_omp_clause): Add cases for
10970         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_LINEAR, OMP_CLAUSE_SAFELEN,
10971         OMP_CLAUSE__SIMDUID_.
10972         (dump_generic_node): Handle OMP_SIMD.
10973         * tree-ssa-ccp.c (likely_value): Handle IFN_GOMP_SIMD*.
10974         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Do not
10975         unroll OMP_SIMD loops here.
10976         * tree-ssa-loop.c (gate_tree_vectorize): Run if has_force_vect_loops.
10977         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Handle
10978         loop->safelen.
10979         (vect_analyze_data_refs): Handle simd loops.
10980         * tree-vect-loop.c (vectorizable_live_operation): Handle
10981         IFN_GOMP_SIMD*.
10982         * tree-vect-stmts.c (vectorizable_call): Handle IFN_GOMP_SIMD_LANE.
10983         (vectorizable_store): Handle STMT_VINFO_SIMD_LANE_ACCESS_P.
10984         (vectorizable_load): Same.
10985         * tree-vectorizer.c: Include hash-table.h and tree-ssa-propagate.h.
10986         (struct simduid_to_vf): New.
10987         (simduid_to_vf::hash): New.
10988         (simduid_to-vf::equal): New.
10989         (struct simd_array_to_simduid): New.
10990         (simd_array_to_simduid::hash): New.
10991         (simd_array_to_simduid::equal): New.
10992         (adjust_simduid_builtins): New.
10993         (struct note_simd_array_uses_struct): New.
10994         (note_simd_array_uses_cb): New.
10995         (note_simd_array_uses): New.
10996         (vectorize_loops): Handle simd hints and adjust simd builtins
10997         accordingly.
10998         * tree-vectorizer.h (struct _stmt_vec_info): Add
10999         simd_lane_access_p field.
11000         (STMT_VINFO_SIMD_LANE_ACCESS_P): New macro.
11001         * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_LINEAR,
11002         OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_UNIFORM.
11003         (omp_clause_code_name): Same.
11004         (walk_tree_1): Handle OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN,
11005         OMP_CLAUSE__SIMDUID_, OMP_CLAUSE_LINEAR.
11006         * tree.def (OMP_SIMD): New entry.
11007         * tree.h (enum omp_clause_code): Add entries for OMP_CLAUSE_LINEAR,
11008         OMP_CLAUSE_UNIFORM, OMP_CLAUSE_SAFELEN, OMP_CLAUSE__SIMDUID_.
11009         (OMP_CLAUSE_DECL): Adjust range for new clauses.
11010         (OMP_CLAUSE_LINEAR_NO_COPYIN): New.
11011         (OMP_CLAUSE_LINEAR_NO_COPYOUT): New.
11012         (OMP_CLAUSE_LINEAR_STEP): New.
11013         (OMP_CLAUSE_SAFELEN_EXPR): New.
11014         (OMP_CLAUSE__SIMDUID__DECL): New.
11015         (find_omp_clause): New prototype.
11017 2013-08-27  H.J. Lu  <hongjiu.lu@intel.com>
11019         * config/i386/driver-i386.c (host_detect_local_cpu): Update
11020         Haswell processor detection.
11022 2013-08-27  Christian Widmer  <shadow@umbrox.de>
11024         PR target/57927
11025         * config/i386/driver-i386.c (host_detect_local_cpu): Add detection
11026         of Ivy Bridge and Haswell processors.  Assume core-avx2 for unknown
11027         AVX2 capable processors.
11029 2013-08-27  Tejas Belagod  <tejas.belagod@arm.com>
11031         * config/aarch64/arm_neon.h: Replace all inline asm implementations
11032         of vget_low_* with implementations in terms of other intrinsics.
11034 2013-08-27  Marc Glisse  <marc.glisse@inria.fr>
11036         PR middle-end/57219
11037         * doc/extend.texi (__builtin_isinf_sign): Restrict the return
11038         values to -1, 0 and 1.
11040 2013-08-27  Vidya Praveen  <vidyapraveen@arm.com>
11042         * config/aarch64/aarch64.md (unspec): Add UNSPEC_SISD_SSHL,
11043         UNSPEC_SISD_USHL, UNSPEC_USHL_2S, UNSPEC_SSHL_2S, UNSPEC_SISD_NEG.
11044         (<optab><mode>3_insn): Remove.
11045         (aarch64_ashl_sisd_or_int_<mode>3): New Pattern.
11046         (aarch64_lshr_sisd_or_int_<mode>3): Likewise.
11047         (aarch64_ashr_sisd_or_int_<mode>3): Likewise.
11048         (define_split for aarch64_lshr_sisd_or_int_di3): Likewise.
11049         (define_split for aarch64_lshr_sisd_or_int_si3): Likewise.
11050         (define_split for aarch64_ashr_sisd_or_int_di3): Likewise.
11051         (define_split for aarch64_ashr_sisd_or_int_si3): Likewise.
11052         (aarch64_sisd_ushl, aarch64_sisd_sshl): Likewise.
11053         (aarch64_ushl_2s, aarch64_sshl_2s, aarch64_sisd_neg_qi): Likewise.
11054         (ror<mode>3_insn): Likewise.
11055         * config/aarch64/predicates.md (aarch64_simd_register): New.
11057 2013-08-27  Richard Biener  <rguenther@suse.de>
11059         PR tree-optimization/57521
11060         * tree-if-conv.c (if_convertible_bb_p): Verify that at least
11061         one edge is non-critical.
11062         (find_phi_replacement_condition): Make sure to use a non-critical
11063         edge.  Cleanup and remove old bug workarounds.
11064         (bb_postdominates_preds): Remove.
11065         (if_convertible_loop_p_1): Do not compute post-dominators.
11066         (combine_blocks): Do not free post-dominators.
11067         (main_tree_if_conversion): Likewise.
11068         (pass_data_if_conversion): Add TODO_verify_ssa.
11070 2013-08-27  DJ Delorie  <dj@redhat.com>
11072         * config/i386/djgpp.h (ASM_DECLARE_FUNCTION_NAME): New.
11074 2013-08-27  Yufeng Zhang  <yufeng.zhang@arm.com>
11076         * function.c (assign_parm_find_data_types): Set passed_mode and
11077         nominal_mode to the TYPE_MODE of nominal_type for the built
11078         pointer type in case of the struct-pass-by-reference.
11080 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11082         * config/avr/avr-stdint.h (INT16_TYPE): Change default to "int".
11083         (UINT16_TYPE): Change default to "unsigned int".
11085         * config/avr/avr.opt (mfract-convert-truncate): New option.
11086         * config/avr/avr.c (avr_out_fract): Unless TARGET_FRACT_CONV_TRUNC
11087         is set, round negative fractional integers according to n1169
11088         when converting to integer types.
11090 2013-08-26  Jan Hubicka  <jh@suse.cz>
11092         * cgraph.c (cgraph_propagate_frequency): Do not assume that virtual
11093         methods can not be called indirectly when their address is not taken.
11095 2013-08-26  Jan Hubicka  <jh@suse.cz>
11097         * gimple-fold.c (gimple_get_virt_method_for_binfo): Use
11098         ctor_for_folding.
11100 2013-08-26  Jan Hubicka  <jh@suse.cz>
11102         * ipa.c (comdat_can_be_unshared_p_1): C++ constructors and destructors
11103         can be unshared.
11105 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11107         * reload.c (find_valid_class): Allow classes that do not include
11108         FIRST_PSEUDO_REGISTER - 1.
11110 2013-08-26  Jan Hubicka  <jh@suse.cz>
11112         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Fix formatting;
11113         fix edge count/frequency when speculation failed; fix type check
11114         for the direct call.
11116 2013-08-26  Jan Hubicka  <jh@suse.cz>
11118         * ipa-prop.c (ipa_print_node_params): Do not ICE during WPA.
11120 2013-08-26  Jan Hubicka  <jh@suse.cz>
11122         * ipa-inline-transform.c (inline_transform): Be ready for basic block
11123         to be changed by edge redirection.
11125 2013-08-26  Jan Hubicka  <jh@suse.cz>
11127         * cgraph.c (cgraph_speculative_call_info): Fix parameter order and
11128         formating; add sanity check.
11129         (cgraph_resolve_speculation): Add FIXME about scaling profiles.
11130         (cgraph_redirect_edge_call_stmt_to_callee): Fix ICE in debug dump.
11131         * ipa-inline.c (heap_edge_removal_hook): Reset node growth cache.
11132         (resolve_noninline_speculation): Update callee keys, too.
11134 2013-08-26  Jan Hubicka  <jh@suse.cz>
11136         * tree.h (tree_decl_with_vis): Add cxx_constructor, cxx_destructor.
11137         (DECL_CXX_CONSTRUCTOR_P, DECL_CXX_DESTRUCTOR_P): New macros.
11139 2013-08-26  Joern Rennecke  <joern.rennecke@embecosm.com>
11141         * config/i386/i386.c (x86_64_elf_select_section): Put ATTRIBUTE_UNUSED
11142         into proper place.
11144 2013-08-26  Uros Bizjak  <ubizjak@gmail.com>
11146         * config/i386/i386.c (ix86_debug_options): Remove prototype.
11147         (x86_64_elf_select_section): Ditto.
11148         (ix86_handle_tm_regparm_attribute): Remove ATTRIBUTE_UNUSED on used
11149         arguments.
11150         (ix86_pass_by_reference): Ditto.
11151         (output_set_got): Ditto.
11152         (ix86_unary_operator_ok): Ditto.
11153         (ix86_expand_builtin): Ditto.
11155 2013-08-23  Jan Hubicka  <jh@suse.cz>
11157         * cgraph.c (cgraph_turn_edge_to_speculative): Fix debug output.
11159 2013-08-23  Jan Hubicka  <jh@suse.cz>
11161         * tree.h (TYPE_FINAL_P, DECL_FINAL_P): New macros.
11162         (tree_decl_with_vis): Add FINAL field.
11164 2013-08-23  Jeff Law  <law@redhat.com>
11166         * tree-ssa-pre.c (do_regular_insertion): Include the expression in
11167         the debugging dump when the expression is fully redundant.
11169 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11171         * diagnostic.c (diagnostic_set_caret_max_width): Use pp_buffer.
11172         * gimple-pretty-print.c (gimple_dump_bb_buff): Likewise.
11173         * pretty-print.c (pp_formatted_text_data): Likewise.
11174         (pp_write_text_to_stream): Likewise.
11175         (pp_write_text_as_dot_label_to_stream): Likewise.
11176         (pp_append_r): Likewise.
11177         (pp_format): Likewise.
11178         (pp_flush): Likewise.
11179         (pp_clear_output_area): Likewise.
11180         (pp_append_text): Likewise.
11181         (pp_formatted_text): Likewise.
11182         (pp_remaining_character_count_for_line): Likewise.
11183         (pp_newline): Likewise.
11184         (pp_character): Likewise.
11185         (output_buffer::~output_buffer): Define.
11186         (pretty_printer::~pretty_printer): Destruct output buffer.
11187         * pretty-print.h (output_buffer::~output_buffer): Declare.
11188         (pretty_printer::~pretty_printer): Declare virtual.
11190 2013-08-24  Marc Glisse  <marc.glisse@inria.fr>
11192         PR other/57324
11193         * hwint.h (HOST_WIDE_INT_UC, HOST_WIDE_INT_1U, HOST_WIDE_INT_M1,
11194         HOST_WIDE_INT_M1U): New macros.
11195         * fold-const.c (sign_bit_p, build_range_check, fold_unary_loc,
11196         fold_binary_loc, fold_ternary_loc): Use the new macros. Use an
11197         unsigned -1 for lshift.
11198         * cse.c (cse_insn): Likewise.
11199         * double-int.c (rshift_double, lshift_double): Likewise.
11200         * builtins.c (fold_builtin_bitop): Likewise.
11201         * combine.c (force_to_mode): Likewise.
11202         * tree.c (integer_pow2p, tree_log2, tree_floor_log2): Likewise.
11203         * simplify-rtx.c (simplify_const_unary_operation,
11204         simplify_const_binary_operation): Likewise.
11205         * tree-stdarg.c (va_list_counter_bump, va_list_ptr_read,
11206         check_va_list_escapes): Likewise.
11207         * rtlanal.c (nonzero_bits1): Likewise.
11208         * expmed.c (expand_smod_pow2): Likewise.
11209         * tree-ssa-structalias.c (UNKNOWN_OFFSET): Use HOST_WIDE_INT_MIN.
11211 2013-08-23  Jan Hubicka  <jh@suse.cz>
11213         * cgraph.c (cgraph_turn_edge_to_speculative): Mark target node
11214         as having address taken.
11216 2013-08-23  Jan Hubicka  <jh@suse.cz>
11218         * ipa-utils.h (method_class_type): Declare.
11219         * ipa-devirt.c (method_class_type): Export.
11221         * cgraphunit.c (analyze_functions): Do basic devirtualization;
11222         do not walk base classes of anonymous types.
11224 2013-08-23  Kaz Kojima  <kkojima@gcc.gnu.org>
11226         PR rtl-optimization/58220
11227         PR regression/58221
11228         * final.c (reemit_insn_block_notes): Use NEXT_INSN to
11229         handle SEQUENCE insns properly.
11231 2013-08-23  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11233         * pretty-print.h (pp_newline_and_flush): Declare.  Remove macro
11234         definition.
11235         (pp_newline_and_indent): Likewise.
11236         (pp_separate_with): Likewise.
11237         * pretty-print.c (pp_newline_and_flush): Define.
11238         (pp_newline_and_indent): Likewise.
11239         (pp_separate_with): Likewise.
11241 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11243         PR target/58218
11244         * config/i386/x86-64.h (TARGET_SECTION_TYPE_FLAGS): Define.
11245         * config/i386/i386.c (x86_64_elf_section_type_flags): New function.
11247 2013-08-23  Kirill Yukhin  <kirill.yukhin@intel.com>
11249         * config/i386/predicates.md (ext_sse_reg_operand): New.
11250         * config/i386/i386.md (*movti_internal): Use
11251         predicate to determine if EVEX is needed.
11252         (*movsi_internal): Ditto.
11253         (*movdf_internal): Ditto.
11254         (*movsf_internal): Ditto.
11255         * config/i386/mmx.md (*mov<mode>_internal): Ditto.
11257 2013-08-23  Jakub Jelinek  <jakub@redhat.com>
11259         PR tree-optimization/58209
11260         * tree-tailcall.c (process_assignment): Handle POINTER_PLUS_EXPR.
11261         (find_tail_calls): Give up for pointer result types if m is non-NULL.
11262         (adjust_return_value_with_ops): For PLUS_EXPR and pointer result type
11263         emit POINTER_PLUS_EXPR.
11264         (create_tailcall_accumulator): For pointer result type accumulate in
11265         sizetype type.
11267 2013-08-22  Paolo Carlini  <paolo.carlini@oracle.com>
11269         * configure.ac: Add backslashes missing from the last change.
11270         * configure: Regenerate.
11272 2013-08-22  Jan Hubicka  <jh@susue.cz>
11274         * ipa.c (function_and_variable_visibility): First remember function
11275         was global and then make it local.
11277 2013-08-22  Julian Brown  <julian@codesourcery.com>
11279         * configure.ac: Add aarch64 to list of arches which use "nop" in
11280         debug_line test.
11281         * configure: Regenerate.
11283 2013-08-22  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
11285         * config/s390/linux.h (TARGET_LIBC_HAS_FUNCTION): Define as
11286         gnu_libc_has_function.
11287         * config/s390/tpf.h: Likewise.
11289 2013-08-22  Jan Hubicka  <jh@susue.cz>
11291         * timevar.c (validate_phases): Add cast.
11293 2013-08-22  Jan Hubicka  <jh@susue.cz>
11295         * timevar.c (validate_phases): Use size_t for memory.
11296         * timevar.h (struct timevar_time_def): Use size_t for ggc_mem.
11298 2013-08-22  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11300         * pretty-print.h (output_buffer::output_buffer): Declare.
11301         (pretty_printer::pretty_printer): Likewise.
11302         (pp_construct): Remove.
11303         * pretty-print.c (output_buffer::output_buffer): Define.
11304         (pretty_printer::pretty_printer): Rename from pp_construct.  Simplify.
11305         * gimple-pretty-print.c (print_gimple_stmt): Do not call pp_construct.
11306         (print_gimple_expr): Likewise.
11307         (print_gimple_seq): Likewise.
11308         (gimple_dump_bb): Likewise.
11309         * sched-vis.c (dump_value_slim): Likewise.
11310         (dump_insn_slim): Likewise.
11311         (dump_rtl_slim): Likewise.
11312         (str_pattern_slim): Likewise.
11313         * tree-mudflap.c (mf_varname_tree): Likewise.
11314         * graph.c (print_graph_cfg): Likewise.
11315         (start_graph_dump): Likewise.
11316         * tree-pretty-print.c (maybe_init_pretty_print): Likewise.  Use
11317         placement-new.
11318         * diagnostic.c (diagnostic_initialize): Simplify early diagnostic
11319         pretty printer initialization.
11320         * coretypes.h (diagnostic_context): Remove superflous type alias
11321         declaration.
11322         (pretty_printer): Likewise.  Declare directly as a class.
11323         (pretty_print_info): Remove declaration as class.
11324         * asan.c (asan_emit_stack_protection): Remove call to pp_construct
11325         and pp_clear_output_area.
11326         (asan_add_global): Likewise.
11328 2013-08-22  Jan Hubicka  <jh@suse.cz>
11330         * cgraphunit.c (analyze_functions) Use update_type_inheritance_graph.
11331         * ipa-utils.h (update_type_inheritance_graph): Declare.
11332         (possible_polymorphic_call_target_p): Declare.
11333         (possible_polymorphic_call_target_p): New.
11334         * ipa-devirt.c: Update toplevel comments.
11335         (cached_polymorphic_call_targets): Move up.
11336         (odr_type_d): Move ID down.
11337         (polymorphic_type_binfo_p): Update comment.
11338         (odr_hasher::remove): Likewise;
11339         (get_odr_type): Set anonymous_namespace.
11340         (dump_odr_type): Dump it.
11341         (dump_type_inheritance_graph): Do not ICE when there are no ODR types.
11342         (maybe_record_node): Record node in cached_polymorphic_call_targets.
11343         (record_binfo): Add comment.
11344         (free_polymorphic_call_targets_hash): Do not ICE when cache is not
11345         built.
11346         (devirt_node_removal_hook): Do not iCE when cache is freed.
11347         (possible_polymorphic_call_target_p): New predicate.
11348         (update_type_inheritance_graph): New function.
11350 2013-08-22  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11351             Maxim Kuznetsov  <maxim.kuznetsov@intel.com>
11352             Sergey Lega  <sergey.s.lega@intel.com>
11353             Anna Tikhonova  <anna.tikhonova@intel.com>
11354             Ilya Tocar  <ilya.tocar@intel.com>
11355             Andrey Turetskiy  <andrey.turetskiy@intel.com>
11356             Ilya Verbin  <ilya.verbin@intel.com>
11357             Kirill Yukhin  <kirill.yukhin@intel.com>
11358             Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
11360         * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_SET): New.
11361         (OPTION_MASK_ISA_AVX512CD_SET): Ditto.
11362         (OPTION_MASK_ISA_AVX512PF_SET): Ditto.
11363         (OPTION_MASK_ISA_AVX512ER_SET): Ditto.
11364         (OPTION_MASK_ISA_AVX2_UNSET): Update.
11365         (OPTION_MASK_ISA_AVX512F_UNSET): New.
11366         (OPTION_MASK_ISA_AVX512CD_UNSET): Ditto.
11367         (OPTION_MASK_ISA_AVX512PF_UNSET): Ditto.
11368         (OPTION_MASK_ISA_AVX512ER_UNSET): Ditto.
11369         (ix86_handle_option): Handle OPT_mavx512f, OPT_mavx512cd,
11370         OPT_mavx512pf, OPT_mavx512er cases.
11371         * config/i386/constraints.md (v): New constraint.
11372         (Yi, Yj): Replace SSE_REGS with ALL_SSE_REGS.
11373         * config/i386/cpuid.h (bit_AVX512F, bit_AVX512PF, bit_AVX512ER)
11374         (bit_AVX512CD): New.
11375         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
11376         AVX512F, AVX512ER, AVX512PF, AVX512CD features.
11377         * config/i386/i386-c.c (ix86_target_macros_internal):
11378         Conditionally define __AVX512F__, __AVX512ER__, __AVX512CD__,
11379         __AVX512PF__.
11380         * config/i386/i386-modes.def (VECTOR_MODES (INT, 128))
11381         (VECTOR_MODES (FLOAT, 128), INT_MODE (XI, 64)): New modes.
11382         * config/i386/i386.c (regclass_map, dbx_register_map)
11383         (dbx64_register_map, svr4_dbx_register_map): Add new SSE registers.
11384         (gate_insert_vzeroupper): Disable vzeroupper for TARGET_AVX512F.
11385         (ix86_target_string): Define -mavx512f, -mavx512er, -mavx512cd,
11386         -mavx512pf options.
11387         (ix86_option_override_internal): Define PTA_AVX512F, PTA_AVX512ER,
11388         PTA_AVX512PF, PTA_AVX512CD.  Handle -mavx512f, -mavx512er, -mavx512cd,
11389         -mavx512pf options.  Fix formatting.
11390         (ix86_conditional_register_usage): Squash EXT_REX_SSE_REGs for 32-bit
11391         targets.  Squash EVEX_SSE_REGS if AVX512F is disabled.
11392         (ix86_valid_target_attribute_inner_p): Handle -mavx512f, -mavx512er,
11393         -mavx512cd, -mavx512pf options.
11394         (standard_sse_constant_opcode): Add vpternlogd for 512-bit modes.
11395         (print_reg, ix86_print_operand): Handle 'g' to output 512-bit operands.
11396         (ix86_preferred_output_reload_class): Replace SSE_REGS with
11397         ALL_SSE_REGS.
11398         (ix86_hard_regno_mode_ok): Support 512-bit registers.
11399         (ix86_set_reg_reg_cost): Ditto.
11400         (x86_order_regs_for_local_alloc): Ditto.
11401         (MAX_VECT_LEN): Extend to 64-byte.
11402         (ix86_spill_class): Replace SSE_REGS with ALL_SSE_REGS.
11403         * config/i386/i386.h (TARGET_AVX512F, TARGET_AVX512PF)
11404         (TARGET_AVX512ER, TARGET_AVX512CD): New.
11405         (BIGGEST_ALIGNMENT): Extend to 512-bits.
11406         (FIRST_PSEUDO_REGISTER, FIXED_REGISTERS): Add new registers.
11407         (CALL_USED_REGISTERS, REG_ALLOC_ORDER): Likewise.
11408         (VALID_AVX512F_SCALAR_MODE, VALID_AVX512F_REG_MODE): New.
11409         (SSE_REG_MODE_P): Support new modes.
11410         (FIRST_MMX_REG, FIRST_REX_INT_REG, FIRST_REX_SSE_REG): Add comments.
11411         (FIRST_EXT_REX_SSE_REG, LAST_EXT_REX_SSE_REG): New.
11412         (reg_class, REG_CLASS_NAMES): Add EVEX_SSE_REGS, ALL_SSE_REGS.
11413         (SSE_CLASS_P, MAYBE_SSE_CLASS_P): Replace SSE_REGS with ALL_SSE_REGS.
11414         (REG_CLASS_CONTENTS): Add new registers.
11415         (SSE_REGNO_P, SSE_REGNO, HARD_REGNO_RENAME_OK): Support new registers.
11416         (EXT_REX_SSE_REGNO_P): New.
11417         (HI_REGISTER_NAMES): Add new registers.
11418         * config/i386/i386.md: Define constants for new registers.
11419         (mode): Add new 512-bit modes.
11420         (prefix): Support evex prefix.
11421         (isa): Support avx512f, noavx512f, fma_avx512f.
11422         (ssemodesuffix): Add new 512-bit modes.
11423         (movxi): New.
11424         (*movxi_internal_avx512f): Ditto.
11425         (*movdi_internal): Replace constraint "x" with the new constraint "v".
11426         Support MODE_XI.
11427         (*movsi_internal): Likewise.
11428         (*movdf_internal): Likewise.
11429         (*movsf_internal): Likewise.
11430         (*fop_<mode>_comm_sse): Replace constraint "x" with new constraint "v".
11431         (<code><mode>3): Likewise.
11432         * config/i386/i386.opt (mavx512f, mavx512pf, mavx512er, mavx512cd):
11433         New.
11434         * config/i386/mmx.md (*mov<mode>_internal): Replace constraint "x"
11435         with the new constraint "v".
11436         * config/i386/sse.md (*mov<mode>_internal): Support new registers and
11437         modes.
11438         (<sse>_loadu<ssemodesuffix><avxsizesuffix>): Replace constraint "x"
11439         with the new constraint "v".
11440         (<sse2>_loaddqu<avxsizesuffix>): Likewise.
11441         (<sse2>_storedqu<avxsizesuffix>): Likewise.
11442         (*<plusminus_insn><mode>3): Likewise.
11443         (<sse>_vm<plusminus_insn><mode>3): Likewise.
11444         (*mul<mode>3): Likewise.
11445         (<sse>_vmmul<mode>3): Likewise.
11446         (<sse>_div<mode>3): Likewise.
11447         (<sse>_vmdiv<mode>3): Likewise.
11448         (<sse>_sqrt<mode>2): Likewise.
11449         (<sse>_vmsqrt<mode>2): Likewise.
11450         (*<code><mode>3_finite): Likewise.
11451         (*<code><mode>3) <smaxmin>: Likewise.
11452         (<sse>_vm<code><mode>3): Likewise.
11453         (*<code><mode>3) <any_logic>: Likewise.
11454         (*fma_fmadd_<mode>): Likewise.
11455         (*fma_fmsub_<mode>): Likewise.
11456         (*fma_fnmadd_<mode>): Likewise.
11457         (*fma_fnmsub_<mode>): Likewise.
11458         (*fma_fmaddsub_<mode>): Likewise.
11459         (*fma_fmsubadd_<mode>): Likewise.
11460         (*fmai_fmadd_<mode>): Likewise.
11461         (*fmai_fmsub_<mode>): Likewise.
11462         (*fmai_fnmadd_<mode>): Likewise.
11463         (*fmai_fnmsub_<mode>): Likewise.
11464         (sse_cvtsi2ss): Likewise.
11465         (sse_cvtsi2ssq): Likewise.
11466         (sse_cvtss2si): Likewise.
11467         (sse_cvtss2si_2): Likewise.
11468         (sse_cvtss2siq): Likewise.
11469         (sse_cvtss2siq_2): Likewise.
11470         (sse_cvttss2si): Likewise.
11471         (sse_cvtss2siq_2): Likewise.
11472         (float<sseintvecmodelower><mode>2): Likewise.
11473         (sse2_cvtsd2si_2): Likewise.
11474         (sse2_cvtsd2siq_2): Likewise.
11475         (*<plusminus_insn><mode>3): Likewise.
11476         (*<sse2_avx2>_<plusminus_insn><mode>3): Likewise.
11477         (*<sse4_1_avx2>_mul<mode>3): Likewise.
11478         (ashr<mode>3): Likewise.
11479         (<shift_insn><mode>3): Likewise.
11480         (avx2_<code><mode>3): Likewise.
11481         (*avx2_<code><mode>3): Likewise.
11482         (*andnot<mode>3): Likewise.
11483         (*<code><mode>3) <any_logic>: Likewise.
11484         (abs<mode>2): Likewise.
11485         (avx2_permvar<mode>): Likewise.
11486         (avx2_perm<mode>_1): Likewise.
11487         (*avx_vpermilp<mode>): Likewise.
11488         (avx_vpermilvar<mode>3): Likewise.
11489         (avx2_ashrv<mode>): Likewise.
11490         (avx2_<shift_insn>v<mode>): Likewise.
11491         * doc/invoke.texi: Document -mavx512f, -mavx512pf, -mavx512er,
11492         -mavx512cd.
11493         * doc/rtl.texi: Document XImode.
11495 2013-08-21  Jeff Law  <law@redhat.com>
11497         * tree-flow.h (register_jump_thread): Pass vector of edges
11498         instead of each important edge.
11499         * tree-ssa-threadedge.c (thread_across_edge): Build the jump
11500         thread path into a vector and pass that to register_jump_thread.
11501         * tree-ssa-threadupdate.c (register_jump_thread): Conver the
11502         passed in edge vector to the current 3-edge form.
11504         Revert:
11505         2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11507         * dce.c (fini_dce): Call df_analyze again just in case
11508         delete_unmarked_insns removed anything.
11510 2013-08-21  Joern Rennecke  <joern.rennecke@embecosm.com>
11512         * reload.h (struct reg_equivs): Rename to ..
11513         (struct reg_equivs_s): .. this.
11515 2013-08-20  Martin Liska  <marxin.liska@gmail.com>
11517         * ipa.c (ipa_profile_read_summary): Fix buffer overflow.
11519 2013-08-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
11521         * config/sol2-10.h (TARGET_LIBC_HAS_FUNCTION): Don't nest comment.
11523 2013-08-21  Jeff Law  <law@redhat.com>
11525         * tree-vrp.c (simplify_stmt_for_jump_threading): Try to
11526         simplify assignments too.  If the RHS collapses to a singleton
11527         range, then return the value for the range.
11529 2013-08-21  Kirill Yukhin  <kirill.yukhin@intel.com>
11531         * config/i386/sse.md (V16): Rename to...
11532         (VMOVE): this.
11533         (mov<mode>): Update iterator name.
11534         (*mov<mode>_internal): Ditto.
11535         (push<mode>1): Ditto.
11536         (movmisalign<mode>): Ditto.
11538 2013-08-20  Jan Hubicka  <jh@suse.cz>
11540         PR bootstrap/58186
11541         * cgraph.c (cgraph_add_edge_to_call_site_hash): Overwrite hash
11542         entry for direct edges.
11543         (cgraph_turn_edge_to_speculative): Fix setting of can_throw_external.
11545 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11547         Revert my last two changes, r201865 and r201864:
11549         Revert r201865:
11550         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11552         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11553         instances can own GC refs.
11555         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11556         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11557         (gcc::context::gt_pch_nx): Likewise.
11558         (gcc::context::gt_pch_nx):  Likewise.
11559         * ggc.h (gt_ggc_mx <T>): New.
11560         (gt_pch_nx_with_op <T>): New.
11561         (gt_pch_nx <T>): New.
11562         * passes.c (opt_pass::gt_ggc_mx): New.
11563         (opt_pass::gt_pch_nx): New.
11564         (opt_pass::gt_pch_nx_with_op): New.
11565         (pass_manager::gt_ggc_mx): New.
11566         (pass_manager::gt_pch_nx): New.
11567         (pass_manager::gt_pch_nx_with_op): New.
11568         (pass_manager::operator new): Use
11569         ggc_internal_cleared_alloc_stat rather than xcalloc.
11570         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11571         (pass_manager::gt_ggc_mx): New.
11572         (pass_manager::gt_pch_nx): New.
11573         (pass_manager::gt_pch_nx_with_op): New.
11574         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11575         (opt_pass::operator new): New.
11576         (opt_pass::gt_ggc_mx): New.
11577         (opt_pass::gt_pch_nx): New.
11578         (opt_pass::gt_pch_nx_with_op): New.
11580         Revert r201864:
11581         2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11583         * Makefile.in (GTFILES): Add context.h.
11584         * context.c (gcc::context::operator new): New.
11585         (gcc::context::gt_ggc_mx): New.
11586         (gcc::context::gt_pch_nx): New.
11587         (gcc::context::gt_pch_nx): New.
11588         * context.h (gcc::context): Add GTY((user)) marking.
11589         (gcc::context::operator new): New.
11590         (gcc::context::gt_ggc_mx): New.
11591         (gcc::context::gt_pch_nx): New.
11592         (gcc::context::gt_pch_nx): New.
11593         (g): Add GTY marking.
11594         (gt_ggc_mx (gcc::context *)): New.
11595         (gt_pch_nx (gcc::context *)): New.
11596         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11597         void *cookie)): New.
11598         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11600 2013-08-20  Alexey Makhalov  <makhaloff@gmail.com>
11602         * dce.c (fini_dce): Call df_analyze again just in case
11603         delete_unmarked_insns removed anything.
11605 2013-08-20  Teresa Johnson  <tejohnson@google.com>
11607         PR rtl-optimizations/57451
11608         * final.c (reemit_insn_block_notes): Prevent lexical blocks
11609         from crossing split section boundaries.
11611 2013-08-20  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
11613         * config/arm/linux-elf.h (MULTILIB_DEFAULTS): Remove definition.
11614         * config/arm/t-linux-eabi (MULTILIB_OPTIONS): Document association
11615         with MULTLIB_DEFAULTS.
11617 2013-08-20  Nick Clifton  <nickc@redhat.com>
11619         * target.def (narrow_volatile_bitfield): Note that the default
11620         value is false, not !TARGET_STRICT_ALIGN.
11621         * doc/tm.texi: Regenerate.
11623 2013-08-20  Pavel Chupin  <pavel.v.chupin@intel.com>
11625         Fix LIB_SPEC for systems without libpthread.
11627         * config/gnu-user.h: Introduce GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC.
11628         * config/arm/linux-eabi.h: Use GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC
11629         for Android.
11630         * config/i386/linux-common.h: Likewise.
11631         * config/mips/linux-common.h: Likewise.
11633 2013-08-20  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
11635         * tree-ssa-ccp.c (get_default_value): Remove redundant condition
11636         checks.
11638 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11640         Make opt_pass and gcc::pass_manager be GC-managed, so that pass
11641         instances can own GC refs.
11643         * Makefile.in (GTFILES): Add pass_manager.h and tree-pass.h.
11644         * context.c (gcc::context::gt_ggc_mx): Traverse passes_.
11645         (gcc::context::gt_pch_nx): Likewise.
11646         (gcc::context::gt_pch_nx):  Likewise.
11647         * ggc.h (gt_ggc_mx <T>): New.
11648         (gt_pch_nx_with_op <T>): New.
11649         (gt_pch_nx <T>): New.
11650         * passes.c (opt_pass::gt_ggc_mx): New.
11651         (opt_pass::gt_pch_nx): New.
11652         (opt_pass::gt_pch_nx_with_op): New.
11653         (pass_manager::gt_ggc_mx): New.
11654         (pass_manager::gt_pch_nx): New.
11655         (pass_manager::gt_pch_nx_with_op): New.
11656         (pass_manager::operator new): Use
11657         ggc_internal_cleared_alloc_stat rather than xcalloc.
11658         * pass_manager.h (class pass_manager): Add GTY((user)) marking.
11659         (pass_manager::gt_ggc_mx): New.
11660         (pass_manager::gt_pch_nx): New.
11661         (pass_manager::gt_pch_nx_with_op): New.
11662         * tree-pass.h (class opt_pass): Add GTY((user)) marking.
11663         (opt_pass::operator new): New.
11664         (opt_pass::gt_ggc_mx): New.
11665         (opt_pass::gt_pch_nx): New.
11666         (opt_pass::gt_pch_nx_with_op): New.
11668 2013-08-20  David Malcolm  <dmalcolm@redhat.com>
11670         * Makefile.in (GTFILES): Add context.h.
11671         * context.c (gcc::context::operator new): New.
11672         (gcc::context::gt_ggc_mx): New.
11673         (gcc::context::gt_pch_nx): New.
11674         (gcc::context::gt_pch_nx): New.
11675         * context.h (gcc::context): Add GTY((user)) marking.
11676         (gcc::context::operator new): New.
11677         (gcc::context::gt_ggc_mx): New.
11678         (gcc::context::gt_pch_nx): New.
11679         (gcc::context::gt_pch_nx): New.
11680         (g): Add GTY marking.
11681         (gt_ggc_mx (gcc::context *)): New.
11682         (gt_pch_nx (gcc::context *)): New.
11683         (gt_pch_nx (gcc::context *ctxt, gt_pointer_operator op,
11684         void *cookie)): New.
11685         * gengtype.c (open_base_files) <ifiles>: Add context.h.
11687 2013-08-20  Alan Modra  <amodra@gmail.com>
11689         PR target/57865
11690         * config/rs6000/rs6000.c (rs6000_emit_prologue): Correct ool_adjust.
11691         (rs6000_emit_epilogue): Likewise.
11693 2013-08-19  Dehao Chen  <dehao@google.com>
11695         * value-prof.c (gimple_ic): Fix the bug of adding EH edge.
11697 2013-08-19  Peter Bergner  <bergner@vnet.ibm.com>
11698             Jakub Jelinek  <jakub@redhat.com>
11700         * builtins.def (BUILT_IN_FABSD32): New DFP ABS builtin.
11701         (BUILT_IN_FABSD64): Likewise.
11702         (BUILT_IN_FABSD128): Likewise.
11703         * builtins.c (expand_builtin): Add support for new DFP ABS builtins.
11704         (fold_builtin_1): Likewise.
11705         * config/rs6000/dfp.md (*negtd2_fpr): Handle non-overlapping
11706         destination and source operands.
11707         (*abstd2_fpr): Likewise.
11708         (*nabstd2_fpr): Likewise.
11710 2013-08-19  Richard Sandiford  <rdsandiford@googlemail.com>
11712         * config/mips/mips.c (mips_adjust_insn_length): Add checks for
11713         JUMP_P and INSN_P.
11715 2013-08-19  Aldy Hernandez  <aldyh@redhat.com>
11717         * doc/invoke.texi (-fcilkplus): Clarify that implementation is
11718         incomplete.
11720 2013-08-19  Alexander Ivchenko  <alexander.ivchenko@intel.com>
11722         * target.def (TARGET_LIBC_HAS_FUNCTION): New target hook.
11723         * builtins.c (default_libc_has_function): New.
11724         (gnu_libc_has_function): Ditto.
11725         (no_c99_libc_has_function): Ditto.
11726         (expand_builtin_cexpi): Using new target hook TARGET_LIBC_HAS_FUNCTION
11727         instead of TARGET_HAS_SINCOS and TARGET_C99_FUNCTIONS.
11728         (fold_builtin_sincos): Likewise.
11729         (fold_builtin_cexp): Likewise.
11730         * builtins.def (DEF_C94_BUILTIN): Likewise.
11731         (DEF_C99_BUILTIN): Likewise.
11732         (DEF_C99_C90RES_BUILTIN): Likewise.
11733         (DEF_C99_COMPL_BUILTIN): New define. Change all complex c99 builtin
11734         definitions to using this define.
11735         * config/darwin-protos.h (darwin_libc_has_function): New.
11736         * config/darwin.c (darwin_libc_has_function): Ditto.
11737         * config/alpha/linux.h: Remove TARGET_C99_FUNCTIONS and
11738         TARGET_HAS_SINCOS. Redefine TARGET_LIBC_HAS_FUNCTION.
11739         * config/darwin.h: Ditto.
11740         * config/elfos.h: Ditto.
11741         * config/freebsd.h: Ditto.
11742         * config/i386/cygming.h: Ditto.
11743         * config/i386/djgpp.h: Ditto.
11744         * config/i386/i386-interix.h: Ditto.
11745         * config/microblaze/microblaze.h: Ditto.
11746         * config/mmix/mmix.h: Ditto.
11747         * config/gnu-user.h: Ditto.
11748         * config/ia64/hpux.h: Ditto.
11749         * config/pa/pa-hpux.h: Ditto.
11750         * config/pdp11/pdp11.h: Ditto.
11751         * config/picochip/picochip.h: Ditto.
11752         * config/linux.h: Ditto.
11753         * config/netbsd.h: Ditto.
11754         * config/openbsd.h: Ditto.
11755         * config/rs6000/aix43.h: Ditto.
11756         * config/rs6000/aix51.h: Ditto.
11757         * config/rs6000/aix52.h: Ditto.
11758         * config/rs6000/aix53.h: Ditto.
11759         * config/rs6000/aix61.h: Ditto.
11760         * config/rs6000/darwin.h: Ditto.
11761         * config/rs6000/linux.h: Ditto.
11762         * config/rs6000/linux64.h: Ditto.
11763         * config/s390/tpf.h: Ditto.
11764         * config/sol2-10.h: Ditto.
11765         * config/sol2.h: Ditto.
11766         * config/vms/vms.h: Ditto.
11767         * config/vxworks.h: Ditto.
11768         * config/linux-android.c (linux_android_libc_has_function):
11769         New linux-specific implementation of TARGET_LIBC_HAS_FUNCTION.
11770         * config/linux-protos.h (linux_android_libc_has_function):
11771         New declaration.
11772         * config/i386/i386.c (ix86_libc_has_function): New.
11773         * config/i386/i386-protos.h
11774         (ix86_libc_has_function): New declaration.
11775         * config/i386/i386.md
11776         ("isinfxf2"): Change condition for TARGET_LIBC_HAS_FUNCTION.
11777         ("isinf<mode>2): Likewise.
11778         * convert.c (convert_to_integer): Using new target hook
11779         TARGET_LIBC_HAS_FUNCTION istead of TARGET_HAS_SINCOS and
11780         TARGET_C99_FUNCTIONS.
11781         * fortran/f95-lang.c (gfc_init_builtin_functions): Ditto.
11782         * tree-ssa-math-opts.c (execute_cse_sincos): Ditto.
11783         * coretypes.h (function_class): New enum for different
11784         classes of functions.
11785         * defaults.h: Remove TARGET_C99_FUNCTIONS and TARGET_HAS_SINCOS.
11786         * doc/tm.texi.in (TARGET_C99_FUNCTIONS): Remove documentation.
11787         (TARGET_HAS_SINCOS): Likewise.
11788         (TARGET_LIBC_HAS_FUNCTION): New.
11789         * doc/tm.texi: Regenerated.
11790         * targhooks.h (default_libc_has_function): New declaration.
11791         (no_c99_libc_has_function): Ditto.
11792         (gnu_libc_has_function): Ditto.
11793         * system.h: Add the poisoning of TARGET_C99_FUNCTIONS
11794         and TARGET_HAS_SINCOS.
11796 2013-08-18  Jan Hubicka  <jh@suse.cz>
11798         * Makeifle-in (ipa-devirt.o): New.
11799         (GTFILES): Add ipa-utils.h and ipa-devirt.c
11800         * cgraphunit.c (decide_is_symbol_needed): Do not care about virtuals.
11801         (analyze_functions): Look into possible targets of polymorphic call.
11802         * dumpfile.c (dump_files): Add type-inheritance dump.
11803         * dumpfile.h (TDI_inheritance): New.
11804         * ipa-devirt.c: New file.
11805         * ipa-utils.h (odr_type_d): Forward declare.
11806         (odr_type): New type.
11807         (build_type_inheritance_graph): Declare.
11808         (possible_polymorphic_call_targets): Declare and introduce inline
11809         variant when only edge is pased.
11810         (dump_possible_polymorphic_call_targets): Likewise.
11811         * timevar.def (TV_IPA_INHERITANCE, TV_IPA_VIRTUAL_CALL): New.
11812         * tree.c (type_in_anonymous_namespace_p): Break out from ...
11813         (types_same_for_odr): ... here.
11814         * tree.h (type_in_anonymous_namespace_p): Declare.
11816 2013-08-18  Jakub Jelinek  <jakub@redhat.com>
11818         PR tree-optimization/58006
11819         * tree-parloops.c (take_address_of): Don't ICE if get_name
11820         returns NULL.
11821         (eliminate_local_variables_stmt): Remove clobber stmts.
11823 2013-08-18  Eric Botcazou  <ebotcazou@adacore.com>
11825         * cgraphunit.c (handle_alias_pairs): Reset the alias flag after the
11826         error message is issued for an alias to undefined symbol.
11828 2013-08-18  Jan Hubicka  <jh@suse.cz>
11830         * cgraph.c (cgraph_create_indirect_edge): Discover
11831         polymorphic calls and record basic info into indirect_info.
11832         * gimple-fold.c (gimple_fold_call): When doing BINFO based
11833         devirtualization, ignore objc function calls.
11834         * ipa-cp.c (initialize_node_lattices): Be ready for polymorphic
11835         call with no parm index info.
11836         * ipa-prop.c (ipa_analyze_call_uses): Likewise.
11837         * tree.c (virtual_method_call_p): New function.
11838         * tree.h (virtual_method_call_p): Declare.
11840 2013-08-16  Jan Hubicka  <jh@suse.cz>
11842         PR middle-end/58179
11843         * tree.c (obj_type_ref_class): Do not ICE on non-method calls.
11845 2013-08-16  David Edelsohn  <dje.gcc@gmail.com>
11847         * config/rs6000/rs6000.md (rs6000_get_timebase_ppc32): Add length
11848         attribute.
11850 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11852         * gengtype.c (type_for_name): Add special-case support for
11853         locating types within the "gcc::" namespace.
11854         (open_base_files): Emit a "using namespace gcc" directive.
11856 2013-08-16  Michael Meissner  <meissner@linux.vnet.ibm.com>
11858         PR target/58160
11859         * config/rs6000/predicates.md (fusion_gpr_mem_load): Allow the
11860         memory rtx to contain ZERO_EXTEND and SIGN_EXTEND.
11862         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): Pass operands
11863         array instead of each individual operand as a separate argument.
11864         (emit_fusion_gpr_load): Likewise.
11865         (expand_fusion_gpr_load): Add new function declaration.
11867         * config/rs6000/rs6000.c (fusion_gpr_load_p): Change the calling
11868         signature to have the operands passed as an array, instead of as
11869         separate arguments.  Allow ZERO_EXTEND to be in the memory
11870         address, and also SIGN_EXTEND if -mpower8-fusion-sign.  Do not
11871         depend on the register live/dead flags when peepholes are run.
11872         (expand_fusion_gpr_load): New function to be called from the
11873         peephole2 pass, to change the register that addis sets to be the
11874         target register.
11875         (emit_fusion_gpr_load): Change the calling signature to have the
11876         operands passed as an array, instead of as separate arguments.
11877         Allow ZERO_EXTEND to be in the memory address, and also
11878         SIGN_EXTEND if -mpower8-fusion-sign.
11880         * config/rs6000/rs6000.md (UNSPEC_FUSION_GPR): Delete unused
11881         unspec enumeration.
11882         (power8 fusion peephole/peephole2): Rework the fusion peepholes to
11883         adjust the register addis loads up in the peephole2 pass.  Do not
11884         depend on the register live/dead state when the peephole pass is done.
11886 2013-08-16  David Malcolm  <dmalcolm@redhat.com>
11888         * gengtype.c (create_user_defined_type): Ensure that the kind
11889         is set to TYPE_USER_STRUCT, fixing a bug seen when an incomplete
11890         declaration is seen before the GTY((user)) marking.
11892 2013-08-16  Bernd Edlinger  <bernd.edlinger@hotmail.de>
11894         PR target/58105
11895         * config/i386/i386.c (make_resolver_func): Set DECL_UNINLINABLE.
11897 2013-08-16  Jan Hubicka  <jh@suse.cz>
11899         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Add new
11900         arugment expected_type.
11901         (gimple_fold_call): Use it.
11902         * gimple.h (gimple_extract_devirt_binfo_from_cst): Update prototype.
11903         * ipa-cp.c (ipa_get_indirect_edge_target_1): Update.
11904         * ipa-prop.c (ipa_analyze_virtual_call_uses): Use obj_type_ref_class.
11905         (try_make_edge_direct_virtual_call): Likewise.
11906         * tree.c (obj_type_ref_class): New.
11907         * tree.h (obj_type_ref_class): Use it.
11909 2013-08-16  Gabriel Dos Reis  <gdr@integrable-solutions.net>
11911         * sched-vis.c (rtl_slim_pp_initialized): Remove.
11912         (rtl_slim_pp): Likewise.
11913         (init_rtl_slim_pretty_print): Likewise.
11914         (dump_value_slim):  Don't call it.  Use local pretty printer.
11915         (dump_insn_slim): Likewise.
11916         (dump_rtl_slim): Likewise.
11917         (str_pattern_slim): Likewise.
11918         * tree-mudflap.c (mf_varname_tree): Use local pretty printer.
11919         Simplify.
11921 2013-08-16  Jakub Jelinek  <jakub@redhat.com>
11923         PR tree-optimization/58164
11924         * gimple.c (walk_stmt_load_store_addr_ops): For visit_addr
11925         walk gimple_goto_dest of GIMPLE_GOTO.
11927         PR tree-optimization/58165
11928         * tree-call-cdce.c (shrink_wrap_one_built_in_call): If
11929         bi_call must be the last stmt in a bb, don't split_block, instead
11930         use fallthru edge from it and give up if there is none.
11931         Release conds vector when returning early.
11933 2013-08-14  Xinliang David Li  <davidxl@google.com>
11935         * config/i386/i386.c (ix86_option_override_internal):
11936         Remove unused variable and field.
11938 2013-08-14  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
11940         PR target/57949
11941         * doc/invoke.texi: Add documentation of mcompat-align-parm option.
11942         * config/rs6000/rs6000.opt: Add mcompat-align-parm option.
11943         * config/rs6000/rs6000.c (rs6000_function_arg_boundary): For AIX
11944         and Linux, correct BLKmode alignment when 128-bit alignment is
11945         required and compatibility flag is not set.
11946         (rs6000_gimplify_va_arg): For AIX and Linux, honor specified alignment
11947         for zero-size arguments when compatibility flag is not set.
11949 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11951         PR tree-optimization/58145
11952         * tree-sra.c (build_ref_for_offset): If prev_base has
11953         TREE_THIS_VOLATILE or TREE_SIDE_EFFECTS, propagate it to MEM_REF.
11955 2013-08-14  Xinliang David Li  <davidxl@google.com>
11957         * config/i386/i386.c (ix86_option_override_internal):
11958         Fix uninitialized variable error.
11960 2013-08-14  Xinliang David Li  <davidxl@google.com>
11962         * config/i386/i386.opt: Define two new options.
11963         * config/i386/x86-tune.def: Add arch selector field in macros.
11964         * config/i386/i386.h: Adjust macro definition.
11965         * config/i386/i386.c (ix86_option_override_internal):
11966         Refactor the code.
11967         (parse_mtune_ctrl_str): New function.
11968         (set_ix86_tune_features): New function.
11969         (ix86_function_specific_restore): Call the new helper function.
11971 2013-08-14  Andrey Belevantsev  <abel@ispras.ru>
11973         PR rtl-optimization/57662
11974         * sel-sched.c (code_motion_process_successors): When the current insn
11975         is removed after the recursive traversal, break from the loop.
11976         Add comments and debug printouts.
11978 2013-08-14  Jakub Jelinek  <jakub@redhat.com>
11979             Alexandre Oliva  <aoliva@redhat.com>
11981         PR target/58067
11982         * config/i386/i386.c (ix86_delegitimize_address): For CM_MEDIUM_PIC
11983         and CM_LARGE_PIC ix86_cmodel fall thru into the -m32 code, handle
11984         there also UNSPEC_PLTOFF.
11986 2013-08-14  Marek Polacek  <polacek@redhat.com>
11988         * ipa-inline-analysis.c (add_clause): Avoid shifting integer
11989         NUM_CONDITIONS bit positions.
11991 2013-08-13  Cary Coutant  <ccoutant@google.com>
11993         * dwarf2out.c (CHECKSUM_BLOCK): New macro.
11994         (attr_checksum): Hash vector contents instead of pointer.
11995         (attr_checksum_ordered): Likewise.
11997 2013-08-13  Uros Bizjak  <ubizjak@gmail.com>
11999         * config/i386/sse.md (*sse2_maskmovdqu): Emit addr32 prefix
12000         when Pmode != word_mode.  Add length_address attribute.
12001         (sse3_monitor_<mode>): Merge from sse3_monitor and
12002         sse3_monitor64_<mode> insn patterns.  Emit addr32 prefix when
12003         Pmode != word_mode.  Update insn length attribute.
12004         * config/i386/i386.c (ix86_option_override_internal): Update
12005         ix86_gen_monitor selection for merged sse3_monitor insn.
12007 2013-08-13  Julian Brown  <julian@codesourcery.com>
12009         * config/rs6000/rs6000.c (rs6000_legitimize_reload_address): Don't
12010         perform invalid legitimization on greater-than-word-size modes for
12011         TARGET_E500_DOUBLE.
12013 2013-08-13  Vladimir Makarov  <vmakarov@redhat.com>
12015         * ira.c (setup_class_translate_array): Use aclass instead of cl
12016         for classes not fully covered by allocno classes.
12018 2013-08-13  Jakub Jelinek  <jakub@redhat.com>
12020         PR tree-optimization/57661
12021         * tree-inline.h (struct copy_body_data): Add blocks_to_copy field.
12022         * tree-inline.c (tree_function_versioning): Initialize it.
12023         (remap_gimple_stmt): Return GIMPLE_NOP for MEM_REF lhs clobber stmts
12024         if id->blocks_to_copy and MEM_REF's SSA_NAME is defined in a block
12025         that is not being copied.
12027         PR sanitizer/56417
12028         * asan.c (instrument_strlen_call): Fix typo in comment.
12029         Use char * type even for the lhs of POINTER_PLUS_EXPR.
12031 2013-08-13  Steve Ellcey  <sellcey@mips.com>
12033         * config/mips/mips.md (prefetch): Use lw instead of ld on
12034         loongson in 32bit mode.
12036 2013-08-13  Nick Clifton  <nickc@redhat.com>
12038         * config.gcc: (avr-linux): Allow for tmake_file not being empty.
12040 2013-08-13  Jan Hubicka  <jh@suse.cz>
12042         * cgraph.c (cgraph_turn_edge_to_speculative): Return newly
12043         introduced edge; fix typo in sanity check.
12044         (cgraph_resolve_speculation): Export; improve diagnostic.
12045         (cgraph_redirect_edge_call_stmt_to_callee): Better diagnostic; cancel
12046         speculation at type mismatch.
12047         * cgraph.h (cgraph_turn_edge_to_speculative): Update.
12048         (cgraph_resolve_speculation): Declare.
12049         (symtab_can_be_discarded): New function.
12050         * value-prof.c (gimple_ic_transform): Remove actual transform code.
12051         * ipa-inline-transform.c (speculation_removed): New global var.
12052         (clone_inlined_nodes): See if speculation can be removed.
12053         (inline_call): If speculations was removed, we growths may not match.
12054         * ipa-inline.c (can_inline_edge_p): Add DISREGARD_LIMITS parameter.
12055         (speculation_useful_p): New function.
12056         (resolve_noninline_speculation): New function.
12057         (inline_small_functions): Resolve useless speculations.
12058         * ipa-inline.h (speculation_useful_p): Declare
12059         * ipa.c (can_replace_by_local_alias): Simplify.
12060         (ipa_profile): Produce speculative calls in non-lto, too;
12061         add simple cost model; produce local aliases.
12063 2013-08-13  David Malcolm  <dmalcolm@redhat.com>
12065         * config/i386/t-i386 (i386.o): Rename stray PIPELINE_H to
12066         PASS_MANAGER_H.
12068 2013-08-12  Paolo Carlini  <paolo.carlini@oracle.com>
12070         * config/i386/i386.c (ix86_function_versions): Use error + inform.
12072 2013-08-12  Uros Bizjak  <ubizjak@gmail.com>
12074         * config/i386/i386.md (floatunssi<mode>2 expand): Use MODEF mode
12075         iterator instead of X87MODEF.
12077 2013-08-12  Perez Read  <netfirewall@gmail.com>
12079         PR target/58132
12080         * config/i386/i386.md (*movabs<mode>_1):  Add <ptrsize> PTR before
12081         operand 0 for intel asm alternative.
12082         (*movabs<mode>_2): Ditto for operand 1.
12084 2013-08-12  James Greenhalgh  <james.greenhalgh@arm.com>
12086         * config/aarch64/arm_none.h
12087         (vdup<bhsd>_lane_<su><8,16,32,64>): Fix macro call.
12089 2013-08-12  Nick Clifton  <nickc@redhat.com>
12091         * config.gcc (m32r-linux): Allow for tmake_file not being empty.
12093 2013-08-12  Yuri Rumyantsev  <ysrumyan@gmail.com>
12095         * config/i386/i386.md (floatunssi<mode>2 expand): Add new
12096         expand for QI/HImode operand to produce more effictive code for
12097         unsigned char(short) --> float(double) conversion.
12099 2013-08-12  Alexander Monakov  <amonakov@ispras.ru>
12101         * doc/invoke.texi: Mention that -ftls-model does not force the final
12102         model.
12104 2013-08-12  Marek Polacek  <polacek@redhat.com>
12105             Marc Glisse  <marc.glisse@inria.fr>
12107         PR tree-optimization/57980
12108         * tree-tailcall.c (process_assignment): Call build_minus_one_cst
12109         when creating -1 constant.
12111 2013-08-10  Jan Hubicka  <jh@suse.cz>
12113         Workaround binutils PR14342.
12114         * tree-profile.c (init_ic_make_global_vars): Add LTO path.
12115         (gimple_init_edge_profiler): Likewise.
12116         (gimple_gen_ic_func_profiler): Likewise.
12118 2013-08-09  Jan Hubicka  <jh@suse.cz>
12120         * cgraph.c (cgraph_create_edge_1): Clear speculative flag.
12122 2013-08-09  Xinliang David Li  <davidxl@google.com>
12124         * config/i386/stringop.def: New file.
12125         * config/i386/stringop.opt: New file.
12126         * config/i386/i386-opts.h: Include stringopt.def.
12127         * config/i386/i386.opt: Include stringopt.opt.
12128         * config/i386/i386.c (ix86_option_override_internal):
12129         Override default size based stringop inline strategies with options.
12130         * config/i386/i386.c (ix86_parse_stringop_strategy_string):
12131         New function.
12133 2013-08-09  Jan Hubicka  <jh@suse.cz>
12135         * ipa-ref.c (ipa_clear_stmts_in_references): Clear lto_stmt_uid, too.
12137 2013-08-09  Jan Hubicka  <jh@suse.cz>
12139         * cgraph.c (cgraph_resolve_speculation): Cut frequency to
12140         CGRAPH_FREQ_MAX.
12141         (dump_cgraph_node): Dump profile-id.
12142         * cgraph.h (cgraph_indirect_call_info): Add common_target_id
12143         and common_target_probability.
12144         * lto-cgraph.c (lto_output_edge): Stream common targets.
12145         (lto_output_node): Stream profile ids.
12146         (input_node): Stream profile ids.
12147         (input_edge): Stream common targets.
12148         * lto-streamer-in.c (fixup_call_stmt_edges_1): Fix formatting.
12149         * ipa.c: Include value-prof.h
12150         (ipa_profile_generate_summary): Turn indirect call statement histograms
12151         into common targets.
12152         (ipa_profile): Turn common targets into speculative edges.
12154 2013-08-09  Jan Hubicka  <jh@suse.cz>
12156         * cgraph.h (cgraph_node): Add profile_id.
12157         * value-prof.c (cgraph_node_map): Turn into pointer_map.
12158         (init_node_map): Rewrite to handle hashes increas of incremental IDs.
12159         (del_node_map): Update.
12160         (find_func_by_funcdef_no): Replace by ...
12161         (find_func_by_profile_id): ... this one.
12162         (gimple_ic_transform): Do not remove useful histograms when
12163         speculation is not done; dump info when indirect call removal
12164         can happen at LTO.
12165         * value-prof.h (find_func_by_profile_id, gimple_ic): Declare.
12166         * gcov-io.h (__gcov_indirect_call_profiler): Replace by ...
12167         (__gcov_indirect_call_profiler_v2): .. this one.
12168         * profile.h (init_node_map): Update.
12169         * coverage.c (coverage_compute_profile_id): New function.
12170         * coverage.h (coverage_compute_profile_id): Declare.
12171         * tree-profile.c (init_ic_make_global_vars): Make
12172         __gcov_indirect_call_callee and  __gcov_indirect_call_counters global.
12173         (gimple_init_edge_profiler): Update prototype of
12174         __gcov_indirect_call_profiler.
12175         (gimple_gen_ic_func_profiler): Simplify.
12176         (tree_profiling): Use init_node_map
12178 2013-08-09  Jan Hubicka  <jh@suse.cz>
12180         * cgraphbuild.c (cgraph_rebuild_references): Rebuild only
12181         non-speculative refs.
12182         * cgraph.c (cgraph_update_edge_in_call_site_hash): New function.
12183         (cgraph_add_edge_to_call_site_hash): Deal with speculative calls.
12184         (cgraph_set_call_stmt): Likewise.
12185         (cgraph_create_edge_1): Fix release checking compilatoin;
12186         clear lto_stmt_uid.
12187         (cgraph_free_edge): Free indirect info.
12188         (cgraph_turn_edge_to_speculative): New function.
12189         (cgraph_speculative_call_info): New function.
12190         (cgraph_make_edge_direct): Return direct edge; handle speculation.
12191         (cgraph_redirect_edge_call_stmt_to_callee): Expand speculative edges.
12192         (dump_cgraph_node): Dump speculation.
12193         (verify_edge_count_and_frequency): Accept speculative edges.
12194         (verify_edge_corresponds_to_fndecl): Handle partitioned cgraph.
12195         (verify_cgraph_node): Handle speculation.
12196         * cgraph.h (cgraph_edge): Add SPECULATIVE flag.
12197         (cgraph_set_call_stmt): Update prototype.
12198         (cgraph_make_edge_direct): Update prototype.
12199         (cgraph_speculative_call_info): Declare.
12200         * ipa-cp.c (ipcp_discover_new_direct_edges): Be ready for edge
12201         to change; update call of ipa_find_references.
12202         * ipa-ref.c (ipa_record_reference): Fix return value; clear
12203         lto_stmt_uid and speculative flags.
12204         (ipa_dump_references): Dump speculation.
12205         (ipa_clone_references): Clone speculative flag.
12206         (ipa_clone_referring): Likewise.
12207         (ipa_clone_ref): New function.
12208         (ipa_find_reference): Look into lto_stmt_uids
12209         (ipa_clear_stmts_in_references): Do not clear speculative calls.
12210         * ipa-ref.h (ipa_ref): Add lto_stmt_uid and speculative flags.
12211         (ipa_find_reference): Update declaration.
12212         (ipa_clone_ref): Declare.
12213         * lto-cgraph.c (lto_output_edge): Make lto_stmt_uids start from 0;
12214         stream speculative flag.
12215         (lto_output_ref): Stream statements uids and speculation.
12216         (input_ref): Likewise.
12217         (input_edge): Stream speuclation.
12218         * cgraphclones.c (cgraph_clone_edge): Clone speculation.
12219         (cgraph_set_call_stmt_including_clones): Handle speculation.
12220         * ipa-inline.c (heap_edge_removal_hook): New function.
12221         (inline_small_functions): Register it.
12222         * lto-streamer-in.c (fixup_call_stmt_edges_1): Bounds checking;
12223         also initialize refs.
12224         * ipa-prop.c (ipa_make_edge_direct_to_target): Be ready for
12225         edge to change.
12226         (try_make_edge_direct_simple_call): Likewise.
12227         (try_make_edge_direct_simple_call): Likewise.
12228         (update_indirect_edges_after_inlining): Likewise.
12229         (remove_described_reference): Look proper lto_stmt_uid.
12230         (propagate_controlled_uses): Likewise.
12231         (propagate_controlled_uses): Liekwise.
12232         * tree-inline.c (copy_bb): Copy speculative edges.
12233         (redirect_all_calls): New function.
12234         (copy_cfg_body): Do redirection after loop info is updated.
12235         (delete_unreachable_blocks_update_callgraph): Updadte speculation.
12237 2013-08-09  Jan Hubicka  <jh@suse.cz>
12239         * lto-streamer-out.c (output_function): Renumber PHIs.
12240         * lto-streamer-in.c (input_function): Likewise.
12242 2013-08-09  James Greenhalgh  <james.greenhalgh@arm.com>
12244         * config/aarch64/aarch64-simd-builtins.def (get_lane_signed): Remove.
12245         (get_lane_unsigned): Likewise.
12246         (dup_lane_scalar): Likewise.
12247         (get_lane): enable for VALL.
12248         * config/aarch64/aarch64-simd.md
12249         (aarch64_dup_lane_scalar<mode>): Remove.
12250         (aarch64_get_lane_signed<mode>): Likewise.
12251         (aarch64_get_lane_unsigned<mode>): Likewise.
12252         (aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): New.
12253         (aarch64_get_lane_zero_extendsi<mode>): Likewise.
12254         (aarch64_get_lane<mode>): Enable for all vector modes.
12255         (aarch64_get_lanedi): Remove misleading constraints.
12256         * config/aarch64/arm_neon.h
12257         (__aarch64_vget_lane_any): Define.
12258         (__aarch64_vget<q>_lane_<fpsu><8,16,32,64>): Likewise.
12259         (vget<q>_lane_<fpsu><8,16,32,64>): Use __aarch64_vget_lane macros.
12260         (vdup<bhsd>_lane_<su><8,16,32,64>): Likewise.
12261         * config/aarch64/iterators.md (VDQQH): New.
12262         (VDQQHS): Likewise.
12263         (vwcore): Likewise.
12265 2013-08-09  Eric Botcazou  <ebotcazou@adacore.com>
12267         * configure.ac: Add GAS check for LEON instructions on SPARC.
12268         * configure: Regenerate.
12269         * config.in: Likewise.
12270         * config.gcc (with_cpu): Remove sparc-leon*-* and deal with LEON in the
12271         sparc*-*-* block.
12272         * config/sparc/sparc.opt (LEON, LEON3): New masks.
12273         * config/sparc/sparc.h (ASM_CPU32_DEFAULT_SPEC): Set to AS_LEON_FLAG
12274         for LEON or LEON3.
12275         (ASM_CPU_SPEC): Pass AS_LEON_FLAG if -mcpu=leon or -mcpu=leon3.
12276         (AS_LEON_FLAG): New macro.
12277         * config/sparc/sparc.c (sparc_option_override): Set MASK_LEON for leon
12278         and MASK_LEON3 for leon3 and unset them if HAVE_AS_LEON is not defined.
12279         Deal with LEON and LEON3 for the memory model.
12280         * config/sparc/sync.md (atomic_compare_and_swap<mode>): Enable if LEON3
12281         (atomic_compare_and_swap<mode>_1): Likewise.
12282         (*atomic_compare_and_swap<mode>_1): Likewise.
12284 2013-08-09  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
12286         * config/arm/neon.md (vcond): Fix floating-point vector
12287         comparisons against 0.
12289 2013-08-08  Vladimir Makarov  <vmakarov@redhat.com>
12291         * lra-constraints.c (emit_spill_move): Remove assert.
12292         (process_alt_operands): Add more debugging
12293         output.  Increase reject for spilling into memory.  Decrease
12294         reject for reloading scratch.
12295         (split_reg): Use HARD_REGNO_CALLER_SAVE_MODE.
12297 2013-08-08  Steve Ellcey  <sellcey@mips.com>
12299         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add nan2008.
12300         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Make mips16 and
12301         micromips incompatible.  Add nan2008.
12302         (MULTILIB_DIRNAMES): Add nan2008.
12303         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12304         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Make mips16
12305         and micromips incompatible.  Add nan2008.
12306         (MULTILIB_DIRNAMES): Add nan2008.
12307         (MULTILIB_EXCEPTIONS): Remove mips16/micromips entry.
12309 2013-08-08  Richard Sandiford  <rdsandiford@googlemail.com>
12311         PR rtl-optimization/58079
12312         * combine.c (combine_simplify_rtx): Avoid using SUBST if
12313         simplify_comparison has widened a comparison with an integer.
12315 2013-08-08  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
12317         * config/arm/neon.md (movmisalign<mode>): Disable when we
12318         don't allow unaligned accesses.
12319         (*movmisalign<mode>_neon_store): Likewise.
12320         (*movmisalign<mode>_neon_load): Likewise.
12321         (*movmisalign<mode>_neon_store): Likewise.
12322         (*movmisalign<mode>_neon_load): Likewise.
12324 2013-08-08  Jan Hubicka  <jh@suse.cz>
12326         * cgraphbuild.c (build_cgraph_edges): Do not walk into debugs.
12327         (make_pass_rebuild_cgraph_edges): Also clear references.
12328         * cgraph.c (verify_cgraph_node): Add basic ipa-ref verifier.
12329         * ipa-inline-transform.c (inline_transform): Remove all references
12330         after inlining.
12331         * cgraphunit.c (expand_function): Remove all references after
12332         expansion.
12333         * ipa-ref.c (ipa_ref_has_aliases_p): Fix formatting.
12334         (ipa_find_reference): Rewrite to iterator.
12335         (remove_stmt_references): Likewise.
12336         (ipa_clear_stmts_in_references): New function.
12337         * ipa-ref.h (ipa_clear_stmts_in_references): Declare.
12338         * cgraphclones.c (cgraph_materialize_all_clones): Remove or
12339         clear references.
12340         * ipa-split.c (split_function): Remove references in split function.
12342 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12344         PR target/57431
12345         * config/arm/arm/neon.md (neon_vld1_dupdi): New expand pattern.
12346         (neon_vld1_dup<mode> VD iterator): Iterate over VD not VDX.
12348 2013-08-08  Richard Earnshaw  <rearnsha@arm.com>
12350         PR target/56979
12351         * config/arm/arm.c (aapcs_vfp_allocate): Decompose the argument if the
12352         suggested mode for the assignment isn't compatible with the
12353         registers required.
12355 2013-08-08  Bernd Edlinger  <bernd.edlinger@hotmail.de>
12357         PR target/58065
12358         * config/arm/arm.h (MALLOC_ABI_ALIGNMENT): Define.
12360 2013-08-07  Xinliang David Li  <davidxl@google.com>
12362         * config/i386/i386.opt: New option -mtune-ctrl=.
12363         * config/i386/x86-tune.def: New file.
12364         * config/i386/i386.h: include x86-tune.def.
12365         * config/i386/i386.c (ix86_option_override_internal):
12366         Parsing -mtune-ctrl= option and set tune features.
12368 2013-08-07  Oleg Endo  <olegendo@gcc.gnu.org>
12370         PR other/12081
12371         * config/rs6000/rs6000.c (gen_2arg_fn_t): Remove typedef.
12372         (rs6000_emit_swdiv, rs6000_emit_swrsqrt): Don't cast result of GEN_FCN
12373         to gen_2arg_fn_t.
12375 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12377         * rtl.h (update_alignments): Declare.
12378         * final.c (grow_label_align): New function extracted from...
12379         (shorten_branches): ...here.  Call it.
12380         (update_alignments): New function.
12381         * reorg.c (sibling_labels): New variable.
12382         (get_label_before): Add SIBLING parameter.  If it is non-zero, push
12383         the new label along with it onto the sibling_labels vector.
12384         (fill_simple_delay_slots): Adjust call to get_label_before.
12385         (fill_slots_from_thread): Likewise.
12386         (relax_delay_slots): Likewise.
12387         (make_return_insns): Likewise.
12388         (dbr_schedule): Invoke update_alignment on the sibling_labels vector.
12390 2013-08-07  Eric Botcazou  <ebotcazou@adacore.com>
12392         * diagnostic.c (diagnostic_classify_diagnostic): Accept zero index and
12393         document its semantics.
12394         (diagnostic_report_diagnostic): Adjust accordingly.
12396 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12398         * config/sparc/sparc.c (insert_pass_work_around_errata): Move into...
12399         (sparc_option_override): ...and port to new C++ pass API.
12400         * config/sparc/t-sparc (sparc.o): Add dep on CONTEXT_H
12402 2013-08-07  Peter Bergner  <bergner@vnet.ibm.com>
12404         * config/rs6000/rs6000.c (htm_expand_builtin) <case 0>: Remove.
12406 2013-08-06  Caroline Tice  <cmtice@google.com>
12408         * gcc.c (VTABLE_VERIFICATION_SPEC): New definition.
12409         (LINK_COMMAND_SPEC): Add VTABLE_VERIFICATION_SPEC.
12410         * tree-pass.h: Add pass_vtable_verify.
12411         * varasm.c (assemble_variable): Add code to properly set the comdat
12412         section and name for the .vtable_map_vars section.
12413         (assemble_vtyv_preinit_initializer): New function.
12414         (default_sectin_type_flags):  Make sure .vtable_map_vars section has
12415         LINK_ONCE flag.
12416         * output.h: Add function decl for assemble_vtv_preinit_initializer.
12417         * vtable-verify.c: New file.
12418         * vtable-verify.h: New file.
12419         * flag-types.h (enum vtv_priority): Defintions for flag_vtable_verify
12420         initialiation levels.
12421         * timevar.def (TV_VTABLE_VERIFICATION): New definition.
12422         * passes.def: Insert pass_vtable_verify.
12423         * aclocal.m4: Reorder includes.
12424         * doc/invoke.texi:  Document the -fvtable-verify=, -fvtv-debug, and
12425         -fvtv-counts options.
12426         * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add vtv_start*.o,
12427         as appropriate, if -fvtable-verify=... is used.
12428         (GNU_USER_TARGET_ENDFILE_SPEC): Add vtv_end*.o as appropriate, if
12429         -fvtable-verify=... is used.
12430         * Makefile.in (OBJS):  Add vtable-verify.o to list.
12431         (vtable-verify.o): Add new build rule.
12432         (GTFILES): Add vtable-verify.c to list.
12433         * common.opt (fvtable-verify=): New flag.
12434         (vtv_priority): Values for fvtable-verify= flag.
12435         (fvtv-counts): New flag.
12436         (fvtv-debug): New flag.
12437         * tree.h (save_vtable_map_decl): New extern function decl.
12439 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12441         * config/rl78/rl78.c (rl78_devirt_pass): Convert from a struct to...
12442         (pass_rl78_devirt): ...new subclass of rtl_opt_pass along with...
12443         (pass_data_rl78_devirt): ...new pass_data instance and...
12444         (make_pass_rl78_devirt): ...new function.
12445         (rl78_asm_file_start): Port pass registration to new C++ API.
12447 2013-08-07  David Malcolm  <dmalcolm@redhat.com>
12449         * coretypes.h (rtl_opt_pass): Add.
12450         (gcc::context): Add.
12451         * config/epiphany/epiphany.c (pass_mode_switch_use): New.
12452         (epiphany_init): Port to new C++ pass API.
12453         (epiphany_optimize_mode_switching): Likewise.
12454         * pass_manager.h (pass_manager::get_pass_split_all_insns): New.
12455         (pass_manager::get_pass_mode_switching): New.
12456         (pass_manager::get_pass_peephole2): New.
12457         * mode-switching.c (pass_mode_switching): Add clone method.
12458         * recog.c (pass_peephole2): Add clone method.
12459         (pass_split_all_insns): Add clone method.
12461 2013-08-06  David Malcolm  <dmalcolm@redhat.com>
12463         * config/mips/mips.c (insert_pass_mips_machine_reorg2): Move into...
12464         (mips_option_override): ...here, porting to new C++ API for passes.
12466 2013-08-06  Jan Hubicka  <jh@suse.cz>
12468         * cgraph.c (cgraph_get_body): New function based on lto.c
12469         implementation.
12470         * cgraph.h (cgraph_get_body): Declare.
12471         * cgraphclones.c (cgraph_create_virtual_clone): Commonize WPA and
12472         LTO paths.
12473         * cgraphunit.c (expand_function): Get body prior expanding.
12474         * ipa.c (function_and_variable_visibility): Use gimple_has_body_p test.
12475         * lto-cgraph.c (lto_output_node): Do not stream bodies we don't
12476         really need.
12477         * passes.c (do_per_function_toporder): Get body.
12478         * tree-inline.c (expand_call_inline): Get body prior inlining it.
12479         * tree-ssa-structalias.c (ipa_pta_execute): Get body; skip clones.
12481 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12483         PR fortran/57987
12484         * cgraphunit.c (cgraph_finalize_function): Assert that nested function
12485         is not re-finalized.  Rename second parameter to no_collect.
12487 2013-08-06  Martin Jambor  <mjambor@suse.cz>
12489         PR middle-end/58041
12490         * gimple-ssa-strength-reduction.c (replace_ref): Make sure built
12491         MEM_REF has proper alignment information.
12493 2013-08-05  Oleg Endo  <olegendo@gcc.gnu.org>
12495         PR other/12081
12496         * recog.h (rtx (*insn_gen_fn) (rtx, ...)): Replace typedef with new
12497         class insn_gen_fn.
12498         * expr.c (move_by_pieces_1, store_by_pieces_2): Replace argument
12499         rtx (*) (rtx, ...) with insn_gen_fn.
12500         * genoutput.c (output_insn_data): Cast gen_? function pointers to
12501         insn_gen_fn::stored_funcptr.  Add initializer braces.
12503 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12505         Rewrite how instances of passes are cloned to remove assumptions
12506         about their sizes (thus allowing pass subclasses to have
12507         additional data fields, albeit non-GC-managed ones at this point).
12509         * passes.c (make_pass_instance): Now that passes have clone
12510         methods, rewrite this function to eliminate XNEW and memcpy
12511         calls that used hardcoded sizes.  Since this function no longer
12512         creates pass instances, rename it to...
12513         (add_pass_instance): ...this.  Document the old way that passes were
12514         numbered and flagged, and rework this function to continue using it.
12515         (next_pass_1): Add an initial_pass argument for use by
12516         add_pass_instance.
12517         (position_pass): When adding multiple instances of a pass, use
12518         the pass's clone method, rather than relying on the XNEW/memcpy
12519         within the former make_pass_instance (now add_pass_instance).
12520         (pass_manager::pass_manager): When invoking next_pass_1, also supply
12521         the initial instance of the current pass within the pass manager.
12523 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
12525         This is the automated part of the conversion of passes from C
12526         structs to C++ classes.
12528         Patch autogenerated by refactor_passes.py from
12529         https://github.com/davidmalcolm/gcc-refactoring-scripts
12530         revision 03fe39476a4c4ea450b49e087cfa817b5f92021e
12532         * asan.c (pass_asan): Convert from a global struct to a subclass of
12533         gimple_opt_pass along with...
12534         (pass_data_asan): ...new pass_data instance and...
12535         (make_pass_asan): ...new function.
12536         (pass_asan_O0): Convert from a global struct to a subclass of
12537         gimple_opt_pass along with...
12538         (pass_data_asan_O0): ...new pass_data instance and...
12539         (make_pass_asan_O0): ...new function.
12540         * auto-inc-dec.c (pass_inc_dec): Convert from a global struct to a
12541         subclass of rtl_opt_pass along with...
12542         (pass_data_inc_dec): ...new pass_data instance and...
12543         (make_pass_inc_dec): ...new function.
12544         * bb-reorder.c (pass_reorder_blocks): Convert from a global struct to
12545         a subclass of rtl_opt_pass along with...
12546         (pass_data_reorder_blocks): ...new pass_data instance and...
12547         (make_pass_reorder_blocks): ...new function.
12548         (pass_duplicate_computed_gotos): Convert from a global struct to a
12549         subclass of rtl_opt_pass along with...
12550         (pass_data_duplicate_computed_gotos): ...new pass_data instance and...
12551         (make_pass_duplicate_computed_gotos): ...new function.
12552         (pass_partition_blocks): Convert from a global struct to a subclass of
12553         rtl_opt_pass along with...
12554         (pass_data_partition_blocks): ...new pass_data instance and...
12555         (make_pass_partition_blocks): ...new function.
12556         * bt-load.c (pass_branch_target_load_optimize1): Convert from a global
12557         struct to a subclass of rtl_opt_pass along with...
12558         (pass_data_branch_target_load_optimize1): ...new pass_data instance
12559         and...
12560         (make_pass_branch_target_load_optimize1): ...new function.
12561         (pass_branch_target_load_optimize2): Convert from a global struct to a
12562         subclass of rtl_opt_pass along with...
12563         (pass_data_branch_target_load_optimize2): ...new pass_data instance
12564         and...
12565         (make_pass_branch_target_load_optimize2): ...new function.
12566         * cfgcleanup.c (pass_jump): Convert from a global struct to a subclass
12567         of rtl_opt_pass along with...
12568         (pass_data_jump): ...new pass_data instance and...
12569         (make_pass_jump): ...new function.
12570         (pass_jump2): Convert from a global struct to a subclass of
12571         rtl_opt_pass along with...
12572         (pass_data_jump2): ...new pass_data instance and...
12573         (make_pass_jump2): ...new function.
12574         * cfgexpand.c (pass_expand): Convert from a global struct to a
12575         subclass of rtl_opt_pass along with...
12576         (pass_data_expand): ...new pass_data instance and...
12577         (make_pass_expand): ...new function.
12578         * cfgrtl.c (pass_free_cfg): Convert from a global struct to a subclass
12579         of rtl_opt_pass along with...
12580         (pass_data_free_cfg): ...new pass_data instance and...
12581         (make_pass_free_cfg): ...new function.
12582         (pass_into_cfg_layout_mode): Convert from a global struct to a
12583         subclass of rtl_opt_pass along with...
12584         (pass_data_into_cfg_layout_mode): ...new pass_data instance and...
12585         (make_pass_into_cfg_layout_mode): ...new function.
12586         (pass_outof_cfg_layout_mode): Convert from a global struct to a
12587         subclass of rtl_opt_pass along with...
12588         (pass_data_outof_cfg_layout_mode): ...new pass_data instance and...
12589         (make_pass_outof_cfg_layout_mode): ...new function.
12590         * cgraphbuild.c (pass_build_cgraph_edges): Convert from a global
12591         struct to a subclass of gimple_opt_pass along with...
12592         (pass_data_build_cgraph_edges): ...new pass_data instance and...
12593         (make_pass_build_cgraph_edges): ...new function.
12594         (pass_rebuild_cgraph_edges): Convert from a global struct to a
12595         subclass of gimple_opt_pass along with...
12596         (pass_data_rebuild_cgraph_edges): ...new pass_data instance and...
12597         (make_pass_rebuild_cgraph_edges): ...new function.
12598         (pass_remove_cgraph_callee_edges): Convert from a global struct to a
12599         subclass of gimple_opt_pass along with...
12600         (pass_data_remove_cgraph_callee_edges): ...new pass_data instance
12601         and...
12602         (make_pass_remove_cgraph_callee_edges): ...new function.
12603         * combine-stack-adj.c (pass_stack_adjustments): Convert from a global
12604         struct to a subclass of rtl_opt_pass along with...
12605         (pass_data_stack_adjustments): ...new pass_data instance and...
12606         (make_pass_stack_adjustments): ...new function.
12607         * combine.c (pass_combine): Convert from a global struct to a subclass
12608         of rtl_opt_pass along with...
12609         (pass_data_combine): ...new pass_data instance and...
12610         (make_pass_combine): ...new function.
12611         * compare-elim.c (pass_compare_elim_after_reload): Convert from a
12612         global struct to a subclass of rtl_opt_pass along with...
12613         (pass_data_compare_elim_after_reload): ...new pass_data instance
12614         and...
12615         (make_pass_compare_elim_after_reload): ...new function.
12616         * cprop.c (pass_rtl_cprop): Convert from a global struct to a subclass
12617         of rtl_opt_pass along with...
12618         (pass_data_rtl_cprop): ...new pass_data instance and...
12619         (make_pass_rtl_cprop): ...new function.
12620         * cse.c (pass_cse): Convert from a global struct to a subclass of
12621         rtl_opt_pass along with...
12622         (pass_data_cse): ...new pass_data instance and...
12623         (make_pass_cse): ...new function.
12624         (pass_cse2): Convert from a global struct to a subclass of
12625         rtl_opt_pass along with...
12626         (pass_data_cse2): ...new pass_data instance and...
12627         (make_pass_cse2): ...new function.
12628         (pass_cse_after_global_opts): Convert from a global struct to a
12629         subclass of rtl_opt_pass along with...
12630         (pass_data_cse_after_global_opts): ...new pass_data instance and...
12631         (make_pass_cse_after_global_opts): ...new function.
12632         * dce.c (pass_ud_rtl_dce): Convert from a global struct to a subclass
12633         of rtl_opt_pass along with...
12634         (pass_data_ud_rtl_dce): ...new pass_data instance and...
12635         (make_pass_ud_rtl_dce): ...new function.
12636         (pass_fast_rtl_dce): Convert from a global struct to a subclass of
12637         rtl_opt_pass along with...
12638         (pass_data_fast_rtl_dce): ...new pass_data instance and...
12639         (make_pass_fast_rtl_dce): ...new function.
12640         * df-core.c (pass_df_initialize_opt): Convert from a global struct to
12641         a subclass of rtl_opt_pass along with...
12642         (pass_data_df_initialize_opt): ...new pass_data instance and...
12643         (make_pass_df_initialize_opt): ...new function.
12644         (pass_df_initialize_no_opt): Convert from a global struct to a
12645         subclass of rtl_opt_pass along with...
12646         (pass_data_df_initialize_no_opt): ...new pass_data instance and...
12647         (make_pass_df_initialize_no_opt): ...new function.
12648         (pass_df_finish): Convert from a global struct to a subclass of
12649         rtl_opt_pass along with...
12650         (pass_data_df_finish): ...new pass_data instance and...
12651         (make_pass_df_finish): ...new function.
12652         * dse.c (pass_rtl_dse1): Convert from a global struct to a subclass of
12653         rtl_opt_pass along with...
12654         (pass_data_rtl_dse1): ...new pass_data instance and...
12655         (make_pass_rtl_dse1): ...new function.
12656         (pass_rtl_dse2): Convert from a global struct to a subclass of
12657         rtl_opt_pass along with...
12658         (pass_data_rtl_dse2): ...new pass_data instance and...
12659         (make_pass_rtl_dse2): ...new function.
12660         * dwarf2cfi.c (pass_dwarf2_frame): Convert from a global struct to a
12661         subclass of rtl_opt_pass along with...
12662         (pass_data_dwarf2_frame): ...new pass_data instance and...
12663         (make_pass_dwarf2_frame): ...new function.
12664         * except.c (pass_set_nothrow_function_flags): Convert from a global
12665         struct to a subclass of rtl_opt_pass along with...
12666         (pass_data_set_nothrow_function_flags): ...new pass_data instance
12667         and...
12668         (make_pass_set_nothrow_function_flags): ...new function.
12669         (pass_convert_to_eh_region_ranges): Convert from a global struct to a
12670         subclass of rtl_opt_pass along with...
12671         (pass_data_convert_to_eh_region_ranges): ...new pass_data instance
12672         and...
12673         (make_pass_convert_to_eh_region_ranges): ...new function.
12674         * final.c (pass_compute_alignments): Convert from a global struct to a
12675         subclass of rtl_opt_pass along with...
12676         (pass_data_compute_alignments): ...new pass_data instance and...
12677         (make_pass_compute_alignments): ...new function.
12678         (pass_final): Convert from a global struct to a subclass of
12679         rtl_opt_pass along with...
12680         (pass_data_final): ...new pass_data instance and...
12681         (make_pass_final): ...new function.
12682         (pass_shorten_branches): Convert from a global struct to a subclass of
12683         rtl_opt_pass along with...
12684         (pass_data_shorten_branches): ...new pass_data instance and...
12685         (make_pass_shorten_branches): ...new function.
12686         (pass_clean_state): Convert from a global struct to a subclass of
12687         rtl_opt_pass along with...
12688         (pass_data_clean_state): ...new pass_data instance and...
12689         (make_pass_clean_state): ...new function.
12690         * function.c (pass_instantiate_virtual_regs): Convert from a global
12691         struct to a subclass of rtl_opt_pass along with...
12692         (pass_data_instantiate_virtual_regs): ...new pass_data instance and...
12693         (make_pass_instantiate_virtual_regs): ...new function.
12694         (pass_leaf_regs): Convert from a global struct to a subclass of
12695         rtl_opt_pass along with...
12696         (pass_data_leaf_regs): ...new pass_data instance and...
12697         (make_pass_leaf_regs): ...new function.
12698         (pass_thread_prologue_and_epilogue): Convert from a global struct to a
12699         subclass of rtl_opt_pass along with...
12700         (pass_data_thread_prologue_and_epilogue): ...new pass_data instance
12701         and...
12702         (make_pass_thread_prologue_and_epilogue): ...new function.
12703         (pass_match_asm_constraints): Convert from a global struct to a
12704         subclass of rtl_opt_pass along with...
12705         (pass_data_match_asm_constraints): ...new pass_data instance and...
12706         (make_pass_match_asm_constraints): ...new function.
12707         * fwprop.c (pass_rtl_fwprop): Convert from a global struct to a
12708         subclass of rtl_opt_pass along with...
12709         (pass_data_rtl_fwprop): ...new pass_data instance and...
12710         (make_pass_rtl_fwprop): ...new function.
12711         (pass_rtl_fwprop_addr): Convert from a global struct to a subclass of
12712         rtl_opt_pass along with...
12713         (pass_data_rtl_fwprop_addr): ...new pass_data instance and...
12714         (make_pass_rtl_fwprop_addr): ...new function.
12715         * gcse.c (pass_rtl_pre): Convert from a global struct to a subclass of
12716         rtl_opt_pass along with...
12717         (pass_data_rtl_pre): ...new pass_data instance and...
12718         (make_pass_rtl_pre): ...new function.
12719         (pass_rtl_hoist): Convert from a global struct to a subclass of
12720         rtl_opt_pass along with...
12721         (pass_data_rtl_hoist): ...new pass_data instance and...
12722         (make_pass_rtl_hoist): ...new function.
12723         * gimple-low.c (pass_lower_cf): Convert from a global struct to a
12724         subclass of gimple_opt_pass along with...
12725         (pass_data_lower_cf): ...new pass_data instance and...
12726         (make_pass_lower_cf): ...new function.
12727         * gimple-ssa-strength-reduction.c (pass_strength_reduction): Convert
12728         from a global struct to a subclass of gimple_opt_pass along with...
12729         (pass_data_strength_reduction): ...new pass_data instance and...
12730         (make_pass_strength_reduction): ...new function.
12731         * ifcvt.c (pass_rtl_ifcvt): Convert from a global struct to a subclass
12732         of rtl_opt_pass along with...
12733         (pass_data_rtl_ifcvt): ...new pass_data instance and...
12734         (make_pass_rtl_ifcvt): ...new function.
12735         (pass_if_after_combine): Convert from a global struct to a subclass of
12736         rtl_opt_pass along with...
12737         (pass_data_if_after_combine): ...new pass_data instance and...
12738         (make_pass_if_after_combine): ...new function.
12739         (pass_if_after_reload): Convert from a global struct to a subclass of
12740         rtl_opt_pass along with...
12741         (pass_data_if_after_reload): ...new pass_data instance and...
12742         (make_pass_if_after_reload): ...new function.
12743         * init-regs.c (pass_initialize_regs): Convert from a global struct to
12744         a subclass of rtl_opt_pass along with...
12745         (pass_data_initialize_regs): ...new pass_data instance and...
12746         (make_pass_initialize_regs): ...new function.
12747         * ipa-cp.c (pass_ipa_cp): Convert from a global struct to a subclass
12748         of ipa_opt_pass_d along with...
12749         (pass_data_ipa_cp): ...new pass_data instance and...
12750         (make_pass_ipa_cp): ...new function.
12751         * ipa-inline-analysis.c (pass_inline_parameters): Convert from a
12752         global struct to a subclass of gimple_opt_pass along with...
12753         (pass_data_inline_parameters): ...new pass_data instance and...
12754         (make_pass_inline_parameters): ...new function.
12755         * ipa-inline.c (pass_early_inline): Convert from a global struct to a
12756         subclass of gimple_opt_pass along with...
12757         (pass_data_early_inline): ...new pass_data instance and...
12758         (make_pass_early_inline): ...new function.
12759         (pass_ipa_inline): Convert from a global struct to a subclass of
12760         ipa_opt_pass_d along with...
12761         (pass_data_ipa_inline): ...new pass_data instance and...
12762         (make_pass_ipa_inline): ...new function.
12763         * ipa-pure-const.c (pass_local_pure_const): Convert from a global
12764         struct to a subclass of gimple_opt_pass along with...
12765         (pass_data_local_pure_const): ...new pass_data instance and...
12766         (make_pass_local_pure_const): ...new function.
12767         (pass_ipa_pure_const): Convert from a global struct to a subclass of
12768         ipa_opt_pass_d along with...
12769         (pass_data_ipa_pure_const): ...new pass_data instance and...
12770         (make_pass_ipa_pure_const): ...new function.
12771         * ipa-reference.c (pass_ipa_reference): Convert from a global struct
12772         to a subclass of ipa_opt_pass_d along with...
12773         (pass_data_ipa_reference): ...new pass_data instance and...
12774         (make_pass_ipa_reference): ...new function.
12775         * ipa-split.c (pass_split_functions): Convert from a global struct to
12776         a subclass of gimple_opt_pass along with...
12777         (pass_data_split_functions): ...new pass_data instance and...
12778         (make_pass_split_functions): ...new function.
12779         (pass_feedback_split_functions): Convert from a global struct to a
12780         subclass of gimple_opt_pass along with...
12781         (pass_data_feedback_split_functions): ...new pass_data instance and...
12782         (make_pass_feedback_split_functions): ...new function.
12783         * ipa.c (pass_ipa_function_and_variable_visibility): Convert from a
12784         global struct to a subclass of simple_ipa_opt_pass along with...
12785         (pass_data_ipa_function_and_variable_visibility): ...new pass_data
12786         instance and...
12787         (make_pass_ipa_function_and_variable_visibility): ...new function.
12788         (pass_ipa_free_inline_summary): Convert from a global struct to a
12789         subclass of simple_ipa_opt_pass along with...
12790         (pass_data_ipa_free_inline_summary): ...new pass_data instance and...
12791         (make_pass_ipa_free_inline_summary): ...new function.
12792         (pass_ipa_whole_program_visibility): Convert from a global struct to a
12793         subclass of ipa_opt_pass_d along with...
12794         (pass_data_ipa_whole_program_visibility): ...new pass_data instance
12795         and...
12796         (make_pass_ipa_whole_program_visibility): ...new function.
12797         (pass_ipa_profile): Convert from a global struct to a subclass of
12798         ipa_opt_pass_d along with...
12799         (pass_data_ipa_profile): ...new pass_data instance and...
12800         (make_pass_ipa_profile): ...new function.
12801         (pass_ipa_cdtor_merge): Convert from a global struct to a subclass of
12802         ipa_opt_pass_d along with...
12803         (pass_data_ipa_cdtor_merge): ...new pass_data instance and...
12804         (make_pass_ipa_cdtor_merge): ...new function.
12805         * ira.c (pass_ira): Convert from a global struct to a subclass of
12806         rtl_opt_pass along with...
12807         (pass_data_ira): ...new pass_data instance and...
12808         (make_pass_ira): ...new function.
12809         (pass_reload): Convert from a global struct to a subclass of
12810         rtl_opt_pass along with...
12811         (pass_data_reload): ...new pass_data instance and...
12812         (make_pass_reload): ...new function.
12813         * jump.c (pass_cleanup_barriers): Convert from a global struct to a
12814         subclass of rtl_opt_pass along with...
12815         (pass_data_cleanup_barriers): ...new pass_data instance and...
12816         (make_pass_cleanup_barriers): ...new function.
12817         * loop-init.c (pass_loop2): Convert from a global struct to a subclass
12818         of rtl_opt_pass along with...
12819         (pass_data_loop2): ...new pass_data instance and...
12820         (make_pass_loop2): ...new function.
12821         (pass_rtl_loop_init): Convert from a global struct to a subclass of
12822         rtl_opt_pass along with...
12823         (pass_data_rtl_loop_init): ...new pass_data instance and...
12824         (make_pass_rtl_loop_init): ...new function.
12825         (pass_rtl_loop_done): Convert from a global struct to a subclass of
12826         rtl_opt_pass along with...
12827         (pass_data_rtl_loop_done): ...new pass_data instance and...
12828         (make_pass_rtl_loop_done): ...new function.
12829         (pass_rtl_move_loop_invariants): Convert from a global struct to a
12830         subclass of rtl_opt_pass along with...
12831         (pass_data_rtl_move_loop_invariants): ...new pass_data instance and...
12832         (make_pass_rtl_move_loop_invariants): ...new function.
12833         (pass_rtl_unswitch): Convert from a global struct to a subclass of
12834         rtl_opt_pass along with...
12835         (pass_data_rtl_unswitch): ...new pass_data instance and...
12836         (make_pass_rtl_unswitch): ...new function.
12837         (pass_rtl_unroll_and_peel_loops): Convert from a global struct to a
12838         subclass of rtl_opt_pass along with...
12839         (pass_data_rtl_unroll_and_peel_loops): ...new pass_data instance
12840         and...
12841         (make_pass_rtl_unroll_and_peel_loops): ...new function.
12842         (pass_rtl_doloop): Convert from a global struct to a subclass of
12843         rtl_opt_pass along with...
12844         (pass_data_rtl_doloop): ...new pass_data instance and...
12845         (make_pass_rtl_doloop): ...new function.
12846         * lower-subreg.c (pass_lower_subreg): Convert from a global struct to
12847         a subclass of rtl_opt_pass along with...
12848         (pass_data_lower_subreg): ...new pass_data instance and...
12849         (make_pass_lower_subreg): ...new function.
12850         (pass_lower_subreg2): Convert from a global struct to a subclass of
12851         rtl_opt_pass along with...
12852         (pass_data_lower_subreg2): ...new pass_data instance and...
12853         (make_pass_lower_subreg2): ...new function.
12854         * lto-streamer-out.c (pass_ipa_lto_gimple_out): Convert from a global
12855         struct to a subclass of ipa_opt_pass_d along with...
12856         (pass_data_ipa_lto_gimple_out): ...new pass_data instance and...
12857         (make_pass_ipa_lto_gimple_out): ...new function.
12858         (pass_ipa_lto_finish_out): Convert from a global struct to a subclass
12859         of ipa_opt_pass_d along with...
12860         (pass_data_ipa_lto_finish_out): ...new pass_data instance and...
12861         (make_pass_ipa_lto_finish_out): ...new function.
12862         * mode-switching.c (pass_mode_switching): Convert from a global struct
12863         to a subclass of rtl_opt_pass along with...
12864         (pass_data_mode_switching): ...new pass_data instance and...
12865         (make_pass_mode_switching): ...new function.
12866         * modulo-sched.c (pass_sms): Convert from a global struct to a
12867         subclass of rtl_opt_pass along with...
12868         (pass_data_sms): ...new pass_data instance and...
12869         (make_pass_sms): ...new function.
12870         * omp-low.c (pass_expand_omp): Convert from a global struct to a
12871         subclass of gimple_opt_pass along with...
12872         (pass_data_expand_omp): ...new pass_data instance and...
12873         (make_pass_expand_omp): ...new function.
12874         (pass_lower_omp): Convert from a global struct to a subclass of
12875         gimple_opt_pass along with...
12876         (pass_data_lower_omp): ...new pass_data instance and...
12877         (make_pass_lower_omp): ...new function.
12878         (pass_diagnose_omp_blocks): Convert from a global struct to a subclass
12879         of gimple_opt_pass along with...
12880         (pass_data_diagnose_omp_blocks): ...new pass_data instance and...
12881         (make_pass_diagnose_omp_blocks): ...new function.
12882         * passes.c (pass_early_local_passes): Convert from a global struct to
12883         a subclass of simple_ipa_opt_pass along with...
12884         (pass_data_early_local_passes): ...new pass_data instance and...
12885         (make_pass_early_local_passes): ...new function.
12886         (pass_all_early_optimizations): Convert from a global struct to a
12887         subclass of gimple_opt_pass along with...
12888         (pass_data_all_early_optimizations): ...new pass_data instance and...
12889         (make_pass_all_early_optimizations): ...new function.
12890         (pass_all_optimizations): Convert from a global struct to a subclass
12891         of gimple_opt_pass along with...
12892         (pass_data_all_optimizations): ...new pass_data instance and...
12893         (make_pass_all_optimizations): ...new function.
12894         (pass_all_optimizations_g): Convert from a global struct to a subclass
12895         of gimple_opt_pass along with...
12896         (pass_data_all_optimizations_g): ...new pass_data instance and...
12897         (make_pass_all_optimizations_g): ...new function.
12898         (pass_rest_of_compilation): Convert from a global struct to a subclass
12899         of rtl_opt_pass along with...
12900         (pass_data_rest_of_compilation): ...new pass_data instance and...
12901         (make_pass_rest_of_compilation): ...new function.
12902         (pass_postreload): Convert from a global struct to a subclass of
12903         rtl_opt_pass along with...
12904         (pass_data_postreload): ...new pass_data instance and...
12905         (make_pass_postreload): ...new function.
12906         * postreload-gcse.c (pass_gcse2): Convert from a global struct to a
12907         subclass of rtl_opt_pass along with...
12908         (pass_data_gcse2): ...new pass_data instance and...
12909         (make_pass_gcse2): ...new function.
12910         * postreload.c (pass_postreload_cse): Convert from a global struct to
12911         a subclass of rtl_opt_pass along with...
12912         (pass_data_postreload_cse): ...new pass_data instance and...
12913         (make_pass_postreload_cse): ...new function.
12914         * predict.c (pass_profile): Convert from a global struct to a subclass
12915         of gimple_opt_pass along with...
12916         (pass_data_profile): ...new pass_data instance and...
12917         (make_pass_profile): ...new function.
12918         (pass_strip_predict_hints): Convert from a global struct to a subclass
12919         of gimple_opt_pass along with...
12920         (pass_data_strip_predict_hints): ...new pass_data instance and...
12921         (make_pass_strip_predict_hints): ...new function.
12922         * recog.c (pass_peephole2): Convert from a global struct to a subclass
12923         of rtl_opt_pass along with...
12924         (pass_data_peephole2): ...new pass_data instance and...
12925         (make_pass_peephole2): ...new function.
12926         (pass_split_all_insns): Convert from a global struct to a subclass of
12927         rtl_opt_pass along with...
12928         (pass_data_split_all_insns): ...new pass_data instance and...
12929         (make_pass_split_all_insns): ...new function.
12930         (pass_split_after_reload): Convert from a global struct to a subclass
12931         of rtl_opt_pass along with...
12932         (pass_data_split_after_reload): ...new pass_data instance and...
12933         (make_pass_split_after_reload): ...new function.
12934         (pass_split_before_regstack): Convert from a global struct to a
12935         subclass of rtl_opt_pass along with...
12936         (pass_data_split_before_regstack): ...new pass_data instance and...
12937         (make_pass_split_before_regstack): ...new function.
12938         (pass_split_before_sched2): Convert from a global struct to a subclass
12939         of rtl_opt_pass along with...
12940         (pass_data_split_before_sched2): ...new pass_data instance and...
12941         (make_pass_split_before_sched2): ...new function.
12942         (pass_split_for_shorten_branches): Convert from a global struct to a
12943         subclass of rtl_opt_pass along with...
12944         (pass_data_split_for_shorten_branches): ...new pass_data instance
12945         and...
12946         (make_pass_split_for_shorten_branches): ...new function.
12947         * ree.c (pass_ree): Convert from a global struct to a subclass of
12948         rtl_opt_pass along with...
12949         (pass_data_ree): ...new pass_data instance and...
12950         (make_pass_ree): ...new function.
12951         * reg-stack.c (pass_stack_regs): Convert from a global struct to a
12952         subclass of rtl_opt_pass along with...
12953         (pass_data_stack_regs): ...new pass_data instance and...
12954         (make_pass_stack_regs): ...new function.
12955         (pass_stack_regs_run): Convert from a global struct to a subclass of
12956         rtl_opt_pass along with...
12957         (pass_data_stack_regs_run): ...new pass_data instance and...
12958         (make_pass_stack_regs_run): ...new function.
12959         * regcprop.c (pass_cprop_hardreg): Convert from a global struct to a
12960         subclass of rtl_opt_pass along with...
12961         (pass_data_cprop_hardreg): ...new pass_data instance and...
12962         (make_pass_cprop_hardreg): ...new function.
12963         * reginfo.c (pass_reginfo_init): Convert from a global struct to a
12964         subclass of rtl_opt_pass along with...
12965         (pass_data_reginfo_init): ...new pass_data instance and...
12966         (make_pass_reginfo_init): ...new function.
12967         * regmove.c (pass_regmove): Convert from a global struct to a subclass
12968         of rtl_opt_pass along with...
12969         (pass_data_regmove): ...new pass_data instance and...
12970         (make_pass_regmove): ...new function.
12971         * regrename.c (pass_regrename): Convert from a global struct to a
12972         subclass of rtl_opt_pass along with...
12973         (pass_data_regrename): ...new pass_data instance and...
12974         (make_pass_regrename): ...new function.
12975         * reorg.c (pass_delay_slots): Convert from a global struct to a
12976         subclass of rtl_opt_pass along with...
12977         (pass_data_delay_slots): ...new pass_data instance and...
12978         (make_pass_delay_slots): ...new function.
12979         (pass_machine_reorg): Convert from a global struct to a subclass of
12980         rtl_opt_pass along with...
12981         (pass_data_machine_reorg): ...new pass_data instance and...
12982         (make_pass_machine_reorg): ...new function.
12983         * sched-rgn.c (pass_sched): Convert from a global struct to a subclass
12984         of rtl_opt_pass along with...
12985         (pass_data_sched): ...new pass_data instance and...
12986         (make_pass_sched): ...new function.
12987         (pass_sched2): Convert from a global struct to a subclass of
12988         rtl_opt_pass along with...
12989         (pass_data_sched2): ...new pass_data instance and...
12990         (make_pass_sched2): ...new function.
12991         * stack-ptr-mod.c (pass_stack_ptr_mod): Convert from a global struct
12992         to a subclass of rtl_opt_pass along with...
12993         (pass_data_stack_ptr_mod): ...new pass_data instance and...
12994         (make_pass_stack_ptr_mod): ...new function.
12995         * store-motion.c (pass_rtl_store_motion): Convert from a global struct
12996         to a subclass of rtl_opt_pass along with...
12997         (pass_data_rtl_store_motion): ...new pass_data instance and...
12998         (make_pass_rtl_store_motion): ...new function.
12999         * tracer.c (pass_tracer): Convert from a global struct to a subclass
13000         of gimple_opt_pass along with...
13001         (pass_data_tracer): ...new pass_data instance and...
13002         (make_pass_tracer): ...new function.
13003         * trans-mem.c (pass_diagnose_tm_blocks): Convert from a global struct
13004         to a subclass of gimple_opt_pass along with...
13005         (pass_data_diagnose_tm_blocks): ...new pass_data instance and...
13006         (make_pass_diagnose_tm_blocks): ...new function.
13007         (pass_lower_tm): Convert from a global struct to a subclass of
13008         gimple_opt_pass along with...
13009         (pass_data_lower_tm): ...new pass_data instance and...
13010         (make_pass_lower_tm): ...new function.
13011         (pass_tm_init): Convert from a global struct to a subclass of
13012         gimple_opt_pass along with...
13013         (pass_data_tm_init): ...new pass_data instance and...
13014         (make_pass_tm_init): ...new function.
13015         (pass_tm_mark): Convert from a global struct to a subclass of
13016         gimple_opt_pass along with...
13017         (pass_data_tm_mark): ...new pass_data instance and...
13018         (make_pass_tm_mark): ...new function.
13019         (pass_tm_edges): Convert from a global struct to a subclass of
13020         gimple_opt_pass along with...
13021         (pass_data_tm_edges): ...new pass_data instance and...
13022         (make_pass_tm_edges): ...new function.
13023         (pass_tm_memopt): Convert from a global struct to a subclass of
13024         gimple_opt_pass along with...
13025         (pass_data_tm_memopt): ...new pass_data instance and...
13026         (make_pass_tm_memopt): ...new function.
13027         (pass_ipa_tm): Convert from a global struct to a subclass of
13028         simple_ipa_opt_pass along with...
13029         (pass_data_ipa_tm): ...new pass_data instance and...
13030         (make_pass_ipa_tm): ...new function.
13031         * tree-call-cdce.c (pass_call_cdce): Convert from a global struct to a
13032         subclass of gimple_opt_pass along with...
13033         (pass_data_call_cdce): ...new pass_data instance and...
13034         (make_pass_call_cdce): ...new function.
13035         * tree-cfg.c (pass_build_cfg): Convert from a global struct to a
13036         subclass of gimple_opt_pass along with...
13037         (pass_data_build_cfg): ...new pass_data instance and...
13038         (make_pass_build_cfg): ...new function.
13039         (pass_split_crit_edges): Convert from a global struct to a subclass of
13040         gimple_opt_pass along with...
13041         (pass_data_split_crit_edges): ...new pass_data instance and...
13042         (make_pass_split_crit_edges): ...new function.
13043         (pass_warn_function_return): Convert from a global struct to a
13044         subclass of gimple_opt_pass along with...
13045         (pass_data_warn_function_return): ...new pass_data instance and...
13046         (make_pass_warn_function_return): ...new function.
13047         (pass_warn_function_noreturn): Convert from a global struct to a
13048         subclass of gimple_opt_pass along with...
13049         (pass_data_warn_function_noreturn): ...new pass_data instance and...
13050         (make_pass_warn_function_noreturn): ...new function.
13051         (pass_warn_unused_result): Convert from a global struct to a subclass
13052         of gimple_opt_pass along with...
13053         (pass_data_warn_unused_result): ...new pass_data instance and...
13054         (make_pass_warn_unused_result): ...new function.
13055         * tree-cfgcleanup.c (pass_merge_phi): Convert from a global struct to
13056         a subclass of gimple_opt_pass along with...
13057         (pass_data_merge_phi): ...new pass_data instance and...
13058         (make_pass_merge_phi): ...new function.
13059         * tree-complex.c (pass_lower_complex): Convert from a global struct to
13060         a subclass of gimple_opt_pass along with...
13061         (pass_data_lower_complex): ...new pass_data instance and...
13062         (make_pass_lower_complex): ...new function.
13063         (pass_lower_complex_O0): Convert from a global struct to a subclass of
13064         gimple_opt_pass along with...
13065         (pass_data_lower_complex_O0): ...new pass_data instance and...
13066         (make_pass_lower_complex_O0): ...new function.
13067         * tree-eh.c (pass_lower_eh): Convert from a global struct to a
13068         subclass of gimple_opt_pass along with...
13069         (pass_data_lower_eh): ...new pass_data instance and...
13070         (make_pass_lower_eh): ...new function.
13071         (pass_refactor_eh): Convert from a global struct to a subclass of
13072         gimple_opt_pass along with...
13073         (pass_data_refactor_eh): ...new pass_data instance and...
13074         (make_pass_refactor_eh): ...new function.
13075         (pass_lower_resx): Convert from a global struct to a subclass of
13076         gimple_opt_pass along with...
13077         (pass_data_lower_resx): ...new pass_data instance and...
13078         (make_pass_lower_resx): ...new function.
13079         (pass_lower_eh_dispatch): Convert from a global struct to a subclass
13080         of gimple_opt_pass along with...
13081         (pass_data_lower_eh_dispatch): ...new pass_data instance and...
13082         (make_pass_lower_eh_dispatch): ...new function.
13083         (pass_cleanup_eh): Convert from a global struct to a subclass of
13084         gimple_opt_pass along with...
13085         (pass_data_cleanup_eh): ...new pass_data instance and...
13086         (make_pass_cleanup_eh): ...new function.
13087         * tree-emutls.c (pass_ipa_lower_emutls): Convert from a global struct
13088         to a subclass of simple_ipa_opt_pass along with...
13089         (pass_data_ipa_lower_emutls): ...new pass_data instance and...
13090         (make_pass_ipa_lower_emutls): ...new function.
13091         * tree-if-conv.c (pass_if_conversion): Convert from a global struct to
13092         a subclass of gimple_opt_pass along with...
13093         (pass_data_if_conversion): ...new pass_data instance and...
13094         (make_pass_if_conversion): ...new function.
13095         * tree-into-ssa.c (pass_build_ssa): Convert from a global struct to a
13096         subclass of gimple_opt_pass along with...
13097         (pass_data_build_ssa): ...new pass_data instance and...
13098         (make_pass_build_ssa): ...new function.
13099         * tree-loop-distribution.c (pass_loop_distribution): Convert from a
13100         global struct to a subclass of gimple_opt_pass along with...
13101         (pass_data_loop_distribution): ...new pass_data instance and...
13102         (make_pass_loop_distribution): ...new function.
13103         * tree-mudflap.c (pass_mudflap_1): Convert from a global struct to a
13104         subclass of gimple_opt_pass along with...
13105         (pass_data_mudflap_1): ...new pass_data instance and...
13106         (make_pass_mudflap_1): ...new function.
13107         (pass_mudflap_2): Convert from a global struct to a subclass of
13108         gimple_opt_pass along with...
13109         (pass_data_mudflap_2): ...new pass_data instance and...
13110         (make_pass_mudflap_2): ...new function.
13111         * tree-nomudflap.c (pass_mudflap_1): Convert from a global struct to a
13112         subclass of gimple_opt_pass along with...
13113         (pass_data_mudflap_1): ...new pass_data instance and...
13114         (make_pass_mudflap_1): ...new function.
13115         (pass_mudflap_2): Convert from a global struct to a subclass of
13116         gimple_opt_pass along with...
13117         (pass_data_mudflap_2): ...new pass_data instance and...
13118         (make_pass_mudflap_2): ...new function.
13119         * tree-nrv.c (pass_nrv): Convert from a global struct to a subclass of
13120         gimple_opt_pass along with...
13121         (pass_data_nrv): ...new pass_data instance and...
13122         (make_pass_nrv): ...new function.
13123         (pass_return_slot): Convert from a global struct to a subclass of
13124         gimple_opt_pass along with...
13125         (pass_data_return_slot): ...new pass_data instance and...
13126         (make_pass_return_slot): ...new function.
13127         * tree-object-size.c (pass_object_sizes): Convert from a global struct
13128         to a subclass of gimple_opt_pass along with...
13129         (pass_data_object_sizes): ...new pass_data instance and...
13130         (make_pass_object_sizes): ...new function.
13131         * tree-optimize.c (pass_cleanup_cfg_post_optimizing): Convert from a
13132         global struct to a subclass of gimple_opt_pass along with...
13133         (pass_data_cleanup_cfg_post_optimizing): ...new pass_data instance
13134         and...
13135         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13136         (pass_fixup_cfg): Convert from a global struct to a subclass of
13137         gimple_opt_pass along with...
13138         (pass_data_fixup_cfg): ...new pass_data instance and...
13139         (make_pass_fixup_cfg): ...new function.
13140         * tree-pass.h (pass_mudflap_1): Replace declaration with that of...
13141         (make_pass_mudflap_1): ...new function.
13142         (pass_mudflap_2): Replace declaration with that of...
13143         (make_pass_mudflap_2): ...new function.
13144         (pass_asan): Replace declaration with that of...
13145         (make_pass_asan): ...new function.
13146         (pass_asan_O0): Replace declaration with that of...
13147         (make_pass_asan_O0): ...new function.
13148         (pass_tsan): Replace declaration with that of...
13149         (make_pass_tsan): ...new function.
13150         (pass_tsan_O0): Replace declaration with that of...
13151         (make_pass_tsan_O0): ...new function.
13152         (pass_lower_cf): Replace declaration with that of...
13153         (make_pass_lower_cf): ...new function.
13154         (pass_refactor_eh): Replace declaration with that of...
13155         (make_pass_refactor_eh): ...new function.
13156         (pass_lower_eh): Replace declaration with that of...
13157         (make_pass_lower_eh): ...new function.
13158         (pass_lower_eh_dispatch): Replace declaration with that of...
13159         (make_pass_lower_eh_dispatch): ...new function.
13160         (pass_lower_resx): Replace declaration with that of...
13161         (make_pass_lower_resx): ...new function.
13162         (pass_build_cfg): Replace declaration with that of...
13163         (make_pass_build_cfg): ...new function.
13164         (pass_early_tree_profile): Replace declaration with that of...
13165         (make_pass_early_tree_profile): ...new function.
13166         (pass_cleanup_eh): Replace declaration with that of...
13167         (make_pass_cleanup_eh): ...new function.
13168         (pass_sra): Replace declaration with that of...
13169         (make_pass_sra): ...new function.
13170         (pass_sra_early): Replace declaration with that of...
13171         (make_pass_sra_early): ...new function.
13172         (pass_early_ipa_sra): Replace declaration with that of...
13173         (make_pass_early_ipa_sra): ...new function.
13174         (pass_tail_recursion): Replace declaration with that of...
13175         (make_pass_tail_recursion): ...new function.
13176         (pass_tail_calls): Replace declaration with that of...
13177         (make_pass_tail_calls): ...new function.
13178         (pass_tree_loop): Replace declaration with that of...
13179         (make_pass_tree_loop): ...new function.
13180         (pass_tree_loop_init): Replace declaration with that of...
13181         (make_pass_tree_loop_init): ...new function.
13182         (pass_lim): Replace declaration with that of...
13183         (make_pass_lim): ...new function.
13184         (pass_tree_unswitch): Replace declaration with that of...
13185         (make_pass_tree_unswitch): ...new function.
13186         (pass_predcom): Replace declaration with that of...
13187         (make_pass_predcom): ...new function.
13188         (pass_iv_canon): Replace declaration with that of...
13189         (make_pass_iv_canon): ...new function.
13190         (pass_scev_cprop): Replace declaration with that of...
13191         (make_pass_scev_cprop): ...new function.
13192         (pass_empty_loop): Replace declaration with that of...
13193         (make_pass_empty_loop): ...new function.
13194         (pass_record_bounds): Replace declaration with that of...
13195         (make_pass_record_bounds): ...new function.
13196         (pass_graphite): Replace declaration with that of...
13197         (make_pass_graphite): ...new function.
13198         (pass_graphite_transforms): Replace declaration with that of...
13199         (make_pass_graphite_transforms): ...new function.
13200         (pass_if_conversion): Replace declaration with that of...
13201         (make_pass_if_conversion): ...new function.
13202         (pass_loop_distribution): Replace declaration with that of...
13203         (make_pass_loop_distribution): ...new function.
13204         (pass_vectorize): Replace declaration with that of...
13205         (make_pass_vectorize): ...new function.
13206         (pass_slp_vectorize): Replace declaration with that of...
13207         (make_pass_slp_vectorize): ...new function.
13208         (pass_complete_unroll): Replace declaration with that of...
13209         (make_pass_complete_unroll): ...new function.
13210         (pass_complete_unrolli): Replace declaration with that of...
13211         (make_pass_complete_unrolli): ...new function.
13212         (pass_parallelize_loops): Replace declaration with that of...
13213         (make_pass_parallelize_loops): ...new function.
13214         (pass_loop_prefetch): Replace declaration with that of...
13215         (make_pass_loop_prefetch): ...new function.
13216         (pass_iv_optimize): Replace declaration with that of...
13217         (make_pass_iv_optimize): ...new function.
13218         (pass_tree_loop_done): Replace declaration with that of...
13219         (make_pass_tree_loop_done): ...new function.
13220         (pass_ch): Replace declaration with that of...
13221         (make_pass_ch): ...new function.
13222         (pass_ccp): Replace declaration with that of...
13223         (make_pass_ccp): ...new function.
13224         (pass_phi_only_cprop): Replace declaration with that of...
13225         (make_pass_phi_only_cprop): ...new function.
13226         (pass_build_ssa): Replace declaration with that of...
13227         (make_pass_build_ssa): ...new function.
13228         (pass_build_alias): Replace declaration with that of...
13229         (make_pass_build_alias): ...new function.
13230         (pass_build_ealias): Replace declaration with that of...
13231         (make_pass_build_ealias): ...new function.
13232         (pass_dominator): Replace declaration with that of...
13233         (make_pass_dominator): ...new function.
13234         (pass_dce): Replace declaration with that of...
13235         (make_pass_dce): ...new function.
13236         (pass_dce_loop): Replace declaration with that of...
13237         (make_pass_dce_loop): ...new function.
13238         (pass_cd_dce): Replace declaration with that of...
13239         (make_pass_cd_dce): ...new function.
13240         (pass_call_cdce): Replace declaration with that of...
13241         (make_pass_call_cdce): ...new function.
13242         (pass_merge_phi): Replace declaration with that of...
13243         (make_pass_merge_phi): ...new function.
13244         (pass_split_crit_edges): Replace declaration with that of...
13245         (make_pass_split_crit_edges): ...new function.
13246         (pass_pre): Replace declaration with that of...
13247         (make_pass_pre): ...new function.
13248         (pass_profile): Replace declaration with that of...
13249         (make_pass_profile): ...new function.
13250         (pass_strip_predict_hints): Replace declaration with that of...
13251         (make_pass_strip_predict_hints): ...new function.
13252         (pass_lower_complex_O0): Replace declaration with that of...
13253         (make_pass_lower_complex_O0): ...new function.
13254         (pass_lower_complex): Replace declaration with that of...
13255         (make_pass_lower_complex): ...new function.
13256         (pass_lower_vector): Replace declaration with that of...
13257         (make_pass_lower_vector): ...new function.
13258         (pass_lower_vector_ssa): Replace declaration with that of...
13259         (make_pass_lower_vector_ssa): ...new function.
13260         (pass_lower_omp): Replace declaration with that of...
13261         (make_pass_lower_omp): ...new function.
13262         (pass_diagnose_omp_blocks): Replace declaration with that of...
13263         (make_pass_diagnose_omp_blocks): ...new function.
13264         (pass_expand_omp): Replace declaration with that of...
13265         (make_pass_expand_omp): ...new function.
13266         (pass_expand_omp_ssa): Replace declaration with that of...
13267         (make_pass_expand_omp_ssa): ...new function.
13268         (pass_object_sizes): Replace declaration with that of...
13269         (make_pass_object_sizes): ...new function.
13270         (pass_strlen): Replace declaration with that of...
13271         (make_pass_strlen): ...new function.
13272         (pass_fold_builtins): Replace declaration with that of...
13273         (make_pass_fold_builtins): ...new function.
13274         (pass_stdarg): Replace declaration with that of...
13275         (make_pass_stdarg): ...new function.
13276         (pass_early_warn_uninitialized): Replace declaration with that of...
13277         (make_pass_early_warn_uninitialized): ...new function.
13278         (pass_late_warn_uninitialized): Replace declaration with that of...
13279         (make_pass_late_warn_uninitialized): ...new function.
13280         (pass_cse_reciprocals): Replace declaration with that of...
13281         (make_pass_cse_reciprocals): ...new function.
13282         (pass_cse_sincos): Replace declaration with that of...
13283         (make_pass_cse_sincos): ...new function.
13284         (pass_optimize_bswap): Replace declaration with that of...
13285         (make_pass_optimize_bswap): ...new function.
13286         (pass_optimize_widening_mul): Replace declaration with that of...
13287         (make_pass_optimize_widening_mul): ...new function.
13288         (pass_warn_function_return): Replace declaration with that of...
13289         (make_pass_warn_function_return): ...new function.
13290         (pass_warn_function_noreturn): Replace declaration with that of...
13291         (make_pass_warn_function_noreturn): ...new function.
13292         (pass_cselim): Replace declaration with that of...
13293         (make_pass_cselim): ...new function.
13294         (pass_phiopt): Replace declaration with that of...
13295         (make_pass_phiopt): ...new function.
13296         (pass_forwprop): Replace declaration with that of...
13297         (make_pass_forwprop): ...new function.
13298         (pass_phiprop): Replace declaration with that of...
13299         (make_pass_phiprop): ...new function.
13300         (pass_tree_ifcombine): Replace declaration with that of...
13301         (make_pass_tree_ifcombine): ...new function.
13302         (pass_dse): Replace declaration with that of...
13303         (make_pass_dse): ...new function.
13304         (pass_nrv): Replace declaration with that of...
13305         (make_pass_nrv): ...new function.
13306         (pass_rename_ssa_copies): Replace declaration with that of...
13307         (make_pass_rename_ssa_copies): ...new function.
13308         (pass_sink_code): Replace declaration with that of...
13309         (make_pass_sink_code): ...new function.
13310         (pass_fre): Replace declaration with that of...
13311         (make_pass_fre): ...new function.
13312         (pass_check_data_deps): Replace declaration with that of...
13313         (make_pass_check_data_deps): ...new function.
13314         (pass_copy_prop): Replace declaration with that of...
13315         (make_pass_copy_prop): ...new function.
13316         (pass_vrp): Replace declaration with that of...
13317         (make_pass_vrp): ...new function.
13318         (pass_uncprop): Replace declaration with that of...
13319         (make_pass_uncprop): ...new function.
13320         (pass_return_slot): Replace declaration with that of...
13321         (make_pass_return_slot): ...new function.
13322         (pass_reassoc): Replace declaration with that of...
13323         (make_pass_reassoc): ...new function.
13324         (pass_rebuild_cgraph_edges): Replace declaration with that of...
13325         (make_pass_rebuild_cgraph_edges): ...new function.
13326         (pass_remove_cgraph_callee_edges): Replace declaration with that of...
13327         (make_pass_remove_cgraph_callee_edges): ...new function.
13328         (pass_build_cgraph_edges): Replace declaration with that of...
13329         (make_pass_build_cgraph_edges): ...new function.
13330         (pass_local_pure_const): Replace declaration with that of...
13331         (make_pass_local_pure_const): ...new function.
13332         (pass_tracer): Replace declaration with that of...
13333         (make_pass_tracer): ...new function.
13334         (pass_warn_unused_result): Replace declaration with that of...
13335         (make_pass_warn_unused_result): ...new function.
13336         (pass_diagnose_tm_blocks): Replace declaration with that of...
13337         (make_pass_diagnose_tm_blocks): ...new function.
13338         (pass_lower_tm): Replace declaration with that of...
13339         (make_pass_lower_tm): ...new function.
13340         (pass_tm_init): Replace declaration with that of...
13341         (make_pass_tm_init): ...new function.
13342         (pass_tm_mark): Replace declaration with that of...
13343         (make_pass_tm_mark): ...new function.
13344         (pass_tm_memopt): Replace declaration with that of...
13345         (make_pass_tm_memopt): ...new function.
13346         (pass_tm_edges): Replace declaration with that of...
13347         (make_pass_tm_edges): ...new function.
13348         (pass_split_functions): Replace declaration with that of...
13349         (make_pass_split_functions): ...new function.
13350         (pass_feedback_split_functions): Replace declaration with that of...
13351         (make_pass_feedback_split_functions): ...new function.
13352         (pass_strength_reduction): Replace declaration with that of...
13353         (make_pass_strength_reduction): ...new function.
13354         (pass_ipa_lower_emutls): Replace declaration with that of...
13355         (make_pass_ipa_lower_emutls): ...new function.
13356         (pass_ipa_function_and_variable_visibility): Replace declaration with
13357         that of...
13358         (make_pass_ipa_function_and_variable_visibility): ...new function.
13359         (pass_ipa_tree_profile): Replace declaration with that of...
13360         (make_pass_ipa_tree_profile): ...new function.
13361         (pass_early_local_passes): Replace declaration with that of...
13362         (make_pass_early_local_passes): ...new function.
13363         (pass_ipa_whole_program_visibility): Replace declaration with that
13364         of...
13365         (make_pass_ipa_whole_program_visibility): ...new function.
13366         (pass_ipa_lto_gimple_out): Replace declaration with that of...
13367         (make_pass_ipa_lto_gimple_out): ...new function.
13368         (pass_ipa_increase_alignment): Replace declaration with that of...
13369         (make_pass_ipa_increase_alignment): ...new function.
13370         (pass_ipa_inline): Replace declaration with that of...
13371         (make_pass_ipa_inline): ...new function.
13372         (pass_ipa_free_lang_data): Replace declaration with that of...
13373         (make_pass_ipa_free_lang_data): ...new function.
13374         (pass_ipa_free_inline_summary): Replace declaration with that of...
13375         (make_pass_ipa_free_inline_summary): ...new function.
13376         (pass_ipa_cp): Replace declaration with that of...
13377         (make_pass_ipa_cp): ...new function.
13378         (pass_ipa_reference): Replace declaration with that of...
13379         (make_pass_ipa_reference): ...new function.
13380         (pass_ipa_pure_const): Replace declaration with that of...
13381         (make_pass_ipa_pure_const): ...new function.
13382         (pass_ipa_pta): Replace declaration with that of...
13383         (make_pass_ipa_pta): ...new function.
13384         (pass_ipa_lto_finish_out): Replace declaration with that of...
13385         (make_pass_ipa_lto_finish_out): ...new function.
13386         (pass_ipa_tm): Replace declaration with that of...
13387         (make_pass_ipa_tm): ...new function.
13388         (pass_ipa_profile): Replace declaration with that of...
13389         (make_pass_ipa_profile): ...new function.
13390         (pass_ipa_cdtor_merge): Replace declaration with that of...
13391         (make_pass_ipa_cdtor_merge): ...new function.
13392         (pass_cleanup_cfg_post_optimizing): Replace declaration with that
13393         of...
13394         (make_pass_cleanup_cfg_post_optimizing): ...new function.
13395         (pass_init_datastructures): Replace declaration with that of...
13396         (make_pass_init_datastructures): ...new function.
13397         (pass_fixup_cfg): Replace declaration with that of...
13398         (make_pass_fixup_cfg): ...new function.
13399         (pass_expand): Replace declaration with that of...
13400         (make_pass_expand): ...new function.
13401         (pass_instantiate_virtual_regs): Replace declaration with that of...
13402         (make_pass_instantiate_virtual_regs): ...new function.
13403         (pass_rtl_fwprop): Replace declaration with that of...
13404         (make_pass_rtl_fwprop): ...new function.
13405         (pass_rtl_fwprop_addr): Replace declaration with that of...
13406         (make_pass_rtl_fwprop_addr): ...new function.
13407         (pass_jump): Replace declaration with that of...
13408         (make_pass_jump): ...new function.
13409         (pass_jump2): Replace declaration with that of...
13410         (make_pass_jump2): ...new function.
13411         (pass_lower_subreg): Replace declaration with that of...
13412         (make_pass_lower_subreg): ...new function.
13413         (pass_cse): Replace declaration with that of...
13414         (make_pass_cse): ...new function.
13415         (pass_fast_rtl_dce): Replace declaration with that of...
13416         (make_pass_fast_rtl_dce): ...new function.
13417         (pass_ud_rtl_dce): Replace declaration with that of...
13418         (make_pass_ud_rtl_dce): ...new function.
13419         (pass_rtl_dce): Replace declaration with that of...
13420         (make_pass_rtl_dce): ...new function.
13421         (pass_rtl_dse1): Replace declaration with that of...
13422         (make_pass_rtl_dse1): ...new function.
13423         (pass_rtl_dse2): Replace declaration with that of...
13424         (make_pass_rtl_dse2): ...new function.
13425         (pass_rtl_dse3): Replace declaration with that of...
13426         (make_pass_rtl_dse3): ...new function.
13427         (pass_rtl_cprop): Replace declaration with that of...
13428         (make_pass_rtl_cprop): ...new function.
13429         (pass_rtl_pre): Replace declaration with that of...
13430         (make_pass_rtl_pre): ...new function.
13431         (pass_rtl_hoist): Replace declaration with that of...
13432         (make_pass_rtl_hoist): ...new function.
13433         (pass_rtl_store_motion): Replace declaration with that of...
13434         (make_pass_rtl_store_motion): ...new function.
13435         (pass_cse_after_global_opts): Replace declaration with that of...
13436         (make_pass_cse_after_global_opts): ...new function.
13437         (pass_rtl_ifcvt): Replace declaration with that of...
13438         (make_pass_rtl_ifcvt): ...new function.
13439         (pass_into_cfg_layout_mode): Replace declaration with that of...
13440         (make_pass_into_cfg_layout_mode): ...new function.
13441         (pass_outof_cfg_layout_mode): Replace declaration with that of...
13442         (make_pass_outof_cfg_layout_mode): ...new function.
13443         (pass_loop2): Replace declaration with that of...
13444         (make_pass_loop2): ...new function.
13445         (pass_rtl_loop_init): Replace declaration with that of...
13446         (make_pass_rtl_loop_init): ...new function.
13447         (pass_rtl_move_loop_invariants): Replace declaration with that of...
13448         (make_pass_rtl_move_loop_invariants): ...new function.
13449         (pass_rtl_unswitch): Replace declaration with that of...
13450         (make_pass_rtl_unswitch): ...new function.
13451         (pass_rtl_unroll_and_peel_loops): Replace declaration with that of...
13452         (make_pass_rtl_unroll_and_peel_loops): ...new function.
13453         (pass_rtl_doloop): Replace declaration with that of...
13454         (make_pass_rtl_doloop): ...new function.
13455         (pass_rtl_loop_done): Replace declaration with that of...
13456         (make_pass_rtl_loop_done): ...new function.
13457         (pass_web): Replace declaration with that of...
13458         (make_pass_web): ...new function.
13459         (pass_cse2): Replace declaration with that of...
13460         (make_pass_cse2): ...new function.
13461         (pass_df_initialize_opt): Replace declaration with that of...
13462         (make_pass_df_initialize_opt): ...new function.
13463         (pass_df_initialize_no_opt): Replace declaration with that of...
13464         (make_pass_df_initialize_no_opt): ...new function.
13465         (pass_reginfo_init): Replace declaration with that of...
13466         (make_pass_reginfo_init): ...new function.
13467         (pass_inc_dec): Replace declaration with that of...
13468         (make_pass_inc_dec): ...new function.
13469         (pass_stack_ptr_mod): Replace declaration with that of...
13470         (make_pass_stack_ptr_mod): ...new function.
13471         (pass_initialize_regs): Replace declaration with that of...
13472         (make_pass_initialize_regs): ...new function.
13473         (pass_combine): Replace declaration with that of...
13474         (make_pass_combine): ...new function.
13475         (pass_if_after_combine): Replace declaration with that of...
13476         (make_pass_if_after_combine): ...new function.
13477         (pass_ree): Replace declaration with that of...
13478         (make_pass_ree): ...new function.
13479         (pass_partition_blocks): Replace declaration with that of...
13480         (make_pass_partition_blocks): ...new function.
13481         (pass_match_asm_constraints): Replace declaration with that of...
13482         (make_pass_match_asm_constraints): ...new function.
13483         (pass_regmove): Replace declaration with that of...
13484         (make_pass_regmove): ...new function.
13485         (pass_split_all_insns): Replace declaration with that of...
13486         (make_pass_split_all_insns): ...new function.
13487         (pass_fast_rtl_byte_dce): Replace declaration with that of...
13488         (make_pass_fast_rtl_byte_dce): ...new function.
13489         (pass_lower_subreg2): Replace declaration with that of...
13490         (make_pass_lower_subreg2): ...new function.
13491         (pass_mode_switching): Replace declaration with that of...
13492         (make_pass_mode_switching): ...new function.
13493         (pass_sms): Replace declaration with that of...
13494         (make_pass_sms): ...new function.
13495         (pass_sched): Replace declaration with that of...
13496         (make_pass_sched): ...new function.
13497         (pass_ira): Replace declaration with that of...
13498         (make_pass_ira): ...new function.
13499         (pass_reload): Replace declaration with that of...
13500         (make_pass_reload): ...new function.
13501         (pass_clean_state): Replace declaration with that of...
13502         (make_pass_clean_state): ...new function.
13503         (pass_branch_prob): Replace declaration with that of...
13504         (make_pass_branch_prob): ...new function.
13505         (pass_value_profile_transformations): Replace declaration with that
13506         of...
13507         (make_pass_value_profile_transformations): ...new function.
13508         (pass_postreload_cse): Replace declaration with that of...
13509         (make_pass_postreload_cse): ...new function.
13510         (pass_gcse2): Replace declaration with that of...
13511         (make_pass_gcse2): ...new function.
13512         (pass_split_after_reload): Replace declaration with that of...
13513         (make_pass_split_after_reload): ...new function.
13514         (pass_branch_target_load_optimize1): Replace declaration with that
13515         of...
13516         (make_pass_branch_target_load_optimize1): ...new function.
13517         (pass_thread_prologue_and_epilogue): Replace declaration with that
13518         of...
13519         (make_pass_thread_prologue_and_epilogue): ...new function.
13520         (pass_stack_adjustments): Replace declaration with that of...
13521         (make_pass_stack_adjustments): ...new function.
13522         (pass_peephole2): Replace declaration with that of...
13523         (make_pass_peephole2): ...new function.
13524         (pass_if_after_reload): Replace declaration with that of...
13525         (make_pass_if_after_reload): ...new function.
13526         (pass_regrename): Replace declaration with that of...
13527         (make_pass_regrename): ...new function.
13528         (pass_cprop_hardreg): Replace declaration with that of...
13529         (make_pass_cprop_hardreg): ...new function.
13530         (pass_reorder_blocks): Replace declaration with that of...
13531         (make_pass_reorder_blocks): ...new function.
13532         (pass_branch_target_load_optimize2): Replace declaration with that
13533         of...
13534         (make_pass_branch_target_load_optimize2): ...new function.
13535         (pass_leaf_regs): Replace declaration with that of...
13536         (make_pass_leaf_regs): ...new function.
13537         (pass_split_before_sched2): Replace declaration with that of...
13538         (make_pass_split_before_sched2): ...new function.
13539         (pass_compare_elim_after_reload): Replace declaration with that of...
13540         (make_pass_compare_elim_after_reload): ...new function.
13541         (pass_sched2): Replace declaration with that of...
13542         (make_pass_sched2): ...new function.
13543         (pass_stack_regs): Replace declaration with that of...
13544         (make_pass_stack_regs): ...new function.
13545         (pass_stack_regs_run): Replace declaration with that of...
13546         (make_pass_stack_regs_run): ...new function.
13547         (pass_df_finish): Replace declaration with that of...
13548         (make_pass_df_finish): ...new function.
13549         (pass_compute_alignments): Replace declaration with that of...
13550         (make_pass_compute_alignments): ...new function.
13551         (pass_duplicate_computed_gotos): Replace declaration with that of...
13552         (make_pass_duplicate_computed_gotos): ...new function.
13553         (pass_variable_tracking): Replace declaration with that of...
13554         (make_pass_variable_tracking): ...new function.
13555         (pass_free_cfg): Replace declaration with that of...
13556         (make_pass_free_cfg): ...new function.
13557         (pass_machine_reorg): Replace declaration with that of...
13558         (make_pass_machine_reorg): ...new function.
13559         (pass_cleanup_barriers): Replace declaration with that of...
13560         (make_pass_cleanup_barriers): ...new function.
13561         (pass_delay_slots): Replace declaration with that of...
13562         (make_pass_delay_slots): ...new function.
13563         (pass_split_for_shorten_branches): Replace declaration with that of...
13564         (make_pass_split_for_shorten_branches): ...new function.
13565         (pass_split_before_regstack): Replace declaration with that of...
13566         (make_pass_split_before_regstack): ...new function.
13567         (pass_convert_to_eh_region_ranges): Replace declaration with that
13568         of...
13569         (make_pass_convert_to_eh_region_ranges): ...new function.
13570         (pass_shorten_branches): Replace declaration with that of...
13571         (make_pass_shorten_branches): ...new function.
13572         (pass_set_nothrow_function_flags): Replace declaration with that of...
13573         (make_pass_set_nothrow_function_flags): ...new function.
13574         (pass_dwarf2_frame): Replace declaration with that of...
13575         (make_pass_dwarf2_frame): ...new function.
13576         (pass_final): Replace declaration with that of...
13577         (make_pass_final): ...new function.
13578         (pass_rtl_seqabstr): Replace declaration with that of...
13579         (make_pass_rtl_seqabstr): ...new function.
13580         (pass_release_ssa_names): Replace declaration with that of...
13581         (make_pass_release_ssa_names): ...new function.
13582         (pass_early_inline): Replace declaration with that of...
13583         (make_pass_early_inline): ...new function.
13584         (pass_inline_parameters): Replace declaration with that of...
13585         (make_pass_inline_parameters): ...new function.
13586         (pass_update_address_taken): Replace declaration with that of...
13587         (make_pass_update_address_taken): ...new function.
13588         (pass_convert_switch): Replace declaration with that of...
13589         (make_pass_convert_switch): ...new function.
13590         * tree-profile.c (pass_ipa_tree_profile): Convert from a global struct
13591         to a subclass of simple_ipa_opt_pass along with...
13592         (pass_data_ipa_tree_profile): ...new pass_data instance and...
13593         (make_pass_ipa_tree_profile): ...new function.
13594         * tree-sra.c (pass_sra_early): Convert from a global struct to a
13595         subclass of gimple_opt_pass along with...
13596         (pass_data_sra_early): ...new pass_data instance and...
13597         (make_pass_sra_early): ...new function.
13598         (pass_sra): Convert from a global struct to a subclass of
13599         gimple_opt_pass along with...
13600         (pass_data_sra): ...new pass_data instance and...
13601         (make_pass_sra): ...new function.
13602         (pass_early_ipa_sra): Convert from a global struct to a subclass of
13603         gimple_opt_pass along with...
13604         (pass_data_early_ipa_sra): ...new pass_data instance and...
13605         (make_pass_early_ipa_sra): ...new function.
13606         * tree-ssa-ccp.c (pass_ccp): Convert from a global struct to a
13607         subclass of gimple_opt_pass along with...
13608         (pass_data_ccp): ...new pass_data instance and...
13609         (make_pass_ccp): ...new function.
13610         (pass_fold_builtins): Convert from a global struct to a subclass of
13611         gimple_opt_pass along with...
13612         (pass_data_fold_builtins): ...new pass_data instance and...
13613         (make_pass_fold_builtins): ...new function.
13614         * tree-ssa-copy.c (pass_copy_prop): Convert from a global struct to a
13615         subclass of gimple_opt_pass along with...
13616         (pass_data_copy_prop): ...new pass_data instance and...
13617         (make_pass_copy_prop): ...new function.
13618         * tree-ssa-copyrename.c (pass_rename_ssa_copies): Convert from a
13619         global struct to a subclass of gimple_opt_pass along with...
13620         (pass_data_rename_ssa_copies): ...new pass_data instance and...
13621         (make_pass_rename_ssa_copies): ...new function.
13622         * tree-ssa-dce.c (pass_dce): Convert from a global struct to a
13623         subclass of gimple_opt_pass along with...
13624         (pass_data_dce): ...new pass_data instance and...
13625         (make_pass_dce): ...new function.
13626         (pass_dce_loop): Convert from a global struct to a subclass of
13627         gimple_opt_pass along with...
13628         (pass_data_dce_loop): ...new pass_data instance and...
13629         (make_pass_dce_loop): ...new function.
13630         (pass_cd_dce): Convert from a global struct to a subclass of
13631         gimple_opt_pass along with...
13632         (pass_data_cd_dce): ...new pass_data instance and...
13633         (make_pass_cd_dce): ...new function.
13634         * tree-ssa-dom.c (pass_dominator): Convert from a global struct to a
13635         subclass of gimple_opt_pass along with...
13636         (pass_data_dominator): ...new pass_data instance and...
13637         (make_pass_dominator): ...new function.
13638         (pass_phi_only_cprop): Convert from a global struct to a subclass of
13639         gimple_opt_pass along with...
13640         (pass_data_phi_only_cprop): ...new pass_data instance and...
13641         (make_pass_phi_only_cprop): ...new function.
13642         * tree-ssa-dse.c (pass_dse): Convert from a global struct to a
13643         subclass of gimple_opt_pass along with...
13644         (pass_data_dse): ...new pass_data instance and...
13645         (make_pass_dse): ...new function.
13646         * tree-ssa-forwprop.c (pass_forwprop): Convert from a global struct to
13647         a subclass of gimple_opt_pass along with...
13648         (pass_data_forwprop): ...new pass_data instance and...
13649         (make_pass_forwprop): ...new function.
13650         * tree-ssa-ifcombine.c (pass_tree_ifcombine): Convert from a global
13651         struct to a subclass of gimple_opt_pass along with...
13652         (pass_data_tree_ifcombine): ...new pass_data instance and...
13653         (make_pass_tree_ifcombine): ...new function.
13654         * tree-ssa-loop-ch.c (pass_ch): Convert from a global struct to a
13655         subclass of gimple_opt_pass along with...
13656         (pass_data_ch): ...new pass_data instance and...
13657         (make_pass_ch): ...new function.
13658         * tree-ssa-loop.c (pass_tree_loop): Convert from a global struct to a
13659         subclass of gimple_opt_pass along with...
13660         (pass_data_tree_loop): ...new pass_data instance and...
13661         (make_pass_tree_loop): ...new function.
13662         (pass_tree_loop_init): Convert from a global struct to a subclass of
13663         gimple_opt_pass along with...
13664         (pass_data_tree_loop_init): ...new pass_data instance and...
13665         (make_pass_tree_loop_init): ...new function.
13666         (pass_lim): Convert from a global struct to a subclass of
13667         gimple_opt_pass along with...
13668         (pass_data_lim): ...new pass_data instance and...
13669         (make_pass_lim): ...new function.
13670         (pass_tree_unswitch): Convert from a global struct to a subclass of
13671         gimple_opt_pass along with...
13672         (pass_data_tree_unswitch): ...new pass_data instance and...
13673         (make_pass_tree_unswitch): ...new function.
13674         (pass_predcom): Convert from a global struct to a subclass of
13675         gimple_opt_pass along with...
13676         (pass_data_predcom): ...new pass_data instance and...
13677         (make_pass_predcom): ...new function.
13678         (pass_vectorize): Convert from a global struct to a subclass of
13679         gimple_opt_pass along with...
13680         (pass_data_vectorize): ...new pass_data instance and...
13681         (make_pass_vectorize): ...new function.
13682         (pass_graphite): Convert from a global struct to a subclass of
13683         gimple_opt_pass along with...
13684         (pass_data_graphite): ...new pass_data instance and...
13685         (make_pass_graphite): ...new function.
13686         (pass_graphite_transforms): Convert from a global struct to a subclass
13687         of gimple_opt_pass along with...
13688         (pass_data_graphite_transforms): ...new pass_data instance and...
13689         (make_pass_graphite_transforms): ...new function.
13690         (pass_check_data_deps): Convert from a global struct to a subclass of
13691         gimple_opt_pass along with...
13692         (pass_data_check_data_deps): ...new pass_data instance and...
13693         (make_pass_check_data_deps): ...new function.
13694         (pass_iv_canon): Convert from a global struct to a subclass of
13695         gimple_opt_pass along with...
13696         (pass_data_iv_canon): ...new pass_data instance and...
13697         (make_pass_iv_canon): ...new function.
13698         (pass_scev_cprop): Convert from a global struct to a subclass of
13699         gimple_opt_pass along with...
13700         (pass_data_scev_cprop): ...new pass_data instance and...
13701         (make_pass_scev_cprop): ...new function.
13702         (pass_record_bounds): Convert from a global struct to a subclass of
13703         gimple_opt_pass along with...
13704         (pass_data_record_bounds): ...new pass_data instance and...
13705         (make_pass_record_bounds): ...new function.
13706         (pass_complete_unroll): Convert from a global struct to a subclass of
13707         gimple_opt_pass along with...
13708         (pass_data_complete_unroll): ...new pass_data instance and...
13709         (make_pass_complete_unroll): ...new function.
13710         (pass_complete_unrolli): Convert from a global struct to a subclass of
13711         gimple_opt_pass along with...
13712         (pass_data_complete_unrolli): ...new pass_data instance and...
13713         (make_pass_complete_unrolli): ...new function.
13714         (pass_parallelize_loops): Convert from a global struct to a subclass
13715         of gimple_opt_pass along with...
13716         (pass_data_parallelize_loops): ...new pass_data instance and...
13717         (make_pass_parallelize_loops): ...new function.
13718         (pass_loop_prefetch): Convert from a global struct to a subclass of
13719         gimple_opt_pass along with...
13720         (pass_data_loop_prefetch): ...new pass_data instance and...
13721         (make_pass_loop_prefetch): ...new function.
13722         (pass_iv_optimize): Convert from a global struct to a subclass of
13723         gimple_opt_pass along with...
13724         (pass_data_iv_optimize): ...new pass_data instance and...
13725         (make_pass_iv_optimize): ...new function.
13726         (pass_tree_loop_done): Convert from a global struct to a subclass of
13727         gimple_opt_pass along with...
13728         (pass_data_tree_loop_done): ...new pass_data instance and...
13729         (make_pass_tree_loop_done): ...new function.
13730         * tree-ssa-math-opts.c (pass_cse_reciprocals): Convert from a global
13731         struct to a subclass of gimple_opt_pass along with...
13732         (pass_data_cse_reciprocals): ...new pass_data instance and...
13733         (make_pass_cse_reciprocals): ...new function.
13734         (pass_cse_sincos): Convert from a global struct to a subclass of
13735         gimple_opt_pass along with...
13736         (pass_data_cse_sincos): ...new pass_data instance and...
13737         (make_pass_cse_sincos): ...new function.
13738         (pass_optimize_bswap): Convert from a global struct to a subclass of
13739         gimple_opt_pass along with...
13740         (pass_data_optimize_bswap): ...new pass_data instance and...
13741         (make_pass_optimize_bswap): ...new function.
13742         (pass_optimize_widening_mul): Convert from a global struct to a
13743         subclass of gimple_opt_pass along with...
13744         (pass_data_optimize_widening_mul): ...new pass_data instance and...
13745         (make_pass_optimize_widening_mul): ...new function.
13746         * tree-ssa-phiopt.c (pass_phiopt): Convert from a global struct to a
13747         subclass of gimple_opt_pass along with...
13748         (pass_data_phiopt): ...new pass_data instance and...
13749         (make_pass_phiopt): ...new function.
13750         (pass_cselim): Convert from a global struct to a subclass of
13751         gimple_opt_pass along with...
13752         (pass_data_cselim): ...new pass_data instance and...
13753         (make_pass_cselim): ...new function.
13754         * tree-ssa-phiprop.c (pass_phiprop): Convert from a global struct to a
13755         subclass of gimple_opt_pass along with...
13756         (pass_data_phiprop): ...new pass_data instance and...
13757         (make_pass_phiprop): ...new function.
13758         * tree-ssa-pre.c (pass_pre): Convert from a global struct to a
13759         subclass of gimple_opt_pass along with...
13760         (pass_data_pre): ...new pass_data instance and...
13761         (make_pass_pre): ...new function.
13762         (pass_fre): Convert from a global struct to a subclass of
13763         gimple_opt_pass along with...
13764         (pass_data_fre): ...new pass_data instance and...
13765         (make_pass_fre): ...new function.
13766         * tree-ssa-reassoc.c (pass_reassoc): Convert from a global struct to a
13767         subclass of gimple_opt_pass along with...
13768         (pass_data_reassoc): ...new pass_data instance and...
13769         (make_pass_reassoc): ...new function.
13770         * tree-ssa-sink.c (pass_sink_code): Convert from a global struct to a
13771         subclass of gimple_opt_pass along with...
13772         (pass_data_sink_code): ...new pass_data instance and...
13773         (make_pass_sink_code): ...new function.
13774         * tree-ssa-strlen.c (pass_strlen): Convert from a global struct to a
13775         subclass of gimple_opt_pass along with...
13776         (pass_data_strlen): ...new pass_data instance and...
13777         (make_pass_strlen): ...new function.
13778         * tree-ssa-structalias.c (pass_build_alias): Convert from a global
13779         struct to a subclass of gimple_opt_pass along with...
13780         (pass_data_build_alias): ...new pass_data instance and...
13781         (make_pass_build_alias): ...new function.
13782         (pass_build_ealias): Convert from a global struct to a subclass of
13783         gimple_opt_pass along with...
13784         (pass_data_build_ealias): ...new pass_data instance and...
13785         (make_pass_build_ealias): ...new function.
13786         (pass_ipa_pta): Convert from a global struct to a subclass of
13787         simple_ipa_opt_pass along with...
13788         (pass_data_ipa_pta): ...new pass_data instance and...
13789         (make_pass_ipa_pta): ...new function.
13790         * tree-ssa-uncprop.c (pass_uncprop): Convert from a global struct to a
13791         subclass of gimple_opt_pass along with...
13792         (pass_data_uncprop): ...new pass_data instance and...
13793         (make_pass_uncprop): ...new function.
13794         * tree-ssa-uninit.c (pass_late_warn_uninitialized): Convert from a
13795         global struct to a subclass of gimple_opt_pass along with...
13796         (pass_data_late_warn_uninitialized): ...new pass_data instance and...
13797         (make_pass_late_warn_uninitialized): ...new function.
13798         * tree-ssa.c (pass_init_datastructures): Convert from a global struct
13799         to a subclass of gimple_opt_pass along with...
13800         (pass_data_init_datastructures): ...new pass_data instance and...
13801         (make_pass_init_datastructures): ...new function.
13802         (pass_early_warn_uninitialized): Convert from a global struct to a
13803         subclass of gimple_opt_pass along with...
13804         (pass_data_early_warn_uninitialized): ...new pass_data instance and...
13805         (make_pass_early_warn_uninitialized): ...new function.
13806         (pass_update_address_taken): Convert from a global struct to a
13807         subclass of gimple_opt_pass along with...
13808         (pass_data_update_address_taken): ...new pass_data instance and...
13809         (make_pass_update_address_taken): ...new function.
13810         * tree-ssanames.c (pass_release_ssa_names): Convert from a global
13811         struct to a subclass of gimple_opt_pass along with...
13812         (pass_data_release_ssa_names): ...new pass_data instance and...
13813         (make_pass_release_ssa_names): ...new function.
13814         * tree-stdarg.c (pass_stdarg): Convert from a global struct to a
13815         subclass of gimple_opt_pass along with...
13816         (pass_data_stdarg): ...new pass_data instance and...
13817         (make_pass_stdarg): ...new function.
13818         * tree-switch-conversion.c (pass_convert_switch): Convert from a
13819         global struct to a subclass of gimple_opt_pass along with...
13820         (pass_data_convert_switch): ...new pass_data instance and...
13821         (make_pass_convert_switch): ...new function.
13822         * tree-tailcall.c (pass_tail_recursion): Convert from a global struct
13823         to a subclass of gimple_opt_pass along with...
13824         (pass_data_tail_recursion): ...new pass_data instance and...
13825         (make_pass_tail_recursion): ...new function.
13826         (pass_tail_calls): Convert from a global struct to a subclass of
13827         gimple_opt_pass along with...
13828         (pass_data_tail_calls): ...new pass_data instance and...
13829         (make_pass_tail_calls): ...new function.
13830         * tree-vect-generic.c (pass_lower_vector): Convert from a global
13831         struct to a subclass of gimple_opt_pass along with...
13832         (pass_data_lower_vector): ...new pass_data instance and...
13833         (make_pass_lower_vector): ...new function.
13834         (pass_lower_vector_ssa): Convert from a global struct to a subclass of
13835         gimple_opt_pass along with...
13836         (pass_data_lower_vector_ssa): ...new pass_data instance and...
13837         (make_pass_lower_vector_ssa): ...new function.
13838         * tree-vectorizer.c (pass_slp_vectorize): Convert from a global struct
13839         to a subclass of gimple_opt_pass along with...
13840         (pass_data_slp_vectorize): ...new pass_data instance and...
13841         (make_pass_slp_vectorize): ...new function.
13842         (pass_ipa_increase_alignment): Convert from a global struct to a
13843         subclass of simple_ipa_opt_pass along with...
13844         (pass_data_ipa_increase_alignment): ...new pass_data instance and...
13845         (make_pass_ipa_increase_alignment): ...new function.
13846         * tree-vrp.c (pass_vrp): Convert from a global struct to a subclass of
13847         gimple_opt_pass along with...
13848         (pass_data_vrp): ...new pass_data instance and...
13849         (make_pass_vrp): ...new function.
13850         * tree.c (pass_ipa_free_lang_data): Convert from a global struct to a
13851         subclass of simple_ipa_opt_pass along with...
13852         (pass_data_ipa_free_lang_data): ...new pass_data instance and...
13853         (make_pass_ipa_free_lang_data): ...new function.
13854         * tsan.c (pass_tsan): Convert from a global struct to a subclass of
13855         gimple_opt_pass along with...
13856         (pass_data_tsan): ...new pass_data instance and...
13857         (make_pass_tsan): ...new function.
13858         (pass_tsan_O0): Convert from a global struct to a subclass of
13859         gimple_opt_pass along with...
13860         (pass_data_tsan_O0): ...new pass_data instance and...
13861         (make_pass_tsan_O0): ...new function.
13862         * var-tracking.c (pass_variable_tracking): Convert from a global
13863         struct to a subclass of rtl_opt_pass along with...
13864         (pass_data_variable_tracking): ...new pass_data instance and...
13865         (make_pass_variable_tracking): ...new function.
13866         * web.c (pass_web): Convert from a global struct to a subclass of
13867         rtl_opt_pass along with...
13868         (pass_data_web): ...new pass_data instance and...
13869         (make_pass_web): ...new function.
13870         * config/epiphany/epiphany.h (pass_mode_switch_use): Replace
13871         declaration with that of...
13872         (make_pass_mode_switch_use): ...new function.
13873         (pass_resolve_sw_modes): Replace declaration with that of...
13874         (make_pass_resolve_sw_modes): ...new function.
13875         * config/epiphany/mode-switch-use.c (pass_mode_switch_use): Convert
13876         from a global struct to a subclass of rtl_opt_pass along with...
13877         (pass_data_mode_switch_use): ...new pass_data instance and...
13878         (make_pass_mode_switch_use): ...new function.
13879         * config/epiphany/resolve-sw-modes.c (pass_resolve_sw_modes): Convert
13880         from a global struct to a subclass of rtl_opt_pass along with...
13881         (pass_data_resolve_sw_modes): ...new pass_data instance and...
13882         (make_pass_resolve_sw_modes): ...new function.
13883         * config/i386/i386.c (pass_insert_vzeroupper): Convert from a global
13884         struct to a subclass of rtl_opt_pass along with...
13885         (pass_data_insert_vzeroupper): ...new pass_data instance and...
13886         (make_pass_insert_vzeroupper): ...new function.
13887         * config/sparc/sparc.c (pass_work_around_errata): Convert from a
13888         global struct to a subclass of rtl_opt_pass along with...
13889         (pass_data_work_around_errata): ...new pass_data instance and...
13890         (make_pass_work_around_errata): ...new function.
13891         * config/mips/mips.c (pass_mips_machine_reorg2): Convert from a global
13892         struct to a subclass of rtl_opt_pass along with...
13893         (pass_data_mips_machine_reorg2): ...new pass_data instance and...
13894         (make_pass_mips_machine_reorg2): ...new function.
13896 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13898         * passes.c (pass_manager::operator new): New.
13900 2013-08-05  David Malcolm  <dmalcolm@redhat.com>
13902         Handwritten part of conversion of passes to C++ classes.
13904         * Makefile.in (PASS_MANAGER_H): Add dep on pass-instances.def.
13905         (toplev.o): Add dep on PASS_MANAGER_H.
13906         * cgraphunit.c (cgraph_process_new_functions): Rework invocation
13907         of early local pases to reflect this moving from a global to a
13908         member of gcc::pass_manager.
13909         (cgraph_add_new_function): Likewise.
13910         * lto-cgraph.c (lto_output_node): Update for conversion of
13911         struct ipa_opt_pass_d to a C++ subclass of opt_pass.
13912         * passes.c (opt_pass::clone): New.
13913         (opt_pass::gate): New.
13914         (opt_pass::execute): New.
13915         (opt_pass::opt_pass): New.
13916         (pass_manager::execute_early_local_passes): New.
13917         (pass_manager::execute_pass_mode_switching): new.
13918         (finish_optimization_passes): Convert to...
13919         (pass_manager::finish_optimization_passes): ...this.
13920         (finish_optimization_passes): Update for conversion of passes to
13921         C++ classes.
13922         (register_dump_files_1): Use has_gate since we cannot portably
13923         check a vtable entry against NULL.
13924         (dump_one_pass): Likewise.
13925         (ipa_write_summaries_2): Likewise.
13926         (ipa_write_optimization_summaries_1): Likewise.
13927         (ipa_read_summaries_1): Likewise.
13928         (ipa_read_optimization_summaries_1): Likewise.
13929         (execute_ipa_stmt_fixups): Likewise.
13930         (pass_manager::pass_manager): Rewrite pass-creation, invoking
13931         pass-creation functions rather than wiring up globals, and
13932         storing the results in fields of pass_manager generated using
13933         pass-instances.def.
13934         (pass_manager::dump_profile_report): Update for conversion of
13935         passes to C++ classes.
13936         (pass_manager::execute_ipa_summary_passes): Likewise.
13937         (execute_one_ipa_transform_pass): Likewise.
13938         (execute_one_pass): Use has_gate and has_execute since we cannot
13939         portably check a vtable entry against NULL.
13940         * pass_manager.h (pass_manager::finish_optimization_passes): New.
13941         (pass_manager): Use pass-instances.def to add fields for the
13942         various pass instances.
13943         * toplev.c (finalize): Update for move of
13944         finish_optimization_passes to a method of gcc::pass_manager.
13945         * toplev.h (finish_optimization_passes): Move to method of class
13946         pass_manager.
13947         * tree-pass.h (struct pass_data): New.
13948         (opt_pass): Convert to C++ class, make it a subclass of pass_data.
13949         (opt_pass::gate): Convert to virtual function.
13950         (opt_pass::~opt_pass): New.
13951         (opt_pass::clone): New.
13952         (opt_pass::execute): Convert to virtual function.
13953         (opt_pass::opt_pass): New.
13954         (opt_pass::ctxt_): new.
13955         (gimple_opt_pass): Convert to subclass of opt_pass.
13956         (gimple_opt_pass::gimple_opt_pass): New.
13957         (rtl_opt_pass): Convert to subclass of opt_pass.
13958         (rtl_opt_pass::rtl_opt_pass): New.
13959         (ipa_opt_pass_d): Convert to subclass of opt_pass.
13960         (ipa_opt_pass_d::ipa_opt_pass_d): New.
13961         (simple_ipa_opt_pass): Convert to subclass of opt_pass.
13962         (simple_ipa_opt_pass::simple_ipa_opt_pass): New.
13963         * config/i386/i386.c (rest_of_handle_insert_vzeroupper): Rework
13964         invocation of pass_mode_switching to reflect this moving from a
13965         global to a member of gcc::pass_manager.
13966         (ix86_option_override): Rework how pass_insert_vzeroupper is
13967         added to the pass_manager to reflect autogenerated changes.
13968         * config/i386/t-i386 (i386.o) Add deps on CONTEXT_H and PASS_MANAGER_H.
13970 2013-08-05  Richard Earnshaw  <rearnsha@arm.com>
13972         PR rtl-optimization/57708
13973         * recog.c (peep2_find_free_register): Validate all regs in a
13974         multi-reg mode.
13976 2013-08-05  Jan Hubicka  <jh@suse.cz>
13978         PR lto/57602
13979         * cgraph.c (verify_cgraph_node): Accept local flags from other
13980         partitions.
13981         * ipa.c (symtab_remove_unreachable_nodes): Do not clear local flag.
13982         (function_and_variable_visibility): Likewise.
13983         * trans-mem.c (ipa_tm_create_version): TM versions are not local.
13985 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13987         * graph.c (init_graph_slim_pretty_print): Remove.
13988         (print_graph_cfg): Do not call it.  Use local pretty printer.
13989         (start_graph_dump): Likewise.
13991 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
13993         * gimple-pretty-print.c (buffer): Remove.
13994         (initialized): Likewise.
13995         (maybe_init_pretty_print): Likewise.
13996         (print_gimple_stmt): Do not call it.  Use non-static local
13997         pretty_printer variable.
13998         (print_gimple_expr): Likewise.
13999         (print_gimple_seq): Likewise.
14000         (gimple_dump_bb): Likewise.
14002 2013-08-05  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14004         * asan.c (asan_pp): Remove.
14005         (asan_pp_initialized): Likewise.
14006         (asan_pp_initialize): Likewise.
14007         (asan_pp_string): Take a pretty_printer parameter.  Adjust callers.
14008         (asan_emit_stack_protection): Tidy.  Use local pretty printer.
14009         (asan_add_global): Likewise.
14011 2013-08-04  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14013         * pretty-print.h (pp_base): Remove.  Adjust dependent macros.
14014         * diagnostic.h (diagnostic_flush_buffer): Adjust.
14015         * pretty-print.c (pp_formatted_text_data): Likewise.
14016         (pp_indent): Rename from pp_base_indent.
14017         (pp_format): Rename from pp_base_format.
14018         (pp_output_formatted_text): Rename from pp_base_output_formatted_text.
14019         (pp_format_verbatim): Rename from pp_base_format_verbatim.
14020         (pp_flush): Rename from pp_base_flush.
14021         (pp_set_line_maximum_length): Rename from
14022         pp_base_set_line_maximum_length.
14023         (pp_clear_output_area): Rename from pp_base_clear_output_area.
14024         (pp_set_prefix): Rename from pp_base_set_prefix.
14025         (pp_destroy_prefix): Rename from pp_base_destroy_prefix.
14026         (pp_emit_prefix): Rename from pp_base_emit_prefix.
14027         (pp_append_text): Rename from pp_base_append_text.
14028         (pp_formatted_text): Rename from pp_base_formatted_text.
14029         (pp_last_position_in_text): Rename from pp_base_last_position_in_text.
14030         (pp_remaining_character_count_for_line): Rename from
14031         pp_base_remaining_character_count_for_line.
14032         (pp_newline): Rename from pp_base_newline.
14033         (pp_character): Rename from pp_base_character.
14034         (pp_string): Rename from pp_base_string.
14035         (pp_maybe_space): Rename from pp_base_maybe_space.
14036         * asan.c (asan_pp_string): Adjust.
14037         (asan_emit_stack_protection): Likewise.
14038         (asan_add_global): Likewise.
14039         * sched-vis.c (str_pattern_slim): Adjust pretty printer function call.
14040         * tree-mudflap.c (mf_varname_tree): Likewise.
14041         * tree-pretty-print.c (pp_tree_identifier): Rename from
14042         pp_base_tree_identifier.
14043         * tree-pretty-print.h (pp_tree_identifier): Remove macro definition.
14044         Declare as function.
14046 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14048         * pretty-print.h (pp_bar_bar): New.
14049         (pp_ampersand_ampersand): Likewise.
14050         (pp_less_equal): Likewise.
14051         (pp_greater_equal): Likewise.
14052         * gimple-pretty-print.c (dump_ternary_rhs): Use specialized pretty
14053         printer functions instead of pp_string or operators and punctuators.
14054         (dump_gimple_call): Likewise.
14055         (dump_gimple_omp_for): Likewise.
14056         (dump_gimple_transaction): Likewise.
14057         (dump_gimple_phi): Likewise.
14058         (pp_gimple_stmt_1): Likewise.
14059         * sched-vis.c (print_insn): Likewise.
14060         * tree-mudflap.c (mf_varname_tree): Likewise.
14061         * tree-pretty-print.c (dump_block_node): Likewise.
14062         (dump_generic_node): Likewise.
14064 2013-08-02  Jan Hubicka  <jh@suse.cz>
14066         * lto-cgraph.c (compute_ltrans_boundary): Add abstract origins into
14067         boundaries.
14068         * lto-streamer-out.c (tree_is_indexable): Results decls and
14069         parm decls are not indexable.
14070         (DFS_write_tree_body): Do not follow args and results.
14071         (hash_tree): Likewise.
14072         (output_functions): Rearrange so struct function is needed
14073         only when real body is output; be able to also ouptut abstract
14074         functions; output DECL_ARGUMENTS and DECL_RESULT.
14075         (lto_output): When not in WPA, ale store abstract functions.
14076         (write_symbol): Do not care about RESULT_DECL.
14077         (output_symbol_p): Handle correctly sbtract decls.
14078         * lto-streamer-in.c (input_function): Rearrange so struct
14079         function can be NULL at entry; allow streaming of
14080         functions w/o body; store DECL_ARGUMENTS and DECL_RESULT.
14081         * ipa.c (symtab_remove_unreachable_nodes): Silence confused
14082         sanity check during LTO.
14083         * tree-streamer-out.c (write_ts_decl_non_common_tree_pointers): Skip
14084         RESULT_DECl and DECL_ARGUMENTS.
14085         * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14086         Likewise.
14088 2013-08-03  Gabriel Dos Reis  <gdr@integrable-solutions.net>
14090         * pretty-print.h (pp_underscore): New.
14091         (pp_comma): Tidy.
14092         * gimple-pretty-print.c (dump_unary_rhs): Use specialized pretty
14093         printer functions instead of pp_character.
14094         (dump_binary_rhs): Likewise.
14095         (dump_ternary_rhs): Likewise.
14096         (dump_gimple_call_args): Likewise.
14097         (pp_points_to_solution): Likewise.
14098         (dump_gimple_call): Likewise.
14099         (dump_gimple_switch): Likewise.
14100         (dump_gimple_cond): Likewise.
14101         (dump_gimple_bind): Likewise.
14102         (dump_gimple_try): Likewise.
14103         (dump_gimple_omp_for): Likewise.
14104         (dump_gimple_omp_continue): Likewise.
14105         (dump_gimple_omp_single): Likewise.
14106         (dump_gimple_omp_sections): Likewise.
14107         (dump_gimple_omp_block): Likewise.
14108         (dump_gimple_omp_critical): Likewise.
14109         (dump_gimple_transaction): Likewise.
14110         (dump_gimple_asm): Likewise.
14111         (dump_gimple_phi): Likewise.
14112         (dump_gimple_omp_parallel): Likewise.
14113         (dump_gimple_omp_task): Likewise.
14114         (dump_gimple_omp_atomic_load): Likewise.
14115         (dump_gimple_omp_atomic_store): Likewise.
14116         (dump_gimple_mem_ops): Likewise.
14117         (pp_gimple_stmt_1): Likewise.
14118         (pp_cfg_jump): Likewise.
14119         (dump_implicit_edges): Likewise.
14120         (gimple_dump_bb_for_graph): Likewise.
14121         * graph.c (draw_cfg_node): Likewise.
14122         * langhooks.c (lhd_print_error_function): Likewise.
14123         * sched-vis.c (print_exp): Likewise.
14124         (print_value): Likewise.
14125         (print_pattern): Likewise.
14126         (print_insn): Likewise.
14127         (rtl_dump_bb_for_graph): Likewise.
14128         * tree-pretty-print.c (dump_function_declaration): Likewise.
14129         (dump_array_domain): Likewise.
14130         (dump_omp_clause): Likewise.
14131         (dump_location): Likewise.
14132         (dump_generic_node): Likewise.
14133         (print_struct_decl): Likewise.
14134         * diagnostic.c (diagnostic_show_locus): Use pp_space.
14136 2013-08-03  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14138         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Update
14139         candidate table when replacing a candidate statement.
14140         (replace_rhs_if_not_dup): Likewise.
14141         (replace_one_candidate): Likewise.
14143 2013-08-02  Jan Hubicka  <jh@suse.cz>
14144             Martin Liska  <marxin.liska@gmail.com>
14146         * cgraphunit.c (add_new_function): Fix logic when adding from
14147         late IPA pass.
14148         (assemble_thunk): Rename to ...
14149         (expand_thunk); .. this one; export; get it working with
14150         general functions; make produced gimple valid.
14151         * cgraph.h (expand_thunk): Declare.
14153 2013-08-02  Jan Hubicka  <jh@suse.cz>
14155         * ipa-cp.c (gather_context_independent_values): Use
14156         ipa_get_param_move_cost.
14157         (get_replacement_map): Remove PARAM; move parameter folding
14158         into tree-inline.c
14159         (create_specialized_node): Update.
14160         * ipa-prop.c (ipa_populate_param_decls): Do not look for origins;
14161         assert that we have gimple body; update move_cost.
14162         (count_formal_params): Assert that we have gimple body.
14163         (ipa_dump_param): New function.
14164         (ipa_alloc_node_params): Break out from ...
14165         (ipa_initialize_node_params): ... here.
14166         (ipa_get_vector_of_formal_parms): ICE when used in WPA.
14167         (ipa_write_node_info): Stream move costs.
14168         (ipa_read_node_info): Read move costs.
14169         (ipa_update_after_lto_read): Do not recompute node params.
14170         * ipa-prop.h (ipa_param_descriptor): Add move_cost.
14171         (ipa_get_param): Check we are not in WPA.
14172         (ipa_get_param_move_cost): New.
14173         * tree-inline.c (tree_function_versioning): Fold replacement as needed.
14174         * ipa-inline-analysis.c (inline_node_duplication_hook): Expect only
14175         parm numbers to be present.
14177 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14179         PR rtl-optimization/58048
14180         * lra-constraints.c (process_alt_operands): Don't check asm
14181         operand on register.
14183 2013-08-02  Eric Botcazou  <ebotcazou@adacore.com>
14185         * config/sparc/sparc.c (sparc_emit_membar_for_model) <SMM_TSO>: Add
14186         the implied StoreLoad barrier for atomic operations if before.
14188 2013-08-02  Jan Hubicka  <jh@suse.cz>
14189             Martin Liska  <marxin.liska@gmail.com>
14191         * cgraph.c (cgraph_function_body_availability): Do not check
14192         cgraph flags.
14193         * cgraph.h (symtab_for_node_and_aliases, symtab_nonoverwritable_alias,
14194         symtab_node_availability): Declare.
14195         * ipa.c (can_replace_by_local_alias): New.
14196         (function_and_variable_visibility): Use it.
14197         * symtab.c (symtab_for_node_and_aliases,
14198         symtab_nonoverwritable_alias_1, symtab_nonoverwritable_alias): New.
14200 2013-08-02  Vladimir Makarov  <vmakarov@redhat.com>
14202         PR rtl-optimization/57963
14203         * lra-constraints.c (reverse_equiv_p, contains_reloaded_insn_p): New.
14204         (lra_constraints): Use them.
14206 2013-08-02  Sofiane Naci  <sofiane.naci@arm.com>
14208         * config/arm/types.md (define_attr "type"): Add "load_acq"
14209         and "store_rel".
14210         * config/arm/cortex-a53.md (cortex_a53_load1): Update for attribute
14211         changes.
14212         (cortex_a53_store1): Likewise.
14214 2013-08-01  Jan Hubicka  <jh@suse.cz>
14216         * ipa.c (symtab_remove_unreachable_nodes): Nodes in other
14217         partitions are not needed.
14219 2013-08-01  Uros Bizjak  <ubizjak@gmail.com>
14221         * config/i386/i386.h (MAYBE_NON_Q_CLASS_P): New.
14222         * config/i386/i386.c (ix86_secondary_reload): Use INTEGER_CLASS_P and
14223         MAYBE_NON_Q_CLASS_P where appropriate.
14225 2013-08-01  Jan Hubicka  <jh@suse.cz>
14227         * cgraph.h (release_function_body): Declare.
14228         * tree.c (free_lang_data_in_decl): Free, parameters and return values
14229         of unused delcarations.
14231 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14233         * config/arm/arm.md (minmax_arithsi_non_canon): Emit canonical
14234         RTL form when subtracting a constant.
14236 2013-08-01  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14238         * config/arm/arm.md (peepholes for eq (reg1) (reg2/imm)):
14239         Generate canonical plus rtx with negated immediate instead of minus
14240         where appropriate.
14241         * config/arm/arm.c (thumb2_reorg): Handle ADCS <Rd>, <Rn> case.
14243 2013-08-01  Jan Hubicka  <jh@suse.cz>
14245         * cgraph.c (cgraph_release_function_body): Use used_as_abstract_origin.
14246         (cgraph_release_function_body): Likewise.
14247         (cgraph_can_remove_if_no_direct_calls_p): Likewise.
14248         * cgraph.h (cgrpah_node): Rename abstract_and_needed
14249         to used_as_abstract_origin.
14250         * tree-inline-transfrom.c (can_remove_node_now_p_1): Do not remove
14251         symbols used as abstract origins.
14252         * cgraphunit.c (analyze_functions): Update.
14253         * ipa.c (symtab_remove_unreachable_nodes): Recompute
14254         used_as_abstract_origin.
14255         * tree-inline.c (tree_function_versioning): Update
14256         used_as_abstract_origin; be ready for DECL_RESULT and
14257         DECL_ARGUMENTS to be NULL.
14259         * lto-symtab.c (lto_symtab_merge_symbols): Merge duplicated nodes
14260         for abstract functions.
14261         * cgraph.h (symtab_real_symbol_p): Abstract declarations are not
14262         real symbols.
14264 2013-08-01  Jan Hubicka  <jh@suse.cz>
14266         * profile.c (compute_value_histograms): Fix thinko.
14268 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14270         * config.gcc (aarch64*-*-*): Add aarch-common.o to extra_objs.  Add
14271         aarch-common-protos.h to extra_headers.
14272         (aarch64*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14273         * config/aarch64/aarch64.md: Include "../arm/cortex-a53.md".
14274         * config/aarch64/t-aarch64 (aarch-common.o): Define.
14276 2013-08-01  Sofiane Naci  <sofiane.naci@arm.com>
14278         * config/aarch64/aarch64.md (define_attr "type"): Delete.
14279         Include "../arm/types.md".  Define "type" attribute for all patterns.
14280         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update for
14281         attribute changes.
14283 2013-07-31  Michael Meissner  <meissner@linux.vnet.ibm.com>
14285         * config/rs6000/predicates.md (fusion_gpr_addis): New predicates
14286         to support power8 load fusion.
14287         (fusion_gpr_mem_load): Likewise.
14289         * config/rs6000/rs6000-modes.def (PTImode): Update a comment.
14291         * config/rs6000/rs6000-protos.h (fusion_gpr_load_p): New
14292         declarations for power8 load fusion.
14293         (emit_fusion_gpr_load): Likewise.
14295         * config/rs6000/rs6000.c (rs6000_option_override_internal): If
14296         tuning for power8, turn on fusion mode by default.  Turn on sign
14297         extending fusion mode if normal fusion mode is on, and we are at
14298         -O2 or -O3.
14299         (fusion_gpr_load_p): New function, return true if we can fuse an
14300         addis instruction with a dependent load to a GPR.
14301         (emit_fusion_gpr_load): Emit the instructions for power8 load
14302         fusion to GPRs.
14304         * config/rs6000/vsx.md (VSX_M2): New iterator for fusion peepholes.
14305         (VSX load fusion peepholes): New peepholes to fuse together an
14306         addi instruction with a VSX load instruction.
14308         * config/rs6000/rs6000.md (GPR load fusion peepholes): New
14309         peepholes to fuse an addis instruction with a load to a GPR base
14310         register.  If we are supporting sign extending fusions, convert
14311         sign extending loads to zero extending loads and add an explicit
14312         sign extension.
14314 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14316         * config.gcc (arm*-*-*): Add aarch-common.o to extra_objs.  Add
14317         aarch-common-protos.h to extra_headers.
14318         (arm*-*-*): Add arm/aarch-common-protos.h to tm_p_file.
14319         * config/arm/arm.c (arm_early_load_addr_dep): Move from here to ...
14320         (arm_early_store_addr_dep): Likewise.
14321         (arm_no_early_alu_shift_dep): Likewise.
14322         (arm_no_early_alu_shift_value_dep): Likewise.
14323         (arm_no_early_mul_dep): Likewise.
14324         (arm_no_early_store_addr_dep): Likewise.
14325         (arm_mac_accumulator_is_mul_result): Likewise.
14326         (arm_mac_accumulator_is_result): Likewise.
14327         * config/arm/aarch-common.c: ... here.  New file.
14328         * config/arm/arm-protos.h (arm_early_load_addr_dep): Move from
14329         here to ...
14330         (arm_early_store_addr_dep): Likewise.
14331         (arm_no_early_alu_shift_dep): Likewise.
14332         (arm_no_early_alu_shift_value_dep): Likewise.
14333         (arm_no_early_mul_dep): Likewise.
14334         (arm_no_early_store_addr_dep): Likewise.
14335         (arm_mac_accumulator_is_mul_result): Likewise.
14336         (arm_mac_accumulator_is_result): Likewise.
14337         * config/arm/aarch-common-protos.h: ... here.  New file.
14338         * config/arm/t-arm (aarch-common.o): Define.
14340 2013-07-31  Sofiane Naci  <sofiane.naci@arm.com>
14342         * config/arm/arm.md: Include new file "types.md".
14343         (define_attr "type"): Move from here to ...
14344         (define_attr "mul32"): Likewise.
14345         (define_attr "mul64"): Likewise.
14346         * config/arm/types.md: ... here.  New file.
14348 2013-07-31  Sebastian Huber  <sebastian.huber@embedded-brains.de>
14350         * config.gcc (*-*-rtems*): Use __cxa_atexit by default.
14351         * config/rs6000/rtems.h (TARGET_LIBGCC_SDATA_SECTION): Define.
14353 2013-07-31  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
14355         * gen-pass-instances.awk: Fix offset of substr().
14357 2013-07-31  David Malcolm  <dmalcolm@redhat.com>
14359         * Makefile.in (pass-instances.def): New.
14360         (passes.o): Replace dependency on passes.def with one on
14361         pass-instances.def
14363         * gen-pass-instances.awk: New.
14365         * passes.c (pass_manager::pass_manager): Use pass-instances.def
14366         rather than passes.def, updating local definition of NEXT_PASS
14367         macro to add an extra NUM parameter (currently unused).
14369 2013-07-30  David Malcolm  <dmalcolm@redhat.com>
14371         * Makefile.in (PASS_MANAGER_H): New.
14372         (lto-cgraph.o): Depend on CONTEXT_H and PASS_MANAGER_H.
14373         (passes.o): Likewise.
14374         (statistics.o): Likewise.
14375         (cgraphunit.o): Likewise.
14376         (context.o): Depend on PASS_MANAGER_H.
14378         * pass_manager.h: New.
14380         * cgraphunit.c (cgraph_add_new_function): Update for moves
14381         of globals to fields of pass_manager.
14382         (analyze_function): Likewise.
14383         (expand_function): Likewise.
14384         (ipa_passes): Likewise.
14385         (compile): Likewise.
14387         * context.c (context::context): New.
14388         * context.h  (context::context): New.
14389         (context::get_passes): New.
14390         (context::passes_): New.
14392         * lto-cgraph.c (input_node): Update for moves of globals to
14393         fields of pass_manager.
14395         * passes.c (all_passes): Remove, in favor of a field of the
14396         same name within the new class pass_manager.
14397         (all_small_ipa_passes): Likewise.
14398         (all_lowering_passes): Likewise.
14399         (all_regular_ipa_passes): Likewise.
14400         (all_late_ipa_passes): Likewise.
14401         (all_lto_gen_passes): Likewise.
14402         (passes_by_id): Likewise.
14403         (passes_by_id_size): Likewise.
14404         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14405         the new class pass_manager.
14406         (set_pass_for_id): Convert to...
14407         (pass_manager::set_pass_for_id): ...method.
14408         (get_pass_for_id): Convert to...
14409         (pass_manager::get_pass_for_id): ...method.
14410         (register_one_dump_file): Move body of implementation into...
14411         (pass_manager::register_one_dump_file): ...here.
14412         (register_dump_files_1): Convert to...
14413         (pass_manager::register_dump_files_1): ...method.
14414         (register_dump_files): Convert to...
14415         (pass_manager::register_dump_files): ...method.
14416         (create_pass_tab): Update for moves of globals to fields of
14417         pass_manager.
14418         (dump_passes): Move body of implementation into...
14419         (pass_manager::dump_passes): ...here.
14420         (register_pass): Move body of implementation into...
14421         (pass_manager::register_pass): ...here.
14422         (init_optimization_passes): Convert into...
14423         (pass_manager::pass_manager): ...constructor for new
14424         pass_manager class, and initialize the pass_lists array.
14425         (check_profile_consistency): Update for moves of globals to
14426         fields of pass_manager.
14427         (dump_profile_report): Move body of implementation into...
14428         (pass_manager::dump_profile_report): ...here.
14429         (ipa_write_summaries_1): Update for moves of pass lists from
14430         being globals to fields of pass_manager.
14431         (ipa_write_optimization_summaries): Likewise.
14432         (ipa_read_summaries):  Likewise.
14433         (ipa_read_optimization_summaries): Likewise.
14434         (execute_all_ipa_stmt_fixups): Likewise.
14436         * statistics.c (statistics_fini): Update for moves of globals to
14437         fields of pass_manager.
14439         * toplev.c (general_init): Replace call to
14440         init_optimization_passes with construction of the pass_manager
14441         instance.
14443         * tree-pass.h (all_passes): Remove, in favor of a field of the
14444         same name within the new class pass_manager.
14445         (all_small_ipa_passes): Likewise.
14446         (all_lowering_passes): Likewise.
14447         (all_regular_ipa_passes): Likewise.
14448         (all_lto_gen_passes): Likewise.
14449         (all_late_ipa_passes): Likewise.
14450         (passes_by_id): Likewise.
14451         (passes_by_id_size): Likewise.
14452         (gcc_pass_lists): Remove, in favor of "pass_lists" field within
14453         the new class pass_manager.
14454         (get_pass_for_id): Remove.
14456 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14458         * config.gcc (arm): Require 64-bit host-wide-int for all ARM target
14459         configs.
14461 2013-07-30  Richard Earnshaw  <rearnsha@arm.com>
14463         * arm.md (mulhi3): New expand pattern.
14465 2013-07-30  Jan Hubicka  <jh@suse.cz>
14466             Martin Liska  <marxin.liska@gmail.com>
14468         * profile.c (compute_value_histograms): Do not ICE when
14469         there is mismatch only on some counters.
14471 2013-07-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
14473         PR rtl-optimization/57637
14474         * function.c (move_insn_for_shrink_wrap): Also check the
14475         GEN set of the LIVE problem for the liveness analysis
14476         if it exists, otherwise give up.
14478 2013-07-29  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14480         PR tree-optimization/57993
14481         * gimple-ssa-strength-reduction.c (replace_mult_candidate): Record
14482         replaced statement in the candidate table.
14483         (phi_add_costs): Return infinite cost when the hidden basis does
14484         not dominate all phis on which the candidate is dependent.
14485         (replace_one_candidate): Record replaced statement in the
14486         candidate table.
14488 2013-07-29  Joern Rennecke  <joern.rennecke@embecosm.com>
14490         * config/epiphany/epiphany.md (*isub_i+2): New peephole.
14491         (ashlv2si3): New expander.
14492         (*ashlv2si3_i): New define_insn_and_split.
14493         * predicates.md (float_operation): Allow patterns with three
14494         basic sub-patterns.
14496         PR rtl-optimization/58021
14497         * mode-switching.c (create_pre_exit): Always split off preceding
14498         insns if we are not at the basic block head.
14500 2013-07-29  Maciej W. Rozycki  <macro@codesourcery.com>
14502         * config/mips/linux.h (GLIBC_DYNAMIC_LINKER): Handle `-mnan=2008'.
14503         (UCLIBC_DYNAMIC_LINKER): New macro.
14504         * config/mips/linux64.h (GLIBC_DYNAMIC_LINKER32): Handle
14505         `-mnan=2008'.
14506         (GLIBC_DYNAMIC_LINKER64, GLIBC_DYNAMIC_LINKERN32): Likewise.
14507         (UCLIBC_DYNAMIC_LINKER32): Undefine macro first.  Handle
14508         `-mnan=2008'.
14509         (UCLIBC_DYNAMIC_LINKER64): Redefine macro.
14510         (UCLIBC_DYNAMIC_LINKERN32): Likewise.
14511         * config/mips/mips-modes.def: Remove RESET_FLOAT_FORMAT calls
14512         for SF and DF modes.  Use ieee_quad_format for TF mode.
14513         * config/mips/mips-opts.h (mips_ieee_754_setting): New enum.
14514         * config/mips/mips.c (mips_file_start): Output a `.nan' directive.
14515         (mips_option_override): Handle `-mnan=legacy'.
14516         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Handle
14517         `-mabs=2008' and `-mnan=2008'.
14518         (OPTION_DEFAULT_SPECS): Add "nan" default.
14519         (ASM_SPEC): Handle `-mnan='.
14520         [!HAVE_AS_NAN] (HAVE_AS_NAN): New macro.
14521         * config/mips/mips.md (abs<mode>2): Handle `-mabs=2008', update
14522         comment accordingly.
14523         (neg<mode>2): Likewise.
14524         * config/mips/mips.opt (mabs, mnan): New options.
14525         * doc/install.texi (Configuration): Document `--with-nan=' option.
14526         * doc/invoke.texi (Option Summary): List MIPS `-mabs=' and
14527         `-mnan=' options.
14528         (MIPS Options): Document them.
14529         * config.gcc <mips*-*-*>: Handle `--with-nan='.
14530         * configure.ac <mips*-*-*>: Check for GAS `-mnan=2008' support.
14531         * configure: Regenerate.
14532         * config.in: Regenerate.
14534 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14536         * config/i386/i386.md (float post-reload splitters): Do not check
14537         for subregs of SSE registers.
14539 2013-07-29  Uros Bizjak  <ubizjak@gmail.com>
14540             H.J. Lu  <hongjiu.lu@intel.com>
14542         PR target/57954
14543         PR target/57988
14544         * config/i386/i386.md (post-reload splitter
14545         to avoid partial SSE reg dependency stalls): New pattern.
14547 2013-07-29  Dominik Vogt  <vogt@linux.vnet.ibm.com>
14549         * config/s390/s390.md ("movcc"): Swap load and store instructions.
14551 2013-07-27  Joern Rennecke  <joern.rennecke@embecosm.com>
14553         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
14554         Also reserve space for saving UNKNOWN_REGNUM for leaf functions.
14556 2013-07-26  Cary Coutant  <ccoutant@google.com>
14558         * dwarf2out.c (die_checksum_ordered): Don't include template
14559         instantiations in signature.
14560         (is_template_parameter): New function.
14561         (is_template_instantiation): New function.
14562         (generate_skeleton_bottom_up): Don't include template instantiations
14563         in type unit DIE.
14564         (generate_skeleton): Likewise.
14565         (break_out_comdat_types): Move recursive call to break out nested
14566         types earlier.
14567         (prune_unused_types_mark_generic_parms_dies): Call
14568         is_template_parameter.
14570 2013-07-26  Ian Bolton  <ian.bolton@arm.com>
14572         * config/aarch64/aarch64.md (neg<mode>2): Offer alternative that
14573         uses vector registers.
14574         * config/aarch64/iterators.md: Add attributes rtn and vas.
14576 2013-07-26  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14577             Richard Earnshaw  <richard.earnshaw@arm.com>
14579         * combine.c (simplify_comparison): Re-canonicalize operands
14580         where appropriate.
14581         * config/arm/arm.md (movcond_addsi): New splitter.
14583 2013-07-25  Sterling Augustine  <saugustine@google.com>
14585         * dwarf2out.c (size_of_pubnames): Move code to...
14586         (include_pubname_in_output): ...here.  New.
14587         (want_pubnames): Rearrange.
14588         (output_pubnames): Call include_pubname_in_output.  Move assertion.
14590 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14592         * doc/extend.texi: Fix return types for __builtin_ia32_cmp*s builtins.
14594 2013-07-25  Cameron McInally  <cameron.mcinally@nyu.edu>
14596         PR target/38836
14597         * doc/extend.texi: Remove obsolete builtins. Fix
14598         typo for __builtin_ia32_loadss and __builtin_ia32_cmpnltss.
14600 2013-07-25  Jan Hubicka  <jh@suse.cz>
14602         * cgraph.c (release_function_body): Break out from ...
14603         (cgraph_release_function_body): ... this one; also release DECL_RESULT
14604         and DECL_ARGUMENTS.
14605         * ipa-cp.c (get_replacement_map): Add parm_num argument; do not set
14606         old_tree in the map.
14607         (create_specialized_node): Update.
14608         * lto-cgraph.c (output_node_opt_summary): Do not translate old_tree
14609         into index.
14610         * cgraphclones.c (cgraph_create_virtual_clone): Do not copy
14611         DECL_ARGUMENTS, DECL_INITIAL and DECL_RESULT.
14612         * ipa-prop.c (ipa_populate_param_decls): Look for origin of clones.
14613         * tree-inline.c (initialize_cfun): Initialize DECL_ARGUMENTS and
14614         DECL_RESULT.
14616 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14618         * config/arm/arm.md (arm_addsi3, addsi3_carryin_<optab>,
14619         addsi3_carryin_alt2_<optab>): Correct output template.
14621 2013-07-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
14623         * config/arm/arm-fixed.md (ssmulsa3, usmulusa3):
14624         Adjust for arm_restrict_it.
14625         Remove trailing whitespace.
14627 2013-07-25 Â Mark Kettenis Â <kettenis@openbsd.org>
14629         * config/pa/pa.c (pa_trampoline_init): Emit __enable_execute_stack
14630         libcall if HAVE_ENABLE_EXECUTE_STACK is defined.
14632         * config.gcc (hppa-*-openbsd*): Don't set tmake_file.
14634 2013-07-25  Vladimir Makarov  <vmakarov@redhat.com>
14636         PR rtl-optimization/57960
14637         * lra-constraints.c (process_alt_operands): Use the right mode
14638         when checking strict_low.
14640 2013-07-25  Jan Hubicka  <jh@suse.cz>
14642         * lto-symtab.c (lto_cgraph_replace_node): Release function body.
14643         * cgraph.c (cgraph_remove_node): Do not release function body
14644         when in cgraph streaming.
14645         * ipa.c (process_references, symtab_remove_unreachable_nodes): Objects
14646         in other partitions are not considered reachable; fix handling of
14647         clones.
14649 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14651         * config/arm/arm.md (*sibcall_insn): Remove unnecessary space.
14653 2013-07-25  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
14655         PR target/19599
14656         PR target/57731
14657         PR target/57837
14658         * config/arm/arm.md ("*sibcall_insn): Replace use of
14659         Ss with US. Adjust output for v5 and v4t.
14660         (*sibcall_value_insn): Likewise and loosen predicate on operand0.
14662         * config/arm/constraints.md ("Ss"): Rename to US.
14664 2013-07-25  Terry Guo  <terry.guo@arm.com>
14666         * config/arm/arm.c (thumb1_size_rtx_costs): Assign proper cost for
14667         shift_add/shift_sub0/shift_sub1 RTXs.
14669 2013-07-24  Bill Schmidt  <wschmidt@linux.ibm.com>
14670             Anton Blanchard  <anton@au1.ibm.com>
14672         * config/rs6000/altivec.md (altivec_vpkpx): Handle little endian.
14673         (altivec_vpks<VI_char>ss): Likewise.
14674         (altivec_vpks<VI_char>us): Likewise.
14675         (altivec_vpku<VI_char>us): Likewise.
14676         (altivec_vpku<VI_char>um): Likewise.
14678 2013-07-24  David Malcolm  <dmalcolm@redhat.com>
14680         Introduce context class.
14682         * Makefile.in (CONTEXT_H): New.
14683         (OBJS): Add context.o.
14684         (toplev.o): Add CONTEXT_H to dependencies.
14685         (context.o): New.
14687         * toplev.c (general_init): Create the singleton gcc::context instance.
14689         * context.c: New.
14691         * context.h: New.
14693 2013-07-24  Joern Rennecke  <joern.rennecke@embecosm.com>
14695         PR rtl-optimization/57968
14696         * mode-switching.c (create_pre_exit): Allow instructions that
14697         don't set a return register to need a non-exit mode.
14699 2013-07-24  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14700             Anton Blanchard  <anton@au1.ibm.com>
14702         * config/rs6000/vector.md (vec_realign_load_<mode>): Reorder input
14703         operands to vperm for little endian.
14704         * config/rs6000/rs6000.c (rs6000_expand_builtin): Use lvsr instead
14705         of lvsl to create the control mask for a vperm for little endian.
14707 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14708             Anton Blanchard  <anton@au1.ibm.com>
14710         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Reverse
14711         two operands for little-endian.
14713 2013-07-23  Steve Ellcey  <sellcey@mips.com>
14715         * config/mips/mips.c (mips_case_values_threshold): New.
14716         (TARGET_CASE_VALUES_THRESHOLD): Define.
14718 2013-07-23  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
14719             Anton Blanchard  <anton@au1.ibm.com>
14721         * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Correct
14722         selection of field for vector splat in little endian mode.
14724 2013-07-23  Michael Meissner  <meissner@linux.vnet.ibm.com>
14726         * config/rs6000/vector.md (xor<mode>3): Move 128-bit boolean
14727         expanders to rs6000.md.
14728         (ior<mode>3): Likewise.
14729         (and<mode>3): Likewise.
14730         (one_cmpl<mode>2): Likewise.
14731         (nor<mode>3): Likewise.
14732         (andc<mode>3): Likewise.
14733         (eqv<mode>3): Likewise.
14734         (nand<mode>3): Likewise.
14735         (orc<mode>3): Likewise.
14737         * config/rs6000/rs6000-protos.h (rs6000_split_logical): New
14738         declaration.
14740         * config/rs6000/rs6000.c (rs6000_split_logical_inner): Add support
14741         to split multi-word logical operations.
14742         (rs6000_split_logical_di): Likewise.
14743         (rs6000_split_logical): Likewise.
14745         * config/rs6000/vsx.md (VSX_L2): Delete, no longer used.
14746         (vsx_and<mode>3_32bit): Move 128-bit logical insns to rs6000.md,
14747         and allow TImode operations in 32-bit.
14748         (vsx_and<mode>3_64bit): Likewise.
14749         (vsx_ior<mode>3_32bit): Likewise.
14750         (vsx_ior<mode>3_64bit): Likewise.
14751         (vsx_xor<mode>3_32bit): Likewise.
14752         (vsx_xor<mode>3_64bit): Likewise.
14753         (vsx_one_cmpl<mode>2_32bit): Likewise.
14754         (vsx_one_cmpl<mode>2_64bit): Likewise.
14755         (vsx_nor<mode>3_32bit): Likewise.
14756         (vsx_nor<mode>3_64bit): Likewise.
14757         (vsx_andc<mode>3_32bit): Likewise.
14758         (vsx_andc<mode>3_64bit): Likewise.
14759         (vsx_eqv<mode>3_32bit): Likewise.
14760         (vsx_eqv<mode>3_64bit): Likewise.
14761         (vsx_nand<mode>3_32bit): Likewise.
14762         (vsx_nand<mode>3_64bit): Likewise.
14763         (vsx_orc<mode>3_32bit): Likewise.
14764         (vsx_orc<mode>3_64bit): Likewise.
14766         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Always allow vector
14767         logical types in GPRs.
14769         * config/rs6000/altivec.md (altivec_and<mode>3): Move 128-bit
14770         logical insns to rs6000.md, and allow TImode operations in
14771         32-bit.
14772         (altivec_ior<mode>3): Likewise.
14773         (altivec_xor<mode>3): Likewise.
14774         (altivec_one_cmpl<mode>2): Likewise.
14775         (altivec_nor<mode>3): Likewise.
14776         (altivec_andc<mode>3): Likewise.
14778         * config/rs6000/rs6000.md (BOOL_128): New mode iterators and mode
14779         attributes for moving the 128-bit logical operations into
14780         rs6000.md.
14781         (BOOL_REGS_OUTPUT): Likewise.
14782         (BOOL_REGS_OP1): Likewise.
14783         (BOOL_REGS_OP2): Likewise.
14784         (BOOL_REGS_UNARY): Likewise.
14785         (BOOL_REGS_AND_CR0): Likewise.
14786         (one_cmpl<mode>2): Add support for DI logical operations on
14787         32-bit, splitting the operations to 32-bit.
14788         (anddi3): Likewise.
14789         (iordi3): Likewise.
14790         (xordi3): Likewise.
14791         (and<mode>3, 128-bit types): Rewrite 2013-06-06 logical operator
14792         changes to combine the 32/64-bit code, allow logical operations on
14793         TI mode in 32-bit, and to use similar match_operator patterns like
14794         scalar mode uses.  Combine the Altivec and VSX code for logical
14795         operations, and move it here.
14796         (ior<mode>3, 128-bit types): Likewise.
14797         (xor<mode>3, 128-bit types): Likewise.
14798         (one_cmpl<mode>3, 128-bit types): Likewise.
14799         (nor<mode>3, 128-bit types): Likewise.
14800         (andc<mode>3, 128-bit types): Likewise.
14801         (eqv<mode>3, 128-bit types): Likewise.
14802         (nand<mode>3, 128-bit types): Likewise.
14803         (orc<mode>3, 128-bit types): Likewise.
14804         (and<mode>3_internal): Likewise.
14805         (bool<mode>3_internal): Likewise.
14806         (boolc<mode>3_internal1): Likewise.
14807         (boolc<mode>3_internal2): Likewise.
14808         (boolcc<mode>3_internal1): Likewise.
14809         (boolcc<mode>3_internal2): Likewise.
14810         (eqv<mode>3_internal1): Likewise.
14811         (eqv<mode>3_internal2): Likewise.
14812         (one_cmpl1<mode>3_internal): Likewise.
14814 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14816         * config/microblaze/microblaze.c (microblaze_expand_prologue):
14817         Rename flag_stack_usage to flag_stack_usage_info.
14819 2013-07-23  David Holsgrove  <david.holsgrove@xilinx.com>
14821         * config/microblaze/sync.md: New file.
14822         * config/microblaze/microblaze.md: Include sync.md
14823         * config/microblaze/microblaze.c: Add print_operand 'y'.
14824         * config/microblaze/constraints.md: Add memory_contraint
14825         'Q' which is a single register.
14827 2013-07-23  Eric Botcazou  <ebotcazou@adacore.com>
14829         * doc/invoke.texi (SPARC Options): Document new leon3 processor value.
14831 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
14833         * reload.c (find_reloads): Exit loop once we find this operand
14834         cannot be reloaded somehow for this alternative.
14836         * reload.c (find_reloads): Exit loop once we find a hard register.
14838         * rtlanal.c (computed_jump_p): Exit loop once we find label
14839         reference is used.
14841         * i386.c (ix86_pad_returns): Exit loop after setting replace.
14843         * cfgloopmanip.c (remove_path): Exit loop after setting
14844         irred_invalidated.
14846         * gensupport.c (subst_dup): Avoid loop if code is not
14847         MATCH_DUP nor MATCH_OP_DUP.
14849 2013-07-23  Nicklas Bo Jensen  <nbjensen@gmail.com>
14851         * doc/md.texi (Machine-Specific Peephole Optimizers): Fix a typo.
14853 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14855         * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Also return
14856         true for SP_REGNUM if mode == ptr_mode.
14857         * config/aarch64/aarch64.h (ADDITIONAL_REGISTER_NAMES): Add "wsp"
14858         with value R0_REGNUM + 31.
14860 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14862         * config/aarch64/aarch64.c (aarch64_pad_arg_upward): In big-endian,
14863         pad pointer-typed argument downward.
14865 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14867         * config/aarch64/aarch64.h (TARGET_CPU_CPP_BUILTINS): Define _ILP32
14868         and __ILP32__ when the ILP32 model is in use.
14870 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14872         * config/aarch64/aarch64.c (POINTER_BYTES): New define.
14873         (aarch64_load_symref_appropriately): In the case of
14874         SYMBOL_SMALL_ABSOLUTE, use the mode of 'dest' instead of Pmode
14875         to generate new rtx; likewise to the case of SYMBOL_SMALL_GOT.
14876         (aarch64_expand_mov_immediate): In the case of SYMBOL_FORCE_TO_MEM,
14877         change to pass 'ptr_mode' to force_const_mem and zero-extend 'mem'
14878         if 'mode' doesn't equal to 'ptr_mode'.
14879         (aarch64_output_mi_thunk): Add an assertion on the alignment of
14880         'vcall_offset'; change to call aarch64_emit_move differently depending
14881         on whether 'Pmode' equals to 'ptr_mode' or not; use 'POINTER_BYTES'
14882         to calculate the upper bound of 'vcall_offset'.
14883         (aarch64_cannot_force_const_mem): Change to also return true if
14884         mode != ptr_mode.
14885         (aarch64_legitimize_reload_address): In the case of large
14886         displacements, add new local variable 'xmode' and an assertion
14887         based on it; change to use 'xmode' to generate the new rtx and
14888         reload.
14889         (aarch64_asm_trampoline_template): Change to generate the template
14890         differently depending on TARGET_ILP32 or not; change to use
14891         'POINTER_BYTES' in the argument passed to assemble_aligned_integer.
14892         (aarch64_trampoline_size): Removed.
14893         (aarch64_trampoline_init): Add new local constant 'tramp_code_sz'
14894         and replace immediate literals with it.  Change to use 'ptr_mode'
14895         instead of 'DImode' and call convert_memory_address if the mode
14896         of 'fnaddr' doesn't equal to 'ptr_mode'.
14897         (aarch64_elf_asm_constructor): Change to use assemble_aligned_integer
14898         to output symbol.
14899         (aarch64_elf_asm_destructor): Likewise.
14900         * config/aarch64/aarch64.h (TRAMPOLINE_SIZE): Change to be dependent
14901         on TARGET_ILP32 instead of aarch64_trampoline_size.
14902         * config/aarch64/aarch64.md (movsi_aarch64): Add new alternatives
14903         of 'mov' between WSP and W registers as well as 'adr' and 'adrp'.
14904         (loadwb_pair<GPI:mode>_<PTR:mode>): Rename to ...
14905         (loadwb_pair<GPI:mode>_<P:mode>): ... this.  Replace PTR with P.
14906         (storewb_pair<GPI:mode>_<PTR:mode>): Likewise; rename to ...
14907         (storewb_pair<GPI:mode>_<P:mode>): ... this.
14908         (add_losym): Change to 'define_expand' and call gen_add_losym_<mode>
14909         depending on the value of 'mode'.
14910         (add_losym_<mode>): New.
14911         (ldr_got_small_<mode>): New, based on ldr_got_small.
14912         (ldr_got_small): Remove.
14913         (ldr_got_small_sidi): New.
14914         * config/aarch64/iterators.md (P): New.
14915         (PTR): Change to 'ptr_mode' in the condition.
14917 2013-07-23  Yufeng Zhang  <yufeng.zhang@arm.com>
14919         * config.gcc (aarch64*-*-*): Support --with-abi.
14920         (aarch64*-*-elf): Support --with-multilib-list.
14921         (aarch64*-*-linux*): Likewise.
14922         (supported_defaults): Add abi to aarch64*-*-*.
14923         * configure.ac: Mention AArch64 for --with-multilib-list.
14924         * configure: Re-generated.
14925         * config/aarch64/biarchilp32.h: New file.
14926         * config/aarch64/biarchlp64.h: New file.
14927         * config/aarch64/aarch64-elf.h (ENDIAN_SPEC): New define.
14928         (ABI_SPEC): Ditto.
14929         (MULTILIB_DEFAULTS): Ditto.
14930         (DRIVER_SELF_SPECS): Ditto.
14931         (ASM_SPEC): Update to also substitute -mabi.
14932         * config/aarch64/aarch64-elf-raw.h (LINK_SPEC): Add linker script
14933         file whose name depends on -mabi= and -mbig-endian.
14934         * config/aarch64/aarch64.h (LONG_TYPE_SIZE): Change to depend on
14935         TARGET_ILP32.
14936         (POINTER_SIZE): New define.
14937         (POINTERS_EXTEND_UNSIGNED): Ditto.
14938         (enum aarch64_abi_type): New enumeration tag.
14939         (AARCH64_ABI_LP64, AARCH64_ABI_ILP32): New enumerators.
14940         (AARCH64_ABI_DEFAULT): Define to AARCH64_ABI_LP64 if undefined.
14941         (TARGET_ILP32): New define.
14942         * config/aarch64/aarch64.opt (mabi): New.
14943         (aarch64_abi): New.
14944         (ilp32, lp64): New values for -mabi.
14945         * config/aarch64/t-aarch64 (comma): New define.
14946         (MULTILIB_OPTIONS): Ditto.
14947         (MULTILIB_DIRNAMES): Ditto.
14948         * config/aarch64/t-aarch64-linux (MULTIARCH_DIRNAME): New define.
14949         * doc/invoke.texi: Document -mabi for AArch64.
14951 2013-07-23  Georg-Johann Lay  <avr@gjlay.de>
14953         * config/avr/avr.md: Explain asm print modifier 'r' for REG.
14955 2013-07-22  Bill Schmidt  <wschmidt@vnet.linux.ibm.com>
14956             Anton Blanchard  <anton@au1.ibm.com>
14958         * config/rs6000/rs6000.c (rs6000_expand_vector_init): Fix
14959         endianness when selecting field to splat.
14961 2013-07-22  Eric Christopher  <echristo@gmail.com>
14963         * dwarf2out.c (die_odr_checksum): New function to use
14964         CHECKSUM_ macros and ULEB128 for DIE tag.
14965         (generate_type_signature): Use.
14967 2013-07-22  Eric Botcazou  <ebotcazou@adacore.com>
14969         * config.gcc (sparc*-*-*): Accept leon3 processor.
14970         (sparc-leon*-*): Merge with sparc*-*-* and add leon3 support.
14971         * doc/invoke.texi (SPARC Options): Adjust -mfix-ut699 entry.
14972         * config/sparc/sparc-opts.h (enum processor_type): Add PROCESSOR_LEON3.
14973         * config/sparc/sparc.opt (enum processor_type): Add leon3.
14974         (mfix-ut699): Adjust comment.
14975         * config/sparc/sparc.h (TARGET_CPU_leon3): New define.
14976         (CPP_CPU32_DEFAULT_SPEC): Add leon3 support.
14977         (CPP_CPU_SPEC): Likewise.
14978         (ASM_CPU_SPEC): Likewise.
14979         * config/sparc/sparc.c (leon3_cost): New constant.
14980         (sparc_option_override): Add leon3 support.
14981         (mem_ref): New function.
14982         (sparc_gate_work_around_errata): Return true if -mfix-ut699 is enabled.
14983         (sparc_do_work_around_errata): Look into the instruction in the delay
14984         slot and adjust accordingly.  Add fix for the data cache nullify issues
14985         of the UT699.  Change insertion position for the NOP.
14986         * config/sparc/leon.md (leon_fpalu, leon_fpmds, write_buf): Delete.
14987         (leon3_load): New reservation.
14988         (leon_store): Bump latency to 2.
14989         (grfpu): New automaton.
14990         (grfpu_alu): New unit.
14991         (grfpu_ds): Likewise.
14992         (leon_fp_alu): Adjust.
14993         (leon_fp_mult): Delete.
14994         (leon_fp_div): Split into leon_fp_divs and leon_fp_divd.
14995         (leon_fp_sqrt): Split into leon_fp_sqrts and leon_fp_sqrtd.
14996         * config/sparc/sparc.md (cpu): Add leon3.
14997         * config/sparc/sync.md (atomic_exchangesi): Disable if -mfix-ut699.
14998         (swapsi): Likewise.
14999         (atomic_test_and_set): Likewise.
15000         (ldstub): Likewise.
15002 2013-07-22  Jürgen Urban  <JuergenUrban@gmx.de>
15004         * config.gcc (mips*-*-*): Add --with-fpu support.  Make single the
15005         default for R5900 targets.
15006         * config/mips/mips.h (OPTION_DEFAULT_SPECS): Handle --with-fpu.
15007         (ISA_HAS_LDC1_SDC1): Set to false for TARGET_MIPS5900.
15008         * config/mips/mips.c (mips_option_override): Report an error for
15009         -march=r5900 -mhard-float -mdouble-float.  Use spu_single_format
15010         for -march=r5900 -mhard-float.
15012 2013-07-22  Po-Chun Chang  <pchang9@cs.wisc.edu>
15014         * df-problems.c (can_move_insns_across): Exit loop once we
15015         find a non-fixed, non-global register.
15017         * ipa-pure-const.c (propagate_nothrow): Exit loop after
15018         setting can_throw.
15020         * omega.c (omega_eliminate_red): Break after setting red_found.
15021         (omega_problem_has_red_equations): Similarly after setting found.
15022         (omega_query_variable): Similarly after setting coupled.
15024 2013-07-22  Marek Polacek  <polacek@redhat.com>
15026         * gimplify.c: Don't include gimple.h twice.
15028 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15030         * config/arm/constraints.md (Pd): Allow TARGET_THUMB
15031         instead of TARGET_THUMB1.
15032         (Pz): New constraint.
15033         * config/arm/arm.md (arm_addsi3): Add alternatives for 16-bit
15034         encodings.
15035         (compare_negsi_si): Likewise.
15036         (compare_addsi2_op0): Likewise.
15037         (compare_addsi2_op1): Likewise.
15038         (addsi3_carryin_<optab>): Likewise.
15039         (addsi3_carryin_alt2_<optab>): Likewise.
15040         (addsi3_carryin_shift_<optab>): Disable cond_exec variant
15041         for arm_restrict_it.
15042         (subsi3_carryin): Likewise.
15043         (arm_subsi3_insn): Add alternatives for 16-bit encoding.
15044         (minmax_arithsi): Disable for arm_restrict_it.
15045         (minmax_arithsi_non_canon): Adjust for arm_restrict_it.
15046         (satsi_<SAT:code>): Disable cond_exec variant for arm_restrict_it.
15047         (satsi_<SAT:code>_shift): Likewise.
15048         (arm_shiftsi3): Add alternative for 16-bit encoding.
15049         (arm32_movhf): Disable for arm_restrict_it.
15050         (arm_cmpdi_unsigned): Add alternatives for 16-bit encoding.
15051         (arm_movtas_ze): Disable cond_exec variant for arm_restrict_it.
15053 2013-07-22  Sofiane Naci  <sofiane.naci@arm.com>
15055         * config/arm/arm.md (attribute "insn"): Delete.
15056         (attribute "type"): Add "mov_imm", "mov_reg", "mov_shift",
15057         "mov_shift_reg", "mvn_imm", "mvn_reg", "mvn_shift" and "mvn_shift_reg".
15058         (not_shiftsi): Update for attribute change.
15059         (not_shiftsi_compare0): Likewise.
15060         (not_shiftsi_compare0_scratch): Likewise.
15061         (arm_one_cmplsi2): Likewise.
15062         (thumb1_one_cmplsi2): Likewise.
15063         (notsi_compare0): Likewise.
15064         (notsi_compare0_scratch): Likewise.
15065         (thumb1_movdi_insn): Likewise.
15066         (arm_movsi_insn): Likewise.
15067         (movhi_insn_arch4): Likewise.
15068         (movhi_bytes): Likewise.
15069         (arm_movqi_insn): Likewise.
15070         (thumb1_movqi_insn): Likewise.
15071         (arm32_movhf): Likewise.
15072         (thumb1_movhf): Likewise.
15073         (arm_movsf_soft_insn): Likewise.
15074         (thumb1_movsf_insn): Likewise.
15075         (thumb_movdf_insn): Likewise.
15076         (movsicc_insn): Likewise.
15077         (movsfcc_soft_insn): Likewise.
15078         (and_scc): Likewise.
15079         (cond_move): Likewise.
15080         (if_move_not): Likewise.
15081         (if_not_move): Likewise.
15082         (if_shift_move): Likewise.
15083         (if_move_shift): Likewise.
15084         (if_shift_shift): Likewise.
15085         (if_not_arith): Likewise.
15086         (if_arith_not): Likewise.
15087         (cond_move_not): Likewise.
15088         * config/arm/neon.md (neon_mov<mode>): Update for attribute change.
15089         (neon_mov<mode>): Likewise.
15090         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute change.
15091         (thumb2_movsi_vfp): Likewise.
15092         (movsf_vfp): Likewise.
15093         (thumb2_movsf_vfp): Likewise.
15094         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15095         change.
15096         (cortexa7_older_only): Likewise.
15097         (cortexa7_younger): Likewise.
15098         * config/arm/arm1020e.md (1020alu_op): Update for attribute change.
15099         (1020alu_shift_op): Likewise.
15100         (1020alu_shift_reg_op): Likewise.
15101         * config/arm/arm1026ejs.md (alu_op): Update for attribute change.
15102         (alu_shift_op): Likewise.
15103         (alu_shift_reg_op): Likewise.
15104         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute change.
15105         (11_alu_shift_op): Likewise.
15106         (11_alu_shift_reg_op): Likewise.
15107         * config/arm/arm926ejs.md (9_alu_op): Update for attribute change.
15108         (9_alu_shift_reg_op): Likewise.
15109         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15110         change.
15111         (cortex_a15_alu_shift): Likewise.
15112         (cortex_a15_alu_shift_reg): Likewise.
15113         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute change.
15114         (cortex_a5_alu_shift): Likewise.
15115         * config/arm/cortex-a53.md (cortex_a53_alu): Update for attribute
15116         change.
15117         (cortex_a53_alu_shift): Likewise.
15118         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15119         change.
15120         (cortex_a7_alu_reg): Likewise.
15121         (cortex_a7_alu_shift): Likewise.
15122         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15123         (cortex_a8_alu_shift): Likewise.
15124         (cortex_a8_alu_shift_reg): Likewise.
15125         (cortex_a8_mov): Likewise.
15126         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute change.
15127         (cortex_a9_dp_shift): Likewise.
15128         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute change.
15129         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute change.
15130         (cortex_r4_mov): Likewise.
15131         (cortex_r4_alu_shift): Likewise.
15132         (cortex_r4_alu_shift_reg): Likewise.
15133         * config/arm/fa526.md (526_alu_op): Update for attribute change.
15134         (526_alu_shift_op): Likewise.
15135         * config/arm/fa606te.md (606te_alu_op): Update for attribute change.
15136         * config/arm/fa626te.md (626te_alu_op): Update for attribute change.
15137         (626te_alu_shift_op): Likewise.
15138         * config/arm/fa726te.md (726te_shift_op): Update for attribute change.
15139         (726te_alu_op): Likewise.
15140         (726te_alu_shift_op): Likewise.
15141         (726te_alu_shift_reg_op): Likewise.
15142         * config/arm/fmp626.md (mp626_alu_op): Update for attribute change.
15143         (mp626_alu_shift_op): Likewise.
15144         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute change.
15145         (pj4_alu_e1_conds): Likewise.
15146         (pj4_alu): Likewise.
15147         (pj4_alu_conds): Likewise.
15148         (pj4_shift): Likewise.
15149         (pj4_shift_conds): Likewise.
15150         (pj4_alu_shift): Likewise.
15151         (pj4_alu_shift_conds): Likewise.
15153 2013-07-22  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
15155         * config/arm/predicates.md (shiftable_operator_strict_it):
15156         New predicate.
15157         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si):
15158         Disable cond_exec version for arm_restrict_it.
15159         (thumb2_smaxsi3): Convert to generate cond_exec.
15160         (thumb2_sminsi3): Likewise.
15161         (thumb32_umaxsi3): Likewise.
15162         (thumb2_uminsi3): Likewise.
15163         (thumb2_abssi2): Adjust constraints for arm_restrict_it.
15164         (thumb2_neg_abssi2): Likewise.
15165         (thumb2_mov_scc): Add alternative for 16-bit encoding.
15166         (thumb2_movsicc_insn): Adjust alternatives.
15167         (thumb2_mov_negscc): Disable for arm_restrict_it.
15168         (thumb2_mov_negscc_strict_it): New pattern.
15169         (thumb2_mov_notscc_strict_it): New pattern.
15170         (thumb2_mov_notscc): Disable for arm_restrict_it.
15171         (thumb2_ior_scc): Likewise.
15172         (thumb2_ior_scc_strict_it): New pattern.
15173         (thumb2_cond_move): Adjust for arm_restrict_it.
15174         (thumb2_cond_arith): Disable for arm_restrict_it.
15175         (thumb2_cond_arith_strict_it): New pattern.
15176         (thumb2_cond_sub): Adjust for arm_restrict_it.
15177         (thumb2_movcond): Likewise.
15178         (thumb2_extendqisi_v6): Disable cond_exec variant for arm_restrict_it.
15179         (thumb2_zero_extendhisi2_v6): Likewise.
15180         (thumb2_zero_extendqisi2_v6): Likewise.
15181         (orsi_notsi_si): Likewise.
15182         (orsi_not_shiftsi_si): Likewise.
15184 2013-07-22  Georg-Johann Lay  <avr@gjlay.de>
15186         * config/avr/avr.c (avr_out_xload): No SBIS around LPM so that
15187         instruction sequence is 1 byte shorter.
15189 2013-07-22  Uros Bizjak  <ubizjak@gmail.com>
15191         * config/i386/i386.md (nonlocal_goto_receiver): Delete insn if
15192         it is not needed after split.
15194 2013-07-20  Iain Sandoe  <iain@codesourcery.com>
15196         PR target/51784
15197         * config/i386/i386.c (output_set_got) [TARGET_MACHO]: Adjust to emit a
15198         second label for nonlocal goto receivers. Don't output pic base labels
15199         unless we're producing PIC; mark that action unreachable().
15200         (ix86_save_reg): If the function contains a nonlocal label, save the
15201         PIC base reg.
15202         * config/darwin-protos.h (machopic_should_output_picbase_label): New.
15203         * config/darwin.c (emitted_pic_label_num): New GTY.
15204         (update_pic_label_number_if_needed): New.
15205         (machopic_output_function_base_name): Adjust for nonlocal receiver
15206         case.
15207         (machopic_should_output_picbase_label): New.
15208         * config/i386/i386.md (enum unspecv): UNSPECV_NLGR: New.
15209         (nonlocal_goto_receiver): New insn and split.
15211 2013-07-20  James Greenhalgh  <james.greenhalgh@arm.com>
15213         * config/aarch64/aarch64-builtins.c
15214         (aarch64_fold_builtin): Fold abs in all modes.
15215         * config/aarch64/aarch64-simd-builtins.def
15216         (abs): Enable for all modes.
15217         * config/aarch64/arm_neon.h
15218         (vabs<q>_s<8,16,32,64): Rewrite using builtins.
15219         (vabs_f64): Add missing intrinsic.
15221 2013-07-19  Ian Bolton  <ian.bolton@arm.com>
15223         * config/aarch64/arm_neon.h (vabs_s64): New function
15225 2013-07-19  Georg-Johann Lay  <avr@gjlay.de>
15227         PR target/57516
15228         * config/avr/avr-fixed.md (round<mode>3_const): Turn expander to insn.
15229         * config/avr/avr.md (adjust_len): Add `round'.
15230         * config/avr/avr-protos.h (avr_out_round): New prototype.
15231         (avr_out_plus): Add `out_label' argument.
15232         * config/avr/avr.c (avr_out_plus_1): Add `out_label' argument.
15233         (avr_out_plus): Pass down `out_label' to avr_out_plus_1.
15234         Handle the case where `insn' is just a pattern.
15235         (avr_out_bitop): Handle the case where `insn' is just a pattern.
15236         (avr_out_round): New function.
15237         (avr_adjust_insn_length): Handle ADJUST_LEN_ROUND.
15239 2013-07-18  David Holsgrove  <david.holsgrove@xilinx.com>
15241         * config/microblaze/microblaze.c (microblaze_expand_prologue):
15242         Add check for flag_stack_usage to handle -fstack-usage support
15244 2013-07-18  Pat Haugen  <pthaugen@us.ibm.com>
15246         * config/rs6000/rs6000.c (rs6000_option_override_internal): Adjust flag
15247         interaction for new Power8 flags and VSX.
15249 2013-07-18  Sriraman Tallam  <tmsriram@google.com>
15251         PR middle-end/57698
15252         * tree-inline.c (expand_call_inline): Emit errors during
15253         early_inlining only if optimization is not turned on.
15255 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15257         * passes.def: New.
15259         * passes.c (init_optimization_passes): Move the construction of
15260         the pass hierarchy into a new passes.def file.
15262         * Makefile.in (passes.o): Add dependency on passes.def.
15264 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15266         * passes.c (init_optimization_passes): Introduce macros for
15267         constructing the tree of passes (INSERT_PASSES_AFTER,
15268         PUSH_INSERT_PASSES_WITHIN, POP_INSERT_PASSES,
15269         TERMINATE_PASS_LIST).
15271 2013-07-18  Vladimir Makarov  <vmakarov@redhat.com>
15272             Wei Mi  <wmi@google.com>
15274         PR rtl-optimization/57878
15275         * lra-assigns.c (assign_by_spills): Move non_reload_pseudos to the
15276         top.
15277         (reload_pseudo_compare_func): Check nregs first for reload
15278         pseudos.
15280 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15282         * tree-pass.h (pass_ipa_lto_wpa_fixup): Remove redundant decl.
15284 2013-07-18  Po-Chun Chang  <pchang9@cs.wisc.edu>
15286         * read-rtl.c (validate_const_int): Once an invalid character is
15287         seen, quit the loop.
15289         * gengtype.c (write_roots): Similarly once we find the "deletable"
15290         or "if_marked" option.
15292 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15294         * config/arm/arm.md (attribute "insn"): Delete values "mrs", "msr",
15295         "xtab" and "sat".  Move value "clz" from here to ...
15296         (attriubte "type"): ... here.
15297         (satsi_<SAT:code>): Delete "insn" attribute.
15298         (satsi_<SAT:code>_shift): Likewise.
15299         (arm_zero_extendqisi2addsi): Likewise.
15300         (arm_extendqisi2addsi): Likewise.
15301         (clzsi2): Update for attribute changes.
15302         (rbitsi2): Likewise.
15303         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Delete "insn"
15304         attribute.
15305         (arm_usatsihi): Likewise.
15306         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute change.
15308 2013-07-18  Sofiane Naci  <sofiane.naci@arm.com>
15310         * config/arm/arm.md (attribute "type"): Rename "simple_alu_imm" to
15311         "arlo_imm".  Rename "alu_reg" to "arlo_reg".  Rename "simple_alu_shift"
15312         to "extend".  Split "alu_shift" into "shift" and "arlo_shift".  Split
15313         "alu_shift_reg" into "shift_reg" and "arlo_shift_reg".  List types
15314         in alphabetical order.
15315         (attribute "core_cycles"): Update for attribute changes.
15316         (arm_addsi3): Likewise.
15317         (addsi3_compare0): Likewise.
15318         (addsi3_compare0_scratch): Likewise.
15319         (addsi3_compare_op1): Likewise.
15320         (addsi3_compare_op2): Likewise.
15321         (compare_addsi2_op0): Likewise.
15322         (compare_addsi2_op1): Likewise.
15323         (addsi3_carryin_shift_<optab>): Likewise.
15324         (subsi3_carryin_shift): Likewise.
15325         (rsbsi3_carryin_shift): Likewise.
15326         (arm_subsi3_insn): Likewise.
15327         (subsi3_compare0): Likewise.
15328         (subsi3_compare): Likewise.
15329         (arm_andsi3_insn): Likewise.
15330         (thumb1_andsi3_insn): Likewise.
15331         (andsi3_compare0): Likewise.
15332         (andsi3_compare0_scratch): Likewise.
15333         (zeroextractsi_compare0_scratch
15334         (andsi_not_shiftsi_si): Likewise.
15335         (iorsi3_insn): Likewise.
15336         (iorsi3_compare0): Likewise.
15337         (iorsi3_compare0_scratch): Likewise.
15338         (arm_xorsi3): Likewise.
15339         (thumb1_xorsi3_insn): Likewise.
15340         (xorsi3_compare0): Likewise.
15341         (xorsi3_compare0_scratch): Likewise.
15342         (satsi_<SAT:code>_shift): Likewise.
15343         (rrx): Likewise.
15344         (arm_shiftsi3): Likewise.
15345         (shiftsi3_compare0): Likewise.
15346         (not_shiftsi): Likewise.
15347         (not_shiftsi_compare0): Likewise.
15348         (not_shiftsi_compare0_scratch): Likewise.
15349         (arm_one_cmplsi2): Likewise.
15350         (thumb_one_complsi2): Likewise.
15351         (notsi_compare0): Likewise.
15352         (notsi_compare0_scratch): Likewise.
15353         (thumb1_zero_extendhisi2): Likewise.
15354         (arm_zero_extendhisi2): Likewise.
15355         (arm_zero_extendhisi2_v6): Likewise.
15356         (arm_zero_extendhisi2addsi): Likewise.
15357         (thumb1_zero_extendqisi2): Likewise.
15358         (thumb1_zero_extendqisi2_v6): Likewise.
15359         (arm_zero_extendqisi2): Likewise.
15360         (arm_zero_extendqisi2_v6): Likewise.
15361         (arm_zero_extendqisi2addsi): Likewise.
15362         (thumb1_extendhisi2): Likewise.
15363         (arm_extendhisi2): Likewise.
15364         (arm_extendhisi2_v6): Likewise.
15365         (arm_extendqisi): Likewise.
15366         (arm_extendqisi_v6): Likewise.
15367         (arm_extendqisi2addsi): Likewise.
15368         (thumb1_extendqisi2): Likewise.
15369         (thumb1_movdi_insn): Likewise.
15370         (arm_movsi_insn): Likewise.
15371         (movsi_compare0): Likewise.
15372         (movhi_insn_arch4): Likewise.
15373         (movhi_bytes): Likewise.
15374         (arm_movqi_insn): Likewise.
15375         (thumb1_movqi_insn): Likewise.
15376         (arm32_movhf): Likewise.
15377         (thumb1_movhf): Likewise.
15378         (arm_movsf_soft_insn): Likewise.
15379         (thumb1_movsf_insn): Likewise.
15380         (movdf_soft_insn): Likewise.
15381         (thumb_movdf_insn): Likewise.
15382         (arm_cmpsi_insn): Likewise.
15383         (cmpsi_shiftsi): Likewise.
15384         (cmpsi_shiftsi_swp): Likewise.
15385         (arm_cmpsi_negshiftsi_si): Likewise.
15386         (movsicc_insn): Likewise.
15387         (movsfcc_soft_insn): Likewise.
15388         (arith_shiftsi): Likewise.
15389         (arith_shiftsi_compare0
15390         (arith_shiftsi_compare0_scratch
15391         (sub_shiftsi): Likewise.
15392         (sub_shiftsi_compare0
15393         (sub_shiftsi_compare0_scratch
15394         (and_scc): Likewise.
15395         (cond_move): Likewise.
15396         (if_plus_move): Likewise.
15397         (if_move_plus): Likewise.
15398         (if_move_not): Likewise.
15399         (if_not_move): Likewise.
15400         (if_shift_move): Likewise.
15401         (if_move_shift): Likewise.
15402         (if_shift_shift): Likewise.
15403         (if_not_arith): Likewise.
15404         (if_arith_not): Likewise.
15405         (cond_move_not): Likewise.
15406         (thumb1_ashlsi3): Set type attribute.
15407         (thumb1_ashrsi3): Likewise.
15408         (thumb1_lshrsi3): Likewise.
15409         (thumb1_rotrsi3): Likewise.
15410         (shiftsi3_compare0_scratch): Likewise.
15411         * config/arm/neon.md (neon_mov<mode>): Update for attribute changes.
15412         (neon_mov<mode>): Likewise.
15413         * config/arm/thumb2.md (thumb_andsi_not_shiftsi_si): Update for
15414         attribute changes.
15415         (thumb2_movsi_insn): Likewise.
15416         (thumb2_cmpsi_neg_shiftsi): Likewise.
15417         (thumb2_extendqisi_v6): Likewise.
15418         (thumb2_zero_extendhisi2_v6): Likewise.
15419         (thumb2_zero_extendqisi2_v6): Likewise.
15420         (thumb2_shiftsi3_short): Likewise.
15421         (thumb2_addsi3_compare0_scratch): Likewise.
15422         (orsi_not_shiftsi_si): Likewise.
15423         * config/arm/vfp.md (arm_movsi_vfp): Update for attribute changes.
15424         * config/arm/arm-fixed.md (arm_ssatsihi_shift): Update for attribute
15425         changes.
15426         * config/arm/arm1020e.md (1020alu_op): Update for attribute changes.
15427         (1020alu_shift_op): Likewise.
15428         (1020alu_shift_reg_op): Likewise.
15429         * config/arm/arm1026ejs.md (alu_op): Update for attribute changes.
15430         (alu_shift_op): Likewise.
15431         (alu_shift_reg_op): Likewise.
15432         * config/arm/arm1136jfs.md (11_alu_op): Update for attribute changes.
15433         (11_alu_shift_op): Likewise.
15434         (11_alu_shift_reg_op): Likewise.
15435         * config/arm/arm926ejs.md (9_alu_op): Update for attribute changes.
15436         (9_alu_shift_reg_op): Likewise.
15437         * config/arm/cortex-a15.md (cortex_a15_alu): Update for attribute
15438         changes.
15439         (cortex_a15_alu_shift): Likewise.
15440         (cortex_a15_alu_shift_reg): Likewise.
15441         * config/arm/cortex-a5.md (cortex_a5_alu): Update for attribute
15442         changes.
15443         (cortex_a5_alu_shift): Likewise.
15444         * config/arm/cortex-a53.md (cortex_a53_alu) : Update for attribute
15445         changes.
15446         (cortex_a53_alu_shift): Likewise.
15447         * config/arm/cortex-a7.md (cortex_a7_alu_imm): Update for attribute
15448         changes.
15449         (cortex_a7_alu_reg): Likewise.
15450         (cortex_a7_alu_shift): Likewise.
15451         * config/arm/cortex-a8.md (cortex_a8_alu): Update for attribute
15452         changes.
15453         (cortex_a8_alu_shift): Likewise.
15454         (cortex_a8_alu_shift_reg): Likewise.
15455         (cortex_a8_mov): Likewise.
15456         * config/arm/cortex-a9.md (cortex_a9_dp): Update for attribute changes.
15457         (cortex_a9_dp_shift): Likewise.
15458         * config/arm/cortex-m4.md (cortex_m4_alu): Update for attribute
15459         changes.
15460         * config/arm/cortex-r4.md (cortex_r4_alu): Update for attribute
15461         changes.
15462         (cortex_r4_mov): Likewise.
15463         (cortex_r4_alu_shift): Likewise.
15464         (cortex_r4_alu_shift_reg): Likewise.
15465         * config/arm/fa526.md (526_alu_op): Update for attribute changes.
15466         (526_alu_shift_op): Likewise.
15467         * config/arm/fa606te.md (606te_alu_op): Update for attribute changes.
15468         * config/arm/fa626te.md (626te_alu_op): Update for attribute changes.
15469         (626te_alu_shift_op): Likewise.
15470         * config/arm/fa726te.md (726te_shift_op): Update for attribute changes.
15471         (726te_alu_op): Likewise.
15472         (726te_alu_shift_op): Likewise.
15473         (726te_alu_shift_reg_op): Likewise.
15474         * config/arm/fmp626.md (mp626_alu_op): Update for attribute changes.
15475         (mp626_alu_shift_op): Likewise.
15476         * config/arm/marvell-pj4.md (pj4_alu_e1): Update for attribute changes.
15477         (pj4_alu_e1_conds): Likewise.
15478         (pj4_alu): Likewise.
15479         (pj4_alu_conds): Likewise.
15480         (pj4_shift): Likewise.
15481         (pj4_shift_conds): Likewise.
15482         (pj4_alu_shift): Likewise.
15483         (pj4_alu_shift_conds): Likewise.
15484         * config/arm/arm.c (xscale_sched_adjust_cost): Update for attribute
15485         changes.
15486         (cortexa7_older_only): Likewise.
15487         (cortexa7_younger): Likewise.
15489 2013-07-18  David Malcolm  <dmalcolm@redhat.com>
15491         * ipa-pure-const.c (generate_summary): Rename to...
15492         (pure_const_generate_summary): ... this.
15494 2013-07-17  Iain Sandoe  <iain@codesourcery.com>
15496         * config/rs6000/darwin.h (REGISTER_NAMES): Add HTM registers.
15498 2013-07-17  Yvan Roux  <yvan.roux@linaro.org>
15500         PR target/57909
15501         * config/arm/arm.c (gen_movmem_ldrd_strd): Fix unaligned load/store
15502         usage in HI mode.
15504 2013-07-17  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15506         * config/s390/s390.c: (s390_expand_builtin): Allow -mhtm to be
15507         enabled without -march=zEC12.
15508         * config/s390/s390.h (TARGET_HTM): Do not require EC12 machine
15509         flags to be set.
15511 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15513         * config/mips/mips.h (ISA_HAS_FP4): Correct formatting.
15514         (ISA_HAS_FP_MADD4_MSUB4): Also enable for ISA_MIPS32R2.
15515         (ISA_HAS_NMADD4_NMSUB4): Remove the MODE argument; rewrite in
15516         terms of ISA_HAS_FP4, and also enable for ISA_MIPS32R2.
15517         (ISA_HAS_NMADD3_NMSUB3): Remove the MODE argument.
15518         * config/mips/mips.c (mips_rtx_costs) <PLUS>: Check for
15519         ISA_HAS_FP_MADD4_MSUB4 || ISA_HAS_FP_MADD3_MSUB3 rather than
15520         ISA_HAS_FP4.
15521         <MINUS, NEG>: Update according to changes to ISA_HAS_NMADD4_NMSUB4
15522         and ISA_HAS_NMADD3_NMSUB3.
15523         * config/mips/mips.md (nmadd4<mode>, nmadd3<mode>): Likewise.
15524         (nmadd4<mode>_fastmath, nmadd3<mode>_fastmath): Likewise.
15525         (nmsub4<mode>, nmsub3<mode>): Likewise.
15526         (nmsub4<mode>_fastmath, nmsub3<mode>_fastmath): Likewise.
15528 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15530         * config/mips/mips.h (ISA_HAS_NMADD4_NMSUB4): Remove
15531         TARGET_MIPS5400 checking.
15533 2013-07-16  Jakub Jelinek  <jakub@redhat.com>
15534             Peter Bergner  <bergner@vnet.ibm.com>
15536         * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTERS): Mention HTM
15537         registers in the comment.
15538         (DWARF_FRAME_REGISTERS): Subtract also the 3 HTM registers.
15539         (DWARF_REG_TO_UNWIND_COLUMN): Use DWARF_FRAME_REGISTERS
15540         rather than FIRST_PSEUDO_REGISTERS.
15542 2013-07-16  Peter Bergner  <bergner@vnet.ibm.com>
15544         * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15545         enable extra ISA flags with TARGET_HTM.
15547 2013-07-16  Maciej W. Rozycki  <macro@codesourcery.com>
15549         * config/mips/mips.h (ISA_HAS_MULS, ISA_HAS_MSAC, ISA_HAS_MACC):
15550         Fix comment typos.
15552 2013-07-15  Cong Hou  <congh@google.com>
15554         * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use hash function
15555         in compare function for sorting.
15557 2013-07-15  Peter Bergner  <bergner@vnet.ibm.com>
15559         * config.gcc (powerpc*-*-*): Install htmintrin.h and htmxlintrin.h.
15560         * config/rs6000/t-rs6000 (MD_INCLUDES): Add htm.md.
15561         * config/rs6000/rs6000.opt: Add -mhtm option.
15562         * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Add OPTION_MASK_HTM.
15563         (ISA_2_7_MASKS_SERVER): Add OPTION_MASK_HTM.
15564         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
15565         __HTM__ if the HTM instructions are available.
15566         * config/rs6000/predicates.md (u3bit_cint_operand, u10bit_cint_operand,
15567         htm_spr_reg_operand): New define_predicates.
15568         * config/rs6000/rs6000.md (define_attr "type"): Add htm.
15569         (TFHAR_REGNO, TFIAR_REGNO, TEXASR_REGNO): New define_constants.
15570         Include htm.md.
15571         * config/rs6000/rs6000-builtin.def (BU_HTM_0, BU_HTM_1, BU_HTM_2,
15572         BU_HTM_3, BU_HTM_SPR0, BU_HTM_SPR1): Add support macros for defining
15573         HTM builtin functions.
15574         * config/rs6000/rs6000.c (RS6000_BUILTIN_H): New macro.
15575         (rs6000_reg_names, alt_reg_names): Add HTM SPR register names.
15576         (rs6000_init_hard_regno_mode_ok): Add support for HTM instructions.
15577         (rs6000_builtin_mask_calculate): Likewise.
15578         (rs6000_option_override_internal): Likewise.
15579         (bdesc_htm): Add new HTM builtin support.
15580         (htm_spr_num): New function.
15581         (htm_spr_regno): Likewise.
15582         (rs6000_htm_spr_icode): Likewise.
15583         (htm_expand_builtin): Likewise.
15584         (htm_init_builtins): Likewise.
15585         (rs6000_expand_builtin): Add support for HTM builtin functions.
15586         (rs6000_init_builtins): Likewise.
15587         (rs6000_invalid_builtin, rs6000_opt_mask): Add support for -mhtm
15588         option.
15589         * config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mhtm.
15590         (TARGET_HTM, MASK_HTM): Define macros.
15591         (FIRST_PSEUDO_REGISTER): Adjust for new HTM SPR registers.
15592         (FIXED_REGISTERS): Likewise.
15593         (CALL_USED_REGISTERS): Likewise.
15594         (CALL_REALLY_USED_REGISTERS): Likewise.
15595         (REG_ALLOC_ORDER): Likewise.
15596         (enum reg_class): Likewise.
15597         (REG_CLASS_NAMES): Likewise.
15598         (REG_CLASS_CONTENTS): Likewise.
15599         (REGISTER_NAMES): Likewise.
15600         (ADDITIONAL_REGISTER_NAMES): Likewise.
15601         (RS6000_BTC_SPR, RS6000_BTC_VOID, RS6000_BTC_32BIT, RS6000_BTC_64BIT,
15602         RS6000_BTC_MISC_MASK, RS6000_BTM_HTM): New macros.
15603         (RS6000_BTM_COMMON): Add RS6000_BTM_HTM.
15604         * config/rs6000/htm.md: New file.
15605         * config/rs6000/htmintrin.h: New file.
15606         * config/rs6000/htmxlintrin.h: New file.
15608 2013-07-15  Marcus Shawcroft  <marcus.shawcroft@arm.com>
15610         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
15611         Define SYMBOL_TINY_GOT, update comment.
15612         * config/aarch64/aarch64.c
15613         (aarch64_load_symref_appropriately): Handle SYMBOL_TINY_GOT.
15614         (aarch64_expand_mov_immediate): Likewise.
15615         (aarch64_print_operand): Likewise.
15616         (aarch64_classify_symbol): Likewise.
15617         * config/aarch64/aarch64.md (UNSPEC_GOTTINYPIC): Define.
15618         (ldr_got_tiny): Define.
15620 2013-07-13  Tobias Grosser  <tobias@grosser.es>
15622         PR tree-optimization/54094
15623         * graphite-clast-to-gimple.c (translate_clast_for_loop): Derive the
15624           scheduling dimension for the parallelism check from the polyhedral
15625           information in the AST.
15626         * graphite-dependences.c (carries_deps): Do not assume the schedule is
15627           in 2D + 1 form.
15629 2013-07-13  Jason Merrill  <jason@redhat.com>
15631         * print-tree.c (debug_vec_tree): Use debug_raw.
15632         (debug_raw (vec<tree, va_gc> &)): New.
15633         (debug_raw (vec<tree, va_gc> *)): New.
15634         * tree.h: Declare them.
15636 2013-07-13  Bin Cheng  <bin.cheng@arm.com>
15638         * ifcvt.c (ifcvt_after_combine): New static variable.
15639         (cheap_bb_rtx_cost_p): Set scale to REG_BR_PROB_BASE when optimizing
15640         for size.
15641         (if_convert): New parameter after_combine.  Set ifcvt_after_combine.
15642         (rest_of_handle_if_conversion, rest_of_handle_if_after_combine,
15643         rest_of_handle_if_after_reload): Pass new argument for if_convert.
15645 2013-07-12  Maciej W. Rozycki  <macro@codesourcery.com>
15647         * config/mips/mips.c (mips_expand_call): Remove empty statement.
15649 2013-07-12  Michael Matz  <matz@suse.de>
15651         PR middle-end/55771
15652         * convert.c (convert_to_real): Reject non-float inner types.
15654 2013-07-12  Tejas Belagod  <tejas.belagod@arm.com>
15656         * config/aarch64/aarch64-protos.h
15657         (aarch64_simd_immediate_valid_for_move): Remove.
15658         * config/aarch64/aarch64.c (simd_immediate_info): New member.
15659         (aarch64_simd_valid_immediate): Recognize idioms for shifting ones
15660         cases.
15661         (aarch64_output_simd_mov_immediate): Print the correct shift specifier.
15663 2013-07-11  Steve Ellcey  <sellcey@mips.com>
15665         * config/mips/mips.c (mips_conditional_register_usage): Do not
15666         use t[0-7] registers in MIPS16 mode when optimizing for size.
15668 2013-07-11  Sriraman Tallam  <tmsriram@google.com>
15670         * config/i386/i386.c (dispatch_function_versions): Fix array
15671         indexing of function_version_info to match actual_versions.
15673 2013-07-11  Teresa Johnson  <tejohnson@google.com>
15675         * vec.h (struct va_gc): Move release out-of-line.
15676         (va_gc::release): Call ggc_free on released vec.
15678 2013-07-11  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
15680         * config/rs6000/rs6000.md (""*tls_gd_low<TLSmode:tls_abi_suffix>"):
15681         Require GOT register as additional operand in UNSPEC.
15682         ("*tls_ld_low<TLSmode:tls_abi_suffix>"): Likewise.
15683         ("*tls_got_dtprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15684         ("*tls_got_tprel_low<TLSmode:tls_abi_suffix>"): Likewise.
15685         ("*tls_gd<TLSmode:tls_abi_suffix>"): Update splitter.
15686         ("*tls_ld<TLSmode:tls_abi_suffix>"): Likewise.
15687         ("tls_got_dtprel_<TLSmode:tls_abi_suffix>"): Likewise.
15688         ("tls_got_tprel_<TLSmode:tls_abi_suffix>"): Likewise.
15690 2013-07-11  Georg-Johann Lay  <avr@gjlay.de>
15692         PR target/57631
15693         * config/avr/avr.c (avr_set_current_function): Sanity-check signal
15694         name seen by assembler/linker rather if available.
15696 2013-07-11  Andreas Schwab  <schwab@suse.de>
15698         * config/aarch64/aarch64-linux.h (CPP_SPEC): Define.
15700 2013-07-10  Vladimir Makarov  <vmakarov@redhat.com>
15702         * lra-constraints.c (curr_insn_transform): Switch off optional reloads.
15704 2013-07-10  Joseph Myers  <joseph@codesourcery.com>
15706         * doc/tm.texi.in: Move hook documentation to ....
15707         * target.def: ... here.
15709         * doc/tm.texi.in (TARGET_CANONICALIZE_COMPARISON): Remove stray
15710         text on @hook line.
15711         * doc/tm.texi: Regenerate.
15713 2013-07-10  Paolo Carlini  <paolo.carlini@oracle.com>
15715         PR c++/57869
15716         * doc/invoke.texi: Document -Wconditionally-supported.
15718 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15720         PR target/57844
15721         * config/avr/avr.c (avr_prologue_setup_frame): Trunk -size to mode
15722         of my_fp.
15724 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15726         PR target/57506
15727         * config/avr/avr-mcus.def (atmega16hva, atmega16hva2, atmega16hvb)
15728         (atmega16m1, atmega16u4, atmega32a, atmega32c1, atmega32hvb)
15729         (atmega32m1, atmega32u4, atmega32u6, atmega64c1, atmega64m1):
15730         Remove duplicate devices.
15731         * config/avr/gen-avr-mmcu-texi.c (print_mcus): Fail on duplicate MCUs.
15732         * config/avr/t-multilib: Regenerate.
15733         * config/avr/avr-tables.opt: Regenerate.
15734         * doc/avr-mmcu.texi: Regenerate.
15736 2013-07-10  Georg-Johann Lay  <avr@gjlay.de>
15738         PR target/56987
15739         * config/avr/avr.opt (Waddr-space-convert): Fix typo.
15741 2013-07-10  Graham Stott  <graham.stott@btinternet.com>
15743         * config/mips/mips.c (mips_rtx_costs): Very slightly increase
15744         the cost of MULT when optimizing for size.
15746 2013-07-10  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
15748         * config/cr16/cr16-protos.h: Don't include target.h.
15750 2013-07-09  Joseph Myers  <joseph@codesourcery.com>
15752         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Only
15753         adjust register size for TDmode and TFmode for VSX registers.
15755 2013-07-08  Kai Tietz  <ktietz@redhat.com>
15757         PR target/56892
15758         * config/i386/i386.c (TARGET_FUNCTION_ATTRIBUTE_INLINABLE_P): Define as
15759         hook_bool_const_tree_true.
15761 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15763         * config/s390/s390.c: Replace F*_REGNUM with FPR*_REGNUM.
15764         * config/s390/s390.h: Remove F*_REGNUM macro definitions.
15765         * config/s390/s390.md: Define FPR*_REGNUM constants.
15766         Fix FPR2_REGNUM constant (18 -> 17).
15767         ("*trunc<BFP:mode><DFP_ALL:mode>2")
15768         ("*trunc<DFP_ALL:mode><BFP:mode>2")
15769         ("trunc<BFP:mode><DFP_ALL:mode>2")
15770         ("trunc<DFP_ALL:mode><BFP:mode>2")
15771         ("*extend<BFP:mode><DFP_ALL:mode>2")
15772         ("*extend<DFP_ALL:mode><BFP:mode>2")
15773         ("extend<BFP:mode><DFP_ALL:mode>2")
15774         ("extend<DFP_ALL:mode><BFP:mode>2"): Replace FPR2_REGNUM with
15775         FPR4_REGNUM.
15777 2013-07-08  Graham Stott  <graham.stott@btinternet.com>
15779         * Makefile.in: (c-family-warn): Define to $(STRICT_WARN)
15781 2013-07-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
15783         * config/s390/s390.c: Rename cfun_set_fpr_bit to cfun_set_fpr_save
15784         and cfun_fpr_bit_p to cfun_fpr_save_p.
15785         (s390_frame_area, s390_register_info, s390_frame_info)
15786         (s390_emit_prologue, s390_emit_epilogue)
15787         (s390_conditional_register_usage): Use the *_REGNUM macros for FPR
15788         register numbers.
15789         * config/s390/s390.h: Define *_REGNUM macros for floating point
15790         register numbers.
15792 2013-07-08  Eric Botcazou  <ebotcazou@adacore.com>
15794         * Makefile.in (tree-ssa-reassoc.o): Add dependency on $(PARAMS_H).
15796 2013-07-08  Po-Chun Chang  <pchang9@cs.wisc.edu>
15798         PR rtl-optimization/57786
15799         * combine.c (distribute_notes) <case REG_DEAD>: Change all_used to bool
15800         and break out of the loop when it is set to false.
15802 2013-07-08  Jakub Jelinek  <jakub@redhat.com>
15804         PR target/57819
15805         * simplify-rtx.c (simplify_unary_operation_1) <case ZERO_EXTEND>:
15806         Simplify (zero_extend:SI (subreg:QI (and:SI (reg:SI)
15807         (const_int 63)) 0)).
15808         * combine.c (make_extraction): Create ZERO_EXTEND or SIGN_EXTEND
15809         using simplify_gen_unary instead of gen_rtx_*_EXTEND.
15810         * config/i386/i386.md (*jcc_bt<mode>_1): New define_insn_and_split.
15812         PR rtl-optimization/57829
15813         * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Ensure that
15814         mask bits outside of mode are just sign-extension from mode to HWI.
15816 2013-07-08  Michael Zolotukhin  <michael.v.zolotukhin@gmail.com>
15818         * config/i386/i386-opts.h (enum stringop_alg): Add vector_loop.
15819         * config/i386/i386.c (expand_set_or_movmem_via_loop): Use
15820         adjust_address instead of change_address to keep info about alignment.
15821         (emit_strmov): Remove.
15822         (emit_memmov): New function.
15823         (expand_movmem_epilogue): Refactor to properly handle bigger sizes.
15824         (expand_movmem_epilogue): Likewise and return updated rtx for
15825         destination.
15826         (expand_constant_movmem_prologue): Likewise and return updated rtx for
15827         destination and source.
15828         (decide_alignment): Refactor, handle vector_loop.
15829         (ix86_expand_movmem): Likewise.
15830         (ix86_expand_setmem): Likewise.
15831         * config/i386/i386.opt (Enum): Add vector_loop to option stringop_alg.
15833 2013-07-07  Uros Bizjak  <ubizjak@gmail.com>
15835         * config/i386/driver-i386.c (host_detect_local_cpu): Do not check
15836         signature_TM2_ebx, it interferes with signature_INTEL_ebx.
15838 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15840         * config/i386/sse.md (sse_movlhps): Change alternative 3
15841         of operand 2 to "m".
15843 2013-07-06  Uros Bizjak  <ubizjak@gmail.com>
15845         PR target/57807
15846         * config/i386/sse.md (iptr): New mode attribute.
15847         (sse2_movq128): Add pointer size overrides for Intel asm dialect.
15848         (<sse>_vm<plusminus_insn><mode>3): Ditto.
15849         (<sse>_vmmul<mode>3): Ditto.
15850         (<sse>_vmdiv<mode>3): Ditto.
15851         (sse_vmrcpv4sf2): Ditto.
15852         (<sse>_vmsqrt<mode>2): Ditto.
15853         (sse_vmrsqrtv4sf2): Ditto.
15854         (<sse>_vm<code><mode>3): Ditto.
15855         (avx_vmcmp<mode>3): Ditto.
15856         (<sse>_vmmaskcmp<mode>3): Ditto.
15857         (<sse>_comi): Ditto.
15858         (<sse>_ucomi): Ditto.
15859         (*xop_vmfrcz_<mode>): Ditto.
15860         (*fmai_fmadd_<mode>): Ditto.
15861         (*fmai_fmsub_<mode>): Ditto.
15862         (*fmai_fnmadd_<mode>): Ditto.
15863         (*fmai_fnmsub_<mode>): Ditto.
15864         (*fma4i_vmfmadd_<mode>): Ditto.
15865         (*fma4i_vmfmsub_<mode>): Ditto.
15866         (*fma4i_vmfnmadd_<mode>): Ditto.
15867         (*fma4i_vmfnmsub_<mode>): Ditto.
15868         (*xop_vmfrcz_<mode>): Ditto.
15869         (sse_cvtps2pi): Ditto.
15870         (sse_cvttps2pi): Ditto.
15871         (sse_cvtss2si): Ditto.
15872         (sse_cvtss2si_2): Ditto.
15873         (sse_cvtss2siq_2): Ditto.
15874         (sse_cvttss2si): Ditto.
15875         (sse_cvttss2siq): Ditto.
15876         (sse_cvtsd2si): Ditto.
15877         (sse_cvtsd2si_2): Ditto.
15878         (sse_cvtsd2siq_2): Ditto.
15879         (sse_cvttsd2si): Ditto.
15880         (sse_cvttsd2siq): Ditto.
15881         (sse_cvtsd2ss): Ditto.
15882         (sse_cvtss2sd): Ditto.
15883         (avx2_pbroadcast<mode>): Ditto.
15884         (avx2_pbroadcast<mode>_1): Ditto.
15885         (*avx_vperm_broadcast_v4sf): Ditto.
15887         (sse_movhlps): Ditto for movlp[sd]/movhp[sd] alternatives.
15888         (sse_movlhps): Ditto.
15889         (sse_storehps): Ditto.
15890         (sse_loadhps): Ditto.
15891         (sse_storelps): Ditto.
15892         (sse_loadlps): Ditto.
15893         (*vec_concatv4sf): Ditto.
15894         (*vec_interleave_highv2df): Ditto.
15895         (*vec_interleave_lowv2df): Ditto.
15896         (*vec_extractv2df_1_sse): Ditto.
15897         (*vec_extractv2df_0_sse): Ditto.
15898         (sse2_storelpd): Ditto.
15899         (sse2_loadlpd): Ditto.
15900         (sse2_movsd): Ditto.
15901         (*vec_concatv4si): Ditto.
15902         (vec_concatv2di): Ditto.
15904         * config/i386/mmx.md (mmx_punpcklbw): Add pointer size overrides
15905         for Intel asm dialect.
15906         (mmx_punpcklwd): Ditto.
15907         (mmx_punpckldq): Ditto.
15909         * config/i386/i386.c (ix86_print_operand) ['H']: Output 'qword ptr'
15910         for intel assembler dialect.
15912 2013-07-06  Jakub Jelinek  <jakub@redhat.com>
15914         PR target/29776
15915         * fold-const.c (tree_call_nonnegative_warnv_p): Return true
15916         for BUILT_IN_C{LZ,LRSB}*.
15917         * tree.h (CASE_INT_FN): Add FN##IMAX case.
15918         * tree-vrp.c (extract_range_basic): Handle
15919         BUILT_IN_{FFS,PARITY,POPCOUNT,C{LZ,TZ,LRSB}}*.  For
15920         BUILT_IN_CONSTANT_P if argument isn't (D) of PARM_DECL,
15921         fall thru to code calling set_value*.
15922         * builtins.c (expand_builtin): Remove *IMAX cases.
15923         (fold_builtin_bitop): For BUILT_IN_CLRSB* return NULL_TREE
15924         if width is bigger than 2*HWI.
15926 2013-07-05  Vladimir Makarov  <vmakarov@redhat.com>
15928         PR rtl-optimization/55342
15929         * lra-int.h (lra_subreg_reload_pseudos): New.
15930         * lra.c: Add undoing optional reloads to the block diagram.
15931         (lra_subreg_reload_pseudos): New.
15932         (lra_optional_reload_pseudos): Change comments.
15933         (lra): Init and clear lra_subreg_reload_pseudos.  Clear
15934         lra_optional_reload_pseudos after undo transformations.
15935         * lra-assigns.c (pseudo_prefix_title): New.
15936         (lra_setup_reg_renumber): Use it.
15937         (spill_for): Ditto.  Check subreg reload pseudos too.
15938         (assign_by_spills): Consider subreg reload pseudos too.
15939         * lra-constraints.c (simplify_operand_subreg): Use
15940         lra_subreg_reload_pseudos instead of lra_optional_reload_pseudos.
15941         (curr_insn_transform): Recognize and do optional reloads.
15942         (undo_optional_reloads): New.
15943         (lra_undo_inheritance): Call undo_optional_reloads.
15945 2013-07-05  Thomas Quinot  <quinot@adacore.com>
15947         * tree-complex.c (expand_complex_operations_1): Fix typo.
15949 2013-07-04  Tejas Belagod  <tejas.belagod@arm.com>
15951         * config/aarch64/aarch64-protos.h (cpu_vector_cost): New.
15952         (tune_params): New member 'const vec_costs'.
15953         * config/aarch64/aarch64.c (generic_vector_cost): New.
15954         (generic_tunings): New member 'generic_vector_cost'.
15955         (aarch64_builtin_vectorization_cost): New.
15956         (aarch64_add_stmt_cost): New.
15957         (TARGET_VECTORIZE_ADD_STMT_COST): New.
15958         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST): New.
15960 2013-07-03  Jakub Jelinek  <jakub@redhat.com>
15962         PR target/57777
15963         * config/i386/predicates.md (vsib_address_operand): Disallow
15964         SYMBOL_REF or LABEL_REF in parts.disp if TARGET_64BIT && flag_pic.
15966 2013-07-03  Hans-Peter Nilsson  <hp@bitrange.com>
15968         PR middle-end/55030
15969         * stmt.c (expand_nl_goto_receiver): Remove almost-copy of
15970         expand_builtin_setjmp_receiver.
15971         (expand_label): Adjust, call expand_builtin_setjmp_receiver
15972         with NULL for the label parameter.
15973         * builtins.c (expand_builtin_setjmp_receiver): Don't clobber
15974         the frame-pointer.  Adjust comments.
15975         [HAVE_builtin_setjmp_receiver]: Emit builtin_setjmp_receiver
15976         only if LABEL is non-NULL.
15978 2013-07-03  Yufeng Zhang  <yufeng.zhang@arm.com>
15980         * config/aarch64/aarch64.h (enum arm_abi_type): Remove.
15981         (ARM_ABI_AAPCS64): Ditto.
15982         (arm_abi): Ditto.
15983         (ARM_DEFAULT_ABI): Ditto.
15985 2013-07-03  James Greenhalgh  <james.greenhalgh@arm.com>
15987         * config/aarch64/aarch64-builtins.c
15988         (aarch64_simd_expand_builtin): Handle AARCH64_SIMD_STORE1.
15989         * config/aarch64/aarch64-simd-builtins.def (ld1): New.
15990         (st1): Likewise.
15991         * config/aarch64/aarch64-simd.md
15992         (aarch64_ld1<VALL:mode>): New.
15993         (aarch64_st1<VALL:mode>): Likewise.
15994         * config/aarch64/arm_neon.h
15995         (vld1<q>_<fpsu><8, 16, 32, 64>): Convert to RTL builtins.
15997 2013-07-02  Sriraman Tallam  <tmsriram@google.com>
15999         * config/i386/i386.c (gate_insert_vzeroupper): Check if
16000         target ISA is AVX.
16001         (ix86_option_override_internal):Turn on all -mavx target flags by
16002         default as they are dependent on AVX anyway.
16004 2013-07-02  Cary Coutant  <ccoutant@google.com>
16006         * dwarf2out.c (loc_checksum): Call hash_loc_operands for a
16007         deterministic hash.
16008         (loc_checksum_ordered): Likewise.
16009         (hash_loc_operands): Remove inline keyword.
16011 2013-07-02  Jakub Jelinek  <jakub@redhat.com>
16013         PR tree-optimization/57741
16014         * tree-vect-loop.c (vect_is_simple_iv_evolution): Disallow
16015         non-INTEGRAL_TYPE_P non-SCALAR_FLOAT_TYPE_P SSA_NAME step_exprs,
16016         or SCALAR_FLOAT_TYPE_P SSA_NAMEs if !flag_associative_math.
16017         Allow REAL_CST step_exprs if flag_associative_math.
16018         (get_initial_def_for_induction): Handle SCALAR_FLOAT_TYPE_P step_expr.
16020 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
16022         * config/aarch64/aarch64-simd.md (absdi2): Support abs for DI mode.
16024 2013-07-02  Ian Bolton  <ian.bolton@arm.com>
16026         * config/aarch64/aarch64.md (*extr_insv_reg<mode>): New pattern.
16028 2013-07-02  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16030         * config/arm/arm.md (arm_andsi3_insn): Add alternatives for 16-bit
16031         encoding.
16032         (iorsi3_insn): Likewise.
16033         (arm_xorsi3): Likewise.
16035 2013-07-01  Sofiane Naci  <sofiane.naci@arm.com>
16037         * arm.md (attribute "wtype"): Delete.  Move attribute values from here
16038         to ...
16039         (attribute "type"): ... here, and prefix with "wmmx_".
16040         (attribute "core_cycles"): Update for attribute changes.
16041         * iwmmxt.md (tbcstv8qi): Update for attribute changes.
16042         (tbcstv4hi): Likewise.
16043         (tbcstv2si): Likewise.
16044         (iwmmxt_iordi3): Likewise.
16045         (iwmmxt_xordi3): Likewise.
16046         (iwmmxt_anddi3): Likewise.
16047         (iwmmxt_nanddi3): Likewise.
16048         (iwmmxt_arm_movdi): Likewise.
16049         (iwmmxt_movsi_insn): Likewise.
16050         (mov<mode>_internal): Likewise.
16051         (and<mode>3_iwmmxt): Likewise.
16052         (ior<mode>3_iwmmxt): Likewise.
16053         (xor<mode>3_iwmmxt): Likewise.
16054         (add<mode>3_iwmmxt): Likewise.
16055         (ssaddv8qi3): Likewise.
16056         (ssaddv4hi3): Likewise.
16057         (ssaddv2si3): Likewise.
16058         (usaddv8qi3): Likewise.
16059         (usaddv4hi3): Likewise.
16060         (usaddv2si3): Likewise.
16061         (sub<mode>3_iwmmxt): Likewise.
16062         (sssubv8qi3): Likewise.
16063         (sssubv4hi3): Likewise.
16064         (sssubv2si3): Likewise.
16065         (ussubv8qi3): Likewise.
16066         (ussubv4hi3): Likewise.
16067         (ussubv2si3): Likewise.
16068         (mulv4hi3_iwmmxt): Likewise.
16069         (smulv4hi3_highpart): Likewise.
16070         (umulv4hi3_highpart): Likewise.
16071         (iwmmxt_wmacs): Likewise.
16072         (iwmmxt_wmacsz): Likewise.
16073         (iwmmxt_wmacu): Likewise.
16074         (iwmmxt_wmacuz): Likewise.
16075         (iwmmxt_clrdi): Likewise.
16076         (iwmmxt_clrv8qi): Likewise.
16077         (iwmmxt_clr4hi): Likewise.
16078         (iwmmxt_clr2si): Likewise.
16079         (iwmmxt_uavgrndv8qi3): Likewise.
16080         (iwmmxt_uavgrndv4hi3): Likewise.
16081         (iwmmxt_uavgv8qi3): Likewise.
16082         (iwmmxt_uavgv4hi3): Likewise.
16083         (iwmmxt_tinsrb): Likewise.
16084         (iwmmxt_tinsrh): Likewise.
16085         (iwmmxt_tinsrw): Likewise.
16086         (iwmmxt_textrmub): Likewise.
16087         (iwmmxt_textrmsb): Likewise.
16088         (iwmmxt_textrmuh): Likewise.
16089         (iwmmxt_textrmsh): Likewise.
16090         (iwmmxt_textrmw): Likewise.
16091         (iwmxxt_wshufh): Likewise.
16092         (eqv8qi3): Likewise.
16093         (eqv4hi3): Likewise.
16094         (eqv2si3): Likewise.
16095         (gtuv8qi3): Likewise.
16096         (gtuv4hi3): Likewise.
16097         (gtuv2si3): Likewise.
16098         (gtv8qi3): Likewise.
16099         (gtv4hi3): Likewise.
16100         (gtv2si3): Likewise.
16101         (smax<mode>3_iwmmxt): Likewise.
16102         (umax<mode>3_iwmmxt): Likewise.
16103         (smin<mode>3_iwmmxt): Likewise.
16104         (umin<mode>3_iwmmxt): Likewise.
16105         (iwmmxt_wpackhss): Likewise.
16106         (iwmmxt_wpackwss): Likewise.
16107         (iwmmxt_wpackdss): Likewise.
16108         (iwmmxt_wpackhus): Likewise.
16109         (iwmmxt_wpackwus): Likewise.
16110         (iwmmxt_wpackdus): Likewise.
16111         (iwmmxt_wunpckihb): Likewise.
16112         (iwmmxt_wunpckihh): Likewise.
16113         (iwmmxt_wunpckihw): Likewise.
16114         (iwmmxt_wunpckilb): Likewise.
16115         (iwmmxt_wunpckilh): Likewise.
16116         (iwmmxt_wunpckilw): Likewise.
16117         (iwmmxt_wunpckehub): Likewise.
16118         (iwmmxt_wunpckehuh): Likewise.
16119         (iwmmxt_wunpckehuw): Likewise.
16120         (iwmmxt_wunpckehsb): Likewise.
16121         (iwmmxt_wunpckehsh): Likewise.
16122         (iwmmxt_wunpckehsw): Likewise.
16123         (iwmmxt_wunpckelub): Likewise.
16124         (iwmmxt_wunpckeluh): Likewise.
16125         (iwmmxt_wunpckeluw): Likewise.
16126         (iwmmxt_wunpckelsb): Likewise.
16127         (iwmmxt_wunpckelsh): Likewise.
16128         (iwmmxt_wunpckelsw): Likewise.
16129         (ror<mode>3): Likewise.
16130         (ashr<mode>3_iwmmxt): Likewise.
16131         (lshr<mode>3_iwmmxt): Likewise.
16132         (ashl<mode>3_iwmmxt): Likewise.
16133         (ror<mode>3_di): Likewise.
16134         (ashr<mode>3_di): Likewise.
16135         (lshr<mode>3_di): Likewise.
16136         (ashl<mode>3_di): Likewise.
16137         (iwmmxt_wmadds): Likewise.
16138         (iwmmxt_wmaddu): Likewise.
16139         (iwmmxt_tmia): Likewise.
16140         (iwmmxt_tmiaph): Likewise.
16141         (iwmmxt_tmiabb): Likewise.
16142         (iwmmxt_tmiatb): Likewise.
16143         (iwmmxt_tmiabt): Likewise.
16144         (iwmmxt_tmiatt): Likewise.
16145         (iwmmxt_tmovmskb): Likewise.
16146         (iwmmxt_tmovmskh): Likewise.
16147         (iwmmxt_tmovmskw): Likewise.
16148         (iwmmxt_waccb): Likewise.
16149         (iwmmxt_wacch): Likewise.
16150         (iwmmxt_waccw): Likewise.
16151         (iwmmxt_waligni): Likewise.
16152         (iwmmxt_walignr): Likewise.
16153         (iwmmxt_walignr0): Likewise.
16154         (iwmmxt_walignr1): Likewise.
16155         (iwmmxt_walignr2): Likewise.
16156         (iwmmxt_walignr3): Likewise.
16157         (iwmmxt_wsadb): Likewise.
16158         (iwmmxt_wsadh): Likewise.
16159         (iwmmxt_wsadbz): Likewise.
16160         (iwmmxt_wsadhz): Likewise.
16161         * iwmmxt2.md (iwmmxt_wabs<mode>3): Update for attribute changes.
16162         (iwmmxt_wabsdiffb): Likewise.
16163         (iwmmxt_wabsdiffh): Likewise.
16164         (iwmmxt_wabsdiffw): Likewise.
16165         (iwmmxt_waddsubhx): Likewise
16166         (iwmmxt_wsubaddhx): Likewise.
16167         (addc<mode>3): Likewise.
16168         (iwmmxt_avg4): Likewise.
16169         (iwmmxt_avg4r): Likewise.
16170         (iwmmxt_wmaddsx): Likewise.
16171         (iwmmxt_wmaddux): Likewise.
16172         (iwmmxt_wmaddsn): Likewise.
16173         (iwmmxt_wmaddun): Likewise.
16174         (iwmmxt_wmulwsm): Likewise.
16175         (iwmmxt_wmulwum): Likewise.
16176         (iwmmxt_wmulsmr): Likewise.
16177         (iwmmxt_wmulumr): Likewise.
16178         (iwmmxt_wmulwsmr): Likewise.
16179         (iwmmxt_wmulwumr): Likewise.
16180         (iwmmxt_wmulwl): Likewise.
16181         (iwmmxt_wqmulm): Likewise.
16182         (iwmmxt_wqmulwm): Likewise.
16183         (iwmmxt_wqmulmr): Likewise.
16184         (iwmmxt_wqmulwmr): Likewise.
16185         (iwmmxt_waddbhusm): Likewise.
16186         (iwmmxt_waddbhusl): Likewise.
16187         (iwmmxt_wqmiabb): Likewise.
16188         (iwmmxt_wqmiabt): Likewise.
16189         (iwmmxt_wqmiatb): Likewise.
16190         (iwmmxt_wqmiatt): Likewise.
16191         (iwmmxt_wqmiabbn): Likewise.
16192         (iwmmxt_wqmiabtn): Likewise.
16193         (iwmmxt_wqmiatbn): Likewise.
16194         (iwmmxt_wqmiattn): Likewise.
16195         (iwmmxt_wmiabb): Likewise.
16196         (iwmmxt_wmiabt): Likewise.
16197         (iwmmxt_wmiatb): Likewise.
16198         (iwmmxt_wmiatt): Likewise.
16199         (iwmmxt_wmiabbn): Likewise.
16200         (iwmmxt_wmiabtn): Likewise.
16201         (iwmmxt_wmiatbn): Likewise.
16202         (iwmmxt_wmiattn): Likewise.
16203         (iwmmxt_wmiawbb): Likewise.
16204         (iwmmxt_wmiawbt): Likewise.
16205         (iwmmxt_wmiawtb): Likewise.
16206         (iwmmxt_wmiawtt): Likewise.
16207         (iwmmxt_wmiawbbn): Likewise.
16208         (iwmmxt_wmiawbtn): Likewise.
16209         (iwmmxt_wmiawtbn): Likewise.
16210         (iwmmxt_wmiawttn): Likewise.
16211         (iwmmxt_wmerge): Likewise.
16212         (iwmmxt_tandc<mode>3): Likewise.
16213         (iwmmxt_torc<mode>3): Likewise.
16214         (iwmmxt_torvsc<mode>3): Likewise.
16215         (iwmmxt_textrc<mode>3): Likewise.
16216         * marvell-f-iwmmxt.md (wmmxt_shift): Update for attribute changes.
16217         (wmmxt_pack): Likewise.
16218         (wmmxt_mult_c1): Likewise.
16219         (wmmxt_mult_c2): Likewise.
16220         (wmmxt_alu_c1): Likewise.
16221         (wmmxt_alu_c2): Likewise.
16222         (wmmxt_alu_c3): Likewise.
16223         (wmmxt_transfer_c1): Likewise.
16224         (wmmxt_transfer_c2): Likewise.
16225         (wmmxt_transfer_c3): Likewise.
16226         (marvell_f_iwmmxt_wstr): Likewise.
16227         (marvell_f_iwmmxt_wldr): Likewise.
16229 2013-06-29  Yufeng Zhang  <yufeng.zhang@arm.com>
16231         * config/aarch64/aarch64.c: Remove junk from the beginning of the file.
16233 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16235         Revert:
16236         2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16237         * lra-constraints.c (need_for_split_p): Check call used hard regs
16238         living through calls.
16240         * lra-constraints.c (inherit_in_ebb): Reset live_hard_regs for
16241         call used regs for call insn.
16243 2013-06-28  Jakub Jelinek  <jakub@redhat.com>
16245         PR target/57736
16246         * config/i386/i386.c (ix86_expand_builtin): If target == NULL and
16247         mode is VOIDmode, don't create a VOIDmode pseudo to copy result into.
16249 2013-06-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
16251         * builtins.def: Fixed the function type of CILKPLUS_BUILTIN.
16253 2013-06-28  Vladimir Makarov  <vmakarov@redhat.com>
16255         * lra-constraints.c (need_for_split_p): Check call used hard regs
16256         living through calls.
16258 2013-06-28  Michael Meissner  <meissner@linux.vnet.ibm.com>
16260         PR target/57744
16261         * config/rs6000/rs6000.h (MODES_TIEABLE_P): Do not allow PTImode
16262         to tie with any other modes.  Eliminate Altivec vector mode tests,
16263         since these are a subset of ALTIVEC or VSX vector modes.  Simplify
16264         code, to return 0 if testing MODE2 for a condition, if we've
16265         already tested MODE1 for the same condition.
16267 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16269         * config/aarch64/aarch64.c (aarch64_cannot_force_const_mem): Adjust
16270         layout.
16272 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16274         * config/aarch64/aarch64-protos.h (aarch64_symbol_type):
16275         Update comment w.r.t SYMBOL_TINY_ABSOLUTE.
16277 2013-06-28  Marcus Shawcroft  <marcus.shawcroft@arm.com>
16279         * config/aarch64/aarch64-protos.h (aarch64_classify_symbol_expression):
16280         Define.
16281         (aarch64_symbolic_constant_p): Remove.
16282         * config/aarch64/aarch64.c (aarch64_classify_symbol_expression): Remove
16283         static.  Fix line length and white space.
16284         (aarch64_symbolic_constant_p): Remove.
16285         * config/aarch64/predicates.md (aarch64_valid_symref):
16286         Use aarch64_classify_symbol_expression.
16288 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16290         * config/arm/constraints.md (Ts): New constraint.
16291         * config/arm/arm.md (arm_movqi_insn): Add alternatives for
16292         16-bit encodings.
16293         (compare_scc): Use "Ts" constraint for operand 0.
16294         (ior_scc_scc): Likewise.
16295         (and_scc_scc): Likewise.
16296         (and_scc_scc_nodom): Likewise.
16297         (ior_scc_scc_cmp): Likewise for operand 7.
16298         (and_scc_scc_cmp): Likewise.
16299         * config/arm/thumb2.md (thumb2_movsi_insn):
16300         Add alternatives for 16-bit encodings.
16301         (thumb2_movhi_insn): Likewise.
16302         (thumb2_movsicc_insn): Likewise.
16303         (thumb2_and_scc): Take 'and' outside cond_exec.  Use "Ts" constraint.
16304         (thumb2_negscc): Use "Ts" constraint.
16305         Move mvn instruction outside cond_exec block.
16306         * config/arm/vfp.md (thumb2_movsi_vfp): Add alternatives
16307         for 16-bit encodings.
16309 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16311         * config/arm/arm.md (arm_mulsi3_v6): Add alternative for 16-bit
16312         encoding.
16313         (mulsi3addsi_v6): Disable predicable variant for arm_restrict_it.
16314         (mulsi3subsi): Likewise.
16315         (mulsidi3adddi): Likewise.
16316         (mulsidi3_v6): Likewise.
16317         (umulsidi3_v6): Likewise.
16318         (umulsidi3adddi_v6): Likewise.
16319         (smulsi3_highpart_v6): Likewise.
16320         (umulsi3_highpart_v6): Likewise.
16321         (mulhisi3tb): Likewise.
16322         (mulhisi3bt): Likewise.
16323         (mulhisi3tt): Likewise.
16324         (maddhisi4): Likewise.
16325         (maddhisi4tb): Likewise.
16326         (maddhisi4tt): Likewise.
16327         (maddhidi4): Likewise.
16328         (maddhidi4tb): Likewise.
16329         (maddhidi4tt): Likewise.
16330         (zeroextractsi_compare0_scratch): Likewise.
16331         (insv_zero): Likewise.
16332         (insv_t2): Likewise.
16333         (anddi_notzesidi_di): Likewise.
16334         (anddi_notsesidi_di): Likewise.
16335         (andsi_notsi_si): Likewise.
16336         (iordi_zesidi_di): Likewise.
16337         (xordi_zesidi_di): Likewise.
16338         (andsi_iorsi3_notsi): Likewise.
16339         (smax_0): Likewise.
16340         (smax_m1): Likewise.
16341         (smin_0): Likewise.
16342         (not_shiftsi): Likewise.
16343         (unaligned_loadsi): Likewise.
16344         (unaligned_loadhis): Likewise.
16345         (unaligned_loadhiu): Likewise.
16346         (unaligned_storesi): Likewise.
16347         (unaligned_storehi): Likewise.
16348         (extv_reg): Likewise.
16349         (extzv_t2): Likewise.
16350         (divsi3): Likewise.
16351         (udivsi3): Likewise.
16352         (arm_zero_extendhisi2addsi): Likewise.
16353         (arm_zero_extendqisi2addsi): Likewise.
16354         (compareqi_eq0): Likewise.
16355         (arm_extendhisi2_v6): Likewise.
16356         (arm_extendqisi2addsi): Likewise.
16357         (arm_movt): Likewise.
16358         (thumb2_ldrd): Likewise.
16359         (thumb2_ldrd_base): Likewise.
16360         (thumb2_ldrd_base_neg): Likewise.
16361         (thumb2_strd): Likewise.
16362         (thumb2_strd_base): Likewise.
16363         (thumb2_strd_base_neg): Likewise.
16364         (arm_negsi2): Add alternative for 16-bit encoding.
16365         (arm_one_cmplsi2): Likewise.
16367 2013-06-28  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
16369         * config/arm/predicates.md (arm_cond_move_operator): New predicate.
16370         * config/arm/arm.md (movsfcc): Use arm_cond_move_operator predicate.
16371         (movdfcc): Likewise.
16372         * config/arm/vfp.md (*thumb2_movsf_vfp):
16373         Disable predication for arm_restrict_it.
16374         (*thumb2_movsfcc_vfp): Disable for arm_restrict_it.
16375         (*thumb2_movdfcc_vfp): Likewise.
16376         (*abssf2_vfp, *absdf2_vfp, *negsf2_vfp, *negdf2_vfp,*addsf3_vfp,
16377         *adddf3_vfp, *subsf3_vfp, *subdf3_vfpc, *divsf3_vfp,*divdf3_vfp,
16378         *mulsf3_vfp, *muldf3_vfp, *mulsf3negsf_vfp, *muldf3negdf_vfp,
16379         *mulsf3addsf_vfp, *muldf3adddf_vfp, *mulsf3subsf_vfp,
16380         *muldf3subdf_vfp, *mulsf3negsfaddsf_vfp, *fmuldf3negdfadddf_vfp,
16381         *mulsf3negsfsubsf_vfp, *muldf3negdfsubdf_vfp, *fma<SDF:mode>4,
16382         *fmsub<SDF:mode>4, *fnmsub<SDF:mode>4, *fnmadd<SDF:mode>4,
16383         *extendsfdf2_vfp, *truncdfsf2_vfp, *extendhfsf2, *truncsfhf2,
16384         *truncsisf2_vfp, *truncsidf2_vfp, fixuns_truncsfsi2, fixuns_truncdfsi2,
16385         *floatsisf2_vfp, *floatsidf2_vfp, floatunssisf2, floatunssidf2,
16386         *sqrtsf2_vfp, *sqrtdf2_vfp, *cmpsf_vfp, *cmpsf_trap_vfp, *cmpdf_vfp,
16387         *cmpdf_trap_vfp, <vrint_pattern><SDF:mode>2):
16388         Disable predication for arm_restrict_it.
16390 2013-06-28  Kirill Yukhin  <kirill.yukhin@intel.com>
16392         * config/i386/bmiintrin.h (_bextr_u32): New.
16393         (_bextr_u64): Ditto.
16395 2013-06-27  Richard Sandiford  <rdsandiford@googlemail.com>
16397         * config.gcc (mips*-mti-elf*, mips*-sde-elf*, mips64r5900-*-elf*)
16398         (mips64r5900el-*-elf*): Include mips/n32-elf.h.
16399         * config/mips/sde.h (LOCAL_LABEL_PREFIX, NO_DOLLAR_IN_LABEL)
16400         (LONG_DOUBLE_TYPE_SIZE, LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Move to...
16401         * config/mips/n32-elf.h: ...this new file.
16403 2013-06-27  Marc Glisse  <marc.glisse@inria.fr>
16405         PR target/57224
16406         * config/i386/i386.c (enum ix86_builtins, bdesc_args): Remove
16407         IX86_BUILTIN_CMPNGTSS and IX86_BUILTIN_CMPNGESS.
16409 2013-06-27  Catherine Moore  <clm@codesourcery.com>
16411         * config/mips/mips-tables.opt: Regenerate.
16412         * config/mips/mips-cpus.def: Add m14ke and m14kec.
16413         * config/mips/mips.h (BASE_DRIVER_SELF_SPECS): m14ke* implies -mdspr2.
16414         * doc/invoke.texi: Add -m14kc.
16416 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16418         PR target/57623
16419         * config/i386/i386.md (bmi_bextr_<mode>): Swap predicates and
16420         constraints of operand 1 and 2.
16422         PR target/57623
16423         * config/i386/i386.md (bmi2_bzhi_<mode>3): Swap AND arguments
16424         to match RTL canonicalization.  Swap predicates and
16425         constraints of operand 1 and 2.
16427 2013-06-27  Vladimir Makarov  <vmakarov@redhat.com>
16429         * lra-constraints.c (inherit_in_ebb): Process static hard regs too.
16430         Process OP_INOUT regs for splitting too.
16432 2013-06-27  Jakub Jelinek  <jakub@redhat.com>
16434         * tree-vect-stmts.c (vectorizable_store): Move ptr_incr var
16435         decl before the loop, initialize to NULL.
16436         (vectorizable_load): Initialize ptr_incr to NULL.
16438 2013-06-27  Martin Jambor  <mjambor@suse.cz>
16440         PR lto/57208
16441         * ipa-ref.h (ipa_maybe_record_reference): Declare.
16442         * ipa-ref.c (ipa_maybe_record_reference): New function.
16443         * cgraphclones.c (cgraph_create_virtual_clone): Use it.
16444         * ipa-cp.c (create_specialized_node): Record potential references from
16445         aggvals.
16446         * Makefile.in (ipa-ref.o): Add IPA_REF_H to dependencies.
16448 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16450         * config/aarch64/aarch64.c (aarch64_force_temporary): Add an extra
16451         parameter 'mode' of type 'enum machine_mode mode'; change to pass
16452         'mode' to force_reg.
16453         (aarch64_add_offset): Update calls to aarch64_force_temporary.
16454         (aarch64_expand_mov_immediate): Likewise.
16456 2013-06-27  Yufeng Zhang  <yufeng.zhang@arm.com>
16458         * config/aarch64/aarch64.c (aarch64_add_offset): Change to pass
16459         'mode' to aarch64_plus_immediate and gen_rtx_PLUS.
16461 2013-06-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16463         * config/s390/s390.c: Rename UNSPEC_CCU_TO_INT to
16464         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16465         (struct machine_function): Add tbegin_p.
16466         (s390_canonicalize_comparison): Fold CC mode compares to
16467         conditional jump if possible.
16468         (s390_emit_jump): Return the emitted jump.
16469         (s390_branch_condition_mask, s390_branch_condition_mnemonic):
16470         Handle CCRAWmode compares.
16471         (s390_option_override): Default to -mhtm if available.
16472         (s390_reg_clobbered_rtx): Handle floating point regs as well.
16473         (s390_regs_ever_clobbered): Use s390_regs_ever_clobbered also for
16474         FPRs instead of df_regs_ever_live_p.
16475         (s390_optimize_nonescaping_tx): New function.
16476         (s390_init_frame_layout): Extend clobbered_regs array to cover
16477         FPRs as well.
16478         (s390_emit_prologue): Call s390_optimize_nonescaping_tx.
16479         (s390_expand_tbegin): New function.
16480         (enum s390_builtin): New enum definition.
16481         (code_for_builtin): New array definition.
16482         (s390_init_builtins): New function.
16483         (s390_expand_builtin): New function.
16484         (TARGET_INIT_BUILTINS): Define.
16485         (TARGET_EXPAND_BUILTIN): Define.
16486         * common/config/s390/s390-common.c (processor_flags_table): Add PF_TX.
16487         * config/s390/predicates.md (s390_comparison): Handle CCRAWmode.
16488         (s390_alc_comparison): Likewise.
16489         * config/s390/s390-modes.def: Add CCRAWmode.
16490         * config/s390/s390.h (processor_flags): Add PF_TX.
16491         (TARGET_CPU_HTM): Define macro.
16492         (TARGET_HTM): Define macro.
16493         (TARGET_CPU_CPP_BUILTINS): Define __HTM__ for htm.
16494         * config/s390/s390.md: Rename UNSPEC_CCU_TO_INT to
16495         UNSPEC_STRCMPCC_TO_INT and UNSPEC_CCZ_TO_INT to UNSPEC_CC_TO_INT.
16496         (UNSPECV_TBEGIN, UNSPECV_TBEGINC, UNSPECV_TEND, UNSPECV_TABORT)
16497         (UNSPECV_ETND, UNSPECV_NTSTG, UNSPECV_PPA): New unspecv enum values.
16498         (TBEGIN_MASK, TBEGINC_MASK): New constants.
16499         ("*cc_to_int"): Move up.
16500         ("*mov<mode>cc", "*cjump_64", "*cjump_31"): Accept integer
16501         constants other than 0.
16502         ("*ccraw_to_int"): New insn and splitter definition.
16503         ("tbegin", "tbegin_nofloat", "tbegin_retry")
16504         ("tbegin_retry_nofloat", "tbeginc", "tend", "tabort")
16505         ("tx_assist"): New expander.
16506         ("tbegin_1", "tbegin_nofloat_1", "*tbeginc_1", "*tend_1")
16507         ("*tabort_1", "etnd", "ntstg", "*ppa"): New insn definition.
16508         * config/s390/s390.opt: Add -mhtm option.
16509         * config/s390/s390-protos.h (s390_emit_jump): Add return type.
16510         * config/s390/htmxlintrin.h: New file.
16511         * config/s390/htmintrin.h: New file.
16512         * config/s390/s390intrin.h: New file.
16513         * doc/extend.texi: Document htm builtins.
16514         * config.gcc: Add the new header files to extra_headers.
16516 2013-06-26  Thomas Schwinge  <thomas@codesourcery.com>
16518         * config/i386/gnu.h [TARGET_LIBC_PROVIDES_SSP]
16519         (TARGET_CAN_SPLIT_STACK, TARGET_THREAD_SPLIT_STACK_OFFSET): Undefine.
16521 2013-06-26  Michael Meissner  <meissner@linux.vnet.ibm.com>
16522             Pat Haugen  <pthaugen@us.ibm.com>
16523             Peter Bergner  <bergner@vnet.ibm.com>
16525         * config/rs6000/power8.md: New.
16526         * config/rs6000/rs6000-cpus.def (RS6000_CPU table): Adjust processor
16527         setting for power8 entry.
16528         * config/rs6000/t-rs6000 (MD_INCLUDES): Add power8.md.
16529         * config/rs6000/rs6000.c (is_microcoded_insn, is_cracked_insn): Adjust
16530         test for Power4/Power5 only.
16531         (insn_must_be_first_in_group, insn_must_be_last_in_group): Add Power8
16532         support.
16533         (force_new_group): Adjust comment.
16534         * config/rs6000/rs6000.md: Include power8.md.
16536 2013-06-26  Greta Yorsh  <Greta.Yorsh@arm.com>
16538         * config/arm/arm.h (MAX_CONDITIONAL_EXECUTE): Define macro.
16539         * config/arm/arm-protos.h (arm_max_conditional_execute): New
16540         declaration.
16541         (tune_params): Update comment.
16542         * config/arm/arm.c (arm_cortex_a15_tune): Set max_cond_insns to 2.
16543         (arm_max_conditional_execute): New function.
16544         (thumb2_final_prescan_insn): Use max_insn_skipped and
16545         MAX_INSN_PER_IT_BLOCK to compute maximum instructions in a block.
16547 2013-06-25  Jakub Jelinek  <jakub@redhat.com>
16549         PR tree-optimization/57705
16550         * tree-vect-loop.c (vect_is_simple_iv_evolution): Allow
16551         SSA_NAME step, provided that it is not defined inside the loop.
16552         (vect_analyze_scalar_cycles_1): Disallow SSA_NAME step in nested loop.
16553         (get_initial_def_for_induction): Handle SSA_NAME IV step.
16555 2013-06-25  Martin Jambor  <mjambor@suse.cz>
16557         PR middle-end/57670
16558         * cgraph.h (cgraph_indirect_call_info): New flag member_ptr.
16559         * ipa-prop.c (ipa_print_node_jump_functions): Mark member pointer
16560         calls in the dump.
16561         (ipa_note_param_call): Initialize member_ptr flag.
16562         (ipa_analyze_indirect_call_uses): Set member_ptr flag.
16563         (ipa_make_edge_direct_to_target): Bail out if member_ptr is set.
16564         (ipa_write_indirect_edge_info): Stream member_ptr flag.
16565         (ipa_read_indirect_edge_info): Likewise.
16567 2013-06-25  Richard Biener  <rguenther@suse.de>
16569         PR middle-end/56977
16570         * passes.c (init_optimization_passes): Move pass_fold_builtins
16571         and pass_dce earlier with -Og.
16573 2013-06-25  Eric Botcazou  <ebotcazou@adacore.com>
16575         * expr.c (expand_expr_real_1) <ARRAY_REF>: Fix formatting glitches.
16576         <BIT_FIELD_REF>: Remove trailing TAB.
16577         * varasm.c (output_constructor_bitfield): Fix formatting glitch and
16578         remove blank line.
16580 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16582         PR tree-optimization/57358
16583         * ipa-prop.c (ipa_func_spec_opts_forbid_analysis_p): New function.
16584         (ipa_compute_jump_functions_for_edge): Bail out if it returns true.
16585         (ipa_analyze_params_uses): Generate pessimistic info when true.
16587 2013-06-24  Martin Jambor  <mjambor@suse.cz>
16589         PR tree-optimization/57539
16590         * cgraphclones.c (cgraph_clone_node): Add parameter new_inlined_to, set
16591         global.inlined_to of the new node to it.  All callers changed.
16592         * ipa-inline-transform.c (clone_inlined_nodes): New variable
16593         inlining_into, pass it to cgraph_clone_node.
16594         * ipa-prop.c (ipa_propagate_indirect_call_infos): Do not call
16595         ipa_free_edge_args_substructures.
16596         (ipa_edge_duplication_hook): Only add edges from inlined nodes to
16597         rdesc linked list.  Do not assert rdesc edges have inlined caller.
16598         Assert we have found an rdesc in the rdesc list.
16600 2013-06-24  Richard Biener  <rguenther@suse.de>
16602         * pointer-set.h (struct pointer_set_t): Move here from pointer-set.c.
16603         (pointer_set_lookup): Declare.
16604         (class pointer_map): New template class implementing a
16605         generic pointer to T map.
16606         (pointer_map<T>::pointer_map, pointer_map<T>::~pointer_map,
16607         pointer_map<T>::contains, pointer_map<T>::insert,
16608         pointer_map<T>::traverse): New functions.
16609         * pointer-set.c (struct pointer_set_t): Moved to pointer-set.h.
16610         (pointer_set_lookup): New function.
16611         (pointer_set_contains): Use pointer_set_lookup.
16612         (pointer_set_insert): Likewise.
16613         (insert_aux): Remove.
16614         (struct pointer_map_t): Embed a pointer_set_t.
16615         (pointer_map_create): Adjust.
16616         (pointer_map_destroy): Likewise.
16617         (pointer_map_contains): Likewise.
16618         (pointer_map_insert): Likewise.
16619         (pointer_map_traverse): Likewise.
16620         * tree-streamer.h (struct streamer_tree_cache_d): Use a
16621         pointer_map<unsigned> instead of a pointer_map_t.
16622         * tree-streamer.c (streamer_tree_cache_insert_1): Adjust.
16623         (streamer_tree_cache_lookup): Likewise.
16624         (streamer_tree_cache_create): Likewise.
16625         (streamer_tree_cache_delete): Likewise.
16626         * lto-streamer.h (struct lto_tree_ref_encoder): Use a
16627         pointer_map<unsigned> instead of a pointer_map_t.
16628         (lto_init_tree_ref_encoder): Adjust.
16629         (lto_destroy_tree_ref_encoder): Likewise.
16630         * lto-section-out.c (lto_output_decl_index): Likewise.
16631         (lto_record_function_out_decl_state): Likewise.
16632         * dominance.c (iterate_fix_dominators): Use pointer_map<int>.
16634 2013-06-24  Richard Biener  <rguenther@suse.de>
16636         PR tree-optimization/57488
16637         * tree-ssa-pre.c (insert): Clear NEW sets before each iteration.
16639 2013-06-24  Alan Modra  <amodra@gmail.com>
16641         * config/rs6000/rs6000.c (vspltis_constant): Correct for little-endian.
16642         (gen_easy_altivec_constant): Likewise.
16643         * config/rs6000/predicates.md (easy_vector_constant_add_self,
16644         easy_vector_constant_msb): Likewise.
16646 2013-06-23  Jakub Jelinek  <jakub@redhat.com>
16648         PR target/57688
16649         * common/config/i386/i386-common.c (ix86_handle_option): For OPT_mlzcnt
16650         add missing return true.
16652 2013-06-23  Oleg Endo  <olegendo@gcc.gnu.org>
16654         PR target/52483
16655         * config/sh/predicates.md (general_extend_operand): Invoke
16656         general_movsrc_operand for memory operands.
16657         (general_movsrc_operand): Allow reg+reg addressing, do not use
16658         general_operand for memory operands.
16660 2013-06-23  Sriraman Tallam  <tmsriram@google.com>
16662         * config/i386/i386.c (ix86_pragma_target_parse):  Restore target
16663         when current target options does not apply.
16664         * config/i386/i386-protos.h (ix86_reset_previous_fndecl): New function.
16665         * config/i386/i386.c (ix86_reset_previous_fndecl): Ditto.
16666         * config/i386/bmiintrin.h: Pass appropriate target
16667         attributes to header.
16668         * config/i386/mmintrin.h: Ditto.
16669         * config/i386/nmmintrin.h: Ditto.
16670         * config/i386/avx2intrin.h: Ditto.
16671         * config/i386/fxsrintrin.h: Ditto.
16672         * config/i386/tbmintrin.h: Ditto.
16673         * config/i386/xsaveintrin.h: Ditto.
16674         * config/i386/f16cintrin.h: Ditto.
16675         * config/i386/xtestintrin.h: Ditto.
16676         * config/i386/xsaveoptintrin.h: Ditto.
16677         * config/i386/bmi2intrin.h: Ditto.
16678         * config/i386/lzcntintrin.h: Ditto.
16679         * config/i386/smmintrin.h: Ditto.
16680         * config/i386/wmmintrin.h: Ditto.
16681         * config/i386/x86intrin.h: Remove all header include guards.
16682         * config/i386/prfchwintrin.h: Ditto.
16683         * config/i386/pmmintrin.h: Ditto.
16684         * config/i386/tmmintrin.h: Ditto.
16685         * config/i386/xmmintrin.h: Ditto.
16686         * config/i386/popcntintrin.h: Ditto.
16687         * config/i386/rdseedintrin.h: Ditto.
16688         * config/i386/ammintrin.h: Ditto.
16689         * config/i386/emmintrin.h: Ditto.
16690         * config/i386/immintrin.h: Remove all header include guards.
16691         * config/i386/fma4intrin.h: Ditto.
16692         * config/i386/lwpintrin.h: Ditto.
16693         * config/i386/xopintrin.h: Ditto.
16694         * config/i386/ia32intrin.h: Ditto.
16695         * config/i386/avxintrin.h: Ditto.
16696         * config/i386/rtmintrin.h: Ditto.
16697         * config/i386/fmaintrin.h: Ditto.
16698         * config/i386/mm3dnow.h: Ditto.
16700 2013-06-22  Sriraman Tallam  <tmsriram@google.com>
16702         * common/config/i386/i386-common.c: Handle LZCNT.
16704 2013-06-22  Andi Kleen  <ak@linux.intel.com>
16706         * doc/extend.texi: Use __atomic_store_n instead of
16707         __atomic_store in HLE example.
16709 2013-06-22  Oleg Endo  <olegendo@gcc.gnu.org>
16711         * config/sh/sh.c: Remove <cstdlib> workaround.
16713 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16715         * doc/extend.texi: Dont use __atomic_clear in HLE example.  Fix typo.
16717 2013-06-21  Andi Kleen  <ak@linux.intel.com>
16719         * doc/extend.texi: Document that __atomic_clear and
16720           __atomic_test_and_set should only be used with bool.
16722 2013-06-20  Jan Hubicka  <jh@suse.cz>
16724         * gimple-fold.c (gimple_extract_devirt_binfo_from_cst): Use
16725         types_same_for_odr.
16726         * tree.c (decls_same_for_odr): New function.
16727         (same_for_edr): New function.
16728         (types_same_for_odr): New function.
16729         (get_binfo_at_offset): Use it.
16730         * tree.h (types_same_for_odr): Declare.
16732 2013-06-20  Oleg Endo  <olegendo@gcc.gnu.org>
16733             Jason Merrill  <jason@redhat.com>
16735         * system.h: Include <cstdlib> as well as <stdlib.h>.
16737 2013-06-20  Uros Bizjak  <ubizjak@gmail.com>
16739         PR target/57655
16740         * config/i386/i386.c (construct_container): Report error if
16741         long double is used with disabled x87 float returns.
16743 2013-06-20  Jan Hubicka  <jh@suse.cz>
16745         * lto-cgraph.c (input_symtab): Do not set cgraph state.
16747 2013-06-20  Joern Rennecke  <joern.rennecke@embecosm.com>
16749         PR rtl-optimization/57425
16750         PR rtl-optimization/57569
16751         * alias.c (write_dependence_p): Remove parameters mem_mode and
16752         canon_mem_addr.  Add parameters x_mode, x_addr and x_canonicalized.
16753         Changed all callers.
16754         (canon_anti_dependence): Get comments and semantics in sync.
16755         Add parameter mem_canonicalized.  Changed all callers.
16756         * rtl.h (canon_anti_dependence): Update prototype.
16758 2013-06-20  Richard Biener  <rguenther@suse.de>
16760         * data-streamer-in.c (streamer_read_uhwi): Optimize single
16761         byte case, inline streamer_read_uchar and defer section
16762         overrun check.
16764 2013-06-20  Richard Biener  <rguenther@suse.de>
16766         PR tree-optimization/57584
16767         * tree-ssa-loop-niter.c (expand_simple_operations): Avoid including
16768         SSA names into the expanded expression that take part in
16769         abnormal coalescing.
16771 2013-06-19  Sharad Singhai  <singhai@google.com>
16773         * gcov.c (print_usage): Handle new option.
16774         (process_args): Ditto.
16775         (get_gcov_intermediate_filename): New function.
16776         (output_intermediate_file): New function.
16777         (output_gcov_file): New function
16778         (generate_results): Handle new option.
16779         (release_function): Relase demangled name.
16780         (read_graph_file): Handle demangled name.
16781         (output_lines): Ditto.
16782         * doc/gcov.texi: Document gcov intermediate format.
16784 2013-06-19  Vladimir Makarov  <vmakarov@redhat.com>
16786         PR bootstrap/57604
16787         * lra.c (emit_add3_insn, emit_add2_insn): New functions.
16788         (lra_emit_add): Use the functions.  Add comment about Y as an
16789         address segment.
16791 2013-06-19  David Edelsohn  <dje.gcc@gmail.com>
16793         PR driver/57652
16794         * collect2.c (collect_atexit): New.
16795         (collect_exit): Delete.
16796         (main): Register collect_atexit with atexit.
16797         (collect_wait): Change collect_exit to exit.
16798         (do_wait): Same.
16799         * collect2.h (collect_exit): Delete.
16800         * tlink.c (do_tlink): Rename exit to ret. Change collect_exit to exit.
16802 2013-06-19  Wei Mi  <wmi@google.com>
16804         PR rtl-optimization/57518
16805         * ira.c (set_paradoxical_subreg): Set pdx_subregs[regno]
16806         if regno is used in paradoxical subreg.
16807         (update_equiv_regs): Check pdx_subregs[regno] before
16808         set a reg to be equivalent with a mem.
16810 2013-06-19  Matthias Klose  <doko@ubuntu.com>
16812         PR driver/57651
16813         * file-find.h (find_a_file): Add a mode parameter.
16814         * file-find.c (find_a_file): Likewise.
16815         * gcc-ar.c (main): Call find_a_file with R_OK for the plugin,
16816         with X_OK for the executables.
16817         * collect2.c (main): Call find_a_file with X_OK.
16819 2013-06-19  Steve Ellcey  <sellcey@mips.com>
16821         PR target/56942
16822         * config/mips/mips.md (casesi_internal_mips16_<mode>):
16823         Use NEXT_INSN instead of next_real_insn.
16825 2013-06-19  Jan Hubicka  <jh@suse.cz>
16827         * cgraph.h (const_value_known_p): Replace by ...
16828         (ctor_for_folding): .. this one.
16829         * cgraphunit.c (process_function_and_variable_attributes): Use it.
16830         * lto-cgraph.c (compute_ltrans_boundary): Use ctor_for_folding.
16831         * expr.c (expand_expr_real_1): Likewise.
16832         (string_constant): Likewise.
16833         * tree-ssa-loop-ivcanon.c (constant_after_peeling): Likewise.
16834         * ipa.c (process_references): Likewise.
16835         (symtab_remove_unreachable_nodes): Likewise.
16836         * ipa-inline-analysis.c (param_change_prob): Likewise.
16837         * gimple-fold.c (canonicalize_constructor_val): Likewise.
16838         (get_base_constructor): Likwise.
16839         * varpool.c (varpool_remove_node): Likewise.
16840         (varpool_remove_initializer): LIkewise.
16841         (dump_varpool_node): LIkwise.
16842         (const_value_known_p): Rewrite to ...
16843         (ctor_for_folding): ... this one.
16845 2013-06-19  Jakub Jelinek  <jakub@redhat.com>
16847         PR driver/57651
16848         * gcc-ar.c (main): If not CROSS_DIRECTORY_STRUCTURE, look for
16849         PERSONALITY in $PATH derived prefixes.
16851 2013-06-19  Jeff Law  <law@redhat.com>
16853         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): Fix typo
16854         in comment.
16856         * tree-ssa-forwprop.c (simplify_bitwise_binary_boolean): New function.
16857         (simplify_bitwise_binary): Use it to simpify certain binary ops on
16858         booleans.
16860 2013-06-19  Sofiane Naci  <sofiane.naci@arm.com>
16862         * config/arm/vfp.md: Move VFP instruction classification documentation
16863         to ...
16864         * config/arm/arm.md: ... here.  Update instruction classification
16865         documentation.
16867 2013-06-19  Richard Earnshaw  <rearnsha@arm.com>
16869         arm.md (split for eq(reg, 0)): Add variants for ARMv5 and Thumb2.
16870         (peepholes for eq(reg, not-0)): Ensure condition register is dead after
16871         pattern.  Use more efficient sequences on ARMv5 and Thumb2.
16873 2013-06-19  Steven Bosscher  <steven@gcc.gnu.org>
16875         PR target/57609
16876         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16877         with NEXT_INSN.  Use tablejump_p to check for jump table data
16878         insns.
16880 2013-06-19  Paolo Carlini  <paolo.carlini@oracle.com>
16882         PR c++/56544
16883         * doc/cpp.texi [Standard Predefined Macros, __cplusplus]: Document
16884         that now in C++ the value is correct per the C++ standards.
16886 2013-06-19  Richard Biener  <rguenther@suse.de>
16888         * expr.c (expand_expr_real_1): Use SCOPE_FILE_SCOPE_P to check
16889         for global context.
16891 2013-06-19  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16893         Revert:
16894         2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
16896         PR target/57609
16897         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
16898         with next_active_insn.
16900 2013-06-18  Sriraman Tallam  <tmsriram@google.com>
16902         * ipa-inline.c (inline_always_inline_functions): Pretend always_inline
16903         functions are inlined during failures to flag an error.
16904         * tree-inline.c (expand_call_inline): Allow the error to be flagged
16905         in early inline pass.
16907 2013-06-18  H.J. Lu  <hongjiu.lu@intel.com>
16909         * config/i386/i386.c (initial_ix86_tune_features): Fix a typo
16910         in comments.
16912 2013-06-18  Julian Brown  <julian@codesourcery.com>
16914         * config/arm/arm.c (neon_vector_mem_operand): Add strict argument.
16915         Permit virtual register pre-reload if !strict.
16916         (coproc_secondary_reload_class): Adjust for neon_vector_mem_operand
16917         change.
16918         * config/arm/arm-protos.h (neon_vector_mem_operand): Adjust
16919         prototype.
16920         * config/arm/neon.md (movmisalign<mode>): Use
16921         neon_perm_struct_or_reg_operand instead of
16922         neon_struct_or_register_operand.
16923         (*movmisalign<mode>_neon_load, *movmisalign<mode>_neon_store): Use
16924         neon_permissive_struct_operand instead of neon_struct_operand.
16925         * config/arm/constraints.md (Un, Um, Us): Adjust calls to
16926         neon_vector_mem_operand.
16927         * config/arm/predicates.md (neon_struct_operand): Adjust call to
16928         neon_vector_mem_operand.
16929         (neon_permissive_struct_operand): New.
16930         (neon_struct_or_register_operand): Rename to...
16931         (neon_perm_struct_or_reg_operand): This. Adjust call to
16932         neon_vector_mem_operand.
16934 2013-06-18  Richard Biener  <rguenther@suse.de>
16936         * Makefile.in (LTO_STREAMER_H): Add pointer-set.h dependency.
16937         * lto-streamer.h: Include pointer-set.h.
16938         (struct lto_decl_slot): Remove.
16939         (struct lto_tree_ref_encoder): Make tree_hash_table a pointer-map.
16940         Remove next_index entry.
16941         (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16942         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16943         (lto_init_tree_ref_encoder): Adjust.
16944         (lto_destroy_tree_ref_encoder): Likewise.
16945         * lto-section-out.c (lto_hash_decl_slot_node, lto_eq_decl_slot_node,
16946         lto_hash_type_slot_node, lto_eq_type_slot_node): Remove.
16947         (lto_output_decl_index): Adjust.
16948         (lto_new_out_decl_state): Likewise.
16949         (lto_record_function_out_decl_state): Likewise.
16950         * lto-streamer-out.c (copy_function): Likewise.
16952 2013-06-18  Richard Biener  <rguenther@suse.de>
16954         * Makefile.in (cgraphunit.o): Add $(CFGLOOP_H) dependency.
16955         * cgraphunit.c: Include cfgloop.h.
16956         (init_lowered_empty_function): Initialize the loop tree.
16957         (assemble_thunk): Insert new BBs into loops.
16959 2013-06-18  Richard Biener  <rguenther@suse.de>
16961         * tree-streamer.h (streamer_tree_cache_create): Adjust prototype.
16962         * tree-streamer.c (streamer_tree_cache_create): Make maintaining
16963         the map from cache entry to cache index optional.
16964         (streamer_tree_cache_replace_tree): Adjust accordingly.
16965         (streamer_tree_cache_append): Likewise.
16966         (streamer_tree_cache_delete): Likewise.
16967         * lto-streamer-in.c (lto_data_in_create): Do not maintain the
16968         streamer cache map from cache entry to cache index.
16969         * lto-streamer-out.c (create_output_block): Adjust.
16971 2013-06-18  Sofiane Naci  <sofiane.naci@arm.com>
16973         * config/arm/arm.md (attribute "insn"): Move multiplication and
16974         division attributes to...
16975         (attribute "type"): ... here.  Remove mult.
16976         (attribute "mul32"): New attribute.
16977         (attribute "mul64"): Add umaal.
16978         (*arm_mulsi3): Update attributes.
16979         (*arm_mulsi3_v6): Likewise.
16980         (*thumb_mulsi3): Likewise.
16981         (*thumb_mulsi3_v6): Likewise.
16982         (*mulsi3_compare0): Likewise.
16983         (*mulsi3_compare0_v6): Likewise.
16984         (*mulsi_compare0_scratch): Likewise.
16985         (*mulsi_compare0_scratch_v6): Likewise.
16986         (*mulsi3addsi): Likewise.
16987         (*mulsi3addsi_v6): Likewise.
16988         (*mulsi3addsi_compare0): Likewise.
16989         (*mulsi3addsi_compare0_v6): Likewise.
16990         (*mulsi3addsi_compare0_scratch): Likewise.
16991         (*mulsi3addsi_compare0_scratch_v6): Likewise.
16992         (*mulsi3subsi): Likewise.
16993         (*mulsidi3adddi): Likewise.
16994         (*mulsi3addsi_v6): Likewise.
16995         (*mulsidi3adddi_v6): Likewise.
16996         (*mulsidi3_nov6): Likewise.
16997         (*mulsidi3_v6): Likewise.
16998         (*umulsidi3_nov6): Likewise.
16999         (*umulsidi3_v6): Likewise.
17000         (*umulsidi3adddi): Likewise.
17001         (*umulsidi3adddi_v6): Likewise.
17002         (*smulsi3_highpart_nov6): Likewise.
17003         (*smulsi3_highpart_v6): Likewise.
17004         (*umulsi3_highpart_nov6): Likewise.
17005         (*umulsi3_highpart_v6): Likewise.
17006         (mulhisi3): Likewise.
17007         (*mulhisi3tb): Likewise.
17008         (*mulhisi3bt): Likewise.
17009         (*mulhisi3tt): Likewise.
17010         (maddhisi4): Likewise.
17011         (*maddhisi4tb): Likewise.
17012         (*maddhisi4tt): Likewise.
17013         (maddhidi4): Likewise.
17014         (*maddhidi4tb): Likewise.
17015         (*maddhidi4tt): Likewise.
17016         (divsi3): Likewise.
17017         (udivsi3): Likewise.
17018         * config/arm/thumb2.md (thumb2_mulsi_short): Update attributes.
17019         (thumb2_mulsi_short_compare0): Likewise.
17020         (thumb2_mulsi_short_compare0_scratch): Likewise.
17021         * config/arm/arm1020e.md (1020mult1): Update attribute change.
17022         (1020mult2): Likewise.
17023         (1020mult3): Likewise.
17024         (1020mult4): Likewise.
17025         (1020mult5): Likewise.
17026         (1020mult6): Likewise.
17027         * config/arm/cortex-a15.md (cortex_a15_mult32): Update attribute
17028         change.
17029         (cortex_a15_mult64): Likewise.
17030         (cortex_a15_sdiv): Likewise.
17031         (cortex_a15_udiv): Likewise.
17032         * config/arm/arm1026ejs.md (mult1): Update attribute change.
17033         (mult2): Likewise.
17034         (mult3): Likewise.
17035         (mult4): Likewise.
17036         (mult5): Likewise.
17037         (mult6): Likewise.
17038         * config/arm/marvell-pj4.md (pj4_ir_mul): Update attribute change.
17039         (pj4_ir_div): Likewise.
17040         * config/arm/arm1136jfs.md (11_mult1): Update attribute change.
17041         (11_mult2): Likewise.
17042         (11_mult3): Likewise.
17043         (11_mult4): Likewise.
17044         (11_mult5): Likewise.
17045         (11_mult6): Likewise.
17046         (11_mult7): Likewise.
17047         * config/arm/cortex-a8.md (cortex_a8_mul): Update attribute change.
17048         (cortex_a8_mla): Likewise.
17049         (cortex_a8_mull): Likewise.
17050         (cortex_a8_smulwy): Likewise.
17051         (cortex_a8_smlald): Likewise.
17052         * config/arm/cortex-m4.md (cortex_m4_alu): Update attribute change.
17053         * config/arm/cortex-r4.md (cortex_r4_mul_4): Update attribute change.
17054         (cortex_r4_mul_3): Likewise.
17055         (cortex_r4_mla_4): Likewise.
17056         (cortex_r4_mla_3): Likewise.
17057         (cortex_r4_smlald): Likewise.
17058         (cortex_r4_mull): Likewise.
17059         (cortex_r4_sdiv): Likewise.
17060         (cortex_r4_udiv): Likewise.
17061         * config/arm/cortex-a7.md (cortex_a7_mul): Update attribute change.
17062         (cortex_a7_idiv): Likewise.
17063         * config/arm/arm926ejs.md (9_mult1): Update attribute change.
17064         (9_mult2): Likewise.
17065         (9_mult3): Likewise.
17066         (9_mult4): Likewise.
17067         (9_mult5): Likewise.
17068         (9_mult6): Likewise.
17069         * config/arm/cortex-a53.md (cortex_a53_mul): Update attribute change.
17070         (cortex_a53_sdiv): Likewise.
17071         (cortex_a53_udiv): Likewise.
17072         * config/arm/fa726te.md (726te_mult_op): Update attribute change.
17073         * config/arm/fmp626.md (mp626_mult1): Update attribute change.
17074         (mp626_mult2): Likewise.
17075         (mp626_mult3): Likewise.
17076         (mp626_mult4): Likewise.
17077         * config/arm/fa526.md (526_mult1): Update attribute change.
17078         (526_mult2): Likewise.
17079         * config/arm/arm-generic.md (mult): Update attribute change.
17080         (mult_ldsched_strongarm): Likewise.
17081         (mult_ldsched): Likewise.
17082         (multi_cycle): Likewise.
17083         * config/arm/cortex-a5.md (cortex_a5_mul): Update attribute change.
17084         * config/arm/fa606te.md (606te_mult1): Update attribute change.
17085         (606te_mult2): Likewise.
17086         (606te_mult3): Likewise.
17087         (606te_mult4): Likewise.
17088         * config/arm/cortex-a9.md (cortex_a9_mult16): Update attribute change.
17089         (cortex_a9_mac16): Likewise.
17090         (cortex_a9_multiply): Likewise.
17091         (cortex_a9_mac): Likewise.
17092         (cortex_a9_multiply_long): Likewise.
17093         * config/arm/fa626te.md (626te_mult1): Update attribute change.
17094         (626te_mult2): Likewise.
17095         (626te_mult3): Likewise.
17096         (626te_mult4): Likewise.
17098 2013-06-18  Richard Biener  <rguenther@suse.de>
17100         PR lto/57334
17101         * lto-symtab.c (lto_symtab_merge_decls): Process nodes properly.
17103 2013-06-18  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17105         PR target/57609
17106         * config/s390/s390.c (s390_chunkify_start): Replace next_real_insn
17107         with next_active_insn.
17109 2013-06-18  Alan Modra  <amodra@gmail.com>
17111         * config/rs6000/rs6000.h (enum data_align): New.
17112         (LOCAL_ALIGNMENT, DATA_ALIGNMENT): Use rs6000_data_alignment.
17113         (DATA_ABI_ALIGNMENT): Define.
17114         (CONSTANT_ALIGNMENT): Correct comment.
17115         * config/rs6000/rs6000-protos.h (rs6000_data_alignment): Declare.
17116         * config/rs6000/rs6000.c (rs6000_data_alignment): New function.
17118 2013-06-17  David Malcolm  <dmalcolm@redhat.com>
17120         * ggc-page.c (ggc_pch_write_object) <d>: Remove erroneous
17121         ATTRIBUTE_UNUSED marking.
17123 2013-06-17  Sofiane Naci  <sofiane.naci@arm.com>
17125         * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>): Add r<-w
17126         alternative and update.
17127         (aarch64_dup_lanedi): Delete.
17128         * config/aarch64/arm_neon.h (vdup<bhsd>_lane_*): Update.
17129         * config/aarch64/aarch64-simd-builtins.def: Update.
17131 2013-06-17  Richard Biener  <rguenther@suse.de>
17133         * lto-streamer.h (enum LTO_tags): Add LTO_tree_scc.
17134         (lto_input_scc): Declare.
17135         (lto_input_tree_1): Likewise.
17136         (struct lto_stats_d): Add num_tree_bodies_output and
17137         num_pickle_refs_output.
17138         * lto-streamer-in.c (lto_read_body): Use streamer_tree_cache_get_tree.
17139         (lto_read_tree_1): Split out from ...
17140         (lto_read_tree): ... this.
17141         (lto_input_scc): New function.
17142         (lto_input_tree_1): Split out from ...
17143         (lto_input_tree): ... this.  Handle LTO_tree_scc.
17144         (lto_data_in_create): Create the streamer cache without hashes.
17145         * lto-streamer-out.c (create_output_block): Create the streamer
17146         cache with hashes when not doing WPA.
17147         (lto_write_tree_1): Split out from ...
17148         (lto_write_tree): ... this.
17149         (get_symbol_initial_value): New function.
17150         (lto_output_tree_1): Split out from ...
17151         (lto_output_tree): ... this.  Write trees as series of SCCs
17152         using a DFS walk via DFS_write_tree.
17153         (struct sccs, struct scc_entry): New types.
17154         (next_dfs_num, sccstack, sccstate, sccstate_obstack): New globals.
17155         (DFS_write_tree_body): New function.
17156         (DFS_write_tree): Likewise.
17157         (hash_tree): Likewise.
17158         (scc_entry_compare): Likewise.
17159         (hash_scc): Likewise.
17160         (tree_is_indexable): DEBUG_EXPR_DECLs are local entities.
17161         * tree-streamer-in.c (lto_input_ts_list_tree_pointers): Stream
17162         TREE_CHAIN as regular reference.
17163         (streamer_read_integer_cst): Remove.
17164         (streamer_get_pickled_tree): Adjust.
17165         * tree-streamer-out.c (streamer_write_chain): Disable streaming
17166         of DECL_EXTERNALs in BLOCK_VARS for now.
17167         (write_ts_list_tree_pointers): Stream TREE_CHAIN as regular
17168         reference.
17169         * tree-streamer.c (streamer_tree_cache_add_to_node_array):
17170         Add hash value argument and record that if hashes are recorded
17171         in the cache.
17172         (streamer_tree_cache_insert_1): Adjust.
17173         (streamer_tree_cache_insert): Likewise.
17174         (streamer_tree_cache_insert_at): Rename to ...
17175         (streamer_tree_cache_replace_tree): ... this and adjust.
17176         (streamer_tree_cache_append): Adjust.
17177         (record_common_node): Likewise.
17178         (streamer_tree_cache_create): Add argument whether to
17179         record hash values together with trees.
17180         (streamer_tree_cache_delete): Adjust.
17181         * tree-streamer.h (struct streamer_tree_cache_d): Add
17182         vector of hashes.
17183         (streamer_read_integer_cst): Remove.
17184         (streamer_tree_cache_insert): Adjust.
17185         (streamer_tree_cache_append): Likewise.
17186         (streamer_tree_cache_insert_at): Rename to ...
17187         (streamer_tree_cache_replace_tree): ... this and adjust.
17188         (streamer_tree_cache_create): Add argument whether to record hashes.
17189         (streamer_tree_cache_get): Rename to ...
17190         (streamer_tree_cache_get_tree): ... this.
17191         (streamer_tree_cache_get_hash): New function.
17192         * tree.c (cache_integer_cst): New function.
17193         * tree.h (cache_integer_cst): Declare.
17194         (ANON_AGGRNAME_FORMAT, ANON_AGGRNAME_P): Move here from cp/cp-tree.h.
17195         * lto-symtab.c (lto_varpool_replace_node): Only release
17196         DECL_INITIAL of non-prevailing decls.
17197         * varpool.c (varpool_remove_initializer): Do not release
17198         DECL_INITIAL when we are still in CGRAPH_LTO_STREAMING.
17200 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17202         * config/mips/mips.h (ISA_HAS_MUL3): Include TARGET_MIPS5900.
17203         (ISA_HAS_MULT, ISA_HAS_DMULT, ISA_HAS_DIV, ISA_HAS_DDIV): New macros.
17204         * config/mips/mips.md (mul<mode>3, mul<mode>3_internal)
17205         (mul<mode>3_r4000): Require ISA_HAS_<D>MULT.
17206         (mul<mode>3_mul3): Handle TARGET_MIPS5900.
17207         (mulsidi3_64bit_dmul): Remove redundant TARGET_64BIT test.
17208         (<su>muldi3_highpart, <su>muldi3_highpart_internal, <u>mulditi3)
17209         (<u>mulditi3_internal, <u>mulditi3_r4000): Require ISA_HAS_DMULT
17210         instead of TARGET_64BIT.
17211         (divmod<mode>4, udivmod<mode>4, <u>divmod<GPR:mode>4_hilo_<HILO:mode>):
17212         Require ISA_HAS_<D>DIV.
17214 2013-06-16  Richard Sandiford  <rdsandiford@googlemail.com>
17216         * config.gcc (mips*-mti-linux*, mips64*-*-linux*, mipsisa64*-*-linux*)
17217         (mips*-*-linux*): Move default with_llsc setting to where other
17218         defaults are set.
17219         (mips*-*-vxworks*): Move with_arch default from with_cpu block to
17220         with_arch block.
17221         (mips64r5900-*-*, mips64r5900el-*-*, mipsr5900-*-*, mipsr5900el-*-*):
17222         Likewise.  Remove default with_tune setting.  Move default float
17223         setting to its own block.  Handle with_llsc in the same block as above.
17225 2013-06-16  Joern Rennecke  <joern.rennecke@embecosm.com>
17227         PR rtl-optimization/57425
17228         PR rtl-optimization/57569
17229         * alias.c (write_dependence_p): Add new parameters mem_mode,
17230         canon_mem_addr and mem_canonicalized.  Change type of writep to bool.
17231         Changed all callers.
17232         (canon_anti_dependence): New function.
17233         * cse.c (check_dependence): Use canon_anti_dependence.
17234         * cselib.c (cselib_invalidate_mem): Likewise.
17235         * rtl.h (canon_anti_dependence): Declare.
17237 2013-06-16  Jürgen Urban  <JuergenUrban@gmx.de>
17239         * config/mips/mips.h (ISA_HAS_LL_SC): Exclude TARGET_MIPS5900.
17240         * config/mips/mips.c (mips_start_ll_sc_sync_block): Output
17241         ".set mips3" for 64-bit targets.
17243 2013-06-15  Dehao Chen  <dehao@google.com>
17245         * tree-flow.h (gimple_check_call_matching_types): Add new argument.
17246         * gimple-low.c (gimple_check_call_matching_types): Likewise.
17247         (gimple_check_call_args): Likewise.
17248         * value-prof.c (check_ic_target): Likewise.
17249         * ipa-inline.c (early_inliner): Likewise.
17250         * ipa-prop.c (update_indirect_edges_after_inlining): Likewise.
17251         * cgraph.c (cgraph_create_edge_1): Likewise.
17252         (cgraph_make_edge_direct): Likewise.
17254 2013-06-14  Michael Meissner  <meissner@linux.vnet.ibm.com>
17256         PR target/57615
17257         * config/rs6000/rs6000.md (mov<mode>_ppc64): Call
17258         rs6000_output_move_128bit to handle emitting quad memory
17259         operations.  Set attribute length to 8 bytes.
17261 2013-06-14  Vidya Praveen  <vidyapraveen@arm.com>
17263         * config/aarch64/aarch64-simd.md (aarch64_<su>mlal_lo<mode>):
17264         New pattern.
17265         (aarch64_<su>mlal_hi<mode>, aarch64_<su>mlsl_lo<mode>): Likewise.
17266         (aarch64_<su>mlsl_hi<mode>, aarch64_<su>mlal<mode>): Likewise.
17267         (aarch64_<su>mlsl<mode>): Likewise.
17269 2013-06-14  Mike Stump  <mikestump@comcast.net>
17271         * Makefile.in (TARGET_H): Add insn-codes.h.
17273 2013-06-14  Alan Modra  <amodra@gmail.com>
17275         PR middle-end/57134
17276         PR middle-end/57586
17277         * expr.c (expand_expr_real_1 <normal_inner_ref>): Pass
17278         EXPAND_MEMORY and EXPAND_WRITE to recursive call.  Don't use
17279         bitfield expansion when EXPAND_MEMORY.
17280         (expand_expr_real_1 <VIEW_CONVERT_EXPR>): Pass modifier likewise.
17282 2013-06-13  Michael Meissner  <meissner@linux.vnet.ibm.com>
17284         * config/rs6000/rs6000.c (rs6000_option_override_internal): Move
17285         test for clearing quad memory on 32-bit later.
17287 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17289         * fold-const.c (negate_expr_p): Handle VECTOR_CST.
17290         (fold_negate_expr): Likewise.
17291         (fold_real_zero_addition_p): Handle vectors.
17292         (fold_binary_loc) <PLUS_EXPR, MINUS_EXPR>: Likewise.
17294 2013-06-14  Alan Modra  <amodra@gmail.com>
17296         * varasm.c (force_const_mem): Revert 2013-06-07 change.
17298 2013-06-13  Jan Hubicka  <jh@suse.cz>
17300         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
17301         Local comdats are not externally visible.
17302         * symtab.c (dump_symtab_base): Dump externally visible.
17303         (verify_symtab_base): Verify back links in the symtab hash.
17305 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17307         * fold-const.c (operand_equal_p): Consider NOP_EXPR and
17308         CONVERT_EXPR as equal nodes.
17310 2013-06-13  Bin Cheng  <bin.cheng@arm.com>
17312         * rtlanal.c (noop_move_p): Check the code to be executed for COND_EXEC.
17314 2013-06-13  Marc Glisse  <marc.glisse@inria.fr>
17316         * tree-ssa-forwprop.c (simplify_bitwise_binary, associate_plusminus):
17317         Generalize to complex and vector.
17318         * tree.c (build_all_ones_cst): New function.
17319         * tree.h (build_all_ones_cst): Declare it.
17321 2013-06-13  Alan Modra  <amodra@gmail.com>
17323         * config/rs6000/rs6000.h (LONG_DOUBLE_LARGE_FIRST): Define.
17324         * config/rs6000/rs6000.md (signbittf2): New insn.
17325         (extenddftf2_internal): Use LONG_DOUBLE_LARGE_FIRST.
17326         (abstf2_internal, cmptf_internal2): Likewise.
17327         * config/rs6000/spe.md (spe_abstf2_cmp, spe_abstf2_tst): Likewise.
17329 2013-06-12  Michael Meissner  <meissner@linux.vnet.ibm.com>
17330             Pat Haugen  <pthaugen@us.ibm.com>
17331             Peter Bergner  <bergner@vnet.ibm.com>
17333         * config/rs6000/rs6000.c (emit_load_locked): Add support for
17334         power8 byte, half-word, and quad-word atomic instructions.
17335         (emit_store_conditional): Likewise.
17336         (rs6000_expand_atomic_compare_and_swap): Likewise.
17337         (rs6000_expand_atomic_op): Likewise.
17339         * config/rs6000/sync.md (larx): Add new modes for power8.
17340         (stcx): Likewise.
17341         (AINT): New mode iterator to include TImode as well as normal
17342         integer modes on power8.
17343         (fetchop_pred): Use int_reg_operand instead of gpc_reg_operand so
17344         that VSX registers are not considered.  Use AINT mode iterator
17345         instead of INT1 to allow inclusion of quad word atomic operations
17346         on power8.
17347         (load_locked<mode>): Likewise.
17348         (store_conditional<mode>): Likewise.
17349         (atomic_compare_and_swap<mode>): Likewise.
17350         (atomic_exchange<mode>): Likewise.
17351         (atomic_nand<mode>): Likewise.
17352         (atomic_fetch_<fetchop_name><mode>): Likewise.
17353         (atomic_nand_fetch<mode>): Likewise.
17354         (mem_thread_fence): Use gen_loadsync_<mode> instead of enumerating
17355         each type.
17356         (ATOMIC): On power8, add QImode, HImode modes.
17357         (load_locked<QHI:mode>_si): Varients of load_locked for QI/HI
17358         modes that promote to SImode.
17359         (load_lockedti): Convert TImode arguments to PTImode, so that we
17360         get a guaranteed even/odd register pair.
17361         (load_lockedpti): Likewise.
17362         (store_conditionalti): Likewise.
17363         (store_conditionalpti): Likewise.
17365         * config/rs6000/rs6000.md (QHI): New mode iterator for power8
17366         atomic load/store instructions.
17367         (HSI): Likewise.
17369 2013-06-12  Richard Sandiford  <rdsandiford@googlemail.com>
17371         * config/mips/mips.md (extended_mips16): Include GOT and constant-pool
17372         loads.
17373         (insn_count): New attribute, with most cases extracted from...
17374         (length): ...here.  Redefine most cases in terms of insn_count.
17375         (single_insn): Delete.
17376         (can_delay): Use insn_count to check for single instructions.
17377         (*mul<mode>3_r4300, mul<mode>3_r4000, *mul_acc_si, *mul_acc_si_r3900)
17378         (*msac_using_macc, *mul_sub_si, <u>mulsidi3_32bit_r4000)
17379         (<u>mulsidi3_64bit_r4000, <su>muldi3_highpart_internal)
17380         (<su>mulsi3_highpart_split, <su>muldi3_highpart_internal)
17381         (<u>mulditi3_r4000, *div<mode>3, *recip<mode>3, divmod<mode>4)
17382         (udivmod<mode>4, sqrt<mode>2, *rsqrt<mode>a, *rsqrt<mode>b)
17383         (fix_truncdfsi2_macro, fix_truncsfsi2_macro, *lea_high64)
17384         (*lea64, cprestore_<mode>, clear_hazard_<mode>, <unnamed insn>)
17385         (casesi_internal_mips16_<mode>, *tls_get_tp_<mode>_split)
17386         (tls_get_tp_mips16, *tls_get_tp_mips16_call_<mode>): Use "insn_count"
17387         rather than "length".
17388         (tls_get_tp_<mode>): Likewise.  Remove redundant "no_delay" attribute.
17389         * config/mips/mips-ps-3d.md (mips_c_cond_4s, mips_cabs_cond_4s):
17390         Use "insn_count" rather than "length".
17391         * config/mips/mips-dsp.md
17392         (mips_l<SHORT:size><u>x_ext<GPR:mode>_<P:mode>)
17393         (mips_l<GPR:size>x_<P:mode>, *mips_lw<u>x_<P:mode>_ext): Remove
17394         length attributes.
17396 2013-06-12  Marc Glisse  <marc.glisse@inria.fr>
17398         PR tree-optimization/57361
17399         * tree-ssa-dse.c (dse_possible_dead_store_p): Handle self-assignment.
17401 2013-06-12  Sofiane Naci  <sofiane.naci@arm.com>
17403         * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Convert
17404         to split.
17405         (aarch64_simd_combine<mode>): New instruction expansion.
17406         * config/aarch64/aarch64-protos.h (aarch64_split_simd_combine): New
17407         function prototype.
17408         * config/aarch64/aarch64.c (aarch64_split_combine): New function.
17409         * config/aarch64/iterators.md (Vdbl): Add entry for DF.
17411 2013-06-12  Jan Hubicka  <jh@suse.cz>
17413         * cgraph.c (verify_edge_corresponds_to_fndecl): Be lax about
17414         decl has when in streaming stage.
17415         * lto-symtab.c (lto_symtab_merge_symbols): Likewise.
17416         * cgraph.h (cgraph_state): Add CGRAPH_LTO_STREAMING.
17418 2013-06-12  Roland Stigge  <stigge@antcom.de>
17420         PR target/57578
17421         * config/rs6000/t-linux (MULTIARCH_DIRNAME): Fix SPE version detection.
17423 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17425         PR tree-optimization/57537
17426         * tree-vect-patterns.c (vect_recog_widen_mult_pattern): If
17427         vect_handle_widen_op_by_const, convert oprnd1 to half_type1.
17429 2013-06-12  Richard Biener  <rguenther@suse.de>
17431         * data-streamer.h (streamer_write_char_stream): CSE
17432         obs->current_pointer.
17433         * data-streamer-out.c (streamer_write_uhwi_stream): Inline
17434         streamer_write_char_stream manually and optimize the resulting loop.
17435         (streamer_write_hwi_stream): Likewise.
17437 2013-06-12  Jan Hubicka  <jh@suse.cz>
17439         * lto-symtab.c (lto_symtab_merge_symbols): Populate symtab hashtable.
17440         * cgraph.h (varpool_create_empty_node): Declare.
17441         * lto-cgraph.c (input_node, input_varpool_node): Forcingly create
17442         duplicated nodes.
17443         * symtab.c (symtab_unregister_node): Be lax about missin entries
17444         in node hash.
17445         (symtab_get_node): Update comment.
17446         * varpool.c (varpool_create_empty_node): Break out from ...
17447         (varpool_node_for_decl): ... here.
17448         * lto-streamer.h (lto_file_decl_data): Add RESOLUTION_MAP.
17450 2013-06-12  Eric Botcazou  <ebotcazou@adacore.com>
17452         * expr.c (expand_expr_real_1) <TARGET_MEM_REF>: Use straight-line flow.
17453         <MEM_REF>: Use 'type' instead of TREE_TYPE (exp) and tidy up the first
17454         part.  Use straight-line flow at the end.
17455         <COMPONENT_REF>: Remove superfluous else.
17456         <VIEW_CONVERT_EXPR>: Use 'type' instead of TREE_TYPE (exp).
17458 2013-06-12  Jakub Jelinek  <jakub@redhat.com>
17460         PR target/56564
17461         * varasm.c (decl_binds_to_current_def_p): Call binds_local_p
17462         target hook even for !TREE_PUBLIC decls.  If no resolution info
17463         is available, return false for common and external decls.
17465 2013-06-12  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
17467         * config/rl78/constraints.md (U): New constraint.
17468         * config/rl78/rl78.md (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): Add
17469         valloc attribute.
17471 2013-06-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
17473         PR target/57589
17474         * config/rs6000/driver-rs6000.c (elf_platform): Make buffer static
17475         to allow returning address to AT_PLATFORM name.
17477 2013-06-11  Jan Hubicka  <jh@suse.cz>
17479         * cgraph.c (cgraph_create_function_alias): Set weakref flag.
17480         * cgraph.h (symtab_node_base): Add weakref flag.
17481         * cgraphunit.c (cgraph_reset_node): Clear weakref flag.
17482         (handle_alias_pairs): Set weakref flag, do not set DECL_EXTERNAL.
17483         (output_weakrefs): Use weakref flag.
17484         * fold-const.c (simple_operand_p): Handle WEAK.
17485         * gimple-fold.c (can_refer_decl_in_current_unit_p): Drop weakref.
17486         * ipa.c (varpool_externally_visible_p): Drop weakref.
17487         (function_and_variable_visibility): Update comment; fix weakref
17488         sanity checks; do not clear DECL_WEAK on them.
17489         * lto-cgraph.c (lto_output_node): update.
17490         (lto_output_varpool_node): Update.
17491         (input_overwrite_node): Update.
17492         (input_node): Update.
17493         (input_varpool_node): Update.
17494         * lto-symtab.c (lto_symtab_symbol_p): Do not special case weakrefs.
17495         (lto_symtab_merge_symbols): Add sanity check.
17496         (lto_symtab_prevailing_decl): Do not special case weakrefs.
17497         * passes.c (rest_of_decl_compilation): Set static flag, too.
17498         * symtab.c (dump_symtab_base): Dump weakref.
17499         (verify_symtab_base): Sanity check weakrefs.
17500         (symtab_make_decl_local): Remove duplicated code.
17501         (symtab_alias_ultimate_target): Simplify.
17502         * varpool.c (varpool_create_variable_alias): Set weakref flag.
17504 2013-06-11  Tom de Vries  <tom@codesourcery.com>
17506         * genautomata.c (gen_regexp_sequence): Handle els_num == -1.  Handle
17507         sequence_vect == NULL.
17509 2013-06-11  DJ Delorie  <dj@redhat.com>
17511         * config/rl78/rl78.c (TARGET_UNWIND_WORD_MODE): Define.
17512         (rl78_unwind_word_mode): New.
17514 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17516         * final.c (debug_prefix_maps): Make static.
17518 2013-06-11  David Malcolm  <dmalcolm@redhat.com>
17520         * function.c (initial_trampoline): Remove stray copy.
17522 2013-06-11  Sofiane Naci  <sofiane.naci@arm.com>
17524         * config/aarch64/aarch64-simd.md (move_lo_quad_<mode>): Update.
17526 2013-06-11  Martin Jambor  <mjambor@suse.cz>
17528         * ipa-cp.c (ipa_get_indirect_edge_target_1): Check that param_index is
17529         within bounds at the beginning of the function.
17531 2013-06-11  Alan Modra  <amodra@gmail.com>
17533         * varasm.c (get_section): Don't die on !DECL_P decl.  Tidy error
17534         reporting.
17535         (get_named_section): Don't NULL !DECL_P decl.
17537 2013-06-11  Igor Zamyatin  <igor.zamyatin@intel.com>
17539         * doc/invoke.texi (core-avx2): Document.
17540         (slm): Likewise.
17541         (atom): Updated with MOVBE.
17543 2013-06-11  Richard Biener  <rguenther@suse.de>
17545         * collect2.c (main): Do not redirect ld stdout/stderr when debugging.
17547 2013-06-11  Anton Blanchard  <anton@samba.org>
17549         * config/rs6000/rs6000.c (rs6000_adjust_atomic_subword): Calculate
17550         correct shift value in little-endian mode.
17552 2013-06-11  Jakub Jelinek  <jakub@redhat.com>
17554         PR target/56564
17555         * varasm.c (get_variable_align): Move #endif to the right place.
17557 2013-06-10  Cary Coutant  <ccoutant@google.com>
17559         * dwarf2out.c (hash_external_ref): Use die_symbol or signature
17560         for hash so that hash table traversal order is deterministic.
17562 2013-06-10  Michael Meissner  <meissner@linux.vnet.ibm.com>
17563             Pat Haugen  <pthaugen@us.ibm.com>
17564             Peter Bergner  <bergner@vnet.ibm.com>
17566         * config/rs6000/vector.md (GPR move splitter): Do not split moves
17567         of vectors in GPRS if they are direct moves or quad word load or
17568         store moves.
17570         * config/rs6000/rs6000-protos.h (rs6000_output_move_128bit): Add
17571         declaration.
17572         (direct_move_p): Likewise.
17573         (quad_load_store_p): Likewise.
17575         * config/rs6000/rs6000.c (enum rs6000_reg_type): Simplify register
17576         classes into bins based on the physical register type.
17577         (reg_class_to_reg_type): Likewise.
17578         (IS_STD_REG_TYPE): Likewise.
17579         (IS_FP_VECT_REG_TYPE): Likewise.
17580         (reload_fpr_gpr): Arrays to determine what insn to use if we can
17581         use direct move instructions.
17582         (reload_gpr_vsx): Likewise.
17583         (reload_vsx_gpr): Likewise.
17584         (rs6000_init_hard_regno_mode_ok): Precalculate the register type
17585         information that is a simplification of register classes.  Also
17586         precalculate direct move reload helpers.
17587         (direct_move_p): New function to return true if the operation can
17588         be done as a direct move instruciton.
17589         (quad_load_store_p): New function to return true if the operation
17590         is a quad memory operation.
17591         (rs6000_legitimize_address): If quad memory, only allow register
17592         indirect for TImode addresses.
17593         (rs6000_legitimate_address_p): Likewise.
17594         (enum reload_reg_type): Delete, replace with rs6000_reg_type.
17595         (rs6000_reload_register_type): Likewise.
17596         (register_to_reg_type): Return register type.
17597         (rs6000_secondary_reload_simple_move): New helper function for
17598         secondary reload and secondary memory needed to identify anything
17599         that is a simple move, and does not need reloading.
17600         (rs6000_secondary_reload_direct_move): New helper function for
17601         secondary reload to identify cases that can be done with several
17602         instructions via the direct move instructions.
17603         (rs6000_secondary_reload_move): New helper function for secondary
17604         reload to identify moves between register types that can be done.
17605         (rs6000_secondary_reload): Add support for quad memory operations
17606         and for direct move.
17607         (rs6000_secondary_memory_needed): Likewise.
17608         (rs6000_debug_secondary_memory_needed): Change argument names.
17609         (rs6000_output_move_128bit): New function to return the move to
17610         use for 128-bit moves, including knowing about the various
17611         limitations of quad memory operations.
17613         * config/rs6000/vsx.md (vsx_mov<mode>): Add support for quad
17614         memory operations.  call rs6000_output_move_128bit for the actual
17615         instruciton(s) to generate.
17616         (vsx_movti_64bit): Likewise.
17618         * config/rs6000/rs6000.md (UNSPEC_P8V_FMRGOW): New unspec values.
17619         (UNSPEC_P8V_MTVSRWZ): Likewise.
17620         (UNSPEC_P8V_RELOAD_FROM_GPR): Likewise.
17621         (UNSPEC_P8V_MTVSRD): Likewise.
17622         (UNSPEC_P8V_XXPERMDI): Likewise.
17623         (UNSPEC_P8V_RELOAD_FROM_VSX): Likewise.
17624         (UNSPEC_FUSION_GPR): Likewise.
17625         (FMOVE128_GPR): New iterator for direct move.
17626         (f32_lv): New mode attribute for load/store of SFmode/SDmode values.
17627         (f32_sv): Likewise.
17628         (f32_dm): Likewise.
17629         (zero_extend<mode>di2_internal1): Add support for power8 32-bit
17630         loads and direct move instructions.
17631         (zero_extendsidi2_lfiwzx): Likewise.
17632         (extendsidi2_lfiwax): Likewise.
17633         (extendsidi2_nocell): Likewise.
17634         (floatsi<mode>2_lfiwax): Likewise.
17635         (lfiwax): Likewise.
17636         (floatunssi<mode>2_lfiwzx): Likewise.
17637         (lfiwzx): Likewise.
17638         (fix_trunc<mode>_stfiwx): Likewise.
17639         (fixuns_trunc<mode>_stfiwx): Likewise.
17640         (mov<mode>_hardfloat, 32-bit floating point): Likewise.
17641         (mov<move>_hardfloat64, 64-bit floating point): Likewise.
17642         (parity<mode>2_cmpb): Set length/type attr.
17643         (unnamed shift right patterns, mov<mode>_internal2): Change type attr
17644         for 'mr.' to fast_compare.
17645         (bpermd_<mode>): Change type attr to popcnt.
17646         (p8_fmrgow_<mode>): New insns for power8 direct move support.
17647         (p8_mtvsrwz_1): Likewise.
17648         (p8_mtvsrwz_2): Likewise.
17649         (reload_fpr_from_gpr<mode>): Likewise.
17650         (p8_mtvsrd_1): Likewise.
17651         (p8_mtvsrd_2): Likewise.
17652         (p8_xxpermdi_<mode>): Likewise.
17653         (reload_vsx_from_gpr<mode>): Likewise.
17654         (reload_vsx_from_gprsf): Likewise.
17655         (p8_mfvsrd_3_<mode>): LIkewise.
17656         (reload_gpr_from_vsx<mode>): Likewise.
17657         (reload_gpr_from_vsxsf): Likewise.
17658         (p8_mfvsrd_4_disf): Likewise.
17659         (multi-word GPR splits): Do not split direct moves or quad memory
17660         operations.
17662 2013-06-10  David Malcolm  <dmalcolm@redhat.com>
17664         * tree-into-ssa.c (interesting_blocks): Make static.
17666 2013-06-10  Jakub Jelinek  <jakub@redhat.com>
17668         PR target/56564
17669         * varasm.c (align_variable): Don't use DATA_ALIGNMENT or
17670         CONSTANT_ALIGNMENT if !decl_binds_to_current_def_p (decl).
17671         Use DATA_ABI_ALIGNMENT for that case instead if defined.
17672         (get_variable_align): New function.
17673         (get_variable_section, emit_bss, emit_common,
17674         assemble_variable_contents, place_block_symbol): Use
17675         get_variable_align instead of DECL_ALIGN.
17676         (assemble_noswitch_variable): Add align argument, use it
17677         instead of DECL_ALIGN.
17678         (assemble_variable): Adjust caller.  Use get_variable_align
17679         instead of DECL_ALIGN.
17680         * config/i386/i386.h (DATA_ALIGNMENT): Adjust x86_data_alignment
17681         caller.
17682         (DATA_ABI_ALIGNMENT): Define.
17683         * config/i386/i386-protos.h (x86_data_alignment): Adjust prototype.
17684         * config/i386/i386.c (x86_data_alignment): Add opt argument.  If
17685         opt is false, only return the psABI mandated alignment increase.
17686         * config/c6x/c6x.h (DATA_ALIGNMENT): Renamed to...
17687         (DATA_ABI_ALIGNMENT): ... this.
17688         * config/mmix/mmix.h (DATA_ALIGNMENT): Renamed to...
17689         (DATA_ABI_ALIGNMENT): ... this.
17690         * config/mmix/mmix.c (mmix_data_alignment): Adjust function comment.
17691         * config/s390/s390.h (DATA_ALIGNMENT): Renamed to...
17692         (DATA_ABI_ALIGNMENT): ... this.
17693         * doc/tm.texi.in (DATA_ABI_ALIGNMENT): Document.
17694         * doc/tm.texi: Regenerated.
17696 2013-06-10  Uros Bizjak  <ubizjak@gmail.com>
17698         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Also use
17699         cmp_code to construct REG_EQUAL note.
17701 2013-06-09  Jakub Jelinek  <jakub@redhat.com>
17703         PR target/57568
17704         * config/i386/i386.md (TARGET_READ_MODIFY_WRITE peepholes): Ensure
17705         that operands[2] doesn't overlap with operands[0].
17707 2013-06-09  David Edelsohn  <dje.gcc@gmail.com>
17708             Jan Hubicka  <jh@suse.cz>
17710         * config/rs6000/rs6000.c (print_operand, 'z'): Remove historical
17711         hack to mark symbols as used.
17713 2013-06-08  Vladimir Makarov  <vmakarov@redhat.com>
17715         PR rtl-optimization/57559
17716         * lra-constraints.c (process_alt_operands): Don't discourage
17717         memory with known offset for offsetable memory constraint.
17718         * lra.c (lra_emit_add): Exchange y and z for 2-op add insn.
17720 2013-06-08  Eric Botcazou  <ebotcazou@adacore.com>
17722         * varasm.c (struct oc_local_state): Reorder fields.
17723         (output_constructor_bitfield): Replace OUTER parameter with BIT_OFFSET
17724         and adjust accordingly.
17725         (output_constructor): Reorder initialization code and adjust call to
17726         output_constructor_bitfield.
17728 2013-06-07  Jan Hubicka  <jh@suse.cz>
17730         * symtab.c (symtab_resolve_alias): Do not remove alias attribute.
17732 2013-06-07  David Malcolm  <dmalcolm@redhat.com>
17734         * tree-object-size.c (unknown): Make const.
17736 2013-06-07  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
17738         * config/s390/s390.md (cpu_facility): Add cpu_zarch.
17739         ("*movmem_short", "*clrmem_short", "*cmpmem_short): Use cpu_zarch
17740         for last alternative in the cpu_facility attribute.
17742 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17744         PR target/56315
17745         * config/arm/arm.md (*xordi3_insn): Change to insn_and_split.
17746         (xordi3): Change operand 2 constraint to arm_xordi_operand.
17747         * config/arm/arm.c (const_ok_for_dimode_op): Handle XOR.
17748         * config/arm/constraints.md (Dg): New constraint.
17749         * config/arm/neon.md (xordi3_neon): Remove.
17750         (neon_veor<mode>): Generate xordi3 instead of xordi3_neon.
17751         * config/arm/predicates.md (arm_xordi_operand): New predicate.
17753 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17755         * config/arm/arm.md (anddi3_insn): Remove duplicate alternatives.
17756         Clean up alternatives.
17758 2013-06-07  Alan Modra  <amodra@gmail.com>
17760         * config/rs6000/rs6000.c (setup_incoming_varargs): Round up
17761         va_list_gpr_size.
17763 2013-06-07  Alan Modra  <amodra@gmail.com>
17765         * varasm.c (force_const_mem): Assert mode is not VOID or BLK.
17767 2013-06-07  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
17769         * config/arm/constraints.md (Df): New constraint.
17770         * config/arm/arm.md (iordi3_insn): Use Df constraint instead of De.
17771         Correct length attribute for last two alternatives.
17773 2013-06-07  Alan Modra  <amodra@gmail.com>
17775         * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
17776         override user -mfp-in-toc.
17777         (offsettable_ok_by_alignment): Consider just the current access
17778         rather than the whole object, unless BLKmode.  Handle
17779         CONSTANT_POOL_ADDRESS_P constants that lack a decl too.
17780         (use_toc_relative_ref): Allow CONSTANT_POOL_ADDRESS_P constants
17781         for -mcmodel=medium.
17782         * config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Don't
17783         override user -mfp-in-toc or -msum-in-toc.  Default to
17784         -mno-fp-in-toc for -mcmodel=medium.
17786 2013-06-06  DJ Delorie  <dj@redhat.com>
17788         * config/rl78/rl78.c (rl78_valid_pointer_mode): New, implements
17789         TARGET_VALID_POINTER_MODE.
17791 2013-06-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
17792             Pat Haugen  <pthaugen@us.ibm.com>
17793             Peter Bergner  <bergner@vnet.ibm.com>
17795         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions):
17796         Document new power8 builtins.
17798         * config/rs6000/vector.md (and<mode>3): Add a clobber/scratch of a
17799         condition code register, to allow 128-bit logical operations to be
17800         done in the VSX or GPR registers.
17801         (nor<mode>3): Use the canonical form for nor.
17802         (eqv<mode>3): Add expanders for power8 xxleqv, xxlnand, xxlorc,
17803         vclz*, and vpopcnt* vector instructions.
17804         (nand<mode>3): Likewise.
17805         (orc<mode>3): Likewise.
17806         (clz<mode>2): LIkewise.
17807         (popcount<mode>2): Likewise.
17809         * config/rs6000/predicates.md (int_reg_operand): Rework tests so
17810         that only the GPRs are recognized.
17812         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17813         support for new power8 builtins.
17815         * config/rs6000/rs6000-builtin.def (xscvspdpn): Add new power8
17816         builtin functions.
17817         (xscvdpspn): Likewise.
17818         (vclz): Likewise.
17819         (vclzb): Likewise.
17820         (vclzh): Likewise.
17821         (vclzw): Likewise.
17822         (vclzd): Likewise.
17823         (vpopcnt): Likewise.
17824         (vpopcntb): Likewise.
17825         (vpopcnth): Likewise.
17826         (vpopcntw): Likewise.
17827         (vpopcntd): Likewise.
17828         (vgbbd): Likewise.
17829         (vmrgew): Likewise.
17830         (vmrgow): Likewise.
17831         (eqv): Likewise.
17832         (eqv_v16qi3): Likewise.
17833         (eqv_v8hi3): Likewise.
17834         (eqv_v4si3): Likewise.
17835         (eqv_v2di3): Likewise.
17836         (eqv_v4sf3): Likewise.
17837         (eqv_v2df3): Likewise.
17838         (nand): Likewise.
17839         (nand_v16qi3): Likewise.
17840         (nand_v8hi3): Likewise.
17841         (nand_v4si3): Likewise.
17842         (nand_v2di3): Likewise.
17843         (nand_v4sf3): Likewise.
17844         (nand_v2df3): Likewise.
17845         (orc): Likewise.
17846         (orc_v16qi3): Likewise.
17847         (orc_v8hi3): Likewise.
17848         (orc_v4si3): Likewise.
17849         (orc_v2di3): Likewise.
17850         (orc_v4sf3): Likewise.
17851         (orc_v2df3): Likewise.
17853         * config/rs6000/rs6000.c (rs6000_option_override_internal): Only
17854         allow power8 quad mode in 64-bit.
17855         (rs6000_builtin_vectorized_function): Add support to vectorize
17856         ISA 2.07 count leading zeros, population count builtins.
17857         (rs6000_expand_vector_init): On ISA 2.07 use xscvdpspn to form
17858         V4SF vectors instead of xscvdpsp to avoid IEEE related traps.
17859         (builtin_function_type): Add vgbbd builtin function which takes an
17860         unsigned argument.
17861         (altivec_expand_vec_perm_const): Add support for new power8 merge
17862         instructions.
17864         * config/rs6000/vsx.md (VSX_L2): New iterator for 128-bit types,
17865         that does not include TImdoe for use with 32-bit.
17866         (UNSPEC_VSX_CVSPDPN): Support for power8 xscvdpspn and xscvspdpn
17867         instructions.
17868         (UNSPEC_VSX_CVDPSPN): Likewise.
17869         (vsx_xscvdpspn): Likewise.
17870         (vsx_xscvspdpn): Likewise.
17871         (vsx_xscvdpspn_scalar): Likewise.
17872         (vsx_xscvspdpn_directmove): Likewise.
17873         (vsx_and<mode>3): Split logical operations into 32-bit and
17874         64-bit. Add support to do logical operations on TImode as well as
17875         VSX vector types.  Allow logical operations to be done in either
17876         VSX registers or in general purpose registers in 64-bit mode.  Add
17877         splitters if GPRs were used. For AND, add clobber of CCmode to
17878         allow use of ANDI on GPRs.  Rewrite nor to use the canonical RTL
17879         encoding.
17880         (vsx_and<mode>3_32bit): Likewise.
17881         (vsx_and<mode>3_64bit): Likewise.
17882         (vsx_ior<mode>3): Likewise.
17883         (vsx_ior<mode>3_32bit): Likewise.
17884         (vsx_ior<mode>3_64bit): Likewise.
17885         (vsx_xor<mode>3): Likewise.
17886         (vsx_xor<mode>3_32bit): Likewise.
17887         (vsx_xor<mode>3_64bit): Likewise.
17888         (vsx_one_cmpl<mode>2): Likewise.
17889         (vsx_one_cmpl<mode>2_32bit): Likewise.
17890         (vsx_one_cmpl<mode>2_64bit): Likewise.
17891         (vsx_nor<mode>3): Likewise.
17892         (vsx_nor<mode>3_32bit): Likewise.
17893         (vsx_nor<mode>3_64bit): Likewise.
17894         (vsx_andc<mode>3): Likewise.
17895         (vsx_andc<mode>3_32bit): Likewise.
17896         (vsx_andc<mode>3_64bit): Likewise.
17897         (vsx_eqv<mode>3_32bit): Add support for power8 xxleqv, xxlnand,
17898         and xxlorc instructions.
17899         (vsx_eqv<mode>3_64bit): Likewise.
17900         (vsx_nand<mode>3_32bit): Likewise.
17901         (vsx_nand<mode>3_64bit): Likewise.
17902         (vsx_orc<mode>3_32bit): Likewise.
17903         (vsx_orc<mode>3_64bit): Likewise.
17905         * config/rs6000/rs6000.h (VLOGICAL_REGNO_P): Update comment.
17907         * config/rs6000/altivec.md (UNSPEC_VGBBD): Add power8 vgbbd
17908         instruction.
17909         (p8_vmrgew): Add power8 vmrgew and vmrgow instructions.
17910         (p8_vmrgow): Likewise.
17911         (altivec_and<mode>3): Add clobber of CCmode to allow AND using
17912         GPRs to be split under VSX.
17913         (p8v_clz<mode>2): Add power8 count leading zero support.
17914         (p8v_popcount<mode>2): Add power8 population count support.
17915         (p8v_vgbbd): Add power8 gather bits by bytes by doubleword
17916         support.
17918         * config/rs6000/rs6000.md (eqv<mode>3): Add support for powerp eqv
17919         instruction.
17921         * config/rs6000/altivec.h (vec_eqv): Add defines to export power8
17922         builtin functions.
17923         (vec_nand): Likewise.
17924         (vec_vclz): Likewise.
17925         (vec_vclzb): Likewise.
17926         (vec_vclzd): Likewise.
17927         (vec_vclzh): Likewise.
17928         (vec_vclzw): Likewise.
17929         (vec_vgbbd): Likewise.
17930         (vec_vmrgew): Likewise.
17931         (vec_vmrgow): Likewise.
17932         (vec_vpopcnt): Likewise.
17933         (vec_vpopcntb): Likewise.
17934         (vec_vpopcntd): Likewise.
17935         (vec_vpopcnth): Likewise.
17936         (vec_vpopcntw): Likewise.
17938 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17940         PR rtl-optimization/57468
17941         * config/i386/i386.c (inline_secondary_memory_needed): Ignore
17942         spilled pseudos.
17944 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17946         PR rtl-optimization/57459
17947         * lra-constraints.c (update_ebb_live_info): Fix typo for operand
17948         type when setting live regs.
17950 2013-06-06  Vladimir Makarov  <vmakarov@redhat.com>
17952         * config/s390/s390.opt (mlra): New option.
17953         * config/s390/s390.c (s390_decompose_address): Check displacement
17954         for all registers for LRA.
17955         (s390_secondary_reload): Don't used secondary reloads for LRA.
17956         (s390_lra_p): New function.
17957         (TARGET_LRA_P): Define.
17958         * config/s390/s390.md (*movmem_short, *clrmem_short): Change value
17959         of attribute cpu_facility to zarch for the last alternative.
17960         (*cmpmem_short): Ditto.
17962 2013-06-06  Eric Botcazou  <ebotcazou@adacore.com>
17964         * config/arm/arm.c (arm_r3_live_at_start_p): New predicate.
17965         (arm_compute_static_chain_stack_bytes): Use it.  Tidy up.
17966         (arm_expand_prologue): Likewise.
17968 2013-06-06  Teresa Johnson  <tejohnson@google.com>
17970         PR c++/53743
17971         * ifcvt.c (find_if_case_1): Replace BB_COPY_PARTITION with assert
17972         as this is now done by redirect_edge_and_branch_force.
17973         * function.c (thread_prologue_and_epilogue_insns): Insert new bb after
17974         barriers, and fix interaction with splitting.
17975         * emit-rtl.c (try_split): Copy REG_CROSSING_JUMP notes.
17976         * cfgcleanup.c (try_forward_edges): Fix early return value to properly
17977         reflect changes made in the routine.
17978         * bb-reorder.c (emit_barrier_after_bb): Move to cfgrtl.c.
17979         (fix_up_fall_thru_edges): Remove incorrect check for bb layout order
17980         since this is called in cfglayout mode, and replace partition fixup
17981         with assert as that is now done by force_nonfallthru_and_redirect.
17982         (add_reg_crossing_jump_notes): Handle the fact that some jumps may
17983         already be marked with region crossing note.
17984         (insert_section_boundary_note): Make non-static, gate on flag
17985         has_bb_partition, rewrite to also check for multiple partitions.
17986         (rest_of_handle_reorder_blocks): Remove call to
17987         insert_section_boundary_note, now done later during free_cfg.
17988         (duplicate_computed_gotos): Don't duplicate partition crossing edge.
17989         * bb-reorder.h (insert_section_boundary_note): Declare.
17990         * Makefile.in (cfgrtl.o): Depend on bb-reorder.h
17991         * cfgrtl.c (rest_of_pass_free_cfg): If partitions exist
17992         invoke insert_section_boundary_note.
17993         (try_redirect_by_replacing_jump): Remove unnecessary
17994         check for region crossing note.
17995         (fixup_partition_crossing): New function.
17996         (rtl_redirect_edge_and_branch): Fixup partition boundaries.
17997         (emit_barrier_after_bb): Move here from bb-reorder.c, handle insertion
17998         in non-cfglayout mode.
17999         (force_nonfallthru_and_redirect): Fixup partition boundaries,
18000         remove old code that tried to do this. Emit barrier correctly
18001         when we are in cfglayout mode.
18002         (last_bb_in_partition): New function.
18003         (rtl_split_edge): Correctly fixup partition boundaries.
18004         (commit_one_edge_insertion): Remove old code that tried to
18005         fixup region crossing edge since this is now handled in
18006         split_block, and set up insertion point correctly since
18007         block may now end in a jump.
18008         (verify_hot_cold_block_grouping): Guard against checking when not in
18009         linearized RTL mode.
18010         (rtl_verify_edges): Add checks for incorrect/missing REG_CROSSING_JUMP
18011         notes.
18012         (rtl_verify_flow_info_1): Move verify_hot_cold_block_grouping to
18013         rtl_verify_flow_info, so not called in cfglayout mode.
18014         (rtl_verify_flow_info): Move verify_hot_cold_block_grouping here.
18015         (fixup_reorder_chain): Remove old code that attempted to fixup region
18016         crossing note as this is now handled in force_nonfallthru_and_redirect.
18017         (duplicate_insn_chain): Don't duplicate switch section notes.
18018         (rtl_can_remove_branch_p): Remove unnecessary check for region crossing
18019         note.
18020         * basic-block.h (emit_barrier_after_bb): Declare.
18022 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18024         * config/arm/arm-fixed.md (add<mode>3,usadd<mode>3,ssadd<mode>3,
18025         sub<mode>3, ussub<mode>3, sssub<mode>3, arm_ssatsihi_shift,
18026         arm_usatsihi): Adjust alternatives for arm_restrict_it.
18028 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18030         * config/arm/arm-ldmstm.ml: Set "predicable_short_it" to "no"
18031         where appropriate.
18032         * config/arm/ldmstm.md: Regenerate.
18034 2013-06-06  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18036         * config/arm/sync.md (atomic_loaddi_1):
18037         Disable predication for arm_restrict_it.
18038         (arm_load_exclusive<mode>): Likewise.
18039         (arm_load_exclusivesi): Likewise.
18040         (arm_load_exclusivedi): Likewise.
18041         (arm_load_acquire_exclusive<mode>): Likewise.
18042         (arm_load_acquire_exclusivesi): Likewise.
18043         (arm_load_acquire_exclusivedi): Likewise.
18044         (arm_store_exclusive<mode>): Likewise.
18045         (arm_store_exclusive<mode>): Likewise.
18046         (arm_store_release_exclusivedi): Likewise.
18047         (arm_store_release_exclusive<mode>): Likewise.
18049 2013-06-06  Richard Biener  <rguenther@suse.de>
18051         * lto-streamer.h (enum LTO_tags): Move LTO_tree_pickle_reference
18052         after LTO_null.
18053         (lto_tag_is_tree_code_p): Adjust.
18054         (lto_tag_is_gimple_code_p): Likewise.
18055         (lto_gimple_code_to_tag): Likewise.
18056         (lto_tag_to_gimple_code): Likewise.
18057         (lto_tree_code_to_tag): Likewise.
18058         (lto_tag_to_tree_code): Likewise.
18059         * data-streamer.h (streamer_write_hwi_in_range): Use
18060         uhwi streaming to stream the normalized range.
18061         (streamer_read_hwi_in_range): Likewise.
18063 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18065         * config/arm/arm.md (enabled_for_depr_it): New attribute.
18066         (predicable_short_it): Likewise.
18067         (predicated): Likewise.
18068         (enabled): Handle above.
18069         (define_cond_exec): Set predicated attribute to yes.
18071 2013-06-05  Mike Stump  <mikestump@comcast.net>
18073         * gdbinit.in (__FUNCTION__): Add.
18075 2013-06-05  Uros Bizjak  <ubizjak@gmail.com>
18077         * config/alpha/alpha.c (alpha_emit_conditional_move): Swap all
18078         GE, GT, GEU and GTU compares, modulo DImode compares with zero.
18080 2013-06-05  Jan Hubicka  <jh@suse.cz>
18082         * varasm.c (mark_decl_referenced): Revert the removal until targets
18083         are fixed.
18085 2013-06-05  David Edelsohn  <dje.gcc@gmail.com>
18087         * config/rs6000/rs6000.c (print_operand, 'z'): Use DECL_PRESERVE_P
18088         instead of mark_decl_referenced.
18090 2013-06-05  Jan Hubicka  <jh@suse.cz>
18092         * cgraph.c (cgraph_remove_node): Clear forced_by_abi.
18093         (cgraph_node_cannot_be_local_p_1): Honnor symbol.forced_by_abi
18094         and symtab_used_from_object_file_p.
18095         (cgraph_make_node_local_1): Clear forced_by_abi.
18096         (cgraph_can_remove_if_no_direct_calls_and): Use forced_by_abi
18097         * cgraph.h (symtab_node_base): Add forced_by_abi.
18098         (decide_is_variable_needed): Remove.
18099         (varpool_can_remove_if_no_refs): Honnor symbol.forced_by_abi.
18100         * cgraphunit.c (cgraph_decide_is_function_needed): Rename to ..
18101         (decide_is_symbol_needed): ... this one; handle symbols in general;
18102         always analyze virtuals; honnor forced_by_abi.
18103         (cgraph_finalize_function): Update.
18104         (varpool_finalize_decl): Update.
18105         (symbol_defined_and_needed): Remove.
18106         (analyze_functions): Update.
18107         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
18108         output_refs, input_overwrite_node): Handle forced_by_abi.
18109         * ipa.c (cgraph_address_taken_from_non_vtable_p): Rename to ...
18110         (address_taken_from_non_vtable_p): ... this one.
18111         (comdat_can_be_unshared_p_1): New function.
18112         (cgraph_comdat_can_be_unshared_p): Rename to ...
18113         (comdat_can_be_unshared_p): ... this one; handle symbols in general.
18114         (varpool_externally_visible_p): Use comdat_can_be_unshared_p.
18115         (function_and_variable_visibility): Clear forced_by_abi as needed.
18116         * trans-mem.c (ipa_tm_mark_forced_by_abi_node): New functoin.
18117         (ipa_tm_create_version_alias, ipa_tm_create_version): Update.
18118         * symtab.c (dump_symtab_base): Dump forced_by_abi.
18119         * varpool.c (decide_is_variable_needed): Remove.
18121 2013-06-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18123         * config/arm/arm.c (MAX_INSN_PER_IT_BLOCK): New macro.
18124         (arm_option_override): Override arm_restrict_it where appropriate.
18125         (thumb2_final_prescan_insn): Use MAX_INSN_PER_IT_BLOCK.
18126         * config/arm/arm.opt (mrestrict-it): New command-line option.
18127         * doc/invoke.texi: Document -mrestrict-it.
18129 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18131         * tsan.c (tsan_atomic_table): Make const.
18133 2013-06-05  Richard Biener  <rguenther@suse.de>
18135         * tree-streamer.c (streamer_tree_cache_insert_1): Update the
18136         index associated with the tree we are supposed to replace.
18137         * tree-streamer-out.c (pack_ts_base_value_fields): Output
18138         TREE_ASM_WRITTEN as zero for everything but SSA names.
18140 2013-06-05  David Malcolm  <dmalcolm@redhat.com>
18142         * tree-ssa-structalias.c (call_stmt_vars): Make static.
18144 2013-06-04  Jan Hubicka  <jh@suse.cz>
18146         * lto-cgraph.c (get_alias_symbol): Remove weakref sanity check.
18147         (input_node, input_varpool_node): Handle correctly external same
18148         body aliases.
18149         * ipa.c (symtab_remove_unreachable_nodes): Do not remove external
18150         nodes at ltrans stage.
18152 2013-06-04  Jan Hubicka  <jh@suse.cz>
18154         * ipa-inline.c (update_caller_keys): Fix availability test.
18155         (update_callee_keys): Likewise.
18156         * symtab.c (symtab_alias_ultimate_target): Make availaiblity logic
18157         to follow ELF standard.
18159 2013-06-04  Jürgen Urban  <JuergenUrban@gmx.de>
18161         * config.gcc (mipsr5900-*-elf*, mipsr5900el-*-elf*, mips64r5900-*-elf*)
18162         (mips64r5900el-*-elf*): New configurations.
18163         * config/mips/mips-cpus.def (r5900): New processor.
18164         * config/mips/mips-tables.opt: Regenerate.
18165         * config/mips/mips.c (mips_rtx_cost_data): Add an R5900 entry.
18166         (mips_issue_rate): Handle PROCESSOR_R5900.
18167         (mips_reorg_process_insns): Force reorder mode for the R5900.
18168         * config/mips/mips.h (TARGET_MIPS5900): Define.
18169         (ISA_HAS_CONDMOVE, ISA_HAS_PREFETCH, ISA_HAS_HILO_INTERLOCKS): Include
18170         TARGET_MIPS5900.
18171         (ISA_HAS_LOAD_DELAY, ISA_HAS_XFER_DELAY, ISA_HAS_FCMP_DELAY): Exclude
18172         TARGET_MIPS5900.
18173         * config/mips/mips.md (processor): Add r5900.
18174         (MOVECC): Disallow CCmode conditions for TARGET_MIPS5900.
18176 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18178         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Call
18179         into function to generate MOVI instruction.
18180         * config/aarch64/aarch64.c (aarch64_simd_container_mode): New function.
18181         (aarch64_preferred_simd_mode): Turn into wrapper.
18182         (aarch64_output_scalar_simd_mov_immediate): New function.
18183         * config/aarch64/aarch64-protos.h: Add prototype for above.
18185 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18187         * config/aarch64/aarch64.c (simd_immediate_info): Remove
18188         element_char member.
18189         (sizetochar): Return signed char.
18190         (aarch64_simd_valid_immediate): Remove elchar and other
18191         unnecessary variables.
18192         (aarch64_output_simd_mov_immediate): Take rtx instead of &rtx.
18193         Calculate element_char as required.
18194         * config/aarch64/aarch64-protos.h: Update and move prototype
18195         for aarch64_output_simd_mov_immediate.
18196         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
18197         Update arguments.
18199 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18201         * config/aarch64/aarch64.c (simd_immediate_info): Struct to hold
18202         information completed by aarch64_simd_valid_immediate.
18203         (aarch64_legitimate_constant_p): Update arguments.
18204         (aarch64_simd_valid_immediate): Work with struct rather than many
18205         pointers.
18206         (aarch64_simd_scalar_immediate_valid_for_move): Update arguments.
18207         (aarch64_simd_make_constant): Update arguments.
18208         (aarch64_output_simd_mov_immediate): Work with struct rather than
18209         many pointers.  Output immediate directly rather than as operand.
18210         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18211         Update prototype.
18212         * config/aarch64/constraints.md (Dn): Update arguments.
18214 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18216         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): No
18217         longer static.
18218         (aarch64_simd_immediate_valid_for_move): Remove.
18219         (aarch64_simd_scalar_immediate_valid_for_move): Update call.
18220         (aarch64_simd_make_constant): Update call.
18221         (aarch64_output_simd_mov_immediate): Update call.
18222         * config/aarch64/aarch64-protos.h (aarch64_simd_valid_immediate):
18223         Add prototype.
18224         * config/aarch64/constraints.md (Dn): Update call.
18226 2013-06-04  Ian Bolton  <ian.bolton@arm.com>
18228         * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Change
18229         return type to bool for prototype.
18230         (aarch64_legitimate_constant_p): Check for true instead of not -1.
18231         (aarch64_simd_valid_immediate): Fix up each return to return a bool.
18232         (aarch64_simd_immediate_valid_for_move): Update retval for bool.
18234 2013-06-04  Catherine Moore  <clm@codesourcery.com>
18236         * config/mips/mips.opt (meva): New.
18237         * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define __mips_eva.
18238         (ASM_SPEC): Handle -meva.
18239         * doc/invoke.texi (meva):  Document.
18241 2013-06-04  Alan Modra  <amodra@gmail.com>
18243         * config/rs6000/rs6000.c (output_toc): Correct little-endian float
18244         constant output.
18246 2013-06-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18248         * rtl.def: Add extra fourth optional field to define_cond_exec.
18249         * gensupport.c (process_one_cond_exec): Process attributes from
18250         define_cond_exec.
18251         * doc/md.texi: Document fourth field in define_cond_exec.
18253 2013-06-04  Eric Botcazou  <ebotcazou@adacore.com>
18255         * expmed.c (extract_bit_field_1): In the larger-than-a-word case, factor
18256         out the processing order as in store_bit_field_1.
18258 2013-06-04  Jan Hubicka  <jh@suse.cz>
18260         PR middle-end/57500
18261         * cgraphunit.c (cgraph_process_same_body_aliases): Create
18262         non-VAR_DECL node if it does not exist yet.
18264 2013-06-03  Richard Sandiford  <rdsandiford@googlemail.com>
18266         * config.gcc (mipsisa64sr71k-*-elf*, mipsisa64sb1-*-elf*)
18267         (mipsisa64sb1el-*-elf*, mips64-*-elf*, mips64el-*-elf*)
18268         (mips64orion-*-elf*, mips64orionel-*-elf*): Remove
18269         target_cpu_default setting.
18271 2013-06-03  Teresa Johnson  <tejohnson@google.com>
18273         * dumpfile.c (opt_info_switch_p): Change -fopt-info
18274         default to -fopt-info=optimized instead of all.
18275         * doc/invoke.texi: Ditto.
18276         * tree-vectorizer.c (vectorize_loops): Emit loop vectorization
18277         success under MSG_OPTIMIZED_LOCATIONS, and use dump_printf_loc.
18278         (execute_vect_slp): Emit BB vectorization success under
18279         MSG_OPTIMIZED_LOCATIONS.
18280         * tree-vect-slp.c (vect_slp_transform_bb): Change
18281         MSG_OPTIMIZED_LOCATIONS to MSG_NOTE.
18282         * tree-vect-loop.c (vect_transform_loop): Ditto.
18284 2013-06-03  Jason Merrill  <jason@redhat.com>
18286         PR c++/57415
18287         * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18288         Use TARGET_EXPR for C++.
18290 2013-06-03  Jakub Jelinek  <jakub@redhat.com>
18292         PR rtl-optimization/57268
18293         * sched-deps.c (sched_analyze_2): Don't flush_pending_lists
18294         if DEBUG_INSN_P (insn).
18296         Reapply
18297         2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18299         PR rtl-optimization/57268
18300         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18301         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18303 2013-06-03  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18305         * config/i386/i386.c (ix86_lea_outperforms): Fix formatting.
18306         (ix86_avoid_lea_for_addr): Likewise.
18307         (exact_dependency_1): Likewise.
18308         (ix86_adjust_cost): Likewise.
18309         (swap_top_of_ready_list): Fix formatting and !reload_completed check
18310         removed.
18311         (do_reorder_for_imul): Fix typo, formatting and
18312         !reload_completed check removed.
18313         (ix86_sched_reorder): Fix typo and formatting.
18314         (fold_builtin_cpu): Move M_INTEL_SLM at the end of processor types
18315         list.
18317 2013-06-03  Sofiane Naci  <sofiane.naci@arm.com>
18319         * config/aarch64/aarch64.md (*movdi_aarch64): Define "simd" attribute.
18321 2013-06-03  Eric Botcazou  <ebotcazou@adacore.com>
18323         * varasm.c (output_constant) <CONSTRUCTOR>: Minor formatting tweak.
18324         <STRING_CST>: Likewise.
18325         <VECTOR_CST>: Likewise.
18327 2013-06-01  Janus Weil  <janus@gcc.gnu.org>
18328             Mikael Morin  <mikael@gcc.gnu.org>
18330         * configure.ac: Add AC_HEADER_TIOCGWINSZ macro.
18331         * config.in: Regenerated.
18332         * configure: Regenerated.
18334 2013-06-01  Jan Hubicka  <jh@suse.cz>
18336         PR middle-end/57366
18337         * cgraphunit.c (compile): When weakref is not supported,
18338         set up transparent aliases before final output pass.
18339         * varasm.c (assemble_alias): Do not try to do it here.
18341 2013-06-01  Jan Hubicka  <jh@suse.cz>
18343         PR middle-end/57467
18344         * passes.c (for_per_function): Skip unanalyzed functions.
18346 2013-06-01  Jan Hubicka  <jh@suse.cz>
18348         * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Rename to ...
18349         (lto_symtab_merge_symbols_1): ... this one.
18350         (lto_symtab_merge_cgraph_nodes): Rename to ...
18351         (lto_symtab_merge_symbols): ... this one; simplify.
18352         * cgraph.c (same_body_aliases_done): Rename to ...
18353         (cpp_implicit_aliases_done): ... this one.
18354         (cgraph_create_function_alias): Update.
18355         (cgraph_same_body_alias): Update.
18356         (dump_cgraph_node): Remove alias dumping; simplify thunk dumping.
18357         (verify_edge_corresponds_to_fndecl): Simplify.
18358         * cgraph.h (symtab_node_base): Add cpp_implicit_alias, alias_target.
18359         (cgraph_node): Remove same_body_alias.
18360         (varpool_node): Remove alias_of and extra_name_alias.
18361         (same_body_aliases_done): Rename to ..
18362         (cpp_implicit_aliases_done): ... this one.
18363         (symtab_alias_ultimate_target): Add default parameter.
18364         (symtab_resolve_alias): New function.
18365         (fixup_same_cpp_alias_visibility): Declare.
18366         (cgraph_function_node): Add default parameter.
18367         (cgraph_node_asm_name): Likewise.
18368         (cgraph_function_or_thunk_node): Add default parameter; do
18369         not ICE when it is NULL.
18370         (varpool_variable_node): Likewise.
18371         * tree-emutls.c (create_emultls_var): Update.
18372         (ipa_lower_emutls): Update.
18373         * cgraphunit.c (cgraph_decide_is_function_needed): Update.
18374         (cgraph_reset_node): Reset alias info.
18375         (cgraph_finalize_function): Update.
18376         (fixup_same_cpp_alias_visibility): Move to symtab.c.
18377         (analyze_function): Simplify.
18378         (cgraph_process_same_body_aliases): Simplify.
18379         (analyze_functions): Fixup same body aliases.
18380         (handle_alias_pairs): Simplify.
18381         (assemble_thunk): Update.
18382         (assemble_thunks_and_aliases): Update.
18383         (output_weakrefs): Rewrite.
18384         * lto-cgraph.c (lto_output_node): Rewrite alias handling.
18385         (lto_output_varpool_node): Likewise.
18386         (compute_ltrans_boundary): Remve assert.
18387         (get_alias_symbol): New functoin.
18388         (input_node): Rewrite alias handling.
18389         (input_varpool_node): Likewise.
18390         * ipa-pure-const.c (propagate_pure_const): Fix formating.
18391         * ipa.c (process_references): Handle weakrefs correctly.
18392         (symtab_remove_unreachable_nodes): Likewise.
18393         * trans-mem.c (get_cg_data): Update.
18394         (ipa_tm_create_version_alias): Update.
18395         (ipa_tm_execute): Update.
18396         * symtab.c (dump_symtab_base): Dump aliases.
18397         (verify_symtab_base): Verify aliases.
18398         (symtab_node_availability): New function.
18399         (symtab_alias_ultimate_target): Simplify.
18400         (fixup_same_cpp_alias_visibility): Move here from cgraphunit.c;
18401         handle all the fixup cases.
18402         (symtab_resolve_alias): New function.
18403         * passes.c (ipa_write_summaries): Handle weakrefs.
18404         * varpool.c (varpool_analyze_node): Simplify.
18405         (assemble_aliases): Update.
18406         (varpool_create_variable_alias): Simplify.
18407         (varpool_extra_name_alias): Simplify.
18408         * lto-streamer.h (lto_symtab_merge_cgraph_nodes): Rename to...
18409         (lto_symtab_merge_symbols): ... this one.
18411 2013-06-01  Dinar Temirbulatov  <dinar@kugelworks.com>
18413         Revert
18414         PR rtl-optimization/57268
18415         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18416         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18418 2013-06-01  Tobias Burnus  <burnus@net-b.de>
18420         Partially reverted:
18421         2013-05-31  Tobias Burnus  <burnus@net-b.de>
18423         PR middle-end/57073
18424         * tree-ssa-math-opts.c (execute_cse_sincos): Move check
18425         further up.
18427 2013-05-31  Dinar Temirbulatov  <dinar@kugelworks.com>
18429         PR rtl-optimization/57268
18430         * sched-deps.c (sched_analyze_2): Flush dependence lists if
18431         the sum of the read and write lists exceeds MAX_PENDING_LIST_LENGTH.
18433 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18435         * config/rs6000/predicates.md (rs6000_cbranch_operator): Accept some
18436         unordered comparison operators when -fno-trapping-math is in effect
18437         on the e500.
18438         * config/rs6000/rs6000.c (rs6000_generate_compare): Remove dead code
18439         and implement unordered comparison operators properly on the e500.
18441 2013-05-31  Eric Botcazou  <ebotcazou@adacore.com>
18443         * simplify-rtx.c (simplify_byte_swapping_operation): Use proper macro
18444         for constant scalar integers.
18445         (simplify_relational_operation_1): Likewise.
18447 2013-05-31  Segher Boessenkool  <segher@kernel.crashing.org>
18449         * config/rs6000/rs6000-opts.h (enum processor_type): Reorder.
18450         * config/rs6000/rs6000.md (cpu): Reorder.  Split long line.
18451         Fix comment.
18453 2013-05-31  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18454             Igor Zamyatin  <igor.zamyatin@intel.com>
18456         Silvermont (SLM) architecture performance tuning.
18457         * config/i386/i386.h (enum ix86_tune_indices): Add
18458         X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS.
18459         (TARGET_SPLIT_MEM_OPND_FOR_FP_CONVERTS): New define.
18461         * config/i386/i386.c (initial_ix86_tune_features)
18462         <X86_TUNE_SPLIT_MEM_OPND_FOR_FP_CONVERTS>: Initialize.
18463         (ix86_lea_outperforms): Handle Silvermont tuning.
18464         (ix86_avoid_lea_for_add): Add new argument to ix86_lea_outperforms
18465         call.
18466         (ix86_use_lea_for_mov): Likewise.
18467         (ix86_avoid_lea_for_addr): Likewise.
18468         (ix86_lea_for_add_ok): Likewise.
18469         (exact_dependency_1): New function.
18470         (exact_store_load_dependency): Likewise.
18471         (ix86_adjust_cost): Handle Silvermont tuning.
18472         (do_reoder_for_imul): Likewise.
18473         (swap_top_of_ready_list): New function.
18474         (ix86_sched_reorder): Changed to handle Silvermont tuning.
18476         * config/i386/i386.md (peepholes that split memory operand in fp
18477         converts): New.
18479 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18481         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
18482         Remove un-necessary braces.
18484 2013-05-31  Marcus Shawcroft  <marcus.shawcroft@arm.com>
18486         * config/aarch64/aarch64.c (aarch64_classify_symbol):
18487         Use SYMBOL_TINY_ABSOLUTE for AARCH64_CMODEL_TINY_PIC.
18489 2013-05-31  Tobias Burnus  <burnus@net-b.de>
18491         PR middle-end/57073
18492         * tree-ssa-math-opts.c (execute_cse_sincos): Move check further up.
18494 2013-05-31  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
18496         PR target/56315
18497         * config/arm/arm.c (const_ok_for_dimode_op): Handle IOR.
18498         * config/arm/arm.md (*iordi3_insn): Change to insn_and_split.
18499         * config/arm/neon.md (iordi3_neon): Remove.
18500         (neon_vorr<mode>): Generate iordi3 instead of iordi3_neon.
18501         * config/arm/predicates.md (imm_for_neon_logic_operand):
18502         Move to earlier in the file.
18503         (neon_logic_op2): Likewise.
18504         (arm_iordi_operand_neon): New predicate.
18506 2013-05-31  Richard Biener  <rguenther@suse.de>
18508         PR tree-optimization/57478
18509         PR tree-optimization/57453
18510         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Uses in PHI nodes
18511         are life as well.
18513 2013-05-31  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
18515         * config/rl78/rl78.md (mulqi3,mulhi3): New define_expands.
18516         (*mulqi3_rl78,*mulhi3_rl78,*mulhi3_g13): New define_insns.
18518 2013-05-30  Tobias Burnus  <burnus@net-b.de>
18519             Thomas Koenig  <tkoenig@gcc.gnu.org>
18521         PR middle-end/57073
18522         * tree-ssa-math-opts.c (execute_cse_sincos): Optimize
18523         powi (-1.0, k) to (k & 1) ? -1.0 : 1.0.
18525 2013-05-30  Steven Bosscher  <steven@gcc.gnu.org>
18527         * rtlanal.c (tablejump_p): Expect table and label to be adjacent.
18529 2013-05-30  Vladimir Makarov  <vmakarov@redhat.com>
18531         * target.def (register_usage_leveling_p): New hook.
18532         * targhooks.c (default_register_usage_leveling_p): New.
18533         * targhooks.h (default_register_usage_leveling_p): New prototype.
18534         * lra-assigns.c (register_usage_leveling_p): Use the hook.
18535         * doc/tm.texi.in (TARGET_REGISTER_USAGE_LEVELING_P): New hook.
18536         * doc/tm.texi: Update.
18537         * config/i386/i386.c (TARGET_REGISTER_USAGE_LEVELING_P): Define.
18539 2013-05-30  Ian Bolton  <ian.bolton@arm.com>
18541         * config/aarch64/aarch64.md (insv<mode>): New define_expand.
18542         (*insv_reg<mode>): New define_insn.
18544 2013-05-30  Joern Rennecke  <joern.rennecke@embecosm.com>
18546         PR rtl-optimization/57439
18547         * postreload.c (move2add_valid_value_p): Check that we have
18548         a zero subreg_regno_offset when accessing the register in
18549         the requested mode.
18551 2013-05-30  Yuri Rumyantsev  <yuri.s.rumyantsev@intel.com>
18552             Igor Zamyatin  <igor.zamyatin@intel.com>
18554         Silvermont (SLM) architecture pipeline model, tuning and
18555         insn selection.
18556         * config.gcc: Add slm config options and target.
18558         * config/i386/slm.md: New.
18560         * config/i386/driver-i386.c (host_detect_local_cpu): Check movbe.
18562         * config/i386/i386-c.c (ix86_target_macros_internal):  New case
18563         PROCESSOR_SLM.
18564         (ix86_target_macros_internal): Likewise.
18566         * config/i386/i386.c (slm_cost): New cost.
18567         (m_SLM): New macro flag.
18568         (initial_ix86_tune_features): Set m_SLM.
18569         (x86_accumulate_outgoing_args): Likewise.
18570         (x86_arch_always_fancy_math_387): Likewise.
18571         (processor_target_table): Add slm cost.
18572         (cpu_names): Add slm cpu name.
18573         (x86_option_override_internal): Set SLM ISA.
18574         (ix86_issue_rate): New case PROCESSOR_SLM.
18575         (ia32_multipass_dfa_lookahead): Likewise.
18576         (fold_builtin_cpu): Add slm.
18578         * config/i386/i386.h (TARGET_SLM): New target macro.
18579         (target_cpu_default): Add TARGET_CPU_DEFAULT_slm.
18580         (processor_type): Add PROCESSOR_SLM.
18582         * config/i386/i386.md (cpu): Add new value "slm".
18583         (slm.md): Include slm.md.
18585 2013-05-30  Bernd Schmidt  <bernds@codesourcery.com>
18586             Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18588         * config/arm/arm-protos.h: Add and update function protos.
18589         * config/arm/arm.c (use_simple_return_p): New added.
18590         (thumb2_expand_return): Check simple_return flag.
18591         * config/arm/arm.md: Add simple_return and conditional simple_return.
18592         * config/arm/iterators.md: Add iterator for return and simple_return.
18594 2013-05-30  Zhenqiang Chen  <zhenqiang.chen@linaro.org>
18596         * config/arm/arm.c (arm_add_cfa_adjust_cfa_note): New added.
18597         (arm_emit_multi_reg_pop): Add REG_CFA_ADJUST_CFA notes.
18598         (arm_emit_vfp_multi_reg_pop): Likewise.
18599         (thumb2_emit_ldrd_pop): Likewise.
18600         (arm_expand_epilogue): Add misc REG_CFA notes.
18601         (arm_unwind_emit): Skip REG_CFA_ADJUST_CFA and REG_CFA_RESTORE.
18603 2013-05-29  Lawrence Crowl  <crowl@google.com>
18605         * config/arm/t-arm: Update for below.
18607         * config/arm/arm.c (arm_libcall_uses_aapcs_base::libcall_htab):
18608         Change type to hash_table.  Update dependent calls and types.
18610         * config/i386/t-cygming: Update for below.
18612         * config/i386/t-interix: Update for below.
18614         * config/i386/winnt.c (i386_pe_section_type_flags::htab):
18615         Change type to hash_table.  Update dependent calls and types.
18616         (i386_find_on_wrapper_list::wrappers): Likewise.
18618         * config/ia64/t-ia64: Update for below.
18620         * config/ia64/ia64.c (bundle_state_table):
18621         Change type to hash_table.  Update dependent calls and types.
18623         * config/mips/mips.c (mips_reorg_process_insns::htab):
18624         Change type to hash_table.  Update dependent calls and types.
18626         * config/sol2.c (solaris_comdat_htab):
18627         Change type to hash_table.  Update dependent calls and types.
18629         * config/t-sol2: Update for above.
18631 2013-05-29  Teresa Johnson  <tejohnson@google.com>
18633         * passes.c (dump_passes): Use FOR_EACH_FUNCTION since
18634         functions are not yet marked as defined.
18636 2013-05-29  Michael Meissner  <meissner@linux.vnet.ibm.com>
18637             Pat Haugen  <pthaugen@us.ibm.com>
18638             Peter Bergner  <bergner@vnet.ibm.com>
18640         * config/rs6000/vector.md (VEC_I): Add support for new power8 V2DI
18641         instructions.
18642         (VEC_A): Likewise.
18643         (VEC_C): Likewise.
18644         (vrotl<mode>3): Likewise.
18645         (vashl<mode>3): Likewise.
18646         (vlshr<mode>3): Likewise.
18647         (vashr<mode>3): Likewise.
18649         * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18650         support for power8 V2DI builtins.
18652         * config/rs6000/rs6000-builtin.def (abs_v2di): Add support for
18653         power8 V2DI builtins.
18654         (vupkhsw): Likewise.
18655         (vupklsw): Likewise.
18656         (vaddudm): Likewise.
18657         (vminsd): Likewise.
18658         (vmaxsd): Likewise.
18659         (vminud): Likewise.
18660         (vmaxud): Likewise.
18661         (vpkudum): Likewise.
18662         (vpksdss): Likewise.
18663         (vpkudus): Likewise.
18664         (vpksdus): Likewise.
18665         (vrld): Likewise.
18666         (vsld): Likewise.
18667         (vsrd): Likewise.
18668         (vsrad): Likewise.
18669         (vsubudm): Likewise.
18670         (vcmpequd): Likewise.
18671         (vcmpgtsd): Likewise.
18672         (vcmpgtud): Likewise.
18673         (vcmpequd_p): Likewise.
18674         (vcmpgtsd_p): Likewise.
18675         (vcmpgtud_p): Likewise.
18676         (vupkhsw): Likewise.
18677         (vupklsw): Likewise.
18678         (vaddudm): Likewise.
18679         (vmaxsd): Likewise.
18680         (vmaxud): Likewise.
18681         (vminsd): Likewise.
18682         (vminud): Likewise.
18683         (vpksdss): Likewise.
18684         (vpksdus): Likewise.
18685         (vpkudum): Likewise.
18686         (vpkudus): Likewise.
18687         (vrld): Likewise.
18688         (vsld): Likewise.
18689         (vsrad): Likewise.
18690         (vsrd): Likewise.
18691         (vsubudm): Likewise.
18693         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Add
18694         support for power8 V2DI instructions.
18696         * config/rs6000/altivec.md (UNSPEC_VPKUHUM): Add support for
18697         power8 V2DI instructions.  Combine pack and unpack insns to use an
18698         iterator for each mode.  Check whether a particular mode supports
18699         Altivec instructions instead of just checking TARGET_ALTIVEC.
18700         (UNSPEC_VPKUWUM): Likewise.
18701         (UNSPEC_VPKSHSS): Likewise.
18702         (UNSPEC_VPKSWSS): Likewise.
18703         (UNSPEC_VPKUHUS): Likewise.
18704         (UNSPEC_VPKSHUS): Likewise.
18705         (UNSPEC_VPKUWUS): Likewise.
18706         (UNSPEC_VPKSWUS): Likewise.
18707         (UNSPEC_VPACK_SIGN_SIGN_SAT): Likewise.
18708         (UNSPEC_VPACK_SIGN_UNS_SAT): Likewise.
18709         (UNSPEC_VPACK_UNS_UNS_SAT): Likewise.
18710         (UNSPEC_VPACK_UNS_UNS_MOD): Likewise.
18711         (UNSPEC_VUPKHSB): Likewise.
18712         (UNSPEC_VUNPACK_HI_SIGN): Likewise.
18713         (UNSPEC_VUNPACK_LO_SIGN): Likewise.
18714         (UNSPEC_VUPKHSH): Likewise.
18715         (UNSPEC_VUPKLSB): Likewise.
18716         (UNSPEC_VUPKLSH): Likewise.
18717         (VI2): Likewise.
18718         (VI_char): Likewise.
18719         (VI_scalar): Likewise.
18720         (VI_unit): Likewise.
18721         (VP): Likewise.
18722         (VP_small): Likewise.
18723         (VP_small_lc): Likewise.
18724         (VU_char): Likewise.
18725         (add<mode>3): Likewise.
18726         (altivec_vaddcuw): Likewise.
18727         (altivec_vaddu<VI_char>s): Likewise.
18728         (altivec_vadds<VI_char>s): Likewise.
18729         (sub<mode>3): Likewise.
18730         (altivec_vsubcuw): Likewise.
18731         (altivec_vsubu<VI_char>s): Likewise.
18732         (altivec_vsubs<VI_char>s): Likewise.
18733         (altivec_vavgs<VI_char>): Likewise.
18734         (altivec_vcmpbfp): Likewise.
18735         (altivec_eq<mode>): Likewise.
18736         (altivec_gt<mode>): Likewise.
18737         (altivec_gtu<mode>): Likewise.
18738         (umax<mode>3): Likewise.
18739         (smax<mode>3): Likewise.
18740         (umin<mode>3): Likewise.
18741         (smin<mode>3): Likewise.
18742         (altivec_vpkuhum): Likewise.
18743         (altivec_vpkuwum): Likewise.
18744         (altivec_vpkshss): Likewise.
18745         (altivec_vpkswss): Likewise.
18746         (altivec_vpkuhus): Likewise.
18747         (altivec_vpkshus): Likewise.
18748         (altivec_vpkuwus): Likewise.
18749         (altivec_vpkswus): Likewise.
18750         (altivec_vpks<VI_char>ss): Likewise.
18751         (altivec_vpks<VI_char>us): Likewise.
18752         (altivec_vpku<VI_char>us): Likewise.
18753         (altivec_vpku<VI_char>um): Likewise.
18754         (altivec_vrl<VI_char>): Likewise.
18755         (altivec_vsl<VI_char>): Likewise.
18756         (altivec_vsr<VI_char>): Likewise.
18757         (altivec_vsra<VI_char>): Likewise.
18758         (altivec_vsldoi_<mode>): Likewise.
18759         (altivec_vupkhsb): Likewise.
18760         (altivec_vupkhs<VU_char>): Likewise.
18761         (altivec_vupkls<VU_char>): Likewise.
18762         (altivec_vupkhsh): Likewise.
18763         (altivec_vupklsb): Likewise.
18764         (altivec_vupklsh): Likewise.
18765         (altivec_vcmpequ<VI_char>_p): Likewise.
18766         (altivec_vcmpgts<VI_char>_p): Likewise.
18767         (altivec_vcmpgtu<VI_char>_p): Likewise.
18768         (abs<mode>2): Likewise.
18769         (vec_unpacks_hi_v16qi): Likewise.
18770         (vec_unpacks_hi_v8hi): Likewise.
18771         (vec_unpacks_lo_v16qi): Likewise.
18772         (vec_unpacks_hi_<VP_small_lc>): Likewise.
18773         (vec_unpacks_lo_v8hi): Likewise.
18774         (vec_unpacks_lo_<VP_small_lc>): Likewise.
18775         (vec_pack_trunc_v8h): Likewise.
18776         (vec_pack_trunc_v4si): Likewise.
18777         (vec_pack_trunc_<mode>): Likewise.
18779         * config/rs6000/altivec.h (vec_vaddudm): Add defines for power8
18780         V2DI builtins.
18781         (vec_vmaxsd): Likewise.
18782         (vec_vmaxud): Likewise.
18783         (vec_vminsd): Likewise.
18784         (vec_vminud): Likewise.
18785         (vec_vpksdss): Likewise.
18786         (vec_vpksdus): Likewise.
18787         (vec_vpkudum): Likewise.
18788         (vec_vpkudus): Likewise.
18789         (vec_vrld): Likewise.
18790         (vec_vsld): Likewise.
18791         (vec_vsrad): Likewise.
18792         (vec_vsrd): Likewise.
18793         (vec_vsubudm): Likewise.
18794         (vec_vupkhsw): Likewise.
18795         (vec_vupklsw): Likewise.
18797 2013-05-29  Jan Hubicka  <jh@suse.cz>
18799         * cgraph.h (symtab_node_base): Add definition, alias and analyzed
18800         flags; reorder rest of fields in more consistent way.
18801         (varpool_node): Remove analyzed, finalized and alias.
18802         (cgraph_ndoe): Likewise.
18803         (symtab_alias_ultimate_target): New function.
18804         (cgraph_function_node): Move offline.
18805         (cgraph_reset_node): Declare.
18806         (cgraph_comdat_can_be_unshared_p): Remove.
18807         (varpool_remove_initializer): Declare.
18808         (varpool_first_defined_variable, varpool_next_defined_variable
18809         cgraph_first_defined_function, cgraph_next_defined_function): Update.
18810         (cgraph_function_with_gimple_body_p): Update.
18811         (varpool_all_refs_explicit_p): Update.
18812         (symtab_alias_target): New function.
18813         (cgraph_alias_aliased_node, varpool_alias_aliased_node): Rename to ...
18814         (cgraph_alias_target, varpool_alias_target): .. this one; simplify.
18815         (cgraph_function_or_thunk_node): Simplify using
18816         symtab_alias_ultimate_target.
18817         (varpool_variable_node): Likewise.
18818         * cgraph.c (cgraph_create_function_alias): Update.
18819         (cgraph_add_thunk): Update.
18820         (cgraph_remove_node): Update.
18821         (dump_cgraph_node): Do not dump removed flags.
18822         (cgraph_function_body_availability): Update.
18823         (cgraph_propagate_frequency): Update.
18824         (verify_cgraph_node): Check sanity of local flag.
18825         (cgraph_function_node): Move here from cgraph.h; revamp for
18826         cgraph_function_or_thunk_node.
18827         * lto-symtab.c (lto_varpool_replace_node): Update.
18828         (lto_symtab_resolve_can_prevail_p): Update.
18829         (lto_symtab_merge_cgraph_nodes): Update.
18830         * ipa-cp.c (determine_versionability, initialize_node_lattices,
18831         propagate_constants_accross_call, devirtualization_time_bonus,
18832         ipcp_propagate_stage): Update.
18833         * tree-emutls.c (create_emultls_var, ipa_lower_emutls): Update.
18834         * ipa-inline-transform.c (clone_inlined_nodes,
18835         preserve_function_body_p): Update.
18836         * ipa-reference.c (propagate): Update.
18837         (write_node_summary_p): Update.
18838         * toplev.c (wrapup_global_declaration_2): Update.
18839         * cgraphunit.c (cgraph_analyze_function): Rename to ...
18840         (analyze_function) ... this one.
18841         (cgraph_process_new_functions): Update.
18842         (cgraph_reset_node): Export.
18843         (cgraph_finalize_function): Update.
18844         (cgraph_add_new_function): Update.
18845         (process_function_and_variable_attributes): Update.
18846         (varpool_finalize_decl): Update.
18847         (symbol_finalized): Remove.
18848         (symbol_finalized_and_needed): Rename to ...
18849         (symbol_defined_and_needed): ... update.
18850         (cgraph_analyze_functions): Update.
18851         (handle_alias_pairs): Update.
18852         (mark_functions_to_output): Update.
18853         (assemble_thunk): Update.
18854         (output_in_order): Update.
18855         (output_weakrefs): Update.
18856         (finalize_compilation_unit): Update.
18857         * lto-cgraph.c (reachable_from_other_partition_p, lto_output_node,
18858         lto_output_varpool_node, compute_ltrans_boundary, input_overwrite_node,
18859         input_node, input_varpool_node): Update.
18860         * dbxout.c (dbxout_expand_expr): Update.
18861         * cgraphclones.c (cgraph_clone_node): Update.
18862         (cgraph_copy_node_for_versioning): Update.
18863         (cgraph_materialize_clone): Update.
18864         (cgraph_materialize_all_clones): Update.
18865         * ipa-pure-const.c (analyze_function, pure_const_write_summary,
18866         propagate_pure_const, propagate_nothrow): Update.
18867         * lto-streamer-out.c (lto_output, write_symbol): Update.
18868         * ipa-utils.c (ipa_reverse_postorder): Update.
18869         * ipa-inline.c (can_inline_edge_p): Update.
18870         (update_caller_keys, ipa_inline): Update.
18871         * dwarf2out.c (reference_to_unused,
18872         premark_types_used_by_global_vars_helper): Update.
18873         * tree-eh.c (tree_could_trap_p): Update.
18874         * ipa-split.c (consider_split, execute_split_functions): Update.
18875         * ipa.c (cgraph_non_local_node_p_1, cgraph_local_node_p,
18876         has_addr_references_p): Update; move ahead in file for better
18877         readability.
18878         (process_references): Simplify.
18879         (symtab_remove_unreachable_nodes): Update; cleanup way function/var
18880         bodies are removed.
18881         (cgraph_comdat_can_be_unshared_p): Make static.
18882         (cgraph_externally_visible_p): Update.
18883         (varpool_externally_visible_p): Update.
18884         (function_and_variable_visibility): Update.
18885         * trans-mem.c (get_cg_data, ipa_tm_mayenterirr_function,
18886         ipa_tm_mark_force_output_node): Update.
18887         * ipa-inline-analysis.c (dump_inline_summary, initialize_inline_failed,
18888         estimate_edge_devirt_benefit, inline_generate_summary,
18889         inline_write_summary): Update.
18890         * gimple-fold.c (can_refer_decl_in_current_unit_p): Update.
18891         * ipa-prop.c (ipa_compute_jump_functions): Update.
18892         (ipa_print_node_params, ipa_prop_read_section,
18893         ipa_update_after_lto_read, read_replacements_section): Update.
18894         * varasm.c (mark_decl_referenced): Update.
18895         (assemble_alias, dump_tm_clone_pairs): Update.
18896         * tree-inline.c (copy_bb): Update.
18897         (estimate_num_insns, optimize_inline_calls, tree_function_versioning):
18898         Update.
18899         * symtab.c (dump_symtab_base): Print new flags.
18900         (verify_symtab_base): Verify new flags.
18901         (symtab_alias_ultimate_target): New function.
18902         * tree-ssa-structalias.c (get_constraint_for_ssa_var,
18903         create_variable_info_for, associate_varinfo_to_alias, ipa_pta_execute):
18904         Update.
18905         * passes.c (ipa_write_summaries, ipa_write_optimization_summaries):
18906         Update.
18907         * i386.c (ix86_get_function_versions_dispatcher,
18908         ix86_generate_version_dispatcher_body): Update.
18909         (fold_builtin_cpu): Use varpool_add_new_variable.
18910         * varpool.c (varpool_remove_initializer): Break out from ...
18911         (varpool_remove_node): ... this one.
18912         (dump_varpool_node, varpool_node_for_asm,
18913         cgraph_variable_initializer_availability, varpool_analyze_node,
18914         varpool_assemble_decl, varpool_remove_unreferenced_decls,
18915         varpool_finalize_named_section_flags, varpool_create_variable_alias):
18916         Update.
18918 2013-05-29  Jan Hubicka  <jh@suse.cz>
18920         * passes.c (init_optimization_passes): Move OMP expansion into lowering.
18922 2013-05-29  Easwaran Raman  <eraman@google.com>
18924         PR tree-optimization/57442
18925         * tree-ssa-reassoc.c (appears_later_in_bb): Return correct value
18926         when control exits the main loop.
18928 2013-05-29  Sandeep Kumar Singh  <Sandeep.Singh2@kpitcummins.com>
18930         * rx/rx.h (TARGET_CPU_CPP_BUILTINS): Add macros for RX100, RX200,
18931         and RX600.
18932         * rx/rx.opt: Add macro for rx100 with string rx100 and value RX100.
18933         * rx/rx-opts.h (rx_cpu_types): Add new cpu type rx100.
18934         * rx/t-rx: Add rx100 under multi library matches option for nofpu
18935         option.
18937 2013-05-29  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
18939         PR tree-optimization/57441
18940         * gimple-ssa-strength-reduction.c (analyze_candidates_and_replace):
18941         Don't limit size of incr_vec to number of candidates.
18943 2013-05-29  Steve Ellcey  <sellcey@imgtec.com>
18945         * config/mips/mti-linux.h (SYSROOT_SUFFIX_SPEC): Add micromips
18946         and mips16 directories.
18947         * config/mips/t-mti-linux (MULTILIB_OPTIONS): Add micromips and mips16.
18948         (MULTILIB_DIRNAMES): Ditto.
18949         (MULTILIB_EXCEPTIONS): Add new exceptions.
18950         * config/mips/t-mti-elf (MULTILIB_OPTIONS): Add micromips.
18951         (MULTILIB_DIRNAMES): Ditto.
18952         (MULTILIB_EXCEPTIONS): Add new exceptions.
18954 2012-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18955             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18957         * config/aarch64/aarch64-protos.h (aarch64_symbol_type): Define
18958         SYMBOL_TINY_ABSOLUTE.
18959         * config/aarch64/aarch64.c (aarch64_load_symref_appropriately): Handle
18960         SYMBOL_TINY_ABSOLUTE.
18961         (aarch64_expand_mov_immediate): Likewise.
18962         (aarch64_classify_symbol): Likewise.
18963         (aarch64_mov_operand_p): Remove ATTRIBUTE_UNUSED.
18964         Permit SYMBOL_TINY_ABSOLUTE.
18965         * config/aarch64/predicates.md (aarch64_mov_operand): Permit CONST.
18967 2013-05-29  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
18968             Marcus Shawcroft  <marcus.shawcroft@arm.com>
18970         * config/aarch64/aarch64.c (aarch64_classify_symbol): Remove comment.
18971         Refactor if/switch.  Replace gcc_assert with if.
18973 2013-05-29  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
18975         * config/i386/i386.c (initial_ix86_tune_features): Enable
18976         FP Reassociation for AMD bdver1 and bdver2.
18978 2013-05-29  Martin Jambor  <mjambor@suse.cz>
18980         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REF, REALPART_EXPR
18981         and IMAGPART_EXPR do not occur within other handled_components.
18983 2013-05-29  Richard Biener  <rguenther@suse.de>
18985         * tree-vect-slp.c (vect_bb_slp_scalar_cost): Guard vinfo
18986         access on whether the use is in the BB we currently try to
18987         vectorize.
18988         (vect_bb_vectorization_profitable_p): Pass the BB we currently
18989         vectorize to vect_bb_slp_scalar_cost.
18991 2013-05-29  Richard Biener  <rguenther@suse.de>
18993         * tree-vect-slp.c (vect_bb_slp_scalar_cost): New function
18994         computing scalar cost offsetted by stmts that are kept live
18995         by scalar uses.
18996         (vect_bb_vectorization_profitable_p): Use vect_bb_slp_scalar_cost
18997         for computation of scalar cost.
18999 2013-05-28  Steve Ellcey  <sellcey@mips.com>
19001         * config/mips/mips-cpus.def (mips32r2): Change processor type.
19003 2013-05-28  Balaji V. Iyer  <balaji.v.iyer@intel.com>
19005         * doc/extend.texi (C Extensions): Added documentation about Cilk Plus
19006         array notation built-in reduction functions.
19007         * doc/passes.texi (Passes): Added documentation about changes done
19008         for Cilk Plus.
19009         * doc/invoke.texi (C Dialect Options): Added documentation about
19010         the -fcilkplus flag.
19011         * Makefile.in (C_COMMON_OBJS): Added c-family/array-notation-common.o.
19012         (BUILTINS_DEF): Depend on cilkplus.def.
19013         * builtins.def: Include cilkplus.def.  Define DEF_CILKPLUS_BUILTIN.
19014         * builtin-types.def: Define BT_FN_INT_PTR_PTR_PTR.
19015         * cilkplus.def: New file.
19017 2013-05-28  Joern Rennecke  <joern.rennecke@embecosm.com>
19019         PR rtl-optimization/57439
19020         * postreload.c (move2add_use_add2_insn): Use gen_lowpart_common.
19022 2013-05-28  Easwaran Raman  <eraman@google.com>
19024         PR tree-optimization/57337
19025         * tree-ssa-reassoc.c (appears_later_in_bb): New function.
19026         (find_insert_point): Correctly identify the insertion point
19027         when two statements with the same UID is compared.
19029 2013-05-28  Richard Biener  <rguenther@suse.de>
19031         PR tree-optimization/56787
19032         * tree-vect-data-refs.c (vect_analyze_data_refs): Drop clobbers
19033         from the list of data references.
19034         * tree-vect-loop.c (vect_determine_vectorization_factor): Skip
19035         clobbers.
19036         (vect_analyze_loop_operations): Likewise.
19037         (vect_transform_loop): Remove clobbers.
19039 2013-05-28  Martin Jambor  <mjambor@suse.cz>
19041         * tree-cfg.c (verify_expr): Verify that BIT_FIELD_REFs, IMAGPART_EXPRs
19042         and REALPART_EXPRs have scalar type.
19044 2013-05-28  Richard Biener  <rguenther@suse.de>
19046         PR tree-optimization/57411
19047         * tree-ssa-copy.c (may_propagate_copy): Cannot propagate
19048         virtual operands.
19049         * tree-ssa-dom.c (eliminate_const_or_copy): Special-case
19050         virtual operand propagation.
19052 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
19054         * config/sparc/sparc.c (sparc_expand_vec_perm_bmask): Use %g0 as
19055         destination register for bmasksi_vis.
19056         (vector_init_bshuffle): Likewise.
19057         * config/sparc/sparc.md (vec_perm_constv8qi): Likewise.
19059 2013-05-28  Eric Botcazou  <ebotcazou@adacore.com>
19061         * doc/invoke.texi (SPARC Options): Document -mfix-ut699.
19062         * builtins.c (expand_builtin_mathfn) <BUILT_IN_SQRT>: Try to widen the
19063         mode if the instruction isn't available in the original mode.
19064         * config/sparc/sparc.opt (mfix-ut699): New option.
19065         * config/sparc/sparc.md (muldf3_extend): Disable if -mfix-ut699.
19066         (divdf3): Turn into expander.
19067         (divdf3_nofix): New insn.
19068         (divdf3_fix): Likewise.
19069         (divsf3): Disable if -mfix-ut699.
19070         (sqrtdf2): Turn into expander.
19071         (sqrtdf2_nofix): New insn.
19072         (sqrtdf2_fix): Likewise.
19073         (sqrtsf2): Disable if -mfix-ut699.
19075 2013-05-27  Richard Biener  <rguenther@suse.de>
19077         PR middle-end/57412
19078         * omp-low.c (expand_omp_atomic_pipeline): Use the correct latch
19079         block for the new loop.
19081 2013-05-27  Richard Biener  <rguenther@suse.de>
19083         PR tree-optimization/57343
19084         * tree-ssa-loop-niter.c (number_of_iterations_ne_max): Do not
19085         use multiple_of_p if not TYPE_OVERFLOW_UNDEFINED.
19086         (number_of_iterations_cond): Do not build the folded tree.
19088 2013-05-27  Richard Biener  <rguenther@suse.de>
19090         Revert
19091         PR middle-end/57381
19092         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19093         OEP_CONSTANT_ADDRESS_OF retained.
19095         PR tree-optimization/57417
19096         * tree-ssa-sccvn.c (vn_reference_fold_indirect): Fix test
19097         for unchanged base.
19098         (set_ssa_val_to): Compare addresses using
19099         get_addr_base_and_unit_offset.
19101 2013-05-27  Joern Rennecke  <joern.rennecke@embecosm.com>
19103         PR rtl-optimization/56833
19104         * postreload.c (move2add_record_mode): New function.
19105         (move2add_record_sym_value, move2add_valid_value_p): Likewise.
19106         (move2add_use_add2_insn): Use move2add_record_sym_value.
19107         (move2add_use_add3_insn): Likewise.
19108         (reload_cse_move2add): Use move2add_valid_value_p and
19109         move2add_record_mode.  Invalidate call-clobbered and REG_INC
19110         affected regs by setting reg_mode to VOIDmode.
19111         (move2add_note_store): Don't pretend the inside of a SUBREG is
19112         the actual destination.  Invalidate single/leading registers by
19113         setting reg_mode to VOIDmode.
19114         Use move2add_record_sym_value, move2add_valid_value_p and
19115         move2add_record_mode.
19117 2013-05-27  Richard Biener  <rguenther@suse.de>
19119         PR tree-optimization/57396
19120         * tree-affine.c (double_int_constant_multiple_p): Properly
19121         return false for val == 0 and div != 0.
19123 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19125         * config/mips/mips.h: Use #elif in preprocessor conditions.
19127 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19129         PR target/53916
19130         * config/mips/constraints.md (kl): New constraint.
19131         * config/mips/mips.md (divmod<mode>4, udivmod<mode>4): Delete.
19132         (divmod<mode>4_internal): Rename to divmod<mode>4.  Use "kl" as the
19133         constraint for operand 0.  Split after CSE for MIPS16.  Emit a move
19134         from LO for MIPS16.
19135         (udivmod<mode>4_internal): Likewise udivmod<mode>4.
19137 2013-05-25  Richard Sandiford  <rdsandiford@googlemail.com>
19139         PR target/55777
19140         * config/mips/mips.c (mips_can_inline_p): New function.
19141         (TARGET_CAN_INLINE_P): Define.
19143 2013-05-25  Steven Bosscher  <steven@gcc.gnu.org>
19145         * sched-int.h (ds_t, dw_t): Make unsigned int.
19146         Fix documentation that describes how all the ds_t bits are used.
19147         Reserve the last bit for delayed-branch scheduling.
19148         (BITS_PER_DEP_STATUS): Move to ds_t typedef.
19149         (BITS_PER_DEP_WEAK): Fix definition and documentation.
19150         (gen_dep_weak_1): Remove prototype.
19151         * sched-deps.c (get_dep_weak_1): Make static.
19152         * target.def (speculate_insn, needs_block_p, gen_spec_check,
19153         get_insn_spec_ds, get_insn_checked_ds): Adjust hook prototypes.
19154         * doc/tm.texi: Regenerate.
19155         * config/ia64/ia64.c (ia64_needs_block_p): Update prototype.
19157 2013-05-24  Steven Bosscher  <steven@gcc.gnu.org>
19159         PR debug/56950
19160         * haifa-sched.c (sched_extend_bb): Ignore DEBUG_INSNs.
19162 2013-05-24  Nathan Sidwell  <nathan@codesourcery.com>
19163             Sandra Loosemore  <sandra@codesourcery.com>
19165         * config.gcc (powerpc-*): Allow native for with-cpu.
19167 2013-05-24  Jeff Law  <law@redhat.com>
19169         PR tree-optimization/57124
19170         * tree-vrp.c (simplify_cond_using_ranges): Only simplify a
19171         conversion feeding a condition if the range has an overflow
19172         if -fstrict-overflow.  Add warnings for when we do make the
19173         transformation.
19175 2013-05-24  Dehao Chen  <dehao@google.com>
19177         * tree-cfg.c (locus_discrim_map): Fix the typo.
19178         (locus_discrim_hasher): Likewise.
19179         (locus_discrim_hasher::hash): Likewise.
19180         (locus_discrim_hasher::equal): Likewise.
19182 2013-05-24  Martin Jambor  <mjambor@suse.cz>
19184         PR tree-optimization/57294
19185         * cgraph.h (ipa_record_stmt_references): Declare.
19186         * cgraphbuild.c (ipa_record_stmt_references): New function.
19187         (build_cgraph_edges): Use ipa_record_stmt_references.
19188         (rebuild_cgraph_edges): Likewise.
19189         (cgraph_rebuild_references): Likewise.
19190         * ipa-prop.c (ipa_modify_call_arguments): Discard references
19191         associated with the old statement and build references from the
19192         newly built statements.
19193         * ipa-ref.c (ipa_remove_stmt_references): New function.
19194         * ipa-ref.h (ipa_remove_stmt_references): Declare.
19196 2013-05-24  Vladimir Makarov  <vmakarov@redhat.com>
19198         * lra-constraints.c (emit_spill_move): Use smaller mode for
19199         mem-mem moves.
19200         (check_and_process_move): Consider mem-reg moves for secondary
19201         too.
19202         (curr_insn_transform): Don't lose insns emitted before for
19203         secondary memory moves.
19204         (inherit_in_ebb): Mark defined reg.  Add usage only if it is not a
19205         reg set up in the current insn.
19207 2013-05-24  Dehao Chen  <dehao@google.com>
19209         * tree-cfg.c (locus_descrim_hasher::hash): Change discriminator
19210         hash function.
19211         (locus_descrim_hasher::equal): Likewise.
19212         (build_gimple_cfg): New discriminator assignment algorithm.
19213         (make_edges): Likewise.
19214         (next_discriminator_for_locus): Likewise.
19215         (same_line_p): Likewise.
19216         (assign_discriminators): Likewise.
19217         (make_cond_expr_edges): Likewise.
19218         (make_gimple_switch_edges): Likewise.
19219         (make_goto_expr_edges): Likewise.
19220         (make_gimple_asm_edges): Likewise.
19222 2013-05-24  Ian Bolton  <ian.bolton@arm.com>
19224         * config/aarch64/aarch64.c (aarch64_print_operand): Change the
19225         X format specifier to only display bottom 16 bits.
19226         * config/aarch64/aarch64.md (insv_imm<mode>): Allow any size of
19227         immediate to match for operand 2, since it will be masked.
19229 2013-05-24  Richard Biener  <rguenther@suse.de>
19231         PR tree-optimization/57287
19232         * tree-ssa-uninit.c (compute_uninit_opnds_pos): Disregard
19233         all SSA names that occur in abnormal PHIs.
19235 2013-05-24  Alexander Ivchenko  <alexander.ivchenko@intel.com>
19237         PR tree-ssa/57385
19238         * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Check
19239         that index is not negative.
19241 2013-05-24  Eric Botcazou  <ebotcazou@adacore.com>
19243         PR rtl-optimization/55177
19244         * simplify-rtx.c (simplify_unary_operation_1) <NOT>: Deal with BSWAP.
19245         (simplify_byte_swapping_operation): New.
19246         (simplify_binary_operation_1): Call it for AND, IOR and XOR.
19247         (simplify_relational_operation_1): Deal with BSWAP.
19249 2013-05-23  Richard Henderson  <rth@redhat.com>
19251         PR target/56742
19252         * config/i386/i386.c (ix86_seh_fixup_eh_fallthru): New.
19253         (ix86_reorg): Call it.
19255 2013-05-23  Uros Bizjak  <ubizjak@gmail.com>
19257         PR target/57379
19258         * config/alpha/alpha.md (unspec): Add UNSPEC_XFLT_COMPARE.
19259         * config/alpha/alpha.c (alpha_emit_xfloating_compare): Construct
19260         REG_EQUAL note as UNSPEC_XFLT_COMPARE unspec.
19262 2013-05-23  Christian Bruel  <christian.bruel@st.com>
19264         PR debug/57351
19265         * config/arm/arm.c (arm_dwarf_register_span): Do not use dbx number.
19267 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19268             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19270         * config/aarch64/aarch64.md (*movdi_aarch64): Replace Usa with S.
19271         * config/aarch64/constraints.md (Usa): Remove.
19272         * doc/md.texi (AArch64 Usa): Remove.
19274 2013-05-23  Chris Schlumberger-Socha  <chris.schlumberger-socha@arm.com>
19275             Marcus Shawcroft  <marcus.shawcroft@arm.com>
19277         * config/aarch64/aarch64-protos.h (aarch64_mov_operand_p): Define.
19278         * config/aarch64/aarch64.c (aarch64_mov_operand_p): Define.
19279         * config/aarch64/predicates.md (aarch64_const_address): Remove.
19280         (aarch64_mov_operand): Use aarch64_mov_operand_p.
19282 2013-05-23  Vidya Praveen  <vidyapraveen@arm.com>
19284         * config/aarch64/aarch64-simd.md (clzv4si2): Support for CLZ
19285         instruction (AdvSIMD).
19286         * config/aarch64/aarch64-builtins.c
19287         (aarch64_builtin_vectorized_function): Handler for BUILT_IN_CLZ.
19288         * config/aarch64/aarch-simd-builtins.def: Entry for CLZ.
19290 2013-05-23  Martin Jambor  <mjambor@suse.cz>
19292         PR middle-end/57347
19293         * tree.h (contains_bitfld_component_ref_p): Declare.
19294         * tree-sra.c (contains_bitfld_comp_ref_p): Move...
19295         * tree.c (contains_bitfld_component_ref_p): ...here.  Adjust its
19296         caller.
19297         * ipa-prop.c (determine_known_aggregate_parts): Check that LHS does
19298         not access a bit-field.  Assert all final offsets are byte-aligned.
19300 2013-05-23  Richard Biener  <rguenther@suse.de>
19302         PR tree-optimization/57380
19303         * tree-ssa-phiprop.c (propagate_with_phi): Do not require at
19304         least one invariant or re-used load.
19305         * passes.c (init_optimization_passes): Move pass_phiprop before
19306         pass_forwprop.
19308 2013-05-23  James Greenhalgh  <james.greenhalgh@arm.com>
19310         * config/aarch64/aarch64-simd.md
19311         (aarch64_cm<optab>di): Add clobber of CC_REGNUM to unsplit pattern.
19313 2013-05-23  Richard Biener  <rguenther@suse.de>
19315         PR middle-end/57381
19316         * fold-const.c (operand_equal_p): Compare FIELD_DECLs with
19317         OEP_CONSTANT_ADDRESS_OF retained.
19319 2013-05-23  Jakub Jelinek  <jakub@redhat.com>
19321         PR middle-end/57344
19322         * expmed.c (store_split_bit_field): If op0 is a REG or SUBREG of a REG,
19323         don't lower unit.  Handle unit not being always BITS_PER_WORD.
19325 2013-05-23  Richard Biener  <rguenther@suse.de>
19327         PR rtl-optimization/57341
19328         * ira.c (validate_equiv_mem_from_store): Use anti_dependence
19329         instead of true_dependence.
19331 2013-05-22  David Malcolm  <dmalcolm@redhat.com>
19333         * bb-reorder.c (branch_threshold): Make const.
19334         (exec_threshold): Ditto.
19336 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19337             Pat Haugen  <pthaugen@us.ibm.com>
19338             Peter Bergner  <bergner@vnet.ibm.com>
19340         * doc/extend.texi (PowerPC AltiVec/VSX Built-in Functions): Add
19341         documentation for the power8 crypto builtins.
19343         * config/rs6000/t-rs6000 (MD_INCLUDES): Add crypto.md.
19345         * config/rs6000/rs6000-builtin.def (BU_P8V_AV_1): Add support
19346         macros for defining power8 builtin functions.
19347         (BU_P8V_AV_2): Likewise.
19348         (BU_P8V_AV_P): Likewise.
19349         (BU_P8V_VSX_1): Likewise.
19350         (BU_P8V_OVERLOAD_1): Likewise.
19351         (BU_P8V_OVERLOAD_2): Likewise.
19352         (BU_CRYPTO_1): Likewise.
19353         (BU_CRYPTO_2): Likewise.
19354         (BU_CRYPTO_3): Likewise.
19355         (BU_CRYPTO_OVERLOAD_1): Likewise.
19356         (BU_CRYPTO_OVERLOAD_2): Likewise.
19357         (XSCVSPDP): Fix typo, point to the correct instruction.
19358         (VCIPHER): Add power8 crypto builtins.
19359         (VCIPHERLAST): Likewise.
19360         (VNCIPHER): Likewise.
19361         (VNCIPHERLAST): Likewise.
19362         (VPMSUMB): Likewise.
19363         (VPMSUMH): Likewise.
19364         (VPMSUMW): Likewise.
19365         (VPERMXOR_V2DI): Likewise.
19366         (VPERMXOR_V4SI: Likewise.
19367         (VPERMXOR_V8HI: Likewise.
19368         (VPERMXOR_V16QI: Likewise.
19369         (VSHASIGMAW): Likewise.
19370         (VSHASIGMAD): Likewise.
19371         (VPMSUM): Likewise.
19372         (VPERMXOR): Likewise.
19373         (VSHASIGMA): Likewise.
19375         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Define
19376         __CRYPTO__ if the crypto instructions are available.
19377         (altivec_overloaded_builtins): Add support for overloaded power8
19378         builtins.
19380         * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19381         support for power8 crypto builtins.
19382         (builtin_function_type): Likewise.
19383         (altivec_init_builtins): Add support for builtins that take vector
19384         long long (V2DI) arguments.
19386         * config/rs6000/crypto.md: New file, define power8 crypto
19387         instructions.
19389 2013-05-22  Michael Meissner  <meissner@linux.vnet.ibm.com>
19390             Pat Haugen  <pthaugen@us.ibm.com>
19391             Peter Bergner  <bergner@vnet.ibm.com>
19393         * doc/invoke.texi (Option Summary): Add power8 options.
19394         (RS/6000 and PowerPC Options): Likewise.
19396         * doc/md.texi (PowerPC and IBM RS6000 constraints): Update to use
19397         constraints.md instead of rs6000.h.  Reorder w* constraints.  Add
19398         wm, wn, wr documentation.
19400         * config/rs6000/constraints.md (wm): New constraint for VSX
19401         registers if direct move instructions are enabled.
19402         (wn): New constraint for no registers.
19403         (wq): New constraint for quad word even GPR registers.
19404         (wr): New constraint if 64-bit instructions are enabled.
19405         (wv): New constraint if power8 vector instructions are enabled.
19406         (wQ): New constraint for quad word memory locations.
19408         * config/rs6000/predicates.md (const_0_to_15_operand): New
19409         constraint for 0..15 for crypto instructions.
19410         (gpc_reg_operand): If VSX allow registers in VSX registers as well
19411         as GPR and floating point registers.
19412         (int_reg_operand): New predicate to match only GPR registers.
19413         (base_reg_operand): New predicate to match base registers.
19414         (quad_int_reg_operand): New predicate to match even GPR registers
19415         for quad memory operations.
19416         (vsx_reg_or_cint_operand): New predicate to allow vector logical
19417         operations in both GPR and VSX registers.
19418         (quad_memory_operand): New predicate for quad memory operations.
19419         (reg_or_indexed_operand): New predicate for direct move support.
19421         * config/rs6000/rs6000-cpus.def (ISA_2_5_MASKS_EMBEDDED):
19422         Inherit from ISA_2_4_MASKS, not ISA_2_2_MASKS.
19423         (ISA_2_7_MASKS_SERVER): New mask for ISA 2.07 (i.e. power8).
19424         (POWERPC_MASKS): Add power8 options.
19425         (power8 cpu): Use ISA_2_7_MASKS_SERVER instead of specifying the
19426         various options.
19428         * config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
19429         Define _ARCH_PWR8 and __POWER8_VECTOR__ for power8.
19431         * config/rs6000/rs6000.opt (-mvsx-timode): Add documentation.
19432         (-mpower8-fusion): New power8 options.
19433         (-mpower8-fusion-sign): Likewise.
19434         (-mpower8-vector): Likewise.
19435         (-mcrypto): Likewise.
19436         (-mdirect-move): Likewise.
19437         (-mquad-memory): Likewise.
19439         * config/rs6000/rs6000.c (power8_cost): Initial definition for power8.
19440         (rs6000_hard_regno_mode_ok): Make PTImode only match even GPR
19441         registers.
19442         (rs6000_debug_reg_print): Print the base register class if -mdebug=reg.
19443         (rs6000_debug_vector_unit): Add p8_vector.
19444         (rs6000_debug_reg_global): If -mdebug=reg, print power8 constraint
19445         definitions.  Also print fusion state.
19446         (rs6000_init_hard_regno_mode_ok): Set up power8 constraints.
19447         (rs6000_builtin_mask_calculate): Add power8 builtin support.
19448         (rs6000_option_override_internal): Add support for power8.
19449         (rs6000_common_init_builtins): Add debugging for skipped builtins
19450         if -mdebug=builtin.
19451         (rs6000_adjust_cost): Add power8 support.
19452         (rs6000_issue_rate): Likewise.
19453         (insn_must_be_first_in_group): Likewise.
19454         (insn_must_be_last_in_group): Likewise.
19455         (force_new_group): Likewise.
19456         (rs6000_register_move_cost): Likewise.
19457         (rs6000_opt_masks): Likewise.
19459         * config/rs6000/rs6000.h (ASM_CPU_POWER8_SPEC): If we don't have a
19460         power8 capable assembler, default to power7 options.
19461         (TARGET_DIRECT_MOVE): Likewise.
19462         (TARGET_CRYPTO): Likewise.
19463         (TARGET_P8_VECTOR): Likewise.
19464         (VECTOR_UNIT_P8_VECTOR_P): Define power8 vector support.
19465         (VECTOR_UNIT_VSX_OR_P8_VECTOR_P): Likewise.
19466         (VECTOR_MEM_P8_VECTOR_P): Likewise.
19467         (VECTOR_MEM_VSX_OR_P8_VECTOR_P): Likewise.
19468         (VECTOR_MEM_ALTIVEC_OR_VSX_P): Likewise.
19469         (TARGET_XSCVDPSPN): Likewise.
19470         (TARGET_XSCVSPDPN): Likewsie.
19471         (TARGET_SYNC_HI_QI): Likewise.
19472         (TARGET_SYNC_TI): Likewise.
19473         (MASK_CRYPTO): Likewise.
19474         (MASK_DIRECT_MOVE): Likewise.
19475         (MASK_P8_FUSION): Likewise.
19476         (MASK_P8_VECTOR): Likewise.
19477         (REG_ALLOC_ORDER): Move fr13 to be lower in priority so that the TFmode
19478         temporary used by some of the direct move instructions to get two FP
19479         temporary registers does not force creation of a stack frame.
19480         (VLOGICAL_REGNO_P): Allow vector logical operations in GPRs.
19481         (MODES_TIEABLE_P): Move the VSX tests above the Altivec tests so
19482         that any VSX registers are tieable, even if they are also an
19483         Altivec vector mode.
19484         (r6000_reg_class_enum): Add wm, wr, wv constraints.
19485         (RS6000_BTM_P8_VECTOR): Power8 builtin support.
19486         (RS6000_BTM_CRYPTO): Likewise.
19487         (RS6000_BTM_COMMON): Likewise.
19489         * config/rs6000/rs6000.md (cpu attribute): Add power8.
19490         * config/rs6000/rs6000-opts.h (PROCESSOR_POWER8): Likewise.
19491         (enum rs6000_vector): Add power8 vector support.
19493 2013-05-22  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19495         PR target/19599
19496         PR target/57340
19497         * config/arm/arm.c (any_sibcall_uses_r3): Rename to ..
19498         (any_sibcall_could_use_r3): this and handle indirect calls.
19499         (arm_get_frame_offsets): Rename use of any_sibcall_uses_r3.
19501 2013-05-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
19503         * config/rs6000/rs6000.h (MALLOC_ABI_ALIGNMENT): New #define.
19505 2013-05-22  Richard Biener  <rguenther@suse.de>
19507         PR middle-end/57349
19508         * profile.c (branch_prob): Do not split blocks that are
19509         abnormally receiving from ECF_RETURNS_TWICE functions.
19511 2013-05-22  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
19513         * recog.c (offsettable_address_addr_space_p): Fix calculation of
19514         address mode.  Move pointer mode initialization to the same place.
19516 2013-05-22  Michael Zolotukhin  <michael.v.zolotukhin@intel.com>
19518         * read-rtl.c (copy_rtx_for_iterators): Continue applying iterators
19519         while it has any effect.
19521 2013-05-21  Easwaran Raman  <eraman@google.com>
19523         PR tree-optimization/57322
19524         * tree-ssa-reassoc.c (build_and_add_sum): If a BB is empty, set the
19525         UID of the statement added to the BB to be 1.
19527 2013-05-21  Jakub Jelinek  <jakub@redhat.com>
19529         PR tree-optimization/57331
19530         * tree-vrp.c (simplify_cond_using_ranges): Don't optimize comparison
19531         of conversion from pointer type to integral type with integer.
19533 2013-05-21  Martin Jambor  <mjambor@suse.cz>
19535         PR lto/57289
19536         * ipa-prop.c (ipa_read_node_info): Process param_used and
19537         controlled_uses in the same order as when writing.
19539 2013-05-21  Magnus Granberg  <baldrick@free.fr>
19541         PR plugins/56754
19542         * Makefile.in (PLUGIN_HEADERS): Add $(TARGET_H).
19544 2013-05-21  Richard Biener  <rguenther@suse.de>
19546         PR tree-optimization/57318
19547         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not
19548         estimate stmts with side-effects as likely eliminated.
19550 2013-05-21  Richard Biener  <rguenther@suse.de>
19552         PR tree-optimization/57330
19553         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Properly
19554         preserve the call stmts fntype.
19556 2013-05-21  Richard Biener  <rguenther@suse.de>
19558         PR tree-optimization/57303
19559         * tree-ssa-sink.c (statement_sink_location): Improve killing
19560         stmt detection and properly handle self-assignments.
19562 2013-05-21  Christian Bruel  <christian.bruel@st.com>
19564         * dwarf2out.c (multiple_reg_loc_descriptor): Use dbx_reg_number for
19565         spanning registers. LEAF_REG_REMAP is supported only for contiguous
19566         registers. Set register size out of the PARALLEL loop.
19568 2013-05-20  Oleg Endo  <olegendo@gcc.gnu.org>
19570         PR target/56547
19571         * config/sh/sh.md (fmasf4): Remove empty constraints strings.
19572         (*fmasf4, *fmasf4_media): New insns.
19574 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19576         * config/mips/mips.h (BASE_INSN_LENGTH, NOP_INSN_LENGTH): New macros.
19577         * config/mips/mips.c (mips_symbol_insns, mips_address_insns)
19578         (mips_const_insns, mips_split_const_insns, mips_load_store_insns)
19579         (mips_idiv_insns): Update the comments to say that the returned
19580         instruction counts are in units of BASE_INSN_LENGTH.
19581         (mips_adjust_insn_length): Multiply the mips_load_label_num_insns
19582         by BASE_INSN_LENGTH rather than 4.  Add the jump separately,
19583         using 2 rather than 4 as the length of indirect MIPS16 and
19584         microMIPS jumps.  Use NOP_INSN_LENGTH rather than 4 as the
19585         length of a NOP.  Don't divide MIPS16 lengths by 2.
19586         (mips16_split_long_branches): Assume a branch is long if the
19587         length is greater than 4 rather than 8.
19588         * config/mips/mips.md (length): Give MIPS16 lengths directly,
19589         rather than multiplying them by 2.  Multiply instruction counts
19590         by BASE_INSN_LENGTH rather than 4.
19591         (*jump_mips16, tls_get_tp_mips16_<mode>)
19592         (*tls_get_tp_mips16_call_<mode>): Divide lengths by 2.
19594 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19596         * config/mips/mips.md (extended_mips16): Remove branch case.
19597         (length): Remove duplicated extended_mips16 test.
19599 2013-05-19  Richard Sandiford  <rdsandiford@googlemail.com>
19601         * config/mips/t-sde: Don't build 64-bit microMIPS multilibs.
19603 2013-05-18  Richard Sandiford  <rdsandiford@googlemail.com>
19605         * recog.h (Recog_data): Rename to...
19606         (recog_data_d): ...this.
19607         (recog_data): Update accordingly.
19608         * recog.c (recog_data): Likewise.
19609         * reload.c (save_recog_data): Likewise.
19610         * config/picochip/picochip.c (picochip_saved_recog_data): Likewise.
19611         (picochip_save_recog_data, picochip_restore_recog_data): Likewise.
19613 2013-05-17  Julian Brown  <julian@codesourcery.com>
19615         * gcse.c (compute_ld_motion_mems): If a non-simple MEM is
19616         found in a REG_EQUAL note, invalidate it.
19618 2013-05-17  Easwaran Raman  <eraman@google.com>
19620         * tree-ssa-reassoc.c (find_insert_point): New function.
19621         (insert_stmt_after): Likewise.
19622         (get_def_stmt): Likewise.
19623         (ensure_ops_are_available): Likewise.
19624         (not_dominated_by): Likewise.
19625         (rewrite_expr_tree): Do not move statements beyond what is
19626         necessary. Remove call to swap_ops_for_binary_stmt...
19627         (reassociate_bb): ... and move it here.
19628         (build_and_add_sum): Assign UIDs for new statements.
19629         (linearize_expr): Likewise.
19630         (do_reassoc): Renumber gimple statement UIDs.
19632 2013-05-17  Jan Hubicka  <jh@suse.cz>
19634         * lto-symtab.c (lto_symtab_merge_cgraph_nodes): Resolve cross module
19635         weakrefs.
19636         * cgraph.c (dump_cgraph_node): Do not ice on unresolved alias.
19637         * cgraphunit.c (handle_alias_pairs): Store target of unresolved
19638         weakrefs.
19639         (output_weakrefs): Update.
19641 2013-05-17  Po-Chun Chang  <pchang9@cs.wisc.edu>
19642             Martin Jambor  <mjambor@suse.cz>
19644         PR middle-end/57276
19645         * ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Break when a
19646         value that corresponds to the given aggval is found in values vector.
19648 2013-05-17  Uros Bizjak  <ubizjak@gmail.com>
19650         * config/i386/driver-i386.c (host_detect_local_cpu): Pass mmx, 3dnow,
19651         sse, sse2, sse3, ssse3 and sse4a flags to options.
19653 2013-05-17  David Malcolm  <dmalcolm@redhat.com>
19655         * gengtype-state.c: (s_expr_writer): New class, to handle
19656         prettifying of output layout of s-expressions.
19657         (state_writer): New class, to write out gtype.state.
19658         (state_written_type_count): Move this variable into member data of
19659         state_writer.
19660         (s_expr_writer::s_expr_writer): New code: constructor for new class
19661         (state_writer::state_writer(): ditto
19662         (s_expr_writer::write_new_line): New function
19663         (s_expr_writer::write_any_indent): ditto
19664         (s_expr_writer::begin_s_expr): ditto
19665         (s_expr_writer::end_s_expr): ditto
19666         (write_state_fileloc): convert to method of state_writer...
19667         (state_writer:: write_state_fileloc): ...and use methods of
19668         s_expr_writer to write indentation into the gtype.state output file
19669         to visually represent the hierarchical structure of the list
19670         structures
19671         (write_state_fields): ditto, renaming to...
19672         (state_writer::write_state_fields)
19673         (write_state_a_string): ditto, renaming to...
19674         (state_writer::write_state_a_string)
19675         (write_state_string_option): ditto, renaming to...
19676         (state_writer::write_state_string_option)
19677         (write_state_type_option): ditto, renaming to...
19678         (state_writer::write_state_type_option)
19679         (write_state_nested_option): ditto, renaming to...
19680         (state_writer::write_state_nested_option)
19681         (write_state_option): ditto, renaming to...
19682         (state_writer::write_state_option)
19683         (write_state_options): ditto, renaming to...
19684         (state_writer::write_state_options)
19685         (write_state_lang_bitmap): ditto, renaming to...
19686         (state_writer::write_state_lang_bitmap)
19687         (write_state_version): ditto, renaming to...
19688         (state_writer::write_state_version)
19689         (write_state_scalar_type): ditto, renaming to...
19690         (state_writer::write_state_scalar_type)
19691         (write_state_string_type): ditto, renaming to...
19692         (state_writer::write_state_string_type)
19693         (write_state_undefined_type): ditto, renaming to...
19694         (state_writer::write_state_undefined_type)
19695         (write_state_struct_union_type): ditto, renaming to...
19696         (state_writer::write_state_struct_union_type)
19697         (write_state_struct_type): ditto, renaming to...
19698         (state_writer::write_state_struct_type)
19699         (write_state_user_struct_type): ditto, renaming to...
19700         (state_writer::write_state_user_struct_type)
19701         (write_state_lang_struct_type): ditto, renaming to...
19702         (state_writer::write_state_lang_struct_type)
19703         (write_state_param_struct_type): ditto, renaming to...
19704         (state_writer::write_state_param_struct_type)
19705         (write_state_pointer_type): ditto, renaming to...
19706         (state_writer::write_state_pointer_type)
19707         (write_state_array_type): ditto, renaming to...
19708         (state_writer::write_state_array_type)
19709         (write_state_gc_used): ditto, renaming to...
19710         (state_writer::write_state_gc_used)
19711         (write_state_common_type_content): ditto, renaming to...
19712         (state_writer::write_state_common_type_content)
19713         (write_state_type): ditto, renaming to...
19714         (state_writer::write_state_type)
19715         (write_state_pair_list): ditto, renaming to...
19716         (state_writer::write_state_pair_list)
19717         (write_state_pair): ditto, renaming to...
19718         (state_writer::write_state_pair)
19719         (write_state_typedefs): ditto, renaming to...
19720         (state_writer::write_state_typedefs)
19721         (write_state_structures): ditto, renaming to...
19722         (state_writer::write_state_structures)
19723         (write_state_param_structs): ditto, renaming to...
19724         (state_writer::write_state_param_structs)
19725         (write_state_variables): ditto, renaming to...
19726         (state_writer::write_state_variables)
19727         (write_state_srcdir): ditto, renaming to...
19728         (state_writer::write_state_srcdir)
19729         (write_state_files_list): ditto, renaming to...
19730         (state_writer::write_state_files_list)
19731         (write_state_languages): ditto, renaming to...
19732         (state_writer::write_state_languages)
19733         (write_state): create a state_writer instance and use it when
19734         writing out the state file
19736 2013-05-17  Mike Stump  <mikestump@comcast.net>
19738         PR rtl-optimization/57304
19739         * web.c (union_match_dups): Ensure that DF_REF_LOC exists before
19740         accessing DF_REF_REAL_LOC.
19742 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19744         PR rtl-optimization/57281
19745         PR rtl-optimization/57300
19746         * config/i386/i386.md (extendsidi2_1 dead reg splitter): Remove.
19747         (extendsidi2_1 peephole2s): Add instead 2 new peephole2s, that undo
19748         what the other splitter did if the registers are dead.
19750 2013-05-17  Richard Biener  <rguenther@suse.de>
19752         * tree-ssa-alias.c (stmt_kills_ref_p_1): Properly compare
19753         MEM_REF offsets.
19755 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19757         * gcc.c (SANITIZER_SPEC): Reject -fsanitize=address -fsanitize=thread
19758         linking.
19760 2013-05-17  Marek Polacek  <polacek@redhat.com>
19762         * tree-ssa-strlen.c (handle_char_store): Don't invalidate cached
19763         length when doing non-zero store of storing '\0' to '\0'.
19765 2013-05-17  Jakub Jelinek  <jakub@redhat.com>
19767         * tree-vect-patterns.c (vect_recog_rotate_pattern): For
19768         vect_external_def oprnd1 with loop_vinfo, try to emit
19769         optional cast, negation and and stmts on the loop preheader
19770         edge instead of into the pattern def seq.
19772         PR tree-optimization/57051
19773         * fold-const.c (const_binop) <case VEC_LSHIFT_EXPR,
19774         case VEC_RSHIFT_EXPR>: Fix BYTES_BIG_ENDIAN handling.
19776 2013-05-16  Nick Clifton  <nickc@redhat.com>
19778         * config/rl78/rl78.c (rl78_attribute_table): Add naked.
19779         (rl78_is_naked_func): New function.
19780         (rl78_expand_prologue): Skip prologue generation for naked functions.
19781         (rl78_expand_epilogue): Skip epilogue generation for naked functions.
19782         * doc/extend.texi (naked): Add RL78 to the list of processors
19783         that supports this attribute.
19785 2013-05-16  Jeff Law  <law@redhat.com>
19787         * Makefile.in (tree-switch-conversion.o): Depend on $(OPTABS_H).
19789 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19791         * config/i386/driver-i386.c (host_detect_local_cpu): Determine
19792         cache parameters using detect_caches_amd also for CYRIX,
19793         NSC and TM2 signatures.
19795 2013-05-16  Uros Bizjak  <ubizjak@gmail.com>
19796             Dzianis Kahanovich  <mahatma@eu.by>
19798         PR target/45359
19799         PR target/46396
19800         * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19801         VIA/Centaur processors and determine their cache parameters
19802         using detect_caches_amd.
19804 2013-05-16  Teresa Johnson  <tejohnson@google.com>
19806         * cfgrtl.c (verify_hot_cold_block_grouping): Return err.
19807         (rtl_verify_edges): New function.
19808         (rtl_verify_bb_insns): Ditto.
19809         (rtl_verify_bb_pointers): Ditto.
19810         (rtl_verify_bb_insn_chain): Ditto.
19811         (rtl_verify_fallthru): Ditto.
19812         (rtl_verify_bb_layout): Ditto.
19813         (rtl_verify_flow_info_1): Outline checks into new functions.
19814         (rtl_verify_flow_info): Ditto.
19816 2013-05-16  Steve Ellcey  <sellcey@imgtec.com>
19818         * cfghooks.c (copy_bbs): Add update_dominance argument.
19819         * cfghooks.h (copy_bbs): Update prototype.
19820         * tree-cfg.c (gimple_duplicate_sese_region):
19821         Add update_dominance argument.
19822         * tree-flow.h (gimple_duplicate_sese_region): Update prototype.
19823         * tree-ssa-loop-ch.c (copy_loop_headers): Update
19824         gimple_duplicate_sese_region call.
19825         * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg):
19826         Update copy_bbs call.
19827         * cfgloopmanip.c (duplicate_loop_to_header_edge): Ditto.
19828         * trans-mem.c (ipa_uninstrument_transaction): Ditto.
19830 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19832         * tree-vectorizer.h (NUM_PATTERNS): Increment.
19833         * tree-vect-patterns.c (vect_vect_recog_func_ptrs): Add
19834         vect_recog_rotate_pattern.
19835         (vect_recog_rotate_pattern): New function.
19837 2013-05-16  Jason Merrill  <jason@redhat.com>
19839         * Makefile.in (LLINKER): New variable.
19840         (mostlyclean): Remove link mutex.
19841         * configure.ac: Handle --enable-link-mutex.
19842         * lock-and-run.sh: New script.
19844 2013-05-16  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19846         PR target/19599
19847         * config/arm/arm.c (arm_function_ok_for_sibcall): Add check
19848         for NULL decl.
19850 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19852         * reorg.c (link_cc0_insns): Wrap in #ifdef HAVE_cc0.
19854 2013-05-16  Greta Yorsh  <Greta.Yorsh@arm.com>
19856         * config/arm/arm-protos.h (gen_movmem_ldrd_strd): New declaration.
19857         * config/arm/arm.c (next_consecutive_mem): New function.
19858         (gen_movmem_ldrd_strd): Likewise.
19859         * config/arm/arm.md (movmemqi): Update condition and code.
19860         (unaligned_loaddi, unaligned_storedi): New patterns.
19862 2013-05-16  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
19864         * config.gcc: Obsolete *-*-solaris2.9*.
19865         * doc/install.texi (Specific, *-*-solaris2*): Document it.
19867 2013-05-16  Richard Biener  <rguenther@suse.de>
19869         * passes.c (init_optimization_passes): Move pass_parallelize_loops
19870         earlier, after GRAPHITE transforms and IV canonicalization.
19872 2013-05-16  Jakub Jelinek  <jakub@redhat.com>
19874         * omp-low.c (extract_omp_for_data): For collapsed loops,
19875         if at least one of the loops is known at compile time to
19876         iterate zero times, set count to 0.
19877         (expand_omp_regimplify_p): New function.
19878         (expand_omp_for_generic): For collapsed loops, if at least
19879         one of the loops isn't known to iterate at least once,
19880         add runtime check with setting count to 0.
19881         (expand_omp_for_static_nochunk, expand_omp_for_static_chunk):
19882         For unsigned types if it isn't known at compile time that
19883         the loop will iterate at least once, add runtime check to bypass
19884         the whole loop if initial condition isn't true.
19886 2013-05-16  Nathan Sidwell  <nathan@codesourcery.com>
19888         * varasm.c (default_use_anchors_for_symbol_p): Use decl_replaceable_p.
19890 2013-05-16  Marc Glisse  <marc.glisse@inria.fr>
19892         PR middle-end/57286
19893         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Disable some
19894         transformations to avoid an infinite loop.
19896 2013-05-16  Marek Polacek  <polacek@redhat.com>
19898         * tree-scalar-evolution.c (scev_const_prop): Add more dumps.
19900 2013-05-15  Leif Ekblad  <leif@rdos.net>
19902         * config/i386/i386.c (ix86_decompose_address): Use
19903         DEFAULT_TLS_SEG_REG to access TLS segment register.
19904         * config/i386/i386.h (DEFAULT_TLS_SEG_REG): New define.
19905         * config/i386/rdos.h (DEFAULT_TLS_SEG_REG): Ditto.
19906         (TARGET_TLS_DIRECT_SEG_REFS_DEFAULT): Ditto.
19908 2013-05-15  Richard Sandiford  <rdsandiford@googlemail.com>
19910         PR target/57260
19911         * config/mips/mips.c (mips_function_ok_for_sibcall): Don't allow
19912         sibling calls to functions that would normally be lazily bound,
19913         unless $gp is call-clobbered.
19915 2013-05-15  Uros Bizjak  <ubizjak@gmail.com>
19917         * config/i386/i386.c (ix86_option_override_internal): Update
19918         processor_alias_table for missing PTA_PRFCHW and PTA_FXSR flags.  Add
19919         PTA_POPCNT to corei7 entry. Do not enable SSE prefetch on
19920         non-SSE 3dNow! targets.  Enable TARGET_PRFCHW for TARGET_3DNOW targets.
19921         * config/i386/i386.md (prefetch): Enable for TARGET_PRFCHW instead
19922         of TARGET_3DNOW.
19923         (*prefetch_3dnow): Enable for TARGET_PRFCHW only.
19925 2013-05-15  Andreas Schwab  <schwab@suse.de>
19927         * config/m68k/m68k.md (*rotlhi3_lowpart, *rotlqi3_lowpart): Name
19928         for rotlhi3+1 and rotlqi3+1, resp.  Fix reference to non-existing
19929         third operand.
19931 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19933         * loop-unroll.c (report_unroll_peel): Check decision before
19934         emitting unroll/peel message.
19936 2013-05-15  Teresa Johnson  <tejohnson@google.com>
19938         * function.h (has_bb_partition): New rtl_data flag.
19939         (bb_reorder_complete): Ditto.
19940         * cfgcleanup.c (try_crossjump_to_edge): Check for has_bb_partition
19941         instead of flag_reorder_blocks_and_partition.
19942         * cfgrtl.c (verify_hot_cold_block_grouping): Moved from bb-reorder.c,
19943         with some enhancements.
19944         (rtl_verify_flow_info_1): Call verify_hot_cold_block_grouping.
19945         * bb-reorder.c (connect_traces): Check for has_bb_partition
19946         instead of flag_reorder_blocks_and_partition.
19947         (verify_hot_cold_block_grouping): Moved to cfgrtl.c.
19948         (reorder_basic_blocks): Set bb_reorder_complete flag, remove call to
19949         verify_hot_cold_block_grouping.
19950         (partition_hot_cold_basic_blocks): Set has_bb_partition.
19952 2013-05-15  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
19954         PR target/19599
19955         * config/arm/predicates.md (call_insn_operand): New predicate.
19956         * config/arm/constraints.md ("Cs", "Ss"):  New constraints.
19957         * config/arm/arm.md (*call_insn, *call_value_insn): Match only
19958         if insn is not a tail call.
19959         (*sibcall_insn, *sibcall_value_insn): Adjust for tailcalling through
19960         registers.
19961         * config/arm/arm.h (enum reg_class): New caller save register class.
19962         (REG_CLASS_NAMES): Likewise.
19963         (REG_CLASS_CONTENTS): Likewise.
19964         * config/arm/arm.c (arm_function_ok_for_sibcall): Allow tailcalling
19965         without decls.
19967 2013-05-15  Richard Biener  <rguenther@suse.de>
19969         * tree-vect-loop.c (vect_transform_loop): Use MSG_NOTE instead
19970         of MSG_OPTIMIZED_LOCATIONS.
19971         * tree-vect-slp.c (vect_make_slp_decision): Likewise.
19972         (vect_slp_transform_bb): Indicate location in MSG_OPTIMIZED_LOCATIONS
19973         message.
19974         * tree-vectorizer.c (vectorize_loops): Use MSG_NOTE instead
19975         of MSG_OPTIMIZED_LOCATIONS.
19976         (execute_vect_slp): Likewise.
19977         * tree-vect-loop-manip.c (vect_do_peeling_for_loop_bound): Likewise.
19978         (vect_create_cond_for_alias_checks): Likewise.
19979         * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
19980         (vect_recog_widen_mult_pattern): Likewise.
19981         (vect_recog_widen_sum_pattern): Likewise.
19982         (vect_recog_over_widening_pattern): Likewise.
19983         (vect_recog_widen_shift_pattern): Likewise.
19984         (vect_recog_vector_vector_shift_pattern): Likewise.
19985         (vect_recog_divmod_pattern): Likewise.
19986         (vect_recog_mixed_size_cond_pattern): Likewise.
19987         (vect_recog_bool_pattern): Likewise.
19988         (vect_pattern_recog_1): Likewise.
19990 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19992         * ipa-prop.c (ipa_make_edge_direct_to_target): Redirect calls to
19993         non-functions to builtin_unreachable.
19994         * ipa-inline-transform.c (inline_call): Do not assert estimates were
19995         correct when new direct edges were discovered.
19997 2013-05-15  Martin Jambor  <mjambor@suse.cz>
19999         * ipa-prop.c (ipa_print_node_jump_functions): Print symbol order in
20000         header, print symbol order instead of node uid, print more information
20001         about indirect edge targets.
20002         (ipa_make_edge_direct_to_target): Print symbol order instead of node
20003         uids.
20004         (ipa_make_edge_direct_to_target): Likewise.
20005         (remove_described_reference): Likewise.
20006         (propagate_controlled_uses): Likewise.
20007         (ipa_print_node_params): Also print symbol order.
20008         (ipcp_transform_function): Print symbol order instead of node uids.
20009         * cgraph.c (cgraph_redirect_edge_call_stmt_to_callee): Likewise.
20010         (cgraph_get_create_real_symbol_node): Likewise.
20011         * ipa-cp.c (print_lattice): Likewise.
20012         (print_all_lattices): Likewise.
20013         (determine_versionability): Likewise.
20014         (initialize_node_lattices): Likewise.
20015         (estimate_local_effects): Likewise.
20016         (update_profiling_info): Likewise.
20017         (create_specialized_node): Likewise.
20018         (perhaps_add_new_callers): Likewise.
20019         (decide_about_value): Likewise.
20020         (decide_whether_version_node): Likewise.
20021         (identify_dead_nodes): Likewise.
20022         * ipa-inline-analysis.c (dump_inline_edge_summary): Likewise.
20023         (dump_inline_summary): Likewise.
20024         (estimate_node_size_and_time): Likewise.
20025         (inline_analyze_function): Likewise.
20026         * ipa-inline.c (report_inline_failed_reason): Likewise.
20027         (want_early_inline_function_p): Likewise.
20028         (edge_badness): Likewise.
20029         (update_edge_key): Likewise.
20030         (inline_small_functions): Likewise.  Add dumping of order to two other
20031         dumps.
20032         * ipa-pure-const.c (pure_const_read_summary): Print symbol order
20033         instead of node uids.
20034         (propagate_pure_const): Likewise.
20035         (propagate_pure_const): Likewise.
20036         * ipa-utils.c (dump_cgraph_node_set): Likewise.
20037         * lto-cgraph.c (input_node): Explicitly specify we dump uid.
20038         * lto-symtab.c (lto_cgraph_replace_node): Print symbol order instead
20039         of node uids.
20040         * tree-pretty-print.c (dump_function_header): Likewise.
20041         * tree-sra.c (convert_callers_for_node): Dump in traditional format.
20042         Print symbol order instead of node uids.
20044 2013-05-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
20046         * config/s390/s390.c (s390_register_move_cost): Don't impose the
20047         FPR<->GPR move cost penalty if ldgr/lgdr can be used.
20049 2013-05-15  Richard Biener  <rguenther@suse.de>
20051         PR tree-optimization/57275
20052         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Fix
20053         return value for fail to do runtime alias checks for gather loads.
20055 2013-05-15  Jan Hubicka  <jh@suse.cz>
20057         PR lto/57038
20058         PR lto/47375
20059         * lto-symtab.c (lto_symtab_symbol_p): Add external symbol;
20060         weakrefs are not external.
20061         (lto_symtab_merge_decls): Fix thinko when dealing with
20062         non-lto_symtab decls.
20063         (lto_symtab_merge_cgraph_nodes): Use lto_symtab_symbol_p.
20064         (lto_symtab_prevailing_decl): Get int sync with lto_symtab_symbol_p.
20065         * varpool.c (dump_varpool_node): Dump more flags.
20067 2013-05-15  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
20069         * config/i386/i386.c (processor_alias_table): Add instruction
20070         FSGSBASE for AMD bdver3 architecture.
20072 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20074         * tree.c (warn_deprecated_use): Print file:line using locus color.
20075         * diagnostic.c (diagnostic_report_current_module): Print file:line
20076         and file:line:column using locus color.
20078 2013-05-14  Mike Stump  <mikestump@comcast.net>
20080         * gdbinit.in: Add __null.
20082 2013-05-14  Mike Stump  <mikestump@comcast.net>
20084         * recog.h: Rename struct recog_data to Recog_data.
20085         * recog.c: Likewise.
20086         * reload.c (can_reload_into): Likewise.
20087         * config/picochip/picochip.c: Likewise.
20089 2013-05-14  Mike Stump  <mikestump@comcast.net>
20091         * web.c (union_match_dups): Also check DF_REF_REAL_LOC.
20093 2013-05-14  Steven Bosscher  <steven@gcc.gnu.org>
20095         * resource.h (struct resources): Remove unch_memory member.
20096         (CLEAR_RESOURCE): Don't clear unch_memory.
20097         * resource.c (mark_referenced_resources): Don't set it.
20098         (mark_set_resources): Likewise.
20099         (mark_target_live_regs): Don't clear it.
20100         (init_resource_info): Likewise.
20101         * reorg.c (resource_conflicts_p): Don't compare it.
20102         (redundant_insn): Don't set it.
20104         * rtl.h (next_label, skip_consecutive_labels, link_cc0_insns):
20105         Remove prototypes.
20106         * emit-rtl.c (next_label): Remove unused function.
20107         (skip_consecutive_labels, link_cc0_insns): Move to ...
20108         * reorg.c (skip_consecutive_labels, link_cc0_insns): ... here, the
20109         only place where these functions are used, and make them static.
20111 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20113         * fold-const.c (fold_negate_expr): Handle vectors.
20114         (fold_truth_not_expr): Make it static.
20115         (fold_invert_truthvalue): New static function.
20116         (invert_truthvalue_loc): Handle vectors. Do not call
20117         fold_truth_not_expr directly.
20118         (fold_unary_loc) <BIT_NOT_EXPR>: Handle comparisons.
20119         <TRUTH_NOT_EXPR>: Do not cast to boolean.
20120         (fold_comparison): Handle vector constants.
20121         (fold_binary_loc) <TRUTH_XOR_EXPR>: Remove redundant code.
20122         (fold_ternary_loc) <VEC_COND_EXPR>: Adapt more COND_EXPR optimizations.
20123         * tree.h (fold_truth_not_expr): Remove declaration.
20125 2013-05-14  James Greenhalgh  <james.greenhalgh@arm.com>
20127         * config/aarch64/aarch64-simd.md
20128         (aarch64_vcond_internal<mode>): Rename to...
20129         (aarch64_vcond_internal<mode><mode>): ...This, for integer modes.
20130         (aarch64_vcond_internal<VDQF_COND:mode><VDQF:mode>): ...This for
20131         float modes. Clarify all iterator modes.
20132         (vcond<mode><mode>): Use new name for vcond expanders.
20133         (vcond<v_cmp_result><mode>): Likewise.
20134         (vcondu<mode><mode>: Likewise.
20135         * config/aarch64/iterators.md (VDQF_COND): New.
20137 2013-05-14  Marc Glisse  <marc.glisse@inria.fr>
20139         PR bootstrap/57266
20140         * fold-const.c (fold_binary_loc) <shift>: Use an unsigned
20141         variable for the shift amount. Check that we shift by non-negative
20142         amounts.
20144 2013-05-14  Chung-Lin Tang  <cltang@codesourcery.com>
20146         PR target/42017
20147         * config/arm/arm.h (EPILOGUE_USES): Only return true
20148         for LR_REGNUM after epilogue_completed.
20150 2013-05-14  Joern Rennecke  <joern.rennecke@embecosm.com>
20152         * config/avr/avr.c (avr_encode_section_info): Bail out if the type
20153         is error_mark_node.
20155 2013-05-14  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
20157         PR target/57261
20158         * configure.ac (gcc_cv_ld_as_needed): Disable before Solaris 11
20159         and Solaris 11+/x86 with gld.
20160         * configure: Regenerate.
20162 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20164         * expmed.c (expand_shift_1): Canonicalize rotates by
20165         constant bitsize / 2 to bitsize - 1.
20166         * simplify-rtx.c (simplify_binary_operation_1) <case ROTATE,
20167         case ROTATERT>: Likewise.
20169         Revert:
20170         2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20172         * config/i386/i386.md (rotateinv): New code attr.
20173         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20174         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20175         roll $31, %eax, etc.
20177 2013-05-14  Richard Biener  <rguenther@suse.de>
20179         PR middle-end/57235
20180         * tree-eh.c (sink_clobbers): Give up for successors with
20181         multiple predecessors and no virtual uses.
20183 2013-05-14  Eric Botcazou  <ebotcazou@adacore.com>
20185         * config/sparc/sp64-elf.h (CPP_SUBTARGET_SPEC): Delete.
20186         * config/sparc/openbsd64.h (CPP_SUBTARGET_SPEC): Likewise.
20188 2013-05-14  Jakub Jelinek  <jakub@redhat.com>
20190         PR middle-end/57251
20191         * expr.c (expand_expr_real_2) <case WIDEN_MULT_EXPR>: Handle
20192         the case when both op0 and op1 have VOIDmode.
20194 2013-05-14  Kaushik Phatak  <kaushik.phatak@kpitcummins.com>
20196         * config/rl78/rl78.md(mulsi3_g13): Add additional 'nop' required
20197         in multiply-accumulate mode.
20199 2013-05-13  Guozhi Wei  <carrot@google.com>
20201         * dwarf2asm.c (dw2_output_indirect_constant_1): Mark new decl STATIC.
20203 2013-05-13  Kai Tietz  <ktietz@redhat.com>
20205         PR target/56975
20206         * config/i386/cygming.h (TARGET_PECOFF): Define as true.
20207         * config/i386/i386.h (TARGET_PECOFF): Define by default as false.
20208         (PIC_OFFSET_TABLE_REGNUM): Use TARGET_PECOFF.
20209         * config/i386/i386.c (ix86_option_override_internal): Likewise.
20210         (ix86_expand_prologue): Likewise.
20211         (ix86_expand_split_stack_prologue): Likewise.
20212         (legitimate_pic_address_disp_p): Likewise.
20213         (legitimize_pic_address): Likewise.
20214         (legitimize_tls_address): Likewise.
20215         (legitimize_pe_coff_symbol): Likewise.
20216         (output_pic_addr_const): Likewise.
20217         (construct_plt_address): Likewise.
20218         (ix86_expand_call): Likewise.
20219         (x86_output_mi_thunk): Likewise.
20220         (x86_function_profiler): Likewise.
20222 2013-05-13  Sofiane Naci  <sofiane.naci@arm.com>
20224         * config/aarch64/aarch64-simd.md (aarch64_simd_mov<mode>): Group
20225         similar switch cases.
20226         (aarch64_simd_mov): Rename to aarch64_split_simd_mov. Update.
20227         (aarch64_simd_mov_to_<mode>low): Delete.
20228         (aarch64_simd_mov_to_<mode>high): Delete.
20229         (move_lo_quad_<mode>): Add w<-r alternative.
20230         (aarch64_simd_move_hi_quad_<mode>): Likewise.
20231         (aarch64_simd_mov_from_*): Update type attribute.
20232         * config/aarch64/aarch64.c (aarch64_split_simd_move): Refacror switch
20233         statement.
20235 2013-05-13  Jan Hubicka  <jh@suse.cz>
20237         * mode-switching.c (optimize_mode_switching): Set correct RTL profile.
20238         * config/i386/i386.c (ix86_compute_frame_layout,
20239         ix86_expand_epilogue, emit_i387_cw_initialization,
20240         ix86_expand_vector_move_misalign, ix86_fp_comparison_strategy,
20241         ix86_local_alignment): Fix use of size/speed predicates.
20243 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20245         PR tree-optimization/45216
20246         PR tree-optimization/57157
20247         * tree-ssa-forwprop.c (simplify_rotate): Only recognize
20248         the (-Y) & (B - 1) variant if OP is |.
20249         * expmed.c (expand_shift_1): For rotations by const0_rtx just
20250         return shifted.  Use (-op1) & (prec - 1) as other_amount
20251         instead of prec - op1.
20253 2013-05-13  Martin Jambor  <mjambor@suse.cz>
20255         PR middle-end/42371
20256         * ipa-prop.h (IPA_UNDESCRIBED_USE): New macro.
20257         (ipa_constant_data): New type.
20258         (ipa_jump_func): Use ipa_constant_data to hold information about
20259         constant jump functions.
20260         (ipa_get_jf_constant): Adjust to jump function type changes.
20261         (ipa_get_jf_constant_rdesc): New function.
20262         (ipa_param_descriptor): New field controlled_uses.
20263         (ipa_get_controlled_uses): New function.
20264         (ipa_set_controlled_uses): Likewise.
20265         * ipa-ref.h (ipa_find_reference): Declare.
20266         * ipa-prop.c (ipa_cst_ref_desc): New type.
20267         (ipa_print_node_jump_functions_for_edge): Adjust for jump function type
20268         changes.
20269         (ipa_set_jf_constant): Likewise.  Also create reference descriptions.
20270         New parameter cs.  Adjust all callers.
20271         (ipa_analyze_params_uses): Detect uncontrolled and controlled uses.
20272         (remove_described_reference): New function.
20273         (jfunc_rdesc_usable): Likewise.
20274         (try_make_edge_direct_simple_call): Decrement controlled use count,
20275         attempt to remove reference if it hits zero.
20276         (combine_controlled_uses_counters): New function.
20277         (propagate_controlled_uses): Likewise.
20278         (ipa_propagate_indirect_call_infos): Call propagate_controlled_uses.
20279         (ipa_edge_duplication_hook): Duplicate reference descriptions.
20280         (ipa_print_node_params): Print described use counter.
20281         (ipa_write_jump_function): Adjust to jump function type changes.
20282         (ipa_read_jump_function): New parameter CS, pass it to
20283         ipa_set_jf_constant.  Adjust caller.
20284         (ipa_write_node_info): Stream controlled use count
20285         (ipa_read_node_info): Likewise.
20286         * cgraph.c (cgraph_mark_address_taken_node): Bail out instead of
20287         asserting.
20288         * ipa-cp.c (ipcp_discover_new_direct_edges): Decrement controlled use
20289         count.  Remove cloning-added reference if it reaches zero.
20290         * ipa-ref.c (ipa_find_reference): New function.
20292 2013-05-13  Ganesh Gopalasubramanian  <Ganesh.Gopalasubramanian@amd.com>
20294         * config/i386/i386.c (processor_target_table): Modified default
20295         alignment values for AMD BD and BT architectures.
20297 2013-05-13  Marc Glisse  <marc.glisse@inria.fr>
20299         * tree-vect-generic.c (uniform_vector_p): Move ...
20300         * tree.c (uniform_vector_p): ... here.
20301         * tree.h (uniform_vector_p): Declare it.
20302         * fold-const.c (fold_binary_loc) <shift>: Turn the second argument
20303         into a scalar.
20305 2013-05-13  Jakub Jelinek  <jakub@redhat.com>
20307         PR tree-optimization/57230
20308         * tree-ssa-strlen.c (handle_char_store): Record length for
20309         array store from STRING_CST.
20311         PR tree-optimization/57230
20312         * tree-ssa-strlen.c (handle_char_store): Add missing integer_zerop
20313         check.
20315 2013-05-12  Joern Rennecke  <joern.rennecke@embecosm.com>
20317         * config/epiphany/epiphany.c (epiphany_init): Check size of
20318         NUM_MODES_FOR_MODE_SWITCHING.
20319         (epiphany_expand_prologue):
20320         Remove CONFIG_REGNUM initial value handling code.
20321         (epiphany_optimize_mode_switching): Handle EPIPHANY_MSW_ENTITY_CONFIG.
20322         (epiphany_mode_needed, epiphany_mode_entry_exit): Likewise.
20323         (emit_set_fp_mode, epiphany_mode_after): Likewise.
20324         (epiphany_mode_needed) <Handle EPIPHANY_MSW_ENTITY_AND>:
20325         Don't return 1 for FP_MODE_NONE.
20326         * config/epiphany/epiphany.h (NUM_MODES_FOR_MODE_SWITCHING):
20327         Add value for EPIPHANY_MSW_ENTITY_CONFIG.
20328         (EPIPHANY_MSW_ENTITY_CONFIG, EPIPHANY_MSW_ENTITY_NUM): Define.
20329         * config/epiphany/epiphany.md (save_config): New pattern.
20331 2013-05-12  Uros Bizjak  <ubizjak@gmail.com>
20333         * config/i386/i386.md (*zero_extendsidi2): Add *x->?r alternative.
20335 2013-05-10  Uros Bizjak  <ubizjak@gmail.com>
20337         * config/i386/i386.md (memory): Handle sseishft1.
20338         * config/i386/sse.md (*vec_extractv4si): Remove memory attribute.
20339         (*vec_extractv2di_1): Ditto.
20341 2013-05-10  Vladimir Makarov  <vmakarov@redhat.com>
20343         * lra-assigns.c (find_hard_regno_for): Add 1 to the cost of call
20344         saved registers.
20346 2013-05-10  Sebastian Huber  <sebastian.huber@embedded-brains.de>
20348         * config/arm/t-rtems-eabi: Remove mthumb/march=armv7 multilib.
20349         Add mthumb/march=armv7-a multilib.
20350         Add mthumb/march=armv7-r multilib.
20351         Add mthumb/march=armv7-a/mfpu=neon/mfloat-abi=hard multilib.
20353 2013-05-10  Ralf Corsépius  <ralf.corsepius@rtems.org>
20355         * config/v850/t-rtems: Add more multilibs.
20357 2013-05-10  Richard Biener  <rguenther@suse.de>
20359         PR tree-optimization/57214
20360         * tree-ssa-loop-ivcanon.c (propagate_constants_for_unrolling): Do
20361         not propagate from SSA names that occur in abnormal PHI nodes.
20363 2013-05-10  Marc Glisse  <marc.glisse@inria.fr>
20365         * stor-layout.c (element_precision): New function.
20366         * machmode.h (element_precision): Declare it.
20367         * tree.c (build_minus_one_cst): New function.
20368         (element_precision): Likewise.
20369         * tree.h (build_minus_one_cst): Declare new function.
20370         (element_precision): Likewise.
20371         * fold-const.c (operand_equal_p): Use element_precision.
20372         (fold_binary_loc): Handle vector types.
20373         * convert.c (convert_to_integer): Use element_precision.
20374         * gimple.c (iterative_hash_canonical_type): Handle complex and vectors
20375         separately.
20377 2013-05-10  Richard Sandiford  <rdsandiford@googlemail.com>
20379         * config/mips/mips-protos.h (m16_uimm3_b, m16_simm4_1, m16_nsimm4_1)
20380         (m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4, m16_simm8_1)
20381         (m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1, m16_uimm8_4)
20382         (m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20383         * config/mips/mips.c (m16_check_op, m16_uimm3_b, m16_simm4_1)
20384         (m16_nsimm4_1, m16_simm5_1, m16_nsimm5_1, m16_uimm5_4, m16_nuimm5_4)
20385         (m16_simm8_1, m16_nsimm8_1, m16_uimm8_1, m16_nuimm8_1, m16_uimm8_m1_1)
20386         (m16_uimm8_4, m16_nuimm8_4, m16_simm8_8, m16_nsimm8_8): Delete.
20387         * config/mips/constraints.md (Udb8, Usb5, Usb8, Usd8, Uub8, Uuw5)
20388         (Uuw8): New constraints.
20389         (Usb4): Move into alphabetical order.
20390         * config/mips/predicates.md (db8_operand, sb5_operand, sb8_operand)
20391         (sd8_operand, ub8_operand, uw8_operand): New predicates.
20392         * config/mips/mips.md (*xor<mode>3, *xor<mode>3_mips16): Name
20393         previously unnamed patterns.
20394         (*add<mode>3_mips16, *xor<mode>3_mips16, *<optab>si3_mips16)
20395         (*ashldi3_mips16, *ashrdi3_mips16, *lshrdi3_mips16)
20396         (*slt<u>_<GPR:mode><GPR2:mode>_mips16)
20397         (*sle<u>_<GPR:mode><GPR2:mode>_mips16): Use constraints instead
20398         of set_attr_alternative/if_then_else.  Use extended_mips16 instead
20399         of specific lengths.
20401 2013-05-10  Jakub Jelinek  <jakub@redhat.com>
20403         * config/i386/i386.md (rotateinv): New code attr.
20404         (*<rotate_insn><mode>3_1, *<rotate_insn>si3_1_zext,
20405         *<rotate_insn>qi3_1_slp): Emit rorl %eax instead of
20406         roll $31, %eax, etc.
20408         PR tree-optimization/45216
20409         PR tree-optimization/57157
20410         * tree-ssa-forwprop.c (simplify_rotate): New function.
20411         (ssa_forward_propagate_and_combine): Call it.
20413 2013-05-10  Richard Biener  <rguenther@suse.de>
20415         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do not
20416         disable peeling when we version for aliasing.
20417         (vector_alignment_reachable_p): Honor explicit user alignment.
20418         (vect_supportable_dr_alignment): Likewise.
20419         * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Use
20420         STMT_VINFO_LOOP_PHI_EVOLUTION_PART instead of recomputing it.
20421         * tree-vect-loop.c (vect_transform_loop): First apply versioning,
20422         then peeling to arrange for the cost-model check to come first.
20424 2013-05-10  Alan Modra  <amodra@gmail.com>
20426         * configure.ac (HAVE_AS_TLS): Swap powerpc64 and powerpc cases.
20427         (HAVE_LD_LARGE_TOC): Don't mention AIX in help text.
20428         * configure: Regenerate.
20430 2013-05-10  Alan Modra  <amodra@gmail.com>
20432         PR target/55033
20433         * varasm.c (default_elf_select_section): Move !DECL_P check..
20434         (get_named_section): ..to here before calling get_section_name.
20435         Adjust assertion.
20436         (default_section_type_flags): Add DECL_P check.
20437         * config/i386/winnt.c (i386_pe_section_type_flags): Likewise.
20438         * config/rs6000/rs6000.c (rs6000_xcoff_section_type_flags): Likewise.
20440 2013-05-09  Joern Rennecke  <joern.rennecke@embecosm.com>
20442         * config/epiphany/epiphany.c (epiphany_expand_prologue):
20443         When using gen_stack_adjust_str with a register offset, add a
20444         REG_FRAME_RELATED_EXPR note.
20446 2013-05-09  Uros Bizjak  <ubizjak@gmail.com>
20448         * config/i386/sse.md (*vec_extractv4si_0_zext): New pattern.
20449         (*vec_extractv4si_zext_mem): Ditto.
20450         (*vec_extractv2di): Add 0->x and x->x alternatives.
20451         * config/i386/mmx.md (*vec_extractv2si_zext_mem): New pattern.
20452         * config/i386/i386.md (*zero_extendsidi2): Add *Yj->?r alternative.
20454 2013-05-09  Jason Merrill  <jason@redhat.com>
20456         N3639 C++1y VLA support
20457         * gimplify.c (gimplify_vla_decl): Don't touch an existing
20458         DECL_VALUE_EXPR.
20460         * tree.c (build_constructor_va): New.
20461         * tree.h: Declare it.
20463 2013-05-09  Martin Jambor  <mjambor@suse.cz>
20465         PR lto/57084
20466         * gimple-fold.c (canonicalize_constructor_val): Call
20467         cgraph_get_create_real_symbol_node instead of cgraph_get_create_node.
20469 2013-05-09  Jan Hubicka  <jh@suse.cz>
20470             Richard Biener  <rguenther@suse.de>
20472         PR lto/54095
20473         * symtab.c (symtab_make_decl_local): Do not add private names.
20475 2013-05-09  Jan Hubicka  <jh@suse.cz>
20477         PR lto/54095
20478         * symtab.c (insert_to_assembler_name_hash): Handle clones.
20479         (unlink_from_assembler_name_hash): Likewise.
20480         (symtab_prevail_in_asm_name_hash, symtab_register_node,
20481         symtab_unregister_node, symtab_initialize_asm_name_hash,
20482         change_decl_assembler_name): Update.
20484 2013-05-09  Sofiane Naci  <sofiane.naci@arm.com>
20486         * config/aarch64/aarch64.md: New movtf split.
20487         (*movtf_aarch64): Update.
20488         (aarch64_movdi_tilow): Handle TF modes and rename to
20489         aarch64_movdi_<mode>low.
20490         (aarch64_movdi_tihigh): Handle TF modes and rename to
20491         aarch64_movdi_<mode>high
20492         (aarch64_movtihigh_di): Handle TF modes and rename to
20493         aarch64_mov<mode>high_di
20494         (aarch64_movtilow_di): Handle TF modes and rename to
20495         aarch64_mov<mode>low_di
20496         (aarch64_movtilow_tilow): Remove spurious whitespace.
20497         * config/aarch64/aarch64.c (aarch64_split_128bit_move): Handle TFmode
20498         splits.
20499         (aarch64_print_operand): Update.
20501 2013-05-09  Alan Modra  <amodra@gmail.com>
20503         * configure.ac (HAVE_AS_TLS): Enable tests for powerpcle and
20504         powerpc64le.
20505         * configure: Regenerate.
20507 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20509         * config/i386/mmx.md (*vec_extract* splitters): Simplify post-reload
20510         splitter preparation statements.
20511         * config/i386/sse.md (*vec_extract* splitters): Ditto.
20512         (*avx_vperm_broadcast_<mode>): Use adjust_address instead of
20513         adjust_address_nv.
20515 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20517         * gimple-ssa-strength-reduction.c (count_candidates): Change
20518         return value to int.
20519         (analyze_candidates_and_replace): Change type of length to int.
20521 2013-05-08  Uros Bizjak  <ubizjak@gmail.com>
20523         * config/i386/sse.md (PEXTR_MODE, PEXTR_MODEx): Remove.
20524         (*vec_extract<mode>): Use VI12_128 mode iterator.
20525         (*vec_extract<mode>_mem): Ditto.
20526         (*vec_extract*_mem splitters): Merge splitters using VI_128 mode
20527         attribute.
20529 2013-05-08  Diego Novillo  <dnovillo@google.com>
20531         PR bootstrap/54659
20533         Revert:
20534         2012-08-17  Diego Novillo  <dnovillo@google.com>
20536         PR bootstrap/54281
20537         * configure.ac: Add libintl.h to AC_CHECK_HEADERS list.
20538         * config.in: Regenerate.
20539         * configure: Regenerate.
20540         * intl.h: Always include libintl.h if HAVE_LIBINTL_H is set.
20542 2013-05-08  Jan Hubicka  <jh@suse.cz>
20544         PR lto/54095
20545         * cgraph.c (cgraph_make_node_local_1): Se unique_name.
20546         * cgraph.h (symtab_node_base): Add unique_name.
20547         * lto-cgraph.c (lto_output_node, lto_output_varpool_node,
20548         input_overwrite_node, input_varpool_node): Stream unique_name.
20549         * cgraphclones.c (cgraph_create_virtual_clone,
20550         cgraph_function_versioning): Set unique_name.
20551         * ipa.c (function_and_variable_visibility): Set unique_name.
20553 2013-05-08  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20555         * gimple-ssa-strength-reduction.c (find_phi_def): Revert former "fix."
20556         (alloc_cand_and_find_basis): Restrict conditional candidate
20557         processing to CAND_MULTs.
20559 2013-05-08  Jan Hubicka  <jh@suse.cz>
20561         PR lto/54095
20562         lto-symtab.c (lto_symtab_symbol_p): New function.
20563         (lto_symtab_resolve_can_prevail_p, lto_symtab_resolve_symbols,
20564         lto_symtab_resolve_symbols, lto_symtab_merge_decls_2,
20565         lto_symtab_merge_decls_1, lto_symtab_merge_cgraph_nodes_1):
20566         Skip static symbols.
20568 2013-05-08  Paolo Carlini  <paolo.carlini@oracle.com>
20570         PR tree-optimization/57200
20571         * tree-ssa-loop-niter.c (do_warn_aggressive_loop_optimizations):
20572         Only call inform if the preceding warning_at returns true.
20574 2013-05-07  Han Shen  <shenhan@google.com>
20576         * cfgexpand.c (record_or_union_type_has_array_p): New function.
20577         (expand_used_vars): Add logic handling '-fstack-protector-strong'.
20578         * common.opt (fstack-protector-strong): New option.
20579         * doc/cpp.texi (__SSP_STRONG__): New builtin "__SSP_STRONG__".
20580         * doc/invoke.texi (Optimization Options): Document
20581         "-fstack-protector-strong".
20582         * gcc.c (LINK_SSP_SPEC): Add 'fstack-protector-strong'.
20584 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20586         * config/mips/mips.c (mips_machine_reorg2): Return 0.
20588 2013-05-07  Vladimir Makarov  <vmakarov@redhat.com>
20590         * ira.c (update_equiv_regs): Add insn having equiv memory even if
20591         it is not lhs of the insn.
20592         (setup_reg_equiv): Remove insn having equiv memory which it is not
20593         lhs of the insn.
20594         * lra-constraints.c (process_address): Try to improve generation
20595         code for address base + disp.
20596         (lra_constraints): Make correct the code for checking insn setting
20597         up backward equivalence.  Remove insn only if it is in the init
20598         insn list.
20599         * lra-eliminations.c (update_reg_eliminate): Change return value.
20600         (lra_eliminate): Use the result.
20602 2013-05-07  Uros Bizjak  <ubizjak@gmail.com>
20604         * config/i386/sse.md (ssescalarnummask): New mode attribute.
20605         (PEXTR_MODE, PEXTR_MODEx): New mode iterators.
20606         (*vec_extract<mode>): Merge from *sse4_1_pextrb_memory and
20607         *sse4_1_pextrw_memory using PEXTR_MODE mode iterator. Handle
20608         register target operands.
20609         (*vec_extractv8hi_sse2): New pattern.
20610         (*vec_extractv16qi_zext): Rename from *sse4_1_pextrb_<mode>.
20611         (*vec_extractv8hi_zext): Rename from *sse2_pextrw_<mode>.
20612         (*vec_extract<mode>_mem): New insn and split pattern.
20614 2013-05-07  Christophe Lyon  <christophe.lyon@linaro.org>
20616         * config/arm/arm.c (arm_asan_shadow_offset): New function.
20617         (TARGET_ASAN_SHADOW_OFFSET): Define.
20618         * config/arm/linux-eabi.h (ASAN_CC1_SPEC): Define.
20619         (LINUX_OR_ANDROID_CC): Add ASAN_CC1_SPEC.
20621 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20623         * gimple-ssa-strength-reduction.c (MAX_INCR_VEC_LEN): New constant.
20624         (incr_vec_index): Return -1 if increment not found.
20625         (create_add_on_incoming_edge): Assert if increment not found.
20626         (record_increment): Limit number of increments recorded.
20627         (all_phi_incrs_profitable): Return false if an increment not found.
20628         (replace_profitable_candidates): Don't process increments that were
20629         not recorded.
20630         (analyze_candidates_and_replace): Limit size of incr_vec.
20632 2013-05-07  Richard Biener  <rguenther@suse.de>
20634         * calls.c (special_function_p): setjmp-like functions are leaf.
20635         * builtins.def (BUILT_IN_SETJMP): setjmp is leaf.
20636         * tree-inline.c (update_ssa_across_abnormal_edges): Remove assert.
20638 2013-05-07  Sofiane Naci  <sofiane.naci@arm.com>
20640         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): call splitter.
20641         (aarch64_simd_mov<mode>): New expander.
20642         (aarch64_simd_mov_to_<mode>low): New instruction pattern.
20643         (aarch64_simd_mov_to_<mode>high): Likewise.
20644         (aarch64_simd_mov_from_<mode>low): Likewise.
20645         (aarch64_simd_mov_from_<mode>high): Likewise.
20646         (aarch64_dup_lane<mode>): Update.
20647         (aarch64_dup_lanedi): New instruction pattern.
20648         * config/aarch64/aarch64-protos.h (aarch64_split_simd_move): New prototype.
20649         * config/aarch64/aarch64.c (aarch64_split_simd_move): New function.
20651 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20653         * gimple-ssa-strength-reduction.c (lazy_create_slsr_reg): Remove.
20654         (replace_mult_candidate): Remove unnecessary argument; remove
20655         unnecessary parameter from call to introduce_cast_before_cand.
20656         (replace_unconditional_candidate): Remove unnecessary parameter
20657         from call to replace_mult_candidate.
20658         (replace_conditional_candidate): Likewise.
20659         (insert_initializers): Use make_temp_ssa_name.
20660         (introduce_cast_before_cand): Remove unnecessary argument; use
20661         make_temp_ssa_name.
20662         (replace_one_candidate): Remove unnecessary argument; remove
20663         unnecessary parameter from calls to introduce_cast_before_cand.
20664         (replace_profitable_candidates): Remove unnecessary parameters
20665         from calls to replace_one_candidate.
20667 2013-05-07  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20669         * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
20670         phi def as possibly hiding a basis for a CAND_ADD whose operands
20671         have been commuted in the analysis.
20672         (alloc_cand_and_find_basis): Add parms to call to find_phi_def.
20674 2013-05-07  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
20676         * config/aarch64/aarch64.md
20677         (cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): Restrict the
20678         shift value between 0-4.
20680 2013-05-07  Richard Biener  <rguenther@suse.de>
20682         * double-int.h (rshift): New overload.
20683         * double-int.c (rshift): New function.
20684         * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Optimize.
20685         (create_reference_ops_from_ref): Remove.
20686         (vn_reference_insert): Use shared ops for constructing the
20687         reference and copy it.
20689 2013-05-07  Richard Biener  <rguenther@suse.de>
20691         PR middle-end/57190
20692         * tree-eh.c (sink_clobbers): Properly propagate
20693         SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
20695 2013-05-07  Jakub Jelinek  <jakub@redhat.com>
20697         PR tree-optimization/57149
20698         * tree-ssa-uninit.c (uninit_undefined_value_p): New inline.
20699         (can_skip_redundant_opnd, compute_uninit_opnds_pos,
20700         collect_phi_def_edges, execute_late_warn_uninitialized): Use
20701         uninit_undefined_value_p instead of ssa_undefined_value_p.
20703         PR debug/57184
20704         * expr.c (expand_expr_addr_expr_1): Handle COMPOUND_LITERAL_EXPR
20705         for modifier == EXPAND_INITIALIZER.
20707 2013-05-07  Anton Blanchard  <anton@samba.org>
20709         * configure.ac (HAVE_LD_LARGE_TOC): Use correct linker emulation
20710         for powerpc64 little endian.
20711         * configure: Regenerate.
20713 2013-05-06  Graham Stott  <grahams@btinternet.com>
20715         * expmed.c (init_expmed_rtl): Remove unused fields reg_fld, plus_fld,
20716         mult_fld, sdiv_fld1, udiv_fld1, sdiv_32_fld1, smod_32_fld1,
20717         wide_mult_fld1, wide_lshr_fld1, shift_fld1, shift_mult_fld1,
20718         shift_add_fld1, shift_sub0_fld1, shift_sub1_fld1.
20720 2013-05-06  Graham Stott  <grahams@btinternet.com>
20722         * gensupport.c (add_predicate_code): Also exclude SCRATCH from rtx
20723         codes which allow non-lvalues.
20725 2013-05-06  Marc Glisse  <marc.glisse@inria.fr>
20727         * tree.c (integer_all_onesp) <COMPLEX_CST>: Test that both
20728         components are all 1s.
20729         (integer_minus_onep): New function.
20730         * tree.h (integer_minus_onep): Declare it.
20731         * fold-const.c (fold_binary_loc) <MULT_EXPR>: Test
20732         integer_minus_onep instead of integer_all_onesp.
20734 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20736         PR target/52933
20737         * config/sh/sh.md (*cmp_div0s_0, *cmp_div0s_1, *movsicc_div0s): Add
20738         variations of these patterns.
20740 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20742         * config/i386/i386.md (isa): Add x64_sse4 member.
20743         (enabled): Handle x64_sse4.
20744         (*movdi_internal): Add *x->?r alternative to emit pextrq $0,%xmm,%reg
20745         instruction for 64bit SSE4_1 targets.  Update insn attributes.
20746         (*movsi_internal): Add *x->?r alternative to emit pextrd $0,%xmm,%reg
20747         instruction for SSE4_1 targets.  Update insn attributes.
20748         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20749         with *sse4_1_pextrd and *sse4_1_pextrq having const_0 selector.
20750         (*vec_extractv2di_1): Merge with *sse4_1_pextrq having
20751         const_1 selector.
20752         (*vec_extractv4si): Rename from *sse4_1_pextrd.
20753         (*vec_extractv4si_zext): Rename from *sse4_1_pextrd_zext.
20754         (*vec_extract<ssevecmodelower>_0 splitters): Merge splitters together.
20756 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20758         PR target/57108
20759         * config/sh/sh.md (tstsi_t_zero_extract_eq): Use QIHISIDI mode iterator.
20761 2013-05-06  Maxim Kuznetsov  <maks.kuznetsov@gmail.com>
20763         * final.c (do_assembler_dialects): Don't handle curly braces and
20764         vertical bar escaped by % as dialect delimiters.
20765         (output_asm_insn): Print curly braces and vertical bar if escaped
20766         by % and ASSEMBLER_DIALECT defined.
20767         * doc/tm.texi.in (ASSEMBLER_DIALECT): Document new standard escapes.
20768         * doc/tm.texi: Regenerated.
20770 2013-05-06  Steven Bosscher  <steven@gcc.gnu.org>
20772         * config/mips/mips.c: Include tree-pass.h.
20773         (mips_reorg): Split in pre- and post-dbr_schedule parts.
20774         (mips_machine_reorg2): Move mips_reorg post-dbr_schedule parts here.
20775         (pass_mips_machine_reorg2): New machine specific pass.
20776         (insert_pass_mips_machine_reorg2): New pass plugin definition.
20777         (mips_option_override): Register the new pass.
20778         * rtl.h (cleanup_barriers): Remove prototype.
20779         (dbr_schedule): Likewise.
20780         * jump.c (cleanup_barriers): Make static.
20781         * reorg.c (dbr_schedule): Likewise.
20783 2013-05-06  Richard Biener  <rguenther@suse.de>
20785         PR tree-optimization/57185
20786         * tree-parloops.c (add_field_for_reduction): Handle anonymous
20787         SSA names properly.
20789 2013-05-06  Uros Bizjak  <ubizjak@gmail.com>
20791         PR target/57106
20792         * config/i386/i386.c (add_parameter_dependencies): Add dependence
20793         between "first_arg" and "insn", not "last" and "insn".
20795 2013-05-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20797         * gimple-ssa-strength-reduction.c (slsr_process_phi): Re-enable.
20798         (find_candidates_in_block): Re-enable slsr_process_phi.
20799         (create_phi_basis): Fix double counting of candidate adjustment.
20801 2013-05-06  Richard Biener  <rguenther@suse.de>
20803         PR middle-end/57147
20804         * tree-cfg.c (gimple_purge_dead_abnormal_call_edges): If
20805         the edge is also fallthru, preserve it and just clear the
20806         abnormal flag.
20807         * tree-cfgcleanup.c (remove_fallthru_edge): If the edge is
20808         also complex, preserve that and just clear the fallthru flag.
20809         * tree-inline.c (update_ssa_across_abnormal_edges): Also
20810         update virtual operands.
20812 2013-05-06  Alan Modra  <amodra@gmail.com>
20814         * config/rs6000/linux.h (DEFAULT_ASM_ENDIAN): Define.
20815         (LINK_OS_LINUX_EMUL): Use ENDIAN_SELECT.
20816         * config/rs6000/linux64.h (DEFAULT_ASM_ENDIAN): Define.
20817         * config/rs6000/sysv4le.h (DEFAULT_ASM_ENDIAN): Define.
20818         (LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20819         * config/rs6000/sysv4.h (DEFAULT_ASM_ENDIAN): Define as -mbig.
20821 2013-05-06  Alan Modra  <amodra@gmail.com>
20823         * config/rs6000/sysv4.h (ENDIAN_SELECT): Define, extracted from
20824         (ASM_SPEC): ..here.  Emit DEFAULT_ASM_ENDIAN too.
20825         (DEFAULT_ASM_ENDIAN): Define.
20826         (CC1_SPEC, LINK_TARGET_SPEC): Use ENDIAN_SELECT.
20827         * config/rs6000/linux64.h (ASM_SPEC32): Remove endian options.
20828         Update -K PIC clause from sysv4.h.
20829         (ASM_SPEC_COMMON): Use ENDIAN_SELECT.
20830         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Likewise.
20832 2013-05-06  Alan Modra  <amodra@gmail.com>
20834         * config/rs6000/rs6000.md (bswapdi 2nd splitter): Don't swap words
20835         twice for little-endian.
20836         (ashrdi3_no_power, ashrdi3): Support little-endian.
20838 2013-05-06  Oleg Endo  <olegendo@gcc.gnu.org>
20840         PR target/55303
20841         * config/sh/sh.c (sh_rtx_costs): Handle SMIN and SMAX cases.
20842         * config/sh/sh.md (*clips, uminsi3, *clipu, clipu_one): New insns and
20843         related expanders.
20844         * config/sh/iterators.md (SMIN_SMAX): New code iterator.
20845         * config/sh/predicates.md (arith_reg_or_0_or_1_operand,
20846         clips_min_const_int, clips_max_const_int, clipu_max_const_int):
20847         New predicates.
20849 2013-05-05  Steven Bosscher  <steven@gcc.gnu.org>
20850             John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
20852         * config.gcc (hppa*-*-*): Remove MASK_BIG_SWITCH from CPU default.
20853         * config/pa/pa.opt: Make mbig-switch a no-op.
20854         * config/pa/pa.h (TARGET_DEFAULT): Remove MASK_BIG_SWITCH.
20855         (CASE_VECTOR_MODE): Always return SImode.
20856         (ASM_OUTPUT_ADDR_VEC_ELT, ASM_OUTPUT_ADDR_DIFF_ELT): Remove code
20857         for the !TARGET_BIG_SWITCH case.
20858         * config/pa/pa-linux.h: Likewise.
20859         * config/pa/pa-openbsd.h: Likewise.
20860         * config/pa/pa-hpux.h: Define TARGET_DEFAULT to 0.
20861         * config/pa/pa.md (short_jump): Remove define_insn.
20862         (casesi): Remove code for the !TARGET_BIG_SWITCH case.
20863         (casesi0): Remove define_insn.
20864         (type): Remove btable_branch.
20865         (pa_combine_type): Likewise.
20866         (in_nullified_branch_delay): Likewise.
20867         (in_call_delay): Likewise.
20868         (define_delay): Likewise.
20869         (define_insn_reservation "Z3"): Likewise.
20870         (define_insn_reservation "Z4"): Likewise.
20871         * config/pa/pa.c  (pa_reorg): Remove code for !TARGET_BIG_SWITCH.
20872         (pa_adjust_insn_length): Remove adjustment for btable branches.
20873         * doc/invoke.texi (HPPA Options): Delete documentation for mbig-switch
20874         and mno-big-switch
20876 2013-05-05  Uros Bizjak  <ubizjak@gmail.com>
20878         * config/i386/sse.md (*vec_extract<ssevecmodelower>_0): Merge
20879         from sse2_stored and *sse2_storeq_rex64 using SWI48 mode iterator.
20880         Add m->r,x alternatives.
20881         (*vec_extract<ssevecmodelower>_0 splitters): Merge V2DI and V4SI
20882         splitters using SWI48x mode iterator.
20883         (*vec_extract_v2di_0_sse): Rename from *sse2_storeq.  Disable for
20884         TARGET_64BIT.  Add m->x alternative.
20885         (*vec_extractv4si_mem): Rename from *vec_ext_v4si_mem.
20886         Add o->x alternative.  Enable for TARGET_SSE.
20887         (sse_storeq): Remove expander.
20888         (*vec_extractv2di_1): Enable for TARGET_SSE. Split alternatives
20889         with memory input operand.
20890         (*vec_extractv2di_1 splitter): New.
20891         (*vec_extractv4sf_mem): Rename from *vec_extract_v4sf_mem.
20892         * config/i386/i386.md (ssevecmodelower): New mode attribute.
20894 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20896         * config/rs6000/rs6000.c (INT_P): Reformat.  Delete obsolete comment.
20897         (INT_LOWPART): Delete.
20898         (extract_MB): Adjust.
20899         (extract_ME): Adjust.
20900         (print_operand): Adjust.
20902 2013-05-04  Segher Boessenkool  <segher@kernel.crashing.org>
20904         * config/rs6000/predicates.md (reg_or_add_cint_operand,
20905         reg_or_sub_cint_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20906         (reg_or_logical_cint_operand, easy_fp_constant,
20907         logical_const_operand): Delete "CONST_DOUBLE" case.
20908         * config/rs6000/rs6000.c (num_insns_constant_wide): Delete
20909         "HOST_BITS_PER_WIDE_INT == 64" test.
20910         (num_insns_constant): Ditto.  Delete CONST_DOUBLE DImode/VOIDmode case.
20911         (build_mask64_2_operands): Delete "HOST_BITS_PER_WIDE_INT >= 64" test.
20912         (rs6000_emit_set_const): Delete CONST_DOUBLE case.
20913         (rs6000_emit_set_long_const): Delete "HOST_BITS_PER_WIDE_INT >= 64"
20914         test.
20915         (includes_rldic_lshift_p, includes_rldicr_lshift_p): Delete
20916         CONST_DOUBLE DImode/VOIDmode case.
20917         (INT_P, INT_LOWPART): Delete CONST_DOUBLE case.
20918         (print_operand): Delete "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20919         CONST_DOUBLE VOIDmode case.
20920         (output_toc): Delete "HOST_BITS_PER_WIDE_INT == 32" case.
20921         (rs6000_rtx_costs): Delete CONST_DOUBLE DImode/VOIDmode case.
20922         * config/rs6000/rs6000.md (iordi3, xordi3, splitter for these):
20923         Delete CONST_DOUBLE case.
20924         (splitters for mov FMOVE64 const_double): Delete
20925         "HOST_BITS_PER_WIDE_INT == 32" case.  Delete
20926         "HOST_BITS_PER_WIDE_INT >= 64" test.
20927         (splitter for mov DI const_int): Delete "HOST_BITS_PER_WIDE_INT == 32"
20928         case.
20929         (mov DI const_double): Delete.
20931 2013-05-04  Jakub Jelinek  <jakub@redhat.com>
20933         * combine.c (combine_simplify_rtx) <case SUBREG>: If nonzero_bits
20934         on op shows all bits zero in mode of a lowpart subreg, return zero.
20936 2013-05-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
20938         PR target/57150
20939         * config/rs6000/rs6000.h (HARD_REGNO_CALLER_SAVE_MODE): Use DFmode
20940         to save TFmode registers and DImode to save TImode registers for
20941         caller save operations.
20942         (HARD_REGNO_CALL_PART_CLOBBERED): TFmode and TDmode do not need to
20943         mark being partially clobbered since they only use the first
20944         double word.
20946         * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): TFmode
20947         and TDmode only use the upper 64-bits of each VSX register.
20949 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20951         * gimple-ssa-strength-reduction.c (slsr_process_phi): Disable.
20952         (find_candidates_in_block): Disable slsr_process_phi.
20954 2013-05-03  Guozhi Wei  <carrot@google.com>
20956         * coverage.c (coverage_obj_init): Move the construction of gcov
20957         constructor to ...
20958         (build_init_ctor): ... here.
20960 2013-05-03  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
20962         * gimple-ssa-strength-reduction.c (cand_kind): Add CAND_PHI.
20963         (slsr_cand_d): Redefine def_phi.
20964         (stride_status, phi_adjust_status, count_phis_status): New enums.
20965         (find_phi_def): New.
20966         (find_basis_for_base_expr): New.
20967         (find_basis_for_candidate): Handle hidden bases.
20968         (alloc_cand_and_find_basis): Handle phi candidates.
20969         (slsr_process_phi): New.
20970         (create_mul_ssa_cand): Exclude phi base candidates; use integer_onep.
20971         (create_mul_imm_cand): Likewise.
20972         (create_add_ssa_cand): Exclude phi base candidates.
20973         (create_add_imm_cand): Likewise.
20974         (slsr_process_cast): Likewise.
20975         (slsr_process_copy): Likewise.
20976         (find_candidates_in_block): Handle phi candidates.
20977         (dump_candidate): Likewise.
20978         (unconditional_cands): Delete.
20979         (unconditional_cands_with_known_stride_p): Delete.
20980         (phi_dependent_cand_p): New.
20981         (cand_increment): Handle phi-dependent candidates.
20982         (replace_dependent): Delete.
20983         (replace_mult_candidate): New.
20984         (replace_unconditional_candidate): New.
20985         (incr_vec_index): Move to avoid forward reference.
20986         (create_add_on_incoming_edge): New.
20987         (create_phi_basis): New.
20988         (replace_dependents): Delete.
20989         (replace_conditional_candidate): New.
20990         (phi_add_costs): New.
20991         (replace_uncond_cands_and_profitable_phis): New.
20992         (record_increment): Handle phi adjustments.
20993         (record_phi_increments): New.
20994         (record_increments): Handle phi adjustments.
20995         (phi_incr_cost): New.
20996         (lowest_cost_path): Handle phis.
20997         (total_savings): Likewise.
20998         (analyze_increments): Likewise.
20999         (ncd_with_phi): New.
21000         (ncd_of_cand_and_phis): New.
21001         (nearest_common_dominator_for_cands): Handle phi increments.
21002         (all_phi_incrs_profitable): New.
21003         (replace_profitable_candidates): Handle phi-dependent candidates.
21004         (analyze_candidates_and_replace): Likewise.
21006 2013-05-03  Teresa Johnson  <tejohnson@google.com>
21008         PR bootstrap/57154
21009         * sched-rgn.c (compute_dom_prob_ps): Ensure accumulated probabilities
21010         do not exceed REG_BR_PROB_BASE.
21012 2013-05-03  Jeff Law  <law@redhat.com>
21014         PR tree-optimization/57144
21015         * tree-vrp.c (simplify_cond_using_ranges): Verify the constant
21016         operand of the condition will bit into the new type when eliminating
21017         a cast feeding a condition.
21019 2013-05-03  Jakub Jelinek  <jakub@redhat.com>
21021         PR rtl-optimization/57130
21022         * combine.c (make_compound_operation) <case SUBREG>: Pass SET instead
21023         of COMPARE as in_code to the recursive call if needed.
21025 2013-05-03  Uros Bizjak  <ubizjak@gmail.com>
21027         * config/i386/i386.md (isa): Add x64_sse4_noavx and x64_avx members.
21028         (enabled): Handle new members.
21029         * config/i386/sse.md (*vec_concatv2si): Merge from
21030         *vec_concatv2si_sse2 and vec_concatv2si_sse.
21031         (vec_concatv2di): Merge with *vec_concatv2di_rex64.
21033 2013-05-03  Joern Rennecke  <joern.rennecke@embecosm.com>
21035         PR tree-optimization/57027
21036         * tree-ssa-math-opts.c (convert_mult_to_fma): When checking
21037         for fnms opportunity, check we got the prerequisite kind
21038         of tree / gimple before using accessor functions.
21040 2013-05-03  Richard Biener  <rguenther@suse.de>
21042         * double-int.h (lshift): New overload without precision
21043         and arith argument.
21044         (operator *=, operator +=, operator -=): Move ...
21045         * double-int.c (operator *=, operator +=, operator -=): ... here
21046         and implement more efficiently.
21047         (mul_double_with_sign): Remove.
21048         (lshift_double): Adjust to take unsinged shift argument, push
21049         dispatching code to callers.
21050         (mul_double_wide_with_sign): Add early out for callers that
21051         are not interested in high parts or overflow.
21052         (lshift): New function.
21053         (lshift, rshift, alshift, arshift, llshift, lrshift): Add
21054         dispatch code here.
21055         (lrotate, rrotate): Use logical shifts.
21056         * expr.c (get_inner_reference): Use lshift.
21057         * fixed-value.c (do_fixed_divide): Likewise.
21058         * tree-dfa.c (get_ref_base_and_extent): Likewise.
21059         * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
21060         (indirect_refs_may_alias_p): Likewise.
21061         (stmt_kills_ref_p_1): Likewise.
21063 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
21065         * config/aarch64/aarch64-simd.md (simd_fabd): Correct the description.
21067 2013-05-03  Vidya Praveen  <vidyapraveen@arm.com>
21069         * config/aarch64/aarch64-simd.md (*fabd_scalar<mode>3): Support
21070         scalar form of FABD instruction.
21072 2013-05-02  Vladimir Makarov  <vmakarov@redhat.com>
21074         * lra-constraints.c (process_alt_operands): Add checking alt
21075         number to choose the best alternative.
21077 2013-05-02  Richard Biener  <rguenther@suse.de>
21079         * tree-eh.c (cleanup_empty_eh_merge_phis): Remove rename_virts
21080         bitmap and its handling.
21081         (pass_cleanup_eh): Set todo_flags_finish to TODO_verify_ssa.
21083 2013-05-02  Richard Biener  <rguenther@suse.de>
21085         PR middle-end/57140
21086         * tree-inline.c (copy_loops): Properly handle removed loops.
21087         (copy_cfg_body): Mark destination loops for fixup if source
21088         loops needed fixup.
21090 2013-05-02  Greta Yorsh  <Greta.Yorsh@arm.com>
21092         PR target/56732
21093         * config/arm/arm.c (arm_expand_epilogue): Check really_return before
21094         generating simple_return for naked functions.
21096 2013-05-02  Martin Jambor  <mjambor@suse.cz>
21098         PR middle-end/56988
21099         * ipa-prop.h (ipa_agg_replacement_value): New flag by_ref.
21100         * ipa-cp.c (ipa_get_indirect_edge_target_1): Also check that by_ref
21101         flags match.
21102         (find_aggregate_values_for_callers_subset): Fill in the by_ref flag of
21103         ipa_agg_replacement_value structures.
21104         (known_aggs_to_agg_replacement_list): Likewise.
21105         * ipa-prop.c (write_agg_replacement_chain): Stream by_ref flag.
21106         (read_agg_replacement_chain): Likewise.
21107         (ipcp_transform_function): Also check that by_ref flags match.
21109 2013-05-02  Richard Biener  <rguenther@suse.de>
21111         * graphds.h (struct graph): Add obstack member.
21112         * graphds.c (new_graph): Initialize obstack and allocate
21113         vertices from it.
21114         (add_edge): Allocate edge from the obstack.
21115         (free_graph): Free the obstack instead of all edges and vertices.
21117 2013-05-02  Teresa Johnson  <tejohnson@google.com>
21119         * loop-unswitch.c (unswitch_loop): Use helper routines with rounding
21120         divides.
21121         * cfg.c (update_bb_profile_for_threading): Ditto.
21122         * tree-inline.c (copy_bb): Ditto.
21123         (copy_edges_for_bb): Ditto.
21124         (initialize_cfun): Ditto.
21125         (copy_cfg_body): Ditto.
21126         (expand_call_inline): Ditto.
21127         * ipa-inline-analysis.c (estimate_edge_size_and_time): Ditto.
21128         (estimate_node_size_and_time): Ditto.
21129         (inline_merge_summary): Ditto.
21130         * cgraphclones.c (cgraph_clone_edge): Ditto.
21131         (cgraph_clone_node): Ditto.
21132         * sched-rgn.c (compute_dom_prob_ps): Ditto.
21133         (compute_trg_info): Ditto.
21135 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21137         * config/aarch64/aarch64.md (movsi_aarch64): Only allow to/from
21138         S reg when fp attribute set.
21139         (movdi_aarch64): Only allow to/from D reg when fp attribute set.
21141 2013-05-02  Ian Bolton  <ian.bolton@arm.com>
21143         * config/aarch64/aarch64.md (*and_one_cmpl<mode>3_compare0):
21144         New pattern.
21145         (*and_one_cmplsi3_compare0_uxtw): Likewise.
21146         (*and_one_cmpl_<SHIFT:optab><mode>3_compare0): Likewise.
21147         (*and_one_cmpl_<SHIFT:optab>si3_compare0_uxtw): Likewise.
21149 2013-05-02  Richard Biener  <rguenther@suse.de>
21151         * tree-scalar-evolution.c (scev_info_hasher): Remove.
21152         (struct instantiate_cache_entry): New type.
21153         (struct instantiate_cache_entry_hasher): New hashtable descriptor.
21154         (struct instantiate_cache_type): New type.
21155         (set_instantiated_value, get_instantiated_value): Remove.
21156         (get_instantiated_value_entry): New function.
21157         (instantiate_scev_name): Use the new cache and adjust.
21158         (instantiate_scev_poly): Adjust.
21159         (instantiate_scev_binary): Likewise.
21160         (instantiate_array_ref): Likewise.
21161         (instantiate_scev_convert): Likewise.
21162         (instantiate_scev_not): Likewise.
21163         (instantiate_scev_3): Likewise.
21164         (instantiate_scev_2): Likewise.
21165         (instantiate_scev_r): Likewise.
21166         (instantiate_scev): Likewise.
21167         (resolve_mixers): Likewise.
21169 2013-05-01  Vladimir Makarov  <vmakarov@redhat.com>
21171         PR target/57091
21172         * lra-constraints.c (best_small_class_operands_num): Remove.
21173         (process_alt_operands): Remove small_class_operands_num.  Take
21174         small classes operands into losers and only if the operand is not
21175         matched.  Modify debugging output.
21176         (curr_insn_transform): Remove best_small_class_operands_num.
21177         Print insn name.
21179 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21181         * config/aarch64/aarch64-builtins.c
21182         (aarch64_gimple_fold_builtin.c): Fold more modes for reduc_splus_.
21183         * config/aarch64/aarch64-simd-builtins.def
21184         (reduc_splus_): Add new modes.
21185         (reduc_uplus_): New.
21186         * config/aarch64/aarch64-simd.md (aarch64_addvv4sf): Remove.
21187         (reduc_uplus_v4sf): Likewise.
21188         (reduc_splus_v4sf): Likewise.
21189         (aarch64_addv<mode>): Likewise.
21190         (reduc_uplus_<mode>): Likewise.
21191         (reduc_splus_<mode>): Likewise.
21192         (aarch64_addvv2di): Likewise.
21193         (reduc_uplus_v2di): Likewise.
21194         (reduc_splus_v2di): Likewise.
21195         (aarch64_addvv2si): Likewise.
21196         (reduc_uplus_v2si): Likewise.
21197         (reduc_splus_v2si): Likewise.
21198         (reduc_<sur>plus_<mode>): New.
21199         (reduc_<sur>plus_v2di): Likewise.
21200         (reduc_<sur>plus_v2si): Likewise.
21201         (reduc_<sur>plus_v4sf): Likewise.
21202         (aarch64_addpv4sf): Likewise.
21203         * config/aarch64/arm_neon.h
21204         (vaddv<q>_<s,u,f><8, 16, 32, 64): Rewrite using builtins.
21205         * config/aarch64/iterators.md (unspec): Remove UNSPEC_ADDV,
21206         add UNSPEC_SADDV, UNSPEC_UADDV.
21207         (SUADDV): New.
21208         (sur): Add UNSPEC_SADDV, UNSPEC_UADDV.
21210 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21212         * config/aarch64/arm_neon.h
21213         (v<max,min><nm><q><v>_<sfu><8, 16, 32, 64>): Rewrite using builtins.
21215 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21217         * config/aarch64/aarch64-builtins
21218         (aarch64_gimple_fold_builtin): Fold reduc_<su><maxmin>_ builtins.
21220 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21222         * config/aarch64/aarch64-simd-builtins.def
21223         (reduc_smax_): New.
21224         (reduc_smin_): Likewise.
21225         (reduc_umax_): Likewise.
21226         (reduc_umin_): Likewise.
21227         (reduc_smax_nan_): Likewise.
21228         (reduc_smin_nan_): Likewise.
21229         (fmax): Remove.
21230         (fmin): Likewise.
21231         (smax): Update for V2SF, V4SF and V2DF modes.
21232         (smin): Likewise.
21233         (smax_nan): New.
21234         (smin_nan): Likewise.
21235         * config/aarch64/aarch64-simd.md (<maxmin><mode>3): Rename to...
21236         (<su><maxmin><mode>3): ...This, refactor.
21237         (s<maxmin><mode>3): New.
21238         (<maxmin_uns><mode>3): Likewise.
21239         (reduc_<maxmin_uns>_<mode>): Refactor.
21240         (reduc_<maxmin_uns>_v4sf): Likewise.
21241         (reduc_<maxmin_uns>_v2si): Likewise.
21242         (aarch64_<fmaxmin><mode>: Remove.
21243         * config/aarch64/arm_neon.h (vmax<q>_f<32,64>): Rewrite to use
21244         new builtin names.
21245         (vmin<q>_f<32,64>): Likewise.
21246         * config/iterators.md (unspec): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21247         (FMAXMIN): New.
21248         (su): Add mappings for smax, smin, umax, umin.
21249         (maxmin): New.
21250         (FMAXMINV): Add UNSPEC_FMAXNMV, UNSPEC_FMINNMV.
21251         (FMAXMIN): Rename as...
21252         (FMAXMIN_UNS): ...This.
21253         (maxminv): Remove.
21254         (fmaxminv): Likewise.
21255         (fmaxmin): Likewise.
21256         (maxmin_uns): New.
21257         (maxmin_uns_op): Likewise.
21259 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21261         * config/aarch64/arm_neon.h
21262         (vac<ge, gt><sd>_f<32, 64>): Rename to...
21263         (vca<ge, gt><sd>_f<32, 64>): ...this, reimpliment in C.
21264         (vca<ge, gt, lt, le><q>_f<32, 64>): Reimpliment in C.
21266 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21268         * config/aarch64/aarch64-simd.md (*aarch64_fac<optab><mode>): New.
21269         * config/aarch64/iterators.md (FAC_COMPARISONS): New.
21271 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21273         * config/aarch64/aarch64-simd.md
21274         (vcond<mode>_internal): Handle special cases for constant masks.
21275         (vcond<mode><mode>): Allow nonmemory_operands for outcome vectors.
21276         (vcondu<mode><mode>): Likewise.
21277         (vcond<v_cmp_result><mode>): New.
21279 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21281         * config/aarch64/aarch64-builtins.c (BUILTIN_VALLDI): Define.
21282         (aarch64_fold_builtin): Add folding for cm<eq,ge,gt,tst>.
21283         * config/aarch64/aarch64-simd-builtins.def
21284         (cmeq): Update to BUILTIN_VALLDI.
21285         (cmgt): Likewise.
21286         (cmge): Likewise.
21287         (cmle): Likewise.
21288         (cmlt): Likewise.
21289         * config/aarch64/arm_neon.h
21290         (vc<eq, lt, le, gt, ge, tst><z><qsd>_<fpsu><8,16,32,64>): Remap
21291         to builtins or C as appropriate.
21293 2013-05-01  James Greenhalgh  <james.greenhalgh@arm.com>
21295         * config/aarch64/aarch64-simd-builtins.def (cmhs): Rename to...
21296         (cmgeu): ...This.
21297         (cmhi): Rename to...
21298         (cmgtu): ...This.
21299         * config/aarch64/aarch64-simd.md
21300         (simd_mode): Add SF.
21301         (aarch64_vcond_internal): Use new names for unsigned comparison insns.
21302         (aarch64_cm<optab><mode>): Rewrite to not use UNSPECs.
21303         * config/aarch64/aarch64.md (*cstore<mode>_neg): Rename to...
21304         (cstore<mode>_neg): ...This.
21305         * config/aarch64/iterators.md
21306         (VALLF): new.
21307         (unspec): Remove UNSPEC_CM<EQ, LE, LT, GE, GT, HS, HI, TST>.
21308         (COMPARISONS): New.
21309         (UCOMPARISONS): Likewise.
21310         (optab): Add missing comparisons.
21311         (n_optab): New.
21312         (cmp_1): Likewise.
21313         (cmp_2): Likewise.
21314         (CMP): Likewise.
21315         (cmp): Remove.
21316         (VCMP_S): Likewise.
21317         (VCMP_U): Likewise.
21318         (V_cmp_result): Add DF, SF modes.
21319         (v_cmp_result): Likewise.
21320         (v): Likewise.
21321         (vmtype): Likewise.
21322         * config/aarch64/predicates.md (aarch64_reg_or_fp_zero): New.
21324 2013-05-01  Greta Yorsh  <Greta.Yorsh@arm.com>
21326         * config/arm/thumb2.md (thumb2_smaxsi3,thumb2_sminsi3): Convert
21327         define_insn to define_insn_and_split.
21328         (thumb32_umaxsi3,thumb2_uminsi3): Likewise.
21329         (thumb2_negdi2,thumb2_abssi2,thumb2_neg_abssi2): Likewise.
21330         (thumb2_mov_scc,thumb2_mov_negscc,thumb2_mov_notscc): Likewise.
21331         (thumb2_movsicc_insn,thumb2_and_scc,thumb2_ior_scc): Likewise.
21332         (thumb2_negscc): Likewise.
21334 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21336         * config/arm/thumb2.md (thumb2_incscc, thumb2_decscc): Delete.
21338 2013-04-30  Greta Yorsh  <Greta.Yorsh@arm.com>
21340         * config/arm/thumb2.md: Remove trailing whitespaces.
21342 2013-04-30  Richard Sandiford  <rsandifo@linux.vnet.ibm.com>
21344         * explow.c (plus_constant): Pass "mode" to immed_double_int_const.
21345         Use gen_int_mode rather than GEN_INT.
21347 2013-04-30  H.J. Lu  <hongjiu.lu@intel.com>
21349         * value-prof.c (stream_in_histogram_value): Remove the strayed
21350         debug_gimple_stmt.
21352 2013-04-30  Richard Biener  <rguenther@suse.de>
21354         PR middle-end/57122
21355         * cfghooks.c (split_edge): Properly check for the loop latch edge.
21357 2013-04-30  Richard Biener  <rguenther@suse.de>
21359         PR middle-end/57107
21360         * tree-eh.c (sink_clobbers): Preserve virtual SSA form.
21362 2013-04-30  Andrey Belevantsev  <abel@ispras.ru>
21364         PR rtl-optimization/56957
21365         PR rtl-optimization/57105
21366         * sel-sched.c (move_op_orig_expr_found): Remove insn_emitted
21367         variable.  Use just INSN_UID for determining whether an insn
21368         should be only disconnected from the insn stream.
21369         * sel-sched-ir.h (EXPR_WAS_CHANGED): Remove.
21371 2013-04-30  Jakub Jelinek  <jakub@redhat.com>
21373         PR tree-optimization/57104
21374         * tsan.c (instrument_expr): Don't instrument accesses to
21375         DECL_HARD_REGISTER VAR_DECLs.
21377 2013-04-30  Richard Biener  <rguenther@suse.de>
21379         * function.h (loops_for_fn): New inline function.
21380         (set_loops_for_fn): Likewise.
21381         * cfgloop.h (place_new_loop): Add struct function parameter.
21382         (get_loop): Likewise.
21383         (get_loops): Likewise.
21384         (number_of_loops): Likewise.
21385         (fel_next): Adjust.
21386         (fel_init): Likewise.
21387         * cfg.c (get_loop_copy): Adjust.
21388         * cfgloop.c (flow_loops_dump): Likewise.
21389         (record_loop_exits): Likewise.
21390         (verify_loop_structure): Likewise.
21391         * cfgloopanal.c (mark_irreducible_loops): Likewise.
21392         (estimate_reg_pressure_cost): Likewise.
21393         (mark_loop_exit_edges): Likewise.
21394         * cfgloopmanip.c (place_new_loop): Likewise.
21395         (add_loop): Likewise.
21396         (duplicate_loop): Likewise.
21397         * graph.c (draw_cfg_nodes): Likewise.
21398         * graphite-clast-to-gimple.c (translate_clast_user): Likewise.
21399         * graphite-sese-to-poly.c (build_scop_scattering): Likewise.
21400         (extract_affine_chrec): Likewise.
21401         (build_scop_iteration_domain): Likewise.
21402         * graphite.c (graphite_initialize): Likewise.
21403         * ira-build.c (create_loop_tree_nodes): Likewise.
21404         (more_one_region_p): Likewise.
21405         (rebuild_regno_allocno_maps): Likewise.
21406         (mark_loops_for_removal): Likewise.
21407         (mark_all_loops_for_removal): Likewise.
21408         (remove_unnecessary_regions): Likewise.
21409         (ira_build): Likewise.
21410         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
21411         * loop-init.c (fix_loop_structure): Likewise.
21412         (gate_rtl_move_loop_invariants): Likewise.
21413         (gate_rtl_unswitch): Likewise.
21414         (gate_rtl_unroll_and_peel_loops): Likewise.
21415         (rtl_doloop): Likewise.
21416         * lto-streamer-in.c (input_cfg): Likewise.
21417         * lto-streamer-out.c (output_cfg): Likewise.
21418         * modulo-sched.c (sms_schedule): Likewise.
21419         * predict.c (tree_estimate_probability): Likewise.
21420         (tree_estimate_probability_driver): Likewise.
21421         (estimate_loops): Likewise.
21422         * tree-cfg.c (fixup_loop_arrays_after_move): Likewise.
21423         (move_sese_region_to_fn): Likewise.
21424         (debug_loop_num): Likewise.
21425         * tree-chrec.c (chrec_evaluate): Likewise.
21426         (hide_evolution_in_other_loops_than_loop): Likewise.
21427         (chrec_component_in_loop_num): Likewise.
21428         (reset_evolution_in_loop): Likewise.
21429         (evolution_function_is_invariant_rec_p): Likewise.
21430         * tree-if-conv.c (main_tree_if_conversion): Likewise.
21431         * tree-inline.c (copy_loops): Likewise.
21432         (copy_cfg_body): Likewise.
21433         (tree_function_versioning): Likewise.
21434         * tree-loop-distribution.c (rdg_flag_loop_exits): Likewise.
21435         * tree-scalar-evolution.c (chrec_contains_symbols_defined_in_loop):
21436         Likewise.
21437         (add_to_evolution_1): Likewise.
21438         (scev_const_prop): Likewise.
21439         * tree-scalar-evolution.h (get_chrec_loop): Likewise.
21440         * tree-ssa-loop-ch.c (copy_loop_headers): Likewise.
21441         * tree-ssa-loop-im.c (analyze_memory_references): Likewise.
21442         (tree_ssa_lim_initialize): Likewise.
21443         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Likewise.
21444         (verify_loop_closed_ssa): Likewise.
21445         * tree-ssa-loop.c  (tree_ssa_loop_init): Likewise.
21446         (tree_ssa_loop_im): Likewise.
21447         (tree_ssa_loop_unswitch): Likewise.
21448         (tree_vectorize): Likewise.
21449         (check_data_deps): Likewise.
21450         (tree_ssa_loop_ivcanon): Likewise.
21451         (tree_ssa_loop_bounds): Likewise.
21452         (tree_complete_unroll): Likewise.
21453         (tree_complete_unroll_inner): Likewise.
21454         (tree_parallelize_loops): Likewise.
21455         (tree_ssa_loop_prefetch): Likewise.
21456         (tree_ssa_loop_ivopts): Likewise.
21457         * tree-ssa.c (execute_update_addresses_taken): Liekwise.
21458         * tree-vectorizer.c (vectorize_loops): Likewise.
21460 2013-04-29  Mike Frysinger  <vapier@gentoo.org>
21462         * config/arm/bpabi.h (EABI_LINK_SPEC): Define.
21463         (BPABI_LINK_SPEC): Use new EABI_LINK_SPEC.
21464         * config/arm/linux-eabi.h (LINK_SPEC): Replace BE8_LINK_SPEC
21465         with EABI_LINK_SPEC.
21467 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21469         PR target/44578
21470         * config/i386/i386.md (*zero_extendsidi2): Add "!" to m->?*y
21471         alternative.
21473 2013-04-29  Vladimir Makarov  <vmakarov@redhat.com>
21475         PR target/57097
21476         * lra-constraints.c (process_alt_operands): Discourage a bit more
21477         using memory for pseudos.  Print cost dump for alternatives.
21478         Modify cost values for conflicts with early clobbers.
21479         (curr_insn_transform): Spill pseudos reassigned to NO_REGS.
21481 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21483         PR target/57098
21484         * config/i386/i386.c (ix86_expand_vec_perm): Validize constant memory.
21486 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21488         * config/aarch64/aarch64.md (movsi_aarch64): Support LDR/STR
21489         from/to S register.
21490         (movdi_aarch64): Support LDR/STR from/to D register.
21492 2013-04-29  Ian Bolton  <ian.bolton@arm.com>
21494         * common/config/aarch64/aarch64-common.c: Enable REE pass at O2
21495         or higher by default.
21497 2013-04-29  Richard Biener  <rguenther@suse.de>
21499         PR middle-end/57075
21500         * tree-inline.c (copy_edges_for_bb): Still split the bbs,
21501         even if not adding abnormal edges for calls that can make
21502         abnormal gotos.
21504 2013-04-29  Richard Biener  <rguenther@suse.de>
21506         PR middle-end/57103
21507         * tree-cfg.c (move_stmt_op): Fix condition under which to update
21508         TREE_BLOCK.
21509         (move_stmt_r): Remove redundant checking.
21511 2013-04-29  Teresa Johnson  <tejohnson@google.com>
21513         PR bootstrap/57077
21514         * basic-block.h (apply_scale): New function.
21515         (apply_probability): Use apply_scale.
21516         * gimple-streamer-in.c (input_bb): Ditto.
21517         * lto-streamer-in.c (input_cfg): Ditto.
21518         * lto-cgraph.c (merge_profile_summaries): Ditto.
21519         * tree-optimize.c (execute_fixup_cfg): Ditto.
21520         * tree-inline.c (copy_bb): Update comment to use apply_scale.
21521         (copy_edges_for_bb): Ditto.
21522         (copy_cfg_body): Ditto.
21524 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21526         * tree-ssa-tail-merge.c (find_same_succ_bb): Skip loop latch bbs.
21527         (replace_block_by): Don't set LOOPS_NEED_FIXUP.
21528         (tail_merge_optimize): Handle current_loops == NULL.
21530 2013-04-26  Jeff Law  <law@redhat.com>
21532         * tree-vrp.c (range_fits_type_p): Move to earlier point in file.
21533         (simplify_cond_using_ranges): Generalize code to simplify
21534         COND_EXPRs where one argument is a constant and the other
21535         is an SSA_NAME created by an integral type conversion.
21537 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21539         * config/arm/arm.md (store_minmaxsi): Use only when
21540         optimize_insn_for_size_p.
21542 2013-04-29  Christian Bruel  <christian.bruel@st.com>
21544         PR target/57108
21545         * sh.md (tstsi_t_zero_extract_eq): Set mode for operand 0.
21547 2013-04-29  Richard Biener  <rguenther@suse.de>
21549         PR middle-end/57089
21550         * omp-low.c (expand_omp_taskreg): If the parent function had a broken
21551         loop tree make sure to schedule a fixup for the child as well.
21552         (expand_omp_for_generic): Properly add loops.
21553         (expand_omp_for_static_nochunk): Likewise.
21554         (expand_omp_for_static_chunk): Likewise.
21555         (expand_omp_for): For the degenerate case fixup loops.
21556         (expand_omp_sections): Fix default bb placement in loops.
21557         (expand_omp_atomic_pipeline): Properly add loops.
21559 2013-04-29  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
21561         * predict.c: Fix typo in comment above #define PROB_VERY_UNLIKELY.
21563 2013-04-29  Tom de Vries  <tom@codesourcery.com>
21565         * tree-ssa-tail-merge.c: Update header comment.
21567 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21569         * config/aarch64/arm_neon.h
21570         (vcvt<sd>_f<32,64>_s<32,64>): Rewrite in C.
21571         (vcvt<q>_f<32,64>_s<32,64>): Rewrite using builtins.
21572         (vcvt_<high_>_f<32,64>_f<32,64>): Likewise.
21573         (vcvt<qsd>_<su><32,64>_f<32,64>): Likewise.
21574         (vcvta<qsd>_<su><32,64>_f<32,64>): Likewise.
21575         (vcvtm<qsd>_<su><32,64>_f<32,64>): Likewise.
21576         (vcvtn<qsd>_<su><32,64>_f<32,64>): Likewise.
21577         (vcvtp<qsd>_<su><32,64>_f<32,64>): Likewise.
21579 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21581         * config/aarch64/aarch64-simd.md
21582         (<optab><VDQF:mode><fcvt_target>2): New, maps to fix, fixuns.
21583         (<fix_trunc_optab><VDQF:mode><fcvt_target>2): New, maps to
21584         fix_trunc, fixuns_trunc.
21585         (ftrunc<VDQF:mode>2): New.
21586         * config/aarch64/iterators.md (optab): Add fix, fixuns.
21587         (fix_trunc_optab): New.
21589 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21591         * config/aarch64/aarch64-builtins.c
21592         (aarch64_builtin_vectorized_function): Vectorize over ifloorf,
21593         iceilf, lround, iroundf.
21595 2013-04-29  Uros Bizjak  <ubizjak@gmail.com>
21597         PR target/54349
21598         * config/i386/i386.h (enum ix86_tune_indices)
21599         <X86_TUNE_INTER_UNIT_MOVES_TO_VEC, X86_TUNE_INTER_UNIT_MOVES_FROM_VEC>:
21600         New, split from X86_TUNE_INTER_UNIT_MOVES.
21601         <X86_TUNE_INTER_UNIT_MOVES>: Remove.
21602         (TARGET_INTER_UNIT_MOVES_TO_VEC): New define.
21603         (TARGET_INTER_UNIT_MOVES_FROM_VEC): Ditto.
21604         (TARGET_INTER_UNIT_MOVES): Remove.
21605         * config/i386/i386.c (initial_ix86_tune_features): Update.
21606         Disable X86_TUNE_INTER_UNIT_MOVES_FROM_VEC for m_ATHLON_K8 only.
21607         (ix86_expand_convert_uns_didf_sse): Use
21608         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21609         (ix86_expand_vector_init_one_nonzero): Ditto.
21610         (ix86_expand_vector_init_interleave): Ditto.
21611         (inline_secondary_memory_needed): Return true for moves from SSE class
21612         registers for !TARGET_INTER_UNIT_MOVES_FROM_VEC targets and for moves
21613         to SSE class registers for !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21614         * config/i386/constraints.md (Yi, Ym): Depend on
21615         TARGET_INTER_UNIT_MOVES_TO_VEC.
21616         (Yj, Yn): New constraints.
21617         * config/i386/i386.md (*movdi_internal): Change constraints of
21618         operand 1 from Yi to Yj and from Ym to Yn.
21619         (*movsi_internal): Ditto.
21620         (*movdf_internal): Ditto.
21621         (*movsf_internal): Ditto.
21622         (*float<SWI48x:mode><X87MODEF:mode>2_1): Use
21623         TARGET_INTER_UNIT_MOVES_TO_VEC instead of TARGET_INTER_UNIT_MOVES.
21624         (*float<SWI48x:mode><X87MODEF:mode>2_1 splitters): Ditto.
21625         (floatdi<X87MODEF:mode>2_i387_with_xmm): Ditto.
21626         (floatdi<X87MODEF:mode>2_i387_with_xmm splitters): Ditto.
21627         * config/i386/sse.md (movdi_to_sse): Ditto.
21628         (sse2_stored): Change constraint of operand 1 from Yi to Yj.
21629         Use TARGET_INTER_UNIT_MOVES_FROM_VEC instead of
21630         TARGET_INTER_UNIT_MOVES.
21631         (sse_storeq_rex64): Change constraint of operand 1 from Yi to Yj.
21632         (sse_storeq_rex64 splitter): Use TARGET_INTER_UNIT_MOVES_FROM_VEC
21633         instead of TARGET_INTER_UNIT_MOVES.
21634         * config/i386/mmx.md (*mov<mode>_internal): Change constraint of
21635         operand 1 from Yi to Yj and from Ym to Yn.
21637 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21639         * config/aarch64/aarch64-simd-builtins.def (vec_unpacks_hi_): New.
21640         (float_truncate_hi_): Likewise.
21641         (float_extend_lo_): Likewise.
21642         (float_truncate_lo_): Likewise.
21643         * config/aarch64/aarch64-simd.md (vec_unpacks_lo_v4sf): New.
21644         (aarch64_float_extend_lo_v2df): Likewise.
21645         (vec_unpacks_hi_v4sf): Likewise.
21646         (aarch64_float_truncate_lo_v2sf): Likewise.
21647         (aarch64_float_truncate_hi_v4sf): Likewise.
21648         (vec_pack_trunc_v2df): Likewise.
21649         (vec_pack_trunc_df): Likewise.
21651 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21653         * config/aarch64/aarch64-builtins.c
21654         (aarch64_fold_builtin): Fold float conversions.
21655         * config/aarch64/aarch64-simd-builtins.def
21656         (floatv2si, floatv4si, floatv2di): New.
21657         (floatunsv2si, floatunsv4si, floatunsv2di): Likewise.
21658         * config/aarch64/aarch64-simd.md
21659         (<optab><fcvt_target><VDQF:mode>2): New, expands to float and floatuns.
21660         * config/aarch64/iterators.md (FLOATUORS): New.
21661         (optab): Add float, floatuns.
21662         (su_optab): Likewise.
21664 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21666         * config/aarch64/aarch64-builtins.c
21667         (aarch64_builtin_vectorized_function): Use new names for
21668         fcvt builtins.
21669         * config/aarch64/aarch64-simd-builtins.def (fcvtzs): Split as...
21670         (lbtruncv2sf, lbtruncv4sf, lbtruncv2df): ...This.
21671         (fcvtzu): Split as...
21672         (lbtruncuv2sf, lbtruncuv4sf, lbtruncuv2df): ...This.
21673         (fcvtas): Split as...
21674         (lroundv2sf, lroundv4sf, lroundv2df, lroundsf, lrounddf): ...This.
21675         (fcvtau): Split as...
21676         (lrounduv2sf, lrounduv4sf, lrounduv2df, lroundusf, lroundudf): ...This.
21677         (fcvtps): Split as...
21678         (lceilv2sf, lceilv4sf, lceilv2df): ...This.
21679         (fcvtpu): Split as...
21680         (lceiluv2sf, lceiluv4sf, lceiluv2df, lceilusf, lceiludf): ...This.
21681         (fcvtms): Split as...
21682         (lfloorv2sf, lfloorv4sf, lfloorv2df): ...This.
21683         (fcvtmu): Split as...
21684         (lflooruv2sf, lflooruv4sf, lflooruv2df, lfloorusf, lfloorudf): ...This.
21685         (lfrintnv2sf, lfrintnv4sf, lfrintnv2df, lfrintnsf, lfrintndf): New.
21686         (lfrintnuv2sf, lfrintnuv4sf, lfrintnuv2df): Likewise.
21687         (lfrintnusf, lfrintnudf): Likewise.
21688         * config/aarch64/aarch64-simd.md
21689         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Convert to
21690         define_insn.
21691         (aarch64_fcvt<frint_suffix><su><mode>): Remove.
21692         * config/aarch64/iterators.md (FCVT): Include UNSPEC_FRINTN.
21693         (fcvt_pattern): Likewise.
21695 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21697         * config/aarch64/aarch64-simd.md
21698         (l<fcvt_pattern><su_optab><fcvt_target><VDQF:mode>2): Rename to...
21699         (l<fcvt_pattern><su_optab><VDQF:mode><fcvt_target>2): ... This.
21701 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21703         * config/aarch64/arm_neon.h (vrndq<a,m,n,p>_f<32, 64>): Rename to...
21704         (vrnd<a,m,n,p>q_f<32, 64>): ...This, implement using builtin.
21705         (vrnd<a,m,n,p>_f32): Implement using builtins.
21706         (vrnd<i,x><q>_f<32, 64>): New.
21708 2013-04-29  James Greenhalgh  <james.greenhalgh@arm.com>
21710         * config/aarch64/aarch64-builtins.c
21711         (aarch64_builtin_vectorized_function): Fold to standard pattern names.
21712         * config/aarch64/aarch64-simd-builtins.def (frintn): New.
21713         (frintz): Rename to...
21714         (btrunc): ...this.
21715         (frintp): Rename to...
21716         (ceil): ...this.
21717         (frintm): Rename to...
21718         (floor): ...this.
21719         (frinti): Rename to...
21720         (nearbyint): ...this.
21721         (frintx): Rename to...
21722         (rint): ...this.
21723         (frinta): Rename to...
21724         (round): ...this.
21725         * config/aarch64/aarch64-simd.md
21726         (aarch64_frint<frint_suffix><mode>): Delete.
21727         (<frint_pattern><mode>2): Convert to insn.
21728         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRINTN.
21729         * config/aarch64/iterators.md (FRINT): Add UNSPEC_FRINTN.
21730         (frint_pattern): Likewise.
21731         (frint_suffix): Likewise.
21733 2013-04-29  Richard Biener  <rguenther@suse.de>
21735         PR tree-optimization/57081
21736         * loop-init.c: Include tree-flow.h.
21737         (loop_optimizer_finalize): Free number of iteration estimates.
21738         * Makefile.in (loop-init.o): Add $(TREE_FLOW_H) dependency.
21740 2013-04-29  Jakub Jelinek  <jakub@redhat.com>
21742         PR tree-optimization/57083
21743         * tree-vrp.c (extract_range_from_binary_expr_1): For LSHIFT_EXPR with
21744         non-singleton shift count range, zero extend low_bound for uns case.
21746         * config/i386/predicates.md (general_vector_operand): New predicate.
21747         * config/i386/i386.c (const_vector_equal_evenodd_p): New function.
21748         (ix86_expand_mul_widen_evenodd): Force op1 resp. op2 into register
21749         if they aren't nonimmediate operands.  If their original values
21750         satisfy const_vector_equal_evenodd_p, don't shift them.
21751         * config/i386/sse.md (mul<mode>3): Use general_vector_operand
21752         predicates.  For the SSE4.1 case force operands[{1,2}] into registers
21753         if not nonimmediate_operand.
21754         (vec_widen_smult_even_v4si): Use nonimmediate_operand predicates
21755         instead of register_operand.
21756         (vec_widen_<s>mult_odd_<mode>): Use general_vector_operand predicates.
21758 2013-04-28  Eric Botcazou  <ebotcazou@adacore.com>
21760         * stor-layout.c (finalize_size_functions): Allocate a structure and
21761         reset cfun before dumping the functions.
21763 2013-04-27  Jakub Jelinek  <jakub@redhat.com>
21765         * config/i386/i386.c (ix86_expand_call): Make cregs_size unsigned.
21767         PR target/56866
21768         * config/i386/i386.c (ix86_expand_mul_widen_evenodd): Don't
21769         use xop_pmacsdqh if uns_p.
21770         * config/i386/sse.md (xop_rotr<mode>3): Fix up computation of
21771         the immediate rotate count.
21773 2013-04-26  Vladimir Makarov  <vmakarov@redhat.com>
21775         * rtl.h (struct rtx_def): Add comment for field jump.
21776         (LRA_SUBREG_P): New macro.
21777         * recog.c (register_operand): Check LRA_SUBREG_P.
21778         * lra.c (lra): Add note at the end of RTL code. Align non-empty
21779         stack frame.
21780         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
21781         (lra_final_code_change): Skip subreg change for operators.
21782         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
21783         if there are no operand changes.
21784         * lra-constraints.c (curr_insn_set): New.
21785         (match_reload): Set LRA_SUBREG_P.
21786         (emit_spill_move): Ditto.
21787         (check_and_process_move): Use curr_insn_set. Process only single
21788         set insns.  Don't initialize sec_mem_p and change_p.
21789         (simplify_operand_subreg): Use LRA_SUBREG_P.
21790         (reg_in_class_p): New function.
21791         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
21792         of #ifdef.  Add code to remove cycling.
21793         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
21794         non-null disp.  Reload inner instead of disp when base and index
21795         are null.  Try to put lo_sum into register.
21796         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
21797         (check_and_process_move): Move code for move cost check to
21798         simple_move_p.  Remove equiv_substitution.
21799         (simple_move_p): New function.
21800         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
21801         curr_insn_set.  Call check_and_process_move only for single set
21802         insns.  Use the new function.  Move call of check_and_process_move
21803         after operand equiv substitution and address process.
21805 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21807         PR go/57045
21808         * tree-ssa-uninit.c (compute_uninit_opnds_pos): In functions
21809         with nonlocal goto receivers or returns twice calls, ignore
21810         unininitialized values from abnormal edges to nl goto receiver
21811         or returns twice call.
21813 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21815         PR tree-optimization/57051
21816         * fold-const.c (const_binop): Handle VEC_LSHIFT_EXPR
21817         and VEC_RSHIFT_EXPR if shift count is a multiple of element
21818         bitsize.
21820 2013-04-26  Richard Biener  <rguenther@suse.de>
21822         * omp-low.c (finalize_task_copyfn): Do not drop PROP_loops.
21823         (expand_omp_taskreg): Likewise.  Mark loops for fixup.
21824         * tree-cfg.c (move_block_to_fn): Remap loop fathers.
21825         (fixup_loop_arrays_after_move): New function.
21826         (move_sese_region_to_fn): Properly outline the loop tree parts
21827         of the SESE region.
21829 2013-04-26  Uros Bizjak  <ubizjak@gmail.com>
21831         * config/i386/i386.md (type, unit): Fix long lines.
21833 2013-04-26  Richard Biener  <rguenther@suse.de>
21835         * Makefile.in (lto-streamer-in.o): Add $(CFGLOOP_H) dependency.
21836         (lto-streamer-out.o): Likewise.
21837         * cfgloop.c (init_loops_structure): Export, add struct function
21838         argument and adjust.
21839         (flow_loops_find): Adjust.
21840         * cfgloop.h (enum loop_estimation): Add EST_LAST.
21841         (init_loops_structure): Declare.
21842         * lto-streamer-in.c: Include cfgloop.h.
21843         (input_cfg): Input the loop tree.
21844         * lto-streamer-out.c: Include cfgloop.h.
21845         (output_cfg): Output the loop tree.
21846         (output_struct_function_base): Do not drop PROP_loops.
21848 2013-03-26  Richard Biener  <rguenther@suse.de>
21850         * tree-cfg.c (execute_build_cfg): Build the loop tree.
21851         (pass_build_cfg): Provide PROP_loops.
21852         (move_sese_region_to_fn): Remove loops that are outlined into fn
21853         for now.
21854         * tree-inline.c: Include cfgloop.h.
21855         (initialize_cfun): Do not drop PROP_loops.
21856         (copy_loops): New function.
21857         (copy_cfg_body): Copy loop structure.
21858         (tree_function_versioning): Initialize destination loop tree.
21859         * tree-ssa-loop.c (pass_tree_loop_init): Do not provide PROP_loops.
21860         (pass_parallelize_loops): Do IL verification.
21861         * loop-init.c (loop_optimizer_init): Fixup loops if required.
21862         * tree-optimize.c (execute_fixup_cfg): If we need to cleanup
21863         the CFG make sure we fixup loops as well.
21864         * tree-ssa-tail-merge.c: Include cfgloop.h.
21865         (replace_block_by): When merging loop latches mark loops for fixup.
21866         * lto-streamer-out.c (output_struct_function_base): Drop
21867         PROP_loops for now.
21868         * tree-ssa-phiopt.c: Include tree-scalar-evolution.h.
21869         (tree_ssa_cs_elim): Initialize the loop optimizer and SCEV.
21870         * ipa-split.c: Include cfgloop.h.
21871         (split_function): Add the new return block to the loop tree root.
21872         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Return
21873         whether we have removed the forwarder block.
21874         (merge_phi_nodes): If we removed a forwarder mark loops for fixup.
21875         * cfgloop.h (place_new_loop): Declare.
21876         * cfgloopmanip.c (place_new_loop): Export.
21877         * Makefile.in (asan.o): Add $(CFGLOOP_H) dependency.
21878         (tree-switch-conversion.o): Likewise.
21879         (tree-complex.o): Likewise.
21880         (tree-inline.o): Likewise.
21881         (tree-ssa-tailmerge.o): Likewise.
21882         (ipa-split.o): Likewise.
21883         (tree-ssa-phiopt.o): Add $(SCEV_H) dependency.
21884         (tree-ssa-copy.o): Likewise.
21885         * tree-switch-conversion.c: Include cfgloop.h
21886         (process_switch): If we emit a bit-test cascade, schedule loops
21887         for fixup.
21888         * tree-complex.c: Include cfgloop.h.
21889         (expand_complex_div_wide): Properly add new basic-blocks to loops.
21890         * asan.c: Include cfgloop.h.
21891         (create_cond_insert_point): Properly add new basic-blocks to
21892         loops, schedule loop fixup.
21893         * cfgloop.c (verify_loop_structure): Check that looks are not
21894         marked for fixup.
21895         * omp-low.c (expand_parallel_call): Properly add new basic-blocks
21896         to loops.
21897         (expand_omp_for_generic): Likewise.
21898         (expand_omp_sections): Likewise.
21899         (expand_omp_atomic_pipeline): Schedule loops for fixup.
21900         * tree-ssa-copy.c: Include tree-scalar-evolution.h.
21901         (fini_copy_prop): Disable DCE in substitute_and_fold if SCEV
21902         is initialized, not when loops are present.
21903         * tree-parloops.c (parallelize_loops): Remove checking here.
21904         * passes.c (init_optimization_passes): Schedule a copy-propagation
21905         pass before complete unrolling of inner loops.
21907 2013-04-26  Jakub Jelinek  <jakub@redhat.com>
21909         * Makefile.in (toplev.o): Depend on diagnostic-color.h.
21910         * diagnostic-color.c (should_colorize): Remove _WIN32 version.
21911         (colorize_init): Add argument to _WIN32 version.
21912         * toplev.c: Include diagnostic-color.h.
21913         (process_options): Default to -fdiagnostics-color=auto if
21914         GCC_COLORS env var is in the environment.
21915         * common.opt (fdiagnostics-color=): Add Var and Init.
21916         * doc/invoke.texi (-fdiagnostics-color=): Document that if GCC_COLORS
21917         env var is in the environment, the default is auto rather than never.
21919         * diagnostic.h (file_name_as_prefix): Add context argument.
21920         * diagnostic.c (file_name_as_prefix): Likewise.  Colorize
21921         the string as locus.
21922         * langhooks.c (lhd_print_error_function): Adjust caller.
21924 2013-04-25  Lawrence Crowl  <crowl@google.com>
21926         * var-tracking.c (shared_hash_def::htab):
21927         Change type to hash_table.  Update dependent calls and types.
21929 2013-04-25  Lawrence Crowl  <crowl@google.com>
21931         * Makefile.in: Update as needed below.
21933         * alloc-pool.c (static hash_table <alloc_pool_hasher> alloc_pool_hash):
21934         Move declaration to after the type's method definitons.
21936         * attribs.c (htab_t scoped_attributes::attribute_hash):
21937         Change type to hash_table.  Update dependent calls and types.
21939         * bitmap.c (htab_t bitmap_desc_hash):
21940         Change type to hash_table.  Update dependent calls and types.
21942         * cselib.c (htab_t cselib_hash_table):
21943         Change type to hash_table.  Update dependent calls and types.
21945         * data-streamer.h (struct string_slot): Move to lto-streamer.h.
21946         (hash_string_slot_node): Move implementation into lto-streamer.h
21947         struct string_slot_hasher.
21948         (eq_string_slot_node): Likewise.
21950         * data-streamer-out.c: Update output_block::string_hash_table
21951         dependent calls and types.
21953         * dwarf2cfi.c (htab_t trace_index):
21954         Change type to hash_table.  Update dependent calls and types.
21956         * dwarf2out.c (htab_t break_out_includes::cu_hash_table):
21957         Change type to hash_table.  Update dependent calls and types.
21958         (htab_t copy_decls_for_unworthy_types::decl_table): Likewise.
21959         (htab_t optimize_external_refs::map): Likewise.
21960         (htab_t output_comp_unit::extern_map): Likewise.
21961         (htab_t output_comdat_type_unit::extern_map): Likewise.
21962         (htab_t output_macinfo::macinfo_htab): Likewise.
21963         (htab_t optimize_location_lists::htab): Likewise.
21964         (htab_t dwarf2out_finish::comdat_type_table): Likewise.
21966         * except.c (htab_t ehspec_hash_type):
21967         Change type to hash_table.  Update dependent calls and types.
21968         (assign_filter_values::ttypes): Likewise.
21969         (assign_filter_values::ehspec): Likewise.
21970         (sjlj_assign_call_site_values::ar_hash): Likewise.
21971         (convert_to_eh_region_ranges::ar_hash): Likewise.
21973         * gcse.c (htab_t pre_ldst_table):
21974         Change type to hash_table.  Update dependent calls and types.
21976         * ggc-common.c (htab_t saving_htab):
21977         Change type to hash_table.  Update dependent calls and types.
21978         (htab_t loc_hash): Likewise.
21979         (htab_t ptr_hash): Likewise.
21980         (call_count): Rename ggc_call_count.
21981         (call_alloc): Rename ggc_call_alloc.
21982         (loc_descriptor): Rename make_loc_descriptor.
21983         (add_statistics): Rename ggc_add_statistics.
21985         * ggc-common.c (saving_htab):
21986         Change type to hash_table.  Update dependent calls and types.
21988         * gimple.h (struct gimplify_ctx): Move to gimplify-ctx.h.
21989         (push_gimplify_context): Likewise.
21990         (pop_gimplify_context): Likewise.
21991         (struct gimple_temp_hash_elt): Added.
21992         (struct gimplify_hasher): Likewise.
21993         (struct gimplify_ctx.temp_htab):
21994         Change type to hash_table.  Update dependent calls and types.
21996         * gimple-fold.c: Include gimplify-ctx.h.
21998         * gimple-ssa-strength-reduction.c (htab_t base_cand_map):
21999         Change type to hash_table.  Update dependent calls and types.
22000         (base_cand_dump_callback): Rename to ssa_base_cand_dump_callback to
22001         avoid potential global name collision.
22003         * gimplify.c: Include gimplify-ctx.h.
22004         (struct gimple_temp_hash_elt): Move to gimplify-ctx.h.
22005         (htab_t gimplify_ctx::temp_htab):
22006         Update dependent calls and types for new type hash_table.
22007         (gimple_tree_hash): Move into gimplify_hasher in gimplify-ctx.h.
22008         (gimple_tree_eq): Move into gimplify_hasher in gimplify-ctx.h.
22010         * gimplify-ctx.h: New.
22011         (struct gimple_temp_hash_elt): Move from gimplify.c.
22012         (class gimplify_hasher): New.
22013         (struct gimplify_ctx): Move from gimple.h.
22014         (htab_t gimplify_ctx::temp_htab):
22015         Change type to hash_table.  Update dependent calls and types.
22017         * graphite-clast-to-gimple.c: Include graphite-htab.h.
22018         (htab_t ivs_params::newivs_index):
22019         Change type to hash_table.  Update dependent calls and types.
22020         (htab_t ivs_params::params_index): Likewise.
22021         (htab_t print_generated_program::params_index): Likewise.
22022         (htab_t gloog::newivs_index): Likewise.
22023         (htab_t gloog::params_index): Likewise.
22025         * graphite.c: Include graphite-htab.h.
22026         4htab_t graphite_transform_loops::bb_pbb_mapping):
22027         Change type to hash_table.  Update dependent calls and types.
22029         * graphite-clast-to-gimple.h: (extern gloog) Move to graphite-htab.h.
22030         (bb_pbb_map_hash): Fold into bb_pbb_htab_type in graphite-htab.h.
22031         (eq_bb_pbb_map): Fold into bb_pbb_htab_type in graphite-htab.h.
22033         * graphite-dependences.c: Include graphite-htab.h.
22034         (loop_is_parallel_p): Change hash table type of parameter.
22036         * graphite-htab.h: New.
22037         (typedef hash_table <bb_pbb_hasher> bb_pbb_htab_type): New.
22038         (extern find_pbb_via_hash): Move from graphite-poly.h.
22039         (extern loop_is_parallel_p): Move from graphite-poly.h.
22040         (extern get_loop_body_pbbs): Move from graphite-poly.h.
22042         * graphite-poly.h (extern find_pbb_via_hash): Move to graphite-htab.h.
22043         (extern loop_is_parallel_p): Move to graphite-htab.h.
22044         (extern get_loop_body_pbbs): Move to graphite-htab.h.
22046         * haifa-sched.c (htab_t delay_htab):
22047         Change type to hash_table.  Update dependent calls and types.
22048         (htab_t delay_htab_i2): Likewise.
22050         * ira-color.c (htab_t allocno_hard_regs_htab):
22051         Change type to hash_table.  Update dependent calls and types.
22053         * ira-costs.c (htab_t cost_classes_htab):
22054         Change type to hash_table.  Update dependent calls and types.
22056         * loop-invariant.c (htab_t merge_identical_invariants::eq):
22057         Change type to hash_table.  Update dependent calls and types.
22059         * loop-iv.c (htab_t bivs):
22060         Change type to hash_table.  Update dependent calls and types.
22062         * loop-unroll.c (htab_t opt_info::insns_to_split):
22063         Change type to hash_table.  Update dependent calls and types.
22064         (htab_t opt_info::insns_with_var_to_expand): Likewise.
22066         * lto-streamer.h (struct string_slot): Move from data-streamer.h
22067         (struct string_slot_hasher): New.
22068         (htab_t output_block::string_hash_table):
22069         Change type to hash_table.  Update dependent calls and types.
22071         * lto-streamer-in.c (freeing_string_slot_hasher): New.
22072         (htab_t file_name_hash_table):
22073         Change type to hash_table.  Update dependent calls and types.
22075         * lto-streamer-out.c: Update output_block::string_hash_table dependent
22076         calls and types.
22078         * lto-streamer.c (htab_t tree_htab):
22079         Change type to hash_table.  Update dependent calls and types.
22081         * omp-low.c: Include gimplify-ctx.h.
22083         * passes.c (htab_t name_to_pass_map):
22084         Change type to hash_table.  Update dependent calls and types.
22085         (pass_traverse): Rename to passes_pass_traverse.
22087         * plugin.c (htab_t event_tab):
22088         Change type to hash_table.  Update dependent calls and types.
22090         * postreload-gcse.c (htab_t expr_table):
22091         Change type to hash_table.  Update dependent calls and types.
22092         (dump_hash_table_entry): Rename dump_expr_hash_table_entry.
22094         * sese.c (debug_rename_map_1): Make extern.
22095         (htab_t copy_bb_and_scalar_dependences::rename_map):
22096         Change type to hash_table.  Update dependent calls and types.
22098         * sese.h (extern debug_rename_map): Move to .c file.
22100         * store-motion.c (htab_t store_motion_mems_table):
22101         Change type to hash_table.  Update dependent calls and types.
22103         * trans-mem.c (htab_t tm_new_mem_hash):
22104         Change type to hash_table.  Update dependent calls and types.
22106         * tree-browser.c (htab_t TB_up_ht):
22107         Change type to hash_table.  Update dependent calls and types.
22109         * tree-cfg.c (htab_t discriminator_per_locus):
22110         Change type to hash_table.  Update dependent calls and types.
22112         * tree-complex.c: Include tree-hasher.h
22113         (htab_t complex_variable_components):
22114         Change type to hash_table.  Update dependent calls and types.
22116         * tree-eh.c (htab_t finally_tree):
22117         Change type to hash_table.  Update dependent calls and types.
22119         * tree-flow.h (extern int_tree_map_hash): Moved into tree-hasher
22120         struct int_tree_hasher.
22121         (extern int_tree_map_eq): Likewise.
22122         (uid_decl_map_hash): Removed.
22123         (extern decl_tree_map_eq): Likewise.
22125         * tree-hasher.h: New.
22126         (struct int_tree_hasher): New.
22127         (typedef int_tree_htab_type): New.
22129         * tree-inline.c: Include gimplify-ctx.h.
22131         * tree-mudflap.c: Include gimplify-ctx.h.
22133         * tree-parloops.c: Include tree-hasher.h.
22134         (htab_t eliminate_local_variables_stmt::decl_address):
22135         Change type to hash_table.  Update dependent calls and types.
22136         (htab_t separate_decls_in_region::decl_copies): Likewise.
22138         * tree-scalar-evolution.c (htab_t resolve_mixers::cache):
22139         Change type to hash_table.  Update dependent calls and types.
22141         * tree-sra.c (candidates):
22142         Change type to hash_table.  Update dependent calls and types.
22144         * tree-ssa.c (int_tree_map_eq): Moved into struct int_tree_hasher
22145         in tree-flow.h.
22146         (int_tree_map_hash): Likewise.
22148         * tree-ssa-dom.c (htab_t avail_exprs):
22149         Change type to hash_table.  Update dependent calls and types.
22151         * tree-ssa-live.c (var_map_base_init::tree_to_index):
22152         Change type to hash_table.  Update dependent calls and types.
22154         * tree-ssa-loop-ivopts.c (struct ivopts_data.inv_expr_tab):
22155         Change type to hash_table.  Update dependent calls and types.
22157         * tree-ssa-phiopt.c (seen_ssa_names):
22158         Change type to hash_table.  Update dependent calls and types.
22160         * tree-ssa-strlen.c (decl_to_stridxlist_htab):
22161         Change type to hash_table.  Update dependent calls and types.
22163         * tree-ssa-uncprop.c (equiv):
22164         Change type to hash_table.  Update dependent calls and types.
22166 2013-04-25  Jakub Jelinek  <jakub@redhat.com>
22168         PR rtl-optimization/57003
22169         * regcprop.c (copyprop_hardreg_forward_1): If ksvd.ignore_set_reg,
22170         call note_stores with kill_clobbered_value callback again after
22171         killing regs_invalidated_by_call.
22173 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22175         * config/aarch64/aarch64-simd.md
22176         (aarch64_simd_bsl<mode>_internal): Rewrite RTL to not use UNSPEC_BSL.
22177         (aarch64_simd_bsl<mode>): Likewise.
22178         * config/aarch64/iterators.md (unspec): Remove UNSPEC_BSL.
22180 2013-04-25  Marek Polacek  <polacek@redhat.com>
22182         PR tree-optimization/57066
22183         * builtins.c (fold_builtin_logb): Return +Inf for -Inf.
22185 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22187         * config/aarch64/aarch64-simd.md (neg<mode>2): Use VDQ iterator.
22189 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22191         * config/aarch64/aarch64-builtins.c
22192         (aarch64_fold_builtin): New.
22193         * config/aarch64/aarch64-protos.h (aarch64_fold_builtin): New.
22194         * config/aarch64/aarch64.c (TARGET_FOLD_BUILTIN): Define.
22195         * config/aarch64/aarch64-simd-builtins.def (abs): New.
22196         * config/aarch64/arm_neon.h
22197         (vabs<q>_<f32, 64>): Implement using __builtin_aarch64_fabs.
22199 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22200             Tejas Belagod  <tejas.belagod@arm.com>
22202         * config/aarch64/aarch64-builtins.c
22203         (aarch64_gimple_fold_builtin): New.
22204         * config/aarch64/aarch64-protos.h (aarch64_gimple_fold_builtin): New.
22205         * config/aarch64/aarch64-simd-builtins.def (addv): New.
22206         * config/aarch64/aarch64-simd.md (addpv4sf): New.
22207         (addvv4sf): Update.
22208         * config/aarch64/aarch64.c (TARGET_GIMPLE_FOLD_BUILTIN): Define.
22210 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22212         * config/aarch64/aarch64.md
22213         (*cmp_swp_<optab><ALLX:mode>_shft_<GPI:mode>): New pattern.
22215 2013-04-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22217         * config/aarch64/aarch64.md (*ngc<mode>): New pattern.
22218         (*ngcsi_uxtw): New pattern.
22220 2013-04-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22221             Julian Brown  <julian@codesourcery.com>
22223         * config/arm/arm.c (neon_builtin_type_mode): Add T_V4HF.
22224         (TB_DREG): Add T_V4HF.
22225         (v4hf_UP): New macro.
22226         (neon_itype): Add NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22227         (arm_init_neon_builtins): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22228         Handle initialisation of V4HF. Adjust initialisation of reinterpret
22229         built-ins.
22230         (arm_expand_neon_builtin): Handle NEON_FLOAT_WIDEN, NEON_FLOAT_NARROW.
22231         (arm_vector_mode_supported_p): Handle V4HF.
22232         (arm_mangle_map): Handle V4HFmode.
22233         * config/arm/arm.h (VALID_NEON_DREG_MODE): Add V4HF.
22234         * config/arm/arm_neon_builtins.def: Add entries for
22235         vcvtv4hfv4sf, vcvtv4sfv4hf.
22236         * config/arm/neon.md (neon_vcvtv4sfv4hf): New pattern.
22237         (neon_vcvtv4hfv4sf): Likewise.
22238         * config/arm/neon-gen.ml: Handle half-precision floating point
22239         features.
22240         * config/arm/neon-testgen.ml: Handle Requires_FP_bit feature.
22241         * config/arm/arm_neon.h: Regenerate.
22242         * config/arm/neon.ml (type elts): Add F16.
22243         (type vectype): Add T_float16x4, T_floatHF.
22244         (type vecmode): Add V4HF.
22245         (type features): Add Requires_FP_bit feature.
22246         (elt_width): Handle F16.
22247         (elt_class): Likewise.
22248         (elt_of_class_width): Likewise.
22249         (mode_of_elt): Refactor.
22250         (type_for_elt): Handle F16, fix error messages.
22251         (vectype_size): Handle T_float16x4.
22252         (vcvt_sh): New function.
22253         (ops): Add entries for vcvt_f16_f32, vcvt_f32_f16.
22254         (string_of_vectype): Handle T_floatHF, T_float16, T_float16x4.
22255         (string_of_mode): Handle V4HF.
22256         * doc/arm-neon-intrinsics.texi: Regenerate.
22258 2013-04-25  James Greenhalgh  <james.greenhalgh@arm.com>
22260         * config/aarch64/aarch64.c (aarch64_print_operand): Fix asm_fprintf
22261         format specifier in 'X' case.
22263 2013-04-25  Alan Modra  <amodra@gmail.com>
22265         PR target/57052
22266         * config/rs6000/rs6000.md (rotlsi3_internal7): Rename to
22267         rotlsi3_internal7le and condition on !BYTES_BIG_ENDIAN.
22268         (rotlsi3_internal8be): New BYTES_BIG_ENDIAN insn.
22269         Repeat for many other rotate/shift and mask patterns using subregs.
22270         Name lshiftrt insns.
22271         (ashrdisi3_noppc64): Rename to ashrdisi3_noppc64be and condition
22272         on WORDS_BIG_ENDIAN.
22274 2013-04-25  Alan Modra  <amodra@gmail.com>
22276         * config.gcc: Support little-endian powerpc-linux targets.
22277         * config/rs6000/linux.h (LINK_OS_LINUX_EMUL): Define.
22278         (LINK_OS_LINUX_SPEC): Define.
22279         * config/rs6000/linuxspe.h (TARGET_DEFAULT):
22280         Preserve MASK_LITTLE_ENDIAN.
22281         * config/rs6000/default64.h (TARGET_DEFAULT): Likewise.
22282         * config/rs6000/linuxaltivec.h (TARGET_DEFAULT): Likewise.
22283         * config/rs6000/linux64.h (OPTION_LITTLE_ENDIAN): Don't zero.
22284         (LINK_OS_LINUX_EMUL32, LINK_OS_LINUX_EMUL64): Define.
22285         (LINK_OS_LINUX_SPEC32, LINK_OS_LINUX_SPEC64): Use above.
22286         * config/rs6000/rs6000.c (output_toc): Don't use .tc for TARGET_ELF.
22287         Correct fp word order for little-endian.  Don't shift toc entries
22288         smaller than a word for little-endian.
22289         * config/rs6000/rs6000.md (bswaphi2, bswapsi2 split): Comment.
22290         (bswapdi2 splits): Correct low-part subreg for little-endian.
22291         Remove wrong BYTES_BIG_ENDIAN tests, and rename vars to remove
22292         low/high where such is correct only for be.
22293         * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Allow
22294         little-endian for -mcall-aixdesc.
22296 2013-04-25  Alan Modra  <amodra@gmail.com>
22298         * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Use
22299         replace_equiv_address_nv.
22301 2013-04-25  Alan Modra  <amodra@gmail.com>
22303         * config/rs6000/rs6000.c (rs6000_emit_set_long_const): Tidy.
22305 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22307         Revert:
22308         2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22309         * rtl.h (struct rtx_def): ...
22311 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22313         PR rtl-optimizations/57046
22314         * lra-constraints (split_reg): Set up lra_risky_transformations_p
22315         for multi-reg splits.
22317 2013-04-24  H.J. Lu  <hongjiu.lu@intel.com>
22319         * config/i386/x86-64.h (ASM_SPEC): Support -mx32.
22321 2013-04-24  Sterling Augustine  <saugustine@google.com>
22323         * dwarf2out.c (skeleton_debug_str_hash, add_skeleton_AT_string)
22324         (comp_dir_string, debug_str_dwo_section): New.
22325         (DEBUG_STR_DWO_SECTION): Rename to ...
22326         (DEBUG_DWO_STR_SECTION): ... this.
22327         (DEBUG_NORM_STR_SECTION): Delete.
22328         (DEBUG_STR_SECTION, DEBUG_STR_SECTION_FLAGS): Edit definitions.
22329         (DEBUG_STR_DWO_SECTION_FLAGS): New.
22330         (find_AT_string): Move most logic to ...
22331         (find_AT_string_in_table): ... here.  New.
22332         (add_top_level_skeleton_die_attrs): Call comp_dir_string and
22333         add_skeleton_AT_string.  Delete logic.
22334         (output_skeleton_debug_sections): Remove call to
22335         add_top_level_skeleton_die_attrs.
22336         (add_comp_dir_attribute): Move logic to comp_dir_string.
22337         (dwarf2out_init): Initialize debug_str_dwo_section.
22338         (output_indirect_string): Call find_string_form.
22339         (output_indirect_strings): Rewrite.
22340         (prune_unused_types): Empty skeleton_debug_str_hash.
22341         Call get_skeleton_type_unit and add_top_level_skeleton_die_attrs.
22342         (dwarf2out_finish):  Call output_indirect_strings.
22344 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22346         * doc/cpp.texi: Remove __GXX_EXPERIMENTAL_CXX1Y__.
22348 2013-04-24  Vladimir Makarov  <vmakarov@redhat.com>
22350         * rtl.h (struct rtx_def): Add comment for field jump.
22351         (LRA_SUBREG_P): New macro.
22352         * recog.c (register_operand): Check LRA_SUBREG_P.
22353         * lra.c (lra): Add note at the end of RTL code. Align non-empty
22354         stack frame.
22355         * lra-spills.c (lra_spill): Align stack after spilling pseudos.
22356         (lra_final_code_change): Skip subreg change for operators.
22357         * lra-eliminations.c (eliminate_regs_in_insn): Make return earlier
22358         if there are no operand changes.
22359         * lra-constraints.c (curr_insn_set): New.
22360         (match_reload): Set LRA_SUBREG_P.
22361         (emit_spill_move): Ditto.
22362         (check_and_process_move): Use curr_insn_set. Process only single
22363         set insns.  Don't initialize sec_mem_p and change_p.
22364         (simplify_operand_subreg): Use LRA_SUBREG_P.
22365         (reg_in_class_p): New function.
22366         (process_alt_operands): Use it.  Use #if HAVE_ATTR_enabled instead
22367         of #ifdef.  Add code to remove cycling.
22368         (process_address): Check EXTRA_CONSTRAINT_STR. Process even if
22369         non-null disp.  Reload inner instead of disp when base and index
22370         are null.  Try to put lo_sum into register.
22371         (EBB_PROBABILITY_CUTOFF): Redefine probability in percents.
22372         (check_and_process_move): Move code for move cost check to
22373         simple_move_p.  Remove equiv_substitution.
22374         (simple_move_p): New function.
22375         (curr_insn_transform): Initialize sec_mem_p and change_p.  Set up
22376         curr_insn_set.  Call check_and_process_move only for single set
22377         insns.  Use the new function.  Move call of check_and_process_move
22378         after operand equiv substitution and address process.
22380 2013-04-24  James Greenhalgh  <james.greenhalgh@arm.com>
22382         * config/aarch64/arm_neon.h (vld1<q>_lane*): Fix constraints.
22383         (vld1<q>_dup_<sufp><8, 16, 32, 64>): Likewise.
22384         (vld1<q>_<sufp><8, 16, 32, 64>): Likewise.
22386 2013-04-24  Paolo Carlini  <paolo.carlini@oracle.com>
22388         * doc/cpp.texi: Document __GXX_EXPERIMENTAL_CXX1Y__.
22390 2013-04-24  Marek Polacek  <polacek@redhat.com>
22392         * tree-scalar-evolution.h (analyze_scalar_evolution): Remove.
22393         * tree-scalar-evolution.c (get_exit_conditions_rec): Likewise.
22394         (select_loops_exit_conditions): Likewise.
22395         (number_of_iterations_for_all_loops): Likewise.
22396         (analyze_scalar_evolution_for_all_loop_phi_nodes): Likewise.
22397         (scev_analysis): Likewise.
22399 2013-04-02  Catherine Moore  <clm@codesourcery.com>
22400             Chao-ying Fu  <fu@mips.com>
22402         * config/mips/micromips.md (jraddiusp): New pattern.
22403         * config/mips/mips.c (mips_expand_epilogue): Use the JRADDIUSP
22404         instruction if possible.
22406 2013-04-24  Alan Modra  <amodra@gmail.com>
22408         * config/rs6000/driver-rs6000.c (elf_dcachebsize): Fix comment pasto.
22410 2013-04-24  Julian Brown  <julian@codesourcery.com>
22411             Chung-Lin Tang  <cltang@codesourcery.com>
22413         * dwarf2out.c (gen_enumeration_type_die): Fix HOST_BITS_PER_WIDE_INT
22414         dependency behavior in enumeration type DIE generation. Add TODO note
22415         to comments about future DW_FORM_sdata/udata re-work of related code.
22417 2013-04-23  Lawrence Crowl  <crowl@google.com>
22419         * Makefile.in: Update as needed below.
22421         * hash-table.h (class hash_table):
22422         Correct many methods with parameter types compare_type to the correct
22423         value_type.  (Correct code was unlikely to notice the change.)
22424         (hash_table::elements_with_deleted) New.
22425         (class hashtable::iterator): New.
22426         (hashtable::begin()): New.
22427         (hashtable::end()): New.
22428         (FOR_EACH_HASH_TABLE_ELEMENT): New.
22430         * statistics.c (statistics_hashes):
22431         Change type to hash_table.  Update dependent calls and types.
22433         * tree-into-ssa.c (var_infos):
22434         Change type to hash_table.  Update dependent calls and types.
22436         * tree-ssa-coalesce.c (struct coalesce_list_d.list):
22437         Change type to hash_table.  Update dependent calls and types.
22439         * tree-ssa-loop-im.c (struct mem_ref.refs):
22440         Change type to hash_table.  Update dependent calls and types.
22442         * tree-ssa-reassoc.c (undistribute_ops_list::ctable):
22443         Change type to hash_table.  Update dependent calls and types.
22445         * tree-ssa-sccvn.c (vn_tables_s::nary):
22446         Change type to hash_table.  Update dependent calls and types.
22447         (vn_tables_s::phis): Likewise.
22448         (vn_tables_s::references): Likewise.
22450         * tree-ssa-sccvn.h (vn_nary_op_eq): Update parameter and return types.
22451         (vn_reference_eq): Update parameter and return types.
22453         * tree-ssa-structalias.c (pointer_equiv_class_table):
22454         Change type to hash_table.  Update dependent calls and types.
22455         (location_equiv_class_table): Likewise.
22457         * tree-vect-data-refs.c: Consequential changes for making
22458         peeling a hash_table.
22460         * tree-vect-loop.c (new_loop_vec_info): Dependent hash_table update.
22461         (destroy_loop_vec_info): Dependent hash_table update.
22463         * tree-vectorizer.h (peeling_htab):
22464         Change type to hash_table.  Update dependent calls and types.
22466 2013-04-23  Shiva Chen  <shiva0217@gmail.com>
22468         * lra-assigns.c (find_hard_regno_for): Use lra_reg_val_equal_p
22469         to check the register content is equal or not.
22470         * lra-constraints.c (match_reload): Use lra_assign_reg_val
22471         to assign register content record.
22472         * lra-eliminations.c (update_reg_eliminate): Use
22473         lra_update_reg_val_offset to update register content offset.
22474         * lra-int.h (struct lra_reg): Add offset member.
22475         (lra_reg_val_equal_p): New static inline function.
22476         (lra_update_reg_val_offset): New static inline function.
22477         (lra_assign_reg_val): New static inline function.
22478         * lra.c (lra_create_new_reg): Use lra_assign_reg_val
22479         to assign register content record.
22480         (initialize_lra_reg_info_element): Initial offset to zero.
22482 2013-04-23  Catherine Moore  <clm@codesourcery.com>
22484         * config/mips/mips.md (*movhi_internal, *movqi_internal): New
22485         operands.  Record compression.
22487 2013-04-23  Xinliang David Li  <davidxl@google.com>
22489         * cfghhooks.c (dump_bb_for_graph): Support 'slim' graph dump.
22491 2013-04-23  Richard Biener  <rguenther@suse.de>
22493         PR middle-end/57036
22494         * tree-inline.c (copy_edges_for_bb): Add can_make_abnormal_goto
22495         parameter, only add abnormal goto edges from the copied body
22496         if the call could perform abnormal gotos.
22497         (copy_cfg_body): Adjust.
22499 2013-04-23  Sofiane Naci  <sofiane.naci@arm.com>
22501         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add simd attribute.
22503 2013-04-23  Andreas Schwab  <schwab@linux-m68k.org>
22505         * coretypes.h (gimple_stmt_iterator): Add struct to make
22506         compatible with C.
22508 2013-04-23  Richard Biener  <rguenther@suse.de>
22510         PR tree-optimization/57026
22511         * tree-vrp.c (simplify_conversion_using_ranges): Do not propagate
22512         from SSA names occuring in abnormal PHI nodes.
22514 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22516         * lto/lto.c (print_lto_report_1): Fix LTO report names.
22518 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22520         * lto/lto.c (print_lto_report_1): Declare early.
22521         (read_cgraph_and_symbols): Call print_lto_report_1 early.
22523 2013-04-22  Andi Kleen  <ak@linux.intel.com>
22525         * common.opt (-flto-report-wpa): Add.
22526         * doc/invoke.texi (-flto-report-wpa): Add.
22527         * lto/lto.c (do_whole_program_analysis): Check for lto-report-wpa.
22528         (lto_main): dito.
22530 2013-04-22  Xinliang David Li  <davidxl@google.com>
22532         * graph.c (draw_cfg_node_succ_edges): Add branch probility as label.
22533         * cfghhooks.c (dump_bb_for_graph): Dump profile count and frquency.
22534         * Makefile.in: New dependency
22536             David Daney <ddaney.cavm@gmail.com>
22538         * configure.ac (gcc_cv_as_micromips_support):  Use the
22539         --fatal-warnings option.
22540         * configure: Regenerate.
22542 2013-04-22  Marek Polacek  <polacek@redhat.com>
22544         PR sanitizer/56990
22545         * tsan.c (instrument_expr): Don't instrument expression
22546         in case its size is zero.
22548 2013-04-22  Uros Bizjak  <ubizjak@gmail.com>
22550         PR target/57032
22551         Revert:
22552         2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
22554         * config/alpha/alpha.c (TARGET_LRA_P): New define.
22556 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22558         * coretypes.h (gimple_stmt_iterator_d): Forward declare.
22559         (gimple_stmt_iterator): New typedef.
22560         * gimple.h (gimple_stmt_iterator): Rename to...
22561         (gimple_stmt_iterator_d): ... This.
22562         * doc/tm.texi.in (TARGET_FOLD_BUILTIN): Detail restriction that
22563         trees be valid for GIMPLE and GENERIC.
22564         (TARGET_GIMPLE_FOLD_BUILTIN): New.
22565         * gimple-fold.c (gimple_fold_call): Call target hook
22566         gimple_fold_builtin.
22567         * hooks.c (hook_bool_gsiptr_false): New.
22568         * hooks.h (hook_bool_gsiptr_false): New.
22569         * target.def (fold_stmt): New.
22570         * doc/tm.texi: Regenerate.
22572 2013-04-22  Vladimir Makarov  <vmakarov@redhat.com>
22574         PR target/57018
22575         * lra-eliminations.c (mark_not_eliminable): Prevent elimination of
22576         a set sp if no stack realignment.
22578 2013-04-22  Nick Clifton  <nickc@redhat.com>
22580         * config.gcc (tilegx-linux): Extend extra_objs rather than
22581         overwriting it.
22582         (tilepro-linux): Likewise.
22584 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22586         * config/aarch64/aarch64-builtins.c
22587         (CF): Remove.
22588         (CF0, CF1, CF2, CF3, CF4, CF10): New.
22589         (VAR<1-12>): Add MAP parameter.
22590         (BUILTIN_*): Likewise.
22591         * config/aarch64/aarch64-simd-builtins.def: Set MAP parameter.
22592         * config/aarch64/aarch64-simd.md (aarch64_sshl_n<mode>): Remove.
22593         (aarch64_ushl_n<mode>): Likewise.
22594         (aarch64_sshr_n<mode>): Likewise.
22595         (aarch64_ushr_n<mode>): Likewise.
22596         (aarch64_<maxmin><mode>): Likewise.
22597         (aarch64_sqrt<mode>): Likewise.
22598         * config/aarch64/arm_neon.h (vshl<q>_n_*): Use new builtin names.
22599         (vshr<q>_n_*): Likewise.
22601 2013-04-22  James Greenhalgh  <james.greenhalgh@arm.com>
22603         * config/aarch64/aarch64-builtins.c
22604         (aarch64_simd_builtin_type_mode): Handle SF types.
22605         (sf_UP): Define.
22606         (BUILTIN_GPF): Define.
22607         (aarch64_init_simd_builtins): Handle SF types.
22608         * config/aarch64/aarch64-simd-builtins.def (frecpe): Add support.
22609         (frecps): Likewise.
22610         (frecpx): Likewise.
22611         * config/aarch64/aarch64-simd.md
22612         (simd_types): Update simd_frcp<esx> to simd_frecp<esx>.
22613         (aarch64_frecpe<mode>): New.
22614         (aarch64_frecps<mode>): Likewise.
22615         * config/aarch64/aarch64.md (unspec): Add UNSPEC_FRECP<ESX>.
22616         (v8type): Add frecp<esx>.
22617         (aarch64_frecp<FRECP:frecp_suffix><mode>): New.
22618         (aarch64_frecps<mode>): Likewise.
22619         * config/aarch64/iterators.md (FRECP): New.
22620         (frecp_suffix): Likewise.
22621         * config/aarch64/arm_neon.h
22622         (vrecp<esx><qsd>_<fd><32, 64>): Convert to using builtins.
22624 2013-04-22  Christian Bruel  <christian.bruel@st.com>
22626         PR target/56995
22627         * config/sh/sh.h (enum reg_class): Remove DF_HI_REGS.
22628         (REG_CLASS_NAMES): Idem.
22629         (REG_CLASS_CONTENTS): Idem.
22630         (REGCLASS_HAS_FP_REG): Idem.
22631         * config/sh/sh.c (sh_cannot_change_mode_class): Idem.
22632         (sh_conditional_register_usage): Idem.
22634 2013-04-21  Jeff Law  <law@redhat.com>
22636         * tree-ssa-forwprop.c (simplify_conversion_from_bitmask): New function.
22637         (ssa_forward_propagate_and_combine): Use it.
22639 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22641         * lra.c: Update the flow chart diagram.
22643 2013-04-19  Vladimir Makarov  <vmakarov@redhat.com>
22645         PR rtl-optimization/56847
22646         * lra-constraints.c (process_alt_operands): Discourage alternative
22647         with non-matche doffsettable memory constraint fro memory with
22648         known offset.
22650 2013-04-19  Richard Biener  <rguenther@suse.de>
22652         PR tree-optimization/56982
22653         * builtins.def (BUILT_IN_LONGJMP): longjmp is not a leaf
22654         function.
22655         * gimplify.c (gimplify_call_expr): Notice special calls.
22656         (gimplify_modify_expr): Likewise.
22657         * tree-cfg.c (make_abnormal_goto_edges): Handle setjmp-like
22658         abnormal control flow receivers.
22659         (call_can_make_abnormal_goto): Handle cfun->calls_setjmp
22660         in the same way as cfun->has_nonlocal_labels.
22661         (gimple_purge_dead_abnormal_call_edges): Likewise.
22662         (stmt_starts_bb_p): Make setjmp-like abnormal control flow
22663         receivers start a basic-block.
22665 2013-04-19  Richard Biener  <rguenther@suse.de>
22667         * tree-vectorizer.h (struct _slp_instance): Move load_permutation
22668         member ...
22669         (struct _slp_tree): ... here.  Make it a vector of unsigned ints.
22670         (SLP_INSTANCE_LOAD_PERMUTATION): Remove.
22671         (SLP_TREE_LOAD_PERMUTATION): Add.
22672         (vect_transform_slp_perm_load): Adjust prototype.
22673         * tree-vect-slp.c (vect_free_slp_tree): Adjust.
22674         (vect_free_slp_instance): Likewise.
22675         (vect_create_new_slp_node): Likewise.
22676         (vect_supported_slp_permutation_p): Remove.
22677         (vect_slp_rearrange_stmts): Adjust.
22678         (vect_supported_load_permutation_p): Likewise.  Inline
22679         vect_supported_slp_permutation_p here.
22680         (vect_analyze_slp_instance): Compute load permutations per
22681         slp node instead of per instance.
22682         (vect_get_slp_defs): Adjust.
22683         (vect_transform_slp_perm_load): Likewise.
22684         (vect_schedule_slp_instance): Remove redundant code.
22685         (vect_schedule_slp): Remove hack for PR56270, add it ...
22686         * tree-vect-stmts.c (vectorizable_load): ... here, do not
22687         CSE loads for SLP.  Adjust.
22689 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22691         * config/arm/arm.c (load_multiple_sequence, ldm_stm_operation_p): Fix
22692         spelling in two comments.
22694 2013-04-19  Greta Yorsh  <Greta.Yorsh@arm.com>
22696         PR target/56797
22697         * config/arm/arm.c (load_multiple_sequence): Require SP
22698         as base register for loads if SP is in the register list.
22700 2013-04-19  Martin Jambor  <mjambor@suse.cz>
22702         PR tree-optimization/56718
22703         * ipa-cp.c (ipa_value_from_known_type_jfunc): Moved...
22704         * ipa-prop.c (ipa_binfo_from_known_type_jfunc): ...here, renamed
22705         and made public.  Adjusted all callers.
22706         (ipa_intraprocedural_devirtualization): New function.
22707         * ipa-prop.h (ipa_binfo_from_known_type_jfunc): Declare.
22708         (ipa_intraprocedural_devirtualization): Likewise.
22709         * Makefile.in (tree-ssa-pre.o): Add ipa-prop.h to dependencies.
22711 2013-04-19  Richard Biener  <rguenther@suse.de>
22713         PR tree-optimization/57000
22714         * tree-ssa-reassoc.c (pass_reassoc): Add TODO_update_ssa_only_virtuals.
22716 2013-04-19  Terry Guo  <terry.guo@arm.com>
22718         * config/arm/cortex-m4-fpu.md (cortex_m4_v): Delete cpu unit.
22719         Replace with ...
22720         (cortex_m4_v_a,  cortex_m4_v_b): ... new cpu units.
22721         (cortex_m4_v, cortex_m4_exa_va, cortex_m4_exb_vb): New reservations.
22722         (cortex_m4_fmacs): Use new reservations.
22723         (cortex_m4_f_load, cortex_m4_f_store): Likewise.
22725 2013-04-18  Vladimir Makarov  <vmakarov@redhat.com>
22727         PR rtl-optimization/56999
22728         * lra-coalesce.c (coalescable_pseudo_p): Remove 2nd parameter and
22729         related code.
22730         (lra_coalesce): Remove split_origin_bitmap and related code.
22731         * lra.c (lra): Coalesce after undoing inheritance.  Recreate live
22732         ranges if necessary.
22734 2013-04-18  Uros Bizjak  <ubizjak@gmail.com>
22736         * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers):
22737         New array.
22738         (ix86_expand_call): Remove clobbered_registers array and use
22739         x86_64_ms_sysv_extra_clobbered_registers instead.
22740         * config/i386/i386.h (x86_64_ms_sysv_extra_clobbered_registers):
22741         Declare here.
22742         * config/i386/predicates.md (call_rex64_ms_sysv_operation): New
22743         predicate.
22744         * config/i386/i386.md (*call_rex64_ms_sysv): Use
22745         call_rex64_ms_sysv_operation predicate.  Remove explicit clobbers.
22746         (*call_value_rex64_ms_sysv): Ditto.
22748 2013-04-18  Cary Coutant  <ccoutant@google.com>
22750         * dwarf2out.c (output_pubnames): Check die_perennial_p of
22751         parent instead of die_mark.
22753 2013-04-18  Diego Novillo  <dnovillo@google.com>
22755         * gimple.c (create_gimple_tmp): New.
22756         (get_expr_type): New.
22757         (build_assign): New.
22758         (build_type_cast): New.
22759         * gimple.h (enum ssa_mode): Define.
22760         (gimple_seq_set_location): New.
22761         * asan.c (build_check_stmt): Change some gimple_build_* calls
22762         to use build_assign and build_type_cast.
22764 2013-04-18  Richard Biener  <rguenther@suse.de>
22766         * tree-vect-data-refs.c (vect_analyze_group_access): Properly
22767         handle negative step.  Remove redundant checks.
22768         (vect_create_data_ref_ptr): Avoid ICEs with non-constant steps.
22769         * tree-vect-stmts.c (vectorizable_load): Instead of asserting
22770         for negative step and grouped loads fail to vectorize.
22772 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22774         * emit-rtl.c (reset_insn_used_flags): New function.
22775         (reset_all_used_flags): Use it.
22776         (verify_insn_sharing): New function.
22777         (verify_rtl_sharing): Fix verification for SEQUENCEs.
22779 2013-04-18  Jakub Jelinek  <jakub@redhat.com>
22781         PR tree-optimization/56984
22782         * tree-vrp.c (register_edge_assert_for_2): For (x >> M) < N
22783         and (x >> M) >= N don't register any assertion if N << M is the
22784         minimum value.
22786 2013-04-18  Steven Bosscher  <steven@gcc.gnu.org>
22788         * lower-subreg.c (resolve_simple_move): If called self-recursive,
22789         do not delete_insn insns that have not yet been emitted, only
22790         unlink them with remove_insn.
22791         * df-scan.c (df_insn_delete): Revert r197492.
22793 2013-04-17  Steven Bosscher  <steven@gcc.gnu.org>
22795         * emit-rtl.c (link_insn_into_chain): Handle chaining of SEQUENCEs.
22796         * reorg.c (emit_delay_sequence): Simplify with emit-rtl API.
22798 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22800         * config/arm/arm.md (movsicc_insn): Convert define_insn into
22801         define_insn_and_split.
22802         (and_scc,ior_scc,negscc): Likewise.
22803         (cmpsi2_addneg, subsi3_compare): Convert to named patterns.
22805 2013-04-17  Greta Yorsh  <Greta.Yorsh@arm.com>
22807         * config/arm/arm.c (use_return_insn): Return 0 for targets that
22808         can benefit from using a sequence of LDRD instructions in epilogue
22809         instead of a single LDM instruction.
22811 2013-04-17  Manuel López-Ibáñez  <manu@gcc.gnu.org>
22813         PR 45688
22814         * doc/extend.texi: Fix typo.
22816 2013-04-17  Richard Biener  <rguenther@suse.de>
22818         * tree-vect-slp.c (vect_build_slp_tree_1): Split out from ...
22819         (vect_build_slp_tree): ... here.
22820         (vect_build_slp_tree_1): Compute which stmts of the SLP group
22821         match.  Remove special-casing of mismatched complex loads.
22822         (vect_build_slp_tree): Based on the result from vect_build_slp_tree_1
22823         re-try the match with swapped commutative operands.
22824         (vect_supported_load_permutation_p): Remove special-casing of
22825         mismatched complex loads.
22826         (vect_analyze_slp_instance): Adjust.
22828 2013-04-17  Richard Biener  <rguenther@suse.de>
22830         PR rtl-optimization/56921
22831         * cfgloop.h (struct loop): Add simple_loop_desc member.
22832         (struct niter_desc): Mark with GTY(()).
22833         (simple_loop_desc): Do not use aux field but simple_loop_desc.
22834         * loop-iv.c (get_simple_loop_desc): Likewise.
22835         (free_simple_loop_desc): Likewise.
22837         Revert
22838         2013-04-16  Richard Biener  <rguenther@suse.de>
22840         PR rtl-optimization/56921
22841         * loop-init.c (pass_rtl_move_loop_invariants): Add
22842         TODO_do_not_ggc_collect to todo_flags_finish.
22843         (pass_rtl_unswitch): Same.
22844         (pass_rtl_unroll_and_peel_loops): Same.
22845         (pass_rtl_doloop): Same.
22847 2013-04-17  Eric Botcazou  <ebotcazou@adacore.com>
22849         * tree-ssa-alias.c (nonoverlapping_component_refs_of_decl_p): New.
22850         (decl_refs_may_alias_p): Add REF1 and REF2 parameters.
22851         Use nonoverlapping_component_refs_of_decl_p to disambiguate component
22852         references.
22853         (refs_may_alias_p_1): Adjust call to decl_refs_may_alias_p.
22854         * tree-streamer.c (record_common_node): Adjust reference in comment.
22856 2013-04-17  Terry Guo  <terry.guo@arm.com>
22858         * config/arm/cortex-m4.md: Add a new bypass.
22860 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22862         * config/aarch64/aarch64.md (*adds_<optab><mode>_multp2):
22863         New pattern.
22864         (*subs_<optab><mode>_multp2): New pattern.
22865         (*adds_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22866         (*subs_<optab><ALLX:mode>_<GPI:mode>): New pattern.
22868 2013-04-16  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
22870         * config/aarch64/aarch64.md (*adds_mul_imm_<mode>): New pattern.
22871         (*subs_mul_imm_<mode>): New pattern.
22873 2013-04-16  David Edelsohn  <dje.gcc@gmail.com>
22875         PR target/56948
22876         * config/rs6000/vsx.md (vsx_mov<mode>): Add j->r alternative.
22877         (vsx_movti_64bit): Change j->wa to O->wa.  Add n->r alternative.
22878         (vsx_movti_32bit): Change j->wa to O->wa.
22880 2013-04-16  Richard Biener  <rguenther@suse.de>
22882         PR rtl-optimization/56921
22883         * loop-init.c (pass_rtl_move_loop_invariants): Add
22884         TODO_do_not_ggc_collect to todo_flags_finish.
22885         (pass_rtl_unswitch): Same.
22886         (pass_rtl_unroll_and_peel_loops): Same.
22887         (pass_rtl_doloop): Same.
22889 2013-04-16  Greta Yorsh  <Greta.Yorsh@arm.com>
22891         * config/arm/arm.c (emit_multi_reg_push): New declaration
22892         for an existing function.
22893         (arm_emit_strd_push): New function.
22894         (arm_expand_prologue): Used here.
22895         (arm_emit_ldrd_pop): New function.
22896         (arm_expand_epilogue): Used here.
22897         (arm_get_frame_offsets): Update condition.
22898         (arm_emit_multi_reg_pop): Add a special case for load of a single
22899         register with writeback.
22901 2013-04-16  Uros Bizjak  <ubizjak@gmail.com>
22903         * doc/invoke.texi (i386 Option): Reword -mstack-protector-guard
22904         description.
22906 2013-04-16  Richard Biener  <rguenther@suse.de>
22908         PR tree-optimization/56756
22909         * tree-ssa-loop-im.c (struct first_mem_ref_loc_1): New functor.
22910         (first_mem_ref_loc): New.
22911         (execute_sm): Place the load temporarily before a previous
22912         access instead of in the latch edge to ensure its SSA dependencies
22913         are defined at points dominating the load.
22915 2013-04-16  Steven Bosscher  <steven@gcc.gnu.org>
22917         * cfgrtl.c (cfg_layout_merge_blocks): Revert r184005, implement
22918         correct fix by moving header and footer insn to the footer of
22919         the merged basic block.  Clear BB_END of the merged-away block.
22921         PR middle-end/43631
22922         * emit-rtl.c (make_note_raw): New function.
22923         (link_insn_into_chain): New static inline function.
22924         (add_insn): Use it.
22925         (add_insn_before, add_insn_after): Factor insn chain linking code...
22926         (add_insn_before_nobb, add_insn_after_nobb): ...here, new functions
22927         using link_insn_into_chain.
22928         (note_outside_basic_block_p): New helper function for emit_note_after
22929         and emit_note_before.
22930         (emit_note_after): Use nobb variant of add_insn_after if the note
22931         should not be contained in a basic block.
22932         (emit_note_before): Use nobb variant of add_insn_before if the note
22933         should not be contained in a basic block.
22934         (emit_note_copy): Use make_note_raw.
22935         (emit_note): Likewise.
22936         * bb-reorder.c (insert_section_boundary_note): Remove hack to set
22937         BLOCK_FOR_INSN to NULL manually for NOTE_INSN_SWITCH_TEXT_SECTIONS.
22938         * jump.c (cleanup_barriers): Use reorder_insns_nobb to avoid making
22939         the moved barrier the tail of the basic block it follows.
22940         * var-tracking.c (pass_variable_tracking): Add TODO_verify_flow.
22942 2013-04-15  Jakub Jelinek  <jakub@redhat.com>
22944         PR tree-optimization/56962
22945         * gimple-ssa-strength-reduction.c (record_increment): Only set
22946         initializer if gimple_assign_rhs_code is {,POINTER_}PLUS_EXPR and
22947         either rhs1 or rhs2 is equal to c->base_expr.
22949 2013-04-15  Richard Biener  <rguenther@suse.de>
22951         PR tree-optimization/56933
22952         * tree-vectorizer.h (struct _stmt_vec_info): Remove read_write_dep
22953         member.
22954         (GROUP_READ_WRITE_DEPENDENCE): Remove.
22955         (STMT_VINFO_GROUP_READ_WRITE_DEPENDENCE): Likewise.
22956         * tree-vect-data-refs.c (vect_analyze_group_access): Move
22957         dependence check ...
22958         vect_analyze_data_ref_dependence (vect_analyze_data_ref_dependence):
22959         ... here.
22960         * tree-vect-stmts.c (new_stmt_vec_info): Do not initialize
22961         GROUP_READ_WRITE_DEPENDENCE.
22963 2013-04-15  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
22965         * emit-rtl.c (reset_all_used_flags): New function.
22966         (verify_rtl_sharing): Call reset_all_used_flags before and after
22967         performing the checks.
22969 2013-04-15  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
22971         * config/arm/arm.c (const_ok_for_dimode_op): Handle AND case.
22972         * config/arm/arm.md (*anddi3_insn): Change to insn_and_split.
22973         * config/arm/constraints.md (De): New constraint.
22974         * config/arm/neon.md (anddi3_neon): Delete.
22975         (neon_vand<mode>): Expand to standard anddi3 pattern.
22976         * config/arm/predicates.md (imm_for_neon_inv_logic_operand):
22977         Move earlier in the file.
22978         (neon_inv_logic_op2): Likewise.
22979         (arm_anddi_operand_neon): New predicate.
22981 2013-04-15  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
22983         * configure.ac (gcc_cv_ld_as_needed): Set
22984         gcc_cv_ld_as_needed_option, gcc_cv_no_as_needed_option.
22985         Use -z ignore, -z record on *-*-solaris2*.
22986         (HAVE_LD_AS_NEEDED): Update comment.
22987         (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Define.
22988         * configure: Regenerate.
22989         * config.in: Regenerate.
22990         * gcc.c (init_gcc_specs) [USE_LD_AS_NEEDED]: Use
22991         LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION.
22992         * config/sol2.h [HAVE_LD_AS_NEEDED] (USE_LD_AS_NEEDED): Define.
22993         * doc/tm.texi.in (USE_LD_AS_NEEDED): Allow for --as-needed
22994         equivalents.  Fix markup.
22995         * doc/tm.texi: Regenerate.
22997 2013-04-15  Andrew Hsieh  <andrewhsieh.google.com>
22999         * config/i386/i386.opt: New option mstack-protector-guard=.
23000         * config/i386/i386-opts.h: Add enum stack_protector_guard.
23001         * config/i386/i386.h: Define TARGET_SSP_GLOBAL_GUARD and
23002         TARGET_SSP_TLS_GUARD.
23003         * config/i386/i386.c (ix86_option_override_internal): Set
23004         ix86_stack_protector_guard.
23005         * config/i386/i386.md (stack_protect_set): Enable for
23006         TARGET_SSP_TLS_GUARD only.
23007         (stack_protect_set_<mode>): Ditto.
23008         (stack_protect_test): Ditto.
23009         (stack_protect_test_<mode>): Ditto.
23010         * doc/invoke.texi (i386 Option): Document.
23012 2013-04-15  Eric Botcazou  <ebotcazou@adacore.com>
23014         PR target/56890
23015         * config/sparc/sparc.c (enum sparc_mode_class): Add H_MODE value.
23016         (S_MODES): Set H_MODE bit.
23017         (SF_MODES): Set only S_MODE and SF_MODE bits.
23018         (DF_MODES): Set SF_MODES and only D_MODE and DF_MODE bits.
23019         (sparc_init_modes) <MODE_INT>: Set H_MODE bit for sub-word modes.
23020         <MODE_VECTOR_INT>: Do not set SF_MODE for sub-word modes.
23021         <MODE_FLOAT>: Likewise.
23023 2013-04-15  Joey Ye  <joey.ye@arm.com>
23025         * config/arm/arm.c (thumb_far_jump_used_p): Fix typo in comments.
23027 2013-04-15  Joey Ye  <joey.ye@arm.com>
23029         * config/arm/arm.c (thumb1_final_prescan_insn): Assert lr save
23030         for real far jump.
23031         (thumb_far_jump_used_p): Count instruction size and set
23032         far_jump_used.
23034 2013-04-14  Eric Botcazou  <ebotcazou@adacore.com>
23036         * reorg.c (fill_simple_delay_slots): Reindent block of code.
23037         * resource.c (mark_target_live_regs): Reformat conditional block.
23039 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
23041         * sched-deps.c (deps_analyze_insn): Do not check for EH_REGION insn
23042         notes, they are emitted only just before final.
23043         * sched-int.h: Include insn-attr.h before checking INSN_SCHEDULING.
23045 2013-04-13  Steven Bosscher  <steven@gcc.gnu.org>
23047         * emit-rtl.c (remove_insn): Do not call df_insn_delete here.
23048         * cfgrtl.c (delete_insn): Call it here instead.
23049         * lra-spills.c (lra_final_code_change): Use delete_insn.
23050         * haifa-sched.c (sched_remove_insn): Likewise.
23051         * sel-sched-ir.c (return_nop_to_pool): Clear INSN_DELETED_P for nops
23052         returning to the nop pool.
23053         (sel_remove_insn): Simplify the only_disconnect case via remove_insn,
23054         use delete_insn for definitive removal.  Clear BLOCK_FOR_INSN.
23056 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
23058         * doc/tm.texi.in (LOOP_ALIGN): Remove loop note references.
23059         * doc/tm.texi: Regenerated.
23061 2013-04-12  Uros Bizjak  <ubizjak@gmail.com>
23063         * config/i386/i386.c (ix86_hard_regno_mode_ok): Use ANY_QI_REGNO_P in
23064         QImode checks.
23066 2013-04-12  Steven Bosscher  <steven@gcc.gnu.org>
23068         * df-core.c (df_find_def): Compare register numbers.
23069         (df_find_use): Likewise.
23071 2013-04-12  Vladimir Makarov  <vmakarov@redhat.com>
23073         PR target/56903
23074         * config/i386/i386.c (ix86_hard_regno_mode_ok): Add
23075         lra_in_progress for return.
23077 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23079         * config/arm/arm.md (mov_scc,mov_negscc,mov_notscc): Convert
23080         define_insn into define_insn_and_split and emit movsicc patterns.
23082 2013-04-12  Greta Yorsh  <Greta.Yorsh@arm.com>
23084         * config/arm/arm.c (gen_operands_ldrd_strd): Initialize "base".
23086 2013-04-12  Richard Biener  <rguenther@suse.de>
23088         * tree-pass.h (TODO_do_not_ggc_collect): New.
23089         * passes.c (execute_one_ipa_transform_pass): Honor
23090         TODO_do_not_ggc_collect.
23091         (execute_one_pass): Likewise.
23093         Revert
23094         2013-04-10  Richard Biener  <rguenther@suse.de>
23096         * passes.c (init_optimization_passes): Remove reload pass.
23097         * ira.c (do_reload): Merge into ...
23098         (ira): ... this.
23099         (rest_of_handle_reload): Remove.
23100         (pass_reload): Likewise.
23101         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23102         of reload for vzeroupper pass placement.
23104 2013-04-12  Jakub Jelinek  <jakub@redhat.com>
23106         PR tree-optimization/56918
23107         PR tree-optimization/56920
23108         * fold-const.c (int_const_binop_1): Use op1.mul_with_sign (op2, ...)
23109         instead of op1 - op2.  Pass 2 * TYPE_PRECISION (type) as second
23110         argument to rshift method.  For 2 * HOST_BITS_PER_WIDE_INT precision
23111         use wide_mul_with_sign method.
23113 2013-04-12  Richard Biener  <rguenther@suse.de>
23115         * gimple.c (is_gimple_constant): Vector CONSTRUCTORs should
23116         not be considered a gimple constant.
23118 2013-04-12  Marc Glisse  <marc.glisse@inria.fr>
23120         * fold-const.c (const_binop): Handle vector shifts by a scalar.
23121         (fold_binary_loc): Call const_binop also for mixed vector-scalar
23122         operations.
23124 2013-04-12  Manuel López-Ibáñez  <manu@gcc.gnu.org>
23125             Jakub Jelinek  <jakub@redhat.com>
23127         * opts.c: Include diagnostic-color.h.
23128         (common_handle_option): Handle OPT_fdiagnostics_color_.
23129         * Makefile.in (OBJS-libcommon): Add diagnostic-color.o.
23130         (diagnostic.o, opts.o, pretty-print.o): Depend on diagnostic-color.h.
23131         (diagnostic-color.o): New.
23132         * common.opt (fdiagnostics-color, fdiagnostics-color=): New options.
23133         (diagnostic_color_rule): New enum.
23134         * dwarf2out.c (gen_producer_string): Don't print -fdiagnostics-color*.
23135         * langhooks.c (lhd_print_error_function): Add %r "locus" and %R around
23136         the location string.
23137         * diagnostic.def: Add 3rd argument to DEFINE_DIAGNOSTIC_KIND macros,
23138         either NULL, or color kind.
23139         * diagnostic-color.c: New file.
23140         * diagnostic-color.h: New file.
23141         * diagnostic-core.h (DEFINE_DIAGNOSTIC_KIND): Adjust macro for 3
23142         arguments.
23143         * doc/invoke.texi (-fdiagnostics-color): Document.
23144         * pretty-print.h (pp_show_color): Define.
23145         (struct pretty_print_info): Add show_color field.
23146         * diagnostic.c: Include diagnostic-color.h.
23147         (diagnostic_build_prefix): Adjust for 3 argument DEFINE_DIAGNOSTIC_KIND
23148         macros.  Colorize error:, warning: etc. strings and also the location
23149         string.
23150         (diagnostic_show_locus): Colorize the caret line.
23151         * pretty-print.c: Include diagnostic-color.h.
23152         (pp_base_format): Handle %r and %R format specifiers.  Colorize strings
23153         inside of %< %> quotes or quoted through q format modifier.
23155 2013-04-12  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23157         * ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as well.
23159 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23161         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Allow NEG
23162         code in CC_NZ mode.
23163         * config/aarch64/aarch64.md (*neg_<shift><mode>3_compare0): New
23164         pattern.
23166 2013-04-11  Marek Polacek  <polacek@redhat.com>
23168         PR tree-optimization/48184
23169         * params.def (PARAM_ALIGN_THRESHOLD): Increase the minimum value to 1.
23171 2013-04-11  Eric Botcazou  <ebotcazou@adacore.com>
23173         * stor-layout.c (skip_simple_constant_arithmetic): Move to...
23174         * tree.c (skip_simple_constant_arithmetic): ...here and make public.
23175         (skip_simple_arithmetic): Tidy up.
23176         * tree.h (skip_simple_constant_arithmetic): Declare.
23178 2013-04-11  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
23180         * config/aarch64/aarch64.h (REVERSIBLE_CC_MODE): Define.
23182 2013-04-11  Richard Biener  <rguenther@suse.de>
23184         * tree-vect-loop.c (get_initial_def_for_induction): Properly
23185         generate vector constants.
23187 2013-04-11  Richard Biener  <rguenther@suse.de>
23189         PR tree-optimization/56878
23190         * tree-flow.h (outermost_invariant_loop_for_expr): Declare.
23191         * tree-ssa-loop-ivopts.c (outermost_invariant_loop_for_expr):
23192         New function.
23193         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23194         Prefer to align the DR with the most invariant base address.
23196 2013-04-11  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23198         * opts.c (common_handle_option): Fix formatting and add FALLTHRU
23199         comment.
23201 2013-04-11  James Greenhalgh  <james.greenhalgh@arm.com>
23203         * config/aarch64/aarch64-simd.md (aarch64_vcond_internal): Fix
23204         floating-point vector comparisons against 0.
23206 2013-04-11  Jakub Jelinek  <jakub@redhat.com>
23208         PR tree-optimization/56899
23209         * fold-const.c (extract_muldiv_1): Apply distributive law
23210         only if TYPE_OVERFLOW_WRAPS (ctype).
23212 2013-04-11  Bin Cheng  <bin.cheng@arm.com>
23214         PR target/56124
23215         * ira-costs.c (scan_one_insn): Check whether the source rtx of
23216         loading has side effect.
23218 2013-04-10  Steven Bosscher  <steven@gcc.gnu.org>
23220         * config/sparc/sparc.c: Include tree-pass.h.
23221         (TARGET_MACHINE_DEPENDENT_REORG): Do not redefine.
23222         (sparc_reorg): Rename to sparc_do_work_around_errata.  Move to
23223         head of file.  Change return type.  Split off gate function.
23224         (sparc_gate_work_around_errata): New function.
23225         (pass_work_around_errata): New pass definition.
23226         (insert_pass_work_around_errata) New pass insert definition to
23227         insert pass_work_around_errata just after delayed-branch scheduling.
23228         (sparc_option_override): Insert the pass.
23229         * config/sparc/t-sparc (sparc.o): Add TREE_PASS_H dependence.
23231 2013-04-10  David S. Miller  <davem@davemloft.net>
23233         * config/sparc/sparc.h (ASM_CPU_SPEC): Pass -Av8 if -mcpu=supersparc
23234         or -mcpu=hypersparc.
23236         * target.def (cstore_mode): New hook.
23237         * target.h: Include insn-codes.h
23238         * targhooks.c: Likewise.
23239         (default_cstore_mode): New function.
23240         * targhooks.h: Declare it.
23241         * doc/tm.texi.in: New hook slot for TARGET_CSTORE_MODE.
23242         * doc/tm.texi: Rebuild.
23243         * expmed.c (emit_cstore): Obtain cstore boolean result mode using
23244         target hook, rather than inspecting the insn_data.
23245         * config/sparc/sparc.c (sparc_cstore_mode): New function.
23246         (TARGET_CSTORE_MODE): Redefine.
23247         (emit_scc_insn): When TARGET_ARCH64, emit new 64-bit boolean
23248         result patterns.
23249         * config/sparc/predicates.md (cstore_result_operand): New special
23250         predicate.
23251         * config/sparc/sparc.md (cstoresi4, cstoredi4, cstore<F:mode>4):
23252         Use it for operand 0.
23253         (*seqsi_special): Rewrite using 'P' mode iterator on operand 0.
23254         (*snesi_special): Likewise.
23255         (*snesi_zero): Likewise.
23256         (*seqsi_zero): Likewise.
23257         (*sltu_insn): Likewise.
23258         (*sgeu_insn): Likewise.
23259         (*seqdi_special): Make operand 0 and comparison operation be of
23260         DImode.
23261         (*snedi_special): Likewise.
23262         (*snedi_special_vis3): Likewise.
23263         (*neg_snesi_zero): Rename to *neg_snesisi_zero.
23264         (*neg_snesi_sign_extend): Rename to *neg_snesidi_zero.
23265         (*snesi_zero_extend): Delete, covered by 'P' mode iterator.
23266         (*neg_seqsi_zero): Rename to *neg_seqsisi_zero.
23267         (*neg_seqsi_sign_extend): Rename to *neg_seqsidi_zero.
23268         (*seqsi_zero_extend): Delete, covered by 'P' mode iterator.
23269         (*sltu_extend_sp64): Likewise.
23270         (*neg_sltu_insn): Rename to *neg_sltusi_insn.
23271         (*neg_sltu_extend_sp64): Rename to *neg_sltudi_insn.
23272         (*sgeu_extend_sp64): Delete, covered by 'P' mode iterator.
23273         (*neg_sgeu_insn): Rename to *neg_sgeusi_insn.
23274         (*neg_sgeu_extend_sp64): Rename to *neg_sgeudi_insn.
23276 2013-04-10  Yufeng Zhang  <yufeng.zhang@arm.com>
23278         * config/aarch64/aarch64.c (aarch64_print_extension): New function.
23279         (aarch64_start_file): Use the new function.
23281 2013-04-10  Senthil Kumar Selvaraj  <senthil_kumar.selvaraj@atmel.com>
23282             Jason Merrill  <jason@redhat.com>
23284         * common.opt: Add -gdwarf.
23285         * opts.c (common_handle_option): Handle it.
23286         * gcc.c (ASM_DEBUG_SPEC): Don't expect "-2" for DWARF.
23288 2013-04-10  Richard Biener  <rguenther@suse.de>
23290         * passes.c (execute_todo): Do not call ggc_collect conditional here.
23291         (execute_one_ipa_transform_pass): But unconditionally here.
23292         (execute_one_pass): And here.
23293         (init_optimization_passes): Remove reload pass.
23294         * tree-pass.h (TODO_ggc_collect): Remove.
23295         (pass_reload): Likewise.
23296         * ira.c (do_reload): Merge into ...
23297         (ira): ... this.
23298         (rest_of_handle_reload): Remove.
23299         (pass_reload): Likewise.
23300         * config/i386/i386.c (ix86_option_override): Refer to ira instead
23301         of reload for vzeroupper pass placement.
23302         * <everywhere>: Remove TODO_ggc_collect from todo_flags_start
23303         and todo_flags_finish of all passes.
23305 2013-04-10  Richard Biener  <rguenther@suse.de>
23307         * tree-vectorizer.h (struct _slp_oprnd_info): Remove
23308         first_const_oprnd field, rename first_def_type to first_op_type.
23309         * tree-vect-slp.c (vect_create_oprnd_info): Adjust.
23310         (vect_get_and_check_slp_defs): Always use the type of the
23311         operand.  Allow mixed vect_external_def, vect_constant_def types.
23312         (vect_get_constant_vectors): Handle mixed vect_external_def,
23313         vect_constant_def types.
23315 2013-04-10  Joern Rennecke  <joern.rennecke@embecosm.com>
23317         PR tree-optimization/55524
23318         * tree-ssa-math-opts.c
23319         (convert_mult_to_fma): Don't use an fms construct
23320         when we don't have an fms operation, but fnma, and it looks
23321         likely that we'll be able to use the latter.
23323 2013-04-10  Zhouyi Zhou  <yizhouzhou@ict.ac.cn>
23325         * cif-code.def (OVERWRITABLE): Correct the comment for overwritable
23326         function.
23327         * ipa-inline.c (can_inline_edge_p): Let dump mechanism report the
23328         inline fail caused by overwritable functions.
23330 2013-04-10  Chung-Ju Wu  <jasonwucj@gmail.com>
23332         * combine.c (simplify_compare_const): Use GET_MODE_MASK to filter out
23333         unnecessary bits in the constant power of two case.
23335 2013-04-10  Richard Biener  <rguenther@suse.de>
23337         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove
23338         broken code swapping operands.
23339         (vect_build_slp_tree): Do not compute load permutations here.
23340         (vect_analyze_slp_instance): Compute load permutations here,
23341         after building the SLP tree.
23343 2013-04-09  Christian Bruel  <christian.bruel@st.com>
23345         * config/sh/sh.md (barrier_align): Use next/prev_active_insn instead
23346         of next/prev_real_insn.
23348 2013-04-09  Jan Hubicka  <jh@suse.cz>
23350         * ipa.c (cgraph_externally_visible_p, varpool_externally_visible_p):
23351         Drop aliased parameter.
23352         (function_and_variable_visibility): Do not handle alias pairs.
23353         * cgraph.c (varpool_externally_visible_p): Update prototype.
23354         * varpool.c (varpool_add_new_variable): Update.
23356 2013-04-09  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23358         * config/arm/arm.md (minmax_arithsi_non_canon): New pattern.
23360 2013-04-09  Steven Bosscher  <steven@gcc.gnu.org>
23362         * sched-vis.c (print_pattern): Print SEQUENCE of insns as insns.
23364         * config/sparc/sparc.md: Use define_c_enum for "unspec" and "unspecv".
23366 2013-04-09  Marek Polacek  <polacek@redhat.com>
23368         PR tree-optimization/48762
23369         * params.def (PARAM_MAX_CSE_INSNS): Increase the minimum value to 1.
23371 2013-04-09  Richard Biener  <rguenther@suse.de>
23373         * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove code
23374         dealing with cost.
23375         (vect_build_slp_tree): Likewise.
23376         (vect_analyze_slp_cost_1, vect_analyze_slp_cost): New functions
23377         calculating the cost of a SLP instance.
23378         (vect_analyze_slp_instance): Use it from here, after building
23379         the SLP tree.
23381 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23383         PR middle-end/56883
23384         * omp-low.c (expand_omp_for_generic, expand_omp_for_static_nochunk,
23385         expand_omp_for_static_chunk): Use simple_p = true in
23386         force_gimple_operand_gsi calls when assigning to addressable decls.
23388 2013-04-09  Jeff Law  <law@redhat.com>
23390         * tree-vrp.c (simplify_cond_using_ranges): Simplify test of boolean
23391         when the boolean was created by converting a wider object which
23392         had a boolean range.
23394 2013-04-09  Richard Biener  <rguenther@suse.de>
23396         * tree-vectorizer.h (slp_void_p): Remove.
23397         (slp_tree): Typedef before _slp_tree declaration.
23398         (struct _slp_tree): Use a vector of slp_tree as children.
23399         (vect_get_place_in_interleaving_chain): Remove.
23400         * tree-vect-data-refs.c (vect_get_place_in_interleaving_chain):
23401         Move ...
23402         * tree-vect-slp.c (vect_get_place_in_interleaving_chain): ... here
23403         and make static.
23404         (vect_free_slp_tree, vect_print_slp_tree, vect_mark_slp_stmts,
23405         vect_mark_slp_stmts_relevant, vect_slp_rearrange_stmts,
23406         vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations,
23407         vect_schedule_slp_instance, vect_remove_slp_scalar_calls):
23408         Use slp_node instead of slp_void_p and adjust.
23410 2013-04-09  Richard Biener  <rguenther@suse.de>
23412         * tree-ssa-loop-manip.c (rewrite_into_loop_closed_ssa): Avoid
23413         work that is not necessary.
23415 2013-04-09  Jakub Jelinek  <jakub@redhat.com>
23417         PR tree-optimization/56854
23418         * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Don't
23419         forward into clobber stmts if it would change MEM_REF lhs into
23420         non-MEM_REF.
23422 2013-04-09  Maxim Kuvyrkov  <maxim@kugelworks.com>
23424         * tree.c (type_hash_lookup, type_hash_add): Make static.
23425         * tree.h (type_hash_lookup, type_hash_add): Remove global declarations.
23427 2013-04-09  Richard Biener  <rguenther@suse.de>
23429         * tree.h (unsave_expr_now): Remove.
23430         * tree-inline.c (mark_local_for_remap_r): Remove.
23431         (unsave_expr_1): Likewise.
23432         (unsave_r): Likewise.
23433         (unsave_expr_now): Likewise.
23434         * tree-ssa-copy.c (replace_exp_1): Use unshare_expr.
23435         (propagate_tree_value): Likewise.
23437 2013-04-08  Steven Bosscher  <steven@gcc.gnu.org>
23439         * doc/rtl.texi (sequence): Rewrite documentation to match the
23440         current use of SEQUENCE rtl objects.
23441         * rtl.def (SEQUENCE): Likewise.
23443         * doc/rtl.texi (NOTE_INSN_EH_REGION_BEG, NOTE_INSN_EH_REGION_END):
23444         Update documentation.
23445         (NOTE_INSN_LOOP_BEG, NOTE_INSN_LOOP_END, NOTE_INSN_LOOP_CONT,
23446         NOTE_INSN_LOOP_VTOP): Remove documentation for non-existing notes.
23448         * reg-notes.def (REG_EH_CONTEXT): Remove unused note.
23450 2013-04-08  Teresa Johnson  <tejohnson@google.com>
23452         * basic-block.h (GCOV_COMPUTE_SCALE): Define.
23453         * ipa-inline-analysis.c (param_change_prob): Use helper rounding divide
23454         methods.
23455         (estimate_edge_size_and_time): Add comment to suggest using rounding
23456         methods.
23457         (estimate_node_size_and_time): Ditto.
23458         (remap_edge_change_prob): Use helper rounding divide methods.
23459         * value-prof.c (gimple_divmod_fixed_value_transform): Ditto.
23460         (gimple_mod_pow2_value_transform): Ditto.
23461         (gimple_mod_subtract_transform): Ditto.
23462         (gimple_ic_transform): Ditto.
23463         (gimple_stringops_transform): Ditto.
23464         * stmt.c (conditional_probability): Ditto.
23465         (emit_case_dispatch_table): Ditto.
23466         * lto-cgraph.c (merge_profile_summaries): Ditto.
23467         * tree-optimize.c (execute_fixup_cfg): Ditto.
23468         * cfgcleanup.c (try_forward_edges): Ditto.
23469         * cfgloopmanip.c (scale_loop_profile): Ditto.
23470         (loopify): Ditto.
23471         (duplicate_loop_to_header_edge): Ditto.
23472         (lv_adjust_loop_entry_edge): Ditto.
23473         * tree-vect-loop.c (vect_transform_loop): Ditto.
23474         * profile.c (compute_branch_probabilities): Ditto.
23475         * cfgbuild.c (compute_outgoing_frequencies): Ditto.
23476         * lto-streamer-in.c (input_cfg): Ditto.
23477         * gimple-streamer-in.c (input_bb): Ditto.
23478         * ipa-cp.c (update_profiling_info): Ditto.
23479         (update_specialized_profile): Ditto.
23480         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge): Ditto.
23481         * cfg.c (update_bb_profile_for_threading): Add comment to suggest using
23482         rounding methods.
23483         * sched-rgn.c (compute_dom_prob_ps): Ditto.
23484         (compute_trg_info): Ditto.
23485         * cfgrtl.c (force_nonfallthru_and_redirect): Ditto.
23486         (purge_dead_edges): Ditto.
23487         * loop-unswitch.c (unswitch_loop): Ditto.
23488         * cgraphclones.c (cgraph_clone_edge): Ditto.
23489         (cgraph_clone_node): Ditto.
23490         * tree-inline.c (copy_bb): Ditto.
23491         (copy_edges_for_bb): Ditto.
23492         (initialize_cfun): Ditto.
23493         (copy_cfg_body): Ditto.
23494         (expand_call_inline): Ditto.
23496 2013-04-08  Kai Tietz  <ktietz@redhat.com>
23498         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Replaced
23499         TARGET_CYGWIN64 by TARGET_64BIT.
23501 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23503         * config/epiphany/epiphany.md (GPR_1): New constant.
23504         (define_expand "mov<mode>cc): FAIL if gen_compare_reg returned 0.
23505         * config/epiphany/epiphany.c (gen_compare_reg):
23506         For flag_finite_math_only, avoid swapping operands when r0 and/or r1
23507         is already in place.
23508         Use GPR_0 / GPR_1 instead of 0/1 for r0/r1 register numbers.
23509         Don't require being called during rtl expansion; If y operlaps r0,
23510         return 0.
23511         (epiphany_compute_frame_size, epiphany_expand_prologue): Use GPR_1.
23512         (epiphany_expand_epilogue): Likewise.
23514         * config/epiphany/epiphany.c (epiphany_select_cc_mode):
23515         Don't use CC_FPmode for ORDERED / UNORDERED.
23516         * config/epiphany/epiphany.md (cmpsf_ord): Make pattern unconditional.
23518         * config/epiphany/constraints.md (CnL): New constraint.
23519         * config/epiphany/epiphany.md (addsi3_i): Add r/r/CnL alternative.
23520         * config/epiphany/predicates.md (add_operand): Allow 1024.
23522         * config/epiphany/epiphany.md (logical_op): New code iterator.
23523         (op_mnc): New code attribute.
23524         (<op_mnc>_f, mov_f, cstoresi4): New patterns.
23525         (mov_f+1, mov_f+2): New peephole2 patterns.
23527         * config/epiphany/epiphany.md (mov_f+2): New peephole2 pattern.
23528         (cstoresi4): Also allow re-use of zero result when doing a NE
23529         comparison to a non-zero operand.
23530         Use (clobber (scratch)) for first insn if the gpr output is not needed.
23532         * config/epiphany/epiphany.md (<insn_opname>v2si3):
23533         Use gen_addsi3_i / gen_subsi3_i.
23535 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23537         PR c++/34949
23538         PR c++/50243
23539         * tree-eh.c (optimize_clobbers): Only remove clobbers if bb doesn't
23540         contain anything but clobbers, at most one __builtin_stack_restore,
23541         optionally debug stmts and final resx, and if it has at least one
23542         incoming EH edge.  Don't check for SSA_NAME on LHS of a clobber.
23543         (sink_clobbers): Don't check for SSA_NAME on LHS of a clobber.
23544         Instead of moving clobbers with MEM_REF LHS with SSA_NAME address
23545         which isn't defaut definition, remove them.
23546         (unsplit_eh, cleanup_empty_eh): Use single_{pred,succ}_{p,edge}
23547         instead of EDGE_COUNT comparisons or EDGE_{PRED,SUCC}.
23548         * tree-ssa-ccp.c (execute_fold_all_builtins): Remove clobbers
23549         with MEM_REF LHS with SSA_NAME address.
23551 2013-04-08  Jeff Law  <law@redhat.com>
23553         * gimple.c (canonicalize_cond_expr_cond): Rewrite x ^ y into x != y.
23555 2013-04-08  Richard Biener  <rguenther@suse.de>
23557         * gimple-pretty-print.c (debug_gimple_stmt): Do not print
23558         extra newline.
23559         * tree-vect-loop.c (vect_determine_vectorization_factor): Dump
23560         determined vector type.
23561         (vect_analyze_data_refs): Likewise.
23562         (vect_get_new_vect_var): Adjust.
23563         (vect_create_destination_var): Preserve SSA name versions.
23564         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Do
23565         not dump anything here.
23567 2013-04-08  Joern Rennecke  <joern.rennecke@embecosm.com>
23569         * config/epiphany/epiphany.h (struct GTY (()) machine_function):
23570         Add member lr_slot_known.
23571         * config/epiphany/epiphany.md (reload_insi_ra): Compute lr_slot_offs
23572         if necessary.
23573         * config/epiphany/epiphany.c (epiphany_compute_frame_size):
23574         Remove code that sets lr_slot_offset according to what a previous
23575         version of epiphany_emit_save_restore used to do.
23576         (epiphany_emit_save_restore): When doing an lr save or restore,
23577         set/verify lr_slot_known and lr_slot_offset.
23579 2013-04-08  Xinyu Qi  <xyqi@marvell.com>
23581         PR target/54338
23582         * config/arm/arm.h (REG_CLASS_CONTENTS): Include IWMMXT_GR_REGS
23583         in ALL_REGS.
23585 2013-04-08  Richard Biener  <rguenther@suse.de>
23587         * alias.c (find_base_term): Fix thinko in previous change.
23589 2013-04-08  Jakub Jelinek  <jakub@redhat.com>
23591         * tree-loop-distribution.c (const_with_all_bytes_same): New function.
23592         (generate_memset_builtin): Only handle integer_all_onesp as -1 val if
23593         TYPE_PRECISION is equal to mode bitsize.  Use const_with_all_bytes_same
23594         if possible to compute val.
23595         (classify_partition): Verify CONSTRUCTOR doesn't have any elts.
23596         For QImode integers don't require anything about precision.  Use
23597         const_with_all_bytes_same to find out if the constant doesn't have
23598         repeated bytes in it.
23600 2013-04-08  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
23602         * config/s390/s390.c (s390_expand_insv): Only accept insertions
23603         within mode size.
23605 2013-04-08  Marek Polacek  <polacek@redhat.com>
23607         PR rtl-optimization/48182
23608         * params.def (PARAM_MIN_CROSSJUMP_INSNS): Increase the minimum
23609         value to 1.
23611 2013-04-06  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
23613         PR target/55487
23614         * config/pa/pa.c (legitimize_pic_address): Before incrementing label
23615         nuses, make sure we have a label.
23617 2013-04-05  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
23619         PR target/56843
23620         * config/rs6000/rs6000.c (rs6000_emit_swdiv_high_precision): Remove.
23621         (rs6000_emit_swdiv_low_precision): Remove.
23622         (rs6000_emit_swdiv): Rewrite to handle between one and four
23623         iterations of Newton-Raphson generally; modify required number of
23624         iterations for some cases.
23625         * config/rs6000/rs6000.h (RS6000_RECIP_HIGH_PRECISION_P): Remove.
23627 2013-04-05  Steven Bosscher  <steven@gcc.gnu.org>
23629         * bb-reorder.c (fix_crossing_unconditional_branches): Remove a
23630         set-but-unused variable.
23632         * cgraph.c (cgraph_release_function_body): Clear cfun->cfg to make
23633         basic blocks of released function bodies garbage-collectable.
23635         * ree.c (find_and_remove_re): Do not call df_finish_pass here.
23636         (struct rtl_opt_pass): Add TODO_df_finish.
23638         * rtl.def (DEFINE_SUBST, DEFINE_SUBST_ATTR): Add documentation.
23640 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23642         * config/arm/constraints.md (q): New constraint.
23643         * config/arm/ldrdstrd.md: New file.
23644         * config/arm/arm.md (ldrdstrd.md) New include.
23645         (arm_movdi): Use "q" instead of "r" constraint
23646         for double-word memory access.
23647         (movdf_soft_insn): Likewise.
23648         * config/arm/vfp.md (movdi_vfp): Likewise.
23649         * config/arm/t-arm (MD_INCLUDES): Add ldrdstrd.md.
23650         * config/arm/arm-protos.h (gen_operands_ldrd_strd): New declaration.
23651         * config/arm/arm.c (gen_operands_ldrd_strd): New function.
23652         (mem_ok_for_ldrd_strd): Likewise.
23653         (output_move_double): Update assertion.
23655 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23657         * config/arm/arm.md: Comment on splitting Thumb1 patterns.
23659 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23661         * config/arm/arm.md (arm_smax_insn): Convert define_insn into
23662         define_insn_and_split.
23663         (arm_smin_insn,arm_umaxsi3,arm_uminsi3): Likewise.
23665 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23667         * config/arm/arm.md (arm_ashldi3_1bit): Convert define_insn into
23668         define_insn_and_split.
23669         (arm_ashrdi3_1bit,arm_lshrdi3_1bit): Likewise.
23670         (shiftsi3_compare): New pattern.
23671         (rrx): New pattern.
23672         * config/arm/unspecs.md (UNSPEC_RRX): New.
23674 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23676         * config/arm/arm.md (negdi_extendsidi): New pattern.
23677         (negdi_zero_extendsidi): Likewise.
23679 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23681         * config/arm/arm.md (andsi_iorsi3_notsi): Convert define_insn into
23682         define_insn_and_split.
23683         (arm_negdi2,arm_abssi2,arm_neg_abssi2): Likewise.
23684         (arm_cmpdi_insn,arm_cmpdi_unsigned): Likewise.
23686 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23688         * config/arm/arm.md (arm_subdi3): Convert define_insn into
23689         define_insn_and_split.
23690         (subdi_di_zesidi,subdi_di_sesidi): Likewise.
23691         (subdi_zesidi_di,subdi_sesidi_di,subdi_zesidi_zesidi): Likewise.
23693 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23695         * config/arm/arm.md (subsi3_carryin): New pattern.
23696         (subsi3_carryin_const): Likewise.
23697         (subsi3_carryin_compare,subsi3_carryin_compare_const): Likewise.
23698         (subsi3_carryin_shift,rsbsi3_carryin_shift): Likewise.
23700 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23702         * config/arm/arm.md (incscc,arm_incscc,decscc,arm_decscc): Delete.
23704 2013-04-05  Greta Yorsh  <Greta.Yorsh@arm.com>
23706         * config/arm/arm.md (addsi3_carryin_<optab>): Set attribute predicable.
23707         (addsi3_carryin_alt2_<optab>,addsi3_carryin_shift_<optab>): Likewise.
23709 2013-04-05  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23711         * config/arm/arm.c (arm_expand_builtin): Change fcode
23712         type to unsigned int.
23714 2013-04-05  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23716         * doc/invoke.texi (ARM Options): Document cortex-a53 support.
23718 2013-04-04  Ian Lance Taylor  <iant@google.com>
23720         * doc/standards.texi (Standards): The Go frontend supports the Go 1
23721         language standard.
23723 2013-04-04  Steven Bosscher  <steven@gcc.gnu.org>
23725         PR middle-end/56729
23726         * df-scan.c (df_insn_delete): Disable failing assert.
23728 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23730         * config/arm/arm-protos.h (arm_builtin_vectorized_function):
23731         New function prototype.
23732         * config/arm/arm.c (TARGET_VECTORIZE_BUILTINS): Define.
23733         (TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION): Likewise.
23734         (arm_builtin_vectorized_function): New function.
23736 2013-04-04  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23738         * config/arm/arm_neon_builtins.def: New file.
23739         * config/arm/arm.c (neon_builtin_data): Move contents to
23740         arm_neon_builtins.def.
23741         (enum arm_builtins): Include neon builtin definitions.
23742         (ARM_BUILTIN_NEON_BASE): Move from enum to macro.
23743         * config/arm/t-arm (arm.o): Add dependency on arm_neon_builtins.def.
23745 2013-04-04  Marek Polacek  <polacek@redhat.com>
23747         PR tree-optimization/48186
23748         * predict.c (maybe_hot_frequency_p): Return false if
23749         HOT_BB_FREQUENCY_FRACTION is 0.
23750         (cgraph_maybe_hot_edge_p): Likewise.
23752 2013-04-04  Richard Biener  <rguenther@suse.de>
23754         PR tree-optimization/56826
23755         * tree-vect-slp.c (vect_build_slp_tree): Compute ncopies
23756         more accurately.
23758 2013-04-04  Richard Biener  <rguenther@suse.de>
23760         PR tree-optimization/56213
23761         * tree-vect-data-refs.c (vect_check_strided_load): Remove.
23762         (vect_analyze_data_refs): Allow all non-nested loads as strided loads.
23764 2013-04-04  Richard Biener  <rguenther@suse.de>
23766         PR tree-optimization/56837
23767         * tree-loop-distribution.c (classify_partition): For non-zero
23768         values require that the value has the same precision as its
23769         mode to be useful as memset value.
23771 2013-04-03  Nick Clifton  <nickc@redhat.com>
23773         * config/v850/v850e3v5.md (fmasf4): Use fmaf.s on E3V5 architectures.
23774         (fmssf4): Use fmsf.s on E3V5 architectures.
23775         (fnmasf4): Use fnmaf.s on E3V5 architectures.
23776         (fnmssf4): Use fnmsf.s on E3V5 architectures.
23778 2013-04-03  Jeff Law  <law@redhat.com>
23780         * Makefile.in (lra-constraints.o): Depend on $(OPTABS_H).
23781         (lra-eliminations.o): Likewise.
23783 2013-04-03  Teresa Johnson  <tejohnson@google.com>
23785         * gcov-io.c (compute_working_sets): Moved most of body of old
23786         compute_working_sets here from profile.c.
23787         * gcov-io.h (NUM_GCOV_WORKING_SETS): Moved here from profile.c.
23788         (gcov_working_set_t): Moved typedef here from basic-block.h
23789         (compute_working_set): Declare.
23790         * profile.c (NUM_GCOV_WORKING_SETS): Moved to gcov-io.h.
23791         (get_working_sets): Renamed from compute_working_set,
23792         replace most of body with call to new compute_working_sets.
23793         (get_exec_counts): Replace call to compute_working_sets
23794         to get_working_sets.
23795         * profile.h (get_working_sets): Renamed from compute_working_set.
23796         * lto-cgraph.c (input_symtab): Replace call to compute_working_sets
23797         to get_working_sets.
23798         * basic-block.h (gcov_working_set_t): Moved to gcov-io.h.
23799         * gcov-dump.c (dump_working_sets): New function.
23801 2013-04-03  Kenneth Zadeck  <zadeck@naturalbridge.com>
23803         * hwint.c (sext_hwi, zext_hwi): New functions.
23804         * hwint.h (HOST_BITS_PER_HALF_WIDE_INT, HOST_HALF_WIDE_INT,
23805         HOST_HALF_WIDE_INT_PRINT, HOST_HALF_WIDE_INT_PRINT_C,
23806         HOST_HALF_WIDE_INT_PRINT_DEC, HOST_HALF_WIDE_INT_PRINT_DEC_C,
23807         HOST_HALF_WIDE_INT_PRINT_UNSIGNED, HOST_HALF_WIDE_INT_PRINT_HEX,
23808         HOST_HALF_WIDE_INT_PRINT_HEX_PURE): New symbols.
23809         (sext_hwi, zext_hwi): New functions.
23811 2013-04-03  Jeff Law  <law@redhat.com>
23813         PR tree-optimization/56799
23814         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Bring
23815         back test for widening conversion erroneously dropped in prior change.
23817 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23819         PR target/56809
23820         * config/aarch64/aarch64.c (is_jump_table): Use next_active_insn
23821         instead of next_real_insn.
23823 2013-04-03  Marek Polacek  <polacek@redhat.com>
23825         PR sanitizer/55702
23826         * tsan.c (instrument_func_exit): Allow BUILT_IN_RETURN functions.
23828 2013-04-03  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
23830         PR target/56809
23831         * config/arm/arm.c (is_jump_table): Use next_active_insn instead of
23832         next_real_insn.
23833         (thumb1_output_casesi): Likewise.
23834         (thumb2_output_casesi): Likewise.
23836 2013-04-03  Richard Biener  <rguenther@suse.de>
23838         PR tree-optimization/56817
23839         * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely):
23840         Split out ...
23841         (tree_unroll_loops_completely_1): ... new function to manually
23842         walk the loop tree, properly defering outer loops of unrolled
23843         loops to later iterations.
23845 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23847         * tree-vect-stmts.c (vectorizable_store): Accept BIT_FIELD_REF.
23848         (vectorizable_load): Likewise.
23849         * tree-vect-slp.c (vect_build_slp_tree): Likewise.
23850         * tree-vect-data-refs.c (vect_create_data_ref_ptr): Handle VECTOR_TYPE.
23852 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23854         * tree-flow-inline.h (get_addr_base_and_unit_offset_1): Handle
23855         BIT_FIELD_REF.
23857 2013-04-03  Ulrich Weigand  <Ulrich.Weigand@de.ibm.com>
23859         * config/spu/spu.c (emit_nop_for_insn): Handle JUMP_TABLE_DATA.
23861 2013-04-03  Bin Cheng  <bin.cheng@arm.com>
23863         * rtl.h (AUTO_INC_DEC): Fix typo of HAVE_POST_MODIFY_DISP.
23865 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23867         PR tree-optimization/56790
23868         * fold-const.c (fold_ternary_loc) <VEC_COND_EXPR>: Add constant
23869         folding.
23871 2013-04-03  Marc Glisse  <marc.glisse@inria.fr>
23873         * simplify-rtx.c (simplify_binary_operation_1) <VEC_SELECT>:
23874         Handle VEC_MERGE.
23875         (simplify_ternary_operation) <VEC_MERGE>: Use unsigned HOST_WIDE_INT
23876         for masks. Test for side effects. Handle nested VEC_MERGE. Handle
23877         equal arguments.
23879 2013-04-03  Jakub Jelinek  <jakub@redhat.com>
23881         PR c/19449
23882         * tree.h (force_folding_builtin_constant_p): New decl.
23883         * builtins.c (force_folding_builtin_constant_p): New variable.
23884         (fold_builtin_constant_p): Fold immediately also if
23885         force_folding_builtin_constant_p.
23887 2013-04-03  Richard Biener  <rguenther@suse.de>
23889         PR tree-optimization/56812
23890         * tree-vect-data-refs.c (vect_slp_analyze_data_ref_dependence):
23891         DRs of the same interleaving chain are independent.
23893 2013-04-02  Jason Merrill  <jason@redhat.com>
23895         * gdbinit.in (pbb): Use debug fn.
23897 2013-04-02  Lawrence Crowl  <crowl@google.com>
23899         * sese.h (struct ivtype_map_elt_s): Remove unused.
23900         (extern debug_ivtype_map): Remove unused.
23901         (extern eq_ivtype_map_elts): Remove unused.
23902         * sese.c (debug_ivtype_map): Removed unused.
23903         (debug_ivtype_map_1): Removed unused.
23904         (debug_ivtype_elt): Remove unused.
23905         (eq_ivtype_map_elts): Remove unused.
23908 2013-04-02  Kai Tietz  <ktietz@redhat.com>
23910         PR target/52790
23911         * config/i386/cygming.h (SUB_TARGET_RECORD_STUB): New sub-target macro.
23912         * config/i386/i386-protos.h (i386_pe_record_stub): Add new prototype.
23913         * config/i386/i386.c (legitimize_pe_coff_extern_decl): New static
23914         function.
23915         (legitimize_pe_coff_symbol): Likewise.
23916         (is_imported_p): New helper-function.
23917         (ix86_option_override_internal): Make MEDIUM_PIC the default code-model
23918         for Windows x64 targets.
23919         (ix86_expand_prologue): Optimize for pe-coff targets.
23920         (ix86_expand_split_stack_prologue): Adjust for pe-coff targets.
23921         (legitimate_pic_address_disp_p): Adjust for x64 pe-coff to support
23922         medium/large code-model.
23923         (legitimize_pic_address): Likewise.
23924         (legitimize_tls_address): Likewise.
23925         (ix86_expand_call): Likewise.
23926         (x86_output_mi_thunk): Likewise.
23927         (get_dllimport_decl): Add new beimport argument.
23928         (construct_plt_address): Don't assert for x64 pe-coff targets.
23929         * config/i386/i386.h (PIC_OFFSET_TABLE_REGNUM): Adjust for x64 pe-coff
23930         targets.
23931         (SYMBOL_FLAG_STUBVAR): New macro.
23932         (SYMBOL_REF_STUBVAR_P): Likewise.
23933         * config/i386/winnt.c (stub_list): New structure.
23934         (stub_head): New local variable.
23935         (i386_pe_record_stub): New function.
23936         (i386_pe_file_end): Emit refptr-stubs.
23938 2013-04-02  Jakub Jelinek  <jakub@redhat.com>
23940         PR rtl-optimization/56745
23941         * ifcvt.c (cond_exec_find_if_block): Don't try to optimize
23942         if then_bb has no successors and else_bb is EXIT_BLOCK_PTR.
23944         PR c++/34949
23945         * tree-ssa-alias.c (stmt_kills_ref_p_1): If base != ref->base
23946         and both of them are MEM_REFs, just compare first argument for
23947         equality and attempt to deal even with differing offsets.
23949         PR c++/34949
23950         * tree-cfg.c (verify_gimple_assign_single): Allow lhs
23951         of gimple_clobber_p to be MEM_REF.
23952         * gimplify.c (gimplify_modify_expr): Gimplify *to_p of
23953         an assignment from TREE_CLOBBER_P.  Allow it to be MEM_REF
23954         after gimplification.
23955         * asan.c (get_mem_ref_of_assignment): Don't instrument
23956         gimple_clobber_p stmts.
23957         * tree-ssa-dse.c (dse_optimize_stmt): Allow DSE of
23958         gimple_clobber_p stmt if they have MEM_REF lhs and
23959         are dead because of another gimple_clobber_p stmt.
23960         * tree-ssa-live.c (clear_unused_block_pointer): Treat
23961         gimple_clobber_p stmts like debug stmts.
23962         (remove_unused_locals): Remove clobbers with MEM_REF lhs
23963         that refer to unused VAR_DECLs or uninitialized values.
23964         * tree-sra.c (sra_ipa_reset_debug_stmts): Also remove
23965         gimple_clobber_p stmts if they refer to removed parameters.
23966         (get_repl_default_def_ssa_name, sra_ipa_modify_expr): Fix up
23967         formatting.
23969 2013-04-02  Uros Bizjak  <ubizjak@gmail.com>
23971         * config/i386/i386.md (*testqi_ext_3): Merge with *testqi_ext_3_rex64
23972         using SWI48 mode attribute.
23974 2013-04-02  Wei Mi  <wmi@google.com>
23976         * config/i386/i386.c (ix86_rtx_costs): Set proper rtx cost for
23977         ashl<mode>3_mask, *<shift_insn><mode>3_mask and
23978         *<rotate_insn><mode>3_mask in i386.md.
23980 2013-04-02  Alexander Ivchenko  <alexander.ivchenko@intel.com>
23982         * config.gcc (arm*-*-linux-*): Remove duplicate t-linux-android.
23984 2013-04-02  Richard Biener  <rguenther@suse.de>
23986         PR tree-optimization/56778
23987         * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
23988         Runtime alias tests are not supported for gather loads.
23989         * tree-vect-loop-manip.c (vect_loop_versioning): Insert
23990         stmts referenced from SSA operands before updating SSA form.
23992 2013-04-02  Ian Caulfield  <ian.caulfield@arm.com>
23993             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
23995         * config/arm/arm-arches.def (armv8-a): Default to cortex-a53.
23996         * config/arm/t-arm (MD_INCLUDES): Depend on cortex-a53.md.
23997         * config/arm/cortex-a53.md: New file.
23998         * config/arm/bpabi.h (BE8_LINK_SPEC): Handle cortex-a53.
23999         * config/arm/arm.md (generic_sched, generic_vfp): Handle cortex-a53.
24000         * config/arm/arm.c (arm_issue_rate): Likewise.
24001         * config/arm/arm-tune.md: Regenerate
24002         * config/arm/arm-tables.opt: Regenerate.
24003         * config/arm/arm-cores.def: Add cortex-a53.
24005 2013-04-02  Zhenqiang Chen  <zhenqiang.chen@arm.com>
24007         * config/arm/uclinux-elf.h: Add %L to LINK_GCC_C_SEQUENCE_SPEC for
24008         non-static link.
24010 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
24012         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add variants for
24013         scalar load/store operations using B/H registers.
24014         (*zero_extend<SHORT:mode><GPI:mode>2_aarch64): Likewise.
24016 2013-04-02  Sofiane Naci  <sofiane.naci@arm.com>
24018         * config/aarch64/aarch64.md (*mov<mode>_aarch64): Add alternatives for
24019         scalar move.
24020         * config/aarch64/aarch64.c
24021         (aarch64_simd_scalar_immediate_valid_for_move): New.
24022         * config/aarch64/aarch64-protos.h
24023         (aarch64_simd_scalar_immediate_valid_for_move): New.
24024         * config/aarch64/constraints.md (Dh, Dq): New.
24025         * config/aarch64/iterators.md (hq): New.
24027 2013-04-02  Eric Botcazou  <ebotcazou@adacore.com>
24029         * reorg.c (get_branch_condition): Deal with conditional returns.
24030         (fill_simple_delay_slots): Remove dead code dealing with jumps.
24032 2013-04-01  Wei Mi  <wmi@google.com>
24034         * config/i386/i386.md (*ashl<mode>3_mask): Rewrite as define_insn.
24035         Truncate operand 2 using %b asm operand modifier.
24036         (*<shift_insn><mode>3_mask): Ditto.
24037         (*<rotate_insn><mode>3_mask): Ditto.
24039 2013-04-01  Steven Bosscher  <steven@gcc.gnu.org>
24041         PR middle-end/56798
24042         * cfgbuild.c (inside_basic_block_p): Restore check broken at r197234.
24044 2013-03-31  Kaz Kojima  <kkojima@gcc.gnu.org>
24046         * config/sh/sh.md (casesi_worker_1): Use next_active_insn instead
24047         of next_real_insn.
24048         (casesi_worker_2, casesi_shift_media, casesi_load_media): Likewise.
24050 2013-03-30  Lawrence Crowl  <crowl@google.com>
24052         * dse.c (clear_alias_sets): Remove never set.
24053         (disqualified_clear_alias_sets): Remove never set.
24054         (clear_alias_mode_pool): Remove never set.
24055         (dse_step0): Remove condition that is never true.
24056         (canon_address): Remove condition that is never true.
24057         (dse_step7): Remove condition that is never true.
24058         (rest_of_handle_dse): Remove condition that is never true.
24059         (rest_of_handle_dse::did_global): Remove never read from above.
24060         (dse_step2_spill): Remove never called from above.
24061         (dse_step5_spill): Remove never called from above.
24063 2013-03-30  Steven Bosscher  <steven@gcc.gnu.org>
24065         * doc/md.texi (Standard Names) <casesi>: Update documentation for
24066         JUMP_TABLE_DATA changes.
24067         * doc/tm.texi.in (Dispatch Tables) <ASM_OUTPUT_CASE_LABEL>: Likewise.
24068         * doc/rtl.texi (Flags) <INSN_DELETED_P, SCHED_GROUP_P>: Likewise.
24069         (Insns) <jump_table_data>: New entry.
24070         * doc/tm.texi: Regenerate.
24072         * cfgrtl.c (fixup_reorder_chain): Do not emit barriers to BB_FOOTER.
24074         * postreload-gcse.c (bb_has_well_behaved_predecessors): Correct test
24075         for table jump at the end of a basic block using tablejump_p.
24076         * targhooks.c (default_invalid_within_doloop): Likewise.
24077         * config/rs6000/rs6000.c (TARGET_INVALID_WITHIN_DOLOOP): Remove
24078         target hook implementation that is identical to the default hook.
24079         (rs6000_invalid_within_doloop): Remove.
24081         * bb-reorder.c (fix_crossing_unconditional_branches): Remove set but
24082         unused variable from tablejump_p call.
24084         * rtl.def (JUMP_TABLE_DATA): New RTX_INSN object.
24085         * rtl.h (RTX_PREV, RTX_NEXT): Adjust for new JUMP_TABLE_DATA.
24086         (INSN_DELETED_P): Likewise.
24087         (emit_jump_table_data): New prototype.
24088         * gengtype.c (adjust_field_rtx_def): Handle JUMP_TABLE_DATA fields
24089         after 4th as unused.
24090         * print-rtl.c (print_rtl): Handle JUMP_TABLE_DATA.
24091         * sched-vis.c (print_insn): Likewise.
24092         * emit-rtl.c (active_insn_p): Consider JUMP_TABLE_DATA an active
24093         insn for compatibility with back ends that use next_active_insn to
24094         identify jump table data.
24095         (set_insn_deleted): Remove no longer useful JUMP_TABLE_DATA_P check.
24096         (remove_insn): Likewise.
24097         (emit_insn): Do not accept JUMP_TABLE_DATA objects in insn chains
24098         to be emitted.
24099         (emit_debug_insn, emit_jump_insn, emit_call_insn, emit_label): Idem.
24100         (emit_jump_table_data): New function.
24102         * cfgbuild.c (inside_basic_block_p): A JUMP_INSN is always inside a
24103         basic block, a JUMP_TABLE_DATA never is.
24104         (control_flow_insn_p): JUMP_TABLE_DATA is not a control flow insn.
24105         * cfgrtl.c (duplicate_insn_chain): Split handling of JUMP_TABLE_DATA
24106         off from code handling real insns.
24107         * final.c (get_attr_length_1): Simplify for JUMP_INSNs.
24108         * function.c (instantiate_virtual_regs): Remove JUMP_TABLE_DATA_P
24109         test, now redundant because JUMP_TABLE_DATA is not an INSN_P insn.
24110         * gcse.c (insert_insn_end_basic_block): Likewise, JUMP_TABLE_DATA_P
24111         is not a NONDEBUG_INSN_P.
24112         * ira-costs.c (scan_one_insn): Likewise.
24113         * jump.c (mark_all_labels): Likewise.
24114         (mark_jump_label_1): Likewise.
24115         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24116         * lra.c (get_insn_freq): Expect all insns reaching here to be in
24117         a basic block.
24118         (check_rtl): Remove JUMP_TABLE_DATA_P test, not a NONDEBUG_INSN_P insn.
24119         * predict.c (expensive_function_p): Use FOR_BB_INSNS.
24120         * reload1.c (calculate_needs_all_insns): Call set_label_offsets for
24121         JUMP_TABLE_DATA_P insns.
24122         (calculate_elim_costs_all_insns): Likewise.
24123         (set_label_offsets): Recurse on the PATTERN of JUMP_TABLE_DATA insns.
24124         (elimination_costs_in_insn): Remove redundant JUMP_TABLE_DATA_P test.
24125         (delete_output_reload): Code style fixups.
24126         * reorg.c (dbr_schedule): Move JUMP_TABLE_DATA_P up to avoid setting
24127         insn flags on this non-insn.
24128         * sched-rgn.c (add_branch_dependences): Treat JUMP_TABLE_DATA insns
24129         as scheduling barriers, for pre-change compatibility.
24130         * stmt.c (emit_case_dispatch_table): Emit jump table data not as
24131         JUMP_INSN objects but instead as JUMP_TABLE_DATA objects.
24133         * config/alpha/alpha.c (alpha_does_function_need_gp): Remove
24134         redundant JUMP_TABLE_DATA_P test.
24135         * config/arm/arm.c (thumb_far_jump_used_p): Likewise.
24136         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24137         (frv_for_each_packet): Likewise.
24138         * config/i386/i386.c (min_insn_size): Likewise.
24139         (ix86_avoid_jump_mispredicts): Likewise.
24140         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24141         * config/mep/mep.c (mep_reorg_erepeat): Likewise.
24142         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24143         (mips16_insn_length): Robustify.
24144         (mips_has_long_branch_p): Remove redundant JUMP_TABLE_DATA_P test.
24145         (mips16_split_long_branches): Likewise.
24146         * config/pa/pa.c (pa_combine_instructions): Likewise.
24147         * config/rs6000/rs6000.c (get_next_active_insn): Treat
24148         JUMP_TABLE_DATA objects as active insns, like in active_insn_p.
24149         * config/s390/s390.c (s390_chunkify_start): Treat JUMP_TABLE_DATA
24150         as contributing to pool range lengths.
24151         * config/sh/sh.c (find_barrier): Restore check for ADDR_DIFF_VEC.
24152         Remove redundant JUMP_TABLE_DATA_P test.
24153         (sh_loop_align): Likewise.
24154         (split_branches): Likewise.
24155         (sh_insn_length_adjustment): Likewise.
24156         * config/spu/spu.c (get_branch_target): Likewise.
24158 2013-03-29  Jan Hubicka  <jh@suse.cz>
24160         * lto-cgraph.c (output_profile_summary, input_profile_summary): Use
24161         gcov streaming; stream hot bb threshold to ltrans.
24162         * predict.c (get_hot_bb_threshold): Break out from ....
24163         (maybe_hot_count_p): ... here.
24164         (set_hot_bb_threshold): New function.
24165         * lto-section-in.c (lto_section_name): Add profile.
24166         * profile.h (get_hot_bb_threshold, set_hot_bb_threshold): Declare.
24167         * ipa.c: Include hash-table.h, tree-inline.h, profile.h, lto-streamer.h
24168         and data-streamer.h
24169         (histogram_entry): New structure.
24170         (histogram, histogram_pool): New global vars.
24171         (histogram_hash): New structure.
24172         (histogram_hash::hash): New method.
24173         (histogram_hash::equal): Likewise.
24174         (account_time_size): New function.
24175         (cmp_counts): New function.
24176         (dump_histogram): New function.
24177         (ipa_profile_generate_summary): New function.
24178         (ipa_profile_write_summary): New function.
24179         (ipa_profile_read_summary): New function.
24180         (ipa_profile): Decide on threshold.
24181         (pass_ipa_profile): Add ipa_profile_write_summary and
24182         ipa_profile_read_summary.
24183         * Makefile.in (ipa.o): Update dependencies.
24184         * lto-streamer.h (LTO_section_ipa_profile): New section.
24186 2013-03-29  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24188         * tree.h (VAR_P): New.
24190 2013-03-29  Paolo Carlini  <paolo.carlini@oracle.com>
24192         PR lto/56777
24193         * doc/invoke.texi ([-fwhole-program]): Fix typo.
24195 2013-03-29  Steven Bosscher  <steven@gcc.gnu.org>
24197         * cfgbuild.c (inside_basic_block_p): Use JUMP_TABLE_DATA_P in lieu
24198         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24199         (control_flow_insn_p): Likewise.
24200         * cfgrtl.c (duplicate_insn_chain): Likewise.
24201         * final.c (get_attr_length_1): Likewise.
24202         (shorten_branches): Likewise.
24203         (final_scan_insn): Likewise.
24204         * function.c (instantiate_virtual_regs): Likewise.
24205         * gcse.c (insert_insn_end_basic_block): Likewise.
24206         * ira-costs.c (scan_one_insn): Likewise.
24207         * lra-eliminations.c (eliminate_regs_in_insn): Likewise.
24208         * lra.c (check_rtl): Likewise.
24209         * reload1.c (elimination_costs_in_insn): Likewise.
24210         * reorg.c (follow_jumps): Likewise.
24212         * config/arm/arm.c (is_jump_table): Use JUMP_TABLE_DATA_P in lieu
24213         of tests for JUMP_P and a ADDR_DIFF_VEC or ADDR_VEC pattern.
24214         (thumb_far_jump_used_p): Likewise.
24215         * config/bfin/bfin.c (workaround_rts_anomaly): Likewise.
24216         (workaround_speculation): Likewise.
24217         (add_sched_insns_for_speculation): Likewise.
24218         * config/c6x/c6x.c (reorg_emit_nops): Likewise.
24219         * config/frv/frv.c (frv_function_contains_far_jump): Likewise.
24220         (frv_for_each_packet): Likewise.
24221         * config/i386/i386.c (ix86_avoid_jump_mispredicts): Likewise.
24222         * config/ia64/ia64.c (emit_all_insn_group_barriers): Likewise.
24223         (final_emit_insn_group_barriers): Likewise.
24224         * config/m32r/m32r.c (m32r_is_insn): Likewise.
24225         * config/mips/mips.c (USEFUL_INSN_P): Likewise.
24226         (mips16_insn_length): Likewise.
24227         * config/pa/pa.c (pa_reorg): Likewise.
24228         (pa_combine_instructions): Likewise.
24229         * config/rs6000/rs6000.c (rs6000_invalid_within_doloop): Likewise.
24230         * config/sh/sh.c (fixup_addr_diff_vecs): Likewise.
24231         (sh_reorg): Likewise.
24232         (split_branches): Likewise.
24233         * config/spu/spu.c (get_branch_target): Likewise.
24235         * config/s390/s390.c (s390_chunkify_start): Simplify logic using
24236         JUMP_TABLE_DATA_P.
24238 2013-03-29  Kirill Yukhin  <kirill.yukhin@intel.com>
24240         * config/i386/avx2intrin.h (_mm256_broadcastsi128_si256):
24241         Fix declaration name.
24243 2013-03-28  Lawrence Crowl  <crowl@google.com>
24245         * graphds.h (struct graph.indicies): Remove unused.
24246         * graphite-poly.h (struct graph.original_pddrs): Remove unused.
24247         (SCOP_ORIGINAL_PDDRS): Remove unused.
24248         * sese.h (extern insert_loop_close_phis): Removed unused.
24249         (extern insert_guard_phis): Removed unused.
24250         (extern ivtype_map_elt_info): Removed unused.
24251         (new_ivtype_map_elt): Removed unused.
24252         * sese.c (ivtype_map_elt_info): Removed unused.
24254 2013-03-28  Lawrence Crowl  <crowl@google.com>
24256         * Makefile.in: Add several missing include dependences.
24257         (DUMPFILE_H): New.
24258         (test-dump.o): New.  This object is not added to any executable,
24259         but is present for ad-hoc testing.
24260         * bitmap.c
24261         (debug (const bitmap_head_def &)): New.
24262         (debug (const bitmap_head_def *)): New.
24263         * bitmap.h
24264         (extern debug (const bitmap_head_def &)): New.
24265         (extern debug (const bitmap_head_def *)): New.
24266         * cfg.c
24267         (debug (edge_def &)): New.
24268         (debug (edge_def *)): New.
24269         * cfghooks.c
24270         (debug (basic_block_def &)): New.
24271         (debug (basic_block_def *)): New.
24272         * dumpfile.h
24273         (dump_node (const_tree, int, FILE *)): Correct source file.
24274         * dwarf2out.c
24275         (debug (die_struct &)): New.
24276         (debug (die_struct *)): New.
24277         * dwarf2out.h
24278         (extern debug (die_struct &)): New.
24279         (extern debug (die_struct *)): New.
24280         * gimple-pretty-print.c
24281         (debug (gimple_statement_d &)): New.
24282         (debug (gimple_statement_d *)): New.
24283         * gimple-pretty-print.h
24284         (extern debug (gimple_statement_d &)): New.
24285         (extern debug (gimple_statement_d *)): New.
24286         * ira-build.c
24287         (debug (ira_allocno_copy &)): New.
24288         (debug (ira_allocno_copy *)): New.
24289         (debug (ira_allocno &)): New.
24290         (debug (ira_allocno *)): New.
24291         * ira-int.h
24292         (extern debug (ira_allocno_copy &)): New.
24293         (extern debug (ira_allocno_copy *)): New.
24294         (extern debug (ira_allocno &)): New.
24295         (extern debug (ira_allocno *)): New.
24296         * ira-lives.c
24297         (debug (live_range &)): New.
24298         (debug (live_range *)): New.
24299         * lra-int.h
24300         (debug (lra_live_range &)): New.
24301         (debug (lra_live_range *)): New.
24302         * lra-lives.c
24303         (debug (lra_live_range &)): New.
24304         (debug (lra_live_range *)): New.
24305         * omega.c
24306         (debug (omega_pb_d &)): New.
24307         (debug (omega_pb_d *)): New.
24308         * omega.h
24309         (extern debug (omega_pb_d &)): New.
24310         (extern debug (omega_pb_d *)): New.
24311         * print-rtl.c
24312         (debug (const rtx_def &)): New.
24313         (debug (const rtx_def *)): New.
24314         * print-tree.c
24315         (debug_tree (tree): Move within file.
24316         (debug_raw (const tree_node &)): New.
24317         (debug_raw (const tree_node *)): New.
24318         (dump_tree_via_hooks (const tree_node *, int)): New.
24319         (debug (const tree_node &)): New.
24320         (debug (const tree_node *)): New.
24321         (debug_verbose (const tree_node &)): New.
24322         (debug_verbose (const tree_node *)): New.
24323         (debug_head (const tree_node &)): New.
24324         (debug_head (const tree_node *)): New.
24325         (debug_body (const tree_node &)): New.
24326         (debug_body (const tree_node *)): New.
24327         (debug_vec_tree (tree): Move and reimplement in terms of dump.
24328         (debug (vec<tree, va_gc> &)): New.
24329         (debug (vec<tree, va_gc> *)): New.
24330         * rtl.h
24331         (extern debug (const rtx_def &)): New.
24332         (extern debug (const rtx_def *)): New.
24333         * sbitmap.c
24334         (debug_raw (simple_bitmap_def &)): New.
24335         (debug_raw (simple_bitmap_def *)): New.
24336         (debug (simple_bitmap_def &)): New.
24337         (debug (simple_bitmap_def *)): New.
24338         * sbitmap.h
24339         (extern debug (simple_bitmap_def &)): New.
24340         (extern debug (simple_bitmap_def *)): New.
24341         (extern debug_raw (simple_bitmap_def &)): New.
24342         (extern debug_raw (simple_bitmap_def *)): New.
24343         * sel-sched-dump.c
24344         (debug (vinsn_def &)): New.
24345         (debug (vinsn_def *)): New.
24346         (debug_verbose (vinsn_def &)): New.
24347         (debug_verbose (vinsn_def *)): New.
24348         (debug (expr_def &)): New.
24349         (debug (expr_def *)): New.
24350         (debug_verbose (expr_def &)): New.
24351         (debug_verbose (expr_def *)): New.
24352         (debug (vec<rtx> &)): New.
24353         (debug (vec<rtx> *)): New.
24354         * sel-sched-dump.h
24355         (extern debug (vinsn_def &)): New.
24356         (extern debug (vinsn_def *)): New.
24357         (extern debug_verbose (vinsn_def &)): New.
24358         (extern debug_verbose (vinsn_def *)): New.
24359         (extern debug (expr_def &)): New.
24360         (extern debug (expr_def *)): New.
24361         (extern debug_verbose (expr_def &)): New.
24362         (extern debug_verbose (expr_def *)): New.
24363         (extern debug (vec<rtx> &)): New.
24364         (extern debug (vec<rtx> *)): New.
24365         * sel-sched-ir.h
24366         (_list_iter_cond_expr): Make inline instead of static.
24367         * sreal.c
24368         (debug (sreal &)): New.
24369         (debug (sreal *)): New.
24370         * sreal.h
24371         (extern debug (sreal &)): New.
24372         (extern debug (sreal *)): New.
24373         * tree.h
24374         (extern debug_raw (const tree_node &)): New.
24375         (extern debug_raw (const tree_node *)): New.
24376         (extern debug (const tree_node &)): New.
24377         (extern debug (const tree_node *)): New.
24378         (extern debug_verbose (const tree_node &)): New.
24379         (extern debug_verbose (const tree_node *)): New.
24380         (extern debug_head (const tree_node &)): New.
24381         (extern debug_head (const tree_node *)): New.
24382         (extern debug_body (const tree_node &)): New.
24383         (extern debug_body (const tree_node *)): New.
24384         (extern debug (vec<tree, va_gc> &)): New.
24385         (extern debug (vec<tree, va_gc> *)): New.
24386         * tree-cfg.c
24387         (debug (struct loop &)): New.
24388         (debug (struct loop *)): New.
24389         (debug_verbose (struct loop &)): New.
24390         (debug_verbose (struct loop *)): New.
24391         * tree-dump.c: Add header dependence.
24392         * tree-flow.h
24393         (extern debug (struct loop &)): New.
24394         (extern debug (struct loop *)): New.
24395         (extern debug_verbose (struct loop &)): New.
24396         (extern debug_verbose (struct loop *)): New.
24397         * tree-data-ref.c
24398         (debug (data_reference &)): New.
24399         (debug (data_reference *)): New.
24400         (debug (vec<data_reference_p> &)): New.
24401         (debug (vec<data_reference_p> *)): New.
24402         (debug (vec<ddr_p> &)): New.
24403         (debug (vec<ddr_p> *)): New.
24404         * tree-data-ref.h
24405         (extern debug (data_reference &)): New.
24406         (extern debug (data_reference *)): New.
24407         (extern debug (vec<data_reference_p> &)): New.
24408         (extern debug (vec<data_reference_p> *)): New.
24409         (extern debug (vec<ddr_p> &)): New.
24410         (extern debug (vec<ddr_p> *)): New.
24411         * tree-ssa-alias.c
24412         (debug (pt_solution &)): New.
24413         (debug (pt_solution *)): New.
24414         * tree-ssa-alias.h
24415         (extern debug (pt_solution &)): New.
24416         (extern debug (pt_solution *)): New.
24417         * tree-ssa-alias.c
24418         (debug (_var_map &)): New.
24419         (debug (_var_map *)): New.
24420         (debug (tree_live_info_d &)): New.
24421         (debug (tree_live_info_d *)): New.
24422         * tree-ssa-alias.h
24423         (extern debug (_var_map &)): New.
24424         (extern debug (_var_map *)): New.
24425         (extern debug (tree_live_info_d &)): New.
24426         (extern debug (tree_live_info_d *)): New.
24428 2013-03-28  Jan Hubicka  <jh@suse.cz>
24430         * lto-cgraph.c (merge_profile_summaries): Fix overflows.
24432 2013-03-28  Ian Bolton  <ian.bolton@arm.com>
24434         * config/aarch64/aarch64.md (aarch64_can_eliminate): Keep frame
24435         record only when desired or required.
24437 2013-03-28  Uros Bizjak  <ubizjak@gmail.com>
24439         * config/i386/i386.md (*vec_extract2vdi_1): Merge with
24440         *vec_extractv2di_1_rex64.  Use x64 isa attribute.
24442 2013-03-28  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
24444         * config/aarch64/aarch64.md (*and<mode>3_compare0): New pattern.
24445         (*andsi3_compare0_uxtw): New pattern.
24446         (*and_<SHIFT:optab><mode>3_compare0): New pattern.
24447         (*and_<SHIFT:optab>si3_compare0_uxtw): New pattern.
24449 2013-03-28  Jan Hubicka  <jh@suse.cz>
24451         * data-streamer-in.c (streamer_read_gcov_count): New function.
24452         * gimple-streamer-out.c: Include value-prof.h.
24453         (output_gimple_stmt): Output histogram.
24454         (output_bb): Use streamer_write_gcov_count.
24455         * value-prof.c: Include data-streamer.h
24456         (dump_histogram_value): Add HIST_TYPE_MAX.
24457         (stream_out_histogram_value): New function.
24458         (stream_in_histogram_value): New function.
24459         * value-prof.h (enum hist_type): Add HIST_TYPE_MAX.
24460         (stream_out_histogram_value, stream_in_histogram_value): Declare.
24461         * data-streamer-out.c (streamer_write_gcov_count): New function.
24462         (streamer_write_gcov_count_stream): New function.
24463         * lto-cgraph.c (lto_output_edge): Update counter streaming.
24464         (lto_output_node): Likewise.
24465         (input_node, input_edge): Likewise.
24466         * lto-streamer-out.c (output_cfg): Update streaming.
24467         * lto-streamer-in.c (input_cfg): Likewise.
24468         * data-streamer.h (streamer_write_gcov_count,
24469         streamer_write_gcov_count_stream, streamer_read_gcov_count): Declare.
24470         * gimple-streamer-in.c: Include value-prof.h
24471         (input_gimple_stmt): Input histograms.
24472         (input_bb): Update profile streaming.
24474 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24476         * genmodes.c (emit_max_int): New function.
24477         (emit_insn_modes_h): Added call to emit_max_function.
24478         * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_INT, MAX_BITSIZE_MODE_ANY_MODE):
24479         Added doc.
24480         * machmode.def: Fixed comment.
24482 2013-03-28  Kenneth Zadeck  <zadeck@naturalbridge.com>
24484         * combine.c (try_combine): Removed useless assert.
24485         * cselib.c (rtx_equal_for_cselib_1): Removed unnecessary parens.
24487 2013-03-28  Marek Polacek  <polacek@redhat.com>
24488             Richard Biener  <rguenther@suse.de>
24490         PR tree-optimization/56695
24491         * tree-vect-stmts.c (vectorizable_condition): Unconditionally
24492         build signed result of a vector comparison.
24493         * tree-cfg.c (verify_gimple_comparison): Check that a result
24494         of a vector comparison has signed type.
24496 2013-03-28  Richard Biener  <rguenther@suse.de>
24498         PR tree-optimization/37021
24499         * tree-vect-slp.c (vect_build_slp_tree): When not unrolling
24500         do not restrict gaps between groups.
24501         * tree-vect-stmts.c (vectorizable_load): Properly account for
24502         a gap between groups.
24504 2013-03-28  Eric Botcazou  <ebotcazou@adacore.com>
24506         * toplev.c (process_options): Do not disable -fomit-frame-pointer on a
24507         general basis if unwind info is requested and ACCUMULATE_OUTGOING_ARGS
24508         is not enabled.
24510 2013-03-27  Gerald Pfeifer  <gerald@pfeifer.com>
24512         * doc/invoke.texi (AVR Options): Tweak link for AVR-LibC user manual.
24513         * doc/extend.texi (Named Address Spaces): Ditto.
24514         (Variable Attributes): Ditto.
24516 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24518         * config.build: Add support for cygwin x64 target.
24519         * config.gcc: Likewise.
24520         * config.host: Likewise.
24521         * configure.ac: Likewise
24522         * configure: Regenerated.
24524 2013-03-27  Kai Tietz  <ktietz@redhat.com>
24526         * config/i386/cygwin-stdint.h: Add support for cygwin x64 target.
24527         * config/i386/t-cygwin-w64: New file.
24528         * config/i386/cygwin-w64.h: New file.
24529         * config/i386/cygwin.h (EXTRA_OS_CPP_BUILTINS): Extend
24530         and add support for x64-cygwin target.
24531         (CPP_SPEC): Likewise.
24532         (CXX_WRAP_SPEC_LIST): Undefine before define.
24533         (LIBGCJ_SONAME): Use 15 as version.
24535 2013-03-27  Richard Biener  <rguenther@suse.de>
24537         PR tree-optimization/56716
24538         * tree-ssa-structalias.c (perform_var_substitution): Adjust
24539         dumping for ref nodes.
24541 2013-03-27  Martin Jambor  <mjambor@suse.cz>
24543         PR tree-optimization/55334
24544         * ipa-cp.c (initialize_node_lattices): Allow IPA-CP through and to
24545         restricted pointers to arrays.
24547 2013-03-27  Gabriel Dos Reis  <gdr@integrable-solutions.net>
24549         * Makefile.in (.SUFFIXES): Add .cc.
24550         (.c.o): Apply same recipe for implicit rule .cc.o.
24552 2013-03-27  Richard Biener  <rguenther@suse.de>
24554         PR tree-optimization/37021
24555         * tree-vect-data-refs.c (vect_check_strided_load): Allow
24556         REALPART/IMAGPART_EXPRs around the supported refs.
24557         * tree-ssa-structalias.c (find_func_aliases): Assume that
24558         floating-point values are not used to transfer pointers.
24560 2013-03-27  Alexander Ivchenko  <alexander.ivchenko@intel.com>
24562         * target.def (TARGET_HAS_IFUNC_P): New target hook.
24563         * doc/tm.texi.in (TARGET_HAS_IFUNC_P): New.
24564         * doc/tm.texi: Regenerate.
24565         * targhooks.h (default_has_ifunc_p): New.
24566         * targhooks.c (default_has_ifunc_p): Ditto.
24567         * config/linux-protos.h: New file.
24568         * config/linux-android.h (TARGET_HAS_IFUNC_P): Using version of this
24569         hook for linux which disables support of indirect functions in android.
24570         * config/linux-android.c: New file.
24571         * config/t-linux-android.c: Ditto.
24572         * config.gcc: Added new object file linux-android.o.
24573         * config/i386/i386.c (ix86_get_function_versions_dispatcher):
24574         Using TARGET_HAS_IFUNC hook instead of HAVE_GNU_INDIRECT_FUNCTION.
24575         * varasm.c (do_assemble_alias): Likewise.
24576         * configure.ac: Define HAVE_GNU_INDIRECT_FUNCTION as zero if the target
24577         doesn't support indirect functions.
24578         * configure: Regenerate.
24580 2013-03-27  Bin Cheng  <bin.cheng@arm.com>
24582         PR target/56102
24583         * config/arm/arm.c (thumb1_rtx_costs, thumb1_size_rtx_costs): Fix
24584         rtx costs for SET/ASHIFT/ASHIFTRT/LSHIFTRT/ROTATERT patterns with
24585         mult-word mode.
24587 2013-03-27  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
24589         * config/s390/s390.h (TARGET_FLT_EVAL_METHOD): Define.
24591 2013-03-27  Terry Guo  <terry.guo@arm.com>
24593         * config/arm/arm-cores.def: Added core cortex-r7.
24594         * config/arm/arm-tune.md: Regenerated.
24595         * config/arm/arm-tables.opt: Regenerated.
24596         * doc/invoke.texi: Added entry for core cortex-r7.
24598 2013-03-27  Walter Lee  <walt@tilera.com>
24600         * config/tilegx/tilegx.c (tilegx_expand_prologue): Avoid
24601         double-decrement of next_scratch_regno.
24603 2013-03-27  Walter Lee  <walt@tilera.com>
24605         * config/tilegx/tilegx.md (insn_v1mulu): Fix predicates on
24606         input operands.
24607         (insn_v1mulus): Ditto.
24608         (insn_v2muls): Ditto.
24610 2013-03-27  Walter Lee  <walt@tilera.com>
24612         * config/tilegx/tilegx.h (ASM_OUTPUT_ADDR_VEC_ELT): Delete extra tab.
24613         (ASM_OUTPUT_ADDR_DIFF_ELT): Ditto.
24615 2013-03-27  Walter Lee  <walt@tilera.com>
24617         * config/tilegx/tilegx.md (*sibcall_insn): Fix type atribute for jr.
24618         (*sibcall_value): Ditto.
24620 2013-03-27  Walter Lee  <walt@tilera.com>
24622         * config/tilegx/tilegx.md (insn_mnz_<mode>): Replaced by ...
24623         (insn_mnz_v8qi): ... this ...
24624         (insn_mnz_v4hi): ... and this.  Replace (const_int 0) with the
24625         vector equivalent.
24626         (insn_v<n>mnz): Replaced by ...
24627         (insn_v1mnz): ... this ...
24628         (insn_v2mnz): ... and this.  Replace (const_int 0) with the vector
24629         equivalent.
24630         (insn_mz_<mode>): Replaced by ...
24631         (insn_mz_v8qi): ... this ...
24632         (insn_mz_v4hi): ... and this.  Replace (const_int 0) with the
24633         vector equivalent.
24634         (insn_v<n>mz): Replaced by ...
24635         (insn_v1mz): ... this ...
24636         (insn_v2mz): ... and this.  Replace (const_int 0) with the vector
24637         equivalent.
24639 2013-03-26  Eric Botcazou  <ebotcazou@adacore.com>
24641         * doc/invoke.texi (SPARC options): Remove -mlittle-endian.
24643 2013-03-26  Roland McGrath  <mcgrathr@google.com>
24645         * config/arm/arm.c (arm_print_operand: case 'w'): Use fputs rather
24646         than fprintf with a non-constant, non-format string.
24648 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24650         * config/i386/i386.md (*cmpqi_ext_1): Merge with *cmpqi_ext_1_rex64
24651         using nox64 isa attribute.  Use nonimmediate_x86nomem_operand as
24652         operand 0 predicate.
24653         (*cmpqi_ext_3): Merge with *cmpqi_ext_3_rex64 using nox64 isa
24654         attribute.  Use general_x64nomem_operand as operand 1 predicate.
24655         (*movqi_extv_1): Merge with *movqi_extv_1_rex64 using nox64 isa
24656         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24657         (*movqi_extzv_2): Merge with *movqi_extzv_2_rex64 using nox64 isa
24658         attribute.  Use nonimmediate_x64nomem_operand as operand 0 predicate.
24659         (mov<mode>_insv_1): Remove expander.  Merge insn with
24660         movsi_insv_1 using SWI48 mode iterator and nox64 isa attribute.
24661         Use general_x64nomem_operand as operand 1 predicate.
24662         (addqi_ext_1): Merge with *addqi_ext_1_rex64 using nox64 isa attribute.
24663         (*testqi_ext_1): Merge with *testqi_ext_1_rex64 using nox64 isa
24664         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24665         (*andqi_ext_1): Merge with *andqi_ext_1_rex64 using nox64 isa
24666         attribute.  Use nonimmediate_x64nomem_operand as operand 2 predicate.
24667         (*<code>qi_ext_1): Merge with *<code>qi_ext_1_rex64 using nox64 isa
24668         attribute.  Use nonimmediate_x64nomem_operand as operand 1 predicate.
24669         (*xorqi_cc_ext_1): Merge with *xorqi_cc_ext_1_rex64 using nox64
24670         isa attribute.  Use general_x64nomem_operand as operand 2 predicate.
24671         * config/i386/predicates.md (nonimmediate_x64nomem_operand): New.
24672         (general_x64nomem_operand): Ditto.
24674 2013-03-26  Sebastian Huber  <sebastian.huber@embedded-brains.de>
24676         * config/rtems.opt: Add -pthread option.
24678 2013-03-26  Richard Biener  <rguenther@suse.de>
24680         * alias.c (find_base_term): Avoid redundant and not used recursion.
24681         (base_alias_check): Get the initial base term from the caller.
24682         (true_dependence_1): Compute and pass base terms to base_alias_check.
24683         (write_dependence_p): Likewise.
24684         (may_alias_p): Likewise.
24686 2013-03-26  Sofiane Naci  <sofiane.naci@arm.com>
24688         * config/aarch64/aarch64.c (aarch64_classify_address): Support
24689         PC-relative load in SI modes and above only.
24691 2013-03-26  Xinyu Qi  <xyqi@marvell.com>
24693         * config/arm/arm.h (FIRST_IWMMXT_GR_REGNUM): Add comment.
24694         * config/arm/iwmmxt.md (WCGR0): Update.
24695         (WCGR1, WCGR2, WCGR3): Likewise.
24697 2013-03-26  Uros Bizjak  <ubizjak@gmail.com>
24699         * config/i386/i386.md (*movdfcc_1): Merge with *movdfcc_1_rex64.
24700         Use x64 and nox64 isa attributes.
24702 2013-03-26  Richard Biener  <rguenther@suse.de>
24704         * emit-rtl.c (set_mem_attributes_minus_bitpos): Remove
24705         alignment computations and rely on get_object_alignment_1
24706         for the !TYPE_P case.
24707         Commonize DECL/COMPONENT_REF handling in the ARRAY_REF path.
24709 2013-03-26  Walter Lee  <walt@tilera.com>
24711         * config/tilegx/tilegx.h (PROFILE_BEFORE_PROLOGUE): Define.
24712         * config/tilegx/tilepro.h (PROFILE_BEFORE_PROLOGUE): Define.
24714 2013-03-25  Jeff Law  <law@redhat.com>
24716         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Add missing
24717         check for INTEGRAL_TYPE_P that was missing due to checking in
24718         wrong version of prior patch.
24720 2013-03-25  Walter Lee  <walt@tilera.com>
24722         * config/tilegx/tilegx-builtins.h (enum tilegx_builtin): Add
24723         TILEGX_INSN_SHUFFLEBYTES1.
24724         * config/tilegx/tilegx.c (tilegx_builtin_info): Add entry for
24725         shufflebytes1.
24726         (tilegx_builtins): Ditto.
24727         * config/tilegx/tilegx.md (insn_shufflebytes1): New pattern.
24729 2013-03-25  Walter Lee  <walt@tilera.com>
24731         * config/tilegx/tilegx.md (floatsisf2): New pattern.
24732         (floatunssisf2): New pattern.
24733         (floatsidf2): New pattern.
24734         (floatunssidf2): New pattern.
24736 2013-03-25  Walter Lee  <walt@tilera.com>
24738         * config/tilegx/tilegx.c (expand_set_cint64_one_inst): Inline
24739         tests for constraint J, K, N, P.
24741 2013-03-25  Walter Lee  <walt@tilera.com>
24743         * config/tilegx/tilegx.c (tilegx_asm_preferred_eh_data_format):
24744         Use indirect/pcrel encoding.
24745         * config/tilepro/tilepro.c (tilepro_asm_preferred_eh_data_format):
24746         Ditto.
24748 2013-03-25  Steve Ellcey  <sellcey@mips.com>
24750         * config/mips/mmips-cpus.def (74kc, 74kf2_1, 74kf, 74kf, 74kf1_1,
24751         74kfx, 74kx, 74kf3_2): Add PTF_AVOID_IMADD.
24752         * config/mips/mips.c (mips_option_override): Set IMADD default.
24753         * config/mips/mips.h (PTF_AVOID_IMADD): New.
24754         (ISA_HAS_MADD_MSUB): Remove MIPS16 check.
24755         (GENERATE_MADD_MSUB): Remove TUNE_74K check, add MIPS16 check.
24756         * config/mips/mips.md (mimadd): New flag for integer madd/msub.
24757         * doc/invoke.texi (-mimadd/-mno-imadd): New.
24759 2013-03-25  Jeff Law  <law@redhat.com>
24761         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Rework
24762         slightly to avoid creating and folding useless trees.  Simplify
24763         slightly by restricting to INTEGER_CSTs and using int_fits_type_p.
24765 2013-03-25  Uros Bizjak  <ubizjak@gmail.com>
24767         * config/i386/i386.md (*zero_extendsidi2): Merge with
24768         *zero_extendsidi2_rex64.  Use x64 and nox64 isa attributes.
24769         * config/i386/predicates.md (x86_64_zext_operand): Rename from
24770         x86_64_zext_general_operand.  Use nonimmediate_operand on 32bit
24771         targets.  Clarify comment.
24773 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24775         * ipa-prop.c (ipa_write_jump_function): Stream simple and aritmetic
24776         pass-through jump functions differently.
24777         (ipa_read_jump_function): Likewise.  Also use setter functions to set
24778         up jump functions.
24780 2013-03-25  Martin Jambor  <mjambor@suse.cz>
24782         * ipa-cp.c (ipa_get_indirect_edge_target): Renamed to
24783         ipa_get_indirect_edge_target_1, added parameter agg_reps and ability to
24784         process it.
24785         (ipa_get_indirect_edge_target): New function.
24786         (devirtualization_time_bonus): New parameter known_aggs, pass it to
24787         ipa_get_indirect_edge_target.  Update all callers.
24788         (ipcp_discover_new_direct_edges): New parameter aggvals.  Pass it to
24789         ipa_get_indirect_edge_target_1 instead of calling
24790         ipa_get_indirect_edge_target.
24791         (create_specialized_node): Pass aggvlas to
24792         ipcp_discover_new_direct_edges.
24794 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24796         * config/arm/arm.md (f_sels, f_seld): New types.
24797         (*cmov<mode>): New pattern.
24798         * config/arm/predicates.md (arm_vsel_comparison_operator): New
24799         predicate.
24801 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24803         * config/i386/xm-mingw32.h (__USE_MINGW_ANSI_STDIO): Enable
24804         POSIX-printf for mingw-hosted  builds.
24806 2013-03-25  Richard Biener  <rguenther@suse.de>
24808         PR middle-end/56694
24809         * tree-eh.c (lower_eh_must_not_throw): Strip BLOCKs from the
24810         must-not-throw stmt location.
24812 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24814         * config/arm/arm.c (arm_emit_load_exclusive): Add acq parameter.
24815         Emit load-acquire versions when acq is true.
24816         (arm_emit_store_exclusive): Add rel parameter.
24817         Emit store-release versions when rel is true.
24818         (arm_split_compare_and_swap): Use acquire-release instructions
24819         instead.
24820         of barriers when appropriate.
24821         (arm_split_atomic_op): Likewise.
24822         * config/arm/arm.h (TARGET_HAVE_LDACQ): New macro.
24823         * config/arm/unspecs.md (VUNSPEC_LAX): New unspec.
24824         (VUNSPEC_SLX): Likewise.
24825         (VUNSPEC_LDA): Likewise.
24826         (VUNSPEC_STL): Likewise.
24827         * config/arm/sync.md (atomic_load<mode>): New pattern.
24828         (atomic_store<mode>): Likewise.
24829         (arm_load_acquire_exclusive<mode>): Likewise.
24830         (arm_load_acquire_exclusivesi): Likewise.
24831         (arm_load_acquire_exclusivedi): Likewise.
24832         (arm_store_release_exclusive<mode>): Likewise.
24834 2013-03-25  Catherine Moore  <clm@codesourcery.com>
24836         * config/mips/constraints.md (u, Udb7 Uead, Uean, Uesp, Uib3,
24837         Uuw6, Usb4, ZS, ZT, ZU, ZV, ZW): New constraints.
24838         * config/mip/predicates.md (lwsp_swsp_operand,
24839         lw16_sw16_operand, lhu16_sh16_operand, lbu16_operand,
24840         sb16_operand, db4_operand, db7_operand, ib3_operand,
24841         sb4_operand, ub4_operand, uh4_operand, uw4_operand,
24842         uw5_operand, uw6_operand, addiur2_operand, addiusp_operand,
24843         andi16_operand): New predicates.
24844         * config/mips/mips.md (compression): New attribute.
24845         (enabled): New attribute.
24846         (length): Consider compression in computing length.
24847         (shift_compression): New code attribute.
24848         (*add<mode>3): New operands. Record compression.
24849         (sub<mode>3): Likewise.
24850         (one_cmpl<mode>2): Likewise.
24851         (*and<mode>3): Likewise.
24852         (*ior<mode>3): Likewise.
24853         (unnamed pattern for xor): Likewise.
24854         (*zero_extend<SHORT:mode><GPR:mode>2): Likewise.
24855         (*<optab><mode>3): Likewise.
24856         (*mov<mode>_internal: Likewise.
24857         * config/mips/mips-protos.h (mips_signed_immediate_p): New.
24858         (mips_unsigned_immediate_p): New.
24859         (umips_lwsp_swsp_address_p): New.
24860         (m16_based_address_p): New.
24861         * config/mips/mips-protos.h (mips_signed_immediate_p): New prototype.
24862         (mips_unsigned_immediate_p): New prototype.
24863         (lwsp_swsp_address_p): New prototype.
24864         (m16_based_address_p): New prototype.
24865         * config/mips/mips.c (mips_unsigned_immediate_p): New function.
24866         (mips_signed_immediate_p): New function.
24867         (m16_based_address_p): New function.
24868         (lwsp_swsp_address_p): New function.
24869         (mips_print_operand_punctuation): Recognize short delay slot insns
24870         for microMIPS.add<mode>3"
24872 2013-03-25  Kyrylo Tkachov  <kyrylo.tkachov@arm.com>
24874         PR target/56720
24875         * config/arm/iterators.md (v_cmp_result): New mode attribute.
24876         * config/arm/neon.md (vcond<mode><mode>): Handle unordered cases.
24878 2013-03-25  Richard Biener  <rguenther@suse.de>
24880         PR tree-optimization/56689
24881         * tree-vrp.c (execute_vrp): Mark loops for fixup if we removed
24882         any edge.
24884 2013-03-25  Richard Biener  <rguenther@suse.de>
24886         * tree-ssa-loop-im.c (struct mem_ref): Use bitmap_head instead
24887         of bitmap.
24888         (memory_references): Likewise.
24889         (outermost_indep_loop, mem_ref_alloc, mark_ref_stored,
24890         gather_mem_refs_stmt, record_dep_loop, ref_indep_loop_p_1,
24891         ref_indep_loop_p_2, find_refs_for_sm): Adjust.
24892         (gather_mem_refs_in_loops): Fold into ...
24893         (analyze_memory_references): ... this.  Move initialization
24894         to tree_ssa_lim_initialize.
24895         (fill_always_executed_in): Rename to ...
24896         (fill_always_executed_in_1): ... this.
24897         (fill_always_executed_in): Move contains_call computation to
24898         this new function from ...
24899         (tree_ssa_lim_initialize): ... here.
24900         (tree_ssa_lim): Call fill_always_executed_in.
24902 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24904         * postreload.c (reload_combine): Fix code detecting returns.
24906 2013-03-25  Eric Botcazou  <ebotcazou@adacore.com>
24908         * function.c (emit_use_return_register_into_block): On cc0 targets,
24909         do not emit the sequence between cc0 setter and user.
24911 2013-03-25  Kai Tietz  <ktietz@redhat.com>
24913         * config/i386/predicates.md (local_symbolic_operand): Interpret
24914         dll-imported symbols as none-local.
24916 2013-03-25  Richard Biener  <rguenther@suse.de>
24918         * tree-ssa-loop-im.c (struct depend): Remove.
24919         (struct lim_aux_data): Make depends a vec of gimples.
24920         (free_lim_aux_data): Adjust.
24921         (add_dependency): Likewise.
24922         (set_level): Likewise.
24924 2013-03-25  Richard Biener  <rguenther@suse.de>
24926         PR middle-end/56434
24927         * calls.c (expand_call): Use MALLOC_ABI_ALIGNMENT to annotate
24928         the pointer returned by calls with ECF_MALLOC set.
24930 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24932         * config/i386/mmx.md (mov<mode>): Add ?!Ym,r and r,?!Ym alternatives.
24934 2013-03-24  Uros Bizjak  <ubizjak@gmail.com>
24936         * config/i386/mmx.md (mov<mode>): Merge with movv2sf expander
24937         using MMXMODE mode iterator.
24938         (*move<mode>_internal): Merge with *movv2sf_internal and
24939         *movv2sf_internal_rex64 using MMXMODE mode iterator.
24941 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24943         * gcse.c (oprs_unchanged_p): Respect flag_gcse_lm.
24944         (record_last_mem_set_info): Likewise.
24946         * df-core.c (rest_of_handle_df_initialize): Use XCNEWVEC instead
24947         of XNEWVEC followed by memset.
24948         (df_worklist_dataflow): Use XNEWVEC instead of xmalloc with a cast.
24950 2013-03-23  Steven Bosscher  <steven@gcc.gnu.org>
24952         * config/avr/avr.c, config/bfin/bfin.c, config/c6x/c6x.c,
24953         config/epiphany/epiphany.c, config/frv/frv.c, config/ia64/ia64.c,
24954         config/iq2000/iq2000.c, config/mcore/mcore.c, config/mep/mep.c,
24955         config/mmix/mmix.c, config/pa/pa.c, config/rs6000/rs6000.c,
24956         config/s390/s390.c, config/sparc/sparc.c, config/spu/spu.c,
24957         config/stormy16/stormy16.c, config/v850/v850.c, config/xtensa/xtensa.c,
24958         dwarf2out.c, hw-doloop.c, resource.c, rtl.h : Where applicable, use
24959         the predicates NOTE_P, NONJUMP_INSN_P, JUMP_P, CALL_P, LABEL_P, and
24960         BARRIER_P instead of GET_CODE.
24962 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24964         * config/sparc/sparc.c (sparc_emit_probe_stack_range): Fix small
24965         inaccuracy in the probing code.
24967         * config/sparc/sparc.md (ctrapsi4): Add predicate for operand #3.
24968         (ctrapdi4): Likewise.
24970 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24972         * calls.c (expand_call): Add missing guard to code handling return
24973         of non-BLKmode structures in MSB.
24974         * function.c (expand_function_end): Likewise.
24976 2013-03-23  Eric Botcazou  <ebotcazou@adacore.com>
24978         * combine.c (try_combine): Adjust comment.  Do not add the set of
24979         insn #0 if the destination indirectly is set or dies in insn #2.
24980         Tidy up code to distribute a new note.
24982 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24984         * config/i386/i386.md (*movdi_internal): Set prefix_rex attribute
24985         also for alternatives 16 and 17.
24987 2013-03-22  Uros Bizjak  <ubizjak@gmail.com>
24989         * config/i386/sse.md (*mov<mode>_internal): Merge with
24990         *mov<mode>_internal_rex64.  Use x64 and nox64 isa attributes.
24991         Emit insn template depending on type attribute.  Use
24992         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
24993         movd instead of movq mnemonic for interunit moves.  Rewrite mode
24994         attribute calculation.  Remove unit attribute calculation.
24995         Set prefix attribute to maybe_vex for sselog1 and ssemov types.
24996         Set prefix_data16 attribute for DImode ssemov types.
24997         Use Ym instead of y for SSE-MMX conversion alternatives.
24998         Reorder operand constraints.
25000 2013-03-22  Steven Bosscher  <steven@gcc.gnu.org>
25002         * df.h (df_insn_delete): Adjust prototype.
25003         * emit-rtl.c (remove_insn): Pass a basic block to df_insn_delete
25004         and let it decide whether mark the basic block dirty.
25005         (set_insn_deleted): Only pass INSN_P insns to df_insn_delete.
25006         * df-scan.c (df_insn_info_delete): New helper function, split
25007         off from df_insn_delete.
25008         (df_scan_free_bb_info): Use it.
25009         (df_insn_rescan, df_insn_rescan_all, df_process_deferred_rescans):
25010         Likewise.
25011         (df_insn_delete): Likewise.  Take insn rtx as argument.  Verify
25012         that the insn is actually an insn and it has a non-NULL basic block.
25013         Do not mark basic block dirty if only deleting a DEBUG_INSN.
25015 2013-03-22  Richard Biener  <rguenther@suse.de>
25017         * tree-ssa-loop-im.c (struct mem_ref): Remove indep_ref and
25018         dep_ref members.
25019         (mem_ref_alloc): Do not allocate them.
25020         (refs_independent_p): Do not query or maintain a cache.
25022 2013-03-22  Richard Biener  <rguenther@suse.de>
25024         * tree-ssa-loop-im.c (memory_references): Drop all_refs_in_loop.
25025         (gather_mem_refs_in_loops): Do not compute it.
25026         (analyze_memory_references): Do not allocate it.
25027         (tree_ssa_lim_finalize): Do not free it.
25028         (for_all_locs_in_loop): Do not query all_refs_in_loop.
25030 2013-03-22  Richard Biener  <rguenther@suse.de>
25032         * is-a.h (as_a): Use gcc_checking_assert.
25034 2013-03-22  Ian Bolton  <ian.bolton@arm.com>
25036         * config/aarch64/aarch64.c (aarch64_print_operand): New
25037         format specifier for printing a constant in hex.
25038         * config/aarch64/aarch64.md (insv_imm<mode>): Use the X
25039         format specifier for printing second operand.
25041 2013-03-22  Richard Biener  <rguenther@suse.de>
25043         * tree-ssa-loop-im.c (memory_references): Add refs_stored_in_loop
25044         bitmaps.
25045         (gather_mem_refs_in_loops): Perform store accumulation here.
25046         (create_vop_ref_mapping_loop): Remove.
25047         (create_vop_ref_mapping): Likewise.
25048         (analyze_memory_references): Initialize refs_stored_in_loop.
25049         (LOOP_DEP_BIT): New define to map to bits in (in)dep_loop bitmaps.
25050         (record_indep_loop): Remove.
25051         (record_dep_loop): New function.
25052         (ref_indep_loop_p_1): Adjust to only walk over references
25053         in the loop, not its subloops.
25054         (ref_indep_loop_p): Rename to ...
25055         (ref_indep_loop_p_2): ... this and recurse over the loop tree,
25056         maintaining a more fine-grained cache.
25057         (ref_indep_loop_p): Wrap ref_indep_loop_p_2.
25058         (tree_ssa_lim_finalize): Free refs_stored_in_loop.
25060 2013-03-22  Richard Biener  <rguenther@suse.de>
25062         * tree-ssa-loop-im.c (struct mem_ref_locs): Remove.
25063         (struct mem_ref): Make accesses_in_loop a vec of a vec of
25064         aggregate mem_ref_loc.
25065         (free_mem_ref_locs): Inline into ...
25066         (memref_free): ... this and adjust.
25067         (mem_ref_alloc): Adjust.
25068         (mem_ref_locs_alloc): Remove.
25069         (record_mem_ref_loc): Adjust.
25070         (get_all_locs_in_loop): Rewrite into ...
25071         (for_all_locs_in_loop): ... this iterator.
25072         (rewrite_mem_ref_loc): New functor.
25073         (rewrite_mem_refs): Use for_all_locs_in_loop.
25074         (sm_set_flag_if_changed): New functor.
25075         (execute_sm_if_changed_flag_set): Use for_all_locs_in_loop.
25076         (ref_always_accessed): New functor.
25077         (ref_always_accessed_p): Use for_all_locs_in_loop.
25079 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25081         * tree-pass.h (PROP_gimple_lvec): New.
25082         * passes.c (dump_properties): Handle PROP_gimple_lvec.
25083         (init_optimization_passes): Move pass_lower_vector.
25084         * tree-vect-generic.c (gate_expand_vector_operations_ssa): Test
25085         PROP_gimple_lvec.
25086         (pass_lower_vector): Provide PROP_gimple_lvec.
25087         (pass_lower_vector_ssa): Likewise.
25088         * cfgexpand.c (pass_expand): Require PROP_gimple_lvec.
25090 2013-03-21  Mark Wielaard  <mjw@redhat.com>
25092         * dwarf2out.c (size_of_aranges): Skip DECL_IGNORED_P functions.
25094 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25096         * config/i386/i386.md (*movdi_internal): Disparage slightly
25097         all MMX moves to/from memory.  Use Yi instead of x for SSE-MMX
25098         conversion alternatives.
25100 2013-03-21  Jakub Jelinek  <jakub@redhat.com>
25102         PR middle-end/48087
25103         * diagnostic.def (DK_WERROR): New kind.
25104         * diagnostic.h (werrorcount): Define.
25105         * diagnostic.c (diagnostic_report_diagnostic): For DK_WARNING
25106         promoted to DK_ERROR, increment DK_WERROR counter instead of
25107         DK_ERROR counter.
25108         * toplev.c (toplev_main): Call print_ignored_options even if
25109         just werrorcount is non-zero.  Exit with FATAL_EXIT_CODE
25110         even if just werrorcount is non-zero.
25112         PR debug/55608
25113         * dwarf2out.c (tree_add_const_value_attribute): Call ggc_free (array)
25114         on failure.
25115         (resolve_one_addr): Fail if referenced STRING_CST hasn't been written.
25116         (string_cst_pool_decl): New function.
25117         (optimize_one_addr_into_implicit_ptr): New function.
25118         (resolve_addr_in_expr): Optimize DWARF location expression
25119         DW_OP_addr DW_OP_stack_value where DW_OP_addr refers to some variable
25120         which doesn't live in memory, but has DW_AT_location or
25121         DW_AT_const_value, or refers to a string literal, into
25122         DW_OP_GNU_implicit_pointer.
25123         (optimize_location_into_implicit_ptr): New function.
25124         (resolve_addr): If removing DW_AT_location of a variable because
25125         it was DW_OP_addr of address of the variable, but the variable doesn't
25126         live in memory, try to emit const value attribute for the initializer.
25128 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25130         * tree.h (VECTOR_TYPE_P): New macro.
25131         (VECTOR_INTEGER_TYPE_P, VECTOR_FLOAT_TYPE_P, FLOAT_TYPE_P,
25132         TYPE_MODE): Use it.
25133         * fold-const.c (fold_cond_expr_with_comparison): Use build_zero_cst.
25134         VEC_COND_EXPR cannot be lvalues.
25135         (fold_ternary_loc) <VEC_COND_EXPR>: Merge with the COND_EXPR case.
25137 2013-03-21  Marc Glisse  <marc.glisse@inria.fr>
25139         * simplify-rtx.c (simplify_binary_operation_1) <VEC_CONCAT>:
25140         Restrict the transformation to equal modes.
25142 2013-03-21  Richard Biener  <rguenther@suse.de>
25144         PR tree-optimization/39326
25145         * tree-ssa-loop-im.c (UNANALYZABLE_MEM_ID): New define.
25146         (MEM_ANALYZABLE): Adjust.
25147         (record_mem_ref_loc): Move bitmap ops ...
25148         (gather_mem_refs_stmt): ... here.  Use the shared mem-ref for
25149         unanalyzable refs, do not record locations for it.
25150         (analyze_memory_references): Allocate ref zero as shared
25151         unanalyzable ref.
25152         (refs_independent_p): Do not test for unanalyzed mems here.
25153         (ref_indep_loop_p_1): Special-case disambiguation against
25154         the unanalyzed ref.
25155         (ref_indep_loop_p): Assert we are not queried for the unanalyzed mem.
25157 2013-03-21  Christophe Lyon  <christophe.lyon@linaro.org>
25159         * config/arm/arm-protos.h (tune_params): Add
25160         prefer_neon_for_64bits field.
25161         * config/arm/arm.c (prefer_neon_for_64bits): New variable.
25162         (arm_slowmul_tune): Default prefer_neon_for_64bits to false.
25163         (arm_fastmul_tune, arm_strongarm_tune, arm_xscale_tune): Ditto.
25164         (arm_9e_tune, arm_v6t2_tune, arm_cortex_tune): Ditto.
25165         (arm_cortex_a15_tune, arm_cortex_a5_tune): Ditto.
25166         (arm_cortex_a9_tune, arm_v6m_tune, arm_fa726te_tune): Ditto.
25167         (arm_option_override): Handle -mneon-for-64bits new option.
25168         * config/arm/arm.h (TARGET_PREFER_NEON_64BITS): New macro.
25169         (prefer_neon_for_64bits): Declare new variable.
25170         * config/arm/arm.md (arch): Rename neon_onlya8 and neon_nota8 to
25171         avoid_neon_for_64bits and neon_for_64bits. Remove onlya8 and nota8.
25172         (arch_enabled): Handle new arch types. Remove support for onlya8
25173         and nota8.
25174         (one_cmpldi2): Use new arch names.
25175         (zero_extend<mode>di2, extend<mode>di2): Ditto.
25176         * config/arm/arm.opt (mneon-for-64bits): Add option.
25177         * config/arm/neon.md (adddi3_neon, subdi3_neon, iordi3_neon)
25178         (anddi3_neon, xordi3_neon, ashldi3_neon, <shift>di3_neon): Use
25179         neon_for_64bits instead of nota8 and avoid_neon_for_64bits instead
25180         of onlya8.
25181         * doc/invoke.texi (-mneon-for-64bits): Document.
25183 2013-03-21  Richard Biener  <rguenther@suse.de>
25185         PR tree-optimization/39326
25186         * tree-ssa-loop-im.c (bb_loop_postorder): New global static.
25187         (sort_bbs_in_loop_postorder_cmp): New function.
25188         (gather_mem_refs_in_loops): Assign mem-ref IDs in loop postorder.
25190 2013-03-21  Richard Biener  <rguenther@suse.de>
25192         * tree-vect-data-refs.c (vect_update_interleaving_chain): Remove.
25193         (vect_insert_into_interleaving_chain): Likewise.
25194         (vect_drs_dependent_in_basic_block): Inline ...
25195         (vect_slp_analyze_data_ref_dependence): ... here.  New function,
25196         split out from ...
25197         (vect_analyze_data_ref_dependence): ... here.  Simplify.
25198         (vect_check_interleaving): Simplify.
25199         (vect_analyze_data_ref_dependences): Likewise.  Split out ...
25200         (vect_slp_analyze_data_ref_dependences): ... this new function.
25201         (dr_group_sort_cmp): New function.
25202         (vect_analyze_data_ref_accesses): Compute data-reference groups
25203         here instead of in vect_analyze_data_ref_dependence.  Use
25204         a more efficient algorithm.
25205         * tree-vect-slp.c (vect_slp_analyze_bb_1): Use
25206         vect_slp_analyze_data_ref_dependences.  Call
25207         vect_analyze_data_ref_accesses earlier.
25208         * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
25209         * tree-vectorizer.h (vect_analyze_data_ref_dependences): Adjust.
25210         (vect_slp_analyze_data_ref_dependences): New prototype.
25212 2013-03-21  Richard Biener  <rguenther@suse.de>
25214         * tree-ssa-loop-im.c (can_sm_ref_p): Do not test whether
25215         ref is stored in the loop.
25216         (find_refs_for_sm): Walk only over all stores.
25217         (store_motion_loop): Allocate from lim_bitmap_obstack.
25218         (store_motion): Likewise.
25220 2013-03-21  Richard Biener  <rguenther@suse.de>
25222         * tree-vect-loop-manip.c (slpeel_tree_peel_loop_to_edge):
25223         Update virtual SSA form.
25225 2013-03-21  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
25227         * configure.ac (gcc_cv_ld_eh_frame_ciev3): New test.
25228         * configure: Regenerate.
25229         * config.in: Regenerate.
25230         * config/sol2.c (solaris_override_options): Only enforce DWARF 2
25231         if !HAVE_LD_EH_FRAME_CIEV3.
25233 2013-03-21  Richard Biener  <rguenther@suse.de>
25235         * tree-cfg.c (verify_expr_no_block): New function.
25236         (verify_expr_location_1): Verify that neither DECL_DEBUG_EXPR
25237         nor DECL_VALUE_EXPR have locations with associated blocks.
25238         * tree-ssa-live.c (clear_unused_block_pointer_1): Remove.
25239         (clear_unused_block_pointer): Remove code dealing with
25240         blocks in DECL_DEBUG_EXPR locations.
25242 2013-03-21  Richard Biener  <rguenther@suse.de>
25244         * tree.h (DECL_DEBUG_EXPR_IS_FROM): Rename to ...
25245         (DECL_HAS_DEBUG_EXPR_P): ... this.  Guard properly.
25246         * tree.c (copy_node_stat): Do not copy DECL_HAS_DEBUG_EXPR_P.
25247         * dwarf2out.c (add_var_loc_to_decl): Use DECL_HAS_DEBUG_EXPR_P
25248         instead of DECL_DEBUG_EXPR_IS_FROM.
25249         * gimplify.c (gimplify_modify_expr): Likewise.
25250         * tree-cfg.c (verify_expr_location_1): Likewise.
25251         * tree-complex.c (create_one_component_var): Likewise.
25252         * tree-sra.c (create_access_replacement): Likewise.
25253         * tree-ssa-live.c (clear_unused_block_pointer_1): Likewise.
25254         (clear_unused_block_pointer): Likewise.
25255         * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Likewise.
25256         * tree-streamer-out.c (pack_ts_decl_common_value_fields): Likewise.
25257         * var-tracking.c (var_debug_decl): Likewise.
25258         (track_expr_p): Likewise.
25259         * tree-inline.c (add_local_variables): Likewise.  Set
25260         DECL_HAS_DEBUG_EXPR_P after copying it.
25261         * tree-diagnostic.c (default_tree_printer): Use DECL_HAS_DEBUG_EXPR_P
25262         instead of DECL_DEBUG_EXPR_IS_FROM.  Guard properly.
25264 2013-03-21  Uros Bizjak  <ubizjak@gmail.com>
25266         PR bootstrap/56656
25267         * configure.ac (HAVE_AS_IX86_INTERUNIT_MOVQ): New test.
25268         * configure: Regenerate.
25269         * config.in: Regenerate.
25270         * config/i386/i386.md (*movdf_internal): Use
25271         HAVE_AS_IX86_INTERUNIT_MOVQ to handle broken assemblers that require
25272         movd instead of movq mnemonic for interunit moves.
25273         (*movdi_internal): Ditto.
25275 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25277         * config/aarch64/aarch64-simd.md (simd_fabd): New Attribute.
25278         (abd<mode>_3): New pattern.
25279         (aba<mode>_3): New pattern.
25280         (fabd<mode>_3): New pattern.
25282 2013-03-21  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
25284         * config/aarch64/aarch64-elf.h (REGISTER_PREFIX): Remove.
25285         * config/aarch64/aarch64.c (aarch64_print_operand): Remove all
25286         occurrence of REGISTER_PREFIX as its empty string.
25288 2013-03-20  Jeff Law  <law@redhat.com>
25290         * tree-ssa-dom.c (record_equivalences_from_incoming_edge): Record
25291         addititional equivalences for equality comparisons between an SSA_NAME
25292         and a constant where the SSA_NAME was set from a widening conversion.
25294 2013-03-20  Walter Lee  <walt@tilera.com>
25296         * config/tilegx/sync.md (atomic_test_and_set): New pattern.
25298 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25300         * config/i386/i386.md (*movoi_internal_avx): Emit insn template
25301         depending on type attribute.
25302         (*movti_internal): Ditto.
25303         (*movtf_internal): Ditto.
25304         (*movxf_internal): Ditto.
25305         (*movdf_internal): Ditto.
25306         (*movsf_internal): Ditto.
25308 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25310         * config/i386/i386.md (*movti_internal): Set prefix attribute to
25311         maybe_vex for sselog1 and ssemov types.
25312         (*movdi_internal): Reorder operand constraints.
25313         (*movsi_internal): Ditto.  Set prefix attribute to
25314         maybe_vex for sselog1 and ssemov types.
25315         (*movtf_internal): Set prefix attribute to maybe_vex
25316         for sselog1 and ssemov types.
25317         (*movdf_internal): Ditto.  Set prefix_data16 attribute for
25318         DImode ssemov types.  Reorder operand constraints.
25319         (*movsf_internal): Set type of alternatives 3,4 to imov.  Set prefix
25320         attribute to maybe_vex for sselog1 and ssemov types.  Set prefix_data16
25321         attribute for SImode ssemov types.  Reorder operand constraints.
25323 2013-03-20  Martin Jambor  <mjambor@suse.cz>
25325         * params.def (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): New parameter.
25326         * ipa-cp.c (hint_time_bonus): Add abonus for known array indices.
25328 2013-03-20  Pat Haugen  <pthaugen@us.ibm.com>
25330         * config/rs6000/predicates.md (indexed_address, update_address_mem
25331         update_indexed_address_mem): New predicates.
25332         * config/rs6000/vsx.md (vsx_extract_<mode>_zero): Set correct "type"
25333         attribute for load/store instructions.
25334         * config/rs6000/dfp.md (movsd_store): Likewise.
25335         (movsd_load): Likewise.
25336         * config/rs6000/rs6000.md (zero_extend<mode>di2_internal1): Likewise.
25337         (unnamed HI->DI extend define_insn): Likewise.
25338         (unnamed SI->DI extend define_insn): Likewise.
25339         (unnamed QI->SI extend define_insn): Likewise.
25340         (unnamed QI->HI extend define_insn): Likewise.
25341         (unnamed HI->SI extend define_insn): Likewise.
25342         (unnamed HI->SI extend define_insn): Likewise.
25343         (extendsfdf2_fpr): Likewise.
25344         (movsi_internal1): Likewise.
25345         (movsi_internal1_single): Likewise.
25346         (movhi_internal): Likewise.
25347         (movqi_internal): Likewise.
25348         (movcc_internal1): Correct mnemonic for stw insn. Set correct "type"
25349         attribute for load/store instructions.
25350         (mov<mode>_hardfloat): Set correct "type" attribute for load/store
25351         instructions.
25352         (mov<mode>_softfloat): Likewise.
25353         (mov<mode>_hardfloat32): Likewise.
25354         (mov<mode>_hardfloat64): Likewise.
25355         (mov<mode>_softfloat64): Likewise.
25356         (movdi_internal32): Likewise.
25357         (movdi_internal64): Likewise.
25358         (probe_stack_<mode>): Likewise.
25360 2013-03-20  Michael Meissner  <meissner@linux.vnet.ibm.com>
25362         * config/rs6000/vector.md (VEC_R): Add 32-bit integer, binary
25363         floating point, and decimal floating point to reload iterator.
25365         * config/rs6000/constraints.md (wl constraint): New constraints to
25366         return FLOAT_REGS if certain options are used to reduce the number
25367         of separate patterns that exist in the file.
25368         (wx constraint): Likewise.
25369         (wz constraint): Likewise.
25371         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25372         -mdebug=reg, print wg, wl, wx, and wz constraints.
25373         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25374         Initialize the reload functions for 64-bit binary/decimal floating
25375         point types.
25376         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25377         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25378         create the buffer on the stack to overcome not having a 32-bit
25379         load and store.
25380         (rs6000_emit_move): Likewise.
25381         (rs6000_secondary_memory_needed_rtx): Likewise.
25382         (rs6000_alloc_sdmode_stack_slot): Likewise.
25383         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25384         via xxlxor, just like DFmode 0.0.
25386         * config/rs6000/rs6000.h (TARGET_NO_SDMODE_STACK): New macro,
25387         define as 1 if we are running on a power7 or newer.
25388         (enum r6000_reg_class_enum): Add new constraints.
25390         * config/rs6000/dfp.md (movsd): Delete, combine with binary
25391         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25392         with other moves by using conditional constraits (wg).  Use LFIWZX
25393         and STFIWX for loading SDmode on power7.  Use xxlxor to create 0.0f.
25394         (movsd splitter): Likewise.
25395         (movsd_hardfloat): Likewise.
25396         (movsd_softfloat): Likewise.
25398         * config/rs6000/rs6000.md (FMOVE32): New iterators to combine
25399         binary and decimal floating point moves.
25400         (fmove_ok): New attributes to combine binary and decimal floating
25401         point moves, and to combine power6x (mfpgpr) moves along normal
25402         floating moves.
25403         (real_value_to_target): Likewise.
25404         (f32_lr): Likewise.
25405         (f32_lm): Likewise.
25406         (f32_li): Likewise.
25407         (f32_sr): Likewise.
25408         (f32_sm): Likewise.
25409         (f32_si): Likewise.
25410         (movsf): Combine binary and decimal floating point moves.  Combine
25411         power6x (mfpgpr) moves with other moves by using conditional
25412         constraits (wg).  Use LFIWZX and STFIWX for loading SDmode on power7.
25413         (mov<mode> for SFmode/SDmode); Likewise.
25414         (SFmode/SDmode splitters): Likewise.
25415         (movsf_hardfloat): Likewise.
25416         (mov<mode>_hardfloat for SFmode/SDmode): Likewise.
25417         (movsf_softfloat): Likewise.
25418         (mov<mode>_softfloat for SFmode/SDmode): Likewise.
25420         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wl,
25421         wx and wz constraints.
25423         * config/rs6000/constraints.md (wg constraint): New constraint to
25424         return FLOAT_REGS if -mmfpgpr (power6x) was used.
25426         * config/rs6000/rs6000.h (enum r6000_reg_class_enum): Add wg
25427         constraint.
25429         * config/rs6000/rs6000.c (rs6000_debug_reg_global): If
25430         -mdebug=reg, print wg, wl, wx, and wz constraints.
25431         (rs6000_init_hard_regno_mode_ok): Initialize new constraints.
25432         Initialize the reload functions for 64-bit binary/decimal floating
25433         point types.
25434         (reg_offset_addressing_ok_p): If we are on a power7 or later, use
25435         LFIWZX and STFIWX to load/store 32-bit decimal types, and don't
25436         create the buffer on the stack to overcome not having a 32-bit
25437         load and store.
25438         (rs6000_emit_move): Likewise.
25439         (rs6000_secondary_memory_needed_rtx): Likewise.
25440         (rs6000_alloc_sdmode_stack_slot): Likewise.
25441         (rs6000_preferred_reload_class): On VSX, we can create SFmode 0.0f
25442         via xxlxor, just like DFmode 0.0.
25444         * config/rs6000/dfp.md (movdd): Delete, combine with binary
25445         floating point moves in rs6000.md.  Combine power6x (mfpgpr) moves
25446         with other moves by using conditional constraits (wg).  Use LFIWZX
25447         and STFIWX for loading SDmode on power7.
25448         (movdd splitters): Likewise.
25449         (movdd_hardfloat32): Likewise.
25450         (movdd_softfloat32): Likewise.
25451         (movdd_hardfloat64_mfpgpr): Likewise.
25452         (movdd_hardfloat64): Likewise.
25453         (movdd_softfloat64): Likewise.
25455         * config/rs6000/rs6000.md (FMOVE64): New iterators to combine
25456         64-bit binary and decimal floating point moves.
25457         (FMOVE64X): Likewise.
25458         (movdf): Combine 64-bit binary and decimal floating point moves.
25459         Combine power6x (mfpgpr) moves with other moves by using
25460         conditional constraits (wg).
25461         (mov<mode> for DFmode/DDmode): Likewise.
25462         (DFmode/DDmode splitters): Likewise.
25463         (movdf_hardfloat32): Likewise.
25464         (mov<mode>_hardfloat32 for DFmode/DDmode): Likewise.
25465         (movdf_softfloat32): Likewise.
25466         (movdf_hardfloat64_mfpgpr): Likewise.
25467         (movdf_hardfloat64): Likewise.
25468         (mov<mode>_hardfloat64 for DFmode/DDmode): Likewise.
25469         (movdf_softfloat64): Likewise.
25470         (mov<mode>_softfloat64 for DFmode/DDmode): Likewise.
25471         (reload_<mode>_load): Move to later in the file so they aren't in
25472         the middle of the floating point move insns.
25473         (reload_<mode>_store): Likewise.
25475         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wg
25476         constraint.
25478         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print out wg
25479         constraint if -mdebug=reg.
25480         (rs6000_initi_hard_regno_mode_ok): Enable wg constraint if -mfpgpr.
25481         Enable using dd reload support if needed.
25483         * config/rs6000/dfp.md (movtd): Delete, combine with 128-bit
25484         binary and decimal floating point moves in rs6000.md.
25485         (movtd_internal): Likewise.
25487         * config/rs6000/rs6000.md (FMOVE128): Combine 128-bit binary and
25488         decimal floating point moves.
25489         (movtf): Likewise.
25490         (movtf_internal): Likewise.
25491         (mov<mode>_internal, TDmode/TFmode): Likewise.
25492         (movtf_softfloat): Likewise.
25493         (mov<mode>_softfloat, TDmode/TFmode): Likewise.
25495         * config/rs6000/rs6000.md (movdi_mfpgpr): Delete, combine with
25496         movdi_internal64, using wg constraint for move direct operations.
25497         (movdi_internal64): Likewise.
25499         * config/rs6000/rs6000.c (rs6000_debug_reg_global): Print
25500         MODES_TIEABLE_P for selected modes.  Print the numerical value of
25501         the various virtual registers. Use GPR/FPR first/last values,
25502         instead of hard coding the register numbers.  Print which modes
25503         have reload functions registered.
25504         (rs6000_option_override_internal): If -mdebug=reg, trace the options
25505         settings before/after setting cpu, target and subtarget settings.
25506         (rs6000_secondary_reload_trace): Improve the RTL dump for -mdebug=addr
25507         and for secondary reload failures in rs6000_secondary_reload_inner.
25508         (rs6000_secondary_reload_fail): Likewise.
25509         (rs6000_secondary_reload_inner): Likewise.
25511         * config/rs6000/rs6000.md (FIRST_GPR_REGNO): Add convenience
25512         macros for first/last GPR and FPR registers.
25513         (LAST_GPR_REGNO): Likewise.
25514         (FIRST_FPR_REGNO): Likewise.
25515         (LAST_FPR_REGNO): Likewise.
25517         * config/rs6000/vector.md (mul<mode>3): Use the combined macro
25518         VECTOR_UNIT_ALTIVEC_OR_VSX_P instead of separate calls to
25519         VECTOR_UNIT_ALTIVEC_P and VECTOR_UNIT_VSX_P.
25520         (vcond<mode><mode>): Likewise.
25521         (vcondu<mode><mode>): Likewise.
25522         (vector_gtu<mode>): Likewise.
25523         (vector_gte<mode>): Likewise.
25524         (xor<mode>3): Don't allow logical operations on TImode in 32-bit
25525         to prevent the compiler from converting DImode operations to TImode.
25526         (ior<mode>3): Likewise.
25527         (and<mode>3): Likewise.
25528         (one_cmpl<mode>2): Likewise.
25529         (nor<mode>3): Likewise.
25530         (andc<mode>3): Likewise.
25532         * config/rs6000/constraints.md (wt constraint): New constraint
25533         that returns VSX_REGS if TImode is allowed in VSX registers.
25535         * config/rs6000/predicates.md (easy_fp_constant): 0.0f is an easy
25536         constant under VSX.
25538         * config/rs6000/rs6000-modes.def (PTImode): Define, PTImode is
25539         similar to TImode, but it is restricted to being in the GPRs.
25541         * config/rs6000/rs6000.opt (-mvsx-timode): New switch to allow
25542         TImode to occupy a single VSX register.
25544         * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Default to
25545         -mvsx-timode for power7/power8.
25546         (power7 cpu): Likewise.
25547         (power8 cpu): Likewise.
25549         * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Make
25550         sure that TFmode/TDmode take up two registers if they are ever
25551         allowed in the upper VSX registers.
25552         (rs6000_hard_regno_mode_ok): If -mvsx-timode, allow TImode in VSX
25553         registers.
25554         (rs6000_init_hard_regno_mode_ok): Likewise.
25555         (rs6000_debug_reg_global): Add debugging for PTImode and wt
25556         constraint.  Print if LRA is turned on.
25557         (rs6000_option_override_internal): Give an error if -mvsx-timode
25558         and VSX is not enabled.
25559         (invalid_e500_subreg): Handle PTImode, restricting it to GPRs.  If
25560         -mvsx-timode, restrict TImode to reg+reg addressing, and PTImode
25561         to reg+offset addressing.  Use PTImode when checking offset
25562         addresses for validity.
25563         (reg_offset_addressing_ok_p): Likewise.
25564         (rs6000_legitimate_offset_address_p): Likewise.
25565         (rs6000_legitimize_address): Likewise.
25566         (rs6000_legitimize_reload_address): Likewise.
25567         (rs6000_legitimate_address_p): Likewise.
25568         (rs6000_eliminate_indexed_memrefs): Likewise.
25569         (rs6000_emit_move): Likewise.
25570         (rs6000_secondary_reload): Likewise.
25571         (rs6000_secondary_reload_inner): Handle PTImode.  Allow 64-bit
25572         reloads to fpr registers to continue to use reg+offset addressing,
25573         but 64-bit reloads to altivec registers need reg+reg addressing.
25574         Drop test for PRE_MODIFY, since VSX loads/stores no longer support
25575         it.  Treat LO_SUM like a PLUS operation.
25576         (rs6000_secondary_reload_class): If type is 64-bit, prefer to use
25577         FLOAT_REGS instead of VSX_RGS to allow use of reg+offset addressing.
25578         (rs6000_cannot_change_mode_class): Do not allow TImode in VSX
25579         registers to share a register with a smaller sized type, since VSX
25580         puts scalars in the upper 64-bits.
25581         (print_operand): Add support for PTImode.
25582         (rs6000_register_move_cost): Use VECTOR_MEM_VSX_P instead of
25583         VECTOR_UNIT_VSX_P to catch types that can be loaded in VSX
25584         registers, but don't have arithmetic support.
25585         (rs6000_memory_move_cost): Add test for VSX.
25586         (rs6000_opt_masks): Add -mvsx-timode.
25588         * config/rs6000/vsx.md (VSm): Change to use 64-bit aligned moves
25589         for TImode.
25590         (VSs): Likewise.
25591         (VSr): Use wt constraint for TImode.
25592         (VSv): Drop TImode support.
25593         (vsx_movti): Delete, replace with versions for 32-bit and 64-bit.
25594         (vsx_movti_64bit): Likewise.
25595         (vsx_movti_32bit): Likewise.
25596         (vec_store_<mode>): Use VSX iterator instead of vector iterator.
25597         (vsx_and<mode>3): Delete use of '?' constraint on inputs, just put
25598         one '?' on the appropriate output constraint.  Do not allow TImode
25599         logical operations on 32-bit systems.
25600         (vsx_ior<mode>3): Likewise.
25601         (vsx_xor<mode>3): Likewise.
25602         (vsx_one_cmpl<mode>2): Likewise.
25603         (vsx_nor<mode>3): Likewise.
25604         (vsx_andc<mode>3): Likewise.
25605         (vsx_concat_<mode>): Likewise.
25606         (vsx_xxpermdi_<mode>): Fix thinko for non V2DF/V2DI modes.
25608         * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Map from
25609         OPTION_MASK_VSX_TIMODE.
25610         (enum rs6000_reg_class_enum): Add RS6000_CONSTRAINT_wt.
25611         (STACK_SAVEAREA_MODE): Use PTImode instead of TImode.
25613         * config/rs6000/rs6000.md (INT mode attribute): Add PTImode.
25614         (TI2 iterator): New iterator for TImode, PTImode.
25615         (wd mode attribute): Add values for vector types.
25616         (movti_string): Replace TI move operations with operations for TImode
25617         and PTImode.  Add support for TImode being allowed in VSX registers.
25618         (mov<mode>_string, TImode/PTImode): Likewise.
25619         (movti_ppc64): Likewise.
25620         (mov<mode>_ppc64, TImode/PTImode): Likewise.
25621         (TI mode splitters): Likewise.
25623         * doc/md.texi (PowerPC and IBM RS6000 constraints): Document wt
25624         constraint.
25626 2013-03-20  Marc Glisse  <marc.glisse@inria.fr>
25628         PR tree-optimization/56355
25629         * fold-const.c (tree_binary_nonnegative_warnv_p) <MULT_EXPR>:
25630         Also handle integers with undefined overflow.
25632 2013-03-20  Catherine Moore  <clm@codesourcery.com>
25633             Maciej W. Rozycki  <macro@codesourcery.com>
25634             Tom de Vries  <tom@codesourcery.com>
25635             Nathan Sidwell  <nathan@codesourcery.com>
25636             Iain Sandoe  <iain@codesourcery.com>
25637             Nathan Froyd  <froydnj@codesourcery.com>
25638             Chao-ying Fu  <fu@mips.com>
25640         * doc/extend.texi: (micromips, nomicromips, nocompression):
25641         Document new function attributes.
25642         * doc/invoke.texi (minterlink-compressed, mmicromips,
25643         m14k, m14ke, m14kec): Document new options.
25644         (minterlink-mips16): Update documentation.
25645         * doc/md.texi (ZC, ZD): Document new constraints.
25646         * configure.ac (gcc_cv_as_micromips): Check if linker
25647         supports the .set micromips directive.
25648         * configure: Regenerate.
25649         * config.in: Regenerate.
25650         * config/mips/mips-tables.opt: Regenerate.
25651         * config/mips/micromips.md: New file.
25652         * constraints.md (ZC, ZD): New constraints.
25653         * config/mips/predicates.md (movep_src_register): New predicate.
25654         (movep_src_operand): New predicate.
25655         (non_volatile_mem_operand): New predicate.
25656         * config/mips/mips.md (multimem): New type.
25657         (length): Differentiate between 17-bit and 18-bit branch offsets.
25658         (MOVEP1, MOVEP2): New mode iterator.
25659         (mov_<load>l): Use ZC constraint.
25660         (mov_<load>r): Likewise.
25661         (mov_<store>l): Likewise.
25662         (mov_<store>r): Likewise.
25663         (*branch_equality<mode>_inverted): Add microMIPS support.
25664         (*branch_equality<mode>): Likewise.
25665         (*jump_absolute): Likewise.
25666         (indirect_jump_<mode>): Likewise.
25667         (tablejump_<mode>): Likewise.
25668         (<optab>_internal): Likewise.
25669         (sibcall_internal): Likewise.
25670         (sibcall_value_internal): Likewise.
25671         (prefetch): Use constraint ZD.
25672         * config/mips/mips.opt (minterlink-compressed): New option.
25673         (minterlink-mips16): Now an alias for minterlink-compressed.
25674         (mmicromips): New option.
25675         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
25676         (compare_and_swap_12): Likewise.
25677         (sync_add<mode>): Likewise.
25678         (sync_<optab>_12): Likewise.
25679         (sync_old_<optab>_12): Likewise.
25680         (sync_new_<optab>_12): Likewise.
25681         (sync_nand_12): Likewise.
25682         (sync_old_nand_12): Likewise.
25683         (sync_new_nand_12): Likewise.
25684         (sync_sub<mode>): Likewise.
25685         (sync_old_add<mode>): Likewise.
25686         (sync_old_sub<mode>): Likewise.
25687         (sync_new_add<mode>): Likewise.
25688         (sync_new_sub<mode>): Likewise.
25689         (sync_<optab><mode>): Likewise.
25690         (sync_old_<optab><mode>): Likewise.
25691         (sync_new_<optab><mode>): Likewise.
25692         (sync_nand<mode>): Likewise.
25693         (sync_old_nand<mode>): Likewise.
25694         (sync_new_nand<mode>): Likewise.
25695         (sync_lock_test_and_set<mode>): Likewise.
25696         (test_and_set_12): Likewise.
25697         (atomic_compare_and_swap<mode>): Likewise.
25698         (atomic_exchange<mode>_llsc): Likewise.
25699         (atomic_fetch_add<mode>_llsc): Likewise.
25700         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
25701         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
25702         (umips_save_restore_pattern_p): Likewise.
25703         (umips_load_store_pair_p): Likewise.
25704         (umips_output_load_store_pair): Likewise.
25705         (umips_movep_target_p): Likewise.
25706         (umips_12bit_offset_address_p): Likewise.
25707         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
25708         (mips_base_mips16): Rename this...
25709         (mips_base_compression_flags): ...to this. Update all uses.
25710         (mips_attribute_table): Add micromips, nomicromips and nocompression.
25711         (mips_mips16_decl_p): Delete.
25712         (mips_nomips16_decl_p): Delete.
25713         (mips_get_compress_on_flags): New function.
25714         (mips_get_compress_off_flags): New function.
25715         (mips_get_compress_mode): New function.
25716         (mips_get_compress_on_name): New function.
25717         (mips_get_compress_off_name): New function.
25718         (mips_insert_attributes): Support multiple compression types.
25719         (mips_merge_decl_attributes): Likewise.
25720         (umips_12bit_offset_address_p): New function.
25721         (mips_start_function_definition): Emit .set micromips directive.
25722         (mips_call_may_need_jalx_p): New function.
25723         (mips_function_ok_for_sibcall): Add microMIPS support.
25724         (mips_print_operand_punctuation): Support short delay slots and
25725         compact jumps.
25726         (umips_swm_mask, umips_swm_encoding): New.
25727         (umips_build_save_restore): New function.
25728         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
25729         (was_mips16_p): Remove.
25730         (old_compression_mode): New.
25731         (mips_set_compression_mode): New function.
25732         (mips_set_current_function): Add microMIPS support.
25733         (mips_option_override): Likewise.
25734         (umips_save_restore_pattern_p): New function.
25735         (umips_output_save_restore): New function.
25736         (umips_load_store_pair_p_1): New function.
25737         (umips_load_store_pair_p): New function.
25738         (umips_output_load_store_pair_1): New function.
25739         (umips_output_load_store_pair): New function.
25740         (umips_movep_target_p) New function.
25741         (mips_prepare_pch_save): Add microMIPS support.
25742         * config/mips/mips.h (TARGET_COMPRESSION): New.
25743         (TARGET_CPU_CPP_BUILTINS): Update macro
25744         to use new compression flags and to support microMIPS.
25745         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
25746         (MIPS_ARCH_FLOAT_SPEC): Likewise.
25747         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
25748         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
25749         (ASM_SPEC): Support mmicromips and mno-micromips.
25750         (M16STORE_REG_P): New macro.
25751         (MIPS_CALL): Support TARGET_MICROMIPS.
25752         (MICROMIPS_J): New macro.
25753         (mips_base_mips16): Rename this...
25754         (mips_base_compression_flags): ...to this.
25755         (UMIPS_12BIT_OFFSET_P): New macro.
25756         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
25757         (MULTILIB_DIRNAMES): Likewise.
25758 2013-03-20  Richard Biener  <rguenther@suse.de>
25760         PR tree-optimization/56661
25761         * tree-ssa-sccvn.c (visit_use): Only value-number calls if
25762         the result does not have to be distinct.
25764 2013-03-20  Richard Biener  <rguenther@suse.de>
25766         * tree-inline.c (copy_tree_body_r): Sync MEM_REF code with
25767         remap_gimple_op_r.
25769 2013-03-20  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>
25770             Steven Bosscher  <steven@gcc.gnu.org>
25772         PR rtl-optimization/56605
25773         * loop-iv.c (implies_p): Handle equal RTXs and subregs.
25775 2013-03-20  Uros Bizjak  <ubizjak@gmail.com>
25777         PR bootstrap/56656
25778         * config/i386/i386.md (*movdi_internal): Handle broken assemblers
25779         that require movd instead of movq.
25781 2013-03-20  Richard Biener  <rguenther@suse.de>
25783         * tree-ssa-structalias.c (struct variable_info): Add pointer
25784         to the first field of an aggregate with sub-vars.  Make
25785         this and the pointer to the next subfield its ID.
25786         (vi_next): New function.
25787         (nothing_id, anything_id, readonly_id, escaped_id, nonlocal_id,
25788         storedanything_id, integer_id): Increment by one.
25789         (new_var_info, get_call_vi, lookup_call_clobber_vi,
25790         get_call_clobber_vi): Adjust.
25791         (solution_set_expand): Simplify and speedup.
25792         (solution_set_add): Inline into ...
25793         (set_union_with_increment): ... this.  Adjust accordingly.
25794         (do_sd_constraint): Likewise.
25795         (do_ds_constraint): Likewise.
25796         (do_complex_constraint): Simplify.
25797         (build_pred_graph): Adjust.
25798         (solve_graph): Likewise.  Simplify and speedup.
25799         (get_constraint_for_ssa_var, get_constraint_for_ptr_offset,
25800         get_constraint_for_component_ref, get_constraint_for_1,
25801         first_vi_for_offset, first_or_preceding_vi_for_offset,
25802         create_function_info_for, create_variable_info_for_1,
25803         create_variable_info_for, intra_create_variable_infos): Adjust.
25804         (init_base_vars): Push NULL for ID zero.
25805         (compute_points_to_sets): Adjust.
25807 2013-03-20  Richard Biener  <rguenther@suse.de>
25809         * cfgloop.c (verify_loop_structure): Streamline and avoid
25810         ICEing on corrupt loop tree.
25811         * graph.c (draw_cfg_nodes_for_loop): Avoid ICEing on corrupt
25812         loop tree.
25814 2013-03-20  Richard Biener  <rguenther@suse.de>
25816         * tree-vect-loop-manip.c (slpeel_can_duplicate_loop_p): Do not
25817         check whether an SSA update is needed.
25819 2013-03-20  Richard Sandiford  <rdsandiford@googlemail.com>
25821         * config/mips/constraints.md (T): Rename to...
25822         (Yf): ...this.
25823         (U): Rename to...
25824         (Yd): ...this.
25825         * config/mips/mips.md (*movdi_64bit, *movdi_64bit_mips16)
25826         (*mov<mode>_internal, *mov<mode>_mips16): Update accordingly.
25828 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25830         * config/aarch64/aarch64.md (*sub<mode>3_carryin): New pattern.
25831         (*subsi3_carryin_uxtw): Likewise.
25833 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25835         * config/aarch64/aarch64.md (*ror<mode>3_insn): New pattern.
25836         (*rorsi3_insn_uxtw): Likewise.
25838 2013-03-19  Ian Bolton  <ian.bolton@arm.com>
25840         * config/aarch64/aarch64.md (*extr<mode>5_insn): New pattern.
25841         (*extrsi5_insn_uxtw): Likewise.
25843 2013-03-19  Richard Biener  <rguenther@suse.de>
25845         PR tree-optimization/56273
25846         * passes.c (init_optimization_passes): Move second VRP after DOM.
25848 2013-03-19  Uros Bizjak  <ubizjak@gmail.com>
25850         * config/i386/i386.md (*movti_internal): Merge from
25851         *movti_internal_rex64 and *movti_internal_sse.  Use x64 isa attribute.
25852         (*movdi_internal): Merge with *movdi_internal_rex64.  Use x64 and
25853         nox64 isa attributes.
25855 2013-03-18  Richard Biener  <rguenther@suse.de>
25857         * tree-ssa-structalias.c (find): Use gcc_checking_assert.
25858         (unite): Likewise.
25859         (merge_node_constraints): Likewise.
25860         (build_succ_graph): Likewise.
25861         (valid_graph_edge): Inline into single caller.
25862         (unify_nodes): Likewise.  Use bitmap_set_bit return value
25863         and cache varinfo.
25864         (scc_visit): Fix formatting and variable use.
25865         (do_sd_constraint): Use gcc_checking_assert.
25866         (do_ds_constraint): Likewise.
25867         (do_complex_constraint): Likewise.
25868         (condense_visit): Likewise.  Cleanup.
25869         (dump_pred_graph): New function.
25870         (perform_var_substitution): Dump the pred-graph before
25871         variable substitution.
25872         (find_equivalent_node): Use gcc_checking_assert.
25873         (rewrite_constraints): Guard checking loop with ENABLE_CHECKING.
25875 2013-03-18  Richard Biener  <rguenther@suse.de>
25877         * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks):
25878         Remove cond_expr_stmt_list argument and do not gimplify the
25879         built expression.
25880         (vect_loop_versioning): Adjust.
25881         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref):
25882         Cleanup to use less temporaries.
25883         (vect_create_data_ref_ptr): Cleanup.
25885 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25887         PR tree-optimization/56635
25888         * fold-const.c (operand_equal_p): For MEM_REF and TARGET_MEM_REF,
25889         require types_compatible_p types.
25891 2013-03-18  Nick Clifton  <nickc@redhat.com>
25893         * config/stormy16/stormy16.c (xstormy16_expand_prologue): Remove
25894         spurious backslash.
25896         * config/mn10300/mn10300.c (mn10300_get_live_callee_saved_regs):
25897         Add missing line to comment describing function.
25899 2013-03-18  Richard Biener  <rguenther@suse.de>
25901         PR tree-optimization/56210
25902         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25903         Handle string / character search functions.
25904         * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
25906 2013-03-18  Richard Biener  <rguenther@suse.de>
25908         PR middle-end/56483
25909         * cfgexpand.c (expand_gimple_cond): Inline gimple_cond_single_var_p
25910         and implement properly.
25911         * gimple.h (gimple_cond_single_var_p): Remove.
25913 2013-03-18  Richard Biener  <rguenther@suse.de>
25915         * tree-data-ref.h (find_data_references_in_loop): Declare.
25916         * tree-data-ref.c (get_references_in_stmt): Use a stack
25917         vector pre-allocated in the callers.
25918         (find_data_references_in_stmt): Adjust.
25919         (graphite_find_data_references_in_stmt): Likewise.
25920         (create_rdg_vertices): Likewise.
25921         (find_data_references_in_loop): Export.
25922         * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
25923         Compute dependences here...
25924         (vect_analyze_data_refs): ...not here.  When we encounter
25925         a non-vectorizable data reference in basic-block vectorization
25926         truncate the data reference vector.  Do not bother to
25927         fixup data-dependence information for gather loads.
25928         * tree-vect-slp.c (vect_slp_analyze_bb_1): Check the number
25929         of data references, as reported.
25931 2013-03-18  Richard Biener  <rguenther@suse.de>
25933         PR tree-optimization/3713
25934         * tree-ssa-sccvn.c (visit_copy): Simplify.  Always propagate
25935         has_constants and expr.
25936         (stmt_has_constants): Properly valueize SSA names when deciding
25937         whether the stmt has constants.
25939 2013-03-18  Richard Biener  <rguenther@suse.de>
25941         * tree-ssa-loop-manip.c (find_uses_to_rename): Do not scan the
25942         whole function when there is nothing to do.
25943         * tree-ssa-loop.c (pass_vectorize): Remove TODO_update_ssa.
25944         * tree-vectorizer.c (vectorize_loops): Update virtual and
25945         loop-closed SSA once.
25946         * tree-vect-loop.c (vect_transform_loop): Do not update SSA here.
25948 2013-03-18  Richard Biener  <rguenther@suse.de>
25950         PR middle-end/56113
25951         * domwalk.c (bb_postorder): New global static.
25952         (cmp_bb_postorder): New function.
25953         (walk_dominator_tree): Replace scheme imposing an order for
25954         visiting dominator sons by one sorting them at the time they
25955         are pushed on the stack.
25957 2013-03-18  Richard Biener  <rguenther@suse.de>
25959         PR tree-optimization/39326
25960         * tree-ssa-loop-im.c (refs_independent_p): Exploit symmetry.
25961         (struct mem_ref): Replace mem member with ao_ref typed member.
25962         (MEM_ANALYZABLE): Adjust.
25963         (memref_eq): Likewise.
25964         (mem_ref_alloc): Likewise.
25965         (gather_mem_refs_stmt): Likewise.
25966         (mem_refs_may_alias_p): Use the ao_ref to query the alias oracle.
25967         (execute_sm_if_changed_flag_set): Adjust.
25968         (execute_sm): Likewise.
25969         (ref_always_accessed_p): Likewise.
25970         (refs_independent_p): Likewise.
25971         (can_sm_ref_p): Likewise.
25973 2013-03-18  Jakub Jelinek  <jakub@redhat.com>
25975         PR c/56566
25976         * tree.c (tree_int_cst_min_precision): For integer_zerop (value)
25977         return 1 even for !unsignedp.
25979 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25981         * config/i386/i386.md (isa): Add x64 and nox64.
25982         (enabled): Define x64 for TARGET_64BIT and nox64 for !TARGET_64BIT.
25983         (*pushtf): Enable *roF alternative for x64 isa only.
25984         (*pushxf): Merge with *pushxf_nointeger.  Use Yx*r constraint. Set
25985         mode attribute of integer alternatives to DImode for TARGET_64BIT.
25986         (*pushdf): Merge with *pushdf_rex64.  Use x64 and nox64 isa attributes.
25987         (*movtf_internal): Merge from *movtf_internal_rex64 and
25988         *movtf_internal_sse.  Use x64 and nox64 isa attributes.
25989         (*movxf_internal): Merge with *movxf_internal_rex64.  Use x64 and
25990         nox64 isa attributes.
25991         (*movdf_internal): Merge with *movdf_internal_rex64.  Use x64 and
25992         nox64 isa attributes.
25993         * config/i386/constraints.md (Yd): Do not set for TARGET_64BIT.
25995 2013-03-17  Uros Bizjak  <ubizjak@gmail.com>
25997         * config/alpha/alpha.c (TARGET_LRA_P): New define.
25999 2013-03-17  Jakub Jelinek  <jakub@redhat.com>
26001         PR target/56640
26002         * config/arm/arm.h (REG_CLASS_NAMES): Add "SFP_REG" and "AFP_REG"
26003         class names.  Remove trailing comma after "ALL_REGS".
26005 2013-03-16  Jan Hubicka  <jh@suse.cz>
26007         * cgraph.h (cgraph_get_create_real_symbol_node): Declare.
26008         * cgraph.c (cgraph_get_create_real_symbol_node): New function.
26009         * cgrpahbuild.c: Use cgraph_get_create_real_symbol_node instead
26010         of cgraph_get_create_node.
26011         * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
26013 2013-03-16  Jason Merrill  <jason@redhat.com>
26015         PR debug/49090
26016         * dwarf2out.c (gen_generic_params_dies): Indicate default arguments
26017         with DW_AT_default_value.
26019 2013-03-16  Jakub Jelinek  <jakub@redhat.com>
26021         * BASE-VER: Set to 4.9.0.
26023 2013-03-14  Andi Kleen  <ak@linux.intel.com>
26025         PR target/56619
26026         * doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE,
26027         __ATOMIC_HLE_RELEASE. Document __builtin_ia32 TSX intrincs.
26028         Document _x* TSX intrinsics.
26030 2013-03-14  Edgar E. Iglesias  <edgar.iglesias@xilinx.com>
26031             David Holsgrove  <david.holsgrove@xilinx.com>
26033         * configure.ac: Add MicroBlaze TLS support detection.
26034         * configure: Regenerate.
26035         * config/microblaze/microblaze-protos.h
26036         (microblaze_cannot_force_const_mem, microblaze_tls_referenced_p,
26037         symbol_mentioned_p, label_mentioned_p): Add prototypes.
26038         * config/microblaze/microblaze.c (microblaze_address_type): Add
26039         ADDRESS_TLS and tls_reloc address types.
26040         (microblaze_address_info): Add tls_reloc.
26041         (TARGET_HAVE_TLS): Define.
26042         (get_tls_get_addr, microblaze_tls_symbol_p, microblaze_tls_operand_p_1,
26043          microblaze_tls_referenced_p, microblaze_cannot_force_const_mem,
26044          symbol_mentioned_p, label_mentioned_p, tls_mentioned_p,
26045         load_tls_operand,  microblaze_call_tls_get_addr,
26046         microblaze_legitimize_tls_address): New functions.
26047         (microblaze_classify_unspec): Handle UNSPEC_TLS.
26048         (get_base_reg): Use microblaze_tls_symbol_p.
26049         (microblaze_classify_address): Handle TLS.
26050         (microblaze_legitimate_pic_operand): Use symbol_mentioned_p,
26051         label_mentioned_p and microblaze_tls_referenced_p.
26052         (microblaze_legitimize_address): Handle TLS.
26053         (microblaze_address_insns): Handle ADDRESS_TLS.
26054         (pic_address_needs_scratch): Handle TLS.
26055         (print_operand_address): Handle TLS.
26056         (microblaze_expand_prologue): Check TLS_NEEDS_GOT.
26057         (microblaze_expand_move): Handle TLS.
26058         (microblaze_legitimate_constant_p): Check
26059         microblaze_cannot_force_const_mem and microblaze_tls_symbol_p.
26060         (TARGET_CANNOT_FORCE_CONST_MEM): Define.
26061         * config/microblaze/microblaze.h (TLS_NEEDS_GOT): Define
26062         (PIC_OFFSET_TABLE_REGNUM): Set.
26063         * config/microblaze/linux.h (TLS_NEEDS_GOT): Define.
26064         * config/microblaze/microblaze.md (UNSPEC_TLS): Define.
26065         (addsi3, movsi_internal2, movdf_internal): Update constraints
26066         * config/microblaze/predicates.md (arith_plus_operand): Define
26067         (move_operand): Redefine as move_src_operand,
26068         check microblaze_tls_referenced_p.
26070 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
26072         * config/aarch64/aarch64.md: (*and<mode>3nr_compare0): Use CC_NZ.
26073         (*and_<SHIFT:optab><mode>3nr_compare0): Likewise.
26075 2013-03-14  Ian Bolton  <ian.bolton@arm.com>
26077         * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return correct
26078         CC mode for AND.
26080 2013-03-14  Jakub Jelinek  <jakub@redhat.com>
26082         PR tree-optimization/53265
26083         * common.opt (Waggressive-loop-optimizations): New option.
26084         * tree-ssa-loop-niter.c: Include tree-pass.h.
26085         (do_warn_aggressive_loop_optimizations): New function.
26086         (record_estimate): Call it.  Don't add !is_exit bounds to loop->bounds
26087         if number_of_latch_executions returned constant.
26088         (estimate_numbers_of_iterations_loop): Call number_of_latch_executions
26089         early.  If number_of_latch_executions returned constant, set
26090         nb_iterations_upper_bound back to it.
26091         * cfgloop.h (struct loop): Add warned_aggressive_loop_optimizations
26092         field.
26093         * Makefile.in (tree-ssa-loop-niter.o): Depend on $(TREE_PASS_H).
26094         * doc/invoke.texi (-Wno-aggressive-loop-optimizations): Document.
26096         * config/aarch64/t-aarch64-linux (MULTARCH_DIRNAME): Remove.
26097         (MULTILIB_OSDIRNAMES): Set.
26098         * genmultilib: If defaultosdirname doesn't start with :: , set
26099         defaultosdirname2 instead, clear it and emit two . multilib_raw
26100         entries instead of just one.
26102 2013-03-14  Kaz Kojima  <kkojima@gcc.gnu.org>
26104         * config/sh/linux.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26105         (SUBTARGET_OVERRIDE_OPTIONS): Set TARGET_USERMODE as default.
26106         * config/sh/netbsd-elf.h (TARGET_DEFAULT): Remove MASK_USERMODE.
26107         (SUBTARGET_OVERRIDE_OPTIONS): New.
26109 2013-03-13  Oleg Endo  <olegendo@gcc.gnu.org>
26111         PR target/49880
26112         * config/sh/sh.opt (FPU_SINGLE_ONLY): New mask.
26113         (musermode): Convert to Var(TARGET_USERMODE).
26114         * config/sh/sh.h (SELECT_SH2A_SINGLE_ONLY, SELECT_SH4_SINGLE_ONLY,
26115         MASK_ARCH): Add MASK_FPU_SINGLE_ONLY.
26116         * config/sh/sh.c (sh_option_override): Use
26117         TARGET_FPU_DOUBLE || TARGET_FPU_SINGLE_ONLY for call-fp case.
26118         * config/sh/sh.md (udivsi3_i1, divsi3_i1): Remove ! TARGET_SH4
26119         condition.
26120         (udivsi3_i4, divsi3_i4): Use TARGET_FPU_DOUBLE condition instead of
26121         TARGET_SH4.
26122         (udivsi3_i4_single, divsi3_i4_single): Use
26123         TARGET_FPU_SINGLE_ONLY || TARGET_FPU_DOUBLE instead of TARGET_HARD_SH4.
26125 2013-03-13  Dave Korn  <dave.korn.cygwin@gmail.com>
26127         * config/i386/cygwin.h (SHARED_LIBGCC_SPEC): Make shared libgcc the
26128         default setting.
26130 2013-03-13  Richard Biener  <rguenther@suse.de>
26132         PR tree-optimization/56608
26133         * tree-vect-slp.c (vect_schedule_slp): Do not remove scalar
26134         calls when vectorizing basic-blocks.
26136 2013-03-13  Jakub Jelinek  <jakub@redhat.com>
26138         PR plugins/45078
26139         * config.gcc: On arm, mips, sh and sparc add vxworks-dummy.h to
26140         tm_file.
26142 2013-03-12  Jakub Jelinek  <jakub@redhat.com>
26144         * doc/invoke.texi (-Waddr-space-convert): Move into the table earlier.
26146 2013-03-11  Jan Hubicka  <jh@suse.cz>
26148         PR lto/56557
26149         * lto-streamer-out.c (output_symbol_p): Skip references from
26150         constructors of external variables.
26152 2013-03-11  Jan Hubicka  <jh@suse.cz>
26154         PR middle-end/56571
26155         * valtrack.c (cleanup_auto_inc_dec): Unshare clobbers originating
26156         from pseudos.
26157         * emit-rtl.c (verify_rtx_sharing): Likewise.
26158         (copy_insn_1): Likewise.
26159         * rtl.c (copy_rtx): Likewise.
26161 2013-03-11  Georg-Johann Lay  <avr@gjlay.de>
26163         PR target/56591
26164         * config/avr/avr.c (avr_print_operand): Add space after '%c' in
26165         output_operand_lossage message.
26167 2013-03-11  Richard Earnshaw  <rearnsha@arm.com>
26169         PR target/56470
26170         * arm.c (shift_op): Validate RTL pattern on the fly.
26171         (arm_print_operand, case 'S'): Don't use shift_operator to validate
26172         the RTL.
26174 2013-03-10  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
26176         PR target/56347
26177         * config/pa/pa.md (call_value): Check for calls to powf and direct to
26178         new call patterns that clobber %fr12.
26179         (call_val_powf, call_val_powf_pic, call_val_powf_64bit): New insn,
26180         split and postreload patterns.
26181         * config/pa/pa.c (pa_conditional_register_usage): Revert marking
26182         registers %fr12 and %fr12R as call used.
26184 2013-03-09  Steven Bosscher  <steven@gcc.gnu.org>
26186         * dse.c (delete_dead_store_insn): Respect TDF_DETAILS.
26187         (canon_address, record_store, replace_read, check_mem_read_rtx,
26188         scan_insn, dse_step1, dse_step2_init, dse_step2_spill,
26189         dse_step4, dse_step5_nospill, dse_step5_spill, dse_step6,
26190         rest_of_handle_dse): Likewise.
26192 2013-03-09  Richard Sandiford  <rdsandiford@googlemail.com>
26194         PR middle-end/56524
26195         * tree.h (tree_optimization_option): Rename target_optabs to optabs.
26196         Add base_optabs.
26197         (TREE_OPTIMIZATION_OPTABS): Update after previous field change.
26198         (TREE_OPTIMIZATION_BASE_OPTABS): New macro.
26199         (save_optabs_if_changed): Replace with...
26200         (init_tree_optimization_optabs): ...this.
26201         * optabs.c (save_optabs_if_changed): Rename to...
26202         (init_tree_optimization_optabs): ...this.  Take the optimization node
26203         as argument.  Do nothing if the base optabs are already correct.
26204         Reuse the existing TREE_OPTIMIZATION_OPTABS memory if we need
26205         to recompute optabs.
26206         * function.h (function): Remove optabs field.
26207         * function.c (invoke_set_current_function_hook): Call
26208         init_tree_optimization_optabs.  Use the result to initialize
26209         this_fn_optabs.
26211 2013-02-27  Aldy Hernandez  <aldyh@redhat.com>
26213         * trans-mem.c (expand_transaction): Do not set PR_INSTRUMENTEDCODE
26214         if GTMA_HAS_NO_INSTRUMENTATION.
26215         (generate_tm_state): Keep GTMA_HAS_NO_INSTRUMENTATION bit.
26216         (ipa_tm_transform_transaction): Set GTMA_HAS_NO_INSTRUMENTATION.
26217         * gimple.h (GTMA_HAS_NO_INSTRUMENTATION): Define.
26218         * gimple-pretty-print.c (dump_gimple_transaction): Handle
26219         GTMA_HAS_NO_INSTRUMENTATION.
26221 2013-03-08  Jakub Jelinek  <jakub@redhat.com>
26223         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Don't link against
26224         libasan_preinit.o.
26226 2013-03-08  Marek Polacek  <polacek@redhat.com>
26227             Jakub Jelinek  <jakub@redhat.com>
26229         PR tree-optimization/56478
26230         * predict.c (is_comparison_with_loop_invariant_p): Change the
26231         type of loop_step to tree.
26232         (predict_loops): Adjust.
26233         (predict_iv_comparison): Perform the computations on double_ints.
26235 2013-03-08  Richard Biener  <rguenther@suse.de>
26237         PR tree-optimization/56570
26238         * tree-cfg.c (verify_expr_location_1): Verify locations for
26239         DECL_DEBUG_EXPR.
26240         * tree-sra.c (create_access_replacement): Strip locations
26241         from DECL_DEBUG_EXPRs.
26243 2013-03-08  Richard Biener  <rguenther@suse.de>
26245         * tree-inline.c (expand_call_inline): Do not associate
26246         a BLOCK with the location in BLOCK_SOURCE_LOCATION.
26247         * tree-cfg.c (verify_location): Verify BLOCK_SOURCE_LOCATION.
26249 2013-03-08  Richard Biener  <rguenther@suse.de>
26251         * tree-ssa-ter.c (is_replaceable_p): Do not TER across location
26252         or block changes with -Og.  Fix for location / block encoding
26253         changes and PHI arguments with locations.
26255 2013-03-07  Steven Bosscher  <steven@gcc.gnu.org>
26257         * bitmap.c (struct bitmap_descriptor_d): Use unsigned HOST_WIDEST_INT
26258         for all counters.
26259         (struct output_info): Likewise.
26260         (register_overhead): Remove bad gcc_assert.
26261         (bitmap_find_bit): If there is only a single bitmap element, do not
26262         count a miss as a search.
26263         (print_statistics): Update for counter type changes.
26264         (dump_bitmap_statistics): Likewise.  Print headers such that they
26265         are properly lined up with the printed counters.
26267 2013-03-07  Jakub Jelinek  <jakub@redhat.com>
26269         PR tree-optimization/56559
26270         * tree-ssa-reassoc.c (zero_one_operation): When looking at rhs2,
26271         check that it has only a single use.
26273 2013-03-07  Richard Biener  <rguenther@suse.de>
26275         * doc/invoke.texi (fwhole-program): Discourage use in combination
26276         with -flto.
26278 2013-03-06  Jakub Jelinek  <jakub@redhat.com>
26280         * config/arm/t-arm (TM_H, OPTIONS_H_EXTRA): Add arm-cores.def.
26282         PR tree-optimization/56539
26283         * tree-tailcall.c (adjust_return_value_with_ops): Use GSI_SAME_STMT
26284         instead of GSI_CONTINUE_LINKING as last argument to
26285         force_gimple_operand_gsi.  Adjust function comment.
26287         * config/aarch64/t-aarch64 (TM_H, OPTIONS_H_EXTRA): Add
26288         aarch64-cores.def.
26290         PR middle-end/56548
26291         * expr.c (expand_cond_expr_using_cmove): When expanding cmove in
26292         promoted mode, convert the result back to the original mode.
26294 2013-03-06  Richard Biener  <rguenther@suse.de>
26296         PR middle-end/56294
26297         * tree-into-ssa.c (insert_phi_nodes_for): Add dumping.
26298         (insert_updated_phi_nodes_compare_uids): New function.
26299         (update_ssa): Sort symbols_to_rename after UID before
26300         traversing it to insert PHI nodes.
26302 2013-03-06  Richard Biener  <rguenther@suse.de>
26304         PR middle-end/50494
26305         * tree-vect-data-refs.c (vect_can_force_dr_alignment_p):
26306         Do not adjust alignment of DECL_IN_CONSTANT_POOL decls.
26308         Revert
26309         2013-02-13  Richard Biener  <rguenther@suse.de>
26311         PR lto/50494
26312         * varasm.c (output_constant_def_1): Get the decl representing
26313         the constant as argument.
26314         (output_constant_def): Wrap output_constant_def_1.
26315         (make_decl_rtl): Use output_constant_def_1 with the decl
26316         representing the constant.
26317         (build_constant_desc): Optionally re-use a decl already
26318         representing the constant.
26319         (tree_output_constant_def): Adjust.
26321 2013-03-06  Joey Ye  <joey.ye@arm.com>
26323         PR lto/50293
26324         * gcc.c (convert_white_space): New function.
26325         (main): Handles white space in function name.
26327 2013-03-06  Oleg Endo  <olegendo@gcc.gnu.org>
26329         PR target/56529
26330         * config/sh/sh.c (sh_option_override): Check for TARGET_DYNSHIFT
26331         instead of TARGET_SH2 for call-table case.  Do not set sh_div_strategy
26332         to SH_DIV_CALL_TABLE for TARGET_SH2.
26333         * config.gcc (sh_multilibs): Add m2 and m2a to sh*-*-linux* multilib
26334         list.
26335         * doc/invoke.texi (SH options): Document mdiv= call-div1, call-fp,
26336         call-table options.
26338 2013-03-05  Sterling Augustine  <saugustine@google.com>
26339             Cary Coutant  <ccoutant@google.com>
26341         PR debug/55364
26342         * dwarf2out.c (resolve_addr): Don't call
26343         remove_loc_list_addr_table_entries a second time for the same
26344         expression.
26346 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26348         PR debug/56510
26349         * cfgexpand.c (expand_debug_parm_decl): Call copy_rtx on incoming.
26350         (avoid_complex_debug_insns): New function.
26351         (expand_debug_locations): Call it.
26353         PR rtl-optimization/56484
26354         * ifcvt.c (noce_process_if_block): If else_bb is NULL, avoid extending
26355         lifetimes of hard registers on small register class machines.
26357 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26359         * config/microblaze/microblaze-protos.h: Rename
26360         microblaze_is_interrupt_handler to microblaze_is_interrupt_variant.
26361         * config/microblaze/microblaze.c (microblaze_attribute_table): Add
26362         fast_interrupt.
26363         (microblaze_fast_interrupt_function_p): New function.
26364         (microblaze_is_interrupt_handler): Rename to
26365         microblaze_is_interrupt_variant and add fast_interrupt check.
26366         (microblaze_must_save_register): Use microblaze_is_interrupt_variant.
26367         (save_restore_insns): Likewise.
26368         (compute_frame_size): Likewise.
26369         (microblaze_function_prologue): Add FAST_INTERRUPT_NAME.
26370         (microblaze_globalize_label): Likewise.
26371         * config/microblaze/microblaze.h: Define FAST_INTERRUPT_NAME.
26372         * config/microblaze/microblaze.md: Use wrapper
26373         microblaze_is_interrupt_variant.
26375 2013-03-05  Kai Tietz  <ktietz@redhat.com>
26377         * sdbout.c (sdbout_one_type): Switch to current function's section
26378         supporting cold/hot.
26380 2013-03-05  David Holsgrove  <david.holsgrove@xilinx.com>
26382         * doc/invoke.texi (MicroBlaze): Add -mbig-endian, -mlittle-endian,
26383         -mxl-reorder.
26385 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26387         PR middle-end/56461
26388         * ggc-common.c (gt_pch_save): For ENABLE_VALGRIND_CHECKING,
26389         if VALGRIND_GET_VBITS is defined, temporarily make object
26390         memory all defined, and restore previous valgrind addressability
26391         and definability afterwards.  Free this_object at the end.
26393         PR middle-end/56461
26394         * lra.c (lra): Call lra_clear_live_ranges if live_p,
26395         right before calling lra_create_live_ranges, also call it
26396         when clearing live_p.  Only call lra_clear_live_ranges
26397         at the end if live_p.
26399         PR middle-end/56461
26400         * sched-deps.c (delete_dep_node): Free DEP_REPLACE.
26402 2013-03-05  Richard Biener  <rguenther@suse.de>
26404         PR tree-optimization/56521
26405         * tree-ssa-sccvn.c (set_value_id_for_result): Always initialize
26406         value-id.
26408 2013-03-05  Steven Bosscher  <steven@gcc.gnu.org>
26410         PR c++/55135
26411         * except.h (remove_unreachable_eh_regions): New prototype.
26412         * except.c (remove_eh_handler_splicer): New function, split out
26413         of remove_eh_handler.
26414         (remove_eh_handler): Use remove_eh_handler_splicer.  Add comment
26415         warning about running it on many EH regions one at a time.
26416         (remove_unreachable_eh_regions_worker): New function, walk the
26417         EH tree in depth-first order and remove non-marked regions.
26418         (remove_unreachable_eh_regions): New function.
26419         * tree-eh.c (mark_reachable_handlers): New function, split out
26420         from remove_unreachable_handlers.
26421         (remove_unreachable_handlers): Use mark_reachable_handlers and
26422         remove_unreachable_eh_regions.
26423         (remove_unreachable_handlers_no_lp): Use mark_reachable_handlers
26424         and remove_unreachable_eh_regions.
26426 2013-03-05  Richard Biener  <rguenther@suse.de>
26428         PR middle-end/56525
26429         * loop-init.c (fix_loop_structure): Remove loops in two stages,
26430         not freeing them until the end.
26432 2013-03-05  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
26434         * config/s390/s390.h: Define DWARF2_ASM_LINE_DEBUG_INFO.
26436 2013-03-05  Richard Biener  <rguenther@suse.de>
26438         PR tree-optimization/56270
26439         * tree-vect-slp.c (vect_schedule_slp): Clear vectorized stmts
26440         of loads after scheduling an SLP instance.
26442 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26444         * Makefile.in (dg_target_exps): Add aarch64.exp, epiphany.exp and
26445         tic6x.exp.
26446         (check_gcc_parallelize): Run guality.exp as a separate job from
26447         vect.exp with unsorted.exp and $(dg_target_exps) separately from
26448         struct-layout-1.exp with stackalign.exp.
26450         * alias.c (init_alias_analysis): Clear reg_known_equiv_p bitmap.
26452         PR middle-end/56461
26453         * tree-vect-slp.c (vect_supported_load_permutation_p): Free
26454         load_index sbitmap even if some bit in it isn't set.
26456         PR middle-end/56461
26457         * tree-ssa-loop-niter.c (bb_queue): Remove typedef.
26458         (discover_iteration_bound_by_body_walk): Change queues to
26459         vec<vec<basic_block> > and queue to vec<basic_block>.  Fix up
26460         spelling in comment.  Call safe_push on queues[bound_index] directly.
26461         Release queues[queue_index] in every iteration unconditionally.
26462         Release bounds vector.
26464         PR middle-end/56461
26465         * tree-vect-stmts.c (free_stmt_vec_info_vec): Call
26466         free_stmt_vec_info on any left-over stmt_vec_info in the vector.
26467         * tree-vect-loop.c (vect_create_epilog_for_reduction): Release
26468         inner_phis vector.
26470 2013-03-05  Richard Biener  <rguenther@suse.de>
26472         PR lto/56515
26473         * tree-inline.c (remap_blocks_to_null): New function.
26474         (expand_call_inline): When expanding a call stmt without
26475         an associated block inline remap all callee blocks to NULL.
26477 2013-03-05  Jakub Jelinek  <jakub@redhat.com>
26479         PR rtl-optimization/56494
26480         * simplify-rtx.c (simplify_truncation): If C is narrower than A,
26481         optimize (truncate:A (subreg:B (truncate:C X) 0)) into
26482         (subreg:A (truncate:C X) 0) instead of (truncate:A X).
26484         PR middle-end/56461
26485         * sel-sched-ir.c (free_sched_pools): Release
26486         succs_info_pool.stack[succs_info_pool.max_top] vectors too
26487         if succs_info_pool.max_top isn't -1.
26489         PR bootstrap/56509
26490         * opts.c (opts_obstack, opts_concat): Moved to...
26491         * opts-common.c (opts_obstack, opts_concat): ... here.
26493 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26495         PR middle-end/56461
26496         * diagnostic.c (diagnostic_append_note): Save and restore old prefix.
26498 2013-03-04  Martin Jambor  <mjambor@suse.cz>
26500         * tree-dfa.c (get_or_create_ssa_default_def): Use parameter fn in
26501         all appropriate places.
26503 2013-01-04  Eric Botcazou  <ebotcazou@adacore.com>
26505         PR tree-optimization/56424
26506         * ipa-split.c (split_function): Do not set the RSO flag if result is
26507         not by reference and its type is a register type.
26509 2013-03-04  David Holsgrove  <david.holsgrove@xilinx.com>
26511         * config/microblaze/microblaze.c (microblaze_valid_pic_const): New
26512         (microblaze_legitimate_pic_operand): Likewise
26513         * config/microblaze/microblaze.h (LEGITIMATE_PIC_OPERAND_P): calls
26514         new function microblaze_legitimate_pic_operand
26515         * config/microblaze/microblaze-protos.h
26516         (microblaze_legitimate_pic_operand): Declare.
26518 2013-03-04  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26520         * config/microblaze/predicates.md (call_insn_simple_operand):
26521         New predicate for supported rtx code types.
26522         * config/microblaze/microblaze.md (call_internal1): Use
26523         call_insn_simple_operand predicate.
26525 2013-03-04  Jakub Jelinek  <jakub@redhat.com>
26527         PR middle-end/56461
26528         * tree-loop-distribution.c (ldist_gen): Call partition_free after each
26529         partitions.ordered_remove.
26531         PR middle-end/56461
26532         * tree-vect-stmts.c (vectorizable_conversion): Don't call
26533         vec_oprnds0.create (1) for modifier == NONE.
26535         PR middle-end/56461
26536         * tree-vect-stmts.c (vectorizable_shift): Don't call create methods
26537         on vec_oprnds0 or vec_oprnds1 before loop, only call it on
26538         vec_oprnds1 right before pushing anything to it for
26539         scalar_shift_arg.
26541         PR middle-end/56461
26542         * tree-vect-loop.c (destroy_loop_vec_info): For !clean_stmts, just
26543         set nbbs to 0 instead of having separate code path.
26544         (vect_analyze_loop_form): Call destroy_loop_vec_info with true
26545         instead of false as last argument if returning NULL.
26547 2013-03-03  Sandra Loosemore  <sandra@codesourcery.com>
26549         * target.def (TARGET_OPTION_VALID_ATTRIBUTE_P): Update comments;
26550         the attribute is now called "target" instead of "option".
26551         (TARGET_OPTION_PRAGMA_PARSE): Likewise, for the pragma.
26552         * doc/tm.texi.in (Target Attributes):  Likewise document the correct
26553         attribute/pragma name for TARGET_OPTION_VALID_P and
26554         TARGET_OPTION_PRAGMA_PARSE.  Also copy-edit and correct markup.
26555         * doc/tm.texi: Regenerated.
26557 2013-03-02  David Holsgrove  <david.holsgrove@xilinx.com>
26559         * config/microblaze/microblaze.c:
26560         Check mcpu, pcmp requirement and set TARGET_REORDER to 0 if not met.
26561         * config/microblaze/microblaze.h: Add -mxl-reorder to
26562         DRIVER_SELF_SPECS.
26563         * config/microblaze/microblaze.md: New bswapsi2 and bswaphi2.
26564         instructions emitted if TARGET_REORDER.
26565         * config/microblaze/microblaze.opt: New option -mxl-reorder set to 1
26566         or 0 for -m/-mno case, but initialises as 2 to detect default use case
26567         separately.
26569 2013-03-01  Xinliang David Li  <davidxl@google.com>
26571         * tree-ssa-uninit.c (compute_control_dep_chain): Limit post-dom
26572         walk length.
26574 2013-03-01  Jakub Jelinek  <jakub@redhat.com>
26576         PR middle-end/56461
26577         * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Release path
26578         vector even when returning true.  Fix up function comment formatting.
26580         PR middle-end/56461
26581         * ira-build.c (ira_loop_nodes_count): New variable.
26582         (create_loop_tree_nodes): Initialize it.
26583         (finish_loop_tree_nodes): Use it instead of looking at current_loops.
26585         PR middle-end/56461
26586         * tree-vect-data-refs.c (vect_permute_store_chain): Avoid using copy
26587         method on dr_chain and result_chain.
26588         * tree-vect-stmts.c (vectorizable_store): Only call
26589         result_chain.create if j == 0.
26591         PR middle-end/56461
26592         * tree-vect-stmts.c (vect_create_vectorized_promotion_stmts): Call
26593         vec_oprnds0->release (); rather than vec_oprnds0->truncate (0)
26594         before overwriting it.
26596 2013-03-01  Tobias Burnus  <burnus@net-b.de>
26598         * doc/extended.texi (C Extensions): Change order in @menu
26599         to match @node.
26600         (Other MIPS Built-in Functions): Move last MIPS entry before
26601         "picoChip Built-in Functions".
26602         (SH Built-in Functions): Move after RX Built-in Functions.
26603         * doc/gcc.texi (Introduction): Change order in @menu
26604         to match @node.
26605         * doc/md.texi (Constraints): Ditto.
26606         * gty.texi (Type Information): Ditto.
26607         (User-provided marking routines for template types): Make
26608         subsection.
26609         * doc/invoke.texi (AArch64 Options): Move before
26610         "Adapteva Epiphany Options".
26612 2013-02-28  Konstantin Serebryany  <konstantin.s.serebryany@gmail.com>
26613             Jakub Jelinek  <jakub@redhat.com>
26615         PR sanitizer/56454
26616         * asan.c (gate_asan): Lookup no_sanitize_address instead of
26617         no_address_safety_analysis attribute.
26618         * doc/extend.texi (no_address_safety_attribute): Rename to
26619         no_sanitize_address attribute, mention no_address_safety_analysis
26620         attribute as deprecated alias.
26622 2013-02-28  Jakub Jelinek  <jakub@redhat.com>
26624         PR middle-end/56461
26625         * tree-vectorizer.h (vect_get_slp_defs): Change 3rd argument
26626         type to vec<vec<tree> > *.
26627         * tree-vect-slp.c (vect_get_slp_defs): Likewise.  Change vec_defs
26628         to be vec<tree> instead of vec<tree> *, set vec_defs
26629         to vNULL and call vec_defs.create (number_of_vects), adjust other
26630         uses of vec_defs.
26631         * tree-vect-stmts.c (vect_get_vec_defs, vectorizable_call,
26632         vectorizable_condition): Adjust vect_get_slp_defs callers.
26634 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26636         * config/aarch64/aarch64.c
26637         (aarch64_float_const_representable): Remove unused variable.
26639 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26641         * config/aarch64/aarch64.c (aarch64_mangle_type): Make static.
26643 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26645         * config/aarch64/aarch64-builtins.c
26646         (aarch64_init_simd_builtins): Make static.
26648 2013-02-28  James Greenhalgh  <james.greenhalgh@arm.com>
26650         * config/aarch64/aarch64.c
26651         (aarch64_simd_make_constant): Make static.
26653 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26655         * tree-sra.c (load_assign_lhs_subreplacements): Do not put replacements
26656         with no initialization to the RHS of debug statements.
26658 2013-02-28  Martin Jambor  <mjambor@suse.cz>
26660         PR tree-optimization/56294
26661         * tree-sra.c (analyze_access_subtree): Create replacement declarations.
26662         Adjust dumping.
26663         (get_access_replacement): Do not call create_access_replacement.
26664         Assert a replacement exists.
26665         (get_repl_default_def_ssa_name): Create the replacement declaration
26666         itself.
26668 2013-02-28  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
26670         * config/arm/arm.c (arm_output_mi_thunk): Call final_start_function and
26671         final_end_function.
26673 2013-02-28  Marek Polacek  <polacek@redhat.com>
26675         PR rtl-optimization/56466
26676         * loop-unroll.c (unroll_and_peel_loops): Call fix_loop_structure
26677         if we're changing a loop.
26678         (peel_loops_completely): Likewise.
26680 2013-02-28  Paolo Carlini  <paolo.carlini@oracle.com>
26682         PR c++/55813
26683         * doc/invoke.texi ([-Wctor-dtor-privacy]): Complete.
26685 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26687         PR target/56445
26688         * config/avr/avr.c (avr_init_builtins): Use 'n' instead of empty
26689         macro parameters with: FX_FTYPE_FX, FX_FTYPE_FX_INT, INT_FTYPE_FX,
26690         INTX_FTYPE_FX, FX_FTYPE_INTX.
26691         * config/avr/builtins.def: Adjust respective DEF_BUILTIN.
26693 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26695         * avr/avr-mcus.def (ata5272, ata5505, attiny1634, ata6285)
26696         (ata6286, atmega8a, atmega48pa, ata5790, ata5790n, ata5795)
26697         (atmega164pa, atmega165pa, atmega168pa, atmega16hva, atmega16hvb)
26698         (atmega16hvbrevb, atmega16m1, atmega16u4, atmega26hvg, atmega32a)
26699         (atmega32a, atmega3250pa, atmega3290pa, atmega32c1, atmega32m1)
26700         (atmega32u4, atmega32u6, atmega64a, atmega6490a, atmega6490p)
26701         (atmega64c1, atmega64m1, atmega64rfa2, atmega64rfr2, atmega32hvb)
26702         (atmega32hvbrevb, atmega16hva2, atmega48hvf, at90pwm161)
26703         (atmega128a, atmega1284, atmxt112sl, atmxt224, atmxt224e)
26704         (atmxt336s, atxmega16a4u, atxmega16c4, atxmega32a4u, atxmega32c4)
26705         (atxmega32e5, atxmega64a3u, atxmega64a4u, atxmega64b1, atxmega64b3)
26706         (atxmega64c3, atxmega64d4, atxmega128a3u, atxmega128b1)
26707         (atxmega128b3, atxmega128c3, atxmega128d4, atmxt540s, atmxt540sreva)
26708         (atxmega192a3u, atxmega192c3, atxmega256a3u, atxmega256c3)
26709         (atxmega384c3, atxmega384d3, atxmega128a4u): New AVR_MCU.
26710         (avrxmega6): Increase max flash segments from 5 to 6.
26711         * config/avr/t-multilib: Regenerate.
26712         * config/avr/avr-tables.opt: Regenerate.
26713         * doc/avr-mmcu.texi: Regenerate.
26715 2013-02-28  Georg-Johann Lay  <avr@gjlay.de>
26717         * config/avr/avr.h (device_to_arch): Rename to device_to_ld.
26718         (avr_device_to_arch): Rename to avr_device_to_ld.
26719         (avr_device_to_as): New prototype.
26720         (EXTRA_SPEC_FUNCTIONS): Add device_to_as.
26721         (ASM_SPEC): Use device_to_as to get -mmcu= and -mno-skip-bug=.
26722         * config/avr/driver-avr.c (avr_device_to_as): New.
26723         (avr_device_to_arch): Rename to avr_device_to_ld.
26725 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26727         PR middle-end/56461
26728         * tree-vect-data-refs.c (vect_permute_load_chain): Avoid using copy
26729         method on dr_chain and result_chain.
26731         PR middle-end/56461
26732         * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Call
26733         pointer_set_destroy on not_executed_last_iteration.
26735         PR middle-end/56461
26736         * tree-vect-loop.c (vectorizable_reduction): Release vect_defs vector.
26738         PR middle-end/56461
26739         * ipa-pure-const.c (propagate): Use FOR_EACH_FUNCTION instead of
26740         FOR_EACH_DEFINED_FUNCTION when freeing state.
26742         PR middle-end/56461
26743         * df-scan.c (df_insn_delete): Use df_scan_free_mws_vec before
26744         pool_free.
26745         (df_insn_rescan_debug_internal): Use df_scan_free_mws_vec before
26746         overwriting it.
26748         PR middle-end/56461
26749         * ipa-cp.c (decide_whether_version_node): Call vec_free on
26750         known_aggs[i].items and release known_aggs vector.
26752         PR middle-end/56461
26753         * ipa-reference.c (propagate): Free node_info even for alias nodes.
26755 2013-02-27  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
26757         * config/microblaze/microblaze.c (microblaze_emit_compare):
26758         Use xor for EQ/NE comparisions.
26759         * config/microblaze/microblaze.md (cstoresf4): Add constraints
26760         (cbranchsf4): Adjust operator to comparison_operator.
26762 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26764         PR middle-end/56461
26765         * tree-flow.h (edge_var_map_vector): Change into va_heap, vl_embed
26766         vector.
26767         * tree-ssa.c (redirect_edge_var_map_add): Use vec_safe_reserve and
26768         vec_safe_push, always update *slot.
26769         (redirect_edge_var_map_clear): Use vec_free.
26770         (redirect_edge_var_map_dup): Use vec_safe_copy and vec_safe_reserve.
26771         (free_var_map_entry): Use vec_free.
26772         * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Use
26773         FOR_EACH_VEC_SAFE_ELT instead of FOR_EACH_VEC_ELT.
26775 2013-02-27  Andrey Belevantsev  <abel@ispras.ru>
26777         PR middle-end/45472
26778         * sel-sched-ir.c (merge_expr): Also change vinsn of merged expr
26779         when the may_trap_p bit of the exprs being merged differs.
26780         Reorder tests for speculativeness in the logical and operator.
26782 2013-02-27  Jakub Jelinek  <jakub@redhat.com>
26784         * incpath.c (add_standard_paths): Use reconcat instead of concat
26785         where appropriate and avoid leaking memory.
26787         * opts.h: Include obstack.h.
26788         (opts_concat): New prototype.
26789         (opts_obstack): New declaration.
26790         * opts.c (opts_concat): New function.
26791         (opts_obstack): New variable.
26792         (init_options_struct): Call gcc_init_obstack on opts_obstack.
26793         (finish_options): Use opts_concat instead of concat
26794         and XOBNEWVEC instead of XNEWVEC.
26795         * opts-common.c (generate_canonical_option, decode_cmdline_option,
26796         generate_option): Likewise.
26797         * Makefile.in (OPTS_H): Depend on $(OBSTACK_H).
26798         * lto-wrapper.c (main): Call gcc_init_obstack on opts_obstack.
26800         PR target/56455
26801         * stmt.c (expand_switch_as_decision_tree_p): If flag_pic
26802         and ASM_OUTPUT_ADDR_DIFF_ELT isn't defined, return true.
26804 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26806         PR middle-end/56461
26807         * lra-spills.c (lra_spill): Free spill_hard_reg at the end.
26809 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26811         * config/arm/arm.c (const_ok_for_dimode_op): Back out last change.
26812         (arm_block_move_unaligned_straight): Likewise.
26813         (arm_adjust_block_mem): Likewise.
26815 2013-02-26  Joern Rennecke  <joern.rennecke@embecosm.com>
26817         PR target/48901
26818         * config/lm32/lm32.c (gen_int_relational): Remove unused variables
26819         temp, cond and label.
26820         * config/lm32/lm32.md (ashlsi3): Remove unused variable one.
26822         PR target/52500
26823         * config/c6x/c6x.c (dbx_register_map): Change to unsigned.
26824         * config/c6x/c6x.h (dbx_register_map): Update declaration.
26826         PR target/52501
26827         * config/cr16/cr16-protos.h: Move end of RTX_CODE guard below end
26828         of prologue/epilogue functions.
26830         PR target/52550
26831         * config/tilegx/tilegx.c (tilegx_expand_prologue):
26832         Remove unused variable cfa_offset.
26833         * config/tilepro/tilepro.c (tilepro_expand_prologue): Likewise.
26835         PR target/54639
26836         * config/mn10300/mn10300.c (mn10300_expand_epilogue): Avoid offset
26837         type promotion to unsigned.
26839         PR target/54640
26840         * config/arm/arm.c (const_ok_for_dimode_op): Make code consistent
26841         for HOST_WIDE_INT of 32 bit / same size as int.
26842         (arm_block_move_unaligned_straight): Likewise.
26843         (arm_adjust_block_mem): Likewise.
26845         PR target/54662
26846         * config/mep/t-mep (mep-pragma.o): Use ALL_COMPILERFLAGS instead of
26847         ALL_CFLAGS.
26849 2013-02-26  Marek Polacek  <polacek@redhat.com>
26851         PR tree-optimization/56426
26852         * tree-ssa-loop.c (tree_ssa_loop_init): Always call scev_initialize.
26854 2013-02-26  Richard Biener  <rguenther@suse.de>
26856         PR target/56444
26857         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc): Remove
26858         unused variable loops.
26860 2013-02-26  Jakub Jelinek  <jakub@redhat.com>
26862         PR tree-optimization/56448
26863         * fold-const.c (operand_equal_p) <case tcc_reference>: Don't look at
26864         TREE_SIDE_EFFECTS if flags contain OEP_CONSTANT_ADDRESS_OF.
26865         Clear OEP_CONSTANT_ADDRESS_OF from flags before recursing on second or
26866         later operands of the references, or even first operand for
26867         INDIRECT_REF, TARGET_MEM_REF or MEM_REF.
26869         PR tree-optimization/56443
26870         * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): For
26871         overaligned types, pass TYPE_UNSIGNED (scalar_type) as second argument
26872         to type_for_mode langhook.
26874 2013-02-25  Matt Turner  <mattst88@gmail.com>
26876         * doc/invoke.texi: Document r4700.
26878 2013-02-25  Richard Biener  <rguenther@suse.de>
26880         PR tree-optimization/56175
26881         * tree-ssa-forwprop.c (hoist_conversion_for_bitop_p): New predicate,
26882         split out from ...
26883         (simplify_bitwise_binary): ... here.  Also guard the conversion
26884         of (type) X op CST to (type) (X op ((type-x) CST)) with it.
26886 2013-02-25  Catherine Moore  <clm@codesourcery.com>
26888         Revert:
26889         2013-02-24  Catherine Moore  <clm@codesourcery.com>
26890             Maciej W. Rozycki  <macro@codesourcery.com>
26891             Tom de Vries  <tom@codesourcery.com>
26892             Nathan Sidwell  <nathan@codesourcery.com>
26893             Iain Sandoe  <iain@codesourcery.com>
26894             Nathan Froyd  <froydnj@codesourcery.com>
26895             Chao-ying Fu  <fu@mips.com>
26897         * doc/extend.texi: (micromips, nomicromips, nocompression):
26898         Document new function attributes.
26899         * doc/invoke.texi (minterlink-compressed, mmicromips,
26900         m14k, m14ke, m14kec): Document new options.
26901         (minterlink-mips16): Update documentation.
26902         * doc/md.texi (ZC, ZD): Document new constraints.
26903         * configure.ac (gcc_cv_as_micromips): Check if linker
26904         supports the .set micromips directive.
26905         * configure: Regenerate.
26906         * config.in: Regenerate.
26907         * config/mips/mips-tables.opt: Regenerate.
26908         * config/mips/micromips.md: New file.
26909         * constraints.md (ZC, AD): New constraints.
26910         * config/mips/predicates.md (movep_src_register): New predicate.
26911         (movep_src_operand): New predicate.
26912         (non_volatile_mem_operand): New predicate.
26913         * config/mips/mips.md (multimem): New type.
26914         (length): Differentiate between 17-bit and 18-bit branch offsets.
26915         (MOVEP1, MOVEP2): New mode iterator.
26916         (mov_<load>l): Use ZC constraint.
26917         (mov_<load>r): Likewise.
26918         (mov_<store>l): Likewise.
26919         (mov_<store>r): Likewise.
26920         (*branch_equality<mode>_inverted): Add microMIPS support.
26921         (*branch_equality<mode>): Likewise.
26922         (*jump_absolute): Likewise.
26923         (indirect_jump_<mode>): Likewise.
26924         (tablejump_<mode>): Likewise.
26925         (<optab>_internal): Likewise.
26926         (sibcall_internal): Likewise.
26927         (sibcall_value_internal): Likewise.
26928         (prefetch): Use constraint ZD.
26929         * config/mips/mips.opt (minterlink-compressed): New option.
26930         (minterlink-mips16): Now an alias for minterlink-compressed.
26931         (mmicromips): New option.
26932         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
26933         (compare_and_swap_12): Likewise.
26934         (sync_add<mode>): Likewise.
26935         (sync_<optab>_12): Likewise.
26936         (sync_old_<optab>_12): Likewise.
26937         (sync_new_<optab>_12): Likewise.
26938         (sync_nand_12): Likewise.
26939         (sync_old_nand_12): Likewise.
26940         (sync_new_nand_12): Likewise.
26941         (sync_sub<mode>): Likewise.
26942         (sync_old_add<mode>): Likewise.
26943         (sync_old_sub<mode>): Likewise.
26944         (sync_new_add<mode>): Likewise.
26945         (sync_new_sub<mode>): Likewise.
26946         (sync_<optab><mode>): Likewise.
26947         (sync_old_<optab><mode>): Likewise.
26948         (sync_new_<optab><mode>): Likewise.
26949         (sync_nand<mode>): Likewise.
26950         (sync_old_nand<mode>): Likewise.
26951         (sync_new_nand<mode>): Likewise.
26952         (sync_lock_test_and_set<mode>): Likewise.
26953         (test_and_set_12): Likewise.
26954         (atomic_compare_and_swap<mode>): Likewise.
26955         (atomic_exchange<mode>_llsc): Likewise.
26956         (atomic_fetch_add<mode>_llsc): Likewise.
26957         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
26958         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
26959         (umips_save_restore_pattern_p): Likewise.
26960         (umips_load_store_pair_p): Likewise.
26961         (umips_output_load_store_pair): Likewise.
26962         (umips_movep_target_p): Likewise.
26963         (umips_12bit_offset_address_p): Likewise.
26964         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
26965         (mips_base_mips16): Rename this...
26966         (mips_base_compression_flags): ...to this. Update all uses.
26967         (mips_attribute_table): Add micromips, nomicromips and nocompression.
26968         (mips_mips16_decl_p): Delete.
26969         (mips_nomips16_decl_p): Delete.
26970         (mips_get_compress_on_flags): New function.
26971         (mips_get_compress_off_flags): New function.
26972         (mips_get_compress_mode): New function.
26973         (mips_get_compress_on_name): New function.
26974         (mips_get_compress_off_name): New function.
26975         (mips_insert_attributes): Support multiple compression types.
26976         (mips_merge_decl_attributes): Likewise.
26977         (umips_12bit_offset_address_p): New function.
26978         (mips_start_function_definition): Emit .set micromips directive.
26979         (mips_call_may_need_jalx_p): New function.
26980         (mips_function_ok_for_sibcall): Add microMIPS support.
26981         (mips_print_operand_punctuation): Support short delay slots and
26982         compact jumps.
26983         (umips_swm_mask, umips_swm_encoding): New.
26984         (umips_build_save_restore): New function.
26985         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
26986         (was_mips16_p): Remove.
26987         (old_compression_mode): New.
26988         (mips_set_compression_mode): New function.
26989         (mips_set_current_function): Add microMIPS support.
26990         (mips_option_override): Likewise.
26991         (umips_save_restore_pattern_p): New function.
26992         (umips_output_save_restore): New function.
26993         (umips_load_store_pair_p_1): New function.
26994         (umips_load_store_pair_p): New function.
26995         (umips_output_load_store_pair_1): New function.
26996         (umips_output_load_store_pair): New function.
26997         (umips_movep_target_p) New function.
26998         (mips_prepare_pch_save): Add microMIPS support.
26999         * config/mips/mips.h (TARGET_COMPRESSION): New.
27000         (TARGET_CPU_CPP_BUILTINS): Update macro
27001         to use new compression flags and to support microMIPS.
27002         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
27003         (MIPS_ARCH_FLOAT_SPEC): Likewise.
27004         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
27005         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
27006         (ASM_SPEC): Support mmicromips and mno-micromips.
27007         (M16STORE_REG_P): New macro.
27008         (MIPS_CALL): Support TARGET_MICROMIPS.
27009         (MICROMIPS_J): New macro.
27010         (mips_base_mips16): Rename this...
27011         (mips_base_compression_flags): ...to this.
27012         (UMIPS_12BIT_OFFSET_P): New macro.
27013         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
27014         (MULTILIB_DIRNAMES): Likewise.
27016 2013-02-25  Tom de Vries  <tom@codesourcery.com>
27018         PR rtl-optimization/56131
27019         * insn-notes.def (INSN_NOTE_BASIC_BLOCK): Update comment.
27020         * cfgrtl.c (delete_insn): Don't reorder NOTE_INSN_DELETED_LABEL and
27021         NOTE_INSN_BASIC_BLOCK if BLOCK_FOR_INSN == NULL.
27023 2013-02-25  Tobias Burnus  <burnus@net-b.de>
27025         * doc/invoke.texi (-fsanitize=): Move from optimization
27026         to debugging options.
27028 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
27030         * sched-deps.c (sched_analyze_insn): Fix typo in comment.
27032 2013-02-25  Andrey Belevantsev  <abel@ispras.ru>
27033             Alexander Monakov  <amonakov@ispras.ru>
27035         PR middle-end/56077
27036         * sched-deps.c (sched_analyze_insn): When reg_pending_barrier,
27037         flush pending lists also on non-jumps.  Adjust comment.
27039 2013-02-24  Catherine Moore  <clm@codesourcery.com>
27040             Maciej W. Rozycki  <macro@codesourcery.com>
27041             Tom de Vries  <tom@codesourcery.com>
27042             Nathan Sidwell  <nathan@codesourcery.com>
27043             Iain Sandoe  <iain@codesourcery.com>
27044             Nathan Froyd  <froydnj@codesourcery.com>
27045             Chao-ying Fu  <fu@mips.com>
27047         * doc/extend.texi: (micromips, nomicromips, nocompression):
27048         Document new function attributes.
27049         * doc/invoke.texi (minterlink-compressed, mmicromips,
27050         m14k, m14ke, m14kec): Document new options.
27051         (minterlink-mips16): Update documentation.
27052         * doc/md.texi (ZC, ZD): Document new constraints.
27053         * configure.ac (gcc_cv_as_micromips): Check if linker
27054         supports the .set micromips directive.
27055         * configure: Regenerate.
27056         * config.in: Regenerate.
27057         * config/mips/mips-tables.opt: Regenerate.
27058         * config/mips/micromips.md: New file.
27059         * constraints.md (ZC, AD): New constraints.
27060         * config/mips/predicates.md (movep_src_register): New predicate.
27061         (movep_src_operand): New predicate.
27062         (non_volatile_mem_operand): New predicate.
27063         * config/mips/mips.md (multimem): New type.
27064         (length): Differentiate between 17-bit and 18-bit branch offsets.
27065         (MOVEP1, MOVEP2): New mode iterator.
27066         (mov_<load>l): Use ZC constraint.
27067         (mov_<load>r): Likewise.
27068         (mov_<store>l): Likewise.
27069         (mov_<store>r): Likewise.
27070         (*branch_equality<mode>_inverted): Add microMIPS support.
27071         (*branch_equality<mode>): Likewise.
27072         (*jump_absolute): Likewise.
27073         (indirect_jump_<mode>): Likewise.
27074         (tablejump_<mode>): Likewise.
27075         (<optab>_internal): Likewise.
27076         (sibcall_internal): Likewise.
27077         (sibcall_value_internal): Likewise.
27078         (prefetch): Use constraint ZD.
27079         * config/mips/mips.opt (minterlink-compressed): New option.
27080         (minterlink-mips16): Now an alias for minterlink-compressed.
27081         (mmicromips): New option.
27082         * config/mips/sync.md (sync_compare_and_swap<mode>): Use ZR constraint.
27083         (compare_and_swap_12): Likewise.
27084         (sync_add<mode>): Likewise.
27085         (sync_<optab>_12): Likewise.
27086         (sync_old_<optab>_12): Likewise.
27087         (sync_new_<optab>_12): Likewise.
27088         (sync_nand_12): Likewise.
27089         (sync_old_nand_12): Likewise.
27090         (sync_new_nand_12): Likewise.
27091         (sync_sub<mode>): Likewise.
27092         (sync_old_add<mode>): Likewise.
27093         (sync_old_sub<mode>): Likewise.
27094         (sync_new_add<mode>): Likewise.
27095         (sync_new_sub<mode>): Likewise.
27096         (sync_<optab><mode>): Likewise.
27097         (sync_old_<optab><mode>): Likewise.
27098         (sync_new_<optab><mode>): Likewise.
27099         (sync_nand<mode>): Likewise.
27100         (sync_old_nand<mode>): Likewise.
27101         (sync_new_nand<mode>): Likewise.
27102         (sync_lock_test_and_set<mode>): Likewise.
27103         (test_and_set_12): Likewise.
27104         (atomic_compare_and_swap<mode>): Likewise.
27105         (atomic_exchange<mode>_llsc): Likewise.
27106         (atomic_fetch_add<mode>_llsc): Likewise.
27107         * config/mips/mips-cpus.def (m14kc, m14k): New processors.
27108         * config/mips/mips-protos.h (umips_output_save_restore): New prototype.
27109         (umips_save_restore_pattern_p): Likewise.
27110         (umips_load_store_pair_p): Likewise.
27111         (umips_output_load_store_pair): Likewise.
27112         (umips_movep_target_p): Likewise.
27113         (umips_12bit_offset_address_p): Likewise.
27114         * config/mips/mips.c (MIPS_MAX_FIRST_STEP): Update for microMIPS.
27115         (mips_base_mips16): Rename this...
27116         (mips_base_compression_flags): ...to this. Update all uses.
27117         (mips_attribute_table): Add micromips, nomicromips and nocompression.
27118         (mips_mips16_decl_p): Delete.
27119         (mips_nomips16_decl_p): Delete.
27120         (mips_get_compress_on_flags): New function.
27121         (mips_get_compress_off_flags): New function.
27122         (mips_get_compress_mode): New function.
27123         (mips_get_compress_on_name): New function.
27124         (mips_get_compress_off_name): New function.
27125         (mips_insert_attributes): Support multiple compression types.
27126         (mips_merge_decl_attributes): Likewise.
27127         (umips_12bit_offset_address_p): New function.
27128         (mips_start_function_definition): Emit .set micromips directive.
27129         (mips_call_may_need_jalx_p): New function.
27130         (mips_function_ok_for_sibcall): Add microMIPS support.
27131         (mips_print_operand_punctuation): Support short delay slots and
27132         compact jumps.
27133         (umips_swm_mask, umips_swm_encoding): New.
27134         (umips_build_save_restore): New function.
27135         (mips_for_each_saved_gpr_and_fpr): Add microMIPS support.
27136         (was_mips16_p): Remove.
27137         (old_compression_mode): New.
27138         (mips_set_compression_mode): New function.
27139         (mips_set_current_function): Add microMIPS support.
27140         (mips_option_override): Likewise.
27141         (umips_save_restore_pattern_p): New function.
27142         (umips_output_save_restore): New function.
27143         (umips_load_store_pair_p_1): New function.
27144         (umips_load_store_pair_p): New function.
27145         (umips_output_load_store_pair_1): New function.
27146         (umips_output_load_store_pair): New function.
27147         (umips_movep_target_p) New function.
27148         (mips_prepare_pch_save): Add microMIPS support.
27149         * config/mips/mips.h (TARGET_COMPRESSION): New.
27150         (TARGET_CPU_CPP_BUILTINS): Update macro
27151         to use new compression flags and to support microMIPS.
27152         (MIPS_ISA_LEVEL_SPEC): Add m14k processors.
27153         (MIPS_ARCH_FLOAT_SPEC): Likewise.
27154         (ISA_HAS_LWXS): Include TARGET_MICROMIPS.
27155         (ISA_HAS_LOAD_DELAY): Exclude TARGET_MICROMIPS.
27156         (ASM_SPEC): Support mmicromips and mno-micromips.
27157         (M16STORE_REG_P): New macro.
27158         (MIPS_CALL): Support TARGET_MICROMIPS.
27159         (MICROMIPS_J): New macro.
27160         (mips_base_mips16): Rename this...
27161         (mips_base_compression_flags): ...to this.
27162         (UMIPS_12BIT_OFFSET_P): New macro.
27163         * config/mips/t-sde: (MULTILIB_OPTIONS): Add microMIPS.
27164         (MULTILIB_DIRNAMES): Likewise.
27166 2013-02-24  Jakub Jelinek  <jakub@redhat.com>
27168         PR target/52555
27169         * target-globals.c (save_target_globals): For init_reg_sets and
27170         target_reinit remporarily set this_fn_optabs to this_target_optabs.
27172 2013-02-22  James Greenhalgh  <james.greenhalgh@arm.com>
27174         * config/aarch64/aarch64-simd-builtins.def: Add copyright header.
27175         * config/aarch64/t-aarch64
27176         (aarch64-builtins.o): Depend on aarch64-simd-builtins.def.
27178 2013-02-22  Vladimir Makarov  <vmakarov@redhat.com>
27180         PR inline-asm/56148
27181         * lra-constraints.c (process_alt_operands): Reload operand
27182         conflicting with earlier clobber only if no more other conflicting
27183         operands.
27185 2013-02-22  Jakub Jelinek  <jakub@redhat.com>
27187         PR sanitizer/56393
27188         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Link in libasan_preinit.o
27189         if not linking a shared library.
27191 2013-02-22  Seth LaForge  <sethml@google.com>
27193         * config.gcc (arm*-*-eabi*): Treat arm*eb as big-endian.
27195 2013-02-22  Greta Yorsh  <Greta.Yorsh@arm.com>
27197         * config/arm/arm.md (split for extendsidi): Update condition.
27198         (zero_extend<mode>di2,extend<mode>di2): Add an alternative.
27199         * config/arm/iterators.md (qhs_extenddi_cstr): Likewise.
27200         (qhs_zextenddi_cstr): Likewise.
27202 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27204         PR middle-end/56420
27205         * expmed.c (EXACT_POWER_OF_2_OR_ZERO_P): Do subtraction in uhwi, to
27206         avoid signed wrapping.
27207         (expand_mult): Handle properly multiplication by
27208         ((dword_type) -1) << (BITS_PER_WORD - 1).  Improve multiplication by
27209         ((dword_type) 1) << (BITS_PER_WORD - 1).  Avoid undefined behavior
27210         in the compiler if coeff is HOST_WIDE_INT_MIN.
27211         (expand_divmod): Don't make ext_op1 static, change it's type to
27212         uhwi.  Avoid undefined behavior in -INTVAL (op1).
27214         PR rtl-optimization/50339
27215         * lower-subreg.h (struct lower_subreg_choices): Add splitting_ashiftrt
27216         field.
27217         * lower-subreg.c (compute_splitting_shift): Handle ASHIFTRT.
27218         (compute_costs): Call compute_splitting_shift also for ASHIFTRT
27219         into splitting_ashiftrt field.
27220         (find_decomposable_shift_zext, resolve_shift_zext): Handle also
27221         ASHIFTRT.
27222         (dump_choices): Fix up printing LSHIFTRT choices, print ASHIFTRT
27223         choices.
27225 2013-02-20  Aldy Hernandez  <aldyh@redhat.com>
27227         PR middle-end/56108
27228         * trans-mem.c (execute_tm_mark): Do not expand transactions that
27229         are sure to go irrevocable.
27231 2013-02-21  Hans-Peter Nilsson  <hp@axis.com>
27233         * doc/rtl.texi (vec_concat, vec_duplicate): Mention that
27234         scalars are valid operands.
27236 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27238         PR tree-optimization/56310
27239         * ipa-cp.c (agg_replacements_to_vector): New parameter index, copy
27240         only matching indices and non-negative final offsets.
27241         (intersect_aggregates_with_edge): Pass src_idx to
27242         agg_replacements_to_vector.  Pass src_idx insstead of index to
27243         intersect_with_agg_replacements.
27245 2013-02-21  Martin Jambor  <mjambor@suse.cz>
27247         * ipa-cp.c (good_cloning_opportunity_p): Dump the real threshold
27248         instead of hard-wired defaults.
27250 2013-02-21  Maciej W. Rozycki  <macro@codesourcery.com>
27252         * doc/invoke.texi (MIPS Options): Update documentation of the
27253         floating-point multiply-accumulate instruction restrictions.
27255 2013-02-21  Kostya Serebryany  <kcc@google.com>
27257         * config/i386/i386.c (ix86_asan_shadow_offset): Use 0x7fff8000 as
27258         asan_shadow_offset on x86_64 linux.
27260 2013-02-21  Richard Biener  <rguenther@suse.de>
27262         PR tree-optimization/56415
27263         Revert
27264         2013-02-11  Richard Biener  <rguenther@suse.de>
27266         PR tree-optimization/56273
27267         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27268         first VRP run.
27270 2013-02-21  Jakub Jelinek  <jakub@redhat.com>
27272         PR bootstrap/56258
27273         * doc/invoke.texi (-fdump-rtl-pro_and_epilogue): Use @item
27274         instead of @itemx.
27276         PR inline-asm/56405
27277         * expr.c (expand_expr_real_1) <case TARGET_MEM_REF, MEM_REF>: Don't
27278         use movmisalign or extract_bit_field for EXPAND_MEMORY modifier.
27280 2013-02-20  Jan Hubicka  <jh@suse.cz>
27282         PR tree-optimization/56265
27283         * ipa-prop.c (ipa_make_edge_direct_to_target): Fixup callgraph
27284         when target is referenced for first time.
27286 2013-02-20  Richard Biener  <rguenther@suse.de>
27288         * tree-call-cdce.c (tree_call_cdce): Do not remove unused locals.
27289         * tree-ssa-forwprop.c (ssa_forward_propagate_and_combine): Likewise.
27290         * tree-ssa-dce.c (perform_tree_ssa_dce): Likewise.
27291         * tree-ssa-copyrename.c (copy_rename_partition_coalesce): Do
27292         not return anything.
27293         (rename_ssa_copies): Do not remove unused locals.
27294         * tree-ssa-ccp.c (do_ssa_ccp): Likewise.
27295         * tree-ssanames.c (pass_release_ssa_names): Remove unused locals first.
27296         * passes.c (execute_function_todo): Do not schedule unused locals
27297         removal if cleanup_tree_cfg did something.
27298         * tree-ssa-live.c (remove_unused_locals): Dump statistics
27299         about the number of removed locals.
27301 2013-02-20  Richard Biener  <rguenther@suse.de>
27303         PR tree-optimization/56398
27304         * tree-vect-loop-manip.c (adjust_debug_stmts): Skip SSA default defs.
27306 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27308         PR tree-optimization/55334
27309         * ipa-cp.c (initialize_node_lattices): Disable IPA-CP through and to
27310         restricted pointers to arrays.
27312 2013-02-20  Richard Biener  <rguenther@suse.de>
27313             Jakub Jelinek  <jakub@redhat.com>
27315         PR tree-optimization/56396
27316         * tree-ssa-ccp.c (n_const_val): New static variable.
27317         (get_value): Return NULL for SSA names we don't have a lattice
27318         entry for.
27319         (ccp_initialize): Initialize n_const_val.
27320         * tree-ssa-copy.c (n_copy_of): New static variable.
27321         (init_copy_prop): Initialize n_copy_of.
27322         (get_value): Return NULL_TREE for SSA names we don't have a
27323         lattice entry for.
27325 2013-02-20  Martin Jambor  <mjambor@suse.cz>
27327         * ipa-cp.c (initialize_node_lattices): Fix dumping condition.
27329 2013-02-20  Richard Biener  <rguenther@suse.de>
27331         * genpreds.c (write_lookup_constraint): Do not compare first
27332         letter of the constraint again.
27334 2013-02-20  Richard Biener  <rguenther@suse.de>
27336         * tree-ssa-loop-ivopts.c (alloc_use_cost_map): Use bitmap_count_bits
27337         and ceil_log2.
27338         (get_use_iv_cost): Terminate hashtable walk when coming across
27339         an empty entry.
27341 2013-02-20  Igor Zamyatin  <igor.zamyatin@intel.com>
27343         * config/i386/i386.c (initial_ix86_tune_features): Turn on fp
27344         reassociation for avx2 targets.
27346 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27348         * config/microblaze/microblaze.c: microblaze_has_clz = 0
27349         Add version check for v8.10.a to enable microblaze_has_clz
27350         * config/microblaze/microblaze.h: Add TARGET_HAS_CLZ as combined
27351         version and TARGET_PATTERN_COMPARE check
27352         * config/microblaze/microblaze.md: New clzsi2 instruction
27354 2012-02-19  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27356         * config/microblaze/microblaze.md (call_value_intern): Check symbol is
27357         function before branching.
27359 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27361         * sel-sched-dump.c (dump_insn_rtx_flags): Explicitly set
27362         DUMP_INSN_RTX_UID.
27363         (dump_insn_rtx_1): Pass PATTERN (insn) to str_pattern_slim.
27365 2012-02-19  Andrey Belevantsev  <abel@ispras.ru>
27367         PR middle-end/55889
27368         * sel-sched.c: Include ira.h.
27369         (implicit_clobber_conflict_p): New function.
27370         (moveup_expr): Use it.
27371         * Makefile.in (sel-sched.o): Depend on ira.h.
27373 2013-02-19  Richard Biener  <rguenther@suse.de>
27375         PR tree-optimization/56384
27376         * tree-ssa-sccvn.h (struct vn_phi_s): Add type member.
27377         (vn_hash_type): Split out from ...
27378         (vn_hash_constant_with_type): ... here.
27379         * tree-ssa-sccvn.c (vn_phi_compute_hash): Use vn_hash_type.
27380         (vn_phi_eq): Compare types from vn_phi_s structure.
27381         (vn_phi_lookup): Populate vn_phi_s type.
27382         (vn_phi_insert): Likewise.
27384 2013-02-19  Jakub Jelinek  <jakub@redhat.com>
27386         PR tree-optimization/56350
27387         * tree-vect-loop.c (vectorizable_reduction): If orig_stmt, return false
27388         if haven't found reduction or nested cycle operand, rather than
27389         asserting we must find it.
27391         PR tree-optimization/56381
27392         * tree-ssa-pre.c (create_expression_by_pieces): Fix up last argument
27393         to fold_build3.
27395 2013-02-18  Aldy Hernandez  <aldyh@redhat.com>
27396             Jakub Jelinek  <jakub@redhat.com>
27398         PR target/52555
27399         * genopinit.c (raw_optab_handler): Use this_fn_optabs.
27400         (swap_optab_enable): Same.
27401         (init_all_optabs): Use argument instead of global.
27402         * tree.h (struct tree_optimization_option): New field target_optabs.
27403         * expr.h (init_all_optabs): Add argument to prototype.
27404         (TREE_OPTIMIZATION_OPTABS): New.
27405         (save_optabs_if_changed): Protoize.
27406         * optabs.h: Declare this_fn_optabs.
27407         * optabs.c (save_optabs_if_changed): New.
27408         Declare this_fn_optabs.
27409         (init_optabs): Add argument to init_all_optabs() call.
27410         * function.c (invoke_set_current_function_hook): Handle per
27411         function optabs.
27412         * function.h (struct function): New field optabs.
27413         * config/mips/mips.c (mips_set_mips16_mode): Handle when
27414         optimization_current_node has changed.
27415         * target-globals.h (save_target_globals_default_opts): Protoize.
27416         * target-globals.c (save_target_globals_default_opts): New.
27418 2013-02-18  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
27420         PR target/56347
27421         * config/pa/pa.c (pa_conditional_register_usage): On HP-UX, mark
27422         registers %fr12 and %fr12R as call used.
27424         PR target/56214
27425         * config/pa/predicates.md (base14_operand): Except for BLKmode, QImode
27426         and HImode, require all displacements to be an integer multiple of
27427         their mode size.
27428         * config/pa/pa.c (pa_legitimate_address_p): For REG+BASE addresses,
27429         only allow QImode and HImode when reload is in progress and strict is
27430         true.  Likewise for symbolic addresses.  Use base14_operand to check
27431         displacements in REG+BASE addresses.
27433 2013-02-18  Richard Biener  <rguenther@suse.de>
27435         PR tree-optimization/56366
27436         * tree-vect-loop.c (get_initial_def_for_induction): Properly
27437         handle sign-conversion of outer-loop initial induction value.
27439 2013-02-18  Richard Biener  <rguenther@suse.de>
27441         PR middle-end/56349
27442         * cfghooks.c (merge_blocks): If we merge a latch into another
27443         block adjust references to it.
27444         * cfgloop.c (flow_loops_find): Reset latch before recomputing it.
27445         (verify_loop_structure): Verify that a recorded latch is in fact
27446         a latch.
27448 2013-02-18  Richard Biener  <rguenther@suse.de>
27450         PR tree-optimization/56321
27451         * tree-ssa-reassoc.c (propagate_op_to_single_use): Properly
27452         order SSA name release and virtual operand unlinking.
27454 2013-02-17  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27456         * config/microblaze/microblaze.md (save_stack_block): Define.
27457         (restore_stack_block): Likewise.
27459 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27461         * config/microblaze/linux.h (TARGET_SUPPORTS_PIC): Define as 1.
27462         * config/microblaze/microblaze.h (TARGET_SUPPORTS_PIC): Define as 1.
27463         * config/microblaze/microblaze.c (microblaze_option_override):
27464         Bail out early for PIC modes when target does not support PIC.
27466 2013-02-16  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27468         * config/microblaze/microblaze.c (microblaze_asm_trampoline_template):
27469         Replace with a microblaze version.
27470         (microblaze_trampoline_init): Adapt for microblaze.
27471         * config/microblaze/microblaze.h (TRAMPOLINE_SIZE): Adapt for
27472         microblaze.
27474 2013-02-16  Jakub Jelinek  <jakub@redhat.com>
27475             Dodji Seketeli  <dodji@redhat.com>
27477         PR asan/56330
27478         * asan.c (get_mem_refs_of_builtin_call): White space and style cleanup.
27479         (instrument_mem_region_access): Do not forget to always put
27480         instrumentation of the of 'base' and 'base + len' in a "if (len !=
27481         0) statement, even for cases where either 'base' or 'base + len'
27482         are not instrumented -- because they have been previously
27483         instrumented.  Simplify the logic by putting all the statements
27484         instrument 'base + len' inside a sequence, and then insert that
27485         sequence right before the current insertion point.  Then, to
27486         instrument 'base + len', just get an iterator on that statement.
27487         And do not forget to update the pointer to iterator the function
27488         received as argument.
27490 2013-02-15  Vladimir Makarov  <vmakarov@redhat.com>
27492         PR rtl-optimization/56348
27493         * lra-assigns.c (reload_pseudo_compare_func): Prefer bigger pseudos.
27495 2013-02-15  Steven Bosscher  <steven@gcc.gnu.org>
27497         * graph.c (start_graph_dump): Print dumpfile base as digraph label.
27498         (clean_graph_dump_file): Pass base to start_graph_dump.
27500 2013-02-14  Richard Henderson  <rth@redhat.com>
27502         PR target/55941
27503         * lower-subreg.c (simple_move): Check dest mode instead of src mode.
27505 2013-02-14  Steven Bosscher  <steven@gcc.gnu.org>
27507         * collect2-aix.h: Define F_LOADONLY.
27509 2013-02-14  Richard Biener  <rguenther@suse.de>
27511         PR lto/50494
27512         * varasm.c (output_constant_def_1): Get the decl representing
27513         the constant as argument.
27514         (output_constant_def): Wrap output_constant_def_1.
27515         (make_decl_rtl): Use output_constant_def_1 with the decl
27516         representing the constant.
27517         (build_constant_desc): Optionally re-use a decl already
27518         representing the constant.
27519         (tree_output_constant_def): Adjust.
27521 2013-02-14  Dodji Seketeli  <dodji@redhat.com>
27523         Fix an asan crash
27524         * asan.c (instrument_builtin_call):  Really put the length of the
27525         second source argument into src1_len.
27527 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27529         * asan.c (create_cond_insert_point): Add create_then_fallthru_edge
27530         argument.  If it is false, don't create edge from then_bb to
27531         fallthru_bb.
27532         (insert_if_then_before_iter): Pass true to it.
27533         (build_check_stmt): Pass false to it.
27534         (transform_statements): Flush hash table only on extended basic
27535         block boundaries, rather than at the beginning of every bb.
27536         Don't flush hash table on nonfreeing_call_p calls.
27537         * tree-flow.h (nonfreeing_call_p): New prototype.
27538         * tree-ssa-phiopt.c (nonfreeing_call_p): No longer static.
27540 2013-02-13  David S. Miller  <davem@davemloft.net>
27542         * expmed.c (expand_shift_1): Only strip scalar integer subregs.
27544 2013-02-13  Vladimir Makarov  <vmakarov@redhat.com>
27546         PR target/56184
27547         * ira.c (max_regno_before_ira): Move from ...
27548         (ira): ... here.
27549         (fix_reg_equiv_init): Use max_regno_before_ira instead of
27550         vec_safe_length.
27552 2013-02-13  Jakub Jelinek  <jakub@redhat.com>
27554         * config/i386/i386.c (ix86_asan_shadow_offset): Revert last change.
27556 2013-02-13  Richard Biener  <rguenther@suse.de>
27558         PR lto/56295
27559         * gimple-streamer-out.c (output_gimple_stmt): Undo wrapping
27560         globals in MEM_REFs.
27562 2013-02-13  Richard Biener  <rguenther@suse.de>
27564         * loop-init.c (loop_optimizer_init): Clear loop state when
27565         re-initializing preserved loops.
27566         * loop-unswitch.c (unswitch_single_loop): Return whether
27567         we unswitched the loop.  Do not verify loop state here.
27568         (unswitch_loops): When we unswitched a loop discover new loops.
27570 2013-02-13  Kostya Serebryany  <kcc@google.com>
27572         * config/i386/i386.c: Use 0x7fff8000 as asan_shadow_offset
27573         on x86_64 linux.
27574         * sanitizer.def: Rename __asan_init to __asan_init_v1.
27576 2013-02-12  Dodji Seketeli  <dodji@redhat.com>
27578         Avoid instrumenting duplicated memory access in the same basic block
27579         * Makefile.in (asan.o): Add new dependency on hash-table.h
27580         * asan.c (struct asan_mem_ref, struct mem_ref_hasher): New types.
27581         (asan_mem_ref_init, asan_mem_ref_get_end, get_mem_ref_hash_table)
27582         (has_stmt_been_instrumented_p, empty_mem_ref_hash_table)
27583         (free_mem_ref_resources, has_mem_ref_been_instrumented)
27584         (has_stmt_been_instrumented_p, update_mem_ref_hash_table)
27585         (get_mem_ref_of_assignment): New functions.
27586         (get_mem_refs_of_builtin_call): Extract from
27587         instrument_builtin_call and tweak a little bit to make it fit with
27588         the new signature.
27589         (instrument_builtin_call): Use the new
27590         get_mem_refs_of_builtin_call.  Use gimple_call_builtin_p instead
27591         of is_gimple_builtin_call.
27592         (instrument_derefs, instrument_mem_region_access): Insert the
27593         instrumented memory reference into the hash table.
27594         (maybe_instrument_assignment): Renamed instrument_assignment into
27595         this, and change it to advance the iterator when instrumentation
27596         actually happened and return true in that case.  This makes it
27597         homogeneous with maybe_instrument_assignment, and thus give a
27598         chance to callers to be more 'regular'.
27599         (transform_statements): Clear the memory reference hash table
27600         whenever we enter a new BB, when we cross a function call, or when
27601         we are done transforming statements.  Use
27602         maybe_instrument_assignment instead of instrumentation.  No more
27603         need to special case maybe_instrument_assignment and advance the
27604         iterator after calling it; it's now handled just like
27605         maybe_instrument_call.  Update comment.
27607 2013-02-13  Richard Biener  <rguenther@suse.de>
27609         * config/mn10300/mn10300.c (mn10300_scan_for_setlb_lcc):
27610         Fix loop discovery code.
27612 2013-02-12  Vladimir Makarov  <vmakarov@redhat.com>
27614         PR inline-asm/56148
27615         * lra-constraints.c (process_alt_operands): Match early clobber
27616         operand with itself.  Check conflicts with earlyclobber only if
27617         the operand is not reloaded.  Prefer to reload conflicting operand
27618         if earlyclobber and matching operands are the same.
27620 2013-02-12  Richard Biener  <rguenther@suse.de>
27622         PR lto/56297
27623         * lto-streamer-out.c (write_symbol): Do not output symbols
27624         for hard register variables.
27626 2013-02-12  Georg-Johann Lay  <avr@gjlay.de>
27628         PR target/54222
27629         * config/avr/avr-dimode.md (umulsidi3, mulsidi3): New expanders.
27630         (umulsidi3_insn, mulsidi3_insn): New insns.
27632 2013-02-12  Christophe Lyon  <christophe.lyon@linaro.org>
27634         * config/arm/arm-protos.h (struct cpu_vec_costs): New struct type.
27635         (struct tune_params): Add vec_costs field.
27636         * config/arm/arm.c (arm_builtin_vectorization_cost)
27637         (arm_add_stmt_cost): New functions.
27638         (TARGET_VECTORIZE_BUILTIN_VECTORIZATION_COST)
27639         (TARGET_VECTORIZE_ADD_STMT_COST): Define.
27640         (arm_default_vec_cost): New struct of type cpu_vec_costs.
27641         (arm_slowmul_tune, arm_fastmul_tune, arm_strongarm_tune)
27642         (arm_xscale_tune, arm_9e_tune, arm_v6t2_tune, arm_cortex_tune)
27643         (arm_cortex_a15_tune, arm_cortex_a5_tune, arm_cortex_a9_tune)
27644         (arm_v6m_tune, arm_fa726te_tune): Define new vec_costs field.
27646 2013-02-12  Richard Biener  <rguenther@suse.de>
27648         PR lto/56295
27649         * gimple-streamer-in.c (input_gimple_stmt): Strip MEM_REFs off
27650         decls again if possible.
27652 2013-02-12  Richard Biener  <rguenther@suse.de>
27654         PR middle-end/56288
27655         * tree-ssa.c (verify_ssa_name): Fix check, move
27656         SSA_NAME_IN_FREE_LIST check up.
27658 2013-02-12  Jakub Jelinek  <jakub@redhat.com>
27659             Steven Bosscher   <steven@gcc.gnu.org>
27661         PR rtl-optimization/56151
27662         * optabs.c (add_equal_note): Don't return 0 if target is a MEM,
27663         equal to op0 or op1, and last_insn pattern is CODE operation
27664         with MEM dest and one of the operands matches that MEM.
27666 2013-02-11  Sriraman Tallam  <tmsriram@google.com>
27668         * doc/extend.texi: Document Function Multiversioning and "default"
27669         parameter string to target attribute.
27670         * config/i386/i386.c (get_builtin_code_for_version): Return 0 if
27671         target attribute parameter is "default".
27672         (ix86_compare_version_priority): Remove checks for target attribute.
27673         (ix86_mangle_function_version_assembler_name): Change error to sorry.
27674         Remove check for target attribute equal to NULL. Add assert.
27675         (ix86_generate_version_dispatcher_body): Change error to sorry.
27677 2013-02-11  Iain Sandoe  <iain@codesourcery.com>
27678             Jack Howarth  <howarth@bromo.med.uc.edu>
27679             Patrick Marlier  <patrick.marlier@gmail.com>
27681         PR libitm/55693
27682         * config/darwin.h: Replace ENDFILE_SPEC with TM_DESTRUCTOR and
27683         define ENDFILE_SPEC as TM_DESTRUCTOR.
27684         * config/i386/darwin.h (ENDFILE_SPEC): Use TM_DESTRUCTOR.
27686 2013-02-11  Alexander Potapenko  <glider@google.com>
27687             Jack Howarth  <howarth@bromo.med.uc.edu>
27688             Jakub Jelinek  <jakub@redhat.com>
27690         PR sanitizer/55617
27691         * config/darwin.c (cdtor_record): Rename ctor_record.
27692         (sort_cdtor_records): Rename sort_ctor_records.
27693         (finalize_dtors): New routine to sort destructors by
27694         priority before use in assemble_integer.
27695         (machopic_asm_out_destructor): Use finalize_dtors if needed.
27697 2013-02-11  Uros Bizjak  <ubizjak@gmail.com>
27699         PR rtl-optimization/56275
27700         * simplify-rtx.c (avoid_constant_pool_reference): Check that
27701         offset is non-negative and less than cmode size before
27702         calling simplify_subreg.
27704 2013-02-11  Richard Biener  <rguenther@suse.de>
27706         PR tree-optimization/56264
27707         * cfgloop.h (fix_loop_structure): Adjust prototype.
27708         * loop-init.c (fix_loop_structure): Return the number of
27709         newly discovered loops.
27710         * tree-cfgcleanup.c (repair_loop_structures): When new loops
27711         are discovered, do a full loop-closed SSA rewrite.
27713 2013-02-11  Richard Biener  <rguenther@suse.de>
27715         PR tree-optimization/56273
27716         * tree-vrp.c (simplify_cond_using_ranges): Disable for the
27717         first VRP run.
27718         (check_array_ref): Fix missing newline in dumps.
27719         (search_for_addr_array): Likewise.
27721 2013-02-09  David Edelsohn  <dje.gcc@gmail.com>
27723         * config/rs6000/aix61.h (OS_MISSING_ALTIVEC): Undefine.
27725 2013-02-09  Jakub Jelinek  <jakub@redhat.com>
27727         PR target/56256
27728         * config/rs6000/rs6000.h (ASSEMBLER_DIALECT): Define.
27730 2013-02-08  Vladimir Makarov  <vmakarov@redhat.com>
27732         PR rtl-optimization/56246
27733         * lra-constraints.c (simplify_operand_subreg): Try to reuse
27734         reload pseudo.
27735         * lra.c (lra): Clear lra_optional_reload_pseudos only when all
27736         constraints are satisfied.
27738 2013-02-08  Jeff Law  <law@redhat.com>
27740         PR debug/53948
27741         * emit-rtl.c (reg_is_parm_p): New function.
27742         * regs.h (reg_is_parm_p): New prototype.
27743         * ira-conflicts.c (ira_build_conflicts): Allow parameters in
27744         callee-clobbered registers.
27746 2013-02-08  Michael Meissner  <meissner@linux.vnet.ibm.com>
27748         PR target/56043
27749         * config/rs6000/rs6000.c (rs6000_builtin_vectorized_libmass):
27750         If there is no implicit builtin declaration, just return NULL.
27752 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27754         * config/i386/sse.md (FMAMODEM): New mode iterator.
27755         (fma<mode>4, fms<mode>4, fnma<mode>4, fnms<mode>4): Use FMAMODEM
27756         mode iterator. Do not use TARGET_SSE_MATH in insn constraint.
27758 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27760         * config/i386/gnu-user.h (TARGET_CAN_SPLIT_STACK): Define only
27761         when HAVE_GAS_CFI_PERSONALITY_DIRECTIVE is set.
27762         * config/i386/gnu-user64.h (TARGET_CAN_SPLIT_STACK): Ditto.
27764 2013-02-08  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
27766         * config.gcc (microblaze*-linux*): Add TARGET_BIG_ENDIAN_DEFAULT.
27767         (microblaze*-*-elf): Likewise.
27768         * config/microblaze/linux.h: Add -mbig-endian / -mlittle-endian to
27769         LINK_SPEC.
27770         * config/microblaze/microblaze-c.c: Add builtin defines for
27771         _LITTLE_ENDIAN and _BIG_ENDIAN.
27772         * config/microblaze/microblaze.h: Add TARGET_ENDIAN_DEFAULT and
27773         add to TARGET_DEFAULT flags.
27774         Expand ASM_SPEC and LINK_SPEC.
27775         Update BYTES_BIG_ENDIAN and WORDS_BIG_ENDIAN.
27776         * config/microblaze/microblaze.md: Update extendsidi2 and
27777         movdi_internal instructions to use low-order / high-order reg
27778         print_operands.
27779         * config/microblaze/microblaze.opt: Add mbig-endian and mlittle-endian
27780         options and inversemask / mask of LITTLE_ENDIAN.
27781         * config/microblaze/t-microblaze: Expand multilib options to
27782         include mlittle-endian (le) and update exceptions patterns.
27784 2013-02-08  Jakub Jelinek  <jakub@redhat.com>
27786         PR rtl-optimization/56195
27787         * lra-constraints.c (get_reload_reg): Don't reuse regs
27788         if they have smaller mode than requested, if they have
27789         wider mode than requested, try to return a SUBREG.
27791         PR tree-optimization/56250
27792         * fold-const.c (extract_muldiv_1) <case NEGATE_EXPR>: Don't optimize
27793         if type is unsigned and code isn't MULT_EXPR.
27795 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27797         PR tree-optimization/56064
27798         * fixed-value.c (fixed_from_double_int): Sign/zero extend payload
27799         bits according to mode.
27800         * fixed-value.h (fixed_from_double_int)
27801         (const_fixed_from_double_int): Adjust comments.
27803 2013-02-08  Richard Biener  <rguenther@suse.de>
27805         PR lto/56231
27806         * lto-streamer.h (struct data_in): Remove current_file, current_line
27807         and current_col members.
27808         * lto-streamer-out.c (lto_output_location): Stream changed bits
27809         en-block for efficiency.
27810         * lto-streamer-in.c (clear_line_info): Remove.
27811         (lto_input_location): Cache current file, line and column
27812         globally via local statics.  Read changed bits en-block.
27813         (input_function): Do not call clear_line_info.
27814         (lto_read_body): Likewise.
27815         (lto_input_toplevel_asms): Likewise.
27817 2013-02-08  Michael Matz  <matz@suse.de>
27819         PR tree-optimization/52448
27820         * tree-ssa-phiopt.c (struct name_to_bb): Add phase member.
27821         (nt_call_phase): New static.
27822         (add_or_mark_expr): Only mark accesses with newer phase than any
27823         call seen.
27824         (nonfreeing_call_p): New.
27825         (nt_init_block): Update nt_call_phase, mark blocks as visited.
27826         (nt_fini_block): Keep blocks marked as visited.
27827         (get_non_trapping): Initialize nt_call_phase, and reset aux pointer.
27829 2013-02-08  Richard Biener  <rguenther@suse.de>
27831         * ira.c (ira): Free broken dominator information.
27833 2013-02-08  Uros Bizjak  <ubizjak@gmail.com>
27835         * config/i386/i386.c (ix86_spill_class): Use INTEGER_CLASS_P macro.
27837 2013-02-08  Marek Polacek  <polacek@redhat.com>
27839         * cfgloop.c (verify_loop_structure): Add more checking of headers.
27841 2013-02-08  Richard Biener  <rguenther@suse.de>
27843         PR middle-end/56181
27844         * cfgloop.h (flow_loops_find): Adjust.
27845         (bb_loop_header_p): Declare.
27846         * cfgloop.c (bb_loop_header_p): New function split out from ...
27847         (flow_loops_find): ... here.  Adjust function signature,
27848         support incremental loop structure update.
27849         (verify_loop_structure): Cleanup.  Verify a loop is a loop.
27850         * cfgloopmanip.c (fix_loop_structure): Move ...
27851         * loop-init.c (fix_loop_structure): ... here.
27852         (apply_loop_flags): Split out from ...
27853         (loop_optimizer_init): ... here.
27854         (fix_loop_structure): Use apply_loop_flags.  Use flow_loops_find
27855         in incremental mode, only remove dead loops here.
27857 2013-02-08  Georg-Johann Lay  <avr@gjlay.de>
27859         PR target/54222
27860         * config/avr/avr.md (unspec) <UNSPEC_ROUND>: Add.
27861         * config/avr/avr-fixed.md (ALL4QA, ALL124QA): New mode iterators.
27862         (round<mode>3, round<mode>3_const): New expanders for fixed-mode.
27863         (*round<mode>3.libgcc): New insns for fixed-modes.
27864         * config/avr/builtins.def (ABSxx): Use a non-NULL LIBNAME.
27865         (ROUNDxx, COUNTLSxx, BITSxx, xxBITS): New DEF_BUILTINs.
27866         (ROUNDFX, COUNTLSFX, ABSFX): New DEF_BUILTINs.
27867         * config/avr/stdfix.h (absFX, bitsFX, FXbits): Remove inline
27868         implementations.  Define to __builtin_avr_absFX,
27869         __builtin_avr_bitsFX, __builtin_avr_FXbits, respectively.
27870         (roundFX, countlsFX): Define to __builtin_avr_roundFX,
27871         __builtin_avr_countlsFX, respectively.
27872         * config/avr/avr-c.c (target.h): Include it.
27873         (enum avr_builtin_id): New enum.
27874         (avr_resolve_overloaded_builtin): New static function.
27875         (avr_register_target_pragmas): Use it to set
27876         targetm.resolve_overloaded_builtin.
27877         * config/avr/avr.c (avr_init_builtins): Supply myriads of local
27878         tree nodes used by DEF_BUILTIN.
27879         (avr_expand_builtin) <AVR_BUILTIN_ROUNDxx>: Sanity-check them.
27880         (avr_fold_builtin) <AVR_BUILTIN_BITSxx>: Fold to VIEW_COVERT_EXPR.
27881         <AVR_BUILTIN_xxBITS>: Same.
27883 2013-02-08  Richard Biener  <rguenther@suse.de>
27885         * cfgloop.c (verify_loop_structure): Properly handle
27886         a loop exiting to another loop header.
27887         * ira-int.h (ira_loops): Remove.
27888         * ira.c (ira_loops): Remove.
27889         (ira): Use loop_optimizer_init and loop_optimizer_finalize.
27890         (do_reload): Use loop_optimizer_finalize.
27891         * ira-build.c (create_loop_tree_nodes): Use get_loops and
27892         number_of_loops to access the loop tree.
27893         (more_one_region_p): Likewise.
27894         (finish_loop_tree_nodes): Likewise.
27895         (rebuild_regno_allocno_maps): Likewise.
27896         (mark_loops_for_removal): Likewise.
27897         (mark_all_loops_for_removal): Likewise.
27898         (remove_unnecessary_regions): Likewise.
27899         (ira_build): Likewise.
27900         * ira-emit.c (setup_entered_from_non_parent_p): Likewise.
27902 2013-02-08  Richard Biener  <rguenther@suse.de>
27904         * Makefile.in (tree-tailcall.o): Add $(CFGLOOP_H) dependency.
27905         * ipa-pure-const.c (analyze_function): Avoid calling
27906         mark_irreducible_loops twice.
27907         * tree-tailcall.c (tree_optimize_tail_calls_1): Mark loops for fixup.
27909 2013-02-07  David S. Miller  <davem@davemloft.net>
27911         * dwarf2out.c (based_loc_descr): Perform leaf register remapping
27912         on 'reg'.
27913         * var-tracking.c (vt_add_function_parameter): Test the presence of
27914         HAVE_window_save properly and do not remap argument registers when
27915         we have a leaf function.
27917 2013-02-07  Uros Bizjak  <ubizjak@gmail.com>
27919         PR bootstrap/56227
27920         * ggc-page.c (ggc_print_statistics): Use HOST_LONG_LONG_FORMAT
27921         instead of "ll".
27922         * config/i386/i386.c (ix86_print_operand): Ditto.
27924 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27926         * lra-constraints.c (process_alt_operands): Fix recently added comment.
27928 2013-02-07  Vladimir Makarov  <vmakarov@redhat.com>
27930         PR rtl-optimization/56225
27931         * lra-constraints.c (process_alt_operands): Check that reload hard
27932         reg can hold value for strict_low_part.
27934 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27936         PR debug/56154
27937         * dwarf2out.c (dwarf2_debug_hooks): Set end_function hook to
27938         dwarf2out_end_function.
27939         (in_first_function_p, maybe_at_text_label_p,
27940         first_loclabel_num_not_at_text_label): New variables.
27941         (dwarf2out_var_location): In the first function find out
27942         lowest loclabel_num N where .LVLN is known not to be equal to .Ltext0.
27943         (find_empty_loc_ranges_at_text_label, dwarf2out_end_function): New
27944         functions.
27946 2013-02-07  Eric Botcazou  <ebotcazou@adacore.com>
27948         PR rtl-optimization/56178
27949         * cse.c (cse_insn): Do not create a REG_EQUAL note if the source is a
27950         SUBREG of a register.  Tidy up related block of code.
27951         * fwprop.c (forward_propagate_and_simplify): Do not create a REG_EQUAL
27952         note if the source is a register or a SUBREG of a register.
27954 2013-02-07  Jakub Jelinek  <jakub@redhat.com>
27956         PR target/56228
27957         * config/rs6000/rs6000.md (ptrm): New mode attr.
27958         (call_indirect_aix<ptrsize>, call_indirect_aix<ptrsize>_nor11,
27959         call_value_indirect_aix<pttrsize>,
27960         call_value_indirect_aix<pttrsize>_nor11): Use <ptrm> instead of
27961         m in constraints.
27963 2013-02-07  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
27965         * collect2.c (main): Set aix64_flag for -G and -bsvr4 too, disable
27966         if -bnortl. Convert to strcmp and strncmp.
27968 2013-02-07  Alan Modra  <amodra@gmail.com>
27970         PR target/54009
27971         * config/rs6000/rs6000.c (mem_operand_gpr): Check that LO_SUM
27972         addresses won't wrap when offsetting.
27973         (rs6000_secondary_reload): Provide secondary reloads needed for
27974         wrapping LO_SUM addresses.
27976 2013-02-06  Thomas Schwinge  <thomas@codesourcery.com>
27978         * config/gnu.h (GNU_USER_TARGET_OS_CPP_BUILTINS): Never define
27979         MACH, just __MACH__.
27981 2013-02-06  Richard Biener  <rguenther@suse.de>
27983         * tracer.c (tracer): Mark loops with LOOPS_NEED_FIXUP
27984         instead of calling fix_loop_structure.
27986 2013-02-06  Jakub Jelinek  <jakub@redhat.com>
27988         PR middle-end/56217
27989         * omp-low.c (use_pointer_for_field): Return false if
27990         lower_send_shared_vars doesn't generate any copy-out code.
27992 2013-02-06  Tom de Vries  <tom@codesourcery.com>
27994         PR rtl-optimization/56131
27995         * cfgrtl.c (delete_insn): Use NOTE_BASIC_BLOCK instead of BLOCK_FOR_INSN
27996         to get the bb of a NOTE_INSN_BASIC_BLOCK.  Handle the case that the bb
27997         of the label is NULL.  Add comment.
27999 2013-02-05  Jakub Jelinek  <jakub@redhat.com>
28001         * tree.h (struct tree_decl_with_vis): Remove thread_local field.
28003         PR sanitizer/55374
28004         * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Define.
28005         (STATIC_LIBTSAN_LIBS): Likewise.
28006         * gcc.c (ADD_STATIC_LIBTSAN_LIBS, LIBTSAN_EARLY_SPEC): Define.
28007         (LIBTSAN_SPEC): Add ADD_STATIC_LIBTSAN_LIBS, if LIBTSAN_EARLY_SPEC
28008         is defined, don't add anything else beyond that.
28009         (SANITIZER_EARLY_SPEC, SANITIZER_SPEC): Define.
28010         (LINK_COMMAND_SPEC): Use them.
28012         PR tree-optimization/56205
28013         * tree-stdarg.c (check_all_va_list_escapes): Return true if
28014         there are any PHI nodes that set non-va_list_escape_vars SSA_NAME
28015         and some va_list_escape_vars SSA_NAME appears in some PHI argument.
28017 2013-02-05  Richard Biener  <rguenther@suse.de>
28019         PR tree-optimization/53342
28020         PR tree-optimization/53185
28021         * tree-vectorizer.h (vect_check_strided_load): Remove.
28022         * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Do
28023         not disallow peeling for vectorized strided loads.
28024         (vect_check_strided_load): Make static and simplify.
28025         (vect_analyze_data_refs): Adjust.
28026         * tree-vect-stmts.c (vectorizable_load): Handle peeled loops
28027         correctly when vectorizing strided loads.
28029 2013-02-05  Richard Biener  <rguenther@suse.de>
28031         * doc/install.texi: Refer to ISL, not PPL.
28033 2013-02-05  Jan Hubicka  <jh@suse.cz>
28035         PR tree-optimization/55789
28036         * params.def (PARAM_EARLY_INLINER_MAX_ITERATIONS): Drop to 1.
28038 2013-02-05  Jan Hubicka  <jh@suse.cz>
28040         PR tree-optimization/55789
28041         * cgraphclones.c (cgraph_remove_node_and_inline_clones): Remove
28042         the dead call anyway.
28044 2013-02-05  Eric Botcazou  <ebotcazou@adacore.com>
28046         PR sanitizer/55374
28047         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Add missing guard.
28049 2013-02-04  Alexander Potapenko  <glider@google.com>
28050             Jack Howarth  <howarth@bromo.med.uc.edu>
28051             Jakub Jelinek  <jakub@redhat.com>
28053         PR sanitizer/55617
28054         * config/darwin.c (sort_ctor_records): Stabilized qsort
28055         on constructor priority by using original position.
28056         (finalize_ctors): New routine to sort constructors by
28057         priority before use in assemble_integer.
28058         (machopic_asm_out_constructor): Use finalize_ctors if needed.
28060 2013-02-04  Jakub Jelinek  <jakub@redhat.com>
28062         PR libstdc++/54314
28063         * config/i386/winnt.c (i386_pe_assemble_visibility): Don't warn
28064         about visibility on artificial decls.
28065         * config/sol2.c (solaris_assemble_visibility): Likewise.
28067 2013-02-04  Kai Tietz  <ktietz@redhat.com>
28069         PR target/56186
28070         * config/i386/i386.c (function_value_ms_64): Add additional valtype
28071         argument and improve checking of return-argument types for 16-byte
28072         modes.
28073         (ix86_function_value_1): Add additional valtype argument on call
28074         of function_value_64.
28075         (return_in_memory_ms_64): Sync 16-byte sized mode handling with
28076         handling infunction_value_64 function.
28078 2013-02-04  Matthew Gretton-Dann  <matthew.gretton-dann@linaro.org>
28080         * reload.c (subst_reloads): Fix DEBUG_RELOAD build issue.
28082 2013-02-04  Richard Biener  <rguenther@suse.de>
28084         PR tree-optimization/56188
28085         * tree-ssa-structalias.c (label_visit): Consider case with
28086         initially non-empty points-to set.
28087         (perform_var_substitution): Dump node mapping and clean up.
28089 2013-02-04  Richard Guenther  <rguenther@suse.de>
28091         PR lto/56168
28092         * lto-symtab.c (lto_symtab_merge_decls_1): Make non-builtin
28093         node prevail as last resort.
28094         (lto_symtab_merge_decls): Remove guard on LTRANS here.
28095         (lto_symtab_prevailing_decl): Builtins are their own prevailing decl.
28097 2013-02-04  Richard Biener  <rguenther@suse.de>
28099         PR tree-optimization/56113
28100         * tree-ssa-structalias.c (equiv_class_lookup, equiv_class_add):
28101         Merge into ...
28102         (equiv_class_lookup_or_add): ... this.
28103         (label_visit): Adjust and fix error in previous patch.
28104         (perform_var_substitution): Adjust.
28106 2013-02-03  Oleg Endo  <olegendo@gcc.gnu.org>
28108         * config/sh/divtab.c: Fix formatting and comments throughout the file.
28109         * config/sh/sh4-300.md: Likewise.
28110         * config/sh/sh4a.md: Likewise.
28111         * config/sh/constraints.md: Likewise.
28112         * config/sh/sh.md: Likewise.
28113         * config/sh/netbsd-elf.h: Likewise.
28114         * config/sh/predicates.md: Likewise.
28115         * config/sh/sh-protos.h: Likewise.
28116         * config/sh/ushmedia.h: Likewise.
28117         * config/sh/linux.h: Likewise.
28118         * config/sh/sh.c: Likewise.
28119         * config/sh/superh.h: Likewise.
28120         * config/sh/elf.h: Likewise.
28121         * config/sh/sh4.md: Likewise.
28122         * config/sh/sh.h: Likewise.
28124 2013-02-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28126         * config/pa/constraints.md: Adjust unused letters.  Change "T"
28127         constraint to match_test floating_point_store_memory_operand().
28128         * config/pa/predicates.md (reg_plus_base_memory_operand): New.
28129         (base14_operand): New.
28130         (floating_point_store_memory_operand): New.
28131         (integer_store_memory_operand): Revise to use base14_operand and
28132         reg_plus_base_memory_operand.
28133         (move_dest_operand): Allow symbolic_memory_operands.
28134         (symbolic_memory_operand): Check for LO_SOM.
28135         (symbolic_operand): Change default case to break.
28136         * config/pa/pa.md: Remove unamed DFmode and SFmode patterns to force
28137         CONST_DOUBLE values to be reloaded by putting them into memory when
28138         the destination is a floating point register.
28139         (movdf): Remove code to handle CONST_DOUBLE.
28140         (movsf): Likewise.
28141         (reload_indf_r1): New.
28142         (reload_insf_r1): New.
28143         Consistently use "Q" and "T" constraints with integer and floating
28144         point move instructions, respectively.
28145         (movdi): Remove FAIL.
28146         Change predicate for source operand unamed DImode move from
28147         general_operand to move_src_operand.
28148         (umulsidi3): Change predicate for destination operand to
28149         register_operand.
28150         Likewise for similar unamed patterns.
28151         * config/pa/pa-protos.h (pa_legitimize_reload_address): Declare.
28152         * config/pa/pa.c (pa_symbolic_expression_p): Remove extra parenthesis.
28153         (hppa_legitimize_address): Simplify mask calculation.
28154         (pa_emit_move_sequence): Revised handling of secondary reloads from
28155         REG+D addresses for floating point loads and stores.  Directly handle
28156         loading CONST0_RTX (mode) to a floating point register.
28157         (pa_secondary_reload): Handle reloading DF and SFmode constant values
28158         to floating point registers.  Don't restrict secondary reloads to
28159         floating point registers to integer modes.  Revise some comments and
28160         cleanup some code.
28161         (TARGET_LEGITIMATE_ADDRESS_P): Define.
28162         (pa_legitimate_address_p): New.
28163         (pa_legitimize_reload_address): New.
28164         * config/pa/pa.h (STRICT_REG_OK_FOR_INDEX_P): New.
28165         (STRICT_REG_OK_FOR_BASE_P): New.
28166         (GO_IF_LEGITIMATE_ADDRESS): Delete.  Update some related comments.
28167         (LEGITIMIZE_RELOAD_ADDRESS): Revise to use pa_legitimize_reload_address.
28169 2013-02-03  David Edelsohn  <dje.gcc@gmail.com>
28170             Andrew Dixie  <andrewd@gentrack.com>
28172         * collect2.c (GCC_CHECK_HDR): Do not scan objects with F_LOADONLY
28173         flag set.
28175 2013-02-03  Richard Sandiford  <rdsandiford@googlemail.com>
28177         * expmed.c (extract_bit_field_1): Pass the full width of the
28178         structure to get_best_reg_extraction_insn.
28180 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28182         PR target/54601
28183         * configure.ac (use_cxa_atexit): Add AIX.
28184         * configure: Regenerate.
28186         * config/rs6000/aix61.h (STARTFILE_SPEC): Add crtcxa.o.
28188 2013-02-01  Jakub Jelinek  <jakub@redhat.com>
28190         PR debug/54793
28191         * final.c (need_profile_function): New variable.
28192         (final_start_function): Drop ATTRIBUTE_UNUSED from first argument.
28193         If first of NOTE_INSN_BASIC_BLOCK or NOTE_INSN_FUNCTION_BEG
28194         is only preceeded by NOTE_INSN_VAR_LOCATION or NOTE_INSN_DELETED
28195         notes, targetm.asm_out.function_prologue doesn't emit anything,
28196         HAVE_prologue and profiler should be emitted before prologue,
28197         set need_profile_function instead of emitting it.
28198         (final_scan_insn): If need_profile_function, emit
28199         profile_function on the first NOTE_INSN_BASIC_BLOCK or
28200         NOTE_INSN_FUNCTION_BEG note.
28202 2013-02-01  Richard Henderson  <rth@redhat.com>
28204         * config/rs6000/rs6000.md (smulditi3): New.
28205         (umulditi3): New.
28207         * config/alpha/alpha.md (umulditi3): New.
28209 2013-02-01  David Edelsohn  <dje.gcc@gmail.com>
28211         * config/rs6000/xcoff.h (ASM_OUTPUT_ALIGNED_COMMON): Use floor_log2.
28212         (ASM_OUTPUT_ALIGNED_LOCAL): New.
28214 2013-02-01  Richard Biener  <rguenther@suse.de>
28216         PR tree-optimization/56113
28217         * tree-ssa-structalias.c (label_visit): Reduce work for
28218         single-predecessor nodes.
28220 2013-02-01  Eric Botcazou  <ebotcazou@adacore.com>
28222         * fold-const.c (make_range_step) <TRUTH_NOT_EXPR>: Bail out if the
28223         range isn't testing for zero.
28225 2013-01-31  Steven Bosscher  <steven@gcc.gnu.org>
28227         PR middle-end/56113
28228         * fwprop.c (fwprop_init): Set up loops without CFG modifications.
28230 2013-01-31  Hiroyuki Ono  <hiroyuki.ono.jc@renesas.com>
28231             Nick Clifton  <nickc@redhat.com>
28233         * config/v850/constraints.md (Q): Define as a memory constraint.
28234         * config/v850/predicates.md (label_ref_operand): New predicate.
28235         (e3v5_shift_operand): New predicate.
28236         (ior_operator): New predicate.
28237         * config/v850/t-v850: Add e3v5 multilib.
28238         * config/v850/v850-protos.h (v850_adjust_insn_length): Prototype.
28239         (v850_gen_movdi): Prototype.
28240         * config/v850/v850.c: Add support for e3v5 architecture.
28241         Rename all uses of TARGET_V850E || TARGET_V850E2_ALL to
28242         TARGET_V850E_UP.
28243         (construct_save_jarl): Add e3v5 long JARL support.
28244         (v850_adjust_insn_length): New function.  Adjust length of call
28245         insns when using e3v5 instructions.
28246         (v850_gen_movdi): New function: Generate instructions to move a
28247         DImode value.
28248         * config/v850/v850.h (TARGET_CPU_v850e3v5): Define.
28249         (CPP_SPEC): Define __v850e3v5__ as appropriate.
28250         (TARGET_USE_FPU): Enable for e3v5.
28251         (CONST_OK_FOR_W): New macro.
28252         (ADJUST_INSN_LENGTH): Define.
28253         * config/v850/v850.md (UNSPEC_LOOP): Define.
28254         (attr cpu): Add v850e3v5.
28255         Rename all uses of TARGET_V850E2 to TARGET_V850E2V3_UP.
28256         (movdi): New pattern.
28257         (movdi_internal): New pattern.
28258         (cbranchsf4): Conditionalize on TARGET_USE_FPU.
28259         (cbranchdf4): Conditionalize on TARGET_USE_FPU.
28260         (cstoresf4): Likewise.
28261         (cstoredf4): Likewise.
28262         (insv): New pattern.
28263         (rotlso3_a): New pattern.
28264         (rotlsi3_b): New pattern
28265         (rotlsi3_v850e3v5): New pattern.
28266         (doloop_begin): New pattern.
28267         (fix_loop_counter): New pattern.
28268         (doloop_end): New pattern.
28269         (branch_normal): Add e3v5 long branch support.
28270         (branch_invert): Likewise.
28271         (branch_z_normal): Likewise.
28272         (branch_z_invert): Likewise.
28273         (branch_nz_normal): Likewise.
28274         (branch_nz_invert): Likewise.
28275         (call_internal_short): Add e3v5 register-indirect JARL support.
28276         (call_internal_long): Likewise.
28277         (call_value_internal_short): Likewise.
28278         (call_value_internal_long): Likewise.
28279         * config/v850/v850.opt (mv850e3v5, mv850e2v4): New options.
28280         (mloop): New option.
28281         * config.gcc: Add support for configuring v840e3v5 target.
28282         * doc/invoke.texi: Document new v850 specific command line options.
28284 2013-01-31  Paul Koning  <ni1d@arrl.net>
28286         PR debug/55059
28287         PR debug/54508
28288         * dwarf2out.c (prune_unused_types_mark): Mark all of parent's
28289         children if parent is a class.
28290         (prune_unused_types_prune): Don't add DW_AT_declaration.
28292 2013-01-31  Richard Biener  <rguenther@suse.de>
28294         PR tree-optimization/56157
28295         * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
28296         match up operand with SLP child.
28298 2013-01-31  Jason Merrill  <jason@redhat.com>
28300         PR debug/54410
28301         * dwarf2out.c (gen_struct_or_union_type_die): Always schedule template
28302         parameters the first time.
28303         (gen_scheduled_generic_parms_dies): Check completeness here.
28305 2013-01-31  Richard Biener  <rguenther@suse.de>
28307         PR middle-end/53073
28308         * common.opt (faggressive-loop-optimizations): New flag,
28309         enabled by default.
28310         * doc/invoke.texi (faggressive-loop-optimizations): Document.
28311         * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Guard
28312         infer_loop_bounds_from_undefined by it.
28314 2013-01-31  Richard Biener  <rguenther@suse.de>
28316         PR tree-optimization/56150
28317         * tree-ssa-loop-manip.c (find_uses_to_rename_stmt): Do not
28318         visit virtual operands.
28319         (find_uses_to_rename_bb): Likewise.
28321 2013-01-31  Richard Biener  <rguenther@suse.de>
28323         PR tree-optimization/56150
28324         * tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
28325         mixed store non-store stmts.
28327 2013-01-30  Jakub Jelinek  <jakub@redhat.com>
28329         PR sanitizer/55374
28330         * gcc.c (LIBASAN_SPEC): Define just to ADD_STATIC_LIBASAN_LIBS if
28331         LIBASAN_EARLY_SPEC is defined.
28332         (LIBASAN_EARLY_SPEC): Define to empty string if not already defined.
28333         (LINK_COMMAND_SPEC): Add LIBASAN_EARLY_SPEC for -fsanitize=address,
28334         before %o.
28335         * config/gnu-user.h (LIBASAN_EARLY_SPEC): Define.
28337         PR c++/55742
28338         * config/i386/i386.c (ix86_valid_target_attribute_inner_p): Diagnose
28339         invalid args instead of ICEing on it.
28340         (ix86_valid_target_attribute_tree): Return error_mark_node if
28341         ix86_valid_target_attribute_inner_p failed.
28342         (ix86_valid_target_attribute_p): Return false only if
28343         ix86_valid_target_attribute_tree returned error_mark_node.  Allow
28344         target("default") attribute.
28345         (sorted_attr_string): Change argument from const char * to tree,
28346         merge in all target attribute arguments rather than just one.
28347         Formatting fix.  Use XNEWVEC instead of xmalloc and XDELETEVEC
28348         instead of free.  Avoid using strcat.
28349         (ix86_mangle_function_version_assembler_name): Mangle
28350         target("default") as if no target attribute is present.  Adjust
28351         sorted_attr_string caller.  Avoid leaking memory.  Use XNEWVEC
28352         instead of xmalloc and XDELETEVEC instead of free.
28353         (ix86_function_versions): Don't return true if one of the decls
28354         doesn't have target attribute.  If they don't and one of the decls
28355         is DECL_FUNCTION_VERSIONED, report an error.  Adjust
28356         sorted_attr_string caller.  Use XDELETEVEC instead of free.
28357         (ix86_supports_function_versions): Remove.
28358         (make_name): Fix up formatting.
28359         (make_dispatcher_decl): Remove resolver_name and its initialization.
28360         Avoid leaking memory.
28361         (is_function_default_version): Return true if there is
28362         target("default") attribute rather than no target attribute at all.
28363         (make_resolver_func): Avoid leaking memory.
28364         (ix86_generate_version_dispatcher_body): Likewise.
28365         (TARGET_OPTION_SUPPORTS_FUNCTION_VERSIONS): Remove.
28366         * target.def (supports_function_versions): Remove.
28367         * doc/tm.texi.in (SUPPORTS_FUNCTION_VERSIONS): Remove.
28368         * doc/tm.texi: Regenerated.
28370 2013-01-30  Vladimir Makarov  <vmakarov@redhat.com>
28372         PR rtl-optimization/56144
28373         * lra-constraints.c (get_reload_reg): Don't reuse reload pseudo
28374         for values with side effects.
28376 2013-01-30  Richard Biener  <rguenther@suse.de>
28378         * sparseset.h (sparseset_bit_p): Use gcc_checking_assert.
28379         (sparseset_pop): Likewise.
28380         * cfganal.c (compute_idf): Likewise.  Increase work-stack size
28381         to be able to use quick_push in the worker loop.
28383 2013-01-30  Marek Polacek  <polacek@redhat.com>
28385         * cfgcleanup.c (cleanup_cfg): Don't mark affected BBs.
28387 2013-01-30  Richard Biener  <rguenther@suse.de>
28389         PR lto/56147
28390         * lto-symtab.c (lto_symtab_merge_decls_1): Guard DECL_BUILT_IN check.
28392 2013-01-30  Georg-Johann Lay  <avr@gjlay.de>
28394         PR tree-optimization/56064
28395         * fixed-value.c (fixed_from_double_int): New function.
28396         * fixed-value.h (fixed_from_double_int): New prototype.
28397         (const_fixed_from_double_int): New static inline function.
28398         * fold-const.c (native_interpret_fixed): New static function.
28399         (native_interpret_expr) <FIXED_POINT_TYPE>: Use it.
28400         (can_native_interpret_type_p) <FIXED_POINT_TYPE>: Return true.
28401         (native_encode_fixed): New static function.
28402         (native_encode_expr) <FIXED_CST>: Use it.
28403         (native_interpret_int): Move double_int worker code to...
28404         * double-int.c (double_int::from_buffer): ...this new static method.
28405         * double-int.h (double_int::from_buffer): Prototype it.
28407 2013-01-30  Richard Biener  <rguenther@suse.de>
28409         * tree-ssa-structalias.c (final_solutions, final_solutions_obstack):
28410         New pointer-map and obstack.
28411         (init_alias_vars): Allocate pointer-map and obstack.
28412         (delete_points_to_sets): Free them.
28413         (find_what_var_points_to): Cache result.
28414         (find_what_p_points_to): Adjust for changed interface of
28415         find_what_var_points_to.
28416         (compute_points_to_sets): Likewise.
28417         (ipa_pta_execute): Likewise.
28419 2013-01-30  Rainer Orth  <ro@CeBiTec.Uni-Bielefeld.DE>
28421         * configure.ac (HAVE_AS_SPARC_NOBITS): New test.
28422         * configure: Regenerate.
28423         * config.in: Regenerate.
28424         * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section): Emit
28425         #nobits/#progbits if supported.
28427 2013-01-29  Oleg Endo  <olegendo@gcc.gnu.org>
28429         PR target/56121
28430         * config/sh/sh.md (bclr_m2a, bset_m2a, bst_m2a, bld_m2a, bldsign_m2a,
28431         bld_reg, *bld_regqi, band_m2a, bandreg_m2a, bor_m2a, borreg_m2a,
28432         bxor_m2a, bxorreg_m2a): Add satisfies_constraint_K03 condition.
28434 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28436         * config/arm/cortex-a7.md (cortex_a7_neon, cortex_a7_all): Remove.
28437         (cortex_a7_idiv): Use cortex_a7_both instead of cortex_a7_all.
28439 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28441         * config/arm/arm.c (cortexa7_younger): Return true for TYPE_CALL.
28442         * config/arm/cortex-a7.md (cortex_a7_call): Update required units.
28444 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28446         * config/arm/arm-protos.h (arm_mac_accumulator_is_result): New
28447         declaration.
28448         * config/arm/arm.c (arm_mac_accumulator_is_result): New function.
28449         * config/arm/cortex-a7.md: New bypasses using
28450         arm_mac_accumulator_is_result.
28452 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28454         * config/arm/cortex-a7.md (cortex_a7_neon_mul):  New reservation.
28455         (cortex_a7_neon_mla): Likewise.
28456         (cortex_a7_fpfmad): New reservation.
28457         (cortex_a7_fpmacs): Use ffmas and update required units.
28458         (cortex_a7_fpmuld): Update required units and latency.
28459         (cortex_a7_fpmacd): Likewise.
28460         (cortex_a7_fdivs, cortex_a7_fdivd): Likewise.
28461         (cortex_a7_neon). Likewise.
28462         (bypass) Update participating units.
28464 2013-01-29  Greta Yorsh  <Greta.Yorsh@arm.com>
28466         * config/arm/arm.md (type): Add ffmas and ffmad to "type" attribute.
28467         * config/arm/vfp.md (fma,fmsub,fnmsub,fnmadd): Change type
28468         from fmac to ffma.
28469         * config/arm/vfp11.md (vfp_farith): Use ffmas.
28470         (vfp_fmul): Use ffmad.
28471         * config/arm/cortex-r4f.md (cortex_r4_fmacs): Use ffmas.
28472         (cortex_r4_fmacd): Use ffmad.
28473         * config/arm/cortex-m4-fpu.md (cortex_m4_fmacs): Use ffmas.
28474         * config/arm/cortex-a9.md (cortex_a9_fmacs):  Use ffmas.
28475         (cortex_a9_fmacd): Use ffmad.
28476         * config/arm/cortex-a8-neon.md (cortex_a8_vfp_macs): Use ffmas.
28477         (cortex_a8_vfp_macd): Use ffmad.
28478         * config/arm/cortex-a5.md (cortex_a5_fpmacs): Use ffmas.
28479         (cortex_a5_fpmacd): Use ffmad.
28480         * config/arm/cortex-a15-neon.md (cortex_a15_vfp_macs) Use ffmas.
28481         (cortex_a15_vfp_macd): Use ffmad.
28482         * config/arm/arm1020e.md (v10_fmul): Use ffmas and ffmad.
28484 2013-01-29  Jason Merrill  <jason@redhat.com>
28486         PR libstdc++/54314
28487         * varasm.c (default_assemble_visibility): Don't warn about
28488         visibility on artificial decls.
28490 2013-01-29  Richard Biener  <rguenther@suse.de>
28492         PR tree-optimization/56113
28493         * tree-ssa-structalias.c (equiv_class_lookup): Also return
28494         the bitmap leader.
28495         (label_visit): Free duplicate bitmaps and record the leader instead.
28496         (perform_var_substitution): Adjust.
28498 2013-01-29  Richard Biener  <rguenther@suse.de>
28500         PR tree-optimization/55270
28501         * tree-ssa-dom.c (eliminate_degenerate_phis): If we changed
28502         the CFG, schedule loops for fixup.
28504 2013-01-29  Nick Clifton  <nickc@redhat.com>
28506         * config/rl78/rl78.c (rl78_regno_mode_code_ok_for_base_p): Allow
28507         SP_REG.
28509 2013-01-28  Leif Ekblad  <leif@rdos.net>
28511         * config.gcc (i[34567]86-*-rdos*, x86_64-*-rdos*): New targets.
28512         * config/i386/i386.h (TARGET_RDOS): New macro.
28513         (DEFAULT_LARGE_SECTION_THRESHOLD): New macro.
28514         * config/i386/i386.c (ix86_option_override_internal): For 64bit
28515         TARGET_RDOS, set ix86_cmodel to CM_MEDIUM_PIC and flag_pic to 1.
28516         * config/i386/i386.opt (mlarge-data-threshold): Initialize to
28517         DEFAULT_LARGE_SECTION_THRESHOLD.
28518         * config/i386/i386.md (R14_REG, R15_REG): New constants.
28519         * config/i386/rdos.h: New file.
28520         * config/i386/rdos64.h: New file.
28522 2013-01-28  Bernd Schmidt  <bernds@codesourcery.com>
28524         PR other/54814
28525         * reload.c (find_valid_class_1): Use in_hard_reg_set_p instead of
28526         TEST_HARD_REG_BIT.
28528 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28530         PR rtl-optimization/56117
28531         * sched-deps.c (sched_analyze_2) <case PREFETCH>: For use_cselib
28532         call cselib_lookup_from_insn on the MEM before calling
28533         add_insn_mem_dependence.
28535 2013-01-28  Richard Biener  <rguenther@suse.de>
28537         * tree-inline.c (remap_gimple_stmt): Do not assing a BLOCK
28538         to a stmt that didn't have one.
28539         (copy_phis_for_bb): Likewise for PHI arguments.
28540         (copy_debug_stmt): Likewise for debug stmts.
28542 2013-01-28  Richard Biener  <rguenther@suse.de>
28544         PR tree-optimization/56034
28545         * tree-loop-distribution.c (enum partition_kind): Add PKIND_REDUCTION.
28546         (partition_builtin_p): Adjust.
28547         (generate_code_for_partition): Handle PKIND_REDUCTION.  Assert
28548         it is the last partition.
28549         (rdg_flag_uses): Check SSA_NAME_IS_DEFAULT_DEF before looking
28550         up the vertex for the definition.
28551         (classify_partition): Classify whether a partition is a
28552         PKIND_REDUCTION, thus has uses outside of the loop.
28553         (ldist_gen): Inherit PKIND_REDUCTION when merging partitions.
28554         Merge all PKIND_REDUCTION partitions into the last partition.
28555         (tree_loop_distribution): Seed partitions from reductions as well.
28557 2013-01-28  Jakub Jelinek  <jakub@redhat.com>
28559         PR tree-optimization/56125
28560         * tree-ssa-math-opts.c (gimple_expand_builtin_pow): Don't optimize
28561         pow(x,c) into sqrt(x) * powi(x, n/2) or
28562         1.0 / (sqrt(x) * powi(x, abs(n/2))) if c is an integer or when
28563         optimizing for size.
28564         Don't optimize pow(x,c) into powi(x, n/3) * powi(cbrt(x), n%3) or
28565         1.0 / (powi(x, abs(n)/3) * powi(cbrt(x), abs(n)%3)) if 2c is an
28566         integer.
28568         PR tree-optimization/56094
28569         * gimplify.c (force_gimple_operand_1): Temporarily set input_location
28570         to UNKNOWN_LOCATION while gimplifying expr.
28572 2013-01-27  Uros Bizjak  <ubizjak@gmail.com>
28574         PR target/56114
28575         * config/i386/i386.md (*movabs<mode>_1): Add square brackets around
28576         operand 0 in movabs insn template for -masm=intel asm alternative.
28577         (*movabs<mode>_2): Ditto for operand 1.
28579 2013-01-26  David Holsgrove  <david.holsgrove@xilinx.com>
28581         PR target/54663
28582         * config.gcc (microblaze*-linux*): Add tmake_file to allow building
28583         of microblaze-c.o
28585 2013-01-26  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28587         * config.gcc (microblaze*-*-*): Rename microblaze*-*-elf, update
28588         tm_file.
28590 2013-01-25  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
28592         * config/aarch64/aarch64.c (TARGET_FIXED_CONDITION_CODE_REGS):
28593         Undef to avoid warning.
28595 2013-01-25  Michael Haubenwallner  <michael.haubenwallner@salomon.at>
28597         * configure.ac (gcc_cv_ld_static_dynamic): Define for AIX native ld.
28598         * configure: Regenerate.
28600 2013-01-25  Jakub Jelinek  <jakub@redhat.com>
28602         PR tree-optimization/56098
28603         * tree-ssa-phiopt.c (nt_init_block): Don't call add_or_mark_expr
28604         for stmts with volatile ops.
28605         (cond_store_replacement): Don't optimize if assign has volatile ops.
28606         (cond_if_else_store_replacement_1): Don't optimize if either
28607         then_assign or else_assign have volatile ops.
28608         (hoist_adjacent_loads): Don't optimize if either def1 or def2 have
28609         volatile ops.
28611 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28613         * doc/invoke.texi (AVR Built-in Macros): Document __XMEGA__.
28615 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28617         * doc/extend.texi (Example of asm with clobbered asm reg): Fix
28618         missing ':' in asm example.
28620 2013-01-25  Tejas Belagod  <tejas.belagod@arm.com>
28622         * config/aarch64/aarch64-simd-builtins.def: Separate sq<r>dmulh_lane
28623         entries into lane and laneq entries.
28624         * config/aarch64/aarch64-simd.md (aarch64_sq<r>dmulh_lane<mode>):
28625         Remove AdvSIMD scalar modes.
28626         (aarch64_sq<r>dmulh_laneq<mode>): New.
28627         (aarch64_sq<r>dmulh_lane<mode>): New RTL pattern for Scalar AdvSIMD
28628         modes.
28629         * config/aarch64/arm_neon.h: Fix all the vq<r>dmulh_lane* intrinsics'
28630         builtin implementations to relfect changes in RTL in aarch64-simd.md.
28631         * config/aarch64/iterators.md (VCOND): New.
28632         (VCONQ): New.
28634 2013-01-25  Georg-Johann Lay  <avr@gjlay.de>
28636         PR target/54222
28637         * config/avr/builtins.def (DEF_BUILTIN): Add LIBNAME argument.
28638         Add NULL LIBNAME argument to existing definitions.
28639         (ABSHR, ABSR, ABSLR, ABSLLR, ABSHK, ABSK, ABSLK, ABSLLK): New.
28640         * config/avr/avr-c.c (DEF_BUILTIN): Add LIBNAME argument.
28641         * config/avr/avr.c (DEF_BUILTIN): Same.
28642         (avr_init_builtins): Pass down LIBNAME to add_builtin_function.
28643         (avr_expand_builtin): Expand to a vanilla call if a libgcc
28644         implementation is available (DECL_ASSEMBLER_NAME is set).
28645         (avr_fold_absfx): New static function.
28646         (avr_fold_builtin): Use it to handle: AVR_BUILTIN_ABSHR,
28647         AVR_BUILTIN_ABSR, AVR_BUILTIN_ABSLR, AVR_BUILTIN_ABSLLR,
28648         AVR_BUILTIN_ABSHK, AVR_BUILTIN_ABSK, AVR_BUILTIN_ABSLK,
28649         AVR_BUILTIN_ABSLLK.
28650         * config/avr/stdfix.h (abshr, absr, abslr, absllr)
28651         (abshk, absk, abslk, absllk): Provide as static inline functions.
28653 2013-01-25  Marek Polacek  <polacek@redhat.com>
28655         PR tree-optimization/56035
28656         * cfgloopmanip.c (fix_loop_structure): Remove redundant condition.
28658 2012-01-24  Uros Bizjak  <ubizjak@gmail.com>
28660         * config/i386/i386.md (*movti_internal_rex64): Add (o,e) alternative.
28661         (*movtf_internal_rex64): Add (!o,C) alternative
28662         (*movxf_internal_rex64): Ditto.
28663         (*movdf_internal_rex64): Add (?r,C) and (?m,C) alternatives.
28665 2013-01-24  Shenghou Ma  <minux.ma@gmail.com>
28667         * doc/invoke.texi: fix typo.
28668         * doc/objc.texi: fix typo.
28670 2013-01-24  Richard Sandiford  <rdsandiford@googlemail.com>
28672         * config/mips/mips.md (*and<mode>3_mips16): Use the "W" constraint
28673         for the first two alternatives.
28675 2013-01-24  Diego Novillo  <dnovillo@google.com>
28677         * Makefile.in (GGC): Remove.  Replace all instances with ggc-page.o.
28678         (ggc-zone.o): Remove.
28679         * configure.ac: Remove option --with-gc.
28680         * configure: Re-generate.
28681         * doc/install.texi: Remove documentation for --with-gc.
28682         * gengtype.c (write_enum_defn): Remove.  Update all users.
28683         (write_Types_process_field): Remove generation of gt_e_* argument.
28684         (output_type_enum): Remove.  Update all users.
28685         (write_enum_defn): Remove.  Update all users.
28686         (enum alloc_zone): Remove.  Update all users.
28687         (write_splay_tree_allocator_def): Remove generation of gt_e_* argument.
28688         * ggc-common.c (ggc_splay_alloc): Remove first argument.
28689         Update all callers.
28690         (struct ptr_data): Remove field TYPE.  Update all users.
28691         (gt_pch_note_object): Remove argument TYPE.  Update all users.
28692         * ggc-internal.h (ggc_pch_alloc_object): Remove last argument.
28693         Update all users.
28694         * ggc-none.c (ggc_alloc_typed_stat): Remove.
28695         (struct alloc_zone): Remove.
28696         (ggc_internal_alloc_zone_stat): Remove.
28697         (ggc_internal_cleared_alloc_zone_stat): Remove.
28698         * ggc-page.c (ggc_alloc_typed_stat): Remove.
28699         (ggc_pch_count_object): Remove last argument.  Update all users.
28700         (ggc_pch_alloc_object): Remove last argument.  Update all users.
28701         (struct alloc_zone): Remove.
28702         * ggc-zone.c: Remove.
28703         * ggc.h (gt_pch_note_object): Remove last argument.  Update all users.
28704         (struct alloc_zone): Remove.
28705         (ggc_alloc_typed_stat): Remove.
28706         (ggc_alloc_typed): Remove.
28707         (ggc_splay_alloc): Remove first argument.
28708         (rtl_zone): Remove.  Update all users.
28709         (tree_zone): Remove.  Update all users.
28710         (tree_id_zone): Remove.  Update all users.
28711         (ggc_internal_zone_alloc_stat): Remove.  Update all users.
28712         (ggc_internal_zone_cleared_alloc_stat): Remove.  Update all users.
28713         (ggc_internal_zone_vec_alloc_stat): Remove.  Update all users.
28714         * tree-ssanames.c: Remove references to zone allocator in comments.
28716 2013-01-24  Georg-Johann Lay  <avr@gjlay.de>
28718         * config/avr/avr.c (avr_out_fract): Make register numbers that
28719         might be outside of source operand signed.
28721 2013-01-24  Uros Bizjak  <ubizjak@gmail.com>
28723         * config/i386/constraints.md (Yf): New constraint.
28724         * config/i386/i386.md (*movdf_internal_rex64): Use Yf*f instead
28725         of f constraint to conditionaly disable x87 register preferences.
28726         (*movdf_internal): Ditto.
28727         (*movsf_internal): Ditto.
28729 2013-01-24  Steven Bosscher  <steven@gcc.gnu.org>
28731         PR inline-asm/55934
28732         * lra-assigns.c (assign_by_spills): Throw away the pattern of asms
28733         that have operands with impossible constraints.
28734         Add a FIXME for a speed-up opportunity.
28735         * lra-constraints.c (process_alt_operands): Verify that a class
28736         selected from constraints on asms is valid for the operand mode.
28737         (curr_insn_transform): Remove incorrect comment.
28739 2013-01-23  David Edelsohn  <dje.gcc@gmail.com>
28741         * config/rs6000/rs6000.c (rs6000_delegitimize_address): Check that
28742         TOC operand is a valid symbol ref in the constant pool.
28744 2013-01-23  Edgar E. Iglesias  <edgar.iglesias@gmail.com>
28746         * config/microblaze/linux.h: Add TARGET_OS_CPP_BUILTINS
28748 2013-01-23  Georg-Johann Lay  <avr@gjlay.de>
28750         PR target/54222
28751         * config/avr/stdfix.h: New file.
28752         * t-avr (stdfix-gcc.h): New rule to build it.
28753         (EXTRA_HEADERS): Set it to install stdfix.h, stdfix-gcc.h.
28755 2013-01-23  Kostya Serebryany  <kcc@google.com>
28757         * config/darwin.h: remove dependency on
28758         CoreFoundation (asan on Mac OS).
28760 2013-01-23  Jakub Jelinek  <jakub@redhat.com>
28762         PR target/49069
28763         * config/arm/arm.md (cbranchdi4, cstoredi4): Use s_register_operand
28764         instead of cmpdi_operand for first comparison operand.
28765         Don't assert that comparison operands aren't both constants.
28767 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28769         * doc/install.texi (Downloading the Source): Update references to
28770         downloading separate components.
28772 2013-01-22  Jonathan Wakely  <jwakely.gcc@gmail.com>
28774         * doc/extend.texi (__int128): Improve grammar.
28776 2013-01-22  Uros Bizjak  <ubizjak@gmail.com>
28778         PR target/56028
28779         * config/i386/i386.md (*movti_internal_rex64): Change (o,riF)
28780         alternative to (o,r).
28781         (*movdi_internal_rex64): Remove (!o,n) alternative.
28782         (DImode immediate->memory splitter): Remove.
28783         (DImode immediate->memory peephole2): Remove.
28784         (movtf): Enable for TARGET_64BIT || TARGET_SSE.
28785         (*movtf_internal_rex64): Rename from *movtf_internal. Change (!o,F*r)
28786         alternative to (!o,*r).
28787         (*movtf_internal_sse): New pattern.
28788         (*movxf_internal_rex64): New pattern.
28789         (*movxf_internal): Disable for TARGET_64BIT.
28790         (*movdf_internal_rex64): Remove (!o,F) alternative.
28792 2013-01-22  Jakub Jelinek  <jakub@redhat.com>
28794         PR middle-end/56074
28795         * dumpfile.c (dump_loc): Only print loc if LOCATION_LOCUS (loc)
28796         isn't UNKNOWN_LOCATION nor BUILTINS_LOCATION.
28797         * tree-vect-loop-manip.c (find_loop_location): Also ignore
28798         stmt locations where LOCATION_LOCUS of the stmt location is
28799         UNKNOWN_LOCATION or BUILTINS_LOCATION.
28801         PR target/55686
28802         * config/i386/i386.md (UNSPEC_STOS): New.
28803         (strset_singleop, *strsetdi_rex_1, *strsetsi_1, *strsethi_1,
28804         *strsetqi_1): Add UNSPEC_STOS.
28806 2013-01-22  Paolo Carlini  <paolo.carlini@oracle.com>
28808         PR c++/56067
28809         * doc/invoke.texi: Remove left over -Wsynth example.
28811 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28813         PR tree-optimization/56051
28814         * fold-const.c (fold_binary_loc): Don't fold
28815         X < (cast) (1 << Y) into (X >> Y) != 0 if cast is either
28816         a narrowing conversion, or widening conversion from signed
28817         to unsigned.
28819 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28821         PR rtl-optimization/56023
28822         * haifa-sched.c (fix_inter_tick): Do not update ticks of instructions,
28823         dependent on debug instruction.
28825 2013-01-21  Martin Jambor  <mjambor@suse.cz>
28827         PR middle-end/56022
28828         * function.c (allocate_struct_function): Call
28829         invoke_set_current_function_hook earlier.
28831 2013-01-21  Jakub Jelinek  <jakub@redhat.com>
28833         * reload1.c (init_reload): Only initialize reload_obstack
28834         during the first call.
28836 2013-01-21  Marek Polacek  <polacek@redhat.com>
28838         * cfgloop.c (verify_loop_structure): Fix up grammar.
28840 2013-01-21  Yi-Hsiu Hsu  <ahsu@marvell.com>
28842         * config/arm/marvell-pj4.md (pj4_shift_conds, pj4_alu_shift,
28843         pj4_alu_shift_conds, pj4_shift): Handle simple_alu_shift.
28845 2013-01-21  Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28847         PR target/56058
28848         * config/arm/marvell-pj4.md: Update copyright year.
28849         Fix up use of alu to alu_reg and simple_alu_imm.
28851 2013-01-21  Uros Bizjak  <ubizjak@gmail.com>
28853         * config/i386/i386.md (enabled): Do not disable fma4 for TARGET_FMA.
28855 2013-01-20  Vladimir Makarov  <vmakarov@redhat.com>
28857         PR target/55433
28858         * lra-constraints.c (curr_insn_transform): Don't reuse original
28859         insn for secondary memory move when memory mode should be different.
28861 2013-01-20  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
28863         * config/pa/pa.md (atomic_loaddi, atomic_loaddi_1, atomic_storedi,
28864         atomic_storedi_1): New patterns.
28866 2013-01-20  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
28868         btver2 pipeline descriptions.
28869         * config/i386/i386.c: Enable CPU_BTVER2 to use btver2 pipeline
28870         descriptions.
28871         * config/i386/i386.md (btver2_decode): New type attributes.
28872         * config/i386/sse.md (btver2_decode, btver2_sse_attr): New
28873         type attributes.
28874         * config/i386/btver2.md: New file describing btver2 pipelines.
28876 2013-01-19  Andrew Pinski  <apinski@cavium.com>
28878         PR tree-optimization/52631
28879         * tree-ssa-sccvn (visit_use): Before looking up the original
28880         statement, try looking up the simplified expression.
28882 2013-01-19  Anthony Green  <green@moxielogic.com>
28884         * config/moxie/moxie.c (moxie_expand_prologue): Set
28885         current_function_static_stack_size.
28887 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28889         PR tree-optimization/56029
28890         * tree-phinodes.c (reserve_phi_args_for_new_edge): Set
28891         gimple_phi_arg_location for the new arg to UNKNOWN_LOCATION.
28893 2013-01-18  Sharad Singhai  <singhai@google.com>
28895         PR tree-optimization/55995
28896         * dumpfile.c (dump_loc): Print location only if available.
28897         * tree-vectorizer.c (increase_alignment): Intialize vect_location.
28899 2013-01-18  Vladimir Makarov  <vmakarov@redhat.com>
28901         PR target/55433
28902         * lra-constraints.c (curr_insn_transform): Reuse original insn for
28903         secondary memory move.
28904         (inherit_reload_reg): Use rclass instead of cl for
28905         check_secondary_memory_needed_p.
28907 2013-01-18  Jakub Jelinek  <jakub@redhat.com>
28909         PR middle-end/56015
28910         * expr.c (expand_expr_real_2) <case COMPLEX_EXPR>: Handle
28911         the case where writing real complex part of target modifies op1.
28913 2013-01-18  James Greenhalgh  <james.greenhalgh@arm.com>
28915         * config/aarch64/aarch64-simd.md
28916         (aarch64_vcond_internal<mode>): Handle unordered cases.
28917         * config/aarch64/iterators.md (v_cmp_result): New.
28919 2013-01-18  Yi-Hsiu Hsu  <ahsu@marvell.com>
28920             Ramana Radhakrishnan  <ramana.radhakrishnan@arm.com>
28922         * config/arm/marvell-pj4.md: New file.
28923         * config/arm/arm.c (arm_issue_rate): Add marvell_pj4.
28924         * config/arm/arm.md (generic_sched): Add marvell_pj4.
28925         (generic_vfp): Likewise.
28926         * config/arm/arm-cores.def: Add marvell-pj4.
28927         * config/arm/arm-tune.md: Regenerate.
28928         * config/arm/arm-tables.opt: Regenerate.
28929         * config/arm/bpabi.h (BE8_LINK_SPEC): Add marvell_pj4.
28930         * doc/invoke.texi: Document marvell-pj4.
28932 2013-01-18  Tejas Belagod  <tejas.belagod@arm.com>
28934         * config/aarch64/arm_neon.h: Map scalar types to standard types.
28936 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28938         PR debug/54114
28939         PR debug/54402
28940         PR debug/49888
28941         * var-tracking.c (negative_power_of_two_p): New.
28942         (global_get_addr_cache, local_get_addr_cache): New.
28943         (get_addr_from_global_cache, get_addr_from_local_cache): New.
28944         (vt_canonicalize_addr): Rewrite using the above.  Adjust the
28945         heading comment.
28946         (vt_stack_offset_p): Remove.
28947         (vt_canon_true_dep): Always canonicalize loc's address.
28948         (clobber_overlapping_mems): Make sure we have a MEM.
28949         (local_get_addr_clear_given_value): New.
28950         (val_reset): Clear local cached entries.
28951         (compute_bb_dataflow): Create and release the local cache.
28952         Disable duplicate MEMs clobbering.
28953         (emit_notes_in_bb): Clobber MEMs likewise.
28954         (vt_emit_notes): Create and release the local cache.
28955         (vt_initialize, vt_finalize): Create and release the global
28956         cache, respectively.
28957         * alias.c (rtx_equal_for_memref_p): Compare operands of ENTRY_VALUEs.
28959 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28961         PR libmudflap/53359
28962         * tree-mudflap.c (mudflap_finish_file): Skip deferred decls
28963         not found in the symtab.
28965 2013-01-18  Alexandre Oliva  <aoliva@redhat.com>
28967         PR debug/56006
28968         PR rtl-optimization/55547
28969         PR rtl-optimization/53827
28970         PR debug/53671
28971         PR debug/49888
28972         * alias.c (offset_overlap_p): New, factored out of...
28973         (memrefs_conflict_p): ... this.  Use absolute sizes.  Retain
28974         the conservative special case for symbolic constants.  Don't
28975         adjust zero sizes on alignment.
28977 2013-01-18  Bernd Schmidt  <bernds@codesourcery.com>
28979         PR rtl-optimization/52573
28980         * regrename.c (build_def_use): Ignore REG_DEAD notes if there is a
28981         REG_UNUSED for the same register.
28983 2013-01-17  Richard Biener  <rguenther@suse.de>
28984             Marek Polacek  <polacek@redhat.com>
28986         PR rtl-optimization/55833
28987         * loop-unswitch.c (unswitch_loops): Move loop verification...
28988         (unswitch_single_loop): ...here.  Call mark_irreducible_loops.
28989         * cfgloopmanip.c (fix_loop_placement): Add IRRED_INVALIDATED parameter.
28990         Set it to true when we're removing a loop from hierarchy tree in
28991         an irreducible region.
28992         (fix_bb_placements): Adjust caller.
28993         (fix_loop_placements): Likewise.
28995 2013-01-17  Georg-Johann Lay  <avr@gjlay.de>
28997         * config/avr/builtins.def (DEF_BUILTIN): Factor out
28998         "__builtin_avr_" from NAME, turn NAME to an uppercase identifier.
28999         Factor out 'CODE_FOR_' from ICODE, use 'nothing' instead of '-1'.
29000         Remove ID.  Adjust comments.
29001         * config/avr/avr-c.c (avr_builtin_name): Remove.
29002         (avr_cpu_cpp_builtins): Use DEF_BUILTIN instead of for-loop.
29003         * config/avr/avr.c (avr_tolower): New static function.
29004         (DEF_BUILTIN): Remove parameter ID.  Prefix ICODE by 'CODE_FOR_'.
29005         Stringify NAME, prefix it with "__builtin_avr_" and lowercase it.
29006         (avr_expand_builtin): Assert insn_code != CODE_FOR_nothing for
29007         default expansion.
29009 2013-01-17  Jan Hubicka  <jh@suse.cz>
29011         PR tree-optimization/55273
29012         * loop-iv.c (iv_number_of_iterations): Consider zero iteration case.
29014 2013-01-17  Uros Bizjak  <ubizjak@gmail.com>
29016         PR target/55981
29017         * config/i386/sync.md (atomic_store<mode>): Always generate SWImode
29018         store through atomic_store<mode>_1.
29019         (atomic_store<mode>_1): Macroize insn using SWI mode iterator.
29021 2013-01-17  Martin Jambor  <mjambor@suse.cz>
29023         PR tree-optimizations/55264
29024         * ipa-inline-transform.c (can_remove_node_now_p_1): Never return true
29025         for virtual methods.
29026         * ipa.c (symtab_remove_unreachable_nodes): Never return true for
29027         virtual methods before inlining is over.
29028         * cgraph.h (cgraph_only_called_directly_or_aliased_p): Return false for
29029         virtual functions.
29030         * cgraphclones.c (cgraph_create_virtual_clone): Mark clones as
29031         non-virtual.
29033 2013-01-16  Vladimir Makarov  <vmakarov@redhat.com>
29035         PR rtl-optimization/56005
29036         * sched-deps.c (sched_analyze_2): Check deps->readonly for adding
29037         pending reads for prefetch.
29039 2013-01-16  Ian Bolton  <ian.bolton@arm.com>
29041         * config/aarch64/aarch64.md
29042         (*cstoresi_neg_uxtw): New pattern.
29043         (*cmovsi_insn_uxtw): New pattern.
29044         (*<optab>si3_uxtw): New pattern.
29045         (*<LOGICAL:optab>_<SHIFT:optab>si3_uxtw): New pattern.
29046         (*<optab>si3_insn_uxtw): New pattern.
29047         (*bswapsi2_uxtw): New pattern.
29049 2013-01-16  Richard Biener  <rguenther@suse.de>
29051         * tree-inline.c (tree_function_versioning): Remove set but
29052         never used variable.
29054 2013-01-16  Richard Biener  <rguenther@suse.de>
29056         PR tree-optimization/55964
29057         * tree-flow.h (rename_variables_in_loop): Remove.
29058         (rename_variables_in_bb): Likewise.
29059         * tree-loop-distribution.c (update_phis_for_loop_copy): Remove.
29060         (copy_loop_before): Adjust and delete update-ssa status.
29061         * tree-vect-loop-manip.c (rename_variables_in_bb): Make static.
29062         (rename_variables_in_bb): Likewise.  Properly walk over predecessors.
29063         (rename_variables_in_loop): Remove.
29064         (slpeel_update_phis_for_duplicate_loop): Likewise.
29065         (slpeel_tree_duplicate_loop_to_edge_cfg): Handle nested loops,
29066         use available cfg machinery instead of duplicating it.
29067         Update PHI nodes and perform poor-mans SSA update here.
29068         (slpeel_tree_peel_loop_to_edge): Adjust.
29070 2013-01-16  Richard Biener  <rguenther@suse.de>
29072         PR tree-optimization/54767
29073         PR tree-optimization/53465
29074         * tree-vrp.c (vrp_meet_1): Revert original fix for PR53465.
29075         (vrp_visit_phi_node): For PHI arguments coming via backedges
29076         drop all symbolical range information.
29077         (execute_vrp): Compute backedges.
29079 2013-01-16  Richard Biener  <rguenther@suse.de>
29081         * doc/install.texi: Update CLooG and ISL requirements to
29082         0.18.0 and 0.11.1.
29084 2013-01-16  Christian Bruel  <christian.bruel@st.com>
29086         PR target/55301
29087         * config/sh/sh.c (sh_expand_prologue): Postpone new_stack mem symbol.
29088         (broken_move): Handle UNSPECV_SP_SWITCH_B.
29089         * config/sh/sh.md (sp_switch_1): Use set (reg:SI SP_REG).
29091 2013-01-16  DJ Delorie  <dj@redhat.com>
29093         * config/sh/sh.md (UNSPECV_SP_SWITCH_B): New.
29094         (UNSPECV_SP_SWITCH_E): New.
29095         (sp_switch_1): Change to an unspec.
29096         (sp_switch_2): Change to an unspec.  Don't use post-inc when we
29097         replace $r15.
29099 2013-01-16  Uros Bizjak  <ubizjak@gmail.com>
29101         * emit-rtl.c (need_atomic_barrier_p): Mask memory model argument
29102         with MEMMODEL_MASK before comparing with MEMMODEL_* memory types.
29103         * optabs.c (maybe_emit_sync_lock_test_and_set): Ditto.
29104         (expand_mem_thread_fence): Ditto.
29105         (expand_mem_signal_fence): Ditto.
29106         (expand_atomic_load): Ditto.
29107         (expand_atomic_store): Ditto.
29109 2013-01-16  Alexandre Oliva  <aoliva@redhat.com>
29111         PR rtl-optimization/55547
29112         PR rtl-optimization/53827
29113         PR debug/53671
29114         PR debug/49888
29115         * alias.c (memrefs_conflict_p): Set sizes to negative after
29116         AND adjustments.
29118 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29120         PR target/55940
29121         * function.c (thread_prologue_and_epilogue_insns): Always
29122         add crtl->drap_reg to set_up_by_prologue.set, even if
29123         stack_realign_drap is false.
29125 2013-01-15  Jan-Benedict Glaw  <jbglaw@lug-owl.de>
29127         * config/vax/vax.md (add<mode>3, sub<mode>3, mul<mode>3, div<mode>3,
29128         and<mode>3, *and<mode>_const_int, ior<mode>3, xor<mode>3, ashrsi3,
29129         *call): Fix indention.
29131 2013-01-15  Tom de Vries  <tom@codesourcery.com>
29133         PR target/55876
29134         * optabs.c (widen_operand): Use gen_lowpart instead of gen_rtx_SUBREG.
29135         Update comment.
29137 2013-01-15  Vladimir Makarov  <vmakarov@redhat.com>
29139         PR rtl-optimization/55153
29140         * sched-deps.c (sched_analyze_2): Add pending reads for prefetch.
29142 2013-01-15  Martin Jambor  <mjambor@suse.cz>
29144         PR tree-optimization/55920
29145         * tree-sra.c (analyze_access_subtree): Do not mark non-removable
29146         accesses as grp_to_be_debug_replaced.
29148 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29150         PR tree-optimization/55920
29151         * tree-sra.c (sra_modify_assign): If for lacc->grp_to_be_debug_replaced
29152         there is non-useless type conversion needed from debug rhs to lhs,
29153         use build_debug_ref_for_model and/or VIEW_CONVERT_EXPR.
29155 2013-01-15  Joseph Myers  <joseph@codesourcery.com>
29156             Mikael Pettersson  <mikpe@it.uu.se>
29158         PR target/43961
29159         * config/arm/arm.h (ADDR_VEC_ALIGN): Align SImode jump tables for
29160         Thumb.
29161         (ASM_OUTPUT_CASE_LABEL): Remove.
29162         (ASM_OUTPUT_BEFORE_CASE_LABEL): Define to empty.
29163         * final.c (shorten_branches): Update alignment of labels before
29164         jump tables if CASE_VECTOR_SHORTEN_MODE.
29166 2013-01-15  Richard Biener  <rguenther@suse.de>
29168         PR bootstrap/55961
29169         * system.h: Do not include gmp.h for building host tools.
29171 2013-01-15  Richard Biener  <rguenther@suse.de>
29173         PR middle-end/55882
29174         * emit-rtl.c (set_mem_attributes_minus_bitpos): Correctly
29175         account for bitpos when computing alignment.
29177 2013-01-15  Vladimir Yakovlev  <vladimir.b.yakovlev@intel.com>
29179         * config/i386/i386-c.c (ix86_target_macros_internal): New case.
29180         (ix86_target_macros_internal): Likewise.
29182         * config/i386/i386.c (m_CORE2I7): Removed.
29183         (m_CORE_HASWELL): New macro.
29184         (m_CORE_ALL): Likewise.
29185         (initial_ix86_tune_features): m_CORE2I7 is replaced by m_CORE_ALL.
29186         (initial_ix86_arch_features): Likewise.
29187         (processor_target_table): Initializations for Core avx2.
29188         (cpu_names): New names "core-avx2".
29189         (ix86_option_override_internal): Changed PROCESSOR_COREI7 by
29190         PROCESSOR_CORE_HASWELL.
29191         (ix86_issue_rate): New case.
29192         (ia32_multipass_dfa_lookahead): Likewise.
29193         (ix86_sched_init_global): Likewise.
29195         * config/i386/i386.h (TARGET_HASWELL): New macro.
29196         (target_cpu_default): New TARGET_CPU_DEFAULT_haswell.
29197         (processor_type): New PROCESSOR_HASWELL.
29199 2013-01-15  Jakub Jelinek  <jakub@redhat.com>
29201         PR tree-optimization/55955
29202         * tree-vect-loop.c (vectorizable_reduction): Give up early on
29203         *SHIFT_EXPR and *ROTATE_EXPR codes.
29205         PR tree-optimization/48766
29206         * opts.c (common_handle_option): For -fwrapv disable -ftrapv, for
29207         -ftrapv disable -fwrapv.
29209 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29211         PR target/55974
29212         * config/avr/avr-c.c (avr_cpu_cpp_builtins): Define __FLASH
29213         etc. to 1 and not to __flash.
29214         Use LL suffix for __INT24_MAX__ with -mint8.
29215         Use ULL suffix for __UINT24_MAX__ with -mint8.
29217 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29219         * config/avr/avr-arch.h
29220         (struct base_arch_s): Use typedef avr_arch_t instead.
29221         (struct arch_info_s): Use typedef avr_arch_info_t instead.
29222         (struct mcu_type_s): Use typedef avr_mcu_t instead.
29223         * config/avr/avr.c: Same.
29224         * config/avr/avr-devices.c: Same.
29225         * config/avr/driver-avr.c: Same.
29226         * config/avr/gen-avr-mmcu-texi.c: Same.
29227         * config/avr/avr-mcus.def: Adjust comment.
29229 2013-01-14  Tejas Belagod  <tejas.belagod@arm.com>
29231         * config/aarch64/aarch64-simd.md (*aarch64_simd_ld1r<mode>): New.
29232         * config/aarch64/iterators.md (VALLDI): New.
29234 2013-01-14  Uros Bizjak  <ubizjak@gmail.com>
29235             Andi Kleen  <ak@linux.intel.com>
29237         PR target/55948
29238         * config/i386/sync.md (atomic_store<mode>_1): New pattern.
29239         (atomic_store<mode>): Call atomic_store<mode>_1 for IX86_HLE_RELEASE
29240         memmodel flag.
29242 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29244         * config/avr/avr-stdint.h: Remove trailing blanks.
29245         * config/avr/avr-log.h: Same.
29246         * config/avr/avr-arch.h: Same.
29247         * config/avr/avr-devices.c: Same.
29248         * config/avr/avr-dimode.md: Same.
29249         * config/avr/predicates.md: Same.
29250         * config/avr/avr-c.c: Same.  And fix typo.
29252         * config/avr/avr-protos.h: Same.  And:
29253         (function_arg_regno_p): Rename to avr_function_arg_regno_p.
29254         (init_cumulative_args): Rename to avr_init_cumulative_args.
29255         (expand_prologue): Rename to avr_expand_prologue.
29256         (expand_epilogue): Rename to avr_expand_epilogue.
29257         (adjust_insn_length): Rename to avr_adjust_insn_length.
29258         (notice_update_cc): Rename to avr_notice_update_cc.
29259         (final_prescan_insn): Rename to avr_final_prescan_insn.
29260         * config/avr/avr.c: Same.
29261         * config/avr/avr.h: Same.
29262         * config/avr/avr.md: Remove trailing blanks.
29263         (prologue): Use avr_expand_prologue.
29264         (epilogue, sibcall_epilogue): Use avr_expand_epilogue.
29266 2013-01-14  Richard Biener  <rguenther@suse.de>
29268         * tree-cfg.c (verify_expr_location, verify_expr_location_1,
29269         verify_location, collect_subblocks): New functions.
29270         (verify_gimple_in_cfg): Verify that locations only reference
29271         BLOCKs in the functions BLOCK tree.
29273 2013-01-14  Richard Biener  <rguenther@suse.de>
29275         * tree-cfgcleanup.c (remove_forwarder_block): Unshare propagated
29276         PHI argument.
29277         * graphite-sese-to-poly.c (insert_out_of_ssa_copy): Properly
29278         unshare reference.
29279         (insert_out_of_ssa_copy_on_edge): Likewise.
29280         (rewrite_close_phi_out_of_ssa): Likewise.
29281         * tree-ssa.c (insert_debug_temp_for_var_def): Properly unshare
29282         debug expressions.
29283         * tree-ssa-pre.c (insert_into_preds_of_block): Properly unshare
29284         propagated constants.
29285         * tree-cfg.c (tree_node_can_be_shared): Handled component-refs
29286         can not be shared.
29288 2013-01-14  Georg-Johann Lay  <avr@gjlay.de>
29290         * config/avr/avr-modes.def: Add GPL copyright notice.
29292 2013-01-13  Uros Bizjak  <ubizjak@gmail.com>
29294         * config/i386/sync.md (mem_thread_fence): Mask operands[0] with
29295         MEMMODEL_MASK to determine memory model.
29296         (atomic_store<mode>): Ditto from operands[2].
29297         * config/i386/i386.c (ix86_memmodel_check): Declare "strong" as bool.
29299 2013-01-13  Jakub Jelinek  <jakub@redhat.com>
29301         PR fortran/55935
29302         * gimple-fold.c (get_symbol_constant_value): Call unshare_expr.
29303         (fold_gimple_assign): Don't call unshare_expr here.
29304         (fold_ctor_reference): Call unshare_expr.
29306 2013-01-13  Terry Guo  <terry.guo@arm.com>
29308         * Makefile.in (s-mlib): New argument MULTILIB_REUSE.
29309         * doc/fragments.texi: Document MULTILIB_REUSE.
29310         * gcc.c (multilib_reuse): New internal spec.
29311         (set_multilib_dir): Also search multilib from multilib_reuse.
29312         * genmultilib (tmpmultilib3): Refactor code.
29313         (tmpmultilib4): Ditto.
29314         (multilib_reuse): New multilib argument.
29316 2013-01-13  Richard Sandiford  <rdsandiford@googlemail.com>
29318         * Makefile.in: Update copyright.
29320 2013-01-12  Tom de Vries  <tom@codesourcery.com>
29322         PR middle-end/55890
29323         * calls.c (expand_call): Check if arg_nr is valid.
29325 2013-01-11  Michael Meissner  <meissner@linux.vnet.ibm.com>
29327         * doc/extend.texi (X86 Built-in Functions): Add whitespace in
29328         __builtin_ia32_paddb256 and __builtin_ia32_pavgb256
29329         documentation.  Add missing '__' in front of
29330         __builtin_ia32_packssdw256.
29332 2013-01-11  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29334         PR target/55719
29335         * config/s390/s390.c (s390_preferred_reload_class): Do not return
29336         NO_REGS for larl operands.
29337         (s390_reload_larl_operand): Use s390_load_address instead of
29338         emit_move_insn.
29340 2013-01-11  Richard Biener  <rguenther@suse.de>
29342         * tree-cfg.c (verify_node_sharing_1): Split out from ...
29343         (verify_node_sharing): ... here.
29344         (verify_gimple_in_cfg): Use verify_node_sharing_1 for walk_tree.
29346 2013-01-11  Eric Botcazou  <ebotcazou@adacore.com>
29348         * configure.ac (Tree checking): Set TREECHECKING to yes if enabled.
29349         Substitute TREECHECKING.
29350         * configure: Regenerate.
29351         * Makefile.in (TREECHECKING): New.
29353 2013-01-11  Richard Guenther  <rguenther@suse.de>
29355         PR tree-optimization/44061
29356         * tree-vrp.c (extract_range_basic): Compute zero as
29357         value-range for __builtin_constant_p of function parameters.
29359 2013-01-10  Richard Sandiford  <rdsandiford@googlemail.com>
29361         Update copyright years.
29363 2013-01-10  Vladimir Makarov  <vmakarov@redhat.com>
29365         PR rtl-optimization/55672
29366         * lra-eliminations.c (mark_not_eliminable): Permit addition with
29367         const to be eliminable.
29369 2013-01-10  David Edelsohn  <dje.gcc@gmail.com>
29371         * configure.ac (HAVE_AS_TLS): Add check for powerpc-ibm-aix.
29372         * configure: Regenerate.
29374 2013-01-10  Richard Biener  <rguenther@suse.de>
29376         * builtins.c (expand_builtin_init_trampoline): Use set_mem_attributes.
29378 2013-01-10  Richard Biener  <rguenther@suse.de>
29380         PR bootstrap/55792
29381         * tree-into-ssa.c (rewrite_add_phi_arguments): Do not set
29382         locations for virtual PHI arguments.
29383         (rewrite_update_phi_arguments): Likewise.
29385 2013-01-10  Joel Sherrill  <joel.sherrill@OARcorp.com>
29387         * config/v850/rtems.h (ASM_SPEC): Pass -m8byte-align and -mgcc-abi
29388         on to assembler.
29390 2013-01-10  Jakub Jelinek  <jakub@redhat.com>
29392         PR tree-optimization/55921
29393         * tree-complex.c (expand_complex_asm): New function.
29394         (expand_complex_operations_1): Call it for GIMPLE_ASM.
29396 2013-01-10  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
29398         PR target/55718
29399         * config/s390/s390.c (s390_symref_operand_p)
29400         (s390_loadrelative_operand_p): Merge the two functions.
29401         (s390_check_qrst_address, print_operand_address): Add parameters
29402         to s390_loadrelative_operand_p invokation.
29403         (s390_check_symref_alignment): Use s390_loadrelative_operand_p.
29404         (s390_reload_larl_operand, s390_secondary_reload): Use
29405         s390_loadrelative_operand_p instead of s390_symref_operand_p.
29406         (legitimize_pic_address): Handle @GOTENT and @PLT + addend.
29408 2013-01-09  Mike Stump  <mikestump@comcast.net>
29410         * dse.c (record_store): Remove unnecessary assert.
29412 2013-01-09  Jan Hubicka  <jh@suse.cz>
29414         PR tree-optimization/55569
29415         * cfgloopmanip.c (scale_loop_profile): Make ITERATION_BOUND gcov_type.
29416         * cfgloop.h (scale_loop_profile): Likewise.
29418 2013-01-09  Jan Hubicka  <jh@suse.cz>
29420         PR lto/45375
29421         * ipa-inline.c (ipa_inline): Remove extern inlines and virtual
29422         functions.
29423         * cgraphclones.c (cgraph_clone_node): Cpoy also LTO file data.
29425 2013-01-09  Richard Sandiford  <rdsandiford@googlemail.com>
29427         PR middle-end/55114
29428         * expr.h (maybe_emit_group_store): Declare.
29429         * expr.c (maybe_emit_group_store): New function.
29430         * builtins.c (expand_builtin_int_roundingfn): Call it.
29431         (expand_builtin_int_roundingfn_2): Likewise.
29433 2013-01-09  Vladimir Makarov  <vmakarov@redhat.com>
29435         PR rtl-optimization/55829
29436         * lra-constraints.c (match_reload): Add code for absent output.
29437         (curr_insn_transform): Add code for reloads of matched inputs
29438         without output.
29440 2013-01-09  Uros Bizjak  <ubizjak@gmail.com>
29442         * config/i386/sse.md (*vec_interleave_highv2df): Change mode
29443         attribute of movddup insn to DF.
29444         (*vec_interleave_lowv2df): Ditto.
29445         (vec_dupv2df): Ditto.
29447 2013-01-09  Jan Hubicka  <jh@suse.cz>
29449         PR tree-optimiation/55875
29450         * tree-ssa-loop-niter.c (number_of_iterations_cond): Add
29451         EVERY_ITERATION parameter.
29452         (number_of_iterations_exit): Check if exit is executed every iteration.
29453         (idx_infer_loop_bounds): Similarly here.
29454         (n_of_executions_at_most): Simplify
29455         to only test for cases where statement is dominated by the
29456         particular bound; handle correctly the "postdominance" test.
29457         (scev_probably_wraps_p): Use max loop iterations info
29458         as a global bound first.
29460 2013-01-09  Nguyen Duy Dat  <dat.nguyen.yn@rvc.renesas.com>
29461             Nick Clifton  <nickc@redhat.com>
29463         * config/v850/v850.md (cbranchsf4): New pattern.
29464         (cstoresf4): New pattern.
29465         (cbranchdf4): New pattern.
29466         (cstoredf4): New pattern.
29467         (movsicc): Disallow floating point comparisons.
29468         (cmpsf_le_insn): Fix order of operators.
29469         (cmpsf_lt_insn): Likewise.
29470         (cmpsf_eq_insn): Likewise.
29471         (cmpdf_le_insn): Likewise.
29472         (cmpdf_lt_insn): Likewise.
29473         (cmpdf_eq_insn): Likewise.
29474         (cmpsf_ge_insn): Use LE comparison.
29475         (cmpdf_ge_insn): Likewise.
29476         (cmpsf_gt_insn): Use LT comparison.
29477         (cmpdf_gt_insn): Likewise.
29478         (cmpsf_ne_insn): Delete pattern.
29479         (cmpdf_ne_insn): Delete pattern.
29480         * config/v850/v850.c (v850_gen_float_compare): Use
29481         gen_cmpdf_eq_insn for NE comparison.
29482         (v850_float_z_comparison_operator)
29483         (v850_float_nz_comparison_operator): Move from here ...
29484         * config/v850/predicates.md: ... to here.  Move GT and GE
29485         comparisons into v850_float_z_comparison_operator.
29486         * config/v850/v850-protos.h (v850_float_z_comparison_operator):
29487         Delete prototype.
29488         (v850_float_nz_comparison_operator): Likewise.
29490 2013-01-09  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29492         * config/pa/pa.c (pa_emit_move_sequence): Replace calls to gen_insv
29493         with calls to gen_insvsi/gen_insvdi.
29495 2013-01-09  Venkataramanan Kumar  <venkataramanan.kumar@amd.com>
29497         * config/i386/i386.c (initial_ix86_tune_features): Set up
29498         X86_TUNE_AVX128_OPTIMAL for m_BTVER2.
29500 2013-01-09  Steven Bosscher  <steven@gcc.gnu.org>
29501             Jakub Jelinek  <jakub@redhat.com>
29503         PR tree-optimization/48189
29504         * predict.c (predict_loops): If max is 0, don't call compare_tree_int.
29505         If nitercst is 0, don't predict the exit edge.
29507 2013-01-08  Naveen H.S  <Naveen.Hurugalawadi@caviumnetworks.com>
29509         * config/aarch64/aarch64.c (aarch64_print_operand): Replace %r
29510         in asm_fprintf with reg_names.
29511         (aarch64_print_operand_address): Likewise.
29512         (aarch64_return_addr): Likewise.
29513         * config/aarch64/aarch64.h (ASM_FPRINTF_EXTENSIONS): Remove.
29515 2013-01-08  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29517         * config/pa/pa.h (VAL_U6_BITS_P): Define.
29518         (INT_U6_BITS): Likewise.
29519         * config/pa/predicates.md (uint6_operand): New predicate.
29520         (shift5_operand, shift6_operand): Likewise.
29521         * config/pa/pa.md (lshrsi3, rotrsi3): Use shift5_operand instead of
29522         arith32_operand.
29523         (lshrdi3): Use shift6_operand.
29524         (shrpsi4, shrpdi4): New insn patterns.
29525         (extzv): Delete expander.
29526         (extzvsi, extzvdi): New expanders.  Use uint5_operand and uint6_operand
29527         predicates in unamed zero extract patterns.  Tighten common constraint.
29528         (extv): Delete expander.
29529         (extvsi, extvdi): New expanders.  Use uint5_operand and uint6_operand
29530         predicates in unamed sign extract patterns.  Tighten common constraint.
29531         (insv): Delete expander.
29532         (insvsi, insvdi): New expanders.  Use uint5_operand and uint6_operand
29533         predicates in unamed insert patterns.  Tighten common constraint.
29534         Change uint32_operand predicate to uint6_operand predicate in unamed
29535         DImode pattern to insert constant values of type 1...1xxxx.
29537 2013-01-04  Jan Hubicka  <jh@suse.cz>
29539         PR tree-optimization/55823
29540         * ipa-prop.c (update_indirect_edges_after_inlining): Fix ordering
29541         issue.
29543 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29544             Uros Bizjak  <ubizjak@gmail.com>
29546         PR rtl-optimization/55845
29547         * df-problems.c (can_move_insns_across): Stop scanning at
29548         volatile_insn_p source instruction or give up if
29549         across_from .. across_to range contains any volatile_insn_p
29550         instructions.
29552 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29554         * config/aarch64/aarch64-simd.md (vec_init<mode>): New.
29555         * config/aarch64/aarch64-protos.h (aarch64_expand_vector_init):
29556         Declare.
29557         * config/aarch64/aarch64.c (aarch64_simd_dup_constant,
29558         aarch64_simd_make_constant, aarch64_expand_vector_init): New.
29560 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29562         PR fortran/55341
29563         * asan.c (asan_clear_shadow): New function.
29564         (asan_emit_stack_protection): Use it.
29566 2013-01-08  Tejas Belagod  <tejas.belagod@arm.com>
29568         * config/aarch64/aarch64-simd.md (aarch64_simd_vec_<su>mult_lo_<mode>,
29569         aarch64_simd_vec_<su>mult_hi_<mode>): Separate instruction and operand
29570         with tab instead of space.
29572 2013-01-08  Nick Clifton  <nickc@redhat.com>
29574         * config/rl78/rl78.c (rl78_expand_prologue): Always select
29575         register bank 0 at the start of an interrupt handler.
29576         * config/rl78/rl78.md (mulsi3_g13): Correct values for MDBL and
29577         MDBH registers.
29579 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29581         * config/aarch64/aarch64-simd.md
29582         (aarch64_simd_bsl<mode>_internal): Add floating-point modes.
29583         (aarch64_simd_bsl): Likewise.
29584         (aarch64_vcond_internal<mode>): Likewise.
29585         (vcond<mode><mode>): Likewise.
29586         (aarch64_cm<cmp><mode>): Fix constraints, add new modes.
29587         * config/aarch64/iterators.md (V_cmp_result): Add V2DF.
29589 2013-01-08  James Greenhalgh  <james.greenhalgh@arm.com>
29591         * config/aarch64/aarch64-builtins.c
29592         (aarch64_builtin_vectorized_function): Handle sqrt, sqrtf.
29594 2013-01-08  Martin Jambor  <mjambor@suse.cz>
29596         PR debug/55579
29597         * tree-sra.c (analyze_access_subtree): Return true also after
29598         potentially creating a debug-only replacement.
29600 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29602         PR middle-end/55890
29603         * tree-ssa-ccp.c (evaluate_stmt): Use gimple_call_builtin_p.
29605         PR tree-optimization/54120
29606         * tree-vrp.c (range_fits_type_p): Don't allow
29607         src_precision < precision from signed vr to unsigned_p
29608         if vr->min or vr->max is negative.
29609         (simplify_float_conversion_using_ranges): Test can_float_p
29610         against CODE_FOR_nothing.
29612 2013-01-08  Jakub Jelinek  <jakub@redhat.com>
29613             Richard Biener  <rguenther@suse.de>
29615         PR middle-end/55851
29616         * fold-const.c (int_binop_types_match_p): Allow all INTEGRAL_TYPE_P
29617         types instead of just INTEGER_TYPE types.
29619 2013-01-07  Mark Kettenis  <kettenis@openbsd.org>
29621         * config/i386/openbsdelf.h (LIBGCC2_HAS_TF_MODE, LIBGCC2_TF_CEXT,
29622         TF_SIZE): Define.
29624 2013-01-07  Steve Ellcey  <sellcey@mips.com>
29626         PR target/42661
29627         * config/mips/mips.opt: Change mad to mmad to match documentation.
29629 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29631         PR target/55897
29632         * doc/extend.texi (AVR Named Address Spaces): __memx goes into
29633         .progmemx.data now.
29635 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29637         PR target/55897
29638         * config/avr/avr.h (ADDR_SPACE_COUNT): New enum.
29639         (avr_addrspace_t): Add .section_name field.
29640         * config/avr/avr.c (progmem_section): Use ADDR_SPACE_COUNT as
29641         array size.
29642         (avr_addrspace): Same.  Initialize .section_name.  Remove last
29643         NULL entry.  Put __memx into .progmemx.data.
29644         (progmem_section_prefix): Remove.
29645         (avr_asm_init_sections): No need to initialize progmem_section.
29646         (avr_asm_named_section): Use avr_addrspace[].section_name to get
29647         section name prefix.
29648         (avr_asm_select_section): Ditto.  And use get_unnamed_section to
29649         retrieve the progmem section.
29650         * avr-c.c (avr_cpu_cpp_builtins): Use ADDR_SPACE_COUNT as loop
29651         boundary to run over avr_addrspace[].
29652         (avr_register_target_pragmas): Ditto.
29654 2013-01-06  Jakub Jelinek  <jakub@redhat.com>
29656         * varasm.c (output_constant_def_contents): For asan_protect_global
29657         protected strings, adjust DECL_ALIGN if needed, before testing for
29658         anchored symbols.
29659         (place_block_symbol): Adjust size for asan protected STRING_CSTs if
29660         TREE_CONSTANT_POOL_ADDRESS_P.  Increase alignment for asan protected
29661         normal decls.
29662         (output_object_block): For asan protected decls, emit asan padding
29663         after their contents.
29664         * asan.c (asan_protect_global): Don't check TREE_ASM_WRITTEN here.
29665         (asan_finish_file): Test it here instead.
29667 2013-01-07  Nick Clifton  <nickc@redhat.com>
29668             Matthias Klose  <doko@debian.org>
29669             Doug Kwan  <dougkwan@google.com>
29670             H.J. Lu  <hongjiu.lu@intel.com>
29672         PR driver/55470
29673         * collect2.c (main): Support -fuse-ld=bfd and -fuse-ld=gold.
29675         * common.opt: Add fuse-ld=bfd and fuse-ld=gold.
29677         * gcc.c (LINK_COMMAND_SPEC): Pass -fuse-ld=* to collect2.
29679         * opts.c (comman_handle_option): Ignore -fuse-ld=bfd and -fuse-ld=gold.
29681         * doc/invoke.texi: Document -fuse-ld=bfd and -fuse-ld=gold.
29683 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29685         PR target/54461
29686         * doc/install.texi (Cross-Compiler-Specific Options): Document
29687         --with-avrlibc.
29689 2013-01-07  Tejas Belagod  <tejas.belagod@arm.com>
29691         * config/aarch64/arm_neon.h (vmovn_high_is16, vmovn_high_s32,
29692         vmovn_high_s64, vmovn_high_u16, vmovn_high_u32, vmovn_high_u64,
29693         vqmovn_high_s16, vqmovn_high_s32, vqmovn_high_s64, vqmovn_high_u16,
29694         vqmovn_high_u32, vqmovn_high_u64, vqmovun_high_s16, vqmovun_high_s32,
29695         vqmovun_high_s64): Fix source operand number and update copyright.
29697 2013-01-07  Richard Biener  <rguenther@suse.de>
29699         PR middle-end/55890
29700         * gimple.h (gimple_call_builtin_p): New overload.
29701         * gimple.c (validate_call): New function.
29702         (gimple_call_builtin_p): Likewise.
29703         * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
29704         Use gimple_call_builtin_p.
29705         (find_func_clobbers): Likewise.
29706         * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
29707         (strlen_optimize_stmt): Likewise.
29709 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29711         * config/aarch64/arm_neon.h (vld1_dup_*): Make argument const.
29712         (vld1q_dup_*): Likewise.
29713         (vld1_*): Likewise.
29714         (vld1q_*): Likewise.
29715         (vld1_lane_*): Likewise.
29716         (vld1q_lane_*): Likewise.
29718 2013-01-07  Richard Biener  <rguenther@suse.de>
29720         * lto-streamer.h (LTO_minor_version): Bump to 2.
29722 2013-01-07  James Greenhalgh  <james.greenhalgh@arm.com>
29724         * config/aarch64/aarch64-protos.h
29725         (aarch64_const_double_zero_rtx_p): Rename to...
29726         (aarch64_float_const_zero_rtx_p): ...this.
29727         (aarch64_float_const_representable_p): New.
29728         (aarch64_output_simd_mov_immediate): Likewise.
29729         * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>): Refactor
29730         move immediate case.
29731         * config/aarch64/aarch64.c
29732         (aarch64_const_double_zero_rtx_p): Rename to...
29733         (aarch64_float_const_zero_rtx_p): ...this.
29734         (aarch64_print_operand): Allow printing of new constants.
29735         (aarch64_valid_floating_const): New.
29736         (aarch64_legitimate_constant_p): Check for valid floating-point
29737         constants.
29738         (aarch64_simd_valid_immediate): Likewise.
29739         (aarch64_vect_float_const_representable_p): New.
29740         (aarch64_float_const_representable_p): Likewise.
29741         (aarch64_simd_imm_zero_p): Also allow for floating-point 0.0.
29742         (aarch64_output_simd_mov_immediate): New.
29743         * config/aarch64/aarch64.md (*movsf_aarch64): Add new alternative.
29744         (*movdf_aarch64): Likewise.
29745         * config/aarch64/constraints.md (Ufc): New.
29746         (Y): call aarch64_float_const_zero_rtx.
29747         * config/aarch64/predicates.md (aarch64_fp_compare_operand): New.
29749 2013-01-07  Richard Biener  <rguenther@suse.de>
29751         PR tree-optimization/55888
29752         PR tree-optimization/55862
29753         * tree-ssa-pre.c (phi_translate_1): Revert previous change.
29754         (valid_in_sets): Check if a NAME has a leader in AVAIL_OUT,
29755         not if it is contained therein.
29757 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29759         * config/avr/t-avr: Typo.
29761 2013-01-07  Georg-Johann Lay  <avr@gjlay.de>
29763         PR55243
29764         * config/avr/t-avr: Don't automatically rebuild
29765         $(srcdir)/config/avr/t-multilib
29766         $(srcdir)/config/avr/avr-tables.opt
29767         $(srcdir)/doc/avr-mmcu.texi
29768         (avr-mcus): New phony target to build them on request.
29769         (s-avr-mlib, s-avr-mmcu-texi): Remove.
29770         * avr/avr-mcus.def: Adjust comments.
29772 2013-01-07  Uros Bizjak  <ubizjak@gmail.com>
29774         * config/i386/i386.c (DEFAULT_PCC_STRUCT_RETURN): Remove.
29776 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29778         * file-find.c, file-find.h, realmpfr.c: Add FSF as copyright holder.
29780 2013-01-06  Richard Sandiford  <rdsandiford@googlemail.com>
29782         * config/tilepro/gen-mul-tables.cc: Put copyright on one line.
29784 2013-01-05  David Edelsohn  <dje.gcc@gmail.com>
29786         * config/rs6000/aix53.h (LIB_SPEC): Add -lpthreads when compiling
29787         to generate profiling.
29788         * config/rs6000/aix64.h (LIB_SPEC): Same.
29790 2013-01-04  Andrew Pinski  <apinski@cavium.com>
29792         * config/aarch64/aarch64.c (aarch64_fixed_condition_code_regs):
29793         New function.
29794         (TARGET_FIXED_CONDITION_CODE_REGS): Define.
29796 2013-01-04  Uros Bizjak  <ubizjak@gmail.com>
29798         * config/i386/i386.c (ix86_legitimize_address): Call convert_to_mode
29799         unconditionally.
29800         (ix86_expand_move): Ditto.
29801         (ix86_zero_extend_to_Pmode): Ditto.
29802         (ix86_expand_call): Ditto.
29803         (ix86_expand_special_args_builtin): Ditto.
29804         (ix86_expand_builtin): Ditto.
29806 2013-01-04  Richard Biener  <rguenther@suse.de>
29808         PR tree-optimization/55862
29809         * tree-ssa-pre.c (phi_translate_1): Valueize SSA names after
29810         translating them through PHI nodes.
29812 2013-01-04  Martin Jambor  <mjambor@suse.cz>
29814         PR tree-optimization/55755
29815         * tree-sra.c (sra_modify_assign): Do not check that an access has no
29816         children when trying to avoid producing a VIEW_CONVERT_EXPR.
29818 2013-01-04  Marek Polacek  <polacek@redhat.com>
29820         PR middle-end/55859
29821         * opts.c (default_options_optimization): Clarify error message.
29823 2013-01-04  Richard Biener  <rguenther@suse.de>
29825         PR middle-end/55863
29826         * fold-const.c (split_tree): Undo -X - 1 to ~X folding for
29827         reassociation.
29829 2013-01-03  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29831         PR target/53789
29832         * config/pa/pa.md (movsi): Revert previous change.
29833         * config/pa/pa.c (pa_legitimate_constant_p): Reject all TLS symbol
29834         references.
29836 2013-01-03  Richard Henderson  <rth@redhat.com>
29838         * config/i386/i386.c (ix86_expand_move): Always assign to op1
29839         after eliminating TLS symbols.
29841 2013-01-03  Marc Glisse  <marc.glisse@inria.fr>
29843         PR bootstrap/50167
29844         * graphite-interchange.c (pdr_stride_in_loop): Use gmp_fprintf.
29845         * graphite-poly.c (debug_gmp_value): Likewise.
29847 2013-01-03  Uros Bizjak  <ubizjak@gmail.com>
29849         PR target/55712
29850         * config/i386/i386-c.c (ix86_target_macros_internal): Depending on
29851         selected code model, define __code_mode_small__, __code_model_medium__,
29852         __code_model_large__, __code_model_32__ or __code_model_kernel__.
29853         * config/i386/cpuid.h (__cpuid, __cpuid_count) [__i386__]: Prefix
29854         xchg temporary register with %k.  Declare temporary register as
29855         early clobbered.
29856         [__x86_64__]: For medium and large code models, preserve %rbx register.
29858 2013-01-03  Richard Biener  <rguenther@suse.de>
29860         * tree-data-ref.c (dump_conflict_function): Use less vertical spacing.
29861         (dump_subscript): Adjust.
29862         (finalize_ddr_dependent): Do not dump redundant info.
29863         (analyze_siv_subscript): Adjust.
29864         (subscript_dependence_tester): Likewise.
29865         (compute_affine_dependence): Likewise.
29867 2013-01-03  Richard Biener  <rguenther@suse.de>
29869         Revert
29870         2013-01-03  Richard Biener  <rguenther@suse.de>
29872         PR tree-optimization/55857
29873         * tree-vect-stmts.c (vectorizable_load): Do not setup
29874         re-alignment for invariant loads.
29876         2013-01-02  Richard Biener  <rguenther@suse.de>
29878         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29879         invariant load do not generate a vector load from the scalar location.
29881 2013-01-03  Richard Biener  <rguenther@suse.de>
29883         * tree-vect-loop.c (vect_analyze_loop_form): Clarify reason
29884         for not vectorizing.
29885         * tree-vect-data-refs.c (vect_create_addr_base_for_vector_ref): Do
29886         not build INDIRECT_REFs, call get_name once only.
29887         (vect_create_data_ref_ptr): Likewise.  Dump base object kind
29888         based on DR_BASE_OBJECT, not DR_BASE_ADDRESS.
29890 2013-01-03  Richard Biener  <rguenther@suse.de>
29892         PR tree-optimization/55857
29893         * tree-vect-stmts.c (vectorizable_load): Do not setup
29894         re-alignment for invariant loads.
29896 2013-01-03  Richard Biener  <rguenther@suse.de>
29898         PR lto/55848
29899         * lto-symtab.c (lto_symtab_merge_decls_1): As last resort, always
29900         prefer a built-in decl.
29902 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29904         * gcc.c (process_command): Update copyright notice dates.
29905         * gcov.c (print_version): Likewise.
29906         * gcov-dump.c (print_version): Likewise.
29908         PR rtl-optimization/55838
29909         * loop-iv.c (iv_number_of_iterations): Call lowpart_subreg on
29910         iv0.step, iv1.step and step.
29912 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29913             Marc Glisse  <marc.glisse@inria.fr>
29915         PR tree-optimization/55832
29916         * fold-const.c (fold_binary_loc): For ABS_EXPR<x> >= 0 and
29917         ABS_EXPR<x> < 0 folding use constant_boolean_node instead of
29918         integer_{one,zero}_node.
29920 2013-01-03  Jakub Jelinek  <jakub@redhat.com>
29922         PR debug/54402
29923         * params.def (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE): New param.
29924         * var-tracking.c (reverse_op): Don't add reverse ops to
29925         VALUEs that have already
29926         PARAM_VALUE (PARAM_MAX_VARTRACK_REVERSE_OP_SIZE) or longer locs list.
29928 2013-01-02  Gerald Pfeifer  <gerald@pfeifer.com>
29930         * doc/contrib.texi: Note years as release manager for Mark Mitchell.
29932 2013-01-02  Teresa Johnson  <tejohnson@google.com>
29934         * dumpfile.c (dump_loc): Print filename with location.
29935         * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
29936         new location_t parameter to emit complete unroll message with
29937         new dump framework.
29938         (canonicalize_loop_induction_variables): Compute loops location
29939         and pass to try_unroll_loop_completely.
29940         * loop-unroll.c (report_unroll_peel): New function.
29941         (peel_loops_completely): Use new dump format with location
29942         for main dumpfile message, and invoke report_unroll_peel on success.
29943         (decide_unrolling_and_peeling): Ditto.
29944         (decide_peel_once_rolling): Remove old dumpfile message subsumed
29945         by report_unroll_peel.
29946         (decide_peel_completely): Ditto.
29947         (decide_unroll_constant_iterations): Ditto.
29948         (decide_unroll_runtime_iterations): Ditto.
29949         (decide_peel_simple): Ditto.
29950         (decide_unroll_stupid): Ditto.
29951         * cfgloop.c (get_loop_location): New function.
29952         * cfgloop.h (get_loop_location): Declare.
29954 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29956         * config/i386/i386.c (fold_builtin_cpu): Remove unnecessary checks for
29957         NULL.
29959 2013-01-02  John David Anglin  <dave.anglin@nrc-cnrc.gc.ca>
29961         PR middle-end/55198
29962         * expr.c (expand_expr_real_1): Don't use bitfield extraction for non
29963         BLKmode objects when EXPAND_MEMORY is specified.
29965 2013-01-02  Sriraman Tallam  <tmsriram@google.com>
29967         * config/i386/i386.c (ix86_get_function_versions_dispatcher): Fix bug
29968         in loop predicate.
29969         (fold_builtin_cpu): Do not share cpu model decls across statements.
29971 2013-01-02  Jason Merrill  <jason@redhat.com>
29973         PR c++/55804
29974         * tree.c (build_array_type_1): Revert earlier change.
29976 2013-01-02  Yufeng Zhang  <yufeng.zhang@arm.com>
29978         * config/aarch64/aarch64-cores.def: Add entries for "cortex-a53" and
29979         "cortex-a57".
29980         * config/aarch64/aarch64-tune.md: Re-generate.
29982 2013-01-02  Richard Biener  <rguenther@suse.de>
29984         * tree-vect-stmts.c (vectorizable_load): When vectorizing an
29985         invariant load do not generate a vector load from the scalar location.
29987 2013-01-02  Richard Biener  <rguenther@suse.de>
29989         PR bootstrap/55784
29990         * configure.ac: Add $GMPINC to CFLAGS/CXXFLAGS.
29991         * configure: Regenerate.
29993 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
29995         * builtins.c (expand_builtin_mathfn, expand_builtin_mathfn_2)
29996         (expand_builtin_mathfn_ternary, expand_builtin_mathfn_3)
29997         (expand_builtin_int_roundingfn_2): Keep the original target around
29998         for the fallback case.
30000 2013-01-02  Richard Sandiford  <rdsandiford@googlemail.com>
30002         * tree-vrp.c (range_fits_type_p): Require the MSB of the double_int
30003         to be clear for sign changes.
30005 2013-01-01  Jan Hubicka  <jh@suse.cz>
30007         * ipa-inline-analysis.c: Fix formatting.
30009 2013-01-01  Jakub Jelinek  <jakub@redhat.com>
30011         PR tree-optimization/55831
30012         * tree-vect-loop.c (get_initial_def_for_induction): Use
30013         gsi_after_labels instead of gsi_start_bb.
30015 Copyright (C) 2013 Free Software Foundation, Inc.
30017 Copying and distribution of this file, with or without modification,
30018 are permitted in any medium without royalty provided the copyright
30019 notice and this notice are preserved.